Structural and functional analysis of the accessory gene regulators of Staphylococcus aureus and Staphylococcus epidermidis: an in Silico approach

Abstract

Background

Staphylococcus aureus and Staphylococcus epidermidis are tenacious pathogens that cause toxic shock syndrome. Accessory gene regulator (Agr) of Staphylococcus sp. controls the expression of multiple genes that encode virulence properties. Evolutionary covariance of accessory gene regulators of selected strains of two Staphylococcus sp. was entrenched through multiple sequence alignment, relative synonymous codon usage, codon adaptation index and compositional analysis. Artificial intelligence and machine learning based AlphaFold and TrRosetta were used to determine the tertiary structures of the proteins. Structure-based ab initio models could forecast subcellular localization, domain length, molecular docking, and simulation of Agrs in the isolates belonging to Staphylococcus sp.

Results

AT ending codons are preferred over GC ending codons. Besides, the mutational pressure has been found to be one of the causative factors in shaping the codon usage biasness. Topological investigations reveal the existence of AgrA and AgrD in the cytosol, while AgrB and AgrC to reside in the cellular membrane. All Agrs are acidic and stable, except AgrB. Secondary structural studies showed that Agrs mostly consist of α-helix followed by random coils that preferentially remain in the transmembrane region. Protein-protein docking studies using the HDOCK server demonstrated that AgrA has stronger binding affinity with AgrC in S. epidermidis isolates than the same in S. aureus. By analysing the docking potential of AgrB and AgrD, it has been found that S. aureus possesses higher docking score than S. epidermidis.

Conclusion

Such compressive investigations could provide crucial insights into the structural features of Agrs in S. aureus and S. epidermidis, that are actively implicated in quorum sensing signalling-mediated virulence factor regulation and help in the identification of new Agr-dependent quorum sensing inhibitors. AgrA and AgrC are, therefore, appear to be seemingly promising in the management of bacterial infections and are apprehended to be useful therapeutic targets for the discovery of potential antimicrobial drugs.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12866-025-04257-0.

Introduction

Nosocomial infections caused by Staphylococcus aureus and Staphylococcus epidermidis, two of the most common bacteria found in healthcare settings, are becoming increasingly threatening. S. aureus is one of the most lethal bacterium that infects and colonizes healthy people with weakened immune systems as well as the hospitalized patients with reduced immunity [1, 2]. Conventional concept holds that S. epidermidis is a harmless commensal bacterium found on human skin; however, contemporary research works demonstrate that S. epidermidis can also cause bacteraemia and severe sepsis [3]. One of the most common causes of illnesses is their inclination for obstinate adherence and long term persistence on medical devices, and that directs the bacterium to continue its infectivity and subsequent pathogenicity [4]. At least one billion dollar is spent annually by the public health system in the US on S. epidermidis borne infections. Biofilms, which have mosaic polysaccharide structures consisting microbe agglomerations, are known to be responsible for its ability to remain on medical devices [5]. A heterogeneous matrix is formed by the biofilm, that can shield bacteria from the effects of antibiotic treatment, physiologic stress, and possibly even the immune responses from the host [6]. These biofilms can noncovalently interact with host tissue or host proteins in course of infection and as a result, they come across the surfaces of the medical devices and get attache adhered onto their surfaces. To differentiate the harmful bacteria from the normal microbiome, researchers have identified a number of genes that can affect biofilm morphologies, leading subsequently to pathogenicity [7]. One of such factors of Staphylococcus sp. is the accessory gene regulator (Agr), known to regulate quorum sensing that eventually control different virulent gene expressions of these organisms [8]. P2 and P3 promoters govern the two neighbouring transcripts RNAII and RNAIII, respectively, that make up the agr locus. The second RNA subunit consists of the agrB, agrD, agrC, and agrA genes [9]. The autoinducing peptide (AIP), a pro-peptide, is encoded by agrD., while the AIP processor is encoded by the agrB gene. The two-component regulatory system comprised of AgrC and AgrA is encoded by the genes agrC and agrA [10]. The AIP pro-peptide is converted to octapeptide by AgrB and is released into the extracellular space upon Agr activation. The membrane-bound histidine kinase AgrC undergoes autophosphorylation and activation when AIP approaches a threshold, causing AgrA, its cognate response regulator, to be phosphorylated [11]. Phosphorylated AgrA increases AIP production by activating its own RNAII transcript and promoter P3 to enhance RNAIII expression. Pathogenicity is caused by the two primary intracellular effectors, AgrA and RNAIII, which control the expression of virulence factors [12]. Role of a functional Agr in inducing disease in infection models in animals has been extensively studied [11]. Agr malfunction has been linked to thrombin-induced platelet microbicidal protein, reduced sensitivity to vancomycin, and prolonged bacteraemia. In regard to biofilm capacity, Agr dysfunctional strains typically have a larger capacity [13]. The present study focuses on comparing the structural and functional properties of the Agr system between CoNS (S. epidermidis) and CoPS (S. aureus), as they are the most prominent staphylococcal species on the human skin. S. epidermidis accounts for over 90% of the resident aerobic CoNS [14], while S. aureus makes up to 80% of the CoPS population [15]. Since quorum sensing (QS) is a crucial mechanism for microbes, several scientific investigations have been carried out to block QS signaling. Developing methods to inhibit Agrs for supressing QS-mediated virulence features is one of the most promising targets for anti-QS therapies. Moreover, the fact that the toxins produced by AgrA activations promote antibiotic resistance and pathogenecity qualifies Agrs to be an ideal therapeutic target for the management of QS-mediated staphylococcal infections [16, 17]. Thorough studies on the Agr protein in these two medically significant species of bacterial pathogen are required to address the virulence as well as fatality caused by Staphylococci sp. since Agr is critical for staphylococcal infection. Especially the structural analyses of AgrA and AgrC of Staphylococcus sp. have been carried out [18]. However, to solve the entire issue, wider approaches are needed. It is apprehended that computational studies may provide considerably reasonable, reliable as well as prompt solution and most importantly may thrust upon exclusive genomic and proteomic records that happens to be the foundation of every behaviour for living [18]. In the rapidly developing field of artificial intelligence, in silico molecular profiling is significant in terms of gathering information and analyses with the objective to finally make molecular correlations [19]. Agrs have recently been spotted as a mean that can open-up prospective therapeutic avenues in combating biofilm-mediated microbial infections. It is mostly due to conserved architecture and functional modalities in S. aureus and S. epidermidis. Moreover, both these species are well-known models for studying quorum sensing signalling mechanism [20]. Targeting Agr with natural bioactive compounds has been shown in previous studies on S. aureus, including reports from the current research group, to regulate QS quality and biofilm development. Additionally, the current research team is actively investigating a variety of S. aureus issues, such as molecular epidemiology, biofilm signaling, anti-QS development, anti-biofilm tactics, etc [2, 21–23]. It is believed that theoretical techniques can effectively provide distinctive understanding on the fundamental mechanisms prior to laboratory operations, which would ultimately be useful in more effective structuring of the experimental setup. Because some of these Agrs are thought to be transmembrane, it is particularly challenging to do typical X-ray crystallographic research. Since significant intermolecular interactions are required for crystallization, membrane proteins that have been isolated in detergent-bound form provide challenges. Effective protein interactions are hampered by the repulsions of the purification detergents. NMR spectroscopy has the drawback of only being able to study smaller protein (12KDa). It is challenging to study proteins with molecular weights more than 25 kDa using NMR spectroscopy alone. Here, bioinformatic approaches can offer a way of finding out a trustworthy, efficient, and quick solution to this issue [24].

In this study we made an attempt to substantiate certain specific laboratory experiments. Performing experiments in a wet laboratory is a cumbersome process, which can be minimized by using preconceived ideas obtained through theoretical investigation. Therefore, the propositions derived from such theoretical studies can be further substantiated with a justifiable amount of wet lab work. This study limited our focus on the available and most relevant clinical isolates that could readily satisfy our primary objectives in relation to desirable features that also includes obtaining the preliminary ideas from selected isolated strains that are known to be pathogenic by virtue of the previously reported results. The present work aims to analyse and compare the physico-chemical and structural properties of four Agrs, viz., AgrA, AgrB, AgrC, and AgrD, of S. epidermidis ATCC 35984, S. aureus MRSA 252, S. aureus MSSA 476, and S. aureus NCTC 8325 in addition to protein-protein docking studies. The present research group is engaged to understand the molecular mechanisms of biofilm formation and quorum sensing as underlying mystery for the infection caused by two bacterial pathogens, S. aureus and S. epidermidis, beside other bacteria. In this context, development of comprehensive theoretical knowledge by exploiting in silico approaches, that are providentially available at present time, are considered to be extremely significant and applicable. In silico approaches involving the Agr still remain extremely important as it is capable in generating information on the structure and its interactive modalities [21]. Additionally, the docking analyses involving Agr appear most significant as this data-sets are apprehended to be most eloquent and contributory towards drug development approaches targeting management of such life-threatening infectious diseases in human [25].

Methods and materials

Data collection

Amino acid sequence and taxonomic data of four Agr proteins of selected strains of S. aureus (AgrA SA, AgrB SA, AgrC SA, AgrD SA) and S.epidermidis (AgrA SE, AgrB SE, AgrC SE, AgrD SE) were retrieved from UniProt (http://www.uniprot.org) and NCBI (http://www.ncbi.nlm.nih.gov) in FASTA configuration [26]. The genomic and amino acid sequences of Agrs of S. epidermidis ATCC 35984 (isolated during a catheter-associated sepsis outbreak between 1979 and 1980) [27] were analyzed. In case of S. aureus, the following strains were taken into account: MRSA 252 (isolated in 1997 from a 64-year-old female with bacteraemia), MSSA 476 (isolated in 1998 from a 9-year-old boy with primary upper tibial osteomyelitis), and NCTC 8325 (isolated from a sepsis patient in 1960). S. aureus NCTC 8325, despite its origin, is predominantly known and utilized as a laboratory reference strain, having undergone extensive passaging [28]. All the strains were clinical isolates [27, 28] and were commonly used as positive control strains in a variety of research contexts, especially in investigations pertaining to biofilm formation, antibiotic resistance, and QS-mediated pathogenicity evaluations [29–32]. All these strains were found to develop resistance to several commonly used antibiotics and were isolated from clinical settings. Using an agr-dependent QS mechanism, each of these strains controls the initial cell attachment and cell autolysis during biofilm development that produced RNAIII, which is also responsible for encoding the δ-toxin. They were also found to be capable of causing a range of infections, including those related to catheters [32–35].

Multiple sequence alignment studies (MSA)

MSA of different Agrs of S. aureus and S. epidermidis were carried out using ClustalW module of MEGA6.06 programme with default parameters [36]. The MEGA program, which is optimised for repetitive and merged sequence investigations, is designed for comparative analysis of nucleotide and amino acid sequences with the goal of deducing genetic evolutionary trends in different species across time [37]. Jalview can visualise MSA using a single letter amino acid logo [38]. Patterns of alignment can be imported using a number of widely used FASTA formats or obtained from public databases. It enables the importation of alignments produced by alternative applications. Additionally, Jalview’s integrated analytic methods assist in calculating the consensus for every alignment column, which is shown as a histogram beneath the alignment [38].

Physicochemical properties

Different physicochemical properties like aliphatic index (AI), molar extinction coefficient (EC), grand average of hydropathicity (GRAVY), isoelectric point (PI), instability index (II), molecular weight (MW), positive (+ R) and negative (- R) charge of different Agr proteins were analyzed by submitting their FASTA format amino acid sequences, downloaded from the NCBI database, to the ExPASy-ProtParam server (http://web.expasy.org/protparam/) [39]. This service appears with great implications as it can provide certain activities related to proteomics research, such as primary, secondary, and tertiary architecture evaluation [39].

Compositional heterogeneity based on codon usage

Several genomic characteristics, including GC content and relative synonymous codon usage (RCSU), were examined by CAIcal server (https://genomes.urv.es/CAIcal/) [40]. This online tool was used to create A3%, T3%, G3%, C3%, GCs (GC1%, GC2%, GC3%, GC%), and number of codons (Nc) values of Agr in S. aureus and S. epidermidis. The initial and last codons were removed from the computation to limit the impact on codon usage. The RSCU value for each codon was calculated using the method shown below:

1

where, X_ij is the frequency of the j^th codon for the i^th amino acid, and n_i is the number of codons for the i^th amino acid (i^th codon family). The effective number of codons (ENc) was used to calculate the overall frequencies of GC content at the first, second, and third codon places for respective gene sequences to ensure entire synonymous codon usage. The ENc value typically has a strong bias in codon usage and vice versa. An ENc and GC3s plot was generated to emphasise the role of a dominant mutation on codon usage patterns [41]. The expected ENc values for each GC3 were determined utilising the equation shown below.

2

where, ‘s’ denotes GC content at the third codon location [41].

Codon adaptation analysis

Adaptability of the Agr codons to S. aureus and S. epidermidis, were assessed by codon adaptation index (CAI). It was computed by the CAIcal service (https://genomes.urv.es/CAIcal/) [40]. S. aureus and S. epidermidis reference datasets were obtained from the Codon Usage Database (https://www.kazusa.or.jp/codon/) [40]. The original approach put out by Sharp and Li is used to compute CAI [42]. The CAI value ranges from 0 to 1, a greater value is indicative of better adaptation likelihood and gene compatibility in the genome [42]. Detailed procedure has been described in supplementary section.

Subcellular localization and membrane topology

Subcellular localization of different Agrs was analyzed by CELLO v.2.5 (http://cello.life.nctu.edu.tw/) and PSORTb v 3.0.3 (https://www.psort.org/psortb/results.pl) [43, 44]. For both servers, the organism type (Gram-negative or Gram-positive) was initially selected, followed by the submission of query sequences in FASTA format. PSORTb predicts subcellular localization using a 5-fold cross-validation approach to ensure robust and accurate results. Prediction of transmembrane parts and topology of four proteins from the two different species were performed by using different online servers like, TOPCONS (https://topcons.cbr.su.se/) [45], CCTOP (https://cctop.ttk.hu/) [46], Phobius (https://phobius.sbc.su.se/) [47], Octopus (https://github.com/cloudflare/octopus) [48]. Besides, 3D structure of transmembrane part of different Agrs were visualized by using Deep TMHMM (https://dtu.biolib.com/DeepTMHMM) [49]. Presence of signal peptide examined through Signal P v5.0 web server and Deep TMHMM [50].

Crystallization probability

The forecast was produced by adding each crystallisation probabilities to get a single crystallisation score. Based on this score, the protein is given to one of five crystallisation classes: ideal, suboptimal, median, challenging, and extremely challenging. The score of different Agrs from both species was generated by using protein sequences in XtalPred online server (https://xtalpred.godziklab.org/XtalPred-cgi/xtal.pl) [43]. The analysis involved entering the query sequences of amino acid residues in FASTA format and choosing the option “Find close bacterial homologs more likely to crystallize”. XtalPred was evaluated by reexamining the selection of targets from 271 Pfam families that the Joint Center for Structural Genomics had targeted in PSI-2. It was calculated that 30% fewer targets could have been added to the pipeline for protein synthesis and crystallization without reducing the number of families [43].

Secondary structure analysis

Secondary structures (α-helix, β-turn, ramdom coil, extended strands) of Agrs were evaluated through GOR IV (https://npsa-prabi.ibcp.fr/cgi-bin/npsa_automat.pl?page=/NPSA/npsa_gor4.html) [51] and ExPASy SOPMA server (https://npsa-prabi.ibcp.fr/cgi-bin/npsa_automat.pl?page=/NPSA/npsa_sopma.html) [52]. Amino acid sequences were submitted in FASTA format with an output width set to 70 for the analysis. The strategy depends on data science and the idea that all data collected from individual and couple of amino acid residues may be used to estimate the details of the constitutes in a protein. Because of their straightforward physical presumptions The GOR IV and ExPASy SOPMA server have a theoretical edge among various systems like PSIPRED, PHD, and others that rely on algorithms for machine learning [52]. Percentage of amino acid participation in secondary structure was also analyzed by ExPASy-ProtParam server (http://web.expasy.org/protparam/) [39]. Detailed procedure has been described in supplementary section.

Tertiary structure prediction and validation

Tertiary structure of Agrs were derived by using TrRosetta v1.14 (transform-restrained Rosetta; https://yanglab.qd.sdu.edu.cn/trRosetta/) [53] and AlphaFold2 (https://alphafold.ebi.ac.uk/) [54]. The sequences were submitted in FASTA format, excluding the selection of ‘Do not use templates’ and ‘trRosettaX-Single’ options in TrRosetta. The AlphaFold2 protein structure database was utilized in this investigation with its default parameters [54]. TrRosetta v1.14, developed after the 13th Critical Assessment of Protein Structure Prediction (CASP13), uses Rosetta and deep learning to quickly and accurately predict protein structures using default settings. The deep neural network-predicted inter-residue distance and orientation ranges are among the constraints [53]. AlphaFold, DeepMind’s deep learning system, has shared 3D representations with the research community. It is a digital architecture influenced by neuronal circuitry that detects patterns in data [55]. Results from above mentioned server validated through PROCHECK (https://saves.mbi.ucla.edu/) [56] and ERRAT (https://www.doe-mbi.ucla.edu/errat/) [55]. It aids in determining the stereochemical properties of a structure of proteins by creating a series of PostScript plots that analyse its overall and residue-by-residue geometry through generating Ramachandran plot [57]. The improved models were checked using assessment tool, ProSA (https://prosa.services.came.sbg.ac.at/prosa.php) [58]. The Protein Tools server was utilised to identify salt bridges in the predicted structure (https://proteintools.uni-bayreuth.de/) [59].

Flexibility modelling of tertiary structure

Agrs structural flexibility was checked by CABS-flex 2.0 (https://biocomp.chem.uw.edu.pl/CABSflex2) [60]. CABS-flex simulates the dynamics of proteins through a CABS coarse-grained structural framework. The CABS system utilises Monte Carlo dynamics and the asymmetric Metropolis method, which meets the conditions of microscopic reversibility and Boltzmann distribution among produced ensembles. CABS-flex is a complement to existing effective methods for creating structural residue volatility identities, which include sequence-based forecasters that identify protein’s chaotic areas or other coarse-grained techniques. It included over 50 cycles and 50 trajectory segments in 10 ns each, as well as several additional length constraints, such as a universal mass of 1.0. The flexibility of the structures was presented through root-mean-square fluctuations (RMSF). In the standard setting, the CABS-flex employs a set of range constraints and simulation settings provided in the job [60].

Domain and motif evaluation

Different functional domains of the studied proteins were evaluated through InterProScan online tool (https://www.ebi.ac.uk/interpro/search/sequence/) [61]. This server helps to examine the amino acid sequences for matches to the InterPro protein signature libraries. In this chosen method by scanning every record of files, entry data, and program outcomes, the Perl cataloguing framework is used in this server to facilitate simple searching and storage of the basic IDs, project accomplishments, and their current condition. Individuals may view the complete or any particular entry data from the webpage [62]. Amino acid residues were also evaluated through conserved domain search tool (https://www.ncbi.nlm.nih.gov/Structure/cdd/cdd.shtml) in expectation of Pfam data server. Motif of the studied proteins were greeted by MEME server (Multiple Em for Motif Elicitation; https://meme-suite.org/meme/tools/meme) [63].

Protein-protein docking

Docking investigations between AgrA-AgrC and AgrB-AgrD of S. aureus and S. epidermidis were carried out by utilising a protein-protein docking server, HDOCK (http://hdock.phys.hust.edu.cn/) [64] and LZerD (https://kiharalab.org/proteindocking/lzerd.php) [65, 66]. Afterwords, interfacing residues were visualized by PyMOL application [67], PDBsum server (https://www.ebi.ac.uk/thornton-srv/databases/pdbsum/) [68], and LigPlot [69]. An online service called PDBsum offers structural details about the data contained in the Protein Data Bank (PDB). The studies include structural quality of PROCHECK evaluations and graphics-based protein-protein interactions [68]. Binding affinity ΔG (kcal/mol) and dissociation constant (K_d) at 37 °C were studied through the PRODIGY website (https://bianca.science.uu.nl/prodigy/) [70]. Detailed procedure has been described in supplementary section.

Molecular dynamics (MD) simulation studies

MD simulation was performed on server for better understanding on the behaviour of biomolecules and to determine the steadiness of protein-protein docking. To investigate the interaction reliability and adaptability of the protein-protein complex, normal mode analysis was performed using the iMODS (https://imods.iqfr.csic.es/) tool, that generates trajectories using Monte Carlo sampling for 1 atm pressure at 300 K temperature [71] and an elastic network model in internal coordinates. PDB structures were submitted to the iMODS online tool, and outcomes were shown with all by default settings. A 50 ns MD simulation was performed on AgrA-AgrC and AgrB-AgrD docked structures of chosen S. aureus and S. epidermidis strains [72, 73]. Detailed procedure has been described in supplementary section.

Results and discussion

Multiple sequence alignment (MSA) analysis

MSA study helps in classifying the proteins into three regions, viz., fully-conserved, semi-conserved, and non-conserved regions. These locations were investigated in terms of their ancestral evolution including potential protein mutations [38]. MSA studies on distinct Agrs were performed using a single letter symbol for each amino acid residue. Figures S1-3 (supplementary section) demonstrate the MSA of non-conserved and conserved amino acid residues in Agrs of S. aureus and S. epidermidis. AgrA of both the Staphylococcus sp. contain semi-conserved (1^st to 72^nd, 114^th to 141^st, and 190^th to 230^th amino acid residues) and fully-conserved regions (73^rd to 113^th a^nd 142^nd to 189^th amino acid residues) as shown in Figure S1a-b. Only 19^th to 40^th amino acid residues of AgrB were semi-conserved and other parts were non-conserved (Figure S1c-d). Amino acid sequences of AgrC and AgrD from both Staphylococcus sp. were non-conserved (Figure S2a-d and Figure S3). Studied sequences are considered to be variable throughout all Agrs except AgrA. Similar findings have also been reported earlier [74]. N-terminal receiver domains of AgrA of both the species share similarity with the consensus sequence [75]. Two amino acids, His77 and Cys84, were identified to be essential for AgrB proteolytic action [76]. AgrC of S. epidermidis and S. aureus share a similarity of 50%; the homology discovered for the N-terminal and C-terminal area is 28 and 67%, respectively [75]. N-terminal part of AgrB is non-conserved but in case of AgrA it is conserved. AgrA has the most conserved amino acid sequences among all Agrs in both Staphylococcus sp.

Physicochemical properties

Comparative studies on different physiochemical characteristics and amino acids throughout the four Agrs from the selected strains of two distinct species, viz., S. aureus and S. epidermidis were carried out. Table 1 summarizes the results of the investigation of physiochemical characteristics with unique UniProt ID. Number of amino acid residues is same for AgrA and AgrD, i.e., 238 and 46, respectively, in both the species but number of amino acid residues are higher for AgrC (430 amino acid residues) in S. epidermidis and AgrB (194 amino acid residues) in S. aureus. Irrespective of similar length of amino acid residues, all Agrs have different molecular weights. AgrD shows lowest molecular weight; 5130.92 Da for S. epidermidis and 5243.09 Da forS. aureus. In contrast, AgrC shows highest molecular weight of 50252.59 Da for S. epidermidis and 49677.64 Da for S. aureus. Theoretical isoelectric points (pI) of AgrB from strains of both the species, 9.71 in S. aureus and 9.94 in S. epidermidis were greater than 7.00 but other Agrs were found within the range of 4.25 to 6.93.

Table 1.

Physicochemical properties of accessory gene regulators, AgrA, AgrB, AgrC, and AgrD of S. aureus and S. epidermidis

Agr	UniProt ID	No. of amino acids	MW/Da	pI	(-) R	(+) R	ε/M−1cm−1	II	AI	GRAVY
Staphylococcus aureus
AgrA	Q6GF33	238	27905.90	5.78	37	31	15,150	36.25	91.30	−0.379
AgrB	Q6G7S1	187	21801.24	9.71	8	18	23,380	45.06	132.46	0.621
AgrC	B5U0V1	430	49677.64	5.19	45	38	37,040	39.76	124.67	0.501
AgrD	B5U0V0	46	5243.09	4.25	7	2	1490	28.94	118.70	0.337
Staphylococcus epidermidis
AgrA	Q5HMY2	238	27936.97	5.54	39	31	13,660	31.20	91.72	−0.354
AgrB	P0C0N4	194	22573.40	9.94	9	23	12,045	46.77	141.70	0.626
AgrC	Q5HMY3	429	50252.59	6.93	47	47	42,290	35.68	127.39	0.354
AgrD	Q8VSX6	46	5130.92	4.36	6	3	2980	24.02	101.74	−0.046

Theoretical pI values reveal that all Agrs except AgrB are acidic in nature. Numbers of positive charge residues were found higher in AgrB and AgrC, whereas, negative charge residues are higher on AgrA and AgrD for both the species. For values of molar extinction coefficient (ε) for S. aureus, a range from 1490 (AgrD) to 37,040 M⁻¹cm⁻¹ (AgrC) and for S. epidermidis, from 2980 (AgrD) to 42,290 M⁻¹cm⁻¹(AgrC) could be observed. Ε value of a particular protein is determined by the number of tryptophan, cysteine, and tyrosine residues per protein. These residues have a significant impact with the optical density, that can be observed in the wavelength of 276–282 nm. The instability index (II) values were utilised in accessing the stability of protein in laboratory settings. Proteins having II value less than 40 are stable [77]. Except AgrB in S. aureus (45.06) and S. epidermidis (46.77) all Agrs were found as having instability index lesser than 40. Obtained results suggest that except AgrB other Agrs are stable. The aliphatic index (AI) is an additional parameter for evaluating the physicochemical properties of proteins. The AI refers to the fraction of a volume of protein covered by aliphatic amino acid residues [78]. Proteins with higher values are considered to be thermostable, and vice versa. AI values of S. aureus and S. epidermidis varied in between 91.30 (AgrD) to 132.46 (AgrB), and 91.72 (AgrD) to 141.70 (AgrB), respectively. It may be concluded that all the Agrs tested are thermostable. Previous studies also have suggested that Agrs of S. aureus are thermostable [79]. Studies also reveal that Agrs are high in hydrophobic constituents [80]. Furthermore, The GRAVY score is calculated by adding the hydropathy values of all amino acids in the protein and dividing by the number of residues. It ranges from − 2 to + 2, with positive values representing hydrophobic and negative values indicating hydrophilic proteins [81]. GRAVY value was found negative in AgrA for the selected strains from both species and in AgrD for S. epidermidis (−0.046), whereas, other Agrs analyzed show positive GRAVY score (0.337 to 0.626). Consequently, it may be inferred that all studied Agrs in present investigation except AgrA in both Staphylococcus sp. strains and AgrD in S. epidermidis are hydrophobic in nature.

Compositional heterogeneity based on codon usage

GC makeup in codons play crucial role in determining the codon usage bias. Moreover, reports reveal that the 3rd position GC, i.e., GC3, has an immediate influence on codon usage patterns [82]. Occurrence of GC% in the third position was highest and GC% in the first position was lowest among all Agrs of selected both Staphylococcus sp strains. From the analyses it appeared that usage of GC3% for AgrA, AgrB, AgrC, and AgrD are higher in S. epidermidis (43.1, 41.62, 34.64, and 42.55%, respectively) than S. aureus (41.84, 33.17, 33.18, and 34.04%, respectively). Usage of T3% was high in all Agrs except that in AgrD of S. epidermidis. Overall GC% is higher in AgrA (30.96%) and AgrC (27.46%) of S. aureus strains but it was found higher in AgrB (28.09%) and AgrD (31.91%) in S. epidermidis (Table 2). Numerous standpoints of GC percentages indicate the impact of selection pressure on codon usage pattern.

Table 2.

Different positional nucleotides (GC1%, GC2%, GC3%, GC%, A3%, T3%, G3%, C3%, and CAI) values of different accessory gene regulators (Agrs) of S. aureus (SA) and S. epidermidis (SE)

Parameters	AgrA SA	AgrB SA	AgrC SA	AgrD SA	AgrA SE	AgrB SE	AgrC SE	AgrD SE
GC1%	23.85	24.04	23.67	21.28	21.76	17.26	21.71	25.53
GC2%	27.2	25.96	25.52	23.4	26.36	25.38	25.4	27.66
GC3%	41.84	33.17	33.18	34.04	43.1	41.62	34.64	42.55
GC%	30.96	27.72	27.46	26.24	30.4	28.09	27.25	31.91
A3%	32.64	31.73	32.25	38.3	34.73	40.1	31.41	40.43
T3%	43.51	44.23	44.08	40.43	43.51	42.64	46.88	34.04
G3%	9.62	12.5	10.21	6.38	11.72	10.15	11.09	12.77
C3%	14.23	11.54	13.46	14.89	10.04	7.11	10.2	12.77
CAI	0.715	0.650	0.687	0.732	0.746	0.750	0.735	0.769

Effective number of codon (ENc) plot analyses

The correlation between ENc changes and differences in GC contents can be identified by examining the causal connection between ENc and GC3. ENc chart is an effective approach for evaluating codon use trends, and it was utilised to access the effect of GC3s on codon usage bias [83]. ENc was compared against GC3s to determine selection pressure of Agrs (Fig. 1). Traits which appear at or above the predicted curve suggest selection bias, whereas genes that are present below the expected curve but very near to the curve indicate minimal mutational pressure. The ENc values ranging from 44 to 50 were found to be under the predicted curve. Results suggest that minimal mutational pressure is probably acting as the key factor in establishing the effect of genetic incarceration on codon usage bias of Agrs.

Fig. 1.

Modulation of the effective number of codons/number of codons (ENc/Nc) with GC content at the third codon position (GC3s) in AgrA, AgrB, AgrC, and AgrD of S. aureus and S. epidermidis, where the solid line reflects the predicted curve

Relative synonymous codon usage (RSCU) analyses

RSCU is a computationally derived parameter that determines the incorporation of various codons for a particular amino acid in a protein. It is useful in understanding the codon usage pattern of distinct proteins across various organisms [41]. Table S1 (supplementary section) summarizes the values of RSCU for all Agrs. Results showed that all Agrs used more AU ending codons, except for His in AgrD, Ala in AgrB of the S. aureus isolates, and Glu in AgrB of S. epidermidis. Codon inclination was further assessed (Table S1; supplementary section) based on RSCU scores. Synonymous codon use was greater for Leu, Ser, and Thr, average for Ile, Val, Pro, and Ala, and lower for the remaining amino acid residues. Majority of the amino acid residues in all the Agrs preferred AU-ending codons A3 over U3, and vice versa. In accordance with the analysis on chemical composition, characteristics and codon use patterns throughout agr genes, compositional parameters are affected with low mutation pressure. Codon bias variants include biased codon pairs and codon relationship. It is commonly known that codon bias increases translation efficiency by modifying the process’s elongation rate, even though translation initiation is the most crucial step in protein synthesis. Moreover, a variety of biological processes, such as uneven protein production and folding, are impacted by codon bias [84]. Translational preference is a key element that shapes the variance in codon usage among the genes in Streptococci, even if compositional restrictions are the primary factor controlling this usage. Additionally, it has been shown that U-ending codons are favored in quartets of highly expressed genes, but C-ending codons are always chosen in duets [41]. Additionally, there was a bias towards A or U/T in the frequencies of four nucleotides on the third position of codons. An energy-saving tactic by AU-rich codons has been suggested in pathogenic organisms. In a similar vein, the preference for A/U ending codons in Chlamydia trachomatis’ codon use bias implies that knowledge of these patterns may help discover potential therapeutic targets and comprehend the development of the infection [24]. S. aureus has a regulatory system called the GeaRS two-component system (TCS) that allows it to react to changes in its surroundings. GeaRS is AT-rich and comprises a response regulator and a histidine kinase, just like the AgrAC system. The synthesis of virulence factors is one of the many cellular functions that this system controls [85].

Codon adaptation index (CAI) analyses

The expected translation efficiency of four agr genes from S. aureus and S. epidermidis CAI was forecast and evaluated. In the present investigation, the CAI values varied from 0.650 (AgrB of S. aureus) to 0.769 (AgrD of S. epidermidis). Overall CAI values of the four Agrs of selected strains of S. aureus and S. epidermidis were found as 0.696 and 0.750, respectively, indicating considerably greater expression of genes with relatively substantial codon adaptation over an entire sequence (Table 2). CAI was put forth as a quantitative method of estimating a gene’s degree of expression by analyzing its codon sequence [42]. The CAI, the codon bias index, an entropy score pertaining to synonymous codon usage, a TATA-box score, and a pyrimidine bias index were among the various protein features that were used to determine the expression levels. A regression study revealed that the entropy score and the CAI, two metrics pertaining to synonymous codon usage, were the two important model components [86]. Variations in codon usage or codon bias affect gene expression and cellular processes via influencing protein synthesis and folding. Although translation initiation is vital, codon bias adjusts elongation rates, modifies ribosome velocity, and impacts mRNA stability, all of which have an impact on protein folding and abundance. Protein synthesis as well as the structure and functionality of the final proteins are impacted by this bias, which includes preferences for particular codon pairings [84]. Higher values in S. epidermidis compared to S. aureus suggest that the organism’s genes were moderately expressed and the codons were adaptable. All Agrs exhibit CAI values greater than 0.6, suggesting a high level of adaptation. All Agrs have equivalent expression trend, indicating the simultaneous expression in both Staphylococcus sp. The amino acid identity values and the related protein expression that agr encodes are very close [87]. Agrs play significant functions in controlling the expression of the RNAIII gene are suggested by their sequence conservation, and all staphylococcal species most likely share similar regulation mechanisms. Whereas similarity in the expression profiles of agr did not correlate with the expression of staphylococcal exotoxins [88, 89].

Subcellular localization and membrane topology

Two servers, viz., CELLO and PSORTb produced similar results for cellular localization of all the Agrs (Table 3). AgrA and AgrD of both the species were found to be located in cytosol, but AgrB and AgrC were predicted to be transmembrane proteins. This findings appeared consistent with earlier findings [90, 91]. Cytosolic and transmembrane regions of all Agrs are shown in Fig. 2. Research indicates that among the Agrs, AgrA and AgrD are mostly found in the cytosol, whereas AgrB and AgrC are found in the membrane [90, 91]. Length and number of transmembrane regions of AgrB and AgrC are presented in Table S2 (supplementary section). Most interestingly, TOPCON and Deep TMHMM show six, CCTOP and Phobius show five, and Octopus show four transmenbrane regions of AgrB in all S. aureus and S. epidermidis (33 to 184 amino acid residues) strains under present study. AgrB is a transmembrane protein that aids in the digestion of AgrD. AgrB has six transmembrane helices, four hydrophobic and two hydrophilic, with large number of positively charged residues [92]. Transmembrane-nontransmembrane protein interactions have a direct impact on drug absorption and typically hold the role of hub nodes in related pathways because transmembrane proteins are important drug targets. It is true that amino acids outside of a protein’s membrane-spanning domain can serve as therapeutic targets. These areas serve a variety of purposes and can be targeted by medications to alter cellular activity [93, 94]. All servers resulted in the same number of transmembrane regions, i.e., seven for AgrC in strains from both species with length between 7 and 205 amino acid residues. SignalP server suggested that there is no signal peptide for any of the Agrs in both species.

Table 3.

Sub cellular localization of different accessory gene regulators (Agrs) of S. aureus (SA) and S. epidermidis (SE) predicted through several servers. Here, C- cytosolic and M- transmenbrane

Tools	AgrA SA	AgrA SE	AgrB SA	AgrB SE	AgrC SA	AgrC SE	AgrD SA	AgrD SE
CELLO	C	C	M	M	M	M	C	C
PSORTb	C	C	M	M	M	M	C	C

Fig. 2.

Structure of transmembrane helices of accessory gene regulators (Agrs) from S. aureus (SA) and S. epidermidis (SE) strains: (a) MembraneFold online tool was used to visualise transmembrane helices of Agr, (b) Deep TMHMM was used to visualise membrane-spanning portions of particular proteins based on amino acid residues

Crystallisation studies

XtalPred employs a pair of crystallisation categorization methods: expert pool and random forest. The expert pool approach combines multiple protein features like GRAVY, length, insertion value structural disorder, etc., with crystallisation chances into an overall crystallisation value. Scores vary from 1 to 5. Lesser the value, greater the protein’s crystallisation likelihood [43]. Except AgrA, all Agrs in both the species show an expert pool crystallizability value of 5 (Figure S4a1-d1 and S4e1-h1; supplementary section). Aside from that, the random forest approach makes use of acid composition and roughness on protein surfaces, as well as hydrophobicity. A recent research reveals that the X-ray crystal structure of the C-terminal DNA-binding domain of the S. aureus response regulator AgrA in contact with a 15 bp DNA helix comprising unique 9 bp signature interaction motif [95]. The random forest score for all Agrs was more than eight. XtalPred uses expert pool and random forest scores to determine the likelihood of crystallization. The random forest model tests the predictive potential of protein features like surface ruggedness and hydrophobicity of the protein. It is one of the most precise learning algorithms for assessing crystallization propensity utilizing sequence-related attributes. According to earlier research by Jahangiri et al. [101] and Manna et al. [29], a protein is unlikely to crystallize if its score is more than six. Consequently, it is likely that the Agrs under study are difficult to crystallize, so in silico structural investigation is a good substitute for analyzing their physical makeup. For S. aureus strains, AgrB was 8, AgrA was 10, AgrC and AgrD were 11 (Figure S4a2-d2; supplementary section). In S. epidermidis, the AgrB score was 8, AgrA and D were 9, and AgrC was 10 (Fig S4e2-h2; supplementary section). It may, therefore, be inferred that the examined Agrs, with the exception of AgrA in both Staphylococcus sp. are extremely difficult to crystallise, so computational structural analyses are considered as a prospective alternate option for their morphological assessment.

Secondary structure prediction

ExPASy SOPMA and the GOR IV server were utilised to forecast the percentage of various secondary structures in Agr proteins found in S. aureus and S. epidermidis strains in this present investigation. SOPMA analysis indicated the highest frequency of α-helices and the lowest frequency of β-turns in all Agrs (Fig. 3a), whereas GOR IV analysis reveal highest frequency of α-helices and the lack of β-turns in all the AGRs found in both Staphylococcus sp. (Fig. 3b). AgrA from Staphylococcus pseudintermedius is found to be mostly composed of α-helices and is hydrophilic in nature. Previous investigations reveal that AgrA primarily comprises of α-helices with twists and some β-sheets [76], where removal of helix α−1 significantly impacted the structure, DNA binding and rigidity, highlighting the significance of the unique α-helix in a virulence regulator from S. aureus [96]. Studies also show that AgrA has the most extended strands, while AgrC has the most α-helices, and AgrD has the most random coils and β-turns among the Agrs.

Fig. 3.

The proportion of secondary structural characteristics in accessory gene regulators (Agrs) in S. aureus (SA) and S. epidermidis (SE) strains: (a) ExPASy SOPMA and (b) GORIV servers were used to forecast percentage of different secondary structures (α-helix, β-turn, random coil, and extended strand)

Various amino acids in different locations in the structure of a protein have immediate effects on its structural integrity. Furthermore, amino acid variation affects the structure of proteins [97]. In order to comprehend its impact on protein, twenty amino acids that are present in Agrs have been grouped into five categories depending on their charges and polarity: polar uncharged (Cys, His, Asn, Gln, Ser, Thr), negatively charged (Asp, Glu), and positively charged (Arg, Lys), non-polar aliphatic (Ala, Gly, Leu, Ile, Met, Pro, Val), non polar aromatic (Phe, Tyr, Trp) [98]. Percentage distribution patterns of amino acids for all the Agrs of both species are summarised in Figure S5 (supplementary section). Leu and Ile are the most prevalent amino acids in all Agrs. From the Agrs investigated, AgrD of S. epidermidis includes the most Leu (17.4%), whereas AgrC of S. aureus contained the most Ile (16.7%). In AgrC, a putative leucine zipper was discovered, having four leucines separated by six amino acid residues [75]. For all Agrs in this study, Arg frequency is around 5% of the twenty amino acids. A prior research found that arginine inhibits glucose catabolic suppression and stimulates S. aureus growth in anaerobic conditions [99]. Among non-polar aromatic amino acids, Trp has been identified to be comparatively small in concentration, but Tyr and Phe are distributed equally throughout all Agrs except AgrA of S. aureus strains. Among the polar uncharged amino acids, Asn, Ser, and Thr were found to be especially numerous, while Cys was the least prevalent, with just 0.2% present in AgrC of S. epidermidis. In case of polar charged amino acids, the negatively charged Asp and Glu outweighed the positively charged residues. Except AgrD, each Agr contains Arg. These charged amino acid residues contribute in developing the salt bridges that help to stabilize the structures of proteins.

Tertiary structure prediction

TrRosetta and AlphaFold were used for modelling the 3D structure of Agrs for further studies. Models from both online tools were evaluated for each Agr with PROCHECK and ERRAT. PROCHECK uses a Ramachandran plot to identify the computationally permissible areas for backbone dihedral angles ψ vs. φ of amino acid sequences in a particular protein structure [100]. ERRAT evaluates the general quality level of the model. It was observed that the configurations predicted by both the online servers were adequate and suitable for additional investigation [101]. The agr locus from S. epidermidis and S. aureus strains had a 68% total homology [75]. The comprehensive findings from these investigations are shown in Table S3 (supplementary section). This screening approach revealed that the TrRosseta architecture was considerably more trustworthy for AgrA, AgrD of S. aureus strains, and AgrA, AgrB, AgrC of S. epidermidis, whereas Alphafold forecasted the molecular makeup of AgrB, AgrC of S. aureus strains, and AgrD of S. epidermidis. PROCHECK demonstrated an enhanced Ramachandran plot for the Agrs following refining (Table 4). Amino acid residues in the most favoured locations for AgrA, AgrB, AgrC, and AgrD in S. aureus were found as 92.4, 95.4, 95.6, and 90.2%, respectively (Fig. 4i-l), but in S. epidermidis they are 92.9, 95.7, 94.9, and 94.7%, respectively (Fig. 5i-l). On a Ramachandran plot, a high-quality protein structure with good stereochemistry is shown by more than 90% of residues in the most favored region. This represents the optimal protein structure quality [102]. Previous studies on S. aureus reported that 95.8 and 4.2% of the amino acid residues of AgrC (Protein data bank ID-4BXI), Besides that 96.9 and 3.1% amino acid residues of AgrA (Protein data bank ID- 4XYO) were located in the most favored regions and allowed regions, respectively [91, 103].

Table 4.

Percentage of amino acid residue contribution in most favoured regions of Ramachandran plot analyses of different accessory gene regulators (Agrs) in selected strains of S. aureus (SA) and S. epidermidis (SE) through PROCHECK

Amino acid residues (%)	AgrA SA	AgrB SA	AgrC SA	AgrD SA	AgrA SE	AgrB SE	AgrC SE	AgrD SE
Most favoured regions	92.4	95.4	95.6	90.2	92.9	95.5	94.6	94.7
Additionally allowed regions	6.7	4.6	3.9	7.3	5.3	4.5	5.2	5.3
Generously allowed regions	0.4	0	0.5	0	0.6	0	0	0
Disallowed regions	0.4	0	0	2.4	1.2	0	0	0

Fig. 4.

Tertiary structure prediction and evaluation of different accessory gene regulators (Agrs) in S. aureus: Structure of AgrA (a), AgrB (b), AgrC (c), AgrD (d) along with the Z-scores regarding AgrA (e), AgrB (f), AgrC (g), AgrD (h). Evaluation of protein presented by Ramachandran plot of AgrA (i), AgrB (j), AgrC (k), AgrD (l) displayed amino acid locations in favoured [red], additionally allowed [dark yellow], generously allowed [light yellow] and disallowed region [white], and additionally protein flexibility evaluation of AgrA (m, RMSF range: 0-3.5 Å), AgrB (n, RMSF range: 0-3.5 Å), AgrC (o, RMSF range: 0–5 Å), AgrD (p, RMSF range: 0–5 Å) are presented

Fig. 5.

Tertiary structure prediction and evaluation of different accessory gene regulators (Agrs) in S. epidermidis: Structure of AgrA (a), AgrB (b), AgrC (c), AgrD (d) along with the Z-scores regarding AgrA (e), AgrB (f), AgrC (g), AgrD (h). Evaluation of protein presented by Ramachandran plot of AgrA (i), AgrB (j), AgrC (k), AgrD (l) displayed amino acid locations in favoured [red], additionally allowed [dark yellow], generously allowed [light yellow] and disallowed region [white], and additionally protein flexibility evaluation of AgrA (m, RMSF range: 0-3.5 Å), AgrB (n, RMSF range: 0-2.5 Å), AgrC (o, RMSF range: 0–5 Å), AgrD (p, RMSF range: 0–24 Å) are presented

The ProSA analysed the post-refinement attributes of improved models. It assesses total protein makeup using a Z-plot [58]. The Z-scores for the projected configurations of AgrA, AgrB, AgrC, and AgrD in S. aureus were − 8.54, −4.21, −4.35, and − 0.05, respectively (Fig. 4e-h), but in S. epidermidis they were − 7.85, −3.6, −4.4, and − 1.48 (Fig. 5e-h). Except for AgrC in both the species, the Z-score of Agrs fall inside the intersecting area of X-ray crystallography. Salt bridges have a vital role in the stability of protein structures [104]. It was projected that throughout all Agrs, Lys is the most prevalent implicating salt bridge building, whereas His happens to be the least. In AgrA, AgrC, and AgrD, the Lys-Glu salt bridge is the most prevalent one, while the Arg-Asp salt bridge is prominent in AgrB of both species. A salt bridge that connects Asp157 and His208, is positioned in a turn of the 3–10 helix, helps to stabilise the interactions within the sheets. Arg195 (which may be found in the loop connecting α-helix 1 and β-strand 6) and Glu141 (which is found at the starting point of β-strand1) form salt bridges with Asp157. Salt bridges connecting Asp176-Lys223 and His174-Glu226 stabilise the interaction in AgrA [95]. Noticeably, AgrA exhibited the same number of Arg-Lys and Glu-His salt bridges in S. aureus and S. epidermidis (Figure S6, Supplementary section). Charged amino acid residues in trans-membrane helices inhibit appropriate membrane penetration. Energy burden of membrane penetration is decreased by the development of salt bridges [105]. It could be concluded that, all the projected Agrs are of high calibre and might be potentially useful in the investigation of microbial cellular functions.

Flexibility modelling of tertiary structure

Protein elasticity modelling provides knowledge about changes in persisting amino acid residues across time. The CABS-flex 2.0 server was utilized to aid in quick modelling of structural flexibility. It acts as an outstanding simulation server that allows the depiction of large orientation changes in protein structure [106]. Similarly, protein elasticity analyses reveal the existence of functioning amino acid residues, which alter and make proteins extremely flexible in key cellular functions. Yet, extrinsic influences may limit or vary the flexibility qualities [107]. Each Agr model was investigated by using the default settings. Flexibility oscillation was found to be larger in AgrA, AgrB, and AgrC throughout a span of RMSF (Å) score of 0–5. The least oscillation was detected in AgrD with the greatest RMSF (Å) score of 0.5–11 in S. aureus and S. epidermidis strains used under present study. Figure 4m-p (S. aureus) and 5 m-p (S. epidermidis) present the comparison of flexibility modelling plots for each Agr. The RMSF was utilised to assess if the framework is trustworthy throughout simulation or if it departs from original dimensions [108]. Except for AgrD, all other Agrs of both Staphylococcus sp. is steady, in accordance with the RMSF (Figure S15 and S16; Supplementary section).

Domain and motif evaluation

Primary structure analysis of Agr included motif and domain analysis. AgrA of S. aureus and S. epidermidis strains contains two distinct domains. These include the signal transduction response regulator receiver domain, spanning residues 1^st − 125^th, and the LytTR DNA-binding domain, spanning residues 143^rd − 238^th. AgrC of S. aureus contains a histidine kinase domain located between residues 325^th − 427^th. In S. epidermidis, the histidine kinase domain of AgrC spans residues 324^th − 427^th. AgrD of S. aureus includes a quorum-sensing domain ranging 7^th − 42^nd residue, while in S. epidermidis, this domain extends through residues 7^th − 43^rd. Mutations in the histidine kinase domain of S. aureus AgrC alters the synthesis of autoinducing peptide. Different domains in both species are presented in Table 5. The regulatory domain associated with the LytTR domain has been detected in the clinical isolates of S. epidermidis A086 and S. aureus EMRSA1 (ST239-MRSA-III), which is the oldest pandemic MRSA strain that has been circulating in several nations since the 1970 s (Van Wamel et al. 1998). The C-terminal segment of AgrC has been classified as a member of the autophosphorylating histidine protein kinase family in both species. AgrD is the pro-peptide for an AIP in S. aureus RN6911 and S. epidermidis 12228 that when reaches a specific level, starts the Agr cell QS mechanism and thereafter regulates the transcription of virulence genes [109, 110]. Previous studies reported that S. aureus AgrA includes two domains. The first one is the receiver domain (1^st −130^th residues), while the other is the DNA-binding domain (138^th −238^th residues) [103]. There are several drugs currently being found to focus on LytTR DNA-binding domain of AgrA and block its attachment to the P3 promoter [111]. AgrC contains histidine kinase domain from 293 to 430 amino acid residues [82].

Table 5.

Different domains with the corresponding amino acid length of all accessory gene regulators (Agrs) in S. aureus and S. epidermidis evaluated through interpro scan

Organism	Protein	Domain	Amino acid length
Staphylococcus aureus	AgrA	Signal transduction response regulator receiver	1-125
		LytTR DNA-binding domain	143–238
	AgrC	Histidine kinase	325–427
	AgrD	Quorum sensing	7–42
Staphylococcus epidermidis	AgrA	Signal transduction response regulator receiver	1-125
		LytTR DNA-binding domain	143–238
	AgrC	Histidine kinase	324–427
	AgrD	Quorum sensing	7–43

Agr amino acid sequences obtained from the UniProt library were utilised as initial sequences in the MEME server to envisage the position of motifs within the structures of proteins. Three kinds of motifs comprising consensus sequences such as VYVQYDDIMFFESSTKSHRLIAHLDNRQIEFYGNLKELAQLDERFERCHN, NINKLASYIPKHNELDEIQFVTERSGLQL, and CEDDPKQRENMVSIIKNYIMIEEKPMEJALATDDPYEVLEQAKELNDIGC were prevalent throughout the Agrs (Fig. 6a). In S. aureus and S. epidermidis strains under study, it was found that AgrA possess a total of three kinds of motifs, but AgrB, AgrC, and AgrD possess only one. To recognise S. lugdunensis AgrC-1, the C-terminal hydrophobic motif must have a minimum of one aromatic constituent [112]. Interaction of AgrC suppression between S. lugdunensis group-I and group-II, as well as cross-species inhibitory activity between S. epidermidis group-1 and S. lugdunensis group-I is seen, similar to other Staphylococcal sp [112]. The second motif appears in all Agrs, whereas the first and third motifs were only found in AgrA. Motifs act as hallmark sequences and may be used for recognising any kind of protein. The anticipated e-value of the motif indicates the level of its functional accuracy. Their e-values ranged from 1.4 × 10⁻¹⁵ to 1.1 × 10⁻¹³ (Fig. 6b-d). The N-terminal segment of AgrD peptide has an unique amphipathic alpha helical motif which stabilises it and produces mature autoinducer peptide [111].

Fig. 6.

Visualization of motifs in accessory gene regulators (Agrs): (a) prevalence motif conservation; Sequence alignment of different motif sequences of (b) AgrA, (c) all Agrs, (d) AgrA between S. aureus and S. epidermidis

Protein-protein docking studies

In protein-protein docking investigation, it was observed that AgrA and AgrC of S. epidermidis display the highest docking score among all Agrs included in the study (Fig. 7; Table S4, supplementary section). AgrA docked more strongly with AgrC in S. epidermidis (−293.08 kcal mol⁻¹) than that in S. aureus (−288.23 kcal mol⁻¹). Docking between AgrB and AgrD revealed that S. aureus has a higher docking score (−292.82 kcal mol⁻¹) than S. epidermidis (−278.69 kcal mol⁻¹). The confidence level for every protein-protein docking was greater than 0.92. The LGscore between AgrB and AgrD was greater than 1.5, and between AgrA and AgrC it was greater than 4.6, showing that the docking was correct and considerable, respectively in both species (Figure S7, supplementary section). Figure S9 describes the protein-protein docking interface residues for all Agrs. Microbial communities benefit from the ability of the QS machinery to regulate their reactions to external stimuli. Additionally, it controls the synthesis of virulence components, such as biofilm, which aids in the pathophysiology of bacterial infections. Small chemical molecules known as autoinducers are usually responsible for mediating this signaling pathway. Microbes can become nonpathogenic and be easier to eradicate if QS is blocked, for example, by slowing down the manufacture of the signal molecule, breaking it down, or stopping it from attaching to the response regulator. Numerous substances, including furanones and 5-acetyl-4-methyl-2-(3-pyridyl) thiazole, have been suggested as QS inhibitors. In the quest to treat MRSA infections, MD analysis showed encouraging interactions between the QS inhibitors and AgrA in S. aureus, validating their antivirulence and antibiofilm capabilities [113]. A negative score in protein-protein molecular docking indicates a good interaction; higher binding affinities are indicated by more negative values. AgrA and AgrC were shown to have binding affinities of −9.1 and − 9.2 kcal/mol in S. aureus and S. epidermidis, respectively, and to have a higher interaction probability than that in AgrB-AgrD. Consistent with the observed docking score the AgrA-AgrC interaction of S. epidermidis was found to have the most interfacing residues, forty-eight, five hydrogen bonds between Lys57, His68, Glu14, His18, Ser89 of AgrA and Asp169, Ile181, Arg146, and Arg206 of AgrC, respectively (Figure S13 and S14; supplementary section). Glu14 and His18 are associated through hydrogen bond with Arg146. On the other hand, AgrA-AgrC in S. aureus has forty-two interface residues, including two salt bridges between Asp112 and Glu115 of AgrA and Arg134 of AgrC, as well as one hydrogen bond between Glu115 of AgrA and Tyr130 of AgrC. In S. aureus, AgrB-AgrD interaction were through two hydrogen bonds between Tyr51, Tyr143 of AgrB and Met1, Ser26 of AgrD, respectively but in case of S. epidermidis there was no salt bridge or hydrogen bond (Table 6). All the four complexes lack disulfide linkage (Figure S8; supplementary section). The Ramachandran map of docked complexes reveal that only three amino acid residues of S. epidermidis AgrA-AgrC a located in the disallowed area (Figure S10, supplementary section). All the four complexes have four amino acid residues at the end residues region excluding Gly and Pro (Table S5, supplementary section). The equilibrium constant that results from the dissociation of molecules bound together in a complex is represented by the K_d. A chemical complex is said to be more strongly linked or to have a higher binding affinity between its molecules if its dissociation constant value is smaller. Based on the dissociation constant and binding affinity values, AgrB-AgrD has a higher K_d score than AgrA-AgrC in both species. That defines stronger binding affinity between AgrA-AgrC.

Fig. 7.

Protein-protein interacting residues between AgrA-AgrC and AgrB-AgrD of S. aureus (SA) and S. epidermidis (SE) selected strains. Different interactions, viz., salt bridge (red), disulphide bond (yellow), hydrogen bond (blue), and nonbond (orange) between AGRs

Table 6.

Number of interfacing residues and different interaction of AgrA-AgrC and AgrB-AgrD of S. aureus and S. Epidermidis. Binding affinity (ΔG⁰) and dissociation score (K_d) at 37 °C analyzed using PRODIGY server

Protein	No. of interface residues	Interface distance (Å)	ΔG0 (kcal mol−1)	Kd at 37 ℃	Salt bridge	Disulfide bond	Hydrogen bond	Nonbond
AgrA-AgrC (S. aureus)
AgrA	22	1106	−9.1	4.1 × 10−7	2	0	1	192
AgrC	20	1156
AgrA-AgrC (S. epidermidis)
AgrA	23	1261	−9.2	3.2 × 10−7	0	0	5	180
AgrC	25	1202
AgrB-AgrD (S. aureus)
AgrB	18	784	−6.6	2.1 × 10−5	0	0	2	154
AgrD	14	937
AgrB-AgrD (S. epidermidis)
AgrB	16	716	−5.5	1.3 × 10−5	0	0	0	149
AgrD	10	802

The present findings regarding the interacting interfaces between two Agr regulators might serve as very much promising targets for potential antimicrobials from various sources to exert their anti-QS and anti-biofilm effects to successfully combat the infections. Around 90% of Staphylococcus pathogenecity involves skin and soft tissue infection (SSTI) and the existing reports reveal that infection caused by Staphylococcal sp. is predominantly biofilm-based that eventually turns to be the underlying cause for emergence of large number of multi-drug resistant (MDR) pathogens [114]. Besides, Staphylococcus sp., being opportunistic human pathogen in case of medical device/implant-related hospital acquired infections have turned to be an emergent issue in clinical settings responsible for a great deal of morbidity and mortality [84]. Biofilm formation abilities on implant biomaterials play leading roles in persistence of these infectious agents for their adaptive advantages to survive in hostile environments, and overcoming the biocide, and host defense challenges [115]. Although, biofilm forming abilities in S. aureus and S. epidermidis vary distinctively from strains to strains but it has been observed most of the isolates found in clinical settings were high biofilm-formers. Therefore, alternative therapeutic agents to the conventional antibiotics are of high demand in combating these biofilm-based infections [21, 116]. In this context reports have revealed the applicability of the anti-quorum sensing agents as promising therapeutics that may potentially disintegrate biofilm structures as well as combating antimicrobial resistance [117]. It is worth mentioning that Agr is positively allied with this type of infection. Strongly conserved but heterogeneous QS machinery called the Agr is involved in colonisation, pathogenicity, and biofilm formation in Staphylococci sp. that justifies their role in building-up and persistence of the infection and subsequent clinical significance [118]. In an effort to prevent the negative effects of conventional antibiotics on human health and the surroundings, organic compounds have been thoroughly investigated as significant sources of biocompatible antibiofilm agents [119]. For example, plant-derived essential oils, flavonoids, phenolic acids, and terpenoids have shown biofilm-inhibitory effect by interacting with AgrA and AgrC [120]. Furthermore, a growing number of novel antibiofilm targets of natural compounds have been investigated by computer-based computational screening. The docking score in computational screening represents the possible energy shift that takes place when protein-protein interaction occurs. Accordingly, strong binding is indicated by a greater negative score, while weak or nonexistent binding is indicated by a less negative score [121]. The present observation reveals that a superior AgrA-AgrC interaction could be obtained to AgrB-AgrD in Staphylococcus sp. LytTR domain of AgrA is a gene expression regulator that controls the transcription of several virulence and toxins genes, allowing the bacteria to enter host tissues, and eluding the defence system of the human body resulting in serious illness [95]. Agr serves as a key QS regulator in Staphylococcus sp., enabling the secretion of several toxins by the bacteria and enhancing their pathogenicity. By triggering the production of the agrBDCA operon, which codes for the machinery of QS system, the P2 promoter powers autoregulation network of Agr. On the other hand, the P3 promoter uses RNAIII for modulating different toxins [16, 118, 122]. Additionally, AgrA triggers the transcription of RNAIII from the promoter agrP3. Toxins, viz., α-hemolysin (hla), β-hemolysin (hlb), δ-haemolysin (hld) and virulence factors, viz., protein A (spa), toxic shock syndrome toxin-1 (tsst) expression are regulated by RNAIII [111]. PSM (Phenol soluble modulin) - α, β, and γ are tiny, amphipathic, α-helical peptides that belong to the PSM family. AgrA can directly bind to the promoters of the psm in addition to activating the P2 and P3 promoters in the agr locus. The P3 promoter transcribes the δ-toxin gene, which is situated in the RNAIII portion of the agr locus. The cytolytic characteristic of PSM peptides is critical to their pathogenicity. δ-toxin exhibits mild cytolytic activity, PSMβ peptides are non-cytolytic, and PSMα, particularly PSMα3, has a strong capacity to lyse erythrocytes, epithelial cells, and leukocytes [123, 124]. In recent years, several studies target the Agr system for prospective drug designing and development that would interfere with quorum sensing in Staphylococcus sp. Studies have revealed that inhibitors of AgrA have the potential to be a novel class of anti-virulence and anti-biofilm drugs that can limit the production of virulence factors and hence lessen the pathogenicity of the bacterium [118]. In addition to AgrA, AgrC being a receptor of AIP, and sensor domain (residues 1 to 205) of AgrC that interacts with regulator domain (residues 1 to 130) of AgrA, can also be targeted by drug molecules [125]. In the present study, ≥ 20 amino acid residues of AgrA and AgrC, interacting with each other to form a protein-protein complex, might be a potential site of drug targeting to destabilize the biofilm formation in these pathogens (Table 6). The protein-protein interaction investigation contributes to a more thorough comprehension of binding affinity. In AgrA-AgrC complexes, S. epidermidis (−9.2 kcal mol⁻¹) has a greater binding affinity than S. aureus (−9.1 kcal mol⁻¹). However, the AgrB-AgrD interaction demonstrated that S. aureus (−6.6 kcal mol⁻¹) had a greater binding affinity than S. epidermidis (−5.5 kcal mol⁻¹).

The AgrC protein, especially its cytosolic domain, interacts with the AgrA response regulator domain in Staphylococcus sp. The activation of the Agr two-component system, which is involved in QS, and the control of virulence genes, depends on this association [91].Understanding of this QS machinery may hold the ultimate key to an effective therapeutics development. Within a single species, QS-mediated cell-to-cell communication organizes cooperative activity to improve survival under stress and damage immune systems of human host. One of the most extensively researched communication systems of human bacterial pathogens is the agr operon-encoded QS system in Staphylococcus sp. The effectiveness of the agr quorum-sensing system in S. aureus can be improved by a tighter interaction between AgrA and AgrC, which results in quicker and more reliable reactions to AIP signals. As a result, AgrA and AgrC are thought to be a key target for chemotherapeutic research [126–128]. Template-based molecular docking has been used to provide kinase-ligand intricate information for kinase-related drug development. This data, when used to train graph neural networks, produced more accurate binding affinity estimations than models that just employed data on ligand or drug-target interactions [129]. AgrA and AgrC are, therefore, appear to be seemingly promising in the management of bacterial infections and are apprehended to be useful therapeutic targets for the discovery of potential antimicrobial drugs. AgrA may be the ideal therapeutic target because it is found in the cytoplasm. Certain regions outside the spanning area, such as the 130^th −146^th amino acid residues, may function as therapeutic targets since AgrC is present in the transmembrane region.

Molecular dynamics (MD) simulation studies

Even with extremely massive biological macromolecule, iMODS allow for the investigation of the generation of realistic conversion routes between two identical bio molecules. The novel interior construction improves the effectiveness of NMA and widens its applications [130]. Movement of rigidity is represented by the eigen value corresponding to every normal mode analysis. The amount of energy needed to deform the framework defines its value. The eigen value decreases when the distortion becomes simpler [130]. The eigen value of MD simulations indicates how stable the docked complexes are. It is equivalent to the energy needed to cause the structure to deform, and excellent structural stability is linked to low eigenvalues. The relationship between variance and eigenvalues is inverse [24]. Computed eigen values of the AgrA-AgrC interaction of S. aureus and S. epidermidis were found as 2.9 × 10⁻⁷ and 6.1 × ¹⁰⁻⁷, while that for AgrB-AgrD interaction of S. aureus and S. epidermidis were 6.2 × 10⁻⁶ and 1.5 × 10⁻⁵, respectively (Figs. 8c and f and 9c, and f). Given the lower estimated eigenvalues of the complexes, it can be inferred from the above data that the complexes under study, namely AgrA-AgrC, were more stable than AgrB-AgrD for both species. The deformability plot displays the energy necessary to deform the amino acid residues (Figs. 8a and d and 9a, and d). Violet and green bars represent specific and overall (Figure S11a, d, S12a, and d; supplementary section) variances, respectively. The covariance distribution depicts whether correlated, uncorrelated, or anti-correlated through red, white, and blue marks (Figure S11b, e, S12b, and e; supplementary section). Each mark represents an attraction between the corresponding residues, and rigid attraction represented by the darker grey mark and vice versa. Contact map can reveal relationships between amino acid residues and regions by tracking their relative motions (Figure S11c, f, S12c, and f; supplementary section). The B-factor measures the flexibility of a protein and quantifies its ambiguity that is related to the RMSD, as seen in Figs. 8b and e and 9b, and e. These findings indicate sustained binding between respective amino acid residues with tight structure and little fluctuations in protein-protein interaction. The current work used IMODS to perform MD simulation in order to assess the stability and physical mobility of protein-protein interactions. The durability of docked proteins was also examined using NMA. The deformability and B-factor of AgrA-AgrC show peaks that correlate to the deformable areas of the protein; the higher the peak, the higher the deformability. The NMA analysis demonstrated the structural flexibility of the proteins by revealing the substantial mobility of the AgrA-AgrC aggregates. The interaction has a smaller eigen value, making it simpler to deform and more stable. Because ligands can adopt multiple conformations inside the binding pocket and proteins can alter conformation upon ligand binding. Thus flexibility is essentially required for induced fit in a potential drug-target interaction, and that is crucial for druggability [131]. Reports also suggest that deformability peaks close to 1 represents regions with high flexibility [24] and that is also revealed from the Fig. 8a and d in the present study. The observed interactions with high flexibilities in studied proteins suggest that they might be viable therapeutic target.

Fig. 8.

MD simulation of AgrA-AgrC docked complex (a) deformability plot, (b) B-factor, (c) eigen value of S. aureus and (d) deformability plot, (e) B-factor, (f) eigen value of S. epidermidis

Fig. 9.

MD simulation of AgrB-AgrD docked complex (a) deformability plot, (b) B-factor, (c) eigen value of S. aureus and (d) deformability plot, (e) B-factor, (f) eigen value of S. epidermidis

Conclusion

In addition to the virulence mechanism associated with catheters, S. aureus and S. epidermidis primarily cause infections of the skin and soft tissues through QS and biofilm. Because of this, viable long-term solutions for the management of antibiotic-resistant staphylococcal infections are still difficult and ultimately call for alternate strategies. Agr system of Staphylococcus sp. regulates the transcription of multiple factors that encode virulence properties in the current investigation, a variety of computational methods were employed to examine different sorts of structural modifications in four Agrs of S. aureus and S. epidermidis. According to genome study, natural selection is more influential than mutational pressure in determining the habits of codon usage and preferences for AU ending codons, which suggests that these Agrs might be excellent therapeutic targets. Besides that, an energy-saving tactic in pathogenic organisms is suggested by AU-rich codons. All Agrs show codon adaptability index values more than 0.65, indicating a high likelihood of adaptation and expression in a variety of environmental settings. Physicochemical characteristics reveal that AgrB of both the species was predicted to be unstable under in vitro environments. Furthermore, crystallisation study demonstrates that with the exception of AgrA, the others are extremely difficult to be crystallised and thus warrant further structural investigations. Computational investigations are regarded as simple alternative approaches for conducting structural assessments of Agrs. Secondary structural investigations indicate that α-helix is among the most common structure in Agrs, followed by random coils, with Leu and Ile being the most numerous amino acid residues. The current work is thought to have the ability to develop comprehensive research to determine the structural characteristics of these Agrs. AgrA and AgrC are, therefore, appear to be seemingly promising in the management of bacterial infections and are apprehended to be useful therapeutic targets for the discovery of potential antimicrobial drugs. AgrA may be the ideal therapeutic target because it is found in the cytoplasm. Certain regions outside the spanning area, such as the 130^th −146^th amino acid residues, may function as therapeutic targets since AgrC is present in the transmembrane region. However, in order to validate the theoretical data collected here, more thorough analytical techniques are required. To understand the physicochemical properties of these proteins and confirm existing discoveries, modern structural biology methods such as NMR spectroscopy and X-ray crystallography may be helpful.

Abbreviations

Agr

Accessory gene regulator

AI

Aliphatic index

CAI

Codon adaptation index

EC

Molar extinction coefficient

GRAVY

Grand average of hydropathicity

II

Instability index

MD

Molecular dynamics

MEME

Multiple expectation maximizations for motif elicitation

MSA

Multiple sequence alignment

MW

Molecular weight

NCBI

National Center for Biotechnology Information

PI

Isoelectric point

RCSU

Relative synonymous codon usage

(- R)

Negative charge residues

(+ R)

Positive charge residues

Authors' contributions

S.D.: Writing – original draft, Visualization, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. T.M.: Writing – review & editing, Formal analysis. K.C.G.: Writing – review & editing, Formal analysis. M.K.: Writing – review & editing, Formal analysis. D.J.: Writing – review & editing. P.R.: Writing – review & editing. S.B.: Formal analysis. M.M.: Writing – review & editing. S.H.: Writing – review & editing. A.K.P.: Writing – review & editing, Supervision, and Conceptualization C.G.: Writing – review & editing, Supervision, and Conceptualization.

Funding

Subhamoy Dey acknowledges University Grants Commission (UGC), New Delhi, India for providing fellowships to carry out research work.

Data availability

Data is provided within the supplementary information files.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.