Evaluating the effect of SNPs located on some biofilm genes on their protein function and structure in Pseudomonas aeruginosa and Klebsiella pneumonia isolates

Abstract

To evaluate the impact of biofilm-associated SNPs (previously identified) on protein structure and function, we employed a suite of bioinformatics tools. SIFT, Mupro, INPS-3D, and NCBI Conserved Domain Search were utilized to analyze changes in function, stability, secondary structure, phi/psi angles, and Relative Solvent Accessibility (RSA). Furthermore, SwissDock and molecular dynamics simulations were conducted to compare free energy, RMSF, polarity, flexibility, and molecular contacts between wild-type and mutant proteins. In silico analysis indicated that mutations in quorum-sensing (lasI, rhlI) and biofilm-associated genes significantly impact bacterial biofilm formation. Specifically, SIFT and Mupro predicted that lasI/rhlI mutations D39N (0.01/0.2) and D44N (0.5) impair signaling molecule synthesis. Furthermore, the S32N substitution in ndvB was predicted by INPS-3D to alter secondary structure (β-sheet to coil) and torsion angles (ϕ: −99 to 118; ψ: −94 to 5), likely disrupting protein function. Additionally, the R292A mutation in tssc1 was predicted to impact protein stability (SIFT: 0.01; Mupro: −0.9) and increase burial (RSA 37% to 23%, helix to coil). A frame-shift mutation in tssc1 was also identified, suggesting further modulation of biofilm production. Regarding with docking results, D44N in the rhII gene showed a slight change ΔΔG values from −7 to −7.2 and a change in polarity from 16 to 14, however, molecular dynamic simulation showed significant increase in RSMF value from 1.5 to 3.5 and increase the flexibility in mutant compared with wild type which reflects the importance of this SNP in changing the function of the protein mediating biofilm formation.

1. Introduction

Pseudomonas aeruginosa, as well as Klebsiella pneumonia recorded as pathogens with opportunistic infection, associated with many diseases such as cystic fibrosis, sepsis, and urinary tract infection.1, 2, 3 Mutation and horizontal genetic transfer are tools that are used by Pseudomonas aeruginosa and Klebsiella pneumonia to diversify their genome and acquire resistance to antimicrobial agents and causing sever infection.1, 2 In addition to their ability to diversify their genomes,³ Pseudomonas aeruginosa and Klebsiella pneumonia are able to form biofilm as a strategy, which helps to escape the immune system of the host and other killing environmental factors. Biofilm, as a sheath of exopolysaccharide with 90% while other substances are protein and DNA, increases the ability of bacteria to cause infection in hospitals (nosocomial infection) as well as inside the host.4, 5 The colonies of bacteria inside the biofilm possess a high rate of horizontal genetic transfer; therefore, they become multidrug resistant, as seen in Pseudomonas aeruginosa and Klebsiella pneumonia.⁶ Genes mediated biofilm construction are correlated with quorum Sensing, fimbriae, polysaccharides, and efflux pump.⁷ Quorum Sensing is a chemical signal that is produced by the community of bacterial cells to regulate different biological activities, mainly the biofilm formation. These signals are called autoinducers. In Klebsiella pneumonia, one of the quorum-sensing genes is the luxS gene, which plays a vital role in biofilm construction.8, 9, 10 In Pseudomonas aeruginosa, quorum-sensing signaling pathways are controlled by many genes, mainly lasI and rhII genes. These genes produce autoinducers that contribute in biofilm initiation as well as biofilm maturation. Autoinducer production is correlated with the density of bacterial colonies, by that their concentration is low when bacterial colonies are low. High amount of autoinducers production when colonies of bacteria are high, therefore these autoinducers can control gene expression for different genes as seen in biofilm formation stages.11, 12 lasI and rhII genes produce two types of autoinducers (Acyl homoserine lactones) that intact with their receptors to participate in biofilm formation stages.13, 14 Another gene that contributes to biofilm formation is the ndvB gene, this gene encodes to glucosyltransferase that participates in the formation of glycans that serve as a Key element in making a sheath around bacterial colonies, hence biofilm construction.15, 16, 17 It has been reported that the strain luck ndvB gene can still form biofilm, but it becomes sensitive to different antibiotics such as gentamicin and tobramycin as glucan sequesters different antibiotics and prevents them from penetrating the bacterial cells.¹⁸ TssC1 gene contributes to antibiotic resistance in the presence of biofilm. SNPs mediating biofilm formation are more crucial than other SNPs mediating other virulence genes in bacterial infection due to many reasons, mainly due to the function of biofilm as a barrier protecting bacterial cells from antimicrobial attack and immune response. In addition, SNPs mediating biofilm formation may affect traits in comparison with other SNPs, enabling the bacterium to persist for a long time in its host.¹⁹ All mentioned genes contribute directly or indirectly in biofilm formation; therefore, we selected them in our investigation. Many researchers focused on enhancing drug delivery using nanoparticles to treat different diseases related to microorganisms forming biofilm20, 21, 22, 23 while others used microbial extraction to fight against carcinoma.²⁴ Many bioinformatics tools are used by researchers to predict the effect of SNPs in different genes on the function and structure of their protein, especially in humans. However, to date, no record has been announced for evaluating the SNPs' effect on protein structure and function in prokaryotes. Therefore, we intended to visualize the effect of SNPs located on genes mediating biofilm formation on the function and structure of their protein.

2. Material and methods

2.1. Isolates collection and DNA sequencing

Isolation of P. aeruginosa and K. pneumonia from different samples, with 20 isolates using different morphological and biochemical methods for each, was achieved. In addition, the sequencing was carried out on 6 out of 20 isolates (2 weak biofilm formers, 2 moderate biofilm formers, and 2 strong biofilm formers) as mentioned previously in our published papers.25, 26 SNPs were identified in isolates relying on Sanger sequencing technologies based on PCR product sequencing. SNPs that are located within biofilm genes were analyzed separately in both P. aeruginosa and K. pneumonia.

2.2. SIFT program

The idea of the SIFT program is to detect the SNP effect on protein function and structure, relying on the evolutionary aspect. The program started to make alignments for amino acids of protein from a large database hosted on different bioinformatics websites, then looked for the changed amino acid in a particular position in the alignment. Finally, the program calculates the score relying on the physical properties of amino acid change in the alignment (for example, a hydrophilic amino acid change to a hydrophobic amino acid in the position of alignment for the large database is scored low, which means this SNP affects protein function and structure). The program works on SNPs in humans using human databases, but there is the possibility of running the program on prokaryotic databases by submitting alignment from prokaryotic databases along with substitution. We got the database from https://pubmlst.org/organisms/pseudomonas-aeruginosa for a large number of isolates. The database was a DNA sequence with multifasta format that was converted to amino acid sequences through EMBOSS transeq then we submitted an aligned multifasta amino acid database to the SIFT program.²⁷ For the SIFT program default parameters are as follows: scores less than 0.05 are predicted as affecting the function, while scores higher than 0.05 are predicted as not affecting the function of the protein.²⁸

2.3. Mupro program

The program calculates the change in the stability of the protein by measuring ΔΔG value by subtracting Gibbs free energy from the mutant protein and the wild-type protein. The result with a negative value means a decrease in stability, while positive values mean an increase in stability. The amino acid sequence for the protein was submitted with substitution amino acids, and the program is available online at (http://mupro.proteomics.ics.uci.edu).²⁹ For the Mupro program, default parameters are as follows: ΔΔG with a negative value indicates the protein is destabilized, while a positive value indicates the protein is stabilized with value should be more than 0.5.²⁸

2.4. NetSurfP3 program

It calculates the difference in phi/psi angles along with protein secondary structure, disorder, and surface accessibility for substituted amino acids. The amino acid sequence for the protein was submitted for the mutant and wild-type proteins through (https://services.healthtech.dtu.dk/services/NetSurfP-3.0) ³⁰. For NetSurfP3 program default parameters are as follows: change in values of phi/psi angles along with protein secondary structure, disorder, and surface accessibility for substituted amino acids between mutant and wild-type protein indicates the SNPs is affecting protein function. For surface accessibility, a value that is less than 25%, amino acids are considered buried. However, a value that is more than 25%, consider as exposed. For the disorder, a difference in percentage of less than 10% is a minor change, while a difference above this threshold is a major change. The changes in the phi/psi angles and shift in their values on a Ramachandran plot are generally considered a statistically significant indicator.

2.5. NCBI conserved domain databases (CDD program)

CDD program hosted in NCBI (https://www.ncbi.nlm.nih.gov/Structure/cdd/wrpsb.cgi) was used to predict the domain for amino acids sequence of a particular protein and E value was considered to look for the significant result with cutoff 1e −2. The boundaries of domain were predicted and the mutations were manually searched for their location within the domain boundaries.

2.6. Molecular docking

The Swiss docking website (https://www.swissdock.ch/) (AutoDock Vina version1.2.3) was used to carry out the simulation of molecular docking for mutant and wild-type proteins. First, the pdb structure of wild-type was taken from pdb database with PDBID (2WYD) (https://www.rcsb.org/) and then submitted to the Swiss docking website along with a specific ligand (which is meloxicam with compound CID 54677470), after that, the parameters were adjusted to cover the binding site region. Secondly, the mutant amino acid was submitted to the Swiss model (https://swissmodel.expasy.org/) to achieve homology modeling and predict the PDB structure of the mutant, after refinement of pdb structure relying on different criteria, which was submitted along with the ligand to the Swiss docking website. Before submitting receptor to the Swiss docking website, the water molecules were removed and hydrogen bonds were added. Furthermore, missing atom were repaired. Subsequently the grid box size was established to cover almost of protein focusing on biding site (30*30*30). The grid centers' XYZ coordinates was set with default (X = 15.5, Y = 20.1, Z = 5.2) and the docking process was achieved using the Genetic Algorithm (GA) (Exhaustiveness = 8, Number of Modes = 9, Energy Range = 3 kcal/mol). The receptor is rigid while ligand is fully flexible. The result was retrieved focusing on the change in free energy and root mean square deviation (RMSD) between mutant and wild-type proteins. RMSD value refers to the atoms of the mutant protein deviating from their wild-type protein structure (initial structure). Finally, Chimera 1.18 was used to visualize and annotate binding between ligand and protein, highlighting the bonds created between ligand and the binding site of the protein. Formation of different bonds that can be visualized and measured by Chimera between ligand and receptor indicates this ligand is specific for this receptor. In addition, the difference in other parameters such as ΔΔG, RMSD, and polarity values between wild and mutant protein through comparison of docking results reflect change in stability of protein. The default parameters for the ΔΔG should be equal or less than −5 and A change of more than 1–2 kcal/mol is considered relevant to altering binding affinity. The RMSD should be less than 2 and altering the polarity is reflecting the change in hydrogen bonding through comparison wild and muatnt models.³¹ Ligand (meloxicam with compound CID 54677470) was selected based on binding energy (strong binding energy from literature.³² Ligand structures were downloaded in SDF format from PubChem, and 3D conformers were generated using Open Babel. PDBQT files were prepared for AutoDock Vina, with Gasteiger partial charges and defined rotatable bonds.

2.7. Molecular dynamics simulation

The structural dynamics of [PDBID (2WYD) along with ligand] were investigated using the CABS-flex 2.0 web server (http://biocomp.chem.uw.edu.pl/CABSflex2) to simulate conformational fluctuation. The 2WYD structure was uploaded to Pymol to remove all hydrogen atoms and model missing residues. Secondary structure was automatically assigned by the server using the DSSP method, with coil regions allowing for greater flexibility. Simulations were performed using default distance restraints. These restraints are derived to maintain the stability of regular secondary structure elements (α-helices and β-sheets) while allowing flexibility in loop regions. To ensure adequate sampling of the conformational space, the simulations were run for [1000] cycles. Simulations were performed at a reduced temperature of 1.4. The number of cycles and the cycle interval between trajectories were set to 50. The CABS-flex pipeline generated a trajectory of 1000 models, which were subsequently clustered to produce 10 finals, representative cluster medoids. The results were analyzed in terms of Root Mean Square Fluctuation (RMSF) to identify flexible and rigid regions and through residue-residue contact maps to examine conserved intramolecular interactions.³³

3. Result and discussion

3.1. 1 – Biofilm quantification

The ability of P. aeruginosa isolates to form biofilm was superior and exceeded the ability of K. pneumonia, with 80% of isolates forming biofilm, with 45% forming weak biofilm, and 35% forming moderate biofilm. However, 45% of K. pneumonia were non-biofilm producers, while 30% and 25% were weak and moderate biofilm producer respectively. This may indicate that the SNPs detected in genes mediating biofilm formation in P. aeruginosa isolates may be more effective in changing the biofilm schemes among weak and moderate compared with K. pneumonia isolates Fig. 2. (See Fig. 1.)

Fig. 7.

Schematic representation of simulated wild-type and mutant models. Panel A shows the mutant configuration, characterized by a more scattered, flexible distribution. Panel B shows the wild-type, exhibiting a more compact conformational ensemble compared to A. The red sphere indicates the mutation location within the loop domain, which forms part of the active site. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Fig. 2.

Biofilm formation capacity for P. aeruginosa and K. pneumonia isolates. Bar chart indicating a higher prevalence of strong biofilm producers in P. aeruginosa isolates compared to K. pneumonia.

Fig. 1.

Flowchart of the bioinformatics pipeline. The pipeline outlines the process from DNA sequencing, SNPs identification, SNPs evaluation and Molecular docking and simulation.

3.2. 2 - lx gene in K. pneumonia

Investigation of the importance of variation in different genes mediating biofilm formation revealed that variation in the luxS gene in K. pneumonia was synonymous and detected in two positions at 262, thereby (T) changed to (G), and at position 358, thereby (C) changed (T), and their amino acids did not change. Therefore, these variations may have no role in biofilm construction, and we exclude them from our analysis.³⁴Moreover, this may explain why the quantification of biofilm in P. aeruginosa isolates is overestimated in K. pneumonia isolates, as mentioned previously.

3.3. 2 - lasI gene in P. aeruginosa

On the other hand, the variation in the lasI gene in P. aeruginosa isolate was determined with nonsynonymous changes resulting in R31K and D39N mutations, while variation at position 130 of the DNA sequence was synonymous and therefore was excluded from our analysis.²⁵ SIFT program was used to show the importance of these nonsynonymous changes in changing the function and structure of protein coded by lasI gene, the result revealed that substitution at position 31 from K to R is predicted to not effect function of protein with a score of 0.34 while substitution at position 39 from D to N is predicted to effect function of protein with a score of 0.01. Further analysis was achieved to show the effect of these SNPs on protein stability, by which the Mupro program prediction for R31K showed decreased stability with a value of −1.6 which is considered a significant value, while D39N was correlated with increased stability with a value of 0.2 and is generally considered not a significant change. Although the D39N mutation was predicted using the Mupro program with no significant change in stability of the protein, a portion of the stability of the protein is increased; hence, the functionality of the protein increases as well. Therefore, this is the second piece of evidence indicating the D39N mutation is more effective on protein function compared with the R31K mutation. The INPS-3D program was used to predict the solubility, disorder, phi and psi, and secondary structure for each SNP. Results revealed that there was a slight increase in RSA from 34% to 36% which is generally not considered significant and a slight decrease in phi angle from −97 to −98, while other parameters were stable for R31K. Moreover, there was also a slight increase in RSA from 68% to 69% and other parameters were stable; however, phi and psi angles were increased from 53, 34 to 58, 40, respectively, for D39N. In spite of these changes in phi and psi angles is not considered highly significant but it may represent a moderate local structural adjustment in allowed regions of Ramachandran plot. The SNPs also could be predicted as affecting the protein structure and function in case it is located inside the domain (the part responsible for the protein structure and function). Both changes in amino acid sequences at R31K and D39N are located within the lasI superfamily domain.

3.4. 3 - rhII gene in P. aeruginosa

The variation in the rhII gene in P. aeruginosa isolate was determined with a nonsynonymous change in the protein sequence as D44N. SIFT prediction declared that substitution at position 44 from D to N is predicted to be tolerated with a score of 0.82. Moreover, INPS-3D analysis showed that in most parameters with RSA and phi, psi angles were slightly changed, by which RSA increased from 67% to 81% and both those values with more than 25% consider as exposed. Therefore, there is no change from exposed to buried. Regarding with phi, psi angles, phi increased from 57 to 63, and psi also increased from 18 to 22 and is generally not considered a significant conformational change, while other parameters showed stability with no change in their values at all. However, the Mupro program predicted that the SNP affects the stability of the protein by increasing in stability with a value of 0.5 which is considered as significant change in stability of protein. SNP mediating D44N change was located within the domain.

3.5. 4 - ndvB gene in P. aeruginosa

The variation in the ndvB gene in the P. aeruginosa isolate was detected with a nonsynonymous change in the protein sequence as S32N.³⁵ SIFT prediction highlights that substitution at position 32 from S to N is predicted to be tolerated with a score of 0.46. Moreover, the Mupro program predicted that the SNP affects the stability of the protein by increasing in stability with a value of 0.2 and is generally not considered significant. Furthermore, INPS-3D analysis showed that most parameters were changed, by which RSA highly increased from 30% to 41% and both those values with more than 25% are considered as exposed. Therefore, there is no change from exposed to buried. However, phi and psi dramatically decreased from −99, 118 to −94, 5, this change led to a change predicted secondary structure of this particular SNP from B-sheet to coil, while disorder was stable. The variation was located within GH17 family domain.

3.6. 5 - tssC1 gene in P. aeruginosa

Finally, the variation in the tssC1 gene in the P. aeruginosa isolate was detected with a nonsynonymous change in the protein sequence as R292A. SIFT prediction annotated that the substitution at position 292 from R to A is predicted to affect protein function with a score of 0.01. Moreover, the Mupro program predicted that the SNP affects the stability of the protein by decreasing in stability with a value of −0.9 and is generally considered significant. Furthermore, INPS-3D analysis revealed that most parameters were changed, by which RSA highly decreased from 37% to 23% and considered as significant as changing from exposed to buried. Moreover, phi angle decreased from −64 to −98, while psi dramatically increased from −41 to 22. This change led to a change predicted secondary structure of this particular SNP from helix to coil, while disorder was stable with the value of zero. The variation was located within the C0G3517 superfamily domain.

Predict the location of different SNPs in conserved domains using the NCBI conserved domain, and predict the solubility, disorder, phi and psi, and secondary structure for all mutations using the INPS-3D program, as illustrated in Fig. S1 and Fig. S2 (supplementary data).

All Relative solvent accessibility (RSA) and ΔΔG for all changes in different genes can be summarized in Table 1, in addition to phi and psi angles Fig. 3.

Table 1.

Relative solvent accessibility and ΔΔG for all changes in different genes.

Genes	Change type	Change in Relative solvent accessibility	ΔΔG
lasI	R31K	34% to 36%	−1.6
lasI	D39N	68% to 69%	0.2
rhII	D44N	67% to 81%	0.5
ndvB	S32N	30% to 41%	0.2
tssC1	R292A	37% to 23%	−0.9

Fig. 3.

Relative solvent accessibility, ΔΔG and phi, psi angles for all changes in different genes. (A) Relative solvent accessibility (RSA) comparing wild-type and mutants, highlighting the R29A mutant (significant change from exposed to buried). (B) ΔΔG stability changes for each mutation; R31K, D44N, and R292A show significant destabilization (ΔΔG > 0.5 kcal/mol). (C) Ramachandran plot analysis of ϕ and ψ backbone torsion angles in wild-type and mutant structures, showing significant conformational shifts in S32N and R292 mutants.

Additionally, an INDEL was detected at position 807 of the DNA sequence of the ndvB gene that led to a frameshift mutation, which may in turn change all amino acids of the ndvB protein after this position. This frameshift mutation was confirmed to show a change in different amino acids using the SNAP program.³⁶

Regarding the molecular docking, all SNPs of the lasI and ndvB genes are located outside of the binding site; therefore, we assume they do not affect binding energy and RSMD values. However, the D44N change is related to the rhII gene located within the binding site (Fig. 4), hence we achieved molecular docking to visualize the effect of D44N change in the rhII gene on binding energy, and polarity.

Fig. 4.

Three-dimensional structure modeling of rhII protein. (A) Wild-type structure (WT). (B) Mutant structure showing the D44N substitution. The mutated residue is highlighted as a red sphere located on the protein loop, indicating the localized structural change. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

The molecular docking result showed a change in binding energy with the reduction from −7 to −7.2 kcal/mol. Polarity is also decreased from 16 to 14. The interaction between the active site and ligand revealed that the oxygen atom in the ligand forms a hydrogen bond with ARG 71, the carbon with the number 6 atom in the ligand forms a hydrogen bond with ALA 70, the oxygen atom in the ligand forms a hydrogen bond with SER 123, there is a hydrophobic interaction or van der Waals interaction between a protein receptor (LEU 124, carbon atom) and its ligand, and there is a hydrophobic interaction between LEU 102 and a ligand containing an oxygen atom (O₂) Fig. 5.

Fig. 5.

Molecular docking between mutant rhII protein and ligand. (A) Predicted binding pose of the ligand (red) within the active site of the mutant rhII protein, highlighting interacting amino acid residues. Yellow dashed lines indicate intermolecular bonds (e.g., hydrogen bonds, hydrophobic interactions) between the ligand and receptor. (B) Surface representation (spacefill) of the ligand docked within the binding pocket of the mutant rhII protein, demonstrating conformational fit. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Molecular dynamics simulation was carried out on the wild type and the protein carrying the D44N mutation. The flexibility in the structure of both wild-type and mutant types for rhII protein was assessed using the root mean square fluctuation (RMSF) values from the data of a 100 ns simulation trajectory. Overall RMSF values for the mutant model were deviated from the wild-type model, and the highest residual fluctuation for the wild type was noticed in Thr 39, Ser 40, Arg 41, Val 42, Arg 43, and Asp 44 with 3.69 nm, 4.066 nm, 3.781 nm, 4.505 nm, 4.815 nm, and 3.592 nm values, respectively. The highest residual fluctuation was noticed for the position before the mutant D44N position. The deviation in the RMSF values or the mutant model compared with the wild type indicates the impact of deleterious amino acid substitutions in the rhII protein Fig. 6.

Fig. 6.

Schematic representation of RMSF value in nm for wild type and mutant protein with D44N. The graph represents residual flexibility (in nm) calculated over [100 ns] of MD simulation. The D44N mutation causes a significant increase in local flexibility, with the highest deviation in RMSF values observed in the region spanning residues 39 to 44 compared to the wild-type.

The frequency of contact, which indicates the percentage of time two amino acids are in proximity during the simulation trajectory, has a value close to 1 indicates a highly stable/conserved interaction. Both wild type and D44N mutant type showed high stable contact between residues A44 and A42, A44 and A43, A44 and A45, A44 and A46, with a value of 1. However, the contact was changed in the D44N mutant compared with the wild type, and A44 become in contact with A41 with a value of 0.7, instead of A44 and A47 with a value of 0.7 in the wild type. This change reflects the impact of deleterious amino acid substitutions in the rhII protein Table 2.

Table 2.

The frequency of contact between residue A44 with other residue in wild and D44N mutant type.

wild		value	Mutant		value
A44	A14	0.2	A44	A41	0.7
A44	A47	0.7	A44	A47	0.2
A44	A42	1	A44	A42	1
A44	A43	1	A44	A43	1
A44	A45	1	A44	A45	1
A44	A46	1	A44	A46	1

Comparing the 10 different models of molecular dynamic simulation in wild and D44N mutant type revealed that the 10 mutant models are more scattered than the 10 wild models, which indicates that the mutation increases flexibility especially the SNPs located on loop and it is a part of the active site of protein³⁷Fig. 7.

For model validation, the ligand was docked into six snapshots derived from the molecular dynamics (MD) simulation to evaluate binding pose stability via RMSD analysis. Results indicated high stability, with an average pairwise RMSD difference of less than 1 Å. Furthermore, the binding site exhibited low RMSF values (0.12–0.24 Å), confirming structural compactness and minimal conformational changes throughout the simulation.

4. Discussion

Microbial community complexity and host interactions play a role in causing disease in patients with elevated harmful bacteria and decreased beneficial bacteria.³⁸ However, SNPs detection within virulence genes may contribute mainly to the persistence of these pathogenic bacteria in their host and SNP-driven structural changes can influence pathogenic behavior.³⁹ SNP detection can be an important issue if it leads to a change 3D structure of the protein and in turn leads to a change function of the protein. Therefore, we aimed to evaluate all SNPs detected in genes mediating biofilm formation in our previous work, as they change the structure and function of a protein that can permit changes in the capacity of bacteria to construct biofilm. Molecular docking and molecular dynamic simulation were performed using Meloxicam as ligand due to the effect of Meloxicam in enhancing downregulation for quorum-sensing genes in P. aeruginosa and can binds with active sites of P. aeruginosa QS transcriptional regulator proteins as proven by molecular docking studies.32, 40 Meloxicam exhibits synergistic effects with different antibiotics such as ciprofloxacin and inhibits the production of different virulence factors like elastase.⁴⁰ Starting with the luxS gene hosted in K. pneumonia, all the variations were synonymous, which means the protein encoded by the luxS gene does not change at all; therefore, this variation has no role in changing the capacity of bacteria to construct biofilm. However, Tang et al. 2024 showed that the deletion of the luxS gene from some isolates decreases the virulence and the ability of these strains to construct biofilm.⁴¹ Variation in the amino acid sequence was detected in two positions within the protein encoded by the lasI gene in P. aeruginosa isolate at R31K and D39N. These SNPs were located within isolates, forming moderate biofilm compared with non-biofilm producer as reference isolate [25]. Another study showed that mutations in lasI and rhII affect bacterial adhesion and biofilm formation.³⁴ SIFT and Mupro programs predicted that the D39N change was important because it changes the function, as SIFT scored less than 0.05 and Mupro showed an increase in the stability of the protein with a value of 0.2, which serves as a quorum-sensing signal to stimulate the P. aeruginosa to form biofilm. In addition, The INPS-3D program predicted that the D39N mutation affects the phi and psi angles (phi and psi angles were elevated from 53, 34 to 58, 40, respectively). Although the effect was not significant, it may represent a moderate local structural adjustment in allowed regions of the Ramachandran plot. Furthermore, the variation was located within the lasI protein domain. On the other hand, the R31K change was less important in serving as a quorum-sensing signal to stimulate P. aeruginosa to form biofilm as predicted by SIFT with score 0.34, Mupro, and INPS-3D program. Mupro program predicted that the R31K change would significantly decrease the stability of the protein with −1.6 and may affect its functionality. In addition, INPS-3D program revealed no significant change in RSA from 34% to 36% and phi angle from −97 to −98, while other parameters were stable. Regarding with rhII gene, Variation in the amino acid sequence was detected in one position in P. aeruginosa isolate at D44N. This SNP was located within isolate, forming moderate biofilm compared with non-biofilm producer as reference isolate. This gene code for protein also serves as a quorum sensing signal to stimulate the P. aeruginosa to form a biofilm. For the D44N mutation, although SIFT scored higher than 0.05, which is tolerant, and INPS-3D programs showed there was no significant change in RSA (67% to 81%), phi (57 to 63), and psi (18 to 22) angles, and other parameters. In addition, the simulation of molecular docking showed slightly change in binding energy from −7 to −7.2 kcal/mol. Polarity is also decreased from 16 to 14. However, the Mupro program showed an increase in stability of protein with a significant value of 0.5. In addition, molecular dynamics simulation showed deviation in the RMSF values of the mutant model compared with the wild type, (shifting from 1.5 to 3.5) indicating the impact of deleterious amino acid substitutions in the rhII, the protein. Moreover, molecular dynamic simulation revealed that 10 mutant models are more scattered than the 10 wild models and this means the D44N mutation increase in flexibility of the protein. Furthermore, molecular dynamic simulation showed A44 become in contact with A41 with a value of 0.7 instead of A44 and A47 with a value of 0.7 in the wild type, and this may reflect that amino acid substitutions in the rhII protein significantly affect the function. Therefore, we think variation in the rhlI gene and lasI gene was important, as both function as quorum-sensing signal pathways for biofilm production. Molecular Docking and molecular simulation study was carried out on monoterpenoid compounds against lasI and lasR and revealed strong binding through formation of hydrogen bonding at Ile107 of LasI, Tyr47 of LasR and inhibited the Pseudomonas aeruginosa biofilms.⁴² Organic mercury compound was used in molecular docking study against mercuric reductase from Pseudomonas fluorescens, revealed strong binding affinities, with key hydrogen bonds involving residues Gly95, Thr122, and Asp390.⁴³ A molecular docking and simulation study conducted by³¹ showed T240I, D128Y, and F228S mutation were predicted to be cancer-causing nsSNPs. Mutations in the rhII and lasI genes may affect the concentration of the autoinducer signal produced by the enzyme encoded by the rhII gene, hence affecting biofilm construction.⁴⁴,⁴⁵ showed that deletion in lasI/rhII promotes a decrease in the ability of isolates to form biofilm.⁴⁶ showed that the isolates that possess both lasI/rhII were more virulent than the isolates that carried only one gene mediating quorum-sensing signaling. For the ndvB gene, the variation was detected in weak biofilm producer compared with reference isolate. In spite of the fact that the SIFT program showed variation in the ndvB gene does not affect the function of proteins with score 0.46 and RSA increased from 30% to 41% and both those values with more than 25% are considered as exposed. Therefore, there is no change from exposed to buried, however, all other parameters showed dramatic changes in stability with 0.2, phi (−99 to 118), psi (118 to −94, 5), and secondary structure of protein (from B-sheet to coil). Therefore, we think that the change in S32N leads to a change function of the protein, which serves as an enzyme that has a role in the metabolism and transport of carbohydrates that form a sheath around bacterial colonies and increase the ability of bacteria to form biofilm. Another evidence for the importance of the S32N change in biofilm formation is the change located within the enzyme domain.¹⁸ showed that the ndvB gene is responsible for the formation of a glycan that participates in forming a sheath of biofilm around the colonies of bacteria; therefore, loss of this gene may diminish the biofilm productivity. The mutation in this gene may also effect on activity of glucosyltransferase enzyme hence the amount of glycan. and antibiotic resistance. Regarding tssC1 genes, the variation was detected in weak producer isolate compared with reference isolate. SIFT prediction showed that the R292A change led to a change in protein function with score 0.01. In addition, the Mupro program predicted a significant change in stability of the protein with score − 0.9, and INPS-3D analysis predicts the amino acid changed from exposed to buried for RSA (37% to 23%). In addition, psi dramatically increased from −41 to 22, and this leads to the conversion of the conformational state of the amino acid from helix to coil. In addition, the variation was located within the C0G3517 superfamily domain. All these indicate that the R292A change participates in increasing the ability of P. aeruginosa to form biofilm. More importantly, a frame shift mutation occurred within this tssC1 gene, and this led to a change large number of amino acids within this protein.⁴⁷ showed that the genic variation in ndvB and tssC1genes was 1% among isolates collected from different samples in the north of Iraq compared with the database in NCBI. The high variation may offer a mechanism for changing the ability of bacteria to form biofilm through changing proteins that involved with biofilm formation. The tssC1 gene engages with type VI secretion, which is correlated with increasing biofilm formation. It has been reported that a reduction in biofilm formation through deletion of the tssC1 gene in P. aeruginosa isolates.⁴⁸ Finally, the R31K and D39N mutation in lasI protein and D44N mutation in rhII protein were detected through comparing moderate biofilm producer with non-biofilm producer as reference isolate. While, S32N muation in ndvB protein and R292A mutation in tssC1 protein were detected through comparing weak biofilm producer with non-biofilm producer as reference isolate. Therefore, we think the bacteria adapt to stress conditions by changing their ability to form biofilm from non-biofilm producer to moderate or weak biofilm producers through genome variation. Bridging the gap between computational models and clinical results is a major hurdle in computational biology. Our structural and molecular dynamics analyses demonstrate that specific mutations (D44N) in (rhlI), (S32N) in (ndvB), and (R292A) in (tssc1)—drastically disrupt protein stability, structure, and active-site polarity, consistent with known clinical data.⁴⁹ Furthermore, quorum-sensing networks like (lasI) and (rhlI) drive biofilm development. Mutations altering these enzymes disrupt autoinducer production, resulting in the defective biofilms observed in laboratory experiments.⁴⁶ Regarding with clinical diagnosis, identifying these specific SNPs in clinical isolates could serve as rapid molecular markers to predict the virulence and biofilm-forming capacity of bacterial strains, allowing for tailored therapeutic interventions. Moreover, the mutant proteins (e.g., the highly flexible structure of the D44N rhlI mutant) can serve as targets for the rational design of quorum-sensing inhibitors (QSIs) or novel anti-biofilm agents. Our RMSF and (ΔΔG values) predictions provide clear targets for future in vitro site-directed mutagenesis and experimental functional assays (such as HPLC/LC-MS to measure autoinducer concentrations) to validate these computational models. However, this study serves as a hypothesis-generating workflow, prioritizing high-value nsSNPs for future experimental validation via site-directed mutagenesis and functional assays. In addition, the limitation of this study is the lack of evaluation of the effect of SNPs experimentally through measuring the change in biofilm quantification mediating genome variation. We entirely focused on prediction that some SNPs located within biofilm genes strongly affect biofilm formation. Another limitation in the current study is that predictions from different algorithms can sometimes be contradictory; therefore, analysis needs to be carried out using different programs. However, our analysis was carried out on the most famous programs, such as the Mupro program for detecting the stability of proteins.

CRediT authorship contribution statement

Khamaael Mohammed Fayyadh: Writing – review & editing, Writing – original draft, Software, Resources, Methodology, Formal analysis, Conceptualization. Mohammad Abdulrahman Mohammed: Writing – review & editing, Writing – original draft, Software, Resources, Methodology, Formal analysis, Conceptualization. Shaymaa Fouad Rasheed Al-Khazraji: Software, Methodology, Data curation, Conceptualization. Omar Abdulrahman Al-Maeni: Writing – review & editing, Software, Resources, Project administration, Methodology, Data curation, Conceptualization.

Declaration of competing interest

There is no Conflict of Interest.

Appendix A. Supplementary data

Supplementary material