http://orcid.org/0000-0001-6400-9520
Figures
Abstract
Medical devices, such as contact lenses, bring bacteria in direct contact with human cells. Consequences of these host-pathogen interactions include the alteration of mammalian cell surface architecture and induction of cellular death that renders tissues more susceptible to infection. Gram-negative bacteria known to induce cellular blebbing by mammalian cells, Pseudomonas and Vibrio species, do so through a type III secretion system-dependent mechanism. This study demonstrates that a subset of bacteria from the Enterobacteriaceae bacterial family induce cellular death and membrane blebs in a variety of cell types via a type V secretion-system dependent mechanism. Here, we report that ShlA-family cytolysins from Proteus mirabilis and Serratia marcescens were required to induce membrane blebbling and cell death. Blebbing and cellular death were blocked by an antioxidant and RIP-1 and MLKL inhibitors, implicating necroptosis in the observed phenotypes. Additional genetic studies determined that an IgaA family stress-response protein, GumB, was necessary to induce blebs. Data supported a model where GumB and shlBA are in a regulatory circuit through the Rcs stress response phosphorelay system required for bleb formation and pathogenesis in an invertebrate model of infection and proliferation in a phagocytic cell line. This study introduces GumB as a regulator of S. marcescens host-pathogen interactions and demonstrates a common type V secretion system-dependent mechanism by which bacteria elicit surface morphological changes on mammalian cells. This type V secretion-system mechanism likely contributes bacterial damage to the corneal epithelial layer, and enables access to deeper parts of the tissue that are more susceptible to infection.
Author summary
Bacteria must overcome host defenses to cause infection. This is especially true for corneal infections where bacteria must penetrate the epithelium in order to gain access to the stroma where bacteria can rapidly multiply, induce inflammation, and cause vision loss. Members of the Enterobacteriaceae commonly cause contact lens associated infections, but the mechanisms by which they damage corneal cells are largely unknown. Here we present evidence that Serratia marcescens and Proteus mirabilis are able to induce dramatic morphological changes in mammalian corneal cells that correlates with rapid cellular death. Secretion of ShlA-like cytolysins via type V secretion was responsible for this phenotype, and this effect was regulated by the conserved Rcs phosphorelay stress response system, including IgaA-family protein GumB. This study provides a model for stress-mediated regulation of cytolysins that induce epithelial damage and promote ocular infection.
Citation: Brothers KM, Callaghan JD, Stella NA, Bachinsky JM, AlHigaylan M, Lehner KL, et al. (2019) Blowing epithelial cell bubbles with GumB: ShlA-family pore-forming toxins induce blebbing and rapid cellular death in corneal epithelial cells. PLoS Pathog 15(6): e1007825. https://doi.org/10.1371/journal.ppat.1007825
Editor: Eric Oswald, INSERM U1220, FRANCE
Received: November 12, 2018; Accepted: May 7, 2019; Published: June 20, 2019
Copyright: © 2019 Brothers et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: All relevant data are within the manuscript and its Supporting Information files.
Funding: This work was supported by National Institutes of Health (https://www.nih.gov) CORE Grant P30 EY08098, the Eye and Ear Foundation of Pittsburgh (https://eyeandear.org), an unrestricted grant from Research to Prevent Blindness (https://www.rpbusa.org/rpb/?) to the University of Pittsburgh Department of Ophthalmology. NIH grant EY027331 (R.M.Q.S). NIH grant T32EY01727 (K.M.B). This work was also supported, in part, by an Institutional Development Award (IDeA) from the National Institute of General Medical Sciences of the National Institutes of Health (P20GM103434) to JH. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Introduction
Some bacteria induce changes in human cell architecture through expression of virulence factors, which aid in bacterial internalization into cells and provide a more favorable niche for microbial colonization. Cell alterations include the formation of pedestals on intestinal cells by enteropathogenic and enterohaemorrhagic Escherichia coli strains through expression of type III secretion system (T3SS)-associated effector proteins that alter the actin cytoskeleton [1, 2]. Other bacteria create membrane ruffles, alterations that facilitate bacterial invasion into the mammalian cell [3]. Another type of mammalian cell surface alteration, known as bleb formation, appears following cellular injury [4, 5]. The bacterium Pseudomonas aeruginosa can induce bleb formation in airway and ocular cells. These blebs are similar to, but are more stable than, necrotic blebs and require a T3SS encoded by P. aeruginosa, and the ExoS and ExoY T3SS effector proteins [6, 7]. T3SS-dependent induction of membrane blebs on human cells was also reported for Vibrio parahemolyticus [8]. Large necrotic membrane blebs can also be induced by hydrogen peroxide and are thought to be a last-ditch effort by the cells to evade lysis. In this setting, they may serve as a storage unit to sequester damaged cellular components away from the cell or are simply a loss of cell homeostasis [5, 9, 10]. Blebs induced by membrane breach are hypothesized to be a result of the influx of calcium that activates hydrolytic enzymes capable of damaging the cellular cortex [11].
Gram-negative bacteria of the Enterobacteriaceae family, such as Proteus mirabilis and Serratia marcescens, cause nosocomial infections in neonates and immune compromised patients and contact lens associated complications in healthy individuals, including keratitis [12–17]. Microbial keratitis, or infection of the cornea, is a potentially blinding infection with a poor visual outcome, even when effective antibiotics are used to treat the infecting bacterium [14, 18]. Bacteria must overcome the epithelial cell layer in order to cause keratitis, and the killing of ocular surface cells is one mechanism bacteria could use to access the stromal layer that resides under the epithelium [19]. Therefore, we set out to study mechanisms by which keratitis causing bacteria damage the epithelium, which are largely unknown for the Enterobacteriaceae family of bacteria.
In this study, we observed that clinical keratitis isolates of S. marcescens cause bleb formation and cellular death in human ocular cells. However, S. marcescens bacterial genomes rarely encode genes for a T3SS, with strain FS14 [20], isolated from a leaf-cutter ant fungus garden, being the only strain described in the literature with a T3SS. These results suggested that S. marcescens has another mechanism to elicit these structural changes from the human epithelial cell, and we therefore employed a genetic screen to identify bacterial genes required for eliciting bleb formation. The role of the identified genes in bleb formation, cytotoxicity, and virulence was characterized using strains with deletion mutations and their corresponding complements, and a potential regulatory pathway was determined. We demonstrated that the Rcs stress response system controls expression of pore forming toxins secreted by a type V secretion system (T5SS) mechanism. The role of the T5SS-dependent cytolysin in bleb formation was validated using keratitis isolates of P. mirabilis, which suggests a novel conserved mechanism by which bacteria can induce cellular blebs to facilitate pathogenesis at epithelial surfaces.
Results
S. marcescens induces toxic membrane bleb formation by human corneal cells in vitro
Dramatic bleb formation on the cellular surface of a human corneal epithelial cell line (HCLE) was observed by microscopic analysis when S. marcescens contact lens-associated keratitis isolate K904 was co-incubated with the human cells (Fig 1A). Bleb formation was absent in HCLE cells exposed to bacterial growth medium (Mock) without bacteria (0%, n = 969 cells), whereas 69.5 ± 15.0% of HCLE cells challenged with S. marcescens K904 bacteria for 2 h (MOI = 200) produced blebs (n = 920 cells) (p<0.001, Fisher Exact test). Bleb formation frequency remained high when the MOI was 50 (95.4 ± 8.2%, n = 571), but reduced with an MOI = 10, (4.6%, n = 22). Confocal laser scanning microscopy and CellMask fluorescent membrane stain support that the bleb structures are extensions of HCLE plasma membranes and can become almost as large as the cells (Fig 1B). Scanning electron microscopy revealed S. marcescens bacteria associated with the membrane blebs (Fig 1C).
(A) Confocal differential interference contrast (DIC) micrographs of human corneal cell line (HCLE) challenged with LB medium (mock) or with S. marcescens strain K904 (MOI = 200 for 2 h). Yellow arrows indicate a bleb extending from one of the cells. Size bar indicates 50 μm. (B) Confocal micrograph of HCLE cells challenged with S. marcescens strain K904 for 2 h and stained with a fluorescent membrane dye. The image shows a side projection of a z-stack of images. White arrow indicates a surface attached HCLE cell and the yellow arrow indicates a membrane bleb. Size bar is 20 μm. (C) SEM micrograph of blebs (yellow arrows) on a HCLE cell speckled with pseudocolored S. marcescens K904 bacteria (red). Size bar is 10 μm. (D) 2-D area of arbitrarily chosen blebs from 6 independent experiments tracked with video microscopy. (E) Primary corneal cells imaged by confocal microscopy with DIC and fluorescent imaging of the same cells stained with viability dye Calcein AM. Yellow arrows indicate blebs. Bar = 20 μm. Mock n = 48, K904 n = 22. (F) SEM micrograph of porcine corneas that had been exposed to naïve contact lenses (Mock, top) or contact lenses coated with wild-type S. marcescens (K904, bottom) prior to fixation. Blebs are indicated with yellow arrows.
Live microscopic imaging of HCLE cells exposed to S. marcescens K904 produced membrane blebs starting before 120 minutes of co-incubation (compare S Movie 1 for Mock and S Movie 2 for S. marcescens K904, Fig 1D). Blebs were observed to grow over time, and then retract into the cell body as the cells rounded up (Fig 1D).
We tested whether the observed phenomenon was an artifact of using a specific human cell type, bacterial strain, or bacterial species. Testing bacterial strain specificity, we observed that 34 out of 34 S. marcescens strains derived from a variety of sources including environmental and clinical isolates induced bleb formation; these include reference strain Db11 and laboratory strain PIC3611 (S1A Fig). A variety of other species were tested for the ability to induce blebs in HCLE cells (MOI = ~200 for 2 h exposure). We found that Proteus mirabilis and Edwardsiella tarda were able to induce blebs during the 2 h time frame (S1B Fig), but no blebs were induced by tested strains of Acinetobacter baumannii, Citrobacter frundii, coagulase negative staphylococci, Enterobacter aerogenes, Enterococcus faecalis, Klebsiella pneumoniae, Morganella morganii, Staphylococcus aureus MRSA and MSSA, Staphylococcus epidermidis, Streptococcus pneumoniae [21], and Stenotrophomonas maltophilia. With Pseudomonas aeruginosa, keratitis isolate K900 [22] did induce blebs (S1B Fig) but wound isolate PAO1 [23] did not under the tested conditions.
Beyond the HCLE cell line, human primary corneal cells produced blebs in response to S. marcescens strain K904 (Fig 1E). With an MOI of 50, 34% of primary cells had a bleb (n = 22), as compared to 0% without bacterial challenge (n = 48). Calcein AM staining for intact, metabolically active cells suggested that the blebbing cells are no longer viable (Fig 1E). Similarly, S. marcescens strain K904 induced membrane bleb formation in airway epithelial cell line A549 (S2A Fig). S. marcescens causes contact lens associated keratitis, so we tested whether this effect could be seen on corneal tissue exposed to S. marcescens inoculated contact lenses. Strain K904 was introduced onto contact lenses and exposed to pig corneas ex vivo for 3 h. SEM analysis revealed extensive surface changes and membrane bleb formation on the porcine ocular surface on the S. marcescens exposed corneas, but not on control corneas bearing contact lenses without bacteria (Fig 1F).
Genetic screen for mutant strains deficient in bleb induction
The S. marcescens strain K904 genome was sequenced (Genbank PRJNA243053) with no evidence for a type III secretion system (T3SS). Because a T3SS-independent mechanism was therefore implicated, transposon mutagenesis was performed to elucidate the bacterial factors required for this phenotype. 6,920 mutants were screened for the inability to induce bleb formation in HCLE cells and kill the cells as judged by Calcein AM staining. Five mutants were reproducibly defective in bleb induction, which we confirmed with primary corneal epithelial cells. For K904 91.2±4.9% (n = 334) of cells had blebs, compared to 0% for the mutants (n≥160) (Fig 2A). The mutations were mapped to two loci. Three were in the shlBA operon, with two in shlB at base pair 378 and 825 out of the 1680 base pair gene, and one in shlA at base pair 4063 out of the 4824 base pair gene (Fig 2B). Two other mutations mapped to different locations in the gumB gene at base pairs 170 and 957 out of the 2136 base pair gene (Fig 2B). The shlBA operon codes for a type Vb secretion system with ShlB being an outer membrane transporter and ShlA its cognate surface-associated and secreted cytolysin [24]. The gumB gene is a recently described member of the IgaA family involved in bacterial stress response that confers pleiotropic phenotypes when mutated [25]. IgaA is an inner membrane protein that influences Salmonella virulence in rodent infection models [26] and controls the Rcs stress response transcriptional system.
(A) Confocal micrographs of primary human corneal epithelial cells challenged with S. marcescens strain K904 wild type and mutant strains isolated for being unable to induce bleb formation (MOI = 200 for 2 h). Three of five mutants are shown, with the other two having an indistinguishable effect on the corneal cells. Differential interference contrast (DIC) and calcein AM viability stained images are shown. (B) Genetic context of transposon insertion mutations that render S. marcescens unable to induce bleb formation in corneal epithelial cells. Downward facing blue arrows indicate transposon insertion sites. Size bar indicates 1000 base pairs.
shlA and shlB genes are necessary, and ShlA is sufficient, for bleb induction
Fifteen of the 34 bleb-inducing S. marcescens clinical isolates were selected arbitrarily among isolates that had caused different types of ocular infections. These were subject to PCR analysis for the shlA gene. Strain PIC3611 was used as a positive control and an shlBA deletion variant of PIC3611 was used as a negative control. DNA samples from all tested strains, except the deletion mutant, produced an amplicon consistent with the shlA gene (S3 Fig). This result is consistent with shlA being a conserved gene in S. marcescens, and likely responsible for bleb induction by the tested strains.
To further test the necessity of this this operon in bleb induction by S. marcescens, a deletion allele of the shlB gene was constructed in S. marcescens strain K904. The ΔshlB mutant strain was completely defective in the ability to induce blebs and kill HCLE cells in our test conditions (Fig 3A). Addition of the shlBA operon expressed from the nptII promoter on a plasmid (pshlBA) complemented the ΔshlB mutant phenotype supporting that the defect was due to mutation of shlB and not an unknown mutation elsewhere on the chromosome or a polar effect on adjacent genes (Fig 3A). The pMQ591 plasmid (pshlBA::tn), which has the shlBA operon with a transposon insertion in shlA at base pair 4063, was able to restore bleb formation to the S. marcescens ΔshlB strain, providing genetic evidence that the ΔshlB mutation in strain CMS4236 was nonpolar, since the active ShlA protein must come from the chromosomal copy of shlA (Fig 3A). A resazurin fluorescence-based assay was used as a second method to validate the cytotoxic phenotypes demonstrated in this study using calcein AM staining (S4A and S4B Fig). We observed consistent results, including the lack of cytotoxicity caused by the ΔshlB mutant and the restoration of cytotoxicity using complementing plasmids (MOI 10 and 200) (S4A and S4B Fig). Thus, even though shlA may be expressed in the ΔshlB mutant, the ShlA protein is not secreted without the ShlB transporter, as was previously shown in other strains [24].
Confocal micrographs of HCLE cells imaged with differential interference contrast (DIC) and calcein AM viability stain. Yellow arrows indicate blebs extending from corneal cells. The percent of bleb positive cells induced by the indicated treatment are shown. (A) Confocal micrographs of HCLE cells with S. marcescens strain K904 and mutant strains (MOI = 50, 2 h incubation). Vector = pMQ125 or pMQ131; pshlBA = pMQ541; pshlBA::tn = pMQ591. (B) Microscopic evaluation of HCLE cells exposed to E. coli (Top10) with a vector (pMQ175) or shlBA expressing plasmid (pMQ492) at MOI = 50 for 2 h. Cells were alternatively exposed for 3 h to sterile-filtered supernatants from E. coli with the pMQ175 (SUP Control) or pMQ492 (SUP ShlA) plasmids, or partially purified ShlA-containing supernatant fractions from E. coli with the vector negative control (PUR Control) or with pMQ492 (PUR ShlA). Vector = pMQ125; pshlBA = pMQ492.
Importantly, shlBA expression was sufficient to confer bleb-induction ability to the E. coli laboratory strain S17-1 λ-pir that is normally unable to generate blebs (Fig 3B and S2 Fig). This result suggested that ShlA may be sufficient to induce bleb formation. Further evidence supporting that ShlA is sufficient for bleb induction came from observations that partially purified ShlA expressed from E. coli could induce bleb formation; 31±3% of HCLE cells exposed to filtered supernatants (n = 75) and 33±15% (n = 208) of cells challenged with purified ShlA exhibited blebs (Fig 3B). This is in sharp contrast to the absence of blebs in cells challenged with preparations made from E. coli harboring the control vector without shlBA (n≥80 cells, Fig 3B). E. coli expressing pshlBA::tn with the transposon mutation in shlA, noted above to contain a null shlA allele, was unable to induce blebs or kill HCLE cells (S2B Fig), which supports the conclusion that ShlA rather than ShlB is required for bleb induction. Combined, these data support a working model that the T5SS ShlB and cytolysin ShlA are the S. marcescens virulence factors responsible for induction of membrane blebs in mammalian cells and suggest that GumB is a regulator of shlBA expression.
A ShlA-like, T5SS-dependent, cytolysin from P. mirabilis is necessary for bleb induction
Since P. mirabilis is able to induce bleb formation in HCLE cells and its genome contains an shlBA-like virulence operon hpmBA [27], we tested whether this operon could induce bleb formation. HpmA is 44% identical to ShlA at the amino acid level and HpmB/A constitute a Type Vb secretion pair analogous to ShlBA. Induced expression of the hpmBA operon from a plasmid was able to confer the bleb-formation phenotype to E. coli (82±9% blebs, n = 156, Fig 4A). The hpmBA plasmid could complement the ΔshlB mutation in S. marcescens; there were fewer blebs than wild type treated cells (20±17%, n = 287), but the cells appeared to be dead (Fig 4A).
Confocal micrographs of HCLE cells imaged with differential interference contrast (DIC) and calcein AM viability stain. Yellow arrows indicate blebs extending from corneal cells. (A) Confocal micrographs of HCLE cells with E. coli strain (Top10), S. marcescens (S. m.) ΔshlB, and P. mirabilis keratitis isolate K2644 and isogenic hmpA mutant strain (MOI = 50, 2 h incubation, 1 hour with E. coli). (B) As in (A), using P. mirabilis keratitis isolate K2675. phpmBA = pMQ601; vector = pMQ132; phpmA = pMQ602.
To further verify the importance of HpmA in bleb formation, the chromosomal hpmA gene of P. mirabilis was mutated in two clinical keratitis isolates K2644 and K2675, with 56±12% (n = 122) and 39±7% (n = 219) bleb formation, respectively, (Fig 4A). Unlike the wild-type parental strains, isogenic hpmA mutant strains were defective in bleb formation and toxicity, and these phenotypes could be complemented by expression of the hpmA gene from a plasmid. Zero percent bleb formation was observed from corneal cells treated with the hpmA mutants with the vector alone (n≥140). For cells exposed to the hpmA mutants with the hmpA plasmid, ~40% had blebs (39±12%, n = 208 for strain K2644 and 40±10%, n = 187 for strain K2675, Fig 4A and 4B). Together, these data indicate that T5SS secreted cytolysins of the ShlBA family represent a conserved mechanism by which bacteria elicit rapid cell death and dramatic morphological changes in human cells.
GumB is required for bleb formation because of shlBA regulatory activity
Similar to the gumB transposon mutant (Fig 2A), strain K904 with a deletion of the gumB gene was unable to induce blebs or kill primary corneal or HCLE cells (0% cells had blebs, n = 159 cells) (Fig 5A). The same trend was observed when the gumB gene was deleted from the S. marcescens reference and insect pathogen strain Db11 [28] (S1A Fig). Plasmid-based expression (lac promoter) of gumB (85±12%, n = 280) or IgaA-family genes from Escherichia coli (yrfF, 95±1%, n = 175), Salmonella Typhimurium (igaA, 95±5, n = 175), and Klebsiella pneumoniae (kumO, n = 96±1, n = 227) was able to restore bleb-induction ability to the ΔgumB mutant, which supports the notion that the function of IgaA-family proteins is highly conserved (Fig 5B). The umoB gene from P. mirabilis was unable to complement the ΔgumB mutant, suggesting that it differs enough structurally from GumB as to not replace protein-protein interactions necessary for GumB function in S. marcescens (Fig 5B and, 0%, n = 123). Additionally, expression of gumB in E. coli did not enable E. coli to induce blebs (Fig 5A) (0%, n = 82 for vector and 225 for pgumB), suggesting that GumB is necessary, but insufficient for the bleb-induction phenomenon.
(A,B) Confocal micrographs of primary human corneal and HCLE cells imaged with differential interference contrast (DIC) and calcein AM viability stain after exposure to bacteria for 2 h at MOI = 50, except where noted. Yellow arrows indicate blebs extending from corneal cells. (A) Confocal micrographs of HCLE cells with S. marcescens strain K904, mutant strains, or E. coli strain EC100D pir-116. Vector = pMQ132; pshlBA = pMQ541; pgumB = pMQ480. The percent of bleb positive cells induced by select bacteria are shown. (B) HCLE cells exposed to the gumB mutant strain (MOI = 50) with plasmid-borne igaA-family genes from S. marcescens (pgumB = pMQ480), S. enterica (pigaA = pMQ530), K. pneumoniae (pkumO = pMQ529), E. coli (pyrfF = pMQ531), or P. mirabilis (pumoB = pMQ600). Vector = pMQ132. (C) Relative gene expression using the ΔΔCT method depicts shlA transcript levels in the wild type (K904) and ΔgumB mutant strains at OD600 = 3; *p = 0.0286, Mann-Whitney test.
We tested whether shlA expression was reduced in the gumB mutant using qRT-PCR, and observed a 100-fold reduction in transcript (Fig 5C). This finding suggests that gumB mutant strains likely are defective in the ability to induce blebs because they do not produce adequate ShlA cytolysin. To test this model, the constitutive shlBA expression plasmid was introduced into the gumB mutant. The resulting strain induced blebs and was highly cytotoxic (Fig 5A and S4 Fig), which indicates that artificial upregulation of shlBA bypasses the GumB-mediated regulation required for this virulence function.
GumB regulates bleb formation and virulence through the Rcs system
Reports indicate that IgaA-family proteins inhibit the Rcs phosphorelay system in other genera from the Enterobacteriaceae family [29, 30]. The Rcs system is a multicomponent version of a two-component transcription factor system involved in responses to extracellular and envelope stress [31, 32] (Fig 6A). The core Rcs system is composed of sensor histidine kinase RcsC, an intermediate phosphoprotein RcsD, and the RcsB response regulator [32]. Therefore, one would predict that if the Rcs system is derepressed in an IgaA-family protein mutant strain (ΔgumB), then elevated expression of Rcs system components in the wild-type strain could mimic the gumB mutant phenotypes (Fig 6A). To test this prediction, the rcsC gene was placed under control of the E. coli lac promoter on a medium-copy plasmid in the wild-type strain, K904. We observed gumB mutant strain-like phenotypes for the wild type with the rcsC multi-copy plasmid, such as reduction of pigmentation and mucoid colony morphology, which supports that multicopy expression of rcsC was activating the Rcs system akin to mutation of gumB that also prevents pigmentation (S5 Fig). HCLE cells exposed to strain K904 with the rcsC expression plasmid, but not the vector control, were defective in inducing bleb formation and cytotoxicity (Fig 6B). 93±6% bleb formation was observed in cells exposed to the K904 wild-type with vector control (n = 208), whereas those challenged with K904 with prcsC produced no blebs (0%, n = 454).
(A) Model for the regulatory circuit through which GumB functions to regulate epithelial cell bleb formation, based on this study and DiVenanzio, et al [58]. GumB inhibits (red stop bar) the Rcs-phosphorelay system through which the response regulator RcsB inhibits shlBA expression. The ShlA cytolysin is secreted through the outer membrane by ShlB and is maintained on the bacterial outer membrane or released into the environment where it can form pores in mammalian cell membranes and stimulate bleb formation and cellular death. (B) Confocal micrographs of HCLE cells imaged with differential interference contrast (DIC) and calcein AM viability stain after exposure to bacteria for 2 h at MOI = 50. Yellow arrows indicate blebs extending from corneal cells. Multicopy expression of the rcsC histidine sensor kinase gene confers gumB-like phenotypes to the wild type (prcsC = pMQ514). The ΔgumB mutant bleb-phenotype is suppressed by mutation of the rcsB response regulator gene, and this effect can be complemented by the wild-type rcsB gene on a plasmid (prcsB = pMQ614).
The proposed model (Fig 6A) also suggests that inactivation of the Rcs system in a gumB mutant strain should restore toxicity, bleb induction ability, and pigment production. The gene for the RcsB response regulator was mutated in the ΔgumB strain background. In order to interrogate the model and validate the strains, Rcs regulation of pigmentation was analyzed. The rcsB mutation reversed the gumB mutant pigment defect (S5 Fig), which supports that RcsB acts downstream of GumB (S6 Fig). We also observed that the gumB mutant strain pigment defect could be restored through complementation with wild-type rcsB expression from a plasmid (S5 Fig), which further supports the validity of the mutation and plasmid. With regards to host-pathogen interactions, the ΔgumB rcsB double mutant strain was indistinguishable from the S. marcescens K904 parental strain for bleb induction and cytotoxicity to HCLE cells (Fig 6B). Importantly, the ΔgumB rcsB double mutant strain toxicity and bleb inducing phenotypes could be complemented with rcsB on a plasmid (Fig 6B). Zero blebbing cells were counted with the gumB mutant with the vector control (n = 141); 94±6% of cells had blebs in the gumB rcsB with vector group (n = 252), and 0% of cells had blebs when exposed to the gumB rcsB mutant complemented with prcsB (n = 338). Together, these data indicate that bleb formation regulation by GumB requires a functional Rcs system, and that activation of the Rcs system prevents S. marcescens from inducing bleb formation and cytotoxicity to epithelial cells.
S. marcescens-induced epithelial blebbing and toxicity is due to ShlA-mediated pore formation and resultant necroptosis
Next, we investigated the mechanism of cellular death induced by ShlA-like cytolysins in corneal epithelium. Previous studies have shown that osmoprotectants can prevent bacterial T3SS-mediated bleb formation and necroptosis [33, 34]. The osmoprotectant sorbitol (300 mM) was able to reduce bleb formation from 89% (n = 225) for HCLE cells exposed to S. marcescens (strain K904) to 10% (n = 409) for those exposed to S. marcescens and sorbitol (p<0.001). Dextran, a branched polysaccharide, is able to prevent cellular lysis induced by purified streptolysin O and other pore forming toxins, including ShlA, by occluding pores introduced by such toxins [35–38]. Here, dextran was found to reduce cells with blebs from 94.5% (n = 383) with S. marcescens challenge MOI = 50 or 200 to 0 or 4.4% (n = 397/544) with S. marcescens and dextran, a significant reduction (p<0.0001, Fisher’s Exact) (Fig 7A). These data suggest that the membrane pore introduced by ShlA’s pore forming domain is responsible for the bleb and cytotoxicity phenotypes.
Confocal micrographs HCLE cells imaged with differential interference contrast (DIC) and calcein AM viability stain after exposure to bacteria for 2 h at MOI = 50. Yellow arrows indicate blebs extending from corneal cells. HCLE cells were exposed to S. marcescens K904 (MOI = 50 for 2 h) with either (A) dextran sulfate 8000 (30 mM, occludes pores caused by pore-forming toxins), (B) necrostatin 5 (100 μM, an inhibitor of necroptosis), or coenzyme Q10 (CoQ10, 0.1 μM, an antioxidant,) or an equal volume of vehicle DMSO incubated with cells for 1 h prior to challenge.
It has been demonstrated that intracellular S. marcescens can initiate the type of programmed cell death known as necroptosis in macrophages in a ShlA-dependent manner [39]. Oxidative stress plays a major role in necroptosis, so we tested whether the antioxidant coenzyme-Q10 (0.1 μM), had an impact on S. marcescens induced damage. CoQ10 prevented S. marcescens-induced cytotoxicity and bleb formation (0% blebs, n = 146) (Fig 7B), whereas DMSO alone did not alter the ability of S. marcescens to induce blebs (89±6%, n = 121) (Fig 7A).
We tested whether blocking of necroptosis using the RIP-1 inhibitor necrostatin 5 could alleviate S. marcescens-induced phenotypes; a strong reduction in bleb formation (2±3%, n = 316) and cytotoxicity was observed (Fig 7). Necrostatin 5 itself did not produce a bleb or cytotoxicity phenotype (0% blebs, n = 97), nor did the vehicle (DMSO), 0% blebs, n = 115). An inhibitor that targets the major regulator of necroptosis, the mixed lineage kinase domain-like protein (MLKL), was tested [39, 40]. The MLKL inhibitor GW806742X produced a dose-dependent reduction in bleb formation by HCLE cells challenged with wild-type S. marcescens (S7 Fig). Together, these data suggest that S. marcescens-induced necroptosis in response to ShlA-mediated pore formation is responsible for bleb formation and cellular death.
gumB is required for virulence in a ShlA-dependent manner
Whereas the ShlA cytolysin is a known S. marcescens virulence determinant in several pathogenesis models [41–43], the role of GumB is unreported. A Galleria mellonella model of infection was used to test whether GumB is necessary for infection in vivo. When survival was analyzed over time (Fig 8A), larvae that had been injected with S. marcescens strain K904 (200 CFU/larva) started to die just after 20 h post injection. The larvae had a median survival of 23 h, and all larvae were dead by 44 h (Fig 8A).
Shown are survival curves of G. mellonella larvae challenged with S. marcescens. (A) G. mellonella survival over time following injection with 200 CFU is shown (n = 12 for K904, n = 13 for ΔgumB); p<0.001 Log-rank test. (B) Similar survival curves for G. mellonella over time were observed after treatment with either K904 or ΔgumB despite large differences in CFUs injected (n = 14). (C) Survival of larvae injected with the ΔgumB mutant and ΔshlB plasmid with various plasmids as indicated (n = ≥12). Vector = pMQ125; pshlBA = pMQ541.
Strikingly, the ΔgumB mutant strain-injected larvae (200 CFU/larva) were fully viable at 44 h when the experiment reached its endpoint (Fig 8A, p<0.001 Log-rank Test).
In a separate experiment, different doses of S. marcescens (strain K904) and the ΔgumB mutant strain were injected into G. mellonella larvae. A similar survival curve was observed between the two strains with 160,000 CFU of the ΔgumB strain (n = 14) and 10 CFU of the wild-type K904 strain (Fig 8B), which sharply delineates the >10,000-fold difference in the ability of the isogenic strains to kill a host organism.
The ΔgumB strain with shlBA constitutively expressed on a plasmid was used to test whether reduced shlBA expression contributes to the lack of virulence exhibited by the ΔgumB mutant strain in the G. mellonella pathogenesis model. Plasmid-based expression of shlBA increased virulence of the ΔgumB mutant strain compared to the ΔgumB mutant strain carrying the vector negative control; however, it did not confer wild-type levels of virulence (Fig 8C). The isogenic ΔshlB mutant was also defective in virulence compared to the wild type strain and was fully complemented with shlBA expressed from a plasmid. Together these results demonstrated that the gumB mutant is attenuated in virulence relative to the wild type, and suggest that the virulence defect is at least partially due to a loss of ShlA production.
The mechanism for the loss of viability in the G. mellonella model was further analyzed. Bacteria were isolated from larvae before larval death (24 h post-injection), and the CFU were enumerated. There was a ~50-fold reduction in the median CFU of S. marcescens ΔgumB strain CFU isolated from the larvae compared to the K904 wild-type strain (p = 0.029, Mann Whitney test) (Fig 9A). To test whether the ΔgumB mutant strain was less capable of growth on the nutrients available in the larvae, we assessed bacterial growth in inactivated larval homogenates (Fig 9B). The lysates were heat treated to prevent melanization and growth of endogenous bacteria and clarified by centrifugation. The growth rate as assessed by optical density measurement of the ΔgumB mutant and wild type strains were similar in clarified lysates (Fig 9B), and the CFU achieved at 24 h were indistinguishable (p = 0.565, Mann Whitney test), (Fig 9C). We also tested whether the ΔgumB mutant had reduced growth at limiting oxygen concentrations, which could explain its reduced ability to proliferate within the hemolymph of the larvae. The ΔgumB mutant and K904 wild-type strain exhibited qualitatively equivalent colony size on LB agar plates following growth in an anaerobic bag (S5 Fig). Likewise, the ΔgumB mutant strain was similarly tolerant to hydrogen peroxide. Disk diffusion tests indicated that the gumB mutant was no more susceptible than the wild type strain (16.7±0.7 mm diameter of growth inhibition for the wild type strain and 17.4±1.7 for the ΔgumB strain, p = 0.203 Student’s T-test). This suggests that the ΔgumB mutant is more susceptible to immune components such as phagocytizing cells in the larval hemolymph rather than being unable to grow under nutrient- or oxygen-limiting conditions or exposure to reactive oxygen species in the larvae.
(A) Enumeration of S. marcescens K904 wild type and ΔgumB mutant strains 24 h after injection of 103 CFU into G. mellonella. Median and range are shown, n≥3. * indicates significant difference between medians, Mann-Whitney test (p = 0.0286). (B) Growth of K904 and the ΔgumB mutant in G. mellonella homogenates (n = 10). Error bars indicate standard deviation. (C) Enumeration of S. marcescens K904 wild type and ΔgumB mutant strains 24 h growth in heat-treated G. mellonella homogenates, n = 7. Median and interquartile range is shown. (D) Representative experiment describing uptake and proliferation of S. marcescens K904 wild type and ΔgumB mutant strains in RAW macrophage cells (n = 3), mean and standard deviations are shown. *** indicates significant difference by 2-way ANOVA with Tukey’s post-test (p<0.001).
To test whether there is a defect in the ability of the ΔgumB mutant strain to survive interaction with phagocytic immune cells, we tested the ability of the bacteria proliferate in a macrophage-like murine cell line, RAW264.7 cells. The ΔgumB mutant strain was taken up at a reduced rate (~6-fold lower, p<0.05 Student’s t-test) compared to the wild-type K904 strain when CFU within RAW cells was assessed after 2 h of co-culture (Fig 9D). Proliferation within the RAW cells measured at 24 h post-inoculation was also measured. Whereas the wild-type CFU increased almost 7-fold within the RAW cells, there was a ~50% reduction in ΔgumB strain CFU (Fig 9D). Wild-type gumB expression from a plasmid was able to complement these defects in uptake and intracellular proliferation/survival of the gumB mutant, but had no effect on the wild-type K904 strain, as expected (Fig 9D).
We evaluated whether the gumB mutant defect in proliferation within RAW cells was due to reduced shlBA gene expression by expressing shlBA with the nptII promoter from a plasmid in the gumB mutant. We observed a significant, although partial, restoration in bacterial proliferation in RAW cells when shlBA was expressed in the gumB mutant (S8 Fig). Together, these data indicate the GumB is required for virulence and indicate that gumB is required for resistance to phagocytic cell responses of the innate immune system.
Discussion
We report a T5SS-dependent, T3SS-independent mechanism by which Gram-negative bacteria can induce massive morphological changes and cellular death in human cells. The purpose of this study was to characterize and gain mechanistic insight into how Enterobacteriaceae damage the corneal epithelium, a barrier that they must overcome to gain access to the corneal stroma, a niche where they can rapidly replicate. We observed that these bacteria induce formation of blebs in human corneal epithelial cells. S. marcescens-induced blebbing was evident in several types of mammalian epithelial cells and by >30 S. marcescens isolates (100% of isolates tested), which indicates that the effect is broadly conserved and not limited to only a few bacterial isolates or mammalian cell types. The induction of blebs on intact corneas following exposure to bacteria-coated contact lenses implies that contact lens delivery of bacteria or ShlA-like cytolysins may cause damage to the ocular epithelium, possibly whether the bacteria are alive or not. Consistently, approximately 10% of contact lens wearers have an adverse contact lens wear event every year, such as red eye and irritation, and bacteria such as S. marcescens are common contaminants of contact lens cases and lenses [44–46]. Even more important is that contact lenses are a major risk factor for the vision threatening infection, microbial keratitis, with a third to a half of keratitis patients being contact lens users [18, 47].
Notably, the epithelial cell blebs reported here were similar in morphology to those P. aeruginosa-induced blebs described by Fleiszig and colleagues [6, 7, 33]. One difference between these studies and ours is that P. aeruginosa bacteria actively proliferate within the blebs [6, 7, 48], whereas S. marcescens strain K904 was not observed within the epithelial cells. Additionally, the frequency of bleb formation was higher for S. marcescens-exposed cells, with 15–20% of corneal cells exhibiting blebs after treatment with P. aeruginosa at MOI = 100, compared to up to 70% for S. marcescens-exposed cells at MOI = 200, and 26% at MOI = 50. A further difference between P. aeruginosa and S. marcescens is that bleb induction by P. aeruginosa requires a T3SS, a nanomachine largely absent in S. marcescens isolates. This study, rather, demonstrates that S. marcescens requires a T5SS to induce epithelial blebs and cytotoxicity. A subset of T3SS-lacking P. aeruginosa isolates have been described that express the T5SS secreted ExlA toxin [49, 50], and we speculate that these ExlA positive isolates may also induce blebbing and cellular death in a similar manner to P. mirabilis and S. marcescens.
Genetic analysis implicated both the T5SS composed of ShlA and ShlB and the Rcs system regulator GumB in bleb induction, control of toxicity, and facilitation of virulence. Importantly, expression of shlBA in non-pathogenic E. coli strains conferred the ability to induce blebs and kill corneal cells in vitro. This result suggests that this single virulence determinant was sufficient for the observed host-pathogen interactions, a result that was corroborated using partially purified ShlA. Furthermore, the ability of HpmA from P. mirabilis to induce blebs and kill corneal cells supports the model that this widely conserved family of cytolysins are sufficient to cause bleb formation and cell death. Indeed, the other bacteria tested in this study able to cause blebs were from species known to harbor ShlBA-like cytolysins such as E. tarda [51], although we did not prove that the strain we used has this gene. The P. aeruginosa isolate that induced blebbing has not been molecularly characterized, so it is not clear whether it induces blebbing through a T3SS or T5SS mechanism. Notably, ShlA-like proteins are found in a variety of organisms beyond those discussed above, including Chromobacterium violaceum, Haemophilus ducreyi, Photorhabdus luminescens, and Yersinia species [24, 52].
Interestingly, bacteria known to make other types of cytolysins / hemolysins such as S. aureus and S. pneumoniae did not induce bleb formation under our tested conditions, even using strains known to express the respective hemolysins, e.g. S. aureus 8325–4 [53]. Bleb formation by Streptococcus species may be expected since purified streptolysin O has been a key tool in understanding the biology underlying bleb formation [11, 54]. Our observations therefore suggest that many hemolytic bacteria do not generate sufficient pore forming toxins under the tested conditions to induce bleb formation in corneal cells or that the tested cells lack receptors required by the respective toxins. Furthermore, non-pore forming toxins can induce blebs. These include T5SS secreted serine proteases from E. coli, ExpC and Pet, which cause damage and induce bleb formation when added exogenously to epithelial cells [55–57].
This study implicates GumB as a S. marcescens virulence factor and mediator of host-pathogen interactions. GumB was found to be necessary for bleb induction and cytotoxicity by S. marcescens to corneal cells and virulence in G. mellonella. The bleb formation and cytotoxicity phenotypes were complemented by the wild-type gumB gene and several other IgaA-family genes on plasmids. This result implies that GumB is functionally conserved with other IgaA-family proteins, with the possible exception of UmoB from P. mirabilis. Several independent clones of umoB from different P. mirabilis genomes and different plasmid replicons were tested, suggesting that the lack of complementation was not due to a faulty complementation plasmid. The umoB gene codes for a protein that is more distantly related to GumB than the other tested proteins: 42% identity to GumB compared to ≥54% identity for the other tested proteins [25]. These structural differences may account for its inability to complement the ΔgumB mutation.
Since expression of gumB in E. coli was insufficient to induce blebbing in mammalian cells, and it is unlikely that an IgaA-family protein could itself cause damage to mammalian cells, we tested the hypothesis that GumB is required for bleb formation and cytotoxicity through activation of shlBA expression. In support of this model, shlA expression was highly reduced in the ΔgumB mutant strain and ectopic expression of shlBA restored the ability of the ΔgumB mutant to induce blebs and kill cells. These data support the model that GumB is defective in bleb induction because it fails to produce sufficient levels of the ShlA cytolysin.
Because IgaA-family proteins, similar to GumB, regulate the envelope stress response Rcs system [29, 30], we tested whether GumB functions through control of the Rcs system. Multicopy expression of rcsC in the wild type strain phenocopied gumB mutant strain phenotypes, conferring the loss of cytotoxicity and bleb induction. Additionally, mutation of rcsB in the gumB mutant strain restored bleb formation and cytotoxicity phenotypes. These were the expected outcomes if GumB functions to repress Rcs system function and RcsB inhibits shlBA transcription. Together these experiments support the model that GumB regulates shlBA expression indirectly through the Rcs system (Figs 6 and S6). Consistent with this model, the Vescovi group showed that shlBA transcription is directly inhibited by RcsB in S. marcescens strain RM66262, and it was proposed that phosphorylated RcsB binds to the promoters of and directly represses both shlBA and flhDC transcription [58]. Since FlhDC is a positive and direct regulator of shlBA expression, and activated RcsB shuts down flhDC expression, it is clear that RcsB can shut down shlBA transcription both directly and indirectly [58]. It was reported that GumB is necessary for flhDC expression [25] and in this study, for shlBA. This is in agreement with a study by DeVenanzio [58] regarding RcsB control of shlBA, and further supports a role for GumB in Rcs system control.
With respect to virulence, the ΔgumB mutant strain was highly attenuated, as injection of >10,000-fold more ΔgumB than wild-type K904 CFU into the larvae was required to produce similar survival profiles. The ΔgumB strain larvae killing defect was partially restored by multicopy expression of shlBA from a plasmid. Data here demonstrated that shlBA is essential for virulence in a G. mellonella model of infection, extending the host-range in which ShlA is a virulence factor. However, the ΔshlB mutant defect was not as severe as the ΔgumB defect, with 100-fold more CFU of the ΔgumB strain than ΔshlB strain required for complete killing of the larvae (Fig 8C). Together, these data suggested that the lack of shlBA expression by the gumB mutant is partially, but not completely responsible for reduced virulence. In addition to ShlA, other factors regulated by GumB likely contribute to virulence. These could include as metalloproteases [59, 60], FlhDC controlled phospholipase A [61], flagella [62], the hemolytic surfactant serratamolide [63], the biologically active pigment prodigiosin [64], and Rcs system regulated outer membrane vesicle [65] and capsular polysaccharide production [66]. Regardless, GumB-mediated regulation of shlBA accounts for a large portion of S. marcescens virulence activity.
Experiments demonstrated that GumB is necessary for replication within G. mellonella, but the ΔgumB mutant strain is perfectly able to use G. mellonella as a growth substrate. Experiments with the RAW macrophage-like cell line indicated that GumB is required for survival and proliferation after being phagocytized. This result suggests that GumB-regulated factors are required for the bacteria to survive within cells. In support of this notion, there is a growing body of evidence indicating a key role for ShlA in S. marcescens survival within and egress from intracellular vacuoles and regulation of autophagic processes [58, 67–69]. This is somewhat antagonistic to data from E. coli and S. enterica, where partial function alleles of igaA and yrfF increased survival of phagocytized bacteria [70–72]. The different requirements for GumB may be due to fundamental differences in the role of IgaA-family proteins between species. Alternatively, because the yrfF and igaA genes are essential for growth, different results may have resulted from the partial function of the igaA and yrfF alleles used in the previous studies [70–72].
Cellular blebs are generally a sign of impending cellular death, and in this case Enterobacteriaceae that cause contact lens associated keratitis may use this mechanism to damage the corneal epithelium, a key barrier to ocular infections [73]. We speculate that contact lens wear can facilitate contact between bacteria with ShlA-like T5SS and the ocular surface, and that even if the bacteria are killed by cleaning solutions, their surface associated and extracellularly secreted pore-forming toxins of the ShlA family could damage the epithelium. Beyond the eye, S. marcescens causes many types of nosocomial infections [15], and has been implicated in the dysbiosis associated with inflammatory diseases of the human gut [74]. In line with this observation, a recent study has shown that ShlA can cause severe damage to the digestive tract in a Drosophila melanogaster model [42]. ShlA also damages lung tissue and is required for hemorrhagic pneumonia, lung dysfunction, and necroptosis of epithelial cells in animal lung infection models [41, 75].
An additional observation of note was that genetic data noted here suggest that the Rcs system is a regulator of the biologically active red prodigiosin pigment, characteristic to many biotypes of S. marcescens. This conclusion was based upon multicopy expression of rcsC conferring a loss of pigmentation, and mutation of rcsB suppressing the gumB mutant strain pigment defect. Further studies will be required to fully analyze the role of the Rcs system in pigment regulation; however, our current data suggests a model wherein the Rcs system inhibits pigmentation under stressful conditions. This is consistent with a previous study indicating that the alarmone cAMP is used to inhibit prodigiosin biosynthesis under metabolic stress [76].
In conclusion, this study identifies a novel mechanism by which bacteria cause dramatic and lethal morphological changes in host epithelial cells to potentiate pathogenesis on mucosal surfaces, as well as the regulatory pathways underlying this important virulence activity. This ShlB and ShlA-dependent mechanism is highly toxic and employed by a broad range of Gram-negative bacterial pathogens. In the context of bacterial keratitis, this T5SS may enable bacteria to rapidly kill surface epithelial cells, allowing them to penetrate into the corneal stroma, a tissue more permissive to bacterial growth. These findings therefore implicate novel strategies for therapeutic development to prevent this conserved system from causing tissue damage and augmenting disease.
Methods
Ethics statement
De-identified corneas from organ donors were obtained from the Center for Organ Recovery and Education (Pittsburgh, PA) or the National Disease Research Interchange (Philadelphia, PA). Research using de-identified tissue from non-living individuals is not considered human subject research under DHHS regulation 45CFR46, and the use of decedent tissue for this project was approved by the University of Pittsburgh Committee for Oversight of Research and Clinical Training Involving Decedents.
Microbial strains, media, and growth
S. marcescens and P. mirabilis strains are listed in Table 1. Bacteria were grown with on a TC-7 tissue culture roller (New Brunswick) in Lysogeny Broth (LB) medium [77] (0.5% yeast extract, 1% tryptone, 0.5% NaCl) with or without 1.5% agar or in M9 minimal medium [78] supplemented with glucose (0.4%) and casein amino acids (0.06%). Escherichia coli strains used were S17-1 λ-pir [79], WM3064 [80], Top10 (Invitrogen), and EC100D pir-116 (Epicentre). Saccharomyces cerevisiae strain InvSc1 (Invitrogen) was grown with either YPD or SC-uracil media [81]. Antibiotics used in this study include gentamicin (10 μg ml-1), kanamycin (100 μg ml-1), and tetracycline (10 μg ml-1). For growth under oxygen limiting conditions, bacteria were grown on LB plates in a GAS PAK-EZ anaerobe pouch system with indicator (Becton, Dickinson and Company), and incubated at 30°C for 20 h.
Cell culture, bleb formation, cell staining, and cytotoxicity assays
HCLE cells (originally from the Gipson laboratory, Harvard University) [82] were grown in monolayers as previously described [83] in 12 well MatTek glass bottomed dishes (product number P12G-1.5.14-F) that were treated with poly-L-lysine or in tissue culture treated polystyrene 12 well dishes (Costar catalog no. 3513). Cells were grown to confluence in keratinocyte serum-free medium (KSFM) (Gibco cataolog number 10724–011) supplemented with bovine pituitary extract (25 μg/ml) and epidermal growth factor (0.2 ng/ml). Bacteria were grown overnight with aeration at 30°C, washed with phosphate buffered saline (PBS), adjusted to the proper MOI in KSFM in a total volume of 1.5 ml and applied to the MatTek plate. After 2 h of incubation at 37°C with 5% CO2, bacteria were removed by washing cells three times with 37°C PBS, and cells covered in KSFM with or without Calcein AM (0.5 μM, ThermoFisher) for 15 minutes or CellMask plasma membrane stain (100 μM, ThermoFisher), then washed with KSFM and imaged.
Cytotoxicity assays were performed as previously described using the Presto Blue viability assay (ThermoFisher) using bacteria at the described MOI [59].
Primary epithelial cells were obtained using reagents from Gibco and Sigma Aldrich and following the protocol of Chen and colleagues [84] with some modifications. Corneal tissue obtained as noted above from the Center for Organ Recovery and Education (Pittsburgh, PA) was washed three times with Hank’s balanced salt solution supplemented with gentamycin 50μg/ml and amphotericin B 1.25μg/ml. Corneal cells were removed and digested for ~16 h at 4°C using 10 mg/ml dispase II in MESCM (a 1:1 ratio of Dulbecco’s modified Eagle medium and Ham’s F12 medium supplemented with insulin transferrin selenium solution, basic fibroblast growth factor 4 ng/ml, human leukemia inhibitory factor 10 ng/ml, gentamicin 50μg/ml and amphotericin B 1.25μg/ml). Epithelial sheets were removed and incubated with TrypLE protease mixture, neutralized with minimum essential medium supplemented with 20% FBS, and cells were plated into a 12 well plate (~8x105 cells/well).
RAW 264.7 cells were grown and used as previously described [85] using kanamycin protection assays [86] to analyze bacterial proliferation.
Porcine cornea organ culture contact lens model
Porcine corneas were purchased from Sierra Medical (Whittier, CA) and processed as previously described [87]. Corneas and adjacent scleral tissues (~ 3 mm) were excised from eyes (n = 2 per treatment group), rinsed in PBS and placed on supports composed of minimal essential medium (MEM, Gibco), rat tail collagen (1 mg/ml, Sigma), and agarose (1% w/v) in 12 well dishes. MEM was added to cover the tissue up to the limbus. Contact lenses (Air Optix Night and Day Aqua) were incubated in PBS or PBS with S. marcescens strain K904 (OD600 = 1.0; ~2x109 CFU/ml) for 30 minutes and rinsed 2x in PBS to remove unattached bacteria, leaving ~ 1x108 CFU per lens. The control and bacteria-laden lenses were applied to the corneas and together were incubated at 37°C with 5% CO2 for 2.5 h. Lenses were removed and the corneas fixed with glutaraldehyde (2.5%) for 20 h. Corneas were washed with PBS and post-fixed using aqueous osmium tetroxide (1%). The samples were dehydrated using increasing concentrations of ethanol (30%-100%), immersed in hexamethyldisilazane, air dried, and sputter coated with 6 nm of gold/palladium. Corneas were imaged using a JEOL JSM-6335F scanning electron microscope at 3 kV with the secondary electron imaging detector.
Mutagenesis and plasmid construction.
Transposons were introduced into S. marcescens by conjugation as previously described [88] using a Himar-1 based plasposon delivery plasmid pSC189 [89]. Tetracycline (10 μg ml-1) was used to select against donor E. coli growth, and kanamycin (100 μg ml-1) was used to select for S. marcescens with transposon mutations. Transposon insertions were mapped as previously described [89–91].
Cloning was performed using yeast-based recombineering of PCR generated amplicons [92, 93]. PCR amplicons used for cloning were generated using high-fidelity polymerase, Phusion (New England Biolabs) or PrimeSTAR (Clonetech). Clones were analyzed by diagnostic PCR and verified by DNA sequencing (University of Pittsburgh Genomic Research Core). Plasmids are listed in S1 Table. Directed mutagenesis was achieved by two-step allelic replacement or insertional mutagenesis as noted in the text and previously described [92, 93]. Mutations were verified using PCR outside of the cloned region on the mutagenesis plasmid.
Allelic replacement of shlB: To generate the shlB deletion strain, we cloned 698 base pairs upstream of shlB and 604 base pairs downstream of shlB using primer pairs 2619 and 2620 and 2621 and 2622, respectively, into an allelic replacement vector, pMQ460 [91], to generate pMQ473. All primer sequences are listed in S2 Table. The pMQ473 plasmid was introduced into S. marcescens strains K904 [94], PIC3611 [88], and Db11 [28]. Followed by I-SceI expression vector pMQ337 [92] to facilitate recombination. Sucrose resistant isolates were obtained on selective plates (0.5% yeast extract, 1% tryptone, 5.0% sucrose), and mutations were analyzed by PCR.
The shlBA operon was cloned from strain K904 using primers 3464 and 3465 and placed under the control of the E. coli PBAD promoter on plasmid pMQ125 [92] to generate pMQ492. L-arabinose at 0.2% was used to induce expression of shlBA in this study. The pMQ492 plasmid was modified for constitutive expression of shlBA by replacing the PBAD promoter with the E. coli nptII from pSC189 promoter to make pMQ541 (primers 3698 and 3699).
The shlA gene was mutagenized using pMQ524, a Himar-1 ‘phoA delivery plasmid described below. E. coli strain EC100D pir-116 with pMQ492 and pMQ524 were grown for 20 h, then plasmids were harvested. The resulting plasmid DNA was used to transform E. coli strain Top10, which does not support the replication of pMQ524, and kanamycin and gentamicin were used as selection for pMQ492 in which the transposon from pMQ524 had jumped into pMQ492. Two colonies that were blue on plates supplemented with 5-bromo-4-chloro-3-indolyl phosphate (BCIP; Sigma-Aldrich) and L-arabinose were selected, and the transposon was mapped to shlB in one clone and shlA in the other. The resulting plasmids were designated pMQ590 and pMQ591, respectively.
The pMQ524 plasposon delivery plasmid was made by combining a number of amplicons using yeast recombineering as noted above. The amplicons recombined in yeast were 1) a partial phoA gene, without its secretion signal from E. coli strain W3110 [95] (primers 3639 and 3640), 2) yeast replication machinery and URA3 gene from pMQ132 (primers 3647 and 3648) [92], 3) the bla gene, transposase C9 gene, and one inverted repeat from pSC189 [89] (primers 3651 and 3652), and 4) the kanamycin resistance gene nptII, oriR6K and other inverted repeat from pSC189 (primers 3643 and 3644). The resulting plasmid was verified by functional analysis, PCR, and sequencing of junctions.
The rcsB and rcsC and open reading frames were amplified from S. marcescens strain K904 (rcsB, and rcsC) or umoB from P. mirabilis strains (K2644 and K2675) (umoB) and placed under control of the E. coli lac promoter on pMQ132, resulting in plasmids pMQ614, pMQ615, and pMQ600, respectively. No induction was used to express these genes in S. marcescens as this species lacks a lac repressor gene. Primers 3688 and 3689 were used for rcsB, 3691 and 3692 for rcsC, and 3892 and 3893 for umoB.
The pStvZ3 promoter probe plasmid [76] was altered to have a more convenient multicloning site. The pStvZ3 plasmid was cut with BamH1 and used to transform yeast along with oligonucleotides (primers 2664 and 2665) that recombine in yeast to introduce SalI, SacII, SpeI, and BamHI restriction sites to pStvZ3 and generate the plasmid pMQ544 as previously described [92].
The rscB and hpmA genes were mutagenized by targeted insertional mutagenesis. Briefly, a 316 bp long internal region of the rcsB gene and a 621 bp region from hpmA were a cloned into suicide vector pMQ544 and pMQ118, respectively. The resulting rcsB insertion plasmid, pMQ553, and hpmA insertion plasmid, pMQ596, were introduced into recipient by conjugation (as noted above). Primers to amplify the internal region were 3735 and 3736 for rcsB and 3896 and 3897 for hpmA.
The hpmA gene was cloned from strain K2644 into pMQ132 under control of the E. coli lac promoter using primers 3900 and 3901 to make pMQ602, and the hpmBA operon was cloned under control of the E. coli BAD promoter using primers 3919 and 3920 to generate pMQ601.
To generate the shlBA deletion variant of strain PIC3611, lambda red recombineering was used as previously described [96, 97]. A broad host-range delivery plasmid for the lambda red genes was generated by cloning the recombineering machinery from pKD46 [96] using primers 3675 and 3676 into IncQ/RSF1010-based plasmid pMQ397 [98]. After introduction of pMQ538 into strain PIC3611, it was prepared for electroporation and induced with L-arabinose and transformed with 3 μg of a PCR amplicon designed to replace shlBA with a kanamycin resistance cassette from pKD4 [96]. Kanamycin resistant transformants were analyzed for the shlBA deletion using primers that analyze the novel junctions between the S. marcescens chromosome and nptII resistance gene.
Transcriptional analysis
Quantitative reverse transcriptase PCR (qRT-PCR) was used to assess gene expression as previously described [91]. To prepare bacteria for RNA extraction, single colonies were inoculated into 5 ml of LB broth, and the test tubes were incubated 30°C with aeration in 5 ml. After ~16 h, cultures were diluted to OD600 = 0.1 in fresh LB medium, grown to OD600 = 0.5, subcultured to OD600 = 0.1 and then grown to OD600 = 3. RNA and cDNA was prepared and validated to not have chromosomal DNA contamination as previously described [91]. Primers were 2638 and 2639 for the 16S rDNA gene and 4150 and 4151 for shlA sequences.
Partial purification of ShlA
Escherichia coli strain EC100D with pMQ492 (shlB and shlA) and with pMQ175 (empty vector) were grown overnight at 30°C with aeration for 16 h in LB medium supplemented with gentamicin (10μg/ml) and L-arabinose to a final concentration of 0.2% (v/v). Bacteria were removed by centrifugation and filtration (0.22 μm), and supernatants were subject to size fractionation using a 100 kD filter unit (Centricon, Millipore). Protein fractions in PBS (200 μl) were added to HCLE cells (500 μl total volume, 10.3 μM ShlA in the pMQ492 fraction) and incubated for 3 h followed by calcein AM staining and microscopic analysis.
Microscopic analysis
To obtain micrographs, cells on glass bottomed multiwell plates (MatTek) were imaged with a 40X objective using an Olympus IX-81 inverted microscope with an FV-1000 laser scanning confocal system (Olympus) and FluoView FV10-ASW 3.1 imaging software. For live imaging, samples in MatTek dishes were viewed with a Nikon Eclipse Ti microscope equipped with a Photometrics Cascade 1K camera and a 40X 0.30 NA objective. Metamorph software was used to obtain digital images. FIJI software was used to for image analysis [99].
Galleria mellonella infection assays. G. mellonella were infected as previously described [100], with the exception that S. marcescens was suspended in PBS with 10 μg/ml tetracycline. To enumerate S. marcescens, homogenates from individual larvae were generated using a tissue grinder (Corning Pyrex 7725) in PBS with tetracycline. Lysates were serial diluted and plated on LB agar supplemented with ampicillin (150 μg/ml), chloramphenicol (30 μg/ml), and tetracycline (10 μg/ml) to prevent unwanted microbial growth.
To determine bacterial growth in larval homogenates, larvae were homogenized at a ratio of 2 larvae in 1 ml of PBS. When 15 ml of homogenate was obtained, it was centrifuged at 11,000 x g for 10 minutes to clarify the supernatant. The supernatant was heated at 95°C for 60 minutes to kill microbes and prevent melanization, which obscures optical density readings. S. marcescens cultures (1 ml) grown overnight in LB were spun down (13,000 x g for 2 minutes) and washed with PBS and then adjusted to OD600 = 0.05 in the larval homogenate, 150 μl was added to the wells of 96 well plates and were incubated overnight at 30°C. After 20 h, CFU were determined following serial dilution as noted above.
Supporting information
S1 Fig. Effect of select bacterial species on HCLE morphology and viability.
Confocal micrographs of HCLE cells images with DIC and calcein AM viability stain after exposure to bacteria for 2 h at MOI = 200, except where noted. Yellow arrows indicate blebs extending from corneal cells. (A) HCLE cells exposed to S. marcescens strains, including wild type strain Db11 and an isogenic ΔgumB mutant strain. (B). HCLE cells exposed to various bacteria, of which only E. tarda and P. aeruginosa strain K900 induce bleb formation and cytotoxicity.
https://doi.org/10.1371/journal.ppat.1007825.s001
(PDF)
S2 Fig. S. marcescens induces bleb induction in an airway cell line and secretion of ShlA is sufficient for induction of bleb formation and cytotoxicity.
Confocal micrographs of human epithelial cell monolayers images with DIC and fluorescent calcein AM viability stain after challenge with bacteria. Yellow arrows indicate epithelial cell blebs. (A) A549 human airway epithelial cell line exposed to S. marcescens wild type K904 and ΔgumB strains (MOI = 200) for 2 h. (B) HCLE cells exposed to E. coli strain Top10 (MOI = 50, for 1 h) with a control vector, the shlBA expression plasmid, or a version of the shlBA plasmid with a transposon insertion inactivating the shlA gene. The control vector = pMQ125; pshlBA = pMQ492; pshlBA::tn = pMQ591.
https://doi.org/10.1371/journal.ppat.1007825.s002
(PDF)
S3 Fig. PCR analysis for shlA gene in ocular isolates.
All tested strains, from a variety of ocular infections (conjunctivitis, endophthalmitis, and keratitis), were positive for the shlA gene. (A) PCR was performed with degenerate primers due to the variable sequence of the shlA gene. Primer sequences were (5' to 3') gcyaacccgaayggcatcasctg for primer 4722 and yggcstrcatgcygccsags for primer 4725. The predicted amplicon is 367 base pairs. Amplicons and a size standard (SS) were separated on a 0.5% TBE PAGE gel, stained with ethidium bromide, and imaged. Strain PIC3611 was used as a positive control and the same strain with a deletion of the shlBA operon was used as a negative control. Sequence of the PIC3611 amplicon was 100% identical to shlA from several strains of S. marcescens over 267 bp. (B) DNA quality for all strains was verified by spectrophotometry and by PCR using primers for the conserved oxyR gene. Shown are amplicons for PIC3611 and the isogenic ΔshlBA mutant. This data supports that the ΔshlBA mutant is negative for the shlA amplicon because the shlA primers are specific and not because the DNA preparation was defective.
https://doi.org/10.1371/journal.ppat.1007825.s003
(PDF)
S4 Fig. ShlA-mediated cytotoxicity to HCLE cells.
Cytotoxicity was measured using Presto Blue reagent. HCLE monolayers, incubated with bacteria at MOI = 200 (A) or 10 (B) for 2 hours, were analyzed for viability relative to cells treated with detergent (Lysis) or LB medium (Mock). Vector = pMQ125; pshlBA = pMQ541; pgumB = pMQ480.
https://doi.org/10.1371/journal.ppat.1007825.s004
(PDF)
S5 Fig. Pigmentation and anaerobic growth of mutant strains.
(A) Photographs of bacterial pigmentation on an LB plate after growth at 30°C for 24 hours shows that multicopy of expression of rcsC reduces pigmentation almost as severely as mutation of gumB. (B) Photograph depicting that the rcsB mutation suppresses the gumB mutant phenotype and that this can be complemented by wild-type rcsB on a plasmid. Reduced pigmentation of the strain with wild-type rcsB on a plasmid supports the model that RcsB inhibits pigment biosynthesis. (C) Images show growth of the wild-type strain K904 and the ΔgumB mutant (and a ΔgumB rcsC double mutant) on LB agar plates grown at 30°C for 24 hours in a GAS PAK-EZ anaerobe pouch system (left panel) or at ambient oxygen levels (right). The ΔgumB mutant produced colonies of similar size to the wild type under both conditions indicating that the ΔgumB mutant does not have a significant defect for growth under low oxygen conditions.
https://doi.org/10.1371/journal.ppat.1007825.s005
(PDF)
S6 Fig. Model for regulation of shlBA.
Genetic model for regulation of S. marcescens pigment and cytolysin operons. Red bars indicate negative regulation and black arrows indicate activation. Our model predicts that in response to envelope stress, GumB acts as part of the Rcs signal transduction system to modify activity of the RcsB response regulator. In addition to directly inhibiting shlBA expression, RcsB also inhibits expression of the flhDC operon, which codes for a positive transcriptional regulator of shlBA. Expression of the shlBA operon leads to secretion of ShlA. Surface associated and surface-released ShlA forms pores in mammalian cells leading to blebbing and finally necroptosis-associated cell death.
https://doi.org/10.1371/journal.ppat.1007825.s006
(PDF)
S7 Fig. Inhibition of bleb formation by necroptosis inhibitor GWX806742X.
The graph represents data from two experiments with cell counts from n≥6 fields of view (n>80 cells per treatment group). HCLE cells treated with GWX 806742X were challenged with wild-type S. marcescens strain K904 at MOI = 50 and after 2 h cells were imaged and bleb frequency was measured. Mean and SD are shown. ANOVA with Tukey's post-test was used and significance is shown by asterisks. * p<0.05, ** p<0.01, **** p<0.0001. Data suggests specific inhibition of necroptosis mediator MLKL reduces bleb formation.
https://doi.org/10.1371/journal.ppat.1007825.s007
(PDF)
S8 Fig. Role of ShlA in the ΔgumB RAW cell proliferation phenotype.
Uptake and proliferation of S. marcescens K904 wild type and ΔgumB mutant strains with the vector (pMQ132 and shlBA expression plasmid pMQ541) in RAW macrophage cells (n = 4), mean and standard deviations are shown. Asterisks indicate significant difference by 2-way ANOVA with Tukey’s post-test (* = p<0.05, **** = p<0.0001).
https://doi.org/10.1371/journal.ppat.1007825.s008
(PDF)
S1 Movie. Live cell imaging of HCLE cells exposed to mock conditions.
Images of HCLE cells over three h; viewed at 400X.
https://doi.org/10.1371/journal.ppat.1007825.s009
(AVI)
S2 Movie. Live cell imaging of HCLE cells exposed to S. marcescens wild-type strain K904.
Images of HCLE cells over three h exposed to S. marcescens strain K904 at MOI = 50; viewed at 400X.
https://doi.org/10.1371/journal.ppat.1007825.s010
(AVI)
S3 Movie. Live cell imaging of HCLE cells exposed to the S. marcescens ΔgumB mutant strain.
Images of HCLE cells over three h exposed to S. marcescens ΔgumB strain at MOI = 50; viewed at 400X.
https://doi.org/10.1371/journal.ppat.1007825.s011
(AVI)
S2 Table. DNA oligonucleotide primers used in this study.
https://doi.org/10.1371/journal.ppat.1007825.s013
(PDF)
Acknowledgments
The authors would like to thank Kate Davoli for help with cell culture, Jessica Steele for microscopy help, Angelina Passerini and Kristin M. Hunt for technical help, Jeffrey Melvin for critical reading of the manuscript, and Regis P. Kowalski at the University of Pittsburgh, and Carol Kim at the University of Maine for the generous gift of strains.
References
- 1. Nisan I, Wolff C, Hanski E, Rosenshine I. Interaction of enteropathogenic Escherichia coli with host epithelial cells. Folia Microbiol (Praha). 1998;43(3):247–52. pmid:9717251.
- 2. Goosney DL, Gruenheid S, Finlay BB. Gut feelings: enteropathogenic E. coli (EPEC) interactions with the host. Annu Rev Cell Dev Biol. 2000;16:173–89. pmid:11031234.
- 3. Hendricks MR, Bomberger JM. Who's really in control: microbial regulation of protein trafficking in the epithelium. Am J Physiol Cell Physiol. 2014;306(3):C187–97. pmid:24133062; PubMed Central PMCID: PMCPMC3919996.
- 4. Charras GT. A short history of blebbing. J Microsc. 2008;231(3):466–78. pmid:18755002.
- 5. Barros LF, Kanaseki T, Sabirov R, Morishima S, Castro J, Bittner CX, et al. Apoptotic and necrotic blebs in epithelial cells display similar neck diameters but different kinase dependency. Cell Death Differ. 2003;10(6):687–97. pmid:12761577.
- 6. Angus AA, Evans DJ, Barbieri JT, Fleiszig SM. The ADP-ribosylation domain of Pseudomonas aeruginosa ExoS is required for membrane bleb niche formation and bacterial survival within epithelial cells. Infect Immun. 2010;78(11):4500–10. pmid:20732998; PubMed Central PMCID: PMCPMC2976358.
- 7. Hritonenko V, Mun JJ, Tam C, Simon NC, Barbieri JT, Evans DJ, et al. Adenylate cyclase activity of Pseudomonas aeruginosa ExoY can mediate bleb-niche formation in epithelial cells and contributes to virulence. Microb Pathog. 2011;51(5):305–12. PubMed Central PMCID: PMCPMC3213052. pmid:21843628
- 8. Broberg CA, Zhang L, Gonzalez H, Laskowski-Arce MA, Orth K. A Vibrio effector protein is an inositol phosphatase and disrupts host cell membrane integrity. Science. 2010;329(5999):1660–2. pmid:20724587.
- 9. Manjo G, Joris I. Apoptosis, oncosis, and necrosis: and overview of cell death. Am J Pathol. 1995;146(1):3–15. pmid:7856735
- 10. Fink SL, Cookson AL. Apoptosis, pyroptosis, and necrosis: mechanistic description of dead and dying eukryotic cells. Infect Immun. 2005;73(4):1907–16. pmid:15784530
- 11. Babiychuk EB, Monastyrskaya K, Potez S, Draeger A. Blebbing confers resistance against cell lysis. Cell Death Differ. 2011;18(1):80–9. pmid:20596076; PubMed Central PMCID: PMCPMC3131879.
- 12. Hume EB, Willcox MD. Emergence of Serratia marcescens as an ocular surface pathogen. Arch Soc Esp Oftalmol. 2004;79(10):475–7. pmid:15523567.
- 13. Lockhart SR, Abramson MA, Beekmann SE, Gallagher G, Riedel S, Diekema DJ, et al. Antimicrobial resistance among Gram-negative bacilli causing infections in intensive care unit patients in the United States between 1993 and 2004. J Clin Microbiol. 2007;45(10):3352–9. pmid:17715376.
- 14. Mah-Sadorra JH, Najjar DM, Rapuano CJ, Laibson PR, Cohen EJ. Serratia corneal ulcers: a retrospective clinical study. Cornea. 2005;24(7):793–800. pmid:16160494.
- 15. Mahlen SD. Serratia infections: from military experiments to current practice. Clin Microbiol Rev. 2011;24(4):755–91. Epub 2011/10/07. 24/4/755 [pii] pmid:21976608; PubMed Central PMCID: PMC3194826.
- 16. Merkier AK, Rodriguez MC, Togneri A, Brengi S, Osuna C, Pichel M, et al. Outbreak of a cluster with epidemic behavior due to Serratia marcescens after colistin administration in a hospital setting. J Clin Microbiol. 2013;51(7):2295–302. Epub 2013/05/24. JCM.03280-12 [pii] pmid:23698525; PubMed Central PMCID: PMC3697717.
- 17. Zhou R, Zhang R, Sun Y, Platt S, Szczotka-Flynn L, Pearlman E. Innate immune regulation of Serratia marcescens-induced corneal inflammation and infection. Invest Ophthalmol Vis Sci. 2012;53(11):7382–8. Epub 2012/10/04. iovs.12-10238 [pii] pmid:23033384; PubMed Central PMCID: PMC3481604.
- 18. Keay L, Edwards K, Naduvilath T, Taylor HR, Snibson GR, Forde K, et al. Microbial keratitis predisposing factors and morbidity. Ophthalmology. 2006;113(1):109–16. pmid:16360210.
- 19. Evans DJ, Fleiszig SM. Why does the healthy cornea resist Pseudomonas aeruginosa infection? Am J Ophthalmol. 2013;155(6):961–70 e2. pmid:23601656; PubMed Central PMCID: PMCPMC3718454.
- 20. Li P, Kwok AH, Jiang J, Ran T, Xu D, Wang W, et al. Comparative genome analyses of Serratia marcescens FS14 reveals its high antagonistic potential. PLoS One. 2015;10(4):e0123061. pmid:25856195; PubMed Central PMCID: PMCPMC4391916.
- 21. Antic I, Brothers KM, Stolzer M, Lai H, Powell E, Eutsey R, et al. Gene Acquisition by a Distinct Phyletic Group within Streptococcus pneumoniae Promotes Adhesion to the Ocular Epithelium. mSphere. 2017;2(5). pmid:29085912; PubMed Central PMCID: PMCPMC5656748.
- 22. Kowalski RP, Romanowski EG, Mah FS, Shanks RM, Gordon YJ. Topical levofloxacin 1.5% overcomes in vitro resistance in rabbit keratitis models. Acta Ophthalmol. 2010;88(4):e120–5. Epub 2010/05/12. AOS1897 [pii] pmid:20456251.
- 23. Holloway BW. Genetic recombination in Pseudomonas aeruginosa. J Gen Microbiol. 1955;13(3):572–81. pmid:13278508.
- 24. Hertle R. The family of Serratia type pore forming toxins. Curr Protein Pept Sci. 2005;6(4):313–25. pmid:16101433.
- 25. Stella NA, Brothers KM, Callaghan JD, Passerini AM, Sigindere C, Hill PJ, et al. An IgaA/UmoB-family protein from Serratia marcescens regulates motility, capsular polysaccharide, and secondary metabolite production. Appl Environ Microbiol. 2018;84(6):pii: e02575-17. pmid:29305504.
- 26. Dominguez-Bernal G, Pucciarelli MG, Ramos-Morales F, Garcia-Quintanilla M, Cano DA, Casadesus J, et al. Repression of the RcsC-YojN-RcsB phosphorelay by the IgaA protein is a requisite for Salmonella virulence. Mol Microbiol. 2004;53(5):1437–49. Epub 2004/09/25. [pii]. pmid:15387821.
- 27. Welch RA. Pore-forming cytolysins of gram-negative bacteria. Mol Microbiol. 1991;5(3):521–8. pmid:2046545.
- 28. Flyg C, Kenne K, Boman HG. Insect pathogenic properties of Serratia marcescens: phage-resistant mutants with a decreased resistance to Cecropia immunity and a decreased virulence to Drosophila. J Gen Microbiol. 1980;120(1):173–81. pmid:7012273.
- 29. Cano DA, Dominguez-Bernal G, Tierrez A, Garcia-Del Portillo F, Casadesus J. Regulation of capsule synthesis and cell motility in Salmonella enterica by the essential gene igaA. Genetics. 2002;162(4):1513–23. Epub 2003/01/14. pmid:12524328; PubMed Central PMCID: PMC1462382.
- 30. Morgenstein RM, Rather PN. Role of the Umo proteins and the Rcs phosphorelay in the swarming motility of the wild type and an O-antigen (waaL) mutant of Proteus mirabilis. J Bacteriol. 2012;194(3):669–76. Epub 2011/12/06. JB.06047-11 [pii] pmid:22139504; PubMed Central PMCID: PMC3264082.
- 31. Cho SH, Szewczyk J, Pesavento C, Zietek M, Banzhaf M, Roszczenko P, et al. Detecting envelope stress by monitoring beta-barrel assembly. Cell. 2014;159(7):1652–64. Epub 2014/12/20. S0092-8674(14)01522-0 [pii] pmid:25525882.
- 32. Wall E, Majdalani N, Gottesman S. The Complex Rcs Regulatory Cascade. Annu Rev Microbiol. 2018. pmid:29897834.
- 33. Jolly AL, Takawira D, Oke OO, Whiteside SA, Chang SW, Wen ER, et al. Pseudomonas aeruginosa-induced bleb-niche formation in epithelial cells is independent of actinomyosin contraction and enhanced by loss of cystic fibrosis transmembrane-conductance regulator osmoregulatory function. MBio. 2015;6(2):e02533. pmid:25714715; PubMed Central PMCID: PMCPMC4358002.
- 34. LaRocca TJ, Stivison EA, Mal-Sarkar T, Hooven TA, Hod EA, Spitalnik SL, et al. CD59 signaling and membrane pores drive Syk-dependent erythrocyte necroptosis. Cell Death Dis. 2015;6:e1773. pmid:26018734; PubMed Central PMCID: PMCPMC4669712.
- 35. Braun V, Neuss B, Ruan Y, Schiebel E, Schoffler H, Jander G. Identification of the Serratia marcescens hemolysin determinant by cloning into Escherichia coli. J Bacteriol. 1987;169(5):2113–20. pmid:2437098; PubMed Central PMCID: PMCPMC212107.
- 36. Burns FR, Gray RD, Wells JT, Paterson CA. The effect of a synthetic metalloproteinase inhibitor on corneal ulceration in alkali burns and Pseudomonas keratitis. Matrix Suppl. 1992;1:317–8. pmid:1480049.
- 37. Ratner AJ, Hippe KR, Aguilar JL, Bender MH, Nelson AL, Weiser JN. Epithelial cells are sensitive detectors of bacterial pore-forming toxins. J Biol Chem. 2006;281(18):12994–8. pmid:16520379; PubMed Central PMCID: PMCPMC1586115.
- 38. Schonherr R, Hilger M, Broer S, Benz R, Braun V. Interaction of Serratia marcescens hemolysin (ShlA) with artificial and erythrocyte membranes. Demonstration of the formation of aqueous multistate channels. Eur J Biochem. 1994;223(2):655–63. pmid:8055936.
- 39. Gonzalez-Juarbe N, Gilley RP, Hinojosa CA, Bradley KM, Kamei A, Gao G, et al. Pore-Forming Toxins Induce Macrophage Necroptosis during Acute Bacterial Pneumonia. PLoS Pathog. 2015;11(12):e1005337. pmid:26659062; PubMed Central PMCID: PMCPMC4676650.
- 40. Lusthaus M, Mazkereth N, Donin N, Fishelson Z. Receptor-Interacting Protein Kinases 1 and 3, and Mixed Lineage Kinase Domain-Like Protein Are Activated by Sublytic Complement and Participate in Complement-Dependent Cytotoxicity. Front Immunol. 2018;9:306. pmid:29527209; PubMed Central PMCID: PMCPMC5829068.
- 41. Gonzalez-Juarbe N, Mares CA, Hinojosa CA, Medina JL, Cantwell A, Dube PH, et al. Requirement for Serratia marcescens cytolysin in a murine model of hemorrhagic pneumonia. Infect Immun. 2015;83(2):614–24. pmid:25422267; PubMed Central PMCID: PMCPMC4294263.
- 42. Lee KZ, Lestradet M, Socha C, Schirmeier S, Schmitz A, Spenle C, et al. Enterocyte Purge and Rapid Recovery Is a Resilience Reaction of the Gut Epithelium to Pore-Forming Toxin Attack. Cell Host Microbe. 2016;20(6):716–30. pmid:27889464.
- 43. Lin CS, Horng JT, Yang CH, Tsai YH, Su LH, Wei CF, et al. RssAB-FlhDC-ShlBA as a major pathogenesis pathway in Serratia marcescens. Infect Immun. 2010;78(11):4870–81. Epub 2010/08/18. IAI.00661-10 [pii] pmid:20713626; PubMed Central PMCID: PMC2976324.
- 44. Ahanotu EN, Ahearn DG. Association of Pseudomonas aeruginosa and Serratia marcescens with extended-wear soft contact lenses in asymptomatic patients. Clao J. 2002;28(3):157–9. pmid:12144237.
- 45. Holden BA, La Hood D, Grant T, Newton-Howes J, Baleriola-Lucas C, Willcox MD, et al. Gram-negative bacteria can induce contact lens related acute red eye (CLARE) responses. Clao J. 1996;22(1):47–52. pmid:8835069.
- 46. Yung MS, Boost M, Cho P, Yap M. Microbial contamination of contact lenses and lens care accessories of soft contact lens wearers (university students) in Hong Kong. Ophthalmic Physiol Opt. 2007;27(1):11–21. Epub 2007/01/24. OPO427 [pii] pmid:17239186.
- 47. Bourcier T, Thomas F, Borderie V, Chaumeil C, Laroche L. Bacterial keratitis: predisposing factors, clinical and microbiological review of 300 cases. Br J Ophthalmol. 2003;87:834–8. pmid:12812878
- 48. Heimer SR, Evans DJ, Stern ME, Barbieri JT, Yahr T, Fleiszig SM. Pseudomonas aeruginosa utilizes the type III secreted toxin ExoS to avoid acidified compartments within epithelial cells. PLoS One. 2013;8(9):e73111. pmid:24058462; PubMed Central PMCID: PMCPMC3776860.
- 49. Elsen S, Huber P, Bouillot S, Coute Y, Fournier P, Dubois Y, et al. A type III secretion negative clinical strain of Pseudomonas aeruginosa employs a two-partner secreted exolysin to induce hemorrhagic pneumonia. Cell Host Microbe. 2014;15(2):164–76. pmid:24528863.
- 50. Reboud E, Bouillot S, Patot S, Beganton B, Attree I, Huber P. Pseudomonas aeruginosa ExlA and Serratia marcescens ShlA trigger cadherin cleavage by promoting calcium influx and ADAM10 activation. PLoS Pathog. 2017;13(8):e1006579. pmid:28832671; PubMed Central PMCID: PMCPMC5584975.
- 51. Hirono I, Tange N, Aoki T. Iron-regulated haemolysin gene from Edwardsiella tarda. Mol Microbiol. 1997;24(4):851–6. pmid:9194711.
- 52. Hertle R. Serratia type pore forming toxins. Curr Protein Pept Sci. 2000;1(1):75–89. pmid:12369921.
- 53. Bayer AS, Ramos MD, Menzies BE, Yeaman MR, Shen AJ, Cheung AL. Hyperproduction of alpha-toxin by Staphylococcus aureus results in paradoxically reduced virulence in experimental endocarditis: a host defense role for platelet microbicidal proteins. Infect Immun. 1997;65(11):4652–60. pmid:9353046; PubMed Central PMCID: PMCPMC175667.
- 54. Keyel PA, Loultcheva L, Roth R, Salter RD, Watkins SC, Yokoyama WM, et al. Streptolysin O clearance through sequestration into blebs that bud passively from the plasma membrane. J Cell Sci. 2011;124(Pt 14):2414–23. pmid:21693578; PubMed Central PMCID: PMCPMC3124372.
- 55. Betancourt-Sanchez M, Navarro-Garcia F. Pet secretion, internalization and induction of cell death during infection of epithelial cells by enteroaggregative Escherichia coli. Microbiology. 2009;155(Pt 9):2895–906. pmid:19542001.
- 56. Navarro-Garcia F, Canizalez-Roman A, Sui BQ, Nataro JP, Azamar Y. The serine protease motif of EspC from enteropathogenic Escherichia coli produces epithelial damage by a mechanism different from that of Pet toxin from enteroaggregative E. coli. Infect Immun. 2004;72(6):3609–21. pmid:15155671; PubMed Central PMCID: PMCPMC415714.
- 57. Navarro-Garcia F, Sears C, Eslava C, Cravioto A, Nataro JP. Cytoskeletal effects induced by pet, the serine protease enterotoxin of enteroaggregative Escherichia coli. Infect Immun. 1999;67(5):2184–92. pmid:10225873; PubMed Central PMCID: PMCPMC115956.
- 58. Di Venanzio G, Stepanenko TM, Garcia Vescovi E. Serratia marcescens ShlA pore-forming toxin is responsible for early induction of autophagy in host cells and is transcriptionally regulated by RcsB. Infect Immun. 2014;82(9):3542–54. pmid:24914224; PubMed Central PMCID: PMCPMC4187834.
- 59. Brothers KM, Stella NA, Romanowski EG, Kowalski RP, Shanks RM. EepR mediates secreted protein production, desiccation survival, and proliferation in a corneal infection model. Infect Immun. 2015. pmid:26324535.
- 60. Ishii K, Adachi T, Hamamoto H, Sekimizu K. Serratia marcescens suppresses host cellular immunity via the production of an adhesion-inhibitory factor against immunosurveillance cells. J Biol Chem. 2014;289(9):5876–88. Epub 2014/01/09. M113.544536 [pii] pmid:24398686; PubMed Central PMCID: PMC3937657.
- 61. Anderson MT, Mitchell LA, Mobley HLT. Cysteine Biosynthesis Controls Serratia marcescens Phospholipase Activity. J Bacteriol. 2017;199(16). pmid:28559296; PubMed Central PMCID: PMCPMC5527384.
- 62. Wei CF, Tsai YH, Tsai SH, Lin CS, Chang CJ, Lu CC, et al. Cross-talk between bacterial two-component systems drives stepwise regulation of flagellar biosynthesis in swarming development. Biochem Biophys Res Commun. 2017;489(1):70–5. pmid:28522292.
- 63. Shanks RM, Stella NA, Lahr RM, Wang S, Veverka TI, Kowalski RP, et al. Serratamolide is a Hemolytic Factor Produced by Serratia marcescens. PLoS One. 2012;7(5):e36398. Epub 2012/05/23. [pii]. pmid:22615766; PubMed Central PMCID: PMC3353980.
- 64. Perez-Tomas R, Vinas M. New insights on the antitumoral properties of prodiginines. Curr Med Chem. 2010;17(21):2222–31. Epub 2010/05/13. BSP/CMC/E-Pub/ 136 [pii]. pmid:20459382.
- 65. McMahon KJ, Castelli ME, Garcia Vescovi E, Feldman MF. Biogenesis of outer membrane vesicles in Serratia marcescens is thermoregulated and can be induced by activation of the Rcs phosphorelay system. J Bacteriol. 2012;194(12):3241–9. Epub 2012/04/12. JB.00016-12 [pii] pmid:22493021; PubMed Central PMCID: PMC3370869.
- 66. Anderson MT, Mitchell LA, Zhao L, Mobley HLT. Capsule Production and Glucose Metabolism Dictate Fitness during Serratia marcescens Bacteremia. MBio. 2017;8(3):e00740–17. pmid:28536292; PubMed Central PMCID: PMCPMC5442460.
- 67. Hertle R, Schwarz H. Serratia marcescens internalization and replication in human bladder epithelial cells. BMC Infect Dis. 2004;4:16. pmid:15189566.
- 68. Fedrigo GV, Campoy EM, Di Venanzio G, Colombo MI, Garcia Vescovi E. Serratia marcescens is able to survive and proliferate in autophagic-like vacuoles inside non-phagocytic cells. PLoS One. 2011;6(8):e24054. pmid:21901159; PubMed Central PMCID: PMCPMC3162031.
- 69. Di Venanzio G, Lazzaro M, Morales ES, Krapf D, Garcia Vescovi E. A pore-forming toxin enables Serratia a nonlytic egress from host cells. Cell Microbiol. 2017;19(2). pmid:27532510.
- 70. Van Tyne D, Ciolino JB, Wang J, Durand ML, Gilmore MS. Novel Phagocytosis-Resistant Extended-Spectrum beta-Lactamase-Producing Escherichia coli From Keratitis. JAMA Ophthalmol. 2016;134(11):1306–9. pmid:27631542; PubMed Central PMCID: PMCPMC5106311.
- 71. Miskinyte M, Sousa A, Ramiro RS, de Sousa JA, Kotlinowski J, Caramalho I, et al. The genetic basis of Escherichia coli pathoadaptation to macrophages. PLoS Pathog. 2013;9(12):e1003802. Epub 2013/12/19. [pii]. pmid:24348252; PubMed Central PMCID: PMC3861542.
- 72. Cano DA, Martinez-Moya M, Pucciarelli MG, Groisman EA, Casadesus J, Garcia-Del Portillo F. Salmonella enterica serovar Typhimurium response involved in attenuation of pathogen intracellular proliferation. Infect Immun. 2001;69(10):6463–74. Epub 2001/09/13. pmid:11553591; PubMed Central PMCID: PMC98782.
- 73. Evans DJ, McNamara NA, Fleiszig SM. Life at the front: dissecting bacterial-host interactions at the ocular surface. Ocul Surf. 2007;5(3):213–27. pmid:17660895.
- 74. Hoarau G, Mukherjee PK, Gower-Rousseu C, Hager C, Chandra J, Retueto MA, et al. Bacteriome and Mycobiome Interactions Underscore Microbial Dysbiosis in Familial Crohn's Disease. Mbio. 2016;20(7):e01250–16.
- 75. Gonzalez-Juarbe N, Bradley KM, Shenoy AT, Gilley RP, Reyes LF, Hinojosa CA, et al. Pore-forming toxin-mediated ion dysregulation leads to death receptor-independent necroptosis of lung epithelial cells during bacterial pneumonia. Cell Death Differ. 2017; 24(5)917–928. pmid:28387756.
- 76. Kalivoda EJ, Stella NA, Aston MA, Fender JE, Thompson PP, Kowalski RP, et al. Cyclic AMP negatively regulates prodigiosin production by Serratia marcescens. Res Microbiol. 2010;161(2):158–67. Epub 2010/01/05. S0923-2508(09)00256-3 [pii] pmid:20045458; PubMed Central PMCID: PMC2846241.
- 77. Bertani G. Studies on lysogenesis. I. The mode of phage liberation by lysogenic Escherichia coli. J Bacteriol. 1951;62(3):293–300. Epub 1951/09/01. pmid:14888646; PubMed Central PMCID: PMC386127.
- 78.
Adams MH. Bacteriophages. New York, NY: Interscience Publishers Inc.; 1959.
- 79. Miller VL, Mekalanos JJ. A novel suicide vector and its use in construction of insertion mutations: osmoregulation of outer membrane proteins and virulence determinants in Vibrio cholerae requires toxR. J Bacteriol. 1988;170:2575. pmid:2836362
- 80. Caiazza NC, Lies DP, Newman DK. Phototrophic Fe(II) oxidation promotes organic carbon acquisition by Rhodobacter capsulatus SB1003. Appl Environ Microbiol. 2007;73(19):6150–8. pmid:17693559; PubMed Central PMCID: PMCPMC2074999.
- 81.
Burke D. DD, Stearns T. Methods In Yeast Genetics, A Cold Spring Harbor Laboratory Course Manual. Plainview, NY: Cold Harbor laboratory Press; 2000.
- 82. Gipson IK, Spurr-Michaud S, Argüeso P, Tisdale A, Ng TF, Russo CL. Mucin gene expression in immortalized human conreal-limbal and conjunc tival epithelial cell lines. Invest Ophthalmol Vis Sci. 2003;44(6):2496–506. pmid:12766048
- 83. Shanks RM, Davra VR, Romanowski EG, Brothers KM, Stella NA, Godboley D, et al. An Eye to a Kill: Using Predatory Bacteria to Control Gram-Negative Pathogens Associated with Ocular Infections. PLoS One. 2013;8(6):e66723. pmid:23824756; PubMed Central PMCID: PMCPMC3688930.
- 84. Chen SY, Hayashida Y, Chen MY, Xie HT, Tseng SC. A new isolation method of human limbal progenitor cells by maintaining close association with their niche cells. Tissue Eng Part C Methods. 2011;17(5):537–48. pmid:21175372; PubMed Central PMCID: PMCPMC3129703.
- 85. Horzempa J, Carlson PE Jr., O'Dee DM, Shanks RM, Nau GJ. Global transcriptional response to mammalian temperature provides new insight into Francisella tularensis pathogenesis. BMC Microbiol. 2008;8(1):172. pmid:18842136.
- 86. Brett PJ, Burtnick MN, Su H, Nair V, Gherardini FC. iNOS activity is critical for the clearance of Burkholderia mallei from infected RAW 264.7 murine macrophages. Cell Microbiol. 2008;10(2):487–98. pmid:17970762; PubMed Central PMCID: PMCPMC2228653.
- 87. Brothers KM, Stella NA, Hunt KM, Romanowski EG, Liu X, Klarlund JK, et al. Putting on the brakes: Bacterial impediment of wound healing. Sci Rep. 2015;5:14003. Epub 2015/09/15. pmid:26365869; PubMed Central PMCID: PMCPMC4650533.
- 88. Shanks RM, Stella NA, Kalivoda EJ, Doe MR, O'Dee DM, Lathrop KL, et al. A Serratia marcescens OxyR homolog mediates surface attachment and biofilm formation. J Bacteriol. 2007;189(20):7262–72. Epub 2007/08/07. JB.00859-07 [pii] pmid:17675374; PubMed Central PMCID: PMC2168423.
- 89. Chiang SL, Rubin EJ. Construction of a mariner-based transposon for epitope-tagging and genomic targeting. Gene. 2002;296(1–2):179–85. Epub 2002/10/18. S0378111902008569 [pii]. pmid:12383515.
- 90. O'Toole GA, Pratt LA, Watnick PI, Newman DK, Weaver VB, Kolter R. Genetic approaches to study of biofilms. Methods Enzymol. 1999;310:91–109. Epub 1999/11/05. pmid:10547784.
- 91. Stella NA, Lahr RM, Brothers KM, Kalivoda EJ, Hunt KM, Kwak DH, et al. Serratia marcescens cyclic AMP-receptor protein controls transcription of EepR, a novel regulator of antimicrobial secondary metabolites. J Bacteriol. 2015;197(15):2468–78. Epub 2015/04/22. JB.00136-15 [pii] pmid:25897029.
- 92. Shanks RM, Kadouri DE, MacEachran DP, O'Toole GA. New yeast recombineering tools for bacteria. Plasmid. 2009;62(2):88–97. Epub 2009/05/30. pmid:19477196; PubMed Central PMCID: PMC2737453.
- 93. Shanks RM, Stella NA, Arena KE, Fender JE. Mutation of crp mediates Serratia marcescens serralysin and global secreted protein production. Res Microbiol. 2013;164(1):38–45. Epub 2012/10/18. S0923-2508(12)00151-9 [pii] pmid:23072819; PubMed Central PMCID: PMC3534799.
- 94. Kalivoda EJ, Stella NA, O'Dee DM, Nau GJ, Shanks RM. The cyclic AMP-dependent catabolite repression system of Serratia marcescens mediates biofilm formation through regulation of type 1 fimbriae. Appl Environ Microbiol. 2008;74(11):3461–70. Epub 2008/04/22. AEM.02733-07 [pii] pmid:18424546; PubMed Central PMCID: PMC2423026.
- 95. Bachmann BJ. Pedigrees of some mutant strains of Escherichia coli K-12. Bacteriol Rev. 1972;36(4):525–57. pmid:4568763; PubMed Central PMCID: PMCPMC408331.
- 96. Datsenko KA, Wanner BL. One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc Natl Acad Sci U S A. 2000;97(12):6640–5. pmid:10829079.
- 97. Rossi MS, Paquelin A, Ghigo JM, Wandersman C. Haemophore-mediated signal transduction across the bacterial cell envelope in Serratia marcescens: the inducer and the transported substrate are different molecules. Mol Microbiol. 2003;48(6):1467–80. pmid:12791131.
- 98. Mukherjee S, Brothers KM, Shanks RMQ, Kadouri DE. Visualizing Bdellovibrio bacteriovorus by Using the tdTomato Fluorescent Protein. Appl Environ Microbiol. 2015;82(6):1653–61. pmid:26712556; PubMed Central PMCID: PMCPMC4784026.
- 99. Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25 years of image analysis. Nat Methods. 2012;9(7):671–5. pmid:22930834; PubMed Central PMCID: PMCPMC5554542.
- 100. O'Hara JA, Ambe LA, Casella LG, Townsend BM, Pelletier MR, Ernst RK, et al. Activities of vancomycin-containing regimens against colistin-resistant Acinetobacter baumannii clinical strains. Antimicrob Agents Chemother. 2013;57(5):2103–8. Epub 2013/02/21. AAC.02501-12 [pii] pmid:23422916; PubMed Central PMCID: PMC3632926.