Role of SMF-1 and cbl pili in Stenotrophomonas maltophilia biofilm formation

Abstract

Stenotrophomonas maltophilia is an emerging multidrug-resistant, Gram-negative opportunistic pathogen. It causes many healthcare-associated infections such as sepsis, endocarditis, meningitis, and catheter-related urinary tract infections. It also affects individuals with cystic fibrosis, exacerbating their lung condition. S. maltophilia often causes pathogenesis through the formation of biofilms. However, the molecular mechanisms S. maltophilia uses to carry out these pathogenic steps are unclear. The SMF-1 chaperone/usher pilus has been thought to mediate S. maltophilia attachment. To confirm this role, we created an isogenic deletion of the smf-1 pilin gene and observed a defect in biofilm compared to wild type. We also discovered an additional chaperone/usher pilus gene cluster: cbl. Mutation of cbl also affects biofilm levels. Intriguingly, through transmission electron microscopy studies, we found suggestive evidence that the mutation of one pilus (e.g. smf) is not phenotypically compensated by another (e.g. cbl). Additionally, infection of Galleria mellonella larvae revealed increased virulence of an smf-1 deletion mutant and an smf-1 cbl double deletion mutant. Together, these studies show that pili have an important role in switching between acute and chronic infections in conducting S. maltophilia virulence. Understanding their activity may help identify therapeutic targets for this pathogen.

1. Introduction

S. maltophilia is a multidrug-resistant, Gram-negative opportunistic pathogen identified as a leading cause of hospital-acquired infections [[1], [2], [3]]. It readily infects the respiratory tract of individuals suffering from improper pulmonary function [4,5]. For instance, it exacerbates the lung conditions of 11.9–14 % of individuals with cystic fibrosis (CF) [6]. Additionally, S. maltophilia causes bloodstream, wound, and urinary tract infections and infections in many other body sites [2,7,8]. Risk factors for S. maltophilia infection include Foley catheter usage, ventilator support, hemodialysis, neutropenia due to chemotherapy, septic shock, hematologic malignancy, human immunodeficiency virus infection, prolonged ICU stay, heavy antibiotic dependency, significant burns, and other immunocompromised situations [9].

One pathway through which S. maltophilia establishes and maintains its pathogenesis is through cellular adhesion, colonization, and biofilm formation [10,11]. Up to 98.7 % of clinical isolates are capable of producing biofilms [12] as are many environmental isolates, and these biofilms can form on biotic and/or abiotic surfaces [13,14]. The biofilm structure can protect the constituent bacterial cells from adverse environmental conditions and chemicals (such as antimicrobials) as well as the host immune system [15,16]. However, S. maltophilia's biofilm process is poorly understood. Several genes have been linked to biofilm in this microbe. For instance, polysaccharide-related genes rmlA, rmlC, xanB, and spgM impact biofilm formation [17,18], though each were assayed at different or unspecified time points. Additionally, genes controlling iron uptake and involved in producing an intercellular signalling molecule called diffusible signalling factor (DSF) partially regulate biofilm growth and development [19,20]. Our laboratory found that expression of phosphoglycerate mutase gene gpmA enhances S. maltophilia biofilm [21]. While mutation in all the above-mentioned genes decreases biofilm levels at certain time points, the broader implication of the effect of these genes in biofilm development remains unclear. Thus, a better understanding of the molecular mechanisms behind S. maltophilia biofilm formation is needed.

Chaperone/usher pili are extracellular, hair-like structures on the surface of many Gram-negative bacteria that play a key role in adhesion, biofilm, and host colonization [13,17,22]. These pili are assembled through the chaperone/usher pathway, which is essential for pilus biogenesis. In this process, the chaperone protein stabilizes pilus subunits in the periplasm, preventing premature aggregation and ensuring proper folding. The usher protein, located in the outer membrane, facilitates pilus assembly and transport by connecting individual subunits to form the growing pilus structure. Through their adhesive properties, chaperone/usher pili enable bacterial attachment to host cells and surfaces, contributing to biofilm formation and pathogenicity in host environments.

Previous studies have suggested a link between the putative chaperone/usher SMF-1 pilus and attachment and biofilm. One study, which included transmission electron microscopy (TEM), revealed that the SMF-1 pilus is important for attachment of S. maltophilia to biotic and abiotic surfaces, which attachment event can lead to subsequent biofilm formation [13]. However, this study focused on isolating the pilin protein and analyzing antibody reactions raised against this protein, rather than directly investigating the fimbrial gene and its mutants. A transposon mutagenesis suggested a connection between putative SMF-1 pilus operon genes and initial attachment and biofilm development [23]. Additionally, a few studies indicated a widespread distribution of the SMF-1 pilus among S. maltophilia clinical isolates, especially the isolates colonizing the respiratory tract of individuals with CF, based on PCR data, though this does not confirm actual pilus expression [24,25]. Given the above points, while the high prevalence of the smf-1 gene and its potential role in bacterial pathogenesis are suggested, it should be emphasized that no prior studies have specifically characterized mutants lacking smf-1. Furthermore, this high prevalence strongly suggests a connection between pilus production and pathogenesis, yet little is known about how the production of these pili affects infection development.

In this study, we generated an smf-1 isogenic deletion mutant and tested it for biofilm formation in different models to identify the role of the SMF-1 pilus in biofilm. Through bioinformatic analysis, we found an additional chaperone/usher pilus gene cluster that is also in nearly every strain: smlt3830-3833, which we name cbl. Isogenic deletions of the putative pilin genes of this gene cluster was also tested for impact on S. maltophilia biofilm. Finally, we used a Galleria mellonella infection assay to demonstrate that these pili affect virulence. Throughout the paper we will refer to these chaperone/usher pili and their associated pilins simply as “pilus” and “pili” for brevity. It is important to note that there are other types of pili, such as the type IV pili, apart from chaperone/usher pili that contribute to motility, adherence, and biofilm formation [6,26]. As we focus on chaperone/usher pili in this manuscript, though, this shorthand terminology is used exclusively for the chaperone/usher pili discussed in this study.

2. Materials and methods

2.1. Bacterial strains and culture conditions

For our study, we used clinical S. maltophilia isolate K279a, a fully sequenced type strain [27]. Escherichia coli strain S17-1 was used for creation and maintenance of genetic constructs [28]. Bacterial strains were grown in LB medium, with gentamicin as needed to maintain plasmids (10 μg/mL for plasmids in E. coli and 70 μg/mL for plasmids in S. maltophilia).

2.2. Screening of transposon library and identification of genes

Our previously-prepared S. maltophilia transposon library [21] was screened for clones that exhibited decreased biofilm formation, as previously described [21]. The identity of positive hits was performed by arbitrary-primed PCR, as described in that study [21].

2.3. Mutant construction

Generation of an smf-1 isogenic deletion was performed as we previously described [21]. Briefly, using PCR, we generated ∼1000 bp gene fragments of the regions immediately upstream and downstream of smf-1 with primer sets 706LFor/706LRev and 706RFor/706RRev (Supplemental Table 1). These primers incorporated overhangs with homology to cloning vector pMQ30 [29], which acts as a suicide vector in S. maltophilia [21]. These fragments were transformed into Saccharomyces cerevisiae INVSC1 (Invitrogen; Eugene, OR) along with pMQ30 restriction digested with BamHI, as previously described [30,31]. Through the process of recombination in the yeast, plasmid pΔsmf-1 was generated. Subsequently, the plasmid was isolated from S. cerevisiae [30] and transformed by electroporation to E. coli S17-1. The E. coli containing the plasmid pΔsmf-1 was conjugated with wild type S. maltophilia strain K279a. The exconjugants of this process were selected on LB plates with 70 μg/mL gentamicin (to select for bacterial clones with the plasmid) and 5 μg/mL norfloxacin (to select against E. coli) [21,32]. After 24–36 h of incubation, colonies were streaked for confirmation and then grown in LB medium overnight. This culture was then plated on LB agar plates with 10 % sucrose to select for loss of the plasmid, which contains a sacB sucrose sensitivity marker. Deletion mutants were confirmed by PCR with primers 706For/706Rev (Supplemental Table 1) and agarose gel electrophoresis, looking for a fragment of ∼359 bp, instead of ∼868 bp in the wild type strain.

In a similar way, we generated the pΔcblA (with primer sets 3833LFor/3833LRev and 3833RFor/3833RRev) deletion plasmid. This construct was then used to create ΔcblA strain (as described above for Δsmf-1), which was confirmed with 3833For/3833Rev. Using the pΔsmf and pΔcblA deletion plasmid constructs, we similarly generated the double mutant strain Δsmf-1 ΔcblA.

2.4. Complementation plasmid construction

We rescued the attenuated biofilm phenotype by introducing the smf-1 gene in trans. We amplified the full-length smf-1 gene along with a ∼500 bp fragment upstream of the gene with primer pair 706CompNatFor/706CompRev (Supplemental Table 1). Notably this 500bp upstream fragment does not include any predicted open reading frames (ORFs) or small RNAs (sRNAs). These fragments were transformed, along with expression vector pMQ132 that was digested with BamHI, into S. cerevisiae [29]. Through the process of recombination in the yeast, plasmid psmf-1 was generated. Subsequently, the plasmid was isolated from S. cerevisiae [30] and transformed by electroporation to E. coli S17-1. To confirm that the strain had taken up the plasmid, we performed PCR using primers p729/p730A (Supplemental Table 1) and observed the expected band at ∼300 bp on an agarose gel after electrophoresis. E. coli S17-1 with plasmid psmf-1 was then conjugated with S. maltophilia K279a Δsmf-1. The exconjugants of this process were selected on LB plates with 70 μg/mL gentamicin and 5 μg/mL norfloxacin. The pcblA plasmid as similarly created with primers 3833CompNatForNew2 and 3833CompRev (Supplemental Table 1). The ΔcblA/pcblA strain was confirmed using primers p729/p730A.

2.5. Biofilm assays on polystyrene microtiter plate (abiotic surface)

For performing biofilm assays, the indicated strains were grown overnight in 5 mL LB medium at 37 °C to an OD₆₀₀ of 1, and then diluted 1:100 into fresh LB medium [30]. These diluted strains were then added at a volume of 100 μL (2 × 10⁶ CFU/mL final concentration) in individual wells of a 96-well polystyrene microtiter plate (Greiner Bio-One; Monroe, NC) and incubated statically for the indicated time point at 37 °C. Each strain was added in quadruplicate in the plate. After the incubation, the plates were stained with 0.1 % crystal violet (CV) for 15 min to detect bacterial cell attachment and biofilm formation, as described [33]. The wells were washed to remove any unattached stain and dried overnight. Stain was then solubilized using 30 % acetic acid and 125 μL were transferred to a new 96-well polystyrene microtiter plate. The intensity of the stain was then quantified with a SpectraMaxM2 spectrophotometer (Molecular Devices; Sunnyvale, CA) by measuring optical density at 550 nm (OD₅₅₀). We also performed a similar biofilm assay in Synthetic Cystic Fibrosis Sputum Medium (SCFM2) medium that mimics the chemical composition of the CF lung environment [34].

2.6. Biofilm model on cystic fibrosis bronchial epithelial (CFBE) cells (biotic surface)

Biofilm assays were performed in a model system involving immortalized human CFBE that has been used to develop P. aeruginosa biofilms in the context of airway epithelium [35]. In our prior work, we used this system to investigate early biofilm formation of S. maltophilia on airway cells [21]. CFBE cells were seeded in a 24-well tissue culture plate (Falcon; Franklin, NJ) at 2 X 10⁵ cells/well in Minimum Essential Medium (MEM) (Corning inc.; Corning, NY), with 10 % fetal bovine serum (FBS) (Atlanta Biologicals; Lawrenceville, GA), 50 U/mL penicillin, and 50 U/mL streptomycin (Lonza; Walkersville, MD), as described [36]. The cells were allowed to grow for 7–10 days at 37 °C and 5 % CO₂, feeding with new media every 2 days. After 7–10 days, when the cell monolayer was confluent, the cells were washed with 1X phosphate buffered saline (PBS). Bacteria grown overnight in 5 mL LB medium at 37 °C to an OD₆₀₀ of 1 were inoculated to the confluent cells at a multiplicity of infection (MOI) of 30:1 in 500 μL MEM without any phenol red, FBS, or supplementary antibiotics. Infected cells were then incubated for 1 h, after which the medium was aspirated and fresh MEM (500 μL), also without phenol red, FBS, or antibiotics, was added and the cells were further incubated until the indicated time point was reached. After this incubation, the cells were washed with an equal amount of PBS 2–3 times and then treated with 0.1 % Triton X-100 for 15 min. The cells lysates were harvested into microcentrifuge tubes, vortexed for 3 min, serially diluted, and plated on LB agar plates to determine colony forming units (CFU) as a measure of bacterial attachment and biofilm formation. Each bacterial strain was tested in triplicate within this biofilm model.

2.7. TEM studies

Bacterial cultures were grown overnight at in 5 mL LB medium at 37 °C (OD₆₀₀ = 1) then washed twice with 1X PBS and negatively stained with 2 % phosphotungstic acid on carbon-Formvar nickel grids [13]. 3 μL (∼6000 bacterial cells) were then visualized by Tecnai 12 transmission electron microscope (FEI, Hillsboro, OR) with a LaB₆ crystal at 120 kV. Manual quantification of pili was performed on 5 separate micrographs per strain.

2.8. G. mellonella infection

S. maltophilia virulence was tested in vivo in a G. mellonella infection model [37]. Final instar G. melonella larvae were obtained from BugCo (Ham Lake, MN). Bacterial cultures were grown overnight in 5 mL LB medium at 37 °C to an OD₆₀₀ of 1, washed twice in 1X PBS and resuspended in the same. Using 20 μL Hamilton syringes, 10 μL of each bacterial strain at 10⁵ CFU/10 μL was injected into the last prolegs of final instar G. mellonella larvae. Each bacterial strain was injected in 5 larvae. After infection, these larvae were incubated in filter paper lined Petri dishes at 37 °C for 7 days and scored for survival daily. Larvae were scored dead if they showed lack of movement and increased melanization. A control group of larvae was also inoculated with 1X PBS to record any killing of larvae due to mechanical injury; however, all PBS-injected larvae survived over the time course of the experiment. Survival curve was plotted using the Kaplan-Meier method, and log-rank test was calculated to measure significant differences in survival (Graph Pad Prism version 9.0, Software Inc., La Jolla, CA).

2.9. Growth kinetics

Bacterial strains were grown overnight in 5 mL LB medium at 37 °C to an OD₆₀₀ of 1, which were then diluted 1:100 into fresh LB medium, SCFM2, or MEM. 100 μL of each strain was added to wells of a 96-well polystyrene microtiter plate in quadruplicates. The plates were then incubated at 37 °C in a SpectraMax M2 spectrophotometer and the absorbance was read every 30 min for 20 h at OD₆₀₀, with 5 s of shaking before each reading.

2.10. Statistical analysis

Each experiment was performed at least three times with triplicate or quadruplicate samples for each strain. Student's t-test was used to determine statistical significance. A difference was considered statistically significant at a p value of <0.05. As mentioned, a log-rank test was used to determine statistical significance for the G. mellonella infection survival, with p < 0.05 considered significant.

3. Results

3.1. The smf-1 gene is important in biofilm formation

We previously generated an S. maltophilia transposon mutant library of ∼5760 mutants derived from the wild type strain [21]. We screened this library for mutants that exhibited visually decreased biofilm levels compared to wild type and then used arbitrary-primed PCR (as previously described [21]) to identify the location of transposon insertion in hits (Supplemental Table 2). Two of our positive hits mapped to the smf-1 gene, encoding the major pilin protein of SMF-1 pili. Sequence analysis indicated that the smf-1 gene is part of a gene cluster (Fig. 1A) that putatively encodes a chaperone/usher pilus system, SMF-1 ^13,23. In support of this assertion, smf-1 has been found to share homology with fimbrial proteins from other bacteria (Supplemental Fig. 1A). Through bioinformatic analysis, we found another putative chaperone/usher pilus gene cluster: smlt3830-3833 (Fig. 1B). This search was done on the full genome of S. maltophilia strain K279a available in the NCBI Genome database, and the discovery was made by searching the NCBI Genome database using search terms such as “pili”, “pilin”, “chaperone”, “usher” to identify genes potentially involved in pilus assembly and biofilm formation. The locus at smlt3833 has homology to the giant cable pilus, Cbl of Burkholderia cepacia [38,39] (Supplemental Fig. 1B), and we thus name this system CBL, though this designation requires further experimental confirmation. The proximity of the genes in these gene clusters suggest likely operons, and operon prediction software indicates operon structures (http://microbesonline.org/operons). However, this determination also requires further experimentation. The first genes in these putative operons, smf-1 and what we now name cblA, are the likely pilin genes for their respective pilus systems, based on genome annotations and operon structure [22]. We speculate that the CBL pilus systems never came up in our original transposon screen for decreased biofilms because that screen was not exhaustive [21]. That is, once we started identifying repetitive mutants, we focused on exploring their potential contributions to biofilm formation without fully evaluating all possible candidates.

Fig. 1.

Physical maps of putative S. maltophilia chaperone/usher pilus gene clusters. (A) smf-1 (B) cbl. Gene numbers are listed below each arrow, and putative functions as listed in the genome annotation are listed below. Arrow sizes indicate relative length of each gene. The cbl gene cluster is transcribed on the complementary strand.

To identify the role of these genes, we first used allelic replacement to generate isogenic deletions of the smf-1 gene to create the strain Δsmf-1. Similarly, we generated a ΔcblA strain. We then used transmission electron microscopy (TEM) to visualize pilus expression on the bacterial surface. We saw distinct surface expression of pili in the wild type bacteria, falling into 2 categories: one pilus structure appeared as shorter structures, which correlates to the purported SMF-1 (chaperone/usher) pili seen in previous TEM studies [13]; a longer pilus structure also appeared, which likely corresponds to type IV pili, as seen in previous studies [6] (Fig. 2). Although the correct determination of type IV pili and chaperone/usher pilus fibers requires further experimental validation through techniques such as immunogold electron microscopy (EM) or purification and analysis of surface fibers, our qualitative observations suggest that there were marked decreases of surface chaperone/usher pili in both the Δsmf-1 strain and the ΔcblA strain (Fig. 2), though type IV pili were still present. Manual quantification revealed an average of 58 pili on wild type cells, 1 pilus on Δsmf-1 cells, and 8 pili on ΔcblA cells.

Fig. 2.

TEM studies show the presence of chaperone/usher pili (short and hairy) and type IV pili (long) in different bacterial strains. TEM images of WT, Δsmf-1, ΔcblA, and Δsmf-1 ΔcblA strains. Strains are indicated underneath each micrograph. Shorter pili correlate to chaperone/usher pili and longer structures correlate with type IV pili, as seen in previous studies. See text for details. (scale bars are as indicated on the micrographs).

Because SMF-1 has been implicated in biofilm, we wanted to confirm this activity genetically. We complemented the Δsmf-1 strain in trans with plasmid psmf-1 to create the strain Δsmf-1/psmf-1. Biofilm assays with wild type, Δsmf-1, and Δsmf-1/psmf-1 strains were performed as described in Materials and Methods, and we saw that the Δsmf-1 strain produced significantly decreased biofilm levels compared to the wild type (∼90 fold difference at 24 h), whereas complementation restored the wild type phenotype in 96-well polystyrene microtiter plates at both 4 h and 24 h (Fig. 3A). Importantly, growth analysis in LB indicated that the Δsmf-1 strain grows the same as wild type (Supplemental Fig. 2).

Fig. 3.

Isogenic deletion of smf-1 dramatically reduces S. maltophilia biofilm formation. Biofilms of S. maltophilia wild type (WT), Δsmf-1, and complemented strain Δsmf-1/psmf-1 were formed (A) in LB medium, (B) on CFBE cells, and (C) in SCFM2 medium for 4 h (left panels) or 24 h (right panels). ∗p < 0.05. These figures are the combination of three independent experiments (n = 3) carried out in quadruplicate and error bars indicate standard deviations.

3.2. The smf-1 gene is important for biofilm formation in infection-relevant systems

As mentioned, S. maltophilia forms biofilms during infection, such as within the lungs of individuals with CF [40]. Thus, we tested biofilm formation in infection-relevant biofilm systems. Previous studies showed that S. maltophilia can form biofilm on several types of lung cells [14,21]. In a prior publication, we described a model of S. maltophilia biofilm formation on CFBE cells [21]. Using this model, we found, similar to the abiotic biofilm assay, a significant decrease in biofilm of the Δsmf-1 strain compared to the wild type (∼1.943 fold difference) and complemented strain Δsmf-1/psmf-1 (∼1.745 fold difference) (p = 0.008 and p = 0.0019 respectively at 24 h), but no significant difference between wild type and Δsmf-1/psmf-1 (Fig. 3B), though it is unclear if this defect is simply bacterial attachment or biofilm formation. Additionally, we used SCFM2 medium to provide an additional system for determining how our pilus mutants behave in a more infection-relevant environment. SCFM2 has been used previously in biofilm assays, including with S. maltophilia, to mimic the complex environment of the CF lung, incorporating experimentally measured levels of the ions, free amino acids, glucose, lactate, mucin, lipids, proteins, DNA, and other nutrients found in the thick sputum of individuals with CF [[41], [42], [43]]. Using this medium, we saw significant attenuation in biofilm levels for Δsmf-1 compared to the wild type in both early (∼1.315 fold difference) and later time points (∼35.39 fold difference) (Fig. 3C). Complementation partially restored wild type biofilm levels. We conducted growth kinetics in SCFM2 medium, and all strains exhibited planktonic growth similar to wild type (Supplementary Fig. 3).

3.3. Deletion of cblA pilin gene results in decreased biofilm levels on abiotic surfaces

In addition to Δsmf-1 and ΔcblA strains, we created a double mutant strain, Δsmf-1 ΔcblA. Biofilm assays were conducted in 96-well polystyrene microtiter plates with ΔcblA and the Δsmf-1 ΔcblA double mutant strain, using wild type (WT) and Δsmf-1 as controls. Δsmf-1, ΔcblA, and the double mutant strain (Δsmf-1 ΔcblA) produced significantly decreased biofilm levels in LB compared to the wild type (WT) at both early and later biofilm timepoints (Fig. 4A). This result suggests that smf-1 and cblA are both necessary in carrying out biofilm formation. Additionally, growth curves showed that the ΔcblA strain grows similarly to wild type (Supplemental Fig. 4). Interestingly, though we expected that Δsmf-1 ΔcblA double mutant to exhibit at least the same level of biofilm impairment as the smf-1 single mutant, the double mutant strain displayed a phenotype intermediate between the individual deletion mutants (Fig. 4A). This surprising outcome suggests the possibility of a more complex interaction between the pili systems in S. maltophilia. It is plausible that a coordinated or compensatory mechanism between these pili may be at play, mitigating the anticipated increase in biofilm impairment. These findings indicate that biofilm formation in this organism may involve intricate regulatory networks, potentially reflecting the functional overlap or interaction of multiple pili. Further investigation is needed to fully understand these dynamics. Reintroduction of the cblA gene into the ΔcblA strain (with pcblA plasmid) partially restored biofilm formation at 4 h; increase in biofilm at 24 h was not significant, though (Supplemental Fig. 5). The ΔcblA and ΔcblA/pcblA strains grew similarly to wild type in LB (Supplemental Fig. 6).

Fig. 4.

Pilus mutants form decreased biofilm. Biofilms were formed with S. maltophilia wild type (WT), Δsmf-1, ΔcblA, Δsmf-1 ΔcblA (ΔΔ) in 3 models. (A) Biofilms were formed in LB medium for 4 h (left panel) or 24 h (right panel). 4 hr fold differences = 12.36, 3.66, 4.99 for Δsmf-1, ΔcblA, and Δsmf-1 ΔcblA, respectively, compared to WT; 2.47, 1.36 for Δsmf-1 and ΔcblA, respectively, compared to Δsmf-1 ΔcblA. 24 hr fold differences = 5.68, 1.27, 1.83 for Δsmf-1, ΔcblA, and Δsmf-1 ΔcblA, respectively, compared to WT; 3.11, 1.44 for Δsmf-1 and ΔcblA, respectively, compared to Δsmf-1 ΔcblA. (B) Biofilms were formed on CFBE cells for 4 h (left panel) and 14 h (right panel). There was significant difference in biofilm levels in the Δsmf-1 ΔcblA and double mutant compared to the WT at 14 h ∗p < 0.05. 4 hr fold differences = 16.62, 6.01, 11.61 for Δsmf-1, ΔcblA, and Δsmf-1 ΔcblA, respectively, compared to WT; 1.43, 1.93 for Δsmf-1 and ΔcblA, respectively, compared to Δsmf-1 ΔcblA. 24 hr fold differences = 48.93, 2.70, 30.06 for Δsmf-1, ΔcblA, and Δsmf-1 ΔcblA, compared to WT; 1.63, 11.11 for Δsmf-1 and ΔcblA, respectively, compared to Δsmf-1 ΔcblA. (C) Biofilms were formed in SCFM2 medium for 4 h (left panel) or 24 h (right panel). ∗p < 0.05. 4 hr fold differences = 6.4, 1.97, 2.47 for Δsmf-1, ΔcblA, and Δsmf-1 ΔcblA, respectively, compared to WT; 2.59, 1.25 for Δsmf-1 and ΔcblA, respectively, compared to Δsmf-1 ΔcblA. 24 hr fold differences = 10.83, 5.56, 3.65 for Δsmf-1, ΔcblA, and Δsmf-1 ΔcblA, respectively, compared to WT; 2.97, 1.53 for Δsmf-1 and ΔcblA, respectively, compared to Δsmf-1 ΔcblA. These figures are the combination of three independent experiments (n = 3) carried out in quadruplicate and error bars indicate standard deviations.

3.4. Deletion of pilin genes results in decrease in biofilm levels on biotic surfaces

We also studied biofilm levels of wild type, Δsmf-1, ΔcblA, and Δsmf-1 ΔcblA strains at 4 h and 14 h on CFBE cells [30]. We found that by 14 h, the pilus mutant strains exhibited a greater than 2 log unit reduction in biofilm formation, compared to wild type. We saw a similar trend at 4 h, though this difference was not statistically significant (Fig. 4B). Once again, we are unsure of whether this defect is due to reduction in biofilm formation or simply bacterial attachment. We conducted growth kinetics in MEM and all strains exhibited planktonic growth similar to wild type (Supplementary Fig. 4).

3.5. Deletion of pilin genes results in decrease in biofilm levels in SCFM2 medium

We next examined the role of both smf-1 and cblA pili in biofilm formation in SCFM2 medium. We saw that in both early (4 h) and later (24 h) time points, Δsmf-1 mutant showed significantly attenuated biofilm levels compared to that of the wild type (Fig. 4C). For both early and later stages of biofilm development the ΔcblA and double mutant Δsmf-1 ΔcblA had lower biofilm formation with a significant reduction seen in both these strains only in the later stages (∼2.59 and ∼1.25 fold difference between Δsmf-1 and ΔcblA with Δsmf-1 ΔcblA double mutant strain respectively at early stages) (∼2.97 and ∼1.53 fold difference between Δsmf-1 and ΔcblA with Δsmf-1 ΔcblA double mutant strain respectively at later stages). We conducted growth kinetics in SCFM2 medium, and the mutants exhibited planktonic growth similar to wild type (Supplementary Figs. 3 and 8).

3.6. TEM studies reveal that deletion of one pilus gene is not phenotypically compensated by others

As shown previously, deletion of smf-1 and cblA genes decreased chaperone/usher type pili on the surface of the bacterial cell (Fig. 2). Importantly, we also observed a marked decrease in chaperone/usher type pili in the Δsmf-1 ΔcblA strain (Fig. 2), though type IV pili were still present, albeit in lesser numbers in the ΔcblA strain. Manual quantification indicated zero chaperone/usher pili on double mutant cells. Notably, the phenotypic loss of surface chaperone/usher pili in the Δsmf-1 strain was not compensated by the presence of other pilus genes and vice versa. That is, when only one pilus is mutated and the other gene cluster is intact, but there are still no pili present on the surface. We speculate that mutation of one chaperone/usher pilus might influence the production of the others, possible indicating a regulation in pilus expression during growth and biofilm formation. While these TEM results are supportive of the biofilm analysis data, they should be interpreted with caution, particularly because these observations were qualitative and not quantitative.

3.7. Pilus mutants show increased virulence in a G. mellonella infection model

G. mellonella has been used as an infection model for studying host-pathogen interactions of many bacteria, including S. maltophilia [37]. It is cost effective, easy to maintain, and allows for high throughput experiments. The organism shares a similar innate immune system with vertebrates, and while not exactly the same, it has been used as a biologically relevant model for studying host-pathogen interactions [44]. Research using this model has provided insights into the virulence of S. maltophilia and its implications for chronic infections in immunocompromised individuals [45]. We infected G. mellonella larvae with wild type and pilus mutant strains, as described in Materials and Methods. Larvae infected with wild type S. maltophilia resulted in 0 % mortality (Fig. 5). In contrast, infection with Δsmf-1 and Δsmf-1 ΔcblA led to significantly greater mortality: 60 %, and 100 %, respectively (p < 0.05, compared to wild type). We saw the highest overall death rate in larvae infected with the double mutant Δsmf-1 ΔcblA, which showed 80 % mortality in 1 day and 100 % mortality within 7 days of infection (p ≤ 0.05, compared to wild type), proving it to be the most virulent of the strains we tested. Although there was a trend toward decreased survival of larvae infected with ΔcblA, this difference was not statistically different from wild type. Likewise, there was no difference in survival between the wild type and the Δsmf-1/psmf-1 complemented strain (0 % mortality). A log rank test between the Δsmf-1 and Δsmf-1 ΔcblA strains showed a p-value of 0.1936, indicating no statistically significant difference in impairment. However, since our experimental assays suggested that the Δsmf-1 ΔcblA strain is the most virulent among all strains tested, in vivo studies with complementation of ΔcblA strain were not pursued, as the statistical analysis did not support a significant difference requiring additional validation. Importantly, all larvae in our control group (injection with 1X PBS) survived through the course of the infections, demonstrating a lack of death due to mechanical injury.

Fig. 5.

Pilus mutants are highly virulent in the G. mellonella infection assay.G. melonella larvae were infected with wild type (WT), Δsmf-1, ΔcblA, Δsmf-1 ΔcblA, and Δsmf-1/psmf-1 strains, and survival was monitored over 7 days. The graph is the combination of three independent experiments (n = 3) carried out in quintuplicate. ∗p < 0.05 compared to the WT.

4. Discussion

Our results confirmed that the SMF-1 chaperone/usher pilus is important for bacterial attachment and biofilm formation in several systems, both biotic and abiotic (Fig. 3). We also found that additional pilus system, CBL (Fig. 1), is important for biofilm in these models (Fig. 4), while not impacting overall growth (Supplemental Figs. 3-4, 6-8). It is important to note that these pili affect biofilm in infection-relevant models. However, the strain Δsmf-1/psmf-1 shows in partial complementation when biofilm is assayed in SCFM2 medium. The vector backbone of psmf-1 (pMQ132), contains a plac promoter [46] upstream of the psmf-1 construct. We hypothesize that the sugar composition of the SCFM2 medium leads to catabolite repression from this plac promoter, partially inhibiting downstream transcriptional activation of smf-1; this inhibition could subsequently lead to decreased attachment/biofilm. Further investigation would be required to confirm this hypothesis. Additionally, the Δsmf-1 strain appears to be more important for initial attachment in standard LB medium that in SCFM2 medium or CFBE cells. We speculate that this discrepancy could be the results of variability in experimental conditions between the biofilm model systems. For example, in the CFBE model, there are likely additional substrates to which S. maltophilia surface molecules can bind, or the chemical composition of the medium or molecules released from the airway cells could lead to upregulation of additional attachment factors. It will be also important to validate and confirm whether S. maltophilia forms biofilm in this model. This is an intriguing aspect, and one which will need to be explored in future studies.

Comparing all the models tested, single and double mutant strains behave similarly across biofilm systems. Focusing on individual mutants, our data suggest that smf-1 and cblA are impactful on biofilm level in each model. However, the evidence of pilus coordination seen in Fig. 4 indicates that the connection between these pili is complex. Similarly, TEM analysis provided suggestive evidence of piliation phenotypes that cannot be explained by the additive effects of mutations of the individual pilin genes (Fig. 2). Additionally, it is not clear why mutation of one pilin gene is not phenotypically compensated by the presence of the others. We speculate some sort of transcriptional coordination, or some other physical or mechanical connection, and thus a complex regulatory interplay, between these 2 pilus systems. It is possible that mutation of these 2 pili leads to expression of other as-yet-undiscovered pili. It will be important in future studies to follow up on this hypothesis and other pili genes by transcriptional analysis and mutational studies.

Certain bacterial species inversely regulate their acute infection processes (e.g. toxin production) with chronic infection processes (e.g. attachment and biofilm) [47,48] and, based on our results, we hypothesize that a similar activity occurs within S. maltophilia. In the G. mellonella infection model, we found the inverse of the biofilm results: wild type was non-virulent while pilus mutants were highly virulent. We found that strains with the greatest biofilm defect (Δsmf-1, Δsmf-1 ΔcblA) show the maximum virulence within the host organism (Fig. 5). We speculate that, with loss of pilus, the cells are unable to attach and start biofilm development, resulting in systemic spread and rapid killing of G. mellonella. Similar phenomena, of alternately regulating acute toxicity and chronic biofilm formation, have been seen in other bacteria [47]. For instance, in Pseudomonas aeruginosa strain PAK, mutation of retS resulted in reduced activity of Type III secretion system effector molecules and enhanced cellular adhesion to surfaces, thereby increasing its biofilm-forming properties. In this system, the retS mutation inversely regulates two key processes — adhesion/biofilm formation and secretion of virulence factors — ultimately influencing the pathogen's behavior and shifting it from acute cytotoxicity to a chronic infection state. While the mechanisms may differ between P. aeruginosa and S. maltophilia, a comparable regulatory switch may be at play in our study. In S. maltophilia, pilus deficient mutants (Δsmf-1, Δsmf-1 ΔcblA) exhibited heightened virulence in this infection model. This increase in virulence is likely associated with its inability to form biofilms, which could trigger a shift toward systemic spread and the upregulation of toxins, secretion system effector molecules, and other virulence factors. Other examples of this inverse relationship between acute and chronic phenotypes exist [49,50]. This retS example in P. aeruginosa and others provide a conceptual parallel to the potential regulatory trade off we observe in S. maltophilia. We hypothesize that the loss of pili represses biofilm formation while activating pathways that promote acute virulence, mirroring the inverse regulation of biofilm formation and virulence factor secretion seen in P. aeruginosa [51]. Further studies will be needed to unravel the precise mechanisms underlying this phenomenon.

Our results indicate that S. maltophilia produces multiple different chaperone/usher pili, and that these pili each contribute directly or indirectly to attachment and biofilm development of this microorganism. To our knowledge, this is the first report to provide genetic evidence of this phenomenon. It is possible that these multiple pili permit attachment to various surfaces in different environments. It will be important in future studies to identify the receptors for these factors. Additionally, their regulation, and potential co-regulation with each other and with other factors, needs to be addressed. In this manner, we can begin to generate an integrated model for how this deadly pathogen initiates infection and causes disease.

CRediT authorship contribution statement

Radhika Bhaumik: Writing – review & editing, Writing – original draft, Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Alli Beard: Investigation, Formal analysis, Data curation. Oliver Harrigan: Investigation, Formal analysis, Data curation. Layla Ramos-Hegazy: Methodology, Investigation, Formal analysis, Data curation, Conceptualization. Seema Mattoo: Writing – review & editing, Resources, Funding acquisition. Gregory G. Anderson: Writing – review & editing, Supervision, Project administration, Methodology, Investigation, Funding acquisition, Formal analysis, Data curation, Conceptualization.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary data

The following is the Supplementary data to this article.

Data availability

Data will be made available on request.