Present addresses: Jay C.D. Hinton, Department of Microbiology, School of Genetics and Microbiology, Moyne Institute of Preventive Medicine, Trinity College, Dublin 2, Ireland. Edouard E. Galyov, Department of Infection, Immunity & Inflammation, Maurice Shock Building, University of Leicester, PO Box 138, Leicester LE1 9HN, UK.
Editor: Rob Delahay
Correspondence: Abigail N. Layton, Division of Microbiology, Institute for Animal Health, Compton, Berkshire RG20 7NN, UK. Tel.: +44 1635 578411; fax: +44 1635 577237; e-mail: email@example.com
Salmonella enterica serovar Typhimurium is an animal and zoonotic pathogen of worldwide importance. Intestinal colonization, induction of enteritis and systemic translocation by this bacterium requires type III protein secretion. Strategies that target this process have the potential to control infection, pathology and transmission. We defined the global transcriptional response of S. Typhimurium to INP0403, a member of a family of salicylidene acylhydrazides that inhibit type III secretion (T3S). INP0403 treatment was associated with reduced transcription of genes involved in T3S, but also increased transcription of genes associated with iron acquisition. We show that INP0403 restricts iron availability to Salmonella, and that inhibition of T3S system-1 by INP0403 is, at least in part, reversible by exogenous iron and independent of the iron response regulator Fur.
Salmonella enterica produces a spectrum of diseases from asymptomatic carriage through inflammatory diarrhoea to typhoid fever, depending on serovar- and host-specific factors. Salmonella enterica serovar Typhimurium causes acute enteritis in humans and food-producing mammals. Human infections are frequently associated with direct or indirect contact with food-producing animals and strategies are required to limit entry of Salmonella into the food chain and environment. Intestinal colonization, invasion, induction of enteritis and systemic spread by Salmonella requires type III secretion systems (T3SSs; reviewed in Stevens et al., 2009). T3SSs translocate bacterial effector proteins directly into the host cell cytosol where they subvert cellular pathways (reviewed in Galán & Wolf-Watz, 2006). Salmonella possesses three T3SSs (T3SS-1, T3SS-2 and the flagella system) used at distinct stages of infection. The flagella system mediates bacterial motility and influences the induction of innate responses owing to secretion of the Toll-like receptor-5 agonist flagellin. T3SS-1 encoded on Salmonella pathogenicity island (SPI)-1 promotes bacterial entry into intestinal epithelia by subversion of actin dynamics and plays a key role in the induction of enteritis. The SPI-2-encoded T3SS-2 promotes intracellular survival and, in some serovars or hosts, influences intestinal colonization, enteritis and systemic virulence (Stevens et al., 2009).
As structural components of T3SSs are conserved in many pathogenic bacteria, they represent an attractive drug target (Alksne & Projan, 2000; Patel et al., 2005). Targeting virulence factors without affecting viability may offer an advantage over conventional antibiotics as resistance is predicted to be less likely to develop and escape may occur at the cost of virulence factor function or expression. Furthermore, virulence factors are often absent in nonpathogenic bacteria, thereby limiting deleterious effects on endogenous microorganisms. One such class of compounds are salicylidene acylhydrazides, which inhibit T3SSs in Yersinia (Kauppi et al., 2003; Nordfelth et al., 2005), Chlamydia (Muschiol et al., 2006, 2009; Wolf et al., 2006; Bailey et al., 2007), Shigella (Veenendaal et al., 2009), and enterohaemorrhagic E. coli (Tree et al., 2009). Related molecules with a salicylideneaniline moiety inhibit T3S in enteropathogenic Escherichia coli (Gauthier et al., 2005). We and others have shown that several salicylidene acylhydrazides inhibit T3SS-1 in S. Typhimurium in vitro (Hudson et al., 2007; Negrea et al., 2007) and reduce enteritis in a bovine ligated intestinal loop model of infection (Hudson et al., 2007). Here, we sought to determine the effect of a well-studied salicylidene acylhydrazide on the transcriptome of S. Typhimurium and to evaluate the relevance of selected pathways modulated by the drug in the inhibition of T3S.
Materials and methods
INP0403 was prepared as described (Ainscough et al., 1999) by Innate Pharmaceuticals AB (Umeå, Sweden), and was 97% pure as assessed by 1H nuclear magnetic resonance spectroscopy (data not shown). INP0403 was dissolved in dimethyl sulphoxide (DMSO) to a concentration of 100 mM and used in all experiments at a final concentration of 100 μM, unless otherwise stated. DMSO was used as a control at the same concentration as present in INP0403-treated samples (0.1% v/v).
Bacterial strains and culture conditions
A nalidixic acid (Nal)-resistant derivative of S. Typhimurium strain 4/74 (Morgan et al., 2004), a bovine diarrhoea isolate that is the parent of the genome-sequenced hisG derivative SL1344, was used unless otherwise stated. SL1344 derivatives were used to study the effect of inhibitor on transcription of single-copy gfp+ transcriptional fusions to the T3SS-1 gene prgH (prgH′-gfp+; JH3010), the T3SS-2 gene ssaG (ssaG′-gfp+; JH3009), the housekeeping gene rpsM (rpsM′-gfp+; JH3016) and a promoterless gfp+ (JH3008) (Hautefort et al., 2003). In studies to investigate whether inhibition of Salmonella T3SS-1 was dependent on ferric uptake regulator (Fur) regulation of SPI-1, S. Typhimurium SL1344 wild-type and fur deletion mutant (SL1344 Δfur) strains were used (Karavolos et al., 2008).
Bacteria were cultured in Luria–Bertani (LB) media at 37 °C with shaking unless otherwise stated and supplemented with nalidixic acid at 20 μg mL−1 where appropriate. For experiments requiring induction of T3SS-1, bacteria were grown in LB media overnight with shaking at 25 °C, diluted 1 : 10 into fresh LB media and then incubated at 37 °C for 4 h. This temperature-shift method results in elevated secretion of proteins via T3SS-1 into the culture supernatant (Wood et al., 1996). We have previously reported that INP0403 does not affect bacterial viability or growth during culture in LB medium over this time course (Hudson et al., 2007).
Ten millilitres of LB broth supplemented with nalidixic acid was inoculated with fresh single colonies of S. Typhimurium 4/74 NalR and incubated overnight with shaking at 25 °C. Bacteria were collected by centrifugation, resuspended in 10 mL fresh LB, diluted 1 : 10 into LB containing 100 μM INP0403 or 0.1% v/v DMSO and cultured at 37 °C shaking for 90 min. 2.0 OD600 nm units of each culture were incubated in one-fifth culture volume 5% v/v phenol pH 4.3/95% v/v ethanol solution for 30 min on ice to stabilize RNA. RNA was extracted using the SV Total RNA purification kit (Promega, Southampton, UK).
Purified total RNA (10 μg) was labelled with Cy5-dCTP (Amersham Biosciences, Little Chalfont, UK). All hybridizations were performed as indirect comparison experiments, using Cy3-dCTP-labelled S. Typhimurium SL1344 as the common reference as described (Yang & Speed, 2002). SALSA microarrays covering 92% of the genes common between S. Typhimurium LT2 and SL1344 strains were used (Nagy et al., 2006). Fluorescence intensities of scanned microarrays were quantified using genepix pro software, version 6.0 (Axon Instruments Inc., Foster City, CA). Data were filtered and spots showing a reference signal lower than background+2 SDs were discarded. Unequal dye incorporation was compensated by median centering. Three biological replicates were used for the analysis, and significance of the data at P≤0.05 was determined using a parametric test adjusting the individual P-value with the Benjamini and Hochberg false discovery rate multiple test correction (Benjamini & Hochberg, 1995). The filtered INP0403-treated data were analysed with the genespring™gx microarray analysis software (Agilent Technologies, South Queensferry, UK).
Reporter gene assays
Bacterial strains harbouring gfp+ transcriptional fusions to prgH, ssaG or rpsM were grown overnight with shaking at 25 °C, diluted 1 : 10 into fresh LB media containing 100 μM INP0403 or 0.1 v/v DMSO and incubated at 37 °C shaking for 4 h to induce T3SS-1 expression. Bacteria (1 mL) were collected by centrifugation, washed twice in phosphate-buffered saline (PBS), and fixed in 4% v/v formalin/PBS for 1 min. Fixed bacteria were washed three times in PBS, resuspended in 200 μL PBS and transferred to a 96-well flat, clear-bottomed black plate. Each culture was assayed for fluorescence in triplicate. The total fluorescence intensity of each well was determined using a Wallac 1420 VICTOR2 multilabel reader (PerkinElmer, MA) with a fluorescein filter set (excitation 485 nm/emission 535 nm). All PBS solutions used were 0.22-μm-filtered to reduce autofluorescence.
For each experiment, the mean total fluorescence intensity of triplicate samples was determined and the background fluorescence from the promoterless gfp+ strain was subtracted. Experiments were performed on at least four independent occasions, and mean data were expressed ±SEM. Statistical analysis (Welch two-sample t-test) of the mean data was performed, comparing the effect of treatment with INP0403 to the effect of DMSO on the transcription of each gene, using the r statistical software package (version 2.6.2; http://www.R-project.org). P-values ≤0.05 were considered significant.
Preparation and detection of T3SS-1 secreted protein
Bacteria were grown overnight with shaking at 25 °C, diluted 1 : 10 into fresh LB with supplements where appropriate and cultured for 4 h at 37 °C with shaking. Bacteria were pelleted by centrifugation and culture supernatants were passed through a 0.45-μm low-protein binding filter (Millipore, Watford, UK). Secreted proteins were prepared from filtered supernatants using StrataClean™ resin (Agilent Technologies UK Ltd, Stockport, UK) as described (Hudson et al., 2007) and analysed by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE). For studies on Fur regulation of SPI-1, gels were stained with Deep Purple™ total protein stain and fluorescence intensity of the band corresponding to SipC analysed across two biological replicates using a typhoon scanner and imagequant software (GE Healthcare Life Sciences, Little Chalfont, UK). The location of SipC is known from peptide sequencing and Western blot analysis using a SipC-specific monoclonal antibody (Paulin et al., 2007).
Iron-dependent growth assay
Salmonella Typhimurium 4/74 NalR grown on M9 minimal medium agar was used to inoculate 10 mL M9 broth prepared according to Sambrook & Russell (2001), with the modification of 0.8% w/v glucose (M9-0.8% w/v glucose). All solutions were prepared using TraceSelect water (Sigma, Poole, UK), ultrapure reagents and sterile plasticware to minimize iron contamination. The culture was grown overnight at 37 °C shaking and diluted 1 : 1000 into M9-0.8% w/v glucose containing varying concentrations of iron (III) nitrate and 100 μM INP0403 or 0.1% v/v DMSO. Two hundred and fifty microlitres of culture was added per well to a 96-well flat-bottomed plate with a lid and the OD600 nm was recorded every 30 min for 24 h in a Tecan Infinite 200 plate reader (Tecan UK Ltd, Theale, UK), heated to 37 °C. Each sample was assayed in triplicate, and at least four independent biological replicates of the assay were performed. Statistical analysis (Welch two-sample t-test) of the mean data was performed using the r statistical software package, comparing the effect of INP0403 to DMSO alone at each iron (III) nitrate concentration. P-values ≤0.05 were considered significant. To ensure that the growth conditions were strictly iron-dependent, INP0403 was incubated with Chelex100 resin (Bio-Rad, Hemel Hempstead, UK) for 1 h to remove residual iron before use.
Results and discussion
Effect of INP0403 on the transcriptome of S. Typhimurium grown under T3SS-1-inducing conditions
Salicylidene acylhydrazides and related compounds have been reported to impair transcription of T3S loci in Yersinia (Nordfelth et al., 2005), enteropathogenic E. coli (Gauthier et al., 2005) and enterohaemorrhagic E. coli (Tree et al., 2009). In Salmonella, Negrea et al. (2007) proposed that inhibition of secretion via T3SS-1 is due to transcriptional silencing of SPI-1 as reduced expression of chromosomal lacZ fusions to promoters of SPI-1 genes was seen in S. Typhimurium strain TT16729. However, the authors of this report also noted that the inhibitor may impair the secretion competency of T3SS-1 because secretion, but not expression, of a SipB-β-lactamase fusion protein was inhibited, with the SPI-1-encoded fusion protein accumulating intracellularly (Negrea et al., 2007). However transcription of a chromosomal lacZ fusion to sipC in the same operon was repressed approximately 10-fold in the presence of an inhibitor, which is at odds with the absence of effects on SipB fusion protein expression (Negrea et al., 2007). To further investigate the mechanism of salicylidene acylhydrazide-mediated inhibition of Salmonella T3SS-1, we defined the transcriptome of S. Typhimurium under T3SS-1-inducing conditions in the presence or absence of INP0403, which proved to be the most potent inhibitor of T3SS-1 in our previous studies (Hudson et al., 2007). INP0403 is also known as D4 (active against Salmonella T3S; Negrea et al., 2007), compound 11 (active against Yersinia T3S; Nordfelth et al., 2005) and ME0053 (active against E. coli O157:H7 T3S; Tree et al., 2009). The chemical structure of INP0403 has been described (Hudson et al., 2007; Negrea et al., 2007).
Transcript levels of 113 genes differed significantly by at least twofold in INP0403-treated S. Typhimurium compared with the DMSO-treated control (47 genes upregulated; 66 genes downregulated; Fig. 1 and Supporting Information, Tables S1 and S2). The key findings were as follows.
Transcription of genes encoding the T3SS-1 structural apparatus, regulators and chaperones was reduced by INP0403 treatment. For example, invC encoding the T3SS-1 ATPase was reduced 23.8-fold and hilD encoding a regulator of SPI-1 gene expression was reduced 9.7-fold (Table S2). Our data were largely in agreement with those obtained using SPI-1 chromosomal lacZ transcriptional fusions (Negrea et al., 2007), but no T3SS-1 translocators or effectors shared statistically significant changes in transcription (Table S1). When examining the unfiltered data, transcription of sipA, -B, -C and -D, encoding effector/translocator proteins, was reduced four- to fivefold and other T3SS-1 effector genes, including sopA, sopB, sopD and sopE2, were reduced by 1.5–2.5-fold (Table S1). Although transcription of these genes was reduced upon INP0403 treatment they did not meet the stringent filtering criteria. hilA similarly did not show statistically significant regulation by INP0403 (Table S1) for the same reason. HilA is encoded within SPI-1 and is a key transcriptional regulator of SPI-1 genes, non-SPI-1 encoded T3SS-1 effectors and SPI-4 genes (Bajaj et al., 1995, 1996; De Keersmaecker et al., 2005; Morgan et al., 2007; Thijs et al., 2007).
Few T3SS-2 genes were significantly repressed by INP0403 (sseE twofold, ssaL 3.2-fold), likely because the experiments were performed under T3SS-1-inducing conditions, rather than those that induce T3SS-2 (magnesium limitation and phosphate starvation; Deiwick et al., 1999).
Iron transport genes
A quarter of all genes upregulated by more than twofold upon INP0403 treatment were involved in iron acquisition and transport, including feoA encoding ferrous iron transport protein A, exbB and exbD involved in uptake of the siderophore enterochelin and fhuA, B, C and D involved in hydroxymate-dependent iron transport.
A cluster of genes encoding 50S and 30S ribosomal subunits were repressed 2.3–4.5-fold by INP0403, including rplO encoding the 50S ribosomal subunit L15 and rpsB encoding the 30S ribosomal subunit protein S2. It is possible that this may be associated with effects on iron availability, as downregulation of ribosomal proteins in response to iron limitation has been observed in both transcriptome and proteome studies of Francisella tularensis (Deng et al., 2006; Lenco et al., 2007).
Genes encoding various transporters or drug resistance genes were activated, for example nanT encoding sialic acid transport protein and ybhF encoding a putative ABC-type multidrug transport system.
Selected changes in transcript levels were validated using S. Typhimurium SL1344 strains containing single-copy gfp+ transcriptional fusions to promoters of the T3SS-1 gene prgH, the T3SS-2 gene ssaG, the housekeeping gene rpsM and a promoterless gfp+. INP0403 caused a fourfold reduction in the emission of fluorescence from the strain harbouring a T3SS-1 promoter fusion (prgH′-gfp+) compared with the DMSO control (P=0.020; Fig. 2), confirming the transcriptome data. Strain JH3009 harbouring a gfp+ fusion to the T3SS-2 gene ssaG exhibited a threefold decrease in fluorescence in the presence of INP0403 (P=0.023; Fig. 2), supporting the microarray data, although ssaG did not meet the stringent filtering criteria (Table S1). A control strain JH3016 containing a gfp+ fusion to the rpsM promoter showed equivalent levels of fluorescence when treated with DMSO or INP0403 (Fig. 2). Reverse transcriptase-PCR analysis of the same RNA samples used for microarray analysis did not detect prgH transcripts in the INP0403-treated sample, but they were detected in the DMSO-treated sample, while the housekeeping gene, rpoD, was transcribed in equivalent amounts in both INP0403- and DMSO-treated samples (data not shown).
Effect of INP0403 on transcription of known T3SS-1 regulators
Comparison of the INP0403-sensitive transcriptome to the HilA regulon (De Keersmaecker et al., 2005; Thijs et al., 2007) indicated that only one gene (prgH) in the HilA regulon was significantly (at least twofold) repressed, suggesting that inhibition of T3SS-1 by INP0403 may occur in a HilA-independent manner. A large number of positive and negative regulators of Salmonella T3SS-1 exist (reviewed in Altier, 2005; Ellermeier & Slauch, 2007); thus, we sought to determine whether transcription of any of these was affected by INP0403. Only four previously characterized positive regulators of SPI-1 were significantly (P≤0.05) repressed at least twofold in the presence of INP0403 (Table S3). These included RtsA (11-fold), HilC (5.4-fold) and HilD (9.7-fold), all of which are AraC-like transcriptional activators that constitute a feed forward loop that controls hilA expression in S. Typhimurium (Ellermeier et al., 2005). RtsA, HilC and HilD each independently activate the transcription of hilA, as well as each other (Ellermeier et al., 2005). HilD also activates the SPI-2 regulon in a medium- and growth phase-dependent manner (Bustamante et al., 2008). FliZ was also repressed by INP0403 (2.2-fold), and is an FlhD4C2-dependent activator of flagellar Pclass2/middle gene expression (Saini et al., 2008) and a positive regulator of SPI-1 gene expression (Lucas et al., 2000; Iyoda et al., 2001), via post-transcriptional control of HilD (Kage et al., 2008). Although the effect of INP0403 on hilA expression was not statistically significant, it remains feasible that it produces a biologically significant effect on T3S even though transcription of few genes under the control of HilA was significantly modulated.
No other flagellar genes were significantly affected by INP0403 in the filtered dataset (Table S2), but fliA and fliY, which are in an operon with fliZ, were repressed approximately twofold, and most other flagellar genes showed a similar pattern of 1.5–2-fold downregulation (Table S1). This is consistent with previous observations that two salicylidene acylhydrazides caused a modest, but significant, decrease in Salmonella motility and surface expression of flagellin (Negrea et al., 2007).
Addition of iron prevents inhibition of T3SS-1 activity by INP0403
Induction of iron acquisition genes by INP0403 coupled with the observation that exogenous iron reverses the inhibitory effects of salicylidene acylhydrazides on T3S in Chlamydia (Slepenkin et al., 2007) led us to hypothesize that the mechanism of Salmonella T3SS-1 inhibition by INP0403 may involve iron chelation. Secreted proteins were prepared from S. Typhimurium 4/74 NalR grown under T3SS-1-inducing conditions in the presence of INP0403 or DMSO, and iron (II) sulphate, calcium (II) chloride, iron (III) chloride or iron (III) nitrate.
Addition of ferrous iron (Fe2+) partially restored T3SS-1-dependent protein secretion by Salmonella in the presence of INP0403 (Fig. 3a). The ability of iron to reverse inhibition by INP0403 was specific to iron and not other metal cations because addition of 50 μM calcium chloride did not restore T3SS-1-dependent protein secretion in the presence of INP0403 (Fig. 3a), but addition of 50 μM iron in the ferric state [Fe3+; iron (III) chloride or iron (III) nitrate] did (Fig 3b and c). Iron (III) nitrate acted in a dose-dependent manner to prevent inhibition of T3SS-1 activity by INP0403 (Fig. 3c). Furthermore, addition of the iron chelator 2,2′-dipyridyl (200 μM) inhibited secretion of proteins via T3SS-1 as well as INP0403 in the strain used herein (data not shown), supporting the findings of others (Ellermeier & Slauch, 2008). It is noteworthy that recent analysis of the global transcriptional response of E. coli O157:H7 to salicylidene acylhydrazides did not reveal statistically significant effects on iron acquisition genes; however, the transcriptome studies used RNA from bacteria cultured in the presence of exogenous iron [0.25 μM Fe(NO3)2] and it is possible that this may have masked an effect (Tree et al., 2009).
INP0403 restricts iron availability to S. Typhimurium
An iron-dependent growth assay was established to evaluate the ability of INP0403 to restrict iron supply to S. Typhimurium. There was an increase in bacterial growth with increasing concentrations of exogenous iron (Fig. 4a), indicating that growth of S. Typhimurium 4/74 NalR in the assay was iron-dependent. For analysis of the effect of the inhibitor on iron-dependent growth, two different time points were compared; 12 h (logarithmic phase) and 24 h (stationary phase, final OD600 nm reached). At both time-points, INP0403 inhibited iron-dependent growth compared with DMSO (Fig. 4b and c). Between 1 and 10 μM iron (III) nitrate was required to overcome the growth inhibition, confirming that INP0403 restricts iron availability to Salmonella. These observations provide an indirect measure of the effect of INP0403 on iron supply and further studies will be required to determine whether INP0403 directly binds iron.
INP0403 inhibition of T3SS-1 is independent of Fur
Iron regulates T3SSs in Shigella dysenteriae and Pseudomonas syringae (Murphy & Payne, 2007; Bronstein et al., 2008). In Salmonella, the T3SS-1 genes invH and sopA were highly expressed under iron-rich conditions (Bjarnason et al., 2003), and 2,2′ dipyridyl represses expression of the SPI-1 transcriptional activator hilA and subsequent protein secretion via T3SS-1 (Ellermeier & Slauch, 2008; this study). Furthermore, Fur was recently reported to activate hilA expression (Ellermeier & Slauch, 2008). To investigate whether inhibition of Salmonella T3SS-1 is dependent on Fur-regulation of SPI-1, proteins secreted via T3SS-1 were prepared from culture supernatants of S. Typhimurium SL1344 wild-type and SL1344 Δfur strains grown in the presence of INP0403 or DMSO and analysed by SDS-PAGE. Levels of the T3SS-1-secreted protein SipC were quantified by scanning of gels stained with a fluorescent total protein stain (Fig. 5). The location of SipC is known from peptide sequencing of S. Typhimurium secreted proteins and Western blotting (data not shown). Densitometric analysis of secreted SipC in cultures of the wild-type strain indicated a mean fold reduction of 7.97±2.71 in the presence of INP0403 relative to the DMSO-treated control. The Δfur mutant exhibited a reduction in secreted SipC of 3.61±0.67-fold compared with the wild-type in the presence of DMSO, consistent with the role of Fur in the activation of SPI-1 (Ellermeier & Slauch, 2008). In the presence of INP0403, there was a further reduction in SipC secreted by the Δfur mutant of 3.50±0.53-fold relative to DMSO-treated SL1344 Δfur. This indicates that the effect of INP0403 on secretion of SipC occurs, at least in part, independently of Fur. No effect of INP0403 on fur transcription was observed by transcriptome analysis.
In conclusion, inhibition of T3S by a candidate salicylidene acylhydrazide anti-infective agent is associated with modulation of gene expression in a manner that may be linked to iron sequestration. We show that INP0403 is capable of restricting iron supply to Salmonella, and that inhibition of T3SS-1 by INP0403 is reversible by exogenous iron and, at least in part, independent of the iron-response regulator Fur. These data contrast with recent observations that such molecules may impair assembly of the Shigella flexneri T3S needle complex (Veenendaal et al., 2009), and raise the possibility of inhibitor- and species-specific modes of action. Taken together with data on the iron-sensitive activity of salicylidene acylhydrazides against Chlamydia (Slepenkin et al., 2007), our data reinforce the need for future studies on the mode of action of such molecules to address the potential for pleiotropic effects related to iron supply.
The authors gratefully acknowledge the financial support from the Biotechnology and Biological Sciences Research Council (BBSRC), including grant D010632/1 to E.E.G. and M.P.S., and a BBSRC core strategic grant to J.C.D.H. We thank Innate Pharmaceuticals AB for providing inhibitors, and Dr Simon Andrews, University of Reading, for providing S. Typhimurium SL1344 wild-type and Δfur strains and for helpful discussions.