The pagN gene: a dual genetic determinant for biofilm formation and virulence in Salmonella Typhimurium Authors: Asiye Nur Karaca, Nefise Akçelik, Mustafa Akçelik Published: 2026 ABSTRACT Abstract Salmonella enterica serovar Typhimurium persists across environments and causes disease by coordinating biofilm formation and host invasion. Although PagN is a known adhesin and invasin, its role in regulating these processes is unclear. We investigated PagN’s contribution to biofilm development and pathogenicity using a chromosomal pagN deletion mutant (ΔpagN) and a complemented strain. Deletion did not affect growth but significantly reduced biofilm formation on polystyrene at 20°C and 28°C. The mutant showed altered morphotypes, reduced cellulose, impaired pellicle formation, delayed autoaggregation, and restricted motility. In Caco-2 cells, pagN loss reduced adhesion by ∼60% and abolished invasion by >90%. Ectopic expression of pagN successfully abrogated phenotypic shifts, confirming gene specificity. Comparative transcriptomics revealed a niche-specific regulatory footprint; the profound hilA (SPI-1) repression in planktonic cultures was not recapitulated in biofilms. Instead, the mutant exhibited targeted csgD-yaiC attenuation during biofilm development, alongside robust fimF induction, indicating a potential compensatory shift in the adhesive landscape. Network analysis suggests PagN is integrated into global circuits, influencing pathways through the regulator of capsule synthesis (Rcs) system. These findings demonstrate PagN is a key determinant linking biofilm development with virulence regulation in Salmonella Typhimurium, coordinating environmental persistence and host–pathogen interaction. Introduction Salmonella is a leading cause of bacterial gastrointestinal disease in humans and animals and is primarily transmitted through contaminated food, water, or direct contact with infected hosts. Although mammals, birds, reptiles, and plants can serve as hosts, Salmonella also persists in diverse abiotic environments, largely due to its ability to form biofilms. Biofilms are structured microbial communities irreversibly attached to biotic or abiotic surfaces and embedded in a self-produced extracellular polymeric substance (EPS) matrix (Costerton et al. 1999, Donlan 2002, Stewart and Franklin 2008, Hall-Stoodley and Stoodley 2009, Flemming and Wingender 2010, Akçelik and Akçelik 2022). Compared to planktonic cells, bacteria within biofilms exhibit increased tolerance to environmental stresses, including ultraviolet radiation, temperature extremes, osmotic and pH fluctuations, nutrient limitation, and antimicrobial exposure (Yin et al. 2019). Consequently, biofilm formation plays a central role in Salmonella persistence, industrial contamination, and infection. Biofilm development is initiated by surface attachment, a process that also underlies host cell colonization. The initial contact between Salmonella and host epithelial cells is a critical step that precedes protein translocation, cellular entry, and systemic dissemination (Wagner and Hensel 2011). Adhesins are therefore key virulence determinants that facilitate both environmental persistence and host colonization (Hansmeier et al. 2017). Salmonella has evolved a facultative intracellular lifestyle, enabling it to breach host barriers and invade both phagocytic and non-phagocytic cells (Fields et al. 1986, Finlay and Falkow 1989, Ibarra and Steele-Mortimer 2009). Following internalization, the bacterium resides within a modified phagosome known as the Salmonella-containing vacuole (SCV), where it survives and replicates while evading host immune responses (Cossart and Sansonetti 2004, Bernal-Bayard and Ramos-Morales 2018). Efficient host cell entry is therefore a pivotal event for intracellular survival, systemic spread, and transmission (Boumart et al. 2014). The pagN gene encodes PagN, a 25-kDa outer membrane protein that functions as an adhesin and invasin, mediating attachment to and invasion of non-phagocytic epithelial cells (Lambert and Smith 2008). PagN is highly conserved across the Salmonella genus and is encoded within SPI-6 (formerly the Salmonella enterica centisome seven genomic island), which also harbors other adhesion-associated genes, including the safABCD fimbrial cluster and the transcriptional regulator sinR (Mulder et al. 2012, Ray et al. 2022). Unlike many invasion-related loci, pagN is transcribed as a monocistronic unit. PagN expression is regulated by the PhoP/PhoQ two-component system and is induced under restrictive environmental conditions such as acidic pH, magnesium or calcium limitation, and exposure to antimicrobial peptides. PagN mediates host cell entry through a zipper-like mechanism involving interactions with β1 integrin and syndecan-1 receptors, leading to actin cytoskeleton rearrangement and bacterial internalization (Lambert and Smith 2008, Barilleau et al. 2021). Although PagN-dependent invasion requires class I phosphoinositide 3-kinases and tyrosine phosphorylation, its precise intracellular signaling and post-entry roles remain incompletely understood (Barilleau et al. 2021, Ménard et al. 2022). While PagN has been extensively studied in the context of host cell invasion, its contribution to environmental persistence—particularly biofilm formation—and associated regulatory networks has not been systematically investigated. In previous work from our group, transcriptomic analysis revealed that pagN expression was significantly upregulated in a seqA mutant, a global regulator implicated in biofilm regulation (Has et al. 2023). These findings prompted the hypothesis that PagN may play a broader role in Salmonella physiology beyond host cell invasion. In this study, we investigated the role of pagN in biofilm formation, motility, and virulence-associated gene regulation, aiming to define its contribution to Salmonella pathogenicity and environmental persistence. Materials and methods Bacterial strains, chromosomal gene deletion, and complementation Salmonella enterica serovar Typhimurium ATCC 14028 was used as the wild-type (WT) strain in this study. Its isogenic derivatives included a pagN deletion mutant (ΔpagN) and a genetically complemented strain (ΔpagN:: pagN⁺), both constructed for this work. All bacterial strains and plasmids used are listed in Table 1. Table 1Bacterial strains and plasmids used in this study.Strains and plasmidsDescriptionSource/Reference Strains code WT S. enterica serovar Typhimurium ATCC 14028ATCCΔpagNDeletion of pagN gene; cat cassette insertion (CamR)This study pagN::pagN + Complementation of pagN gene; harbouring pBAD24-pagN (CamR, AmpR)This study Plasmids pKD3FRT-flanked cat cassette expressing vector, AmpR, CamR, 2804 bpDatsenko and Wanner (2000)pKD46Red recombinase vector, AmpR, 6329 bpDatsenko and Wanner (2000)pBAD24Bacterial expression vector, AmpR, 4542 bpGuzman et al. (1995) Bacterial strains and plasmids used in this study. Chromosomal deletion of the pagN gene was performed using the λ Red recombination system as described by Datsenko and Wanner (2000). Briefly, the temperature-sensitive Red recombinase plasmid pKD46 was introduced into S. Typhimurium ATCC 14028 by electroporation, and transformants were selected on LB agar supplemented with ampicillin (100 μg ml⁻¹). The chloramphenicol resistance cassette was amplified from plasmid pKD3 using pagN-red primers containing homologous flanking regions of the pagN locus. Following PCR amplification, the template DNA was removed by DpnI digestion. Expression of the λ Red recombinase genes was induced with L-arabinose (10 mM) prior to electroporation of the linear PCR product. Recombinants were selected on LB agar containing chloramphenicol (25 μg ml⁻¹), and successful replacement of the pagN gene was confirmed by PCR using cat-ver primers (Table 2). Table 2Primers used for mutation, complementation, and qPCR analysis.Primers5’-to 3’ sequence pagN-red F: TCTGAAATGAAAAACTTTTTCGCAGTCTGCATCATTCCCCTTGTGGTAGCGTGTAGGCTGGAGCTGCTTC R: TTAAAAGGCGTAAGTAATGCCGAGCATGAAGTCATTGGAGGCAGCCTTTGATGGGAATTAGCCATGGTCCcat-ver F: TCGGCATCTCATCATTATGG R: CAATCGCAGGGGTTCAGGAA pagN-comp F: CCGCGAATTCATGAAAAACTTTTTCGC* R: ACTGCTGCAGTTAAAAGGCGTAAG**comp-ver F: TCTGAAATGAAAAACTTTTTCGAG R: TTAAAAGGCGTAAGTAATGCCGAGC yaiC F: GAGCATGAATCTGATGGGCG R: ACTTACCCAGGCGAACAGAA csgD F: ACGCTACTGAAGACCAGGAAC R: GCATTCGCCACGCAGAATA hilA F: CATGGCTGGTCAGTTGGAG R: CGTAATTCATCGCCTAAACG ssaB F: CTGTGGCTTTCAGTGGTCAG R: TGCGTTGTCCGGTAGTATTTC invA F: CACGCTCTTTCGTCTGGCA R: TACGGTTCCTTTGACGGTGCGA gryA F: CCGTCAATTCCTTTGAGTTT R: AGAGTTTGATCCTGGCTCAG fliF F: GAAGCATTCTGTTCGCCTAT R: TGTAGTGCTCTTTCCGTCTGC rpoS F: GTTGGACGCGACTCAGCTTT R: TTACCACCAGACGCAGGTTGAT fimF F: CCTAACACGTTACCGCTTCA R: CCGGCGTCATATTTTCTGACleavage site is underlined; * EcoRI, **PstI. Primers used for mutation, complementation, and qPCR analysis. Cleavage site is underlined; * EcoRI, **PstI. For genetic complementation, the pagN coding sequence (∼726 bp) was amplified by PCR using primers incorporating EcoRI and PstI restriction sites (Table 2). The purified PCR product and the arabinose-inducible expression vector pBAD24 were digested with EcoRI and PstI and ligated at an insert-to-vector molar ratio of ∼3:1 using T4 DNA ligase at 20°C for 2 h. The ligation mixture was introduced into electrocompetent ΔpagN cells by electroporation. Transformants were selected on LB agar supplemented with ampicillin (100 μg ml⁻¹), and plasmid insertion was confirmed by plasmid isolation and PCR analysis. The ΔpagN mutant was routinely cultured in media supplemented with 25 μg ml⁻¹ chloramphenicol, and for all phenotypic assays, pagN expression in the ΔpagN:: pagN⁺ strain was induced by the addition of 0.01% (w/v) L-arabinose to the culture medium. Growth kinetics A single bacterial colony was first inoculated into 5 ml of LB broth, grown for 18 h, and then diluted 1/100 into 50 ml of fresh LB broth, shaking at 37°C. Hourly optical density (OD600) measurements were recorded. After serial dilution in PBS, aliquots of cultures were inoculated onto Plate Count Agar (PCA) in 100 μl quantities. After 12 h of incubation at 37°C, the colonies on petri plates were counted. This experiment has three technical duplicates. The following formula was used for computations. Evaluation of biofilm formation capacity on polystyrene surfaces Biofilm formation experiments were quantitatively performed in 96-well polystyrene microtiter plates in salt-free LB broth. The optical density (OD595) of the active cultures prepared as indicated was adjusted to 0.2 by diluting with the medium. Thirty microliters of these cultures were transferred to wells containing 100 µl of salt-free LB. Each experimental pattern was set to have three technical replicates. Then, they were incubated under static conditions for 24, 48, 72, 96, and 120 h at 20°C and 28°C. At the end of each incubation period, the medium was removed, and the wells were washed three times with 140 µl of sterile PBS buffer to remove non-adherent cells. One hundred forty microliters of 95% methanol solution was added to the plate wells that stayed at room temperature, and the fixation process was performed by waiting for 15 min. After removing methanol from the medium, the fixed biofilm structure in the completely dried wells was left to be stained for 30 min at room temperature after adding crystal violet (0.1%). In order to remove the dye that did not bind to the matrix, two distilled water washes were done. To dissolve the dye adhering to the cells, 130 µl of glacial acetic acid (33%) was transferred to the wells, and the plates were left at room temperature for ∼45 min. Following the waiting period, the optical density of the dye adhering to the biofilm was determined by measuring it with an ELISA reader (Thermo Multiskan GO) (OD595). The experiments were designed to have two experimental repetitions. The calculation was performed by subtracting the average of the optical density values determined for each experimental group from the negative control optical density values (Stepanovic et al. 2000). Determination of biofilm structural characteristics To determine the biofilm morphotypes of S. Typhimurium 14028 and its mutants, Congo red (40 μg ml⁻¹)-containing medium was used. Twenty microliters of overnight cultures grown in salt-free LB broth were taken and spot-cultured onto Congo red solid medium (Ahmad et al. 2016). The colonies formed on the agar were examined and evaluated in color and structure. The structural properties of biofilm were evaluated in liquid and solid media containing calcofluor white. The active culture was cultured in the medium containing calcofluor dye (40 μg ml⁻¹). For experiments in liquid media, 500 µl of active culture was inoculated into the salt-free LB medium containing calcofluor dye prepared as 4.5 ml. At the end of incubation, they were examined under 366 nm UV light (Biometra TI 2, USA) (Ahmad et al. 2016). To investigate the pellicle formation properties of wild-type and its mutant strains in the liquid-air interface, LB broth without salt was prepared in 4.5 ml volumes. Five hundred microliters of active cultures were added to these tubes. At the end of the incubation, the bacteria were evaluated and compared in terms of pellicle formation and durability. Then, the tubes were emptied, washed twice with distilled water, dried at room temperature, and stained with a crystal violet (0.1%) solution for 10 min. In all experiments, culture media were incubated at 20°C under static conditions for 8 days. Also, all experiments were conducted for each strain with at least three technical replicates and two experimental replicates. Determination of bacterial aggregation For the aggregation experiment, S. Typhimurium wild-type and mutant strains were grown in LB broth and centrifuged in the stationary phase at 10 000 Xg for 10 min. After removing the supernatant, the pellets were washed twice with PBS. The optical density (OD600) of the cultures was adjusted to 0.2 and incubated at 20°C under static conditions. The experimental design was followed for each strain with three technical replicates and two experimental replicates. The optical density value was measured and recorded at 600 nm at 0, 2, 4, 20, and 24 h. Percent aggregation coefficients were determined with the following formula (Merino et al. 2019). OD0: Value measured before incubation. OD0: Value measured before incubation. ODt: Value measured at each incubation period. ODt: Value measured at each incubation period. Monitoring of swimming and swarming motility Swimming, which is an individual movement supported by flagella rotation, and swarming, which is a multicellular surface movement supported by flagella, were examined. A semi-solid medium containing 0.3% agar was used for swimming movement. A medium containing 0.5% agar was used for swarming, and 0.5% glucose was used for cluster chemotaxis movement. Single colonies from overnight culture medium cultivated by the streaking method were placed in the middle of the medium (Ahmad et al. 2016). Plates were incubated at three distinct temperatures for 12 h: 20°C, 28°C, and 37°C. The diameters of the motility zones were measured along two perpendicular axes for each plate in order to assure accuracy and account for potential non-circular expansion. The mean value was recorded. Parallel cultivations were implemented for each experimental group, and the experiment was conducted twice. Caco-2 cell line adhesion and invasion experiments The human colon cell line Caco-2 was utilized to assess and compare the adhesion and invasion capacities of wild-type and mutant strains against human epithelial cells. The Caco-2 cells were cultured in Dulbecco’s Modified Eagle Medium (DMEM) with 10% fetal bovine serum (FBS) and antibiotics and incubated at 37°C in a 5% CO₂ humidified atmosphere. Adhesion and invasion assays were conducted in 24-well plates with medium changes every two days, following the protocol established by Gagnon et al. (2013) and performed in triplicate. Salmonella strains were grown in LB broth at 37°C and collected at early stationary phase for adhesion and invasion assays. After centrifugation of active cultures at 8000 Xg for 5 min, they were rinsed with sterile PBS and suspended in DMEM medium at 2 × 107 CFU/ml (MOI: 1/100) for an adhesion assay. Following the washing of intestinal epithelial cells, bacterial suspensions were added and cultured for 90 min at 37°C with 5% CO₂. Post-incubation, cultures were washed with PBS to remove non-adherent bacteria. Adherent bacterial cells were treated with trypsin-EDTA, and the reaction was halted with FBS-containing media. Colonies were counted on SS agar and incubated overnight. Caco-2 cells were treated with a bacterial suspension, rinsed with PBS, and incubated for 90 min at 37°C. After two PBS washes, DMEM with gentamicin was added to kill extracellular cells, followed by a 60-min incubation. The cells were then washed again to remove noninvasive bacteria and detached using trypsin-EDTA. Triton X-100 was used to lyse the cells, followed by a 10-min incubation to count invasive bacteria, with subsequent serial dilution and cultivation for counting the next day. qPCR analysis Transcriptional profiling was performed under two distinct growth regimes to differentiate between planktonic and biofilm-associated expression patterns. Planktonic cultures were grown in LB broth at 37°C with shaking until mid-exponential phase (OD600 ≈ 0.3). To evaluate the regulatory role of the PagN adhesin during the early stages of biofilm development, cells were harvested from static cultures in salt-free LB broth after 24 h of incubation at 20°C. The GeneMATRIX Universal RNA Purification (EURx, Poland) kit was used for total RNA isolation, and the steps specified by the manufacturer were performed. cDNA synthesis from RNA samples was performed using the smART first-strand cDNA synthesis (EURx, Poland) commercial kit. Quantitative real-time PCR was performed with Hot FIREPol EvaGreen qPCR Supermix (5×, Solis BioDyne, Estonia) on a Roche LightCycler 480 II platform. Relative gene expression levels were determined via the 2−ΔΔCT technique (Livak and Schmittgen 2001), normalized against the housekeeping gene gyrA. In silico protein–protein interaction (PPI) and network topology analysis To elucidate the regulatory hierarchy and functional associations of PagN, a protein–protein interaction (PPI) network was constructed using the STRING database (v12.0) (Szklarczyk et al. 2023). Salmonella enterica serovar Typhimurium LT2 (Taxonomy ID: 99287) was utilized as the reference genome due to its high annotation depth for the proteins under investigation (Shannon et al. 2003). A query list of 21 proteins—specifically PagN, SPI-1 regulators (HilD, HilC, HilA, InvA), biofilm and motility components (CsgD, YaiC, FliF, FlhD, FlhC, FliZ, FimF, FimH, FimA), and global signal transduction factors (PhoP, PhoQ, RpoS, RcsB, RcsC, RcsD, RstA)—was analyzed based on multiple evidence channels with a minimum interaction score of “High Confidence” (0.700) (Assenov et al. 2008). The resulting network, comprising 20 nodes and 33 edges, was exported to Cytoscape (v3.10.1) for visualization and advanced topological evaluation. Network topological parameters, including Node Degree (k), Betweenness Centrality (CB), and the Clustering Coefficient (CC), were calculated using the NetworkAnalyzer plugin to determine the structural and functional importance of each node within the regulatory landscape. To categorize functional modules, the Markov Cluster (MCL) algorithm was applied with an inflation parameter of 3 to identify distinct pathways associated with virulence, motility, and signal transduction (Enright et al. 2002). Statistical analysis of data All statistical analyses and graphical representations were performed using GraphPad Prism software (version 10.0, GraphPad Software, Boston, MA, USA). Data are presented as the mean standard deviation (SD) of at least three independent biological replicates. Bacterial growth kinetics were analyzed using regression analysis. The transcriptional expression statistical significance was determined using multiple unpaired t-tests followed by the Holm–Sidak method to correct for multiple comparisons, with statistical significance. For biofilm formation and autoaggregation assays, data were analyzed using two-way ANOVA followed by Tukey’s multiple comparisons test to determine significant differences at each time point. For adhesion and invasion assays, differences between strains were assessed using one-way ANOVA with Dunnett’s post-hoc test. Results Verification of pagN deletion and its complementation Successful deletion of the pagN gene in S. Typhimurium was confirmed by PCR analysis. Amplification with primers specific for the chloramphenicol resistance cassette yielded the expected ∼384 bp fragment exclusively in the ΔpagN mutant. For genetic complementation, the pagN gene was cloned into the pBAD24 plasmid. PCR using pagN-specific primers produced the expected ∼726 bp amplicon only in the complemented strain (ΔpagN:: pagN⁺), verifying successful restoration of the gene. Effect of pagN deletion on bacterial growth The growth characteristics of WT, ΔpagN, and ΔpagN:: pagN⁺ strains were assessed in LB broth over a 12-h period by measuring OD600 and viable cell counts (log CFU/ml). All strains exhibited comparable growth kinetics and reached similar final cell densities (OD600 ≈ 2.1; log CFU/ml ≈ 6.5–6.6) at 12 h (Fig. 1). In vitro growth analysis of WT, ΔpagN, and ΔpagN:: pagN⁺ strains. (A) Growth monitored by OD600. (B) Viable counts expressed as log CFU/ml. Linear regression analysis of the exponential growth phase revealed growth rates of 2.35 h⁻¹ for WT (R² = 0.93, p < 0.0005), 2.21 h⁻¹ for ΔpagN (R² = 0.93, p < 0.0004), and 2.22 h⁻¹ for the complemented strain (R² = 0.92, p < 0.0007). Although the ΔpagN mutant displayed a slight delay during early exponential growth, this difference was transient and did not affect final biomass accumulation. These results indicate that pagN deletion does not significantly impact bacterial growth or viability under the conditions tested. pagN is required for efficient biofilm formation at different temperatures Biofilm formation on polystyrene surfaces was quantified over 120 h at 20°C and 28°C using the crystal violet assay. At 20°C, the ΔpagN mutant exhibited a pronounced defect in biofilm formation at all time points compared to the WT strain (p < 0.0001). In contrast, WT biofilm biomass increased progressively, reaching a maximum at 72 h. Complementation of pagN fully restored biofilm formation to WT levels, with no significant differences observed (p > 0.05). At 28°C, overall biofilm production was enhanced for all strains; however, the ΔpagN mutant remained significantly impaired relative to WT (p < 0.0001). Restoration of pagN expression rescued the biofilm-deficient phenotype, confirming that the observed defect was directly attributable to the absence of PagN (Fig. 2). These findings demonstrate that PagN plays a critical role in biofilm development across environmentally relevant temperatures. Biofilm formation by WT, ΔpagN, and ΔpagN::pagN⁺ strains at (A) 20°C and (B) 28°C. Statistical significance is indicated as follows: **** p < 0.0001, * p < 0.05, and ns: non-significant (p > 0.05). pagN influences biofilm structural components Biofilm morphotypes were evaluated on Congo red (CR) agar to assess curli and cellulose production. The WT strain displayed the characteristic red, dry, and rough (rdar) morphotype, indicative of co-expression of curli fimbriae and cellulose. In contrast, the ΔpagN mutant exhibited an intermediate morphology resembling the pink, dry, and rough (pdar) phenotype, suggesting impaired curli expression and reduced cellulose production (Fig. 3A). Analysis of biofilm structural elements in WT, ΔpagN, and ΔpagN::pagN⁺ strains. (A) Congo red morphotypes. (B–C) Cellulose detection using calcofluor white on agar and in broth. (D–E) Pellicle formation and crystal violet staining. Consistent with these observations, calcofluor white binding assays revealed markedly reduced fluorescence in the ΔpagN mutant on both solid and liquid media, indicating diminished cellulose synthesis (Fig. 3B and C). Pellicles formed by the mutant were thinner, fragile, and readily disrupted upon agitation, whereas WT pellicles were robust and cohesive (Fig. 3D and E). Together, these results indicate that pagN contributes to the proper assembly and stability of biofilm matrix components. Contribution of PagN to autoaggregation kinetics To investigate the role of PagN in cell-cell interactions, the autoaggregation kinetics of wild-type, pagN mutant, and complemented strains were analyzed over 24 h. The wild-type strain exhibited rapid aggregation, reaching ∼25% within 4 h and peaking at ∼37% by 24 h. In contrast, the pagN deletion mutant displayed a pronounced delay in aggregation. At early time points (2 h and 4 h), its aggregation was significantly lower than that of the wild-type. Although aggregation gradually increased in the mutant, reaching ∼30% at 24 h, it remained consistently below wild-type levels throughout the assay. Complementation of the mutant with a functional pagN gene fully restored the autoaggregation phenotype (Fig. 4). The aggregation kinetics of the complemented strain were essentially identical to those of the wild type, confirming that the defect observed in the mutant was specifically attributable to the loss of PagN. Data points are color-coded by strain for clarity. Statistical significance between WT and ΔpagN was determined by One-way ANOVA (p < 0.05). Autoaggregation kinetics of WT, ΔpagN, and ΔpagN:: pagN⁺ strains over 24 h. Effect of pagN deletion on swimming and swarming motility Motility assays demonstrated that deletion of pagN significantly affected swimming behavior. Compared to WT, swimming motility of the ΔpagN mutant was reduced by 72% at 37°C and 83.3% at 28°C, while no detectable swimming was observed at 20°C. Swarming motility was less affected, showing a minor reduction (1.5%) at 37°C, a modest increase (15%) at 28°C, and a substantial decrease (60%) at 20°C (Fig. 5). These results suggest that PagN contributes primarily to swimming motility in a temperature-dependent manner. Swimming and swarming motility of the strains at different temperatures.(A) Relative percentage of motility diameters normalized to the WT strain and (B) representative images of the motility assays. PagN-mediated adhesion to and invasion of epithelial cells The role of PagN in host-cell interactions was assessed using Caco-2 epithelial cells. Deletion of pagN resulted in a significant reduction in bacterial adhesion, with the ΔpagN mutant associating with ∼4.8% of the initial inoculum compared to ∼11.5% for the WT strain (p < 0.0001). Complementation restored adhesion to WT levels (Fig. 6A). Role of PagN in adhesion to and invasion of Caco-2 cells. (A) Cell association. (B) Intracellular survival following gentamicin protection assay. Statistical significance is indicated as **** p < 0.0001. Invasion assays revealed an even more pronounced defect. While WT bacteria were internalized at a frequency of ∼0.82%, the ΔpagN mutant exhibited negligible invasion (<0.05%), corresponding to a >15-fold reduction (p < 0.0001). Complementation fully restored invasion efficiency (Fig. 6B), demonstrating that PagN is essential for efficient epithelial cell invasion. Transcriptional profiling of PagN-dependent pathways To investigate the transcriptional underpinnings of the PagN-associated phenotypes, gene expression patterns were analyzed across two distinct growth environments. In planktonic culture (Fig. 7A), the ΔpagN mutant exhibited a severe repression of the SPI-1 master regulator hilA, which mirrored the significant reduction in epithelial invasion observed in our assays. This was accompanied by a corresponding decrease in the transcript levels of the downstream effectors invA and ssaB. Furthermore, the observed motility restrictions during this phase coincided with an approximate three-fold reduction in the expression of the flagellar gene fliF. Transcriptional profiling of the ΔpagN mutant in distinct growth environments. (A) Relative expression levels in planktonic cultures (37°C, shaking) and (B) under biofilm-inducing conditions (20°C, static, 24 h). Asterisks denote statistical significance levels: * p < 0.05, ** p < 0.01, *** p < 0.001 **** p < 0.0001 ns, not significant (p > 0.05) determined by multiple unpaired t-tests with Holm–Sidak correction. Conversely, when evaluated under biofilm-inducing conditions (Fig. 7B), the transcriptional profile underwent a notable shift. While the expression of SPI-1 regulators (hilA, invA, and ssaB) remained comparable to wild-type levels, the mutant displayed a significant attenuation of the csgD-yaiC axis. Specifically, the expression of the master biofilm regulator csgD and the diguanylate cyclase yaiC decreased by ∼1.6-fold and 2-fold, respectively, aligning with the matrix-deficient (pdar) morphotype and reduced cellulose production identified in the mutant. Additionally, a distinct downregulation of fliF was observed, consistent with the impaired motility under these conditions. Notably, fimF expression was robustly induced during this early biofilm stage, representing a stark contrast to its profile during planktonic growth. Consistent across both regimes, rpoS expression was not significantly altered, reinforcing the specificity of the PagN-dependent regulatory changes. Relative expression levels in the ΔpagN mutant are presented as fold-change compared to the WT baseline (normalized to 1.0). Protein–protein interaction network and topological analysis of PagN To provide a mechanistic basis for the pleiotropic effects observed in the ΔpagN mutant, a protein–protein interaction (PPI) network was constructed. The resulting network consisted of 20 nodes and 33 edges, with an average clustering coefficient of 0.321 and a network centralization of 0.275 (Fig. 8). Topological analysis identified PagN as a strategic component within the regulatory landscape, showing a direct high-confidence interaction with the master regulator PhoP (combined score: 0.720). The pagN-phoP interaction exhibited an edge betweenness of 38.0, ranking it among the top functional bridges in the network (Table S1). PhoP emerged as a primary hub with a node degree of 8, serving as a critical link between PagN and downstream virulence factors, including HilD and InvA (Table S2). Furthermore, rcsB displayed a remarkably high clustering coefficient of 0.6, indicating its participation in a highly coordinated functional module. This module is central to the envelope stress response, which likely perceives the loss of PagN as a structural defect. The highest Edge Betweenness (71.0) was recorded for the csgD-flhD interaction, which represents the essential axis between biofilm matrix production and flagellar biogenesis. The integration of these topological findings with transcriptional data suggests that PagN acts as a regulatory bottleneck. Its absence disrupts the signal flow through the PhoP/PhoQ and Rcs systems, leading to the dramatic downregulation of the SPI-1 master regulator hilA and the subsequent failure to establish robust biofilm structures. Protein–protein interaction network of PagN in S. Typhimurium. Three functional clusters are identified, representing signal transduction (PhoP/Q, RcsCDB, RpoS), virulence regulation (HilC/D, InvA), and the biofilm-motility axis (CsgD, YaiC, FlhDC, FimAFH). Nodes are grouped into five distinct functional modules: phosphorelay, invasion switch, flagellar assembly, fimbrial attachment, and T3SS structural components. Solid edges denote high-confidence physical interactions derived from experimental or curated data, and dashed edges signify predicted functional associations based on genomic context and co-occurrence Discussion Salmonella enterica is a facultative intracellular pathogen capable of invading epithelial cells via both the trigger mechanism, which induces cytoskeletal rearrangements, and the receptor-mediated zipper mechanism. Following internalization, the bacterium resides within the Salmonella-containing vacuole (SCV), enabling evasion of host innate immune responses (Haraga et al. 2008). Salmonella subsequently exploits macrophages to disseminate to the mesenteric lymph nodes, bloodstream, and systemic organs, including the spleen and liver (Monack et al. 1996, Galán 2021 , Koczerka et al. 2024). Beyond acute infection, Salmonella can persist asymptomatically in organs such as the gallbladder, contributing to long-term carriage and transmission (Crawford et al. 2010). Among the bacterial determinants facilitating epithelial attachment and invasion, PagN has emerged as a critical adhesin/invasin, potentially playing a more pronounced role than other invasins in establishing systemic infection (Koczerka et al. 2024). Previous work from our group demonstrated that a seqA mutant upregulates pagN expression during biofilm development (Uğur et al. 2018, Has et al. 2023), prompting the hypothesis that PagN contributes to biofilm formation in Salmonella. The present study provides comprehensive evidence linking PagN to multiple virulence-associated phenotypes, including biofilm formation, autoaggregation, motility, and epithelial cell invasion. Growth assays (OD600 and Log CFU) revealed that the ΔpagN mutant maintains wild-type-like metabolic fitness, indicating that the observed phenotypic defects are not due to general growth impairment but reflect specific functional consequences of pagN deletion. Biofilm formation is a multi-step process involving initial attachment, microcolony formation, and matrix maturation. The ΔpagN mutant exhibited a severe reduction in biofilm biomass at both 20°C and 28°C, failing to establish significant biofilm even after prolonged incubation (120 h). This phenotype suggests that PagN is critical for early adhesion events and/or the structural integrity of the biofilm matrix. The restoration of wild-type biofilm formation in the complemented strain confirms that this defect is directly attributable to pagN loss rather than secondary mutations. Analysis of biofilm matrix components revealed that PagN quantitatively influences cellulose and curli fimbriae production, key determinants of biofilm architecture. While decreased matrix components contribute to the impaired biofilm, they do not fully explain the defect. Given that PagN functions as an adhesin/invasin, its direct role in surface attachment and intercellular interactions likely underpins the observed biofilm phenotype. Autoaggregation facilitates microcolony formation, a prerequisite for robust biofilm development (Sauer et al. 2022, Kragh et al. 2023). Deletion of pagN significantly delayed autoaggregation kinetics. This contrasts with previous heterologous expression studies in E. coli, which suggested that PagN does not act as an autoagglutinin (Lambert and Smith 2008). Our results in the native S. Typhimurium background indicate that PagN cooperates with Salmonella-specific surface structures to mediate cell-cell interactions. In Caco-2 epithelial cells, the ΔpagN mutant displayed a moderate adhesion defect (∼60% of WT) but a dramatic loss of invasion (>90% reduction). This disproportionate effect suggests that PagN functions primarily as an invasin, facilitating the trigger mechanism of internalization rather than serving solely as an adhesin. Complementation restored both adhesion and invasion, confirming the specificity of the phenotype. These observations are consistent with previous reports showing that PagN expression alone can confer invasive capacity to noninvasive bacteria (Barilleau et al. 2021). While other adhesins, such as fimbriae, may support initial attachment, PagN is essential for the internalization step, likely acting as the trigger for host cell uptake. PagN production is regulated by the PhoP/PhoQ system, a two-component regulatory pathway sensing environmental cues such as acidic pH and low Mg²⁺ and controlling virulence and surface modifications (Groisman et al. 2021). Our data suggest that PhoP/PhoQ-mediated expression of PagN acts as a critical signal for initiating biofilm formation, linking environmental sensing to both extracellular survival and pathogenicity. Although other adhesins exist in Salmonella, the unique contribution of PagN to pathogenicity and biofilm regulation underscores its central role. Motility assays further support this conclusion: the ΔpagN mutant displayed reduced swimming and altered swarming, which likely restrict surface exploration and biofilm maturation. These physiological changes may compound the effects of matrix downregulation, further impairing biofilm establishment. Transcriptional analysis of the ΔpagN mutant revealed that PagN deficiency triggers a specialized regulatory response contingent upon the growth phase and environmental niche. In planktonic cultures, the dramatic repression of the SPI-1 master regulator hilA serves as the primary molecular basis for the %>90 functional loss in epithelial cell invasion. This transcriptional silencing directly restricts the expression of downstream SPI-1 effectors, such as invA and ssaB, thereby neutralizing the pathogen’s internalization mechanism. The role of PagN as both a structural adhesin and a modulator of virulence gene expression aligns with established models where HilA integrates environmental cues to govern the delivery of T3SS-1 effectors (Wagner and Hensel 2011, Galán 2021). During adaptation to biofilm-inducing conditions, the transcriptional focus shifts from SPI-1 toward biofilm matrix components. While SPI-1 regulators remained at wild-type levels, the attenuation of the biofilm master regulator csgD and the diguanylate cyclase yaiC—which is essential for cellulose biosynthesis—correlates with the identified cellulose deficiency and the pdar morphotype. This weakening of the csgD-yaiC axis results in the formation of fragile and unstable pellicles. Notably, the robust induction of fimF expression during the biofilm stage suggests a compensatory adhesion shift, where the bacterium pivots toward fimbriae-based attachment to mitigate the loss of matrix components. This interpretation is supported by evidence establishing matrix components as primary determinants for biofilm architecture (Zogaj et al. 2001). The stability of rpoS expression across both growth regimes confirms that these pleiotropic effects stem from specific regulatory pathways coordinated by PagN rather than generalized metabolic stress This observation is further supported by previous findings demonstrating that the overexpression of rpoS is insufficient to restore the expression of key biofilm regulators like mlrA or compensate for cellulose production defects, indicating that these specific circuits can be governed by niche-specific factors independent of the general stress response (García et al. 2004). Furthermore, these results suggest that PagN deficiency may be perceived as a structural defect in the cell envelope, acting as a signal that necessitates a global transcriptional shift. Literature indicates that signals of envelope stress, such as those induced by outer membrane perturbations or tolB mutations, trigger global regulatory responses to optimize bacterial survival and persistence (Mouslim et al. 2003, Mouslim and Groisman 2003). As a result, the lack of PagN might be seen as a signal that requires a specific change in gene activity. To evaluate the systemic integration of PagN into these global circuits and its potential role as a regulatory bottleneck, we proceeded with a protein–protein interaction (PPI) and network topology analysis. The profound downregulation of hilA and the concomitant loss of invasiveness in the mutant suggest that PagN plays a critical role in maintaining the stability of the SPI-1 regulatory network, potentially by preserving cell envelope integrity. Our PPI network analysis provides a mechanistic framework for this observation, identifying PagN as a strategic bottleneck for signal flow through its high-confidence interaction (0.720) and significant edge betweenness (38.0) with the PhoP hub (k=8). While these PPI network analyses provide a robust predictive framework for regulatory hubs, these in silico interactions represent models that require future experimental validation to confirm direct physical associations. The expression of hilA, which encodes the master regulator of the T3SS-1, is tightly controlled by a feed-forward loop involving the AraC-like regulators HilD, HilC, and RtsA (Ellermeier et al. 2005). Previous studies have established that perturbations in the outer membrane can trigger the RcsCDB phosphorelay system, a sensor of envelope stress that negatively regulates virulence (Arricau et al. 1998). The remarkably high clustering coefficient of rcsB (0.6) in our network identifies it as a central integrator of this stress response. Activation of the Rcs system leads to the phosphorylation of RcsB, which represses hilA transcription directly and indirectly by inhibiting the flhDC flagellar operon, thereby reducing the levels of FliZ, a positive regulator of HilD (Lin et al. 2008). Our data imply that the absence of PagN may be perceived as an envelope stress signal, triggering this Rcs-mediated cascade to silence the metabolically costly SPI-1 apparatus. Furthermore, the pleiotropic defects observed in biofilm formation and motility in the mutant can be mechanistically linked to this global stress response. The significant downregulation of yaiC and the shift to a cellulose-deficient (pdar) morphotype are consistent with the repression of CsgD, the master biofilm regulator. In Salmonella, activated RcsB suppresses csgD expression, subsequently reducing the transcription of yaiC (interaction score: 0.911), a diguanylate cyclase required for the synthesis of cyclic di-GMP and the allosteric activation of cellulose synthase (Zogaj et al. 2001, Latasa et al. 2012). Concurrently, the repression of fliF and the loss of swimming motility observed in our study align with the known role of the Rcs system in repressing flagellar biogenesis to prevent further membrane damage (Arricau et al. 1998). The highest edge betweenness in our network (71.0) was recorded for the csgD-flhD interaction, highlighting the disruption of the central axis between biofilm formation and motility. Collectively, these findings suggest that PagN functions not only as an adhesin but also as a key structural component whose loss shifts the bacterium into a defensive state, sacrificing virulence and biofilm formation for survival. The observed upregulation of pagN in the seqA mutant suggests a potential regulatory connection. SeqA is known to act as a transcriptional silencer of fimbrial operons (Jakomin et al. 2008), and our findings imply that it may similarly influence pagN expression, either directly or indirectly, contributing to coordinated regulation of adhesion, invasion, and biofilm formation. Collectively, our findings demonstrate that PagN is a multifunctional outer membrane protein that integrates environmental sensing, biofilm formation, autoaggregation, motility, and host cell invasion. By modulating SPI-1 expression, biofilm matrix components, and surface organelles, PagN serves as a central determinant of Salmonella virulence and environmental persistence. These results highlight PagN as a potential target for therapeutic intervention aimed at limiting both infection and environmental survival of Salmonella enterica. Supplementary Material