As biofilms former P. aeruginosa strains are one of the major health threats in clinical seating all over the world, therefore, it is very important to assess the biofilm forming potential and the mechanisms of clinical isolates. Different studies on biofilm formation of clinical P. aeruginosa isolates showed variability in the biofilm forming ability around the world [18–20]. Although all of the tested isolates showed some sort of attachment tendency, we found 45% of P. aeruginosa isolates (n = 20) from wound swab, pus, urine, and tracheal aspirate showed strong adherence to the surface, which is similar to the finding of Samad et al., 2019.
Several genes have been found to play important roles in P. aeruginosa biofilm formation. P. aeruginosa DNA microarray analysis revealed only 1% of genes that are differentially expressed in the biofilm growth mode [21]. Four major pathways (cAMP/Vfr signaling, c-di-GMP dependent polysaccharide synthesis, quorum sensing, and the Gac/Rsm pathway) in P. aeruginosa play vital roles in receiving and processing external signals into its regulatory control at the transcriptional, translational, and post-translational levels and thus regulate biofilm formation [22, 23]. In this background, we selected 4 biofilm genes (pelB, lecB, pilT and rhlB) that have different roles (e.g., pel production, lectin binding, twitching motility, quorum sensing) in P. aeruginosa biofilm formation for initial screening of biofilm genes. All of the tested isolates possessed these genes despite their variable biofilm phenotypes. Therefore, complete genome sequences of representative strong, moderate, and weak biofilm former isolates (27b, 20c and 30b) were analyzed to find any sequence variation in biofilm related genes that may have contributed to biofilm formation. Analysis of antibiotic resistant genes and gene cassettes of 27b from its CGS was previously reported by Jahan [24] and genomic diversity and molecular epidemiology of 30b was reported by Hoque [25]. We analyzed the sequences of 88 proteins and regulatory RNA related to these biofilm forming pathways in isolates 27b, 20c, and 30b by comparing them with the reference genome of P. aeruginosa PAO1. Among those proteins and RNAs, only LecB and pel operon proteins showed significant aa sequence divergence.
According to a previous report, LecB from the highly virulent model strain PA14 has a 13% sequence divergence with LecB from the well-characterized PAO1 strain. This difference can also result in differing ligand binding specificities and, ultimately, reduced efficacy of drugs directed towards LecB [26]. Our partial sequence data (Fig. 3e) and corresponding biofilm phenotype analysis showed that isolates that have PA14 like LecB, are strong biofilm formers and isolates that have PAO1 like LecB, are moderate or weak biofilm formers.
To date, no report is available that correlates this sequence divergence with biofilm forming ability. One study regarding this revealed that LecB binds with Mannose-alpha 1,3 mannobiose of Psl and thus has a profound impact on biofilm architecture and biomass in PAO1 [13]. We observed that PA14 like LecB differs from PAO1 like LecB in Psl binding site at positions 24 (PA14 LecB Ser; PAO1 LecB Ala) and 98 (PA14 LecB Ser; PAO1 LecB Gly) (Fig. 3d). It can be noted that PA14 strain itself cannot produce Psl, but other strains that have PA14 like LecB can produce Psl. Variation in the Psl binding site in Psl-producing strains may have an impact on biofilm architecture. Further investigation is therefore necessary to validate this hypothesis.
The pel operon plays a vital role in the synthesis of Pel, which is an aggregative polysaccharide produced by P. aeruginosa. This operon encodes seven enzymes (pelA, pelB, pelC, pelD, pelE, pelF, pelG); however, several genes predicted to be essential for pel production are missing from this operon [27]. Individual BlastN searches in the NCBI database of individual pel operon genes of isolates 27b, 20c, and 30b revealed that all of the pel operon genes of 30b have a certain amount (5–13%) of sequence variation from the majority of other pel operon genes in the database, and only 6 strains share 98–100% sequence identity with Pel operon genes of 30b (Supplementary Table 6). This data suggests that 30b, like Pel operon sequences, is very rare among P. aeruginosa isolates. A BlastP search in the NCBI database revealed that 30b pel operon proteins share sequence homology with P. aeruginosa PA7 pel operon proteins (data not shown). Based on these findings, we propose two types of pel operons: PAO1-like (reference) and PA7-like (variant).
It has been reported that thick pellicle is observed in strains overexpressing the pel operon genes [27]. Recent studies have also described a role for the flagellum regulator FleQ as both a repressor and an activator to control gene expression from the pel operon promoter in response to c-di-GMP [28]. The absence of these FleQ binding sites may play a vital role in the expression of pel operon proteins. In this study, we found that FleQ binding sites are absent in the upstream of the 30b (PA7 like) pel operon. The pellicle forming and Congo red assays also revealed that WBF 30b is unable to form biofilm at the air-liquid interface.
Although, the regulator FleQ binding sites are absent in the upstream of the 30b pel operon, we found the presence of pelB transcripts with a common pelB primer (Supplementary Table 2), which indicates pel operon proteins might have been expressed in 30b. On the other hand, genomic sequence and protein structure analysis of the 30b pel operon revealed significant variation in comparison with the reference PAO1 pel operon. These variations can affect the functions of these individual proteins in a way that can result in failure to Pel production.
It was previously reported that PelA has a tat-dependent signal sequence, suggesting the protein is localized to the periplasm [29]. Our findings indicate that PelA of 30b cannot be transported across the cytoplasmic membrane via the Tat secretion machinery as it does not have the twin arginine residue in its tat recognition motif. It was also predicted that, at least four, and possibly five, distinct domains, three of which have structural similarities to proteins with hydrolase, reductase, and deacetylase activity [29]. aa variations in those protein domains of 30b may affect the function of PelA.
It is known that, PelA and PelB directly interact with each other. The TPR-containing domain of PelB localizes PelA to the Pel polysaccharide secretion apparatus within the periplasm [30]. When pel is deacetylated by pelA, it becomes positively charged. As a result, Pel is drawn toward the electronegatively charged PelC, which guides Pel toward the exit channel formed by PelB. In this proposed model, PelC functions as an electronegative funnel by forming a dodecamer ring around the ß-barrel domain of PelB ([31]. Our 3D structure prediction analysis revealed conformational changes in the ß-barrel of the TPR domain in 30b PelB (PA7 like PelB). We have also shown that, aa variations in 30b PelC (PA7 like PelC) seem to alter the conformation of the protein unit, which affects PelC dodecamer ring formation.
PelD, PelE, PelF, and PelG are responsible for pel polymerization and transport across the cytoplasmic membrane [32, 33]. It was also previously reported that c-di-GMP functions post-translationally in Pel synthesis by modulating the activity of PelD. R161, D367, R370, and R402 are the 4 aa’s that interact with c-di-GMP, which are present in all of our sequenced isolates (27b, 20c, and 30b). On the other hand, PelF uses UDP-glucose as a donor substrate toward the biosynthesis of the Pel exopolysaccharide. PelF's E405, R325 and K330 are proposed to be its UDP glucose binding sites [32]. Our study revealed that 30b PelF (PA7 like PelF) has a lower binding affinity to UDP glucose than PA01 like PelF. Moreover, aa divergence in PA7 like PelD, PelE, PelF, and PelG may affect their interaction. In Fig. 6, we summarized the possible reasons that can adversely affect Pel production machinery in PA7 like pel operon possessing strains. Isolate 27b and 20c have PAO1 like pel operon sequences, and both of them were able to produce Pel polysaccharide. Our findings based on in silico analysis suggest that PA7 like pel operon containing strains are unable to produce Pel polysaccharide as a component of their biofilm matrix.
The main limitation of our study is that we mainly focused on genomic variance and in silico modeling of important biofilm related proteins. In vitro analysis of the proteins we discussed here can give us more conclusive information about the correlation of biofilm forming ability and sequence variation of LecB and pel operon proteins.
Our study suggests that MDR clinical P. aeruginosa isolates from Bangladesh differ in their biofilm phenotypes. Variation in aa sequences in the lectin binding protein LecB and the Pel polysaccharide producing operon proteins in those isolates was found to be related to their biofilm phenotypes. PA14-like LecB protein sequences correlate with increased biofilm formation, while PA7-like pel operon sequences correlate with decreased or impaired Pel polysaccharide production. Strain-family classification of P. aeruginosa is therefore important to understand the multifactorial nature of biofilm formation and to introduce effective therapeutics against them.