Complex c-di-GMP signaling networks mediate transition between virulence properties and biofilm formation in Salmonella enterica serovar Typhimurium

Abstract

Upon Salmonella enterica serovar Typhimurium infection of the gut, an early line of defense is the gastrointestinal epithelium which senses the pathogen and intrusion along the epithelial barrier is one of the first events towards disease. Recently, we showed that high intracellular amounts of the secondary messenger c-di-GMP in S. typhimurium inhibited invasion and abolished induction of a pro-inflammatory immune response in the colonic epithelial cell line HT-29 suggesting regulation of transition between biofilm formation and virulence by c-di-GMP in the intestine. Here we show that highly complex c-di-GMP signaling networks consisting of distinct groups of c-di-GMP synthesizing and degrading proteins modulate the virulence phenotypes invasion, IL-8 production and in vivo colonization in the streptomycin-treated mouse model implying a spatial and timely modulation of virulence properties in S. typhimurium by c-di-GMP signaling. Inhibition of the invasion and IL-8 induction phenotype by c-di-GMP (partially) requires the major biofilm activator CsgD and/or BcsA, the synthase for the extracellular matrix component cellulose. Inhibition of the invasion phenotype is associated with inhibition of secretion of the type three secretion system effector protein SipA, which requires c-di-GMP metabolizing proteins, but not their catalytic activity. Our findings show that c-di-GMP signaling is at least equally important in the regulation of Salmonella-host interaction as in the regulation of biofilm formation at ambient temperature.

Figure 1. Individual GG(D/E)EF/EAL domain proteins affect S. typhimurium UMR1 invasion into HT-29 gastrointestinal epithelial cells.

Invasion of (A) GG(D/E)EF-domain protein mutants, of (B) EAL-domain protein mutants and (C) GG(D/E)EF-EAL-domain protein mutants relative to the wild type S. typhimurium UMR1. Strains were grown under invasion inducing conditions (standing culture, LB+0.3M NaCl) until O.D.600 0.6. HT-29 cells were infected for 1 h with 10⁷ CFU and subsequently incubated for an additional hour with 100 µg/ml gentamicin to kill extracellular bacteria. ΔompR, negative control for invasion. Invasion % is defined as (invasion value of mutant/invasion value of wild type)*100, whereby for each strain invasion is calculated as (CFU recovered inside epithelial cells/CFU at time of inoculation). Bars show mean ± standard deviation from at least five independent biological experiments performed in two technical replicates. Statistical significance is indicated by *P<0.05; **P<0.01; ***P<0.001 as compared with wild type S. typhimurium UMR1 (WT).

Figure 2. Complementation of the invasion phenotype of GG(D/E)EF/EAL protein mutants.

(A) Complementation of the STM4551 mutant with STM4551 and STM4551_E267A demonstrates requirement of the di-guanylate cyclase activity of STM4551 for complementation of the invasion phenotype. The enhanced invasion rate exhibited by the STM4551 mutant was significantly reduced by complementation with wild type STM4551 (p4551+), but not with the non-functional mutant STM4551_E267A (p4551m) with the GGEEF motif altered to GGAEF. WT = wild type S. typhimurium UMR1; VC = vector control pBAD30. (B) Complementation of the STM3611 mutant with STM3611 and STM3611_K179A demonstrates requirement of the c-di-GMP phosphodiesterase activity of STM3611 for complementation of the invasion phenotype. The reduced invasion rate of the STM3611 mutant was significantly enhanced by complementation with wild type STM3611 (p3611+), but not with the non-functional mutant STM3611_K179A (p3611m). WT = wild type S. typhimurium UMR1; VC = vector control pBAD30. (C) The reduced invasion rate of EAL domain protein mutants of S. typhimurium was restored to wild type level by complementation with the EAL-domain only phosphodiesterase STM3611 in plasmid pLAFR3 (p3611+). No complementation was achieved in the STM0468 mutant. WT = wild type S. typhimurium UMR1. VC = vector control pLAFR3. Experimental conditions as in Figure 1. Bars show mean ± standard deviation from at least three independent biological experiments performed in two technical replicates. Statistical significance is indicated by **P<0.01; ***P<0.001 as compared with the corresponding mutant vector control.

Figure 3. GG(D/E)EF/EAL domain proteins affect secretion of the SipA effector protein.

(A) Mutants of the diguanylate cyclases STM1987 and STM4551 show enhanced secretion of a SipA-β-lactamase fusion protein, while secretion was diminished in mutants of the phosphodiesterases STM3611 and STM4264. (B) Enhanced secretion of the SipA-β-lactamase fusion protein in the STM4551 mutant is restored by expression of STM4551 from plasmid pBAD30 and by expression of the catalytically inactive STM4551_E267A mutant. (C) Diminished secretion of the SipA-β-lactamase fusion protein in the STM3611 mutant is restored by expression of STM3611 from plasmid pBAD30 and by expression of the catalytically inactive STM3611_K179A mutant. Positive control = Δ4551-VC. SipA-β-lactamase indicates chromosomal fusion in respective strains. Detection of the SipA-β-lactamase fusion protein by western blot analysis using an anti-β-lactamase antibody. Strain S. typhimurium UMR1 with pBAD30 expressing β-lactamase in the periplasm (WT-VC) served as β-lactamase secretion control. WT = wild type S. typhimurium UMR1; ΔompR, negative control; VC = vector control pBAD30.

Figure 4. Effect of double mutants of GG(D/E)EF or EAL domain proteins on invasion of S. typhimurium into HT-29 cells.

(A) Invasion is not significantly enhanced in the double mutant of the GG(D/E)EF domain proteins STM4551 and STM1987 as compared to the single mutants. (B) Invasion of the double mutant of the EAL proteins STM3611 and STM4264 into HT-29 cells is significantly decreased as compared to the single mutants. Experimental conditions as in Figure 1. WT = wild type S. typhimurium UMR1. Bars show mean ± standard deviation from at least 5 independent biological experiments performed in two technical replicates. **P<0.01 as compared with the corresponding single mutant.

Figure 5. Corresponding di-guanylate cylcases and phosphodiesterases in the regulation of the invasion phenotype.

(A) The decreased invasion phenotype of the STM3611 mutant is partially restored in the STM3611/STM1987 double mutants indicating that STM3611 has a minor effect on the degradation of the c-di-GMP produced by STM1987 that affects invasion. Invasion restauration occurs also in the STM3611/STM4551 double mutant, but a significant effect was not achieved. (B) The decreased invasion phenotype of the STM4264 mutant is restored to wild type levels in the STM4264/STM1987 double mutant indicating that STM4264 is the major phosphodiesterase degrading the c-di-GMP produced by STM1987. STM4551 deletion has a signficant, but less pronounced effect on the restauration of invasion in the STM4264 mutant. (A and B) Invasion of the STM3611 and STM4264 mutant is restored by the deletion of STM1987 and STM4551 indicating that STM1987 and STM4551 are the two major di-guanylate cyclases producing the c-di-GMP that inhibits invasion. Experimental conditions as in Figure 1. WT = wild type S. typhimurium UMR1. Bars show mean ± standard deviation from at least 4 independent biological experiments performed on two technical replicates. Statistical significance is indicated by *p<0.05, **p<0.01, ***p<0.001 as compared with STM4264 (A) and STM3611 mutant (B).

Figure 6. The EAL proteins STM3611 and STM 4264 affect invasion of S. typhimurium through different pathways.

(A) Deletion of the gene encoding transcription regulator of rdar biofilm formation CsgD and the gene encoding the cellulose synthase BcsA restored invasion to wild type levels in the STM4264 mutant. Deletion of csgD or bcsA in the STM3611 mutant did not alter the invasion capability of STM3611. Experimental conditions as in Figure 1. WT = wild type S. typhimurium UMR1. Bars show mean ± standard deviation from at least 4 independent biological experiments performed on two technical replicates. Statistical significance is indicated by ***p<0.001. (B) Deletion of csgD in the STM4264 and STM3611 mutants restored secretion of SipA. Upregulation of SipA secretion was observed also when csgD was deleted in the wild type background, although no upregulated invasion has been observed in the csgD mutant (Fig. S5; [15]). SipA-β-lactamase indicates chromosomal fusion in respective strains. Detection of the SipA-β-lactamase fusion protein by western blot analysis using an anti-β-lactamase antibody. Strain S. typhimurium UMR1 with pBAD30 expressing β-lactamase in the periplasm (WT-VC) served as β-lactamase secretion control. WT = wild type S. typhimurium UMR1; ΔompR, negative control; VC = vector control pBAD30.

Figure 7. Individual GG(D/E)EF/EAL domain proteins affect S. typhimurium induced IL-8 production by the gastrointestinal cell line HT-29.

Relative IL-8 production of HT-29 cells after incubation with individual GG(D/E)EF/EAL mutants. IL-8 production of the wild type (WT) S. typhimurium UMR1 was set 100%. ΔfliCΔfljB, negative control, U = unstimulated cells. Strains were grown under invasion inducing conditions (standing culture, LB+0.3M NaCl) until O.D.₆₀₀ 0.6 and co-incubated with HT-29 cells for 1 h. IL-8 production of HT-29 cells was measured by ELISA. Bars show mean % ± standard deviation from at least three independent biological experiments performed in two technical replicates. Statistical significance is indicated by *P<0.05, **P<0.01, ***P<0.001 as compared with wild type S. typhimurium UMR1.

Figure 8. Complementation of the IL-8 induction phenotype of GG(D/E)EF/EAL protein mutants and double mutant analysis.

(A) Complementation of the STM1283 mutant with STM1283 (p1283+) and STM1283_D425A (p1283m) demonstrates requirement of the di-guanylate cyclase activity of STM1283 for complementation of the IL-8 induction phenotype as the enhanced IL-8 induction rate exhibited by the STM1283 mutant was significantly reduced by complementation with wild type STM1283, but not with the non-functional mutant STM1283_D425A with the GGDEF motif altered to GGAEF. (B) Complementation of the STM4264 mutant with STM3611 (p3611+) and STM3611_K179A (p3611m) demonstrates requirement of the phosphodiesterase activity of STM4264 for complementation of the IL-8 induction phenotype. The reduced IL-8 induction rate exhibited by the STM4264 mutant was increase to wild type levels by complementation with the EAL-only protein STM3611, but not with the catalytically inactive mutant STM3611_K179A. (C) No additive effect on the reduction of IL-8 production is observed when double mutants of STM2503/STM4264 and STM1703/STM4264 were compared to the respective single mutants. (D) Deletion of the gene encoding the phosphodiesterase STM4264 or the putative phosphodiesterase STM2503 has no effect on IL-8 expression in the STM1283 mutant background indicating that STM4264 and STM2503 degrade the c-di-GMP produced by the di-guanylate cyclase STM1283. WT = wild type S. typhimurium UMR1; VC = vector control pBAD30; ΔfliCΔfljB, negative control, U = unstimulated HT-29 cells. Bars show mean ± standard deviation from at least three independent biological experiments performed in two technical replicates. Statistical significance is indicated by *P<0.05, **P<0.01, ***P<0.001 as compared with the respective single mutant.

Figure 9. Impairment of IL-8 production by STM2503, STM1703 and STM4264 mutants is relieved by deletion of the biofilm regulator CsgD.

IL-8 production of HT-29 cells co-incubated with phosphodiesterase EAL-domain proteins/csgD double mutants was monitored. Restoration of immunogenicity was observed when csgD was deleted in combination with EAL-domain proteins. Overexpression of the phosphodiesterase STM3611 (p3611+ = STM3611 in pLAFR3) in the STM4264/csgD double mutant provoked no additional IL-8 production. WT = wild type S. typhimurium UMR1; VC = vector control pLAFR3; U = unstimulated HT-29 cells. Bars show mean ± standard deviation from at least three independent biological experiments performed in two technical replicates. Statistical significance is indicated by *P<0.05, **P<0.01, ***P<0.001 as compared with the respective single mutant.

Figure 10. Presence of GG(D/E)EF/EAL mutants of S. typhimurium UMR1 in fecal pellets of streptomycin treated mice.

Streptomycin treated mice were infected with groups of four strains (wild type and three GG(D/E)EF/EAL mutants). Clearance of mutants from murine intestine was evaluated by CFU counts per gram feces until 30 days after infection. Three of twenty mutants (STM2672, STM3615 and STM4551) were cleared early from the murine intestine. WT = S. typhimurium UMR1.

Figure 11. Working models of the regulatory networks of c-di-GMP signaling in S. typhimurium leading to suppression of invasion and IL-8 induction.

(A) A model of the c-di-GMP signaling network for the suppression of the invasion phenotype by c-di-GMP signaling. Shown are di-guanylate cyclases and phosphodiesterases with the most pronounced effect on invasion. Activity of the di-guanylate cyclases STM1987 and STM4551 and the phosphodiesterases STM3611 and STM4262 create different c-di-GMP pools which subsequently affect target outputs. Upon elevated c-di-GMP, the invasion phenotype is negatively regulated by c-di-GMP signaling through the cellulose synthase BcsA and the biofilm regulator CsgD. CsgD inhibits secretion of TTSS-1 effector protein SipA. SipA secretion is affected by GGDEF and EAL domain proteins. Whether motility affects the invasion phenotype needs to be demonstrated. (B) Model of the c-di-GMP signaling network for the suppression of the IL-8 induction phenotype by c-di-GMP signaling. The c-di-GMP pool created by the di-guanylate cyclase STM1283 is degraded by the phosphodiesterases STM0468, STM1703, STM2503, STM3375 and STM4264 which are shown in the order of affection of the IL-8 induction phenotype. Upon elevation of c-di-GMP the resulting c-di-GMP pool is suggested to stimulate CsgD expression which subsequently represses IL-8 induction.