Postranscriptional regulation od salmonella pathogenicity island 1

DEPARTAMENTO DE GENÉTICA UNIVERSIDAD DE SEVILLA Postranscriptional Regulation of Salmonella Pathogenicity Island 1 Memoria presentada para aspirar al grado de doctor en Biología Javier López Garrido Vº Bº: Dr. Josep Casadesús Pursals Director de la tesis INDEX Page Introduction 1 Chapter 1. “Regulation of Salmonella enterica pathogenicity island 1 (SPI-1) by DNA adenine methylation” 10 Chapter 2. “Genetic cross-talk between Std fimbriae, SPI-1 and flagellum in Salmonella enterica serovar Typhimurium” 53 Chapter 3. “hilD mRNA possesses a long regulatory 3’ unstranslated region involved in the postranscriptional integration of signals for SPI-1 regulation” 92 Chapter 4. “Inhibition of Salmonella enterica invasion by L-arabinose” 118 Discussion 161 Bibliography 173 INTRODUCTION 1 1. The genus Salmonella The genus Salmonella includes facultative anaerobic Gram-negative, rod-shaped bacteria, able to infect a variety of animal hosts, including mammals, birds, reptiles, and amphibians. Most Salmonellae are motile and produce peritrichous flagella. The Salmonella are members of the family Enterobacteriaceae in the γ-proteobacteria subdivision and are close relative to the genera Escherichia, Shigella and Citrobacter. Currently, the genus Salmonella is divided into 2 species, called Salmonella enterica and Salmonella bongori. Salmonella enterica can be further subdivided into 6 subspecies: Salmonella enterica subsp. enterica (subsp. I), Salmonella enterica subsp. salamae (subsp. II), Salmonella enterica subsp. arizonae (subsp. IIIa), Salmonella enterica subsp. diarizonae (subsp. IIIb), Salmonella enterica subsp. enterica (subsp. I), Salmonella enterica subsp. houtenae (subsp. IV), and Salmonella enterica subsp. indica (subsp. VI). Every Salmonella subspecies are classified in serovars. Salmonella serovars are distinguished by antisera to two highly variable surface antigens, the O antigen and the H antigen, reflecting variation in the exposed part of lipopolysaccharide and in flagellin, respectively (Grimont, 2007; McQuiston et al., 2004). There are more than 2500 serovars belonging to the different subspecies (Grimont, 2007). Only serovars of subsp. enterica have the ability to colonize warm-blooded vertebrates, and account for 99 % of all human infections by Salmonella, while members of Salmonella bongori and the rest of Salmonella enterica subspecies rarely infect mammals and birds. More than 1,500 serovars belonging to subsp. enterica have been identified so far (Popoff et al., 2004), and they differ in host specificity and the disease conditions they promote. Some of them are host-restricted, while others can infect a broad variety of animal hosts. The diseases produced by subsp. enterica range from self-limiting gastroenteritis to life-threatening systemic infection, and the outcome of the infection depends on the specific serovar-host combination. For example, the human-restricted serovar Typhi produces typhoid fever. However, serovar Typhimurium produces mild gastroenteritis in humans, but causes a systemic disease similar to human typhoid fever when infecting mice. For that reason, the combination serovar Typhimurium-mouse has been extensively used as a model for typhoid fever in humans, and most of the work done with Salmonella has been carried out with that 2 serovar. In this work, we have used the virulent strain Salmonella enterica subsp. enterica serovar Typhimurium ATCC14028 (see Jarvik et al., 2010, for a deep description of that strain). For simplicity, it will be abbreviated as Salmonella typhimurium 14028, or simply Salmonella typhimurium. 4. Evolution of Salmonella pathogenicity Salmonella and Escherichia coli are close relatives, and diverged from 120 to 160 million years ago (Ochman and Wilson, 1987). Thus, the study of Salmonella has taken advantage of all the information available for Escherichia coli, and basic metabolism is relatively well characterized. What makes Salmonella unique from standard Escherichia coli, of course, is pathogenesis. Almost 25 % of Salmonella genome consists of material that is absent in Escherichia coli (McClelland et al., 2001; Prowllic andMcClelland, 2003). The evolution of Salmonella pathogenicity has involved the sequential acquisition of genetic elements, each one contributing to different aspects of Salmonella virulence (Ochman and Groisman, 1997; Kelly et al., 2009). Amongst those elements are the Salmonella pathogenicity islands (SPIs), which are clusters of virulence genes in Salmonella chromosome. More than 10 SPIs has been described so far (Hensel, 2004), but some of them are serotype-specific. Since those regions are absent in Escherichia coli chromosome, many of them have a different G-C content than the average Salmonella chromosome, and some are flanked by insertion sequences, it is thought that SPIs have been acquired by horizontal gene transfer (Kelly et al., 2009; Prowollik and McClelland, 2003). The 2 better characterized SPIs are Salmonella pathogenicity island 1 (SPI-1), involved in the invasion of intestinal epithelial cells, and Salmonella pathogenicity island 2 (SPI- 2), that allow Salmonella to survive in macrophages and colonize deeper tissues (Ochman et al., 1996). SPI-1 was acquired around 25-40 million years ago by the common Salmonella ancestor, that got then the ability to invade eukaryotic cells and became an intracellular pathogen presumably associated with cold-blooded vertebrates. The next major event was the acquisition of SPI-2. SPI-2 is a defining characteristic of Salmonella enterica, 3 and its acquisition resulted in the split between the 2 Salmonella species. Hence, only members of Salmonella enterica have the ability to reach deeper tissues and produce systemic infections. The subsp. enterica ancestor acquired the capacity to infect warm-blooded vertebrates, and different strains subsequently evolved to colonize a variety of hosts. Even though the mechanism of host specificity is not fully understood, it has been speculated that presence of a virulence plasmid in some subsp. enterica serovars may contribute to that. Another factor that can be involved in host specificity is the presence of different sets of fimbrial operons in different serovars (Townsend et al., 2001). A tentative phylogeny of the Salmonella pointing out the acquisition of main virulence traits is depicted in Figure I1 3. Overview of Salmonella infection As mentioned above, depending on the serovar and the host, Salmonella infections have different outcomes: (i) gastroenteritis; (ii) systemic infection; and (iii) asymptomatic chronic carriage. Salmonella lives primarily in the intestine of animal hosts, and is usually transmitted by the fecal-oral route. Infection normally starts via ingestion of contaminated water or food. Along the digestive track, Salmonella must endure some adverse conditions, such as the acidic pH in the stomach and the presence of bile in the duodenum. pH of the stomach is approximately 3.0, and this acidic environment destroys the majority of bacteria that enter the stomach (McGowan et al., 1996; Tennant et al., 2008). However, S. Typhimurium, which prefers to live and grow at a pH near neutrality, responds to acidic challenges through a complex adaptive system called the acid tolerance response, in which adaptation to mild (pH 5.8) or moderate (pH 4.4) acid conditions enables the cell to endure periods of severe acid stress (pH 3) (Foster & Hall, 1990; Lee et al., 1994; Lee et al., 1995). In the small intestine, Salmonella finds high concentrations bile, secreted in the duodenum during digestion. Bile salts have at least two distinct antibacterial activities: as detergents that disrupt the cell envelope (Gunn, 2000) and as DNA-damaging agents that cause DNA rearrangements and point mutations (Prieto et 4 al., 2004). Nevertheless, enteric bacteria are intrinsically resistant to high concentrations of bile and individual bile salts (Gunn, 2000). When Salmonella reaches the distal small intestine, the pathogen has the ability to penetrate inside epithelial cells in a bacterial-induced phagocytosis-like process known as invasion. Invasion is a critical step for Salmonella infection in both, independently if the final outcome is gastroenteritis or systemic infection. In the case of gastroenteritis, the infection is localized in the intestine, and invasion of intestinal epithelial cells triggers an inflammatory reaction in the intestinal mucosa, what leads to liquid accumulation in the intestinal lumen producing diarrhea. The inflammatory response creates a novel luminal niche, which favors growth of Salmonella over the resident microbiota of the intestine. Remarkably, the cascade of events that takes place as consequence of inflammation produces the accumulation of tetrathionate (S4O62-) in the intestinal lumen (Winter et al., 2010). Since Salmonella can use tetrathionate as terminal electron acceptor (Muller, 1923; Hensel et al., 1999), that series of events enables the pathogen to use tetrathionate respiration to obtain energy for growth in the inflamed gut lumen (Winter et al., 2010), taking advantage over the resident microbiota of the intestine. In the case of systemic infection, the strategy used by Salmonella is different. After invasion, the pathogen crosses the epithelial barrier, and can survive inside macrophages and disseminates through the lymphatic system reaching deeper tissues. Salmonella can colonize target organs, particularly the spleen, liver, gall bladder and bone marrow, where bacteria can proliferate, and eventually causing death. A fraction of individuals recovering from systemic infection become asymptomatic, life-long carriers of Salmonella, acting as reservoirs for future infections. In humans, serovar Typhi can establish chronic carriage in the gall bladder. 4. Type 3 secretion and Salmonella pathogenicity The interaction between Salmonella and host cells involves the delivery of bacterial proteins into host cells cytoplasm through a specialized organelle called type 3 secretion 5 systems (TTSS) (Galan, 1999; Galan and Collmer, 1999; Galan, 2001). These systems are evolutionarily related to the flagellar export apparatus and are present not only in several species of bacteria pathogenic for animals but also in bacteria pathogenic for plants or in symbionts for plants or insects (Cornelis and Van Gijsegem, 2000; Galan, 2001). TTSS are typically composed of approximately 25 proteins (Cornelis and Van Gijsegem, 2000) forming a needle-like complex that spans both the inner and outer bacterial membranes (Kubori et al., 1998). As mentioned above, SPI-1 and SPI-2 encode complete type 3 secretion systems (TTSS), and some proteins, called effectors, that are delivered to the eukaryotic cell cytoplasm through the corresponding TTSS. 5. Salmonella invasion Salmonella has the ability to invade epithelial cells in the animal intestine. That process is induced by the bacteria and requires the expression of genes encoded in SPI-1. Salmonella preferentially invades M cells of Peyer’s patches in the ileum (Carter and Collins, 1974; Jones et al., 1994). Mechanistically, the process is similar to macropinocytosis (Swanson and Watts, 1995): Salmonella induces the formation of membrane ruffles in the epithelial cells. The ruffles are localized in the site of bacterium-host cell interaction (Francis et al., 1999), and its formation involves actin polymerization (Goshima et al., 1984; Yahara et al., 1982; Finlay and Falkow, 1988). Salmonella promotes actin filaments rearrangements by delivering effectors into the target cell cytoplasm through SPI-1 TTSS (Guiney and Lesnick, 2005). Invasion is a critical step in Salmonella infection, independently of the final outcome. 6. Reguation of Salmonella pathogenicity island 1 (SPI-1) 6 Since Salmonella pathogenicity islands have been acquired by horizontal transfer, a critical aspect of Salmonella pathogenesis is achieving a coordinated regulation of virulence genes. In some cases, the islands encode transcriptional regulators of their own expression that serve as link between the genes in the island and ancestral regulatory systems. That is illustrated in the case of SPI-1 and SPI-2 regulation: (i) SPI-2 encodes a two-component regulatory system called SsrA (SpiR)/SsrB, which is responsible for SPI-2 genes expression. SsrA is the predicted integral membrane sensor, and SsrB is the cognate response regulator. SsrB binds to the promoter of all SPI-2 functional gene clusters and is essential for expression of the SPI-2-encoded TTSS and its effectors (Walthers at al., 2007). SsrA/SsrB integrates regulatory inputs by the nucleoid-associated protein H-NS (Bustamante et al., 2008), and the two component systems EnvZ/OmpR (Feng et al., 2003; Lee et al., 2000) and PhoP/PhoQ (Bijlsma and Groisman, 2005), thus contributing to connect SPI-1 expression with global regulators encoded in the core genome. (ii) SPI-1 encodes 4 transcriptional activators of its own expression: HilA, HilC, HilD, and InvF. Those activators form a regulatory network that controls the expression of genes encoding TTSS components and effector proteins. HilA, a member of the OmpR/ToxR family (BAJAJ et al. 1995; LEE et al. 1992) activates transcription of SPI genes that encode components of the secretion apparatus as well as the gene for the InvF transcriptional regulator (BAJAJ et al. 1996). In association with SicA, InvF is necessary to boost transcription of the sicA and sipBCDA transcriptional units (DARWIN and MILLER 1999; EICHELBERG and GALAN 1999). HilC and HilD are members of the AraC/XylS family, and activate transcription from the pinvF and psicA promoters in an apparently redundant manner (AKBAR et al. 2003). Transcriptional activation by HilC and HilD relieves repression of the hilA promoter by the nucleoid proteins H-NS and Hha (OLEKHNOVICH and KADNER 2006). Furthermore, HilC and HilD can activate inv/sicA transcription in the absence of HilA (AKBAR et al. 2003; RAKEMAN et al. 1999). A transcription factor located outside SPI-1, RtsA, is also involved in transcriptional control of SPI-1 (ELLERMEIER and SLAUCH 2003). A diagram of SPI-1 transcriptional regulation is presented in Figure 1. Besides the regulatory actions described above, positive feedback loops are involved in the control of hilD, hilC, and rtsA transcription (ELLERMEIER et al. 2005). 7 SPI-1 is a ~40 kb gene cluster containing at least 37 genes (ALTIER 2005; JONES 2005; LOSTROH and LEE 2001), located at centisome 63 on the Salmonella enterica chromosome (MCCLELLAND et al. 2001). SPI-1 encodes a type 3 secretion system and secreted effectors that interact with proteins inside epithelial cells in the animal intestine (GALAN and CURTISS 1989). SPI-1 genes are organized in 7 or more transcriptional units, whose expression is under the control of four SPI-encoded transcription factors: HilA, HilC, HilD, and InvF (LOSTROH and LEE 2001). HilA, a member of the OmpR/ToxR family (BAJAJ et al. 1995; LEE et al. 1992) activates transcription of SPI genes that encode components of the secretion apparatus as well as the gene for the InvF transcriptional regulator (BAJAJ et al. 1996). In association with SicA, InvF is necessary to boost transcription of the sicA and sipBCDA transcriptional units (DARWIN and MILLER 1999; EICHELBERG and GALAN 1999). HilC and HilD are members of the AraC/XylS family, and activate transcription from the pinvF and psicA promoters in an apparently redundant manner (AKBAR et al. 2003). Transcriptional activation by HilC and HilD relieves repression of the hilA promoter by the nucleoid proteins H-NS and Hha (OLEKHNOVICH and KADNER 2006). Furthermore, HilC and HilD can activate inv/sicA transcription in the absence of HilA (AKBAR et al. 2003; RAKEMAN et al. 1999). A transcription factor located outside SPI-1, RtsA, is also involved in transcriptional control of SPI-1 (ELLERMEIER and SLAUCH 2003). A diagram of SPI-1 transcriptional regulation is presented in Figure 1. Besides the regulatory actions described above, positive feedback loops are involved in the control of hilD, hilC, and rtsA transcription (ELLERMEIER et al. 2005). SPI-1 expression is under the control of additional regulators located outside the island. The ferric uptake regulatory protein, Fur, and the BarA/SirA two-component system are SPI-1 activators (ELLERMEIER and SLAUCH 2008; FORTUNE et al. 2006). In turn, HilE (FAHLEN et al. 2000) and Lon (BODDICKER and JONES 2004; TAKAYA et al. 2003) are negative regulators of SPI-1. Figure 1 In this study, we show that Dam-dependent regulation of SPI-1 has a single target, the hilD gene. However, we present evidence that Dam methylation regulates hilD expression at the postranscriptional level. Because Dam methylase is not known to have functions other than 14 GATC methylation, a reasonable interpretation is that Dam methylation may control transcription of a postranscriptional regulator of hilD expression. 15 MATERIALS AND METHODS Bacterial strains, plasmids, bacteriophages, and strain construction: The Salmonella enterica strains listed in Table 1 belong to serovar Typhimurium, and derive from ATCC 14028. For simplicity, Salmonella enterica serovar Typhimurium is often abbreviated as S. enterica. Luria-Bertani (LB) broth was used as liquid medium. Solid LB contained agar at 1.5% final concentration. Green plates (CHAN et al. 1972) contained methyl blue (Sigma Chemical Co, St. Louis, MO) instead of aniline blue. The indicator for monitoring - galactosidase activity in plate tests was 5-bromo-4-chloro-3-indolyl- -D-galactopyranoside (“X-gal”, Sigma Chemical Co., 40 mg/ml). Antibiotics were used at the concentrations described previously (TORREBLANCA et al. 1999). Targeted gene disruption was achieved using plasmid pKD13 (DATSENKO and WANNER 2000). Antibiotic resistance cassettes introduced during strain construction were excised by recombination with plasmid pCP20 (DATSENKO and WANNER 2000). The oligonucleotides used for disruption (labeled "UP" and "DO") are listed in Table S1, together with the oligonucleotides (labeled "E") used for allele verification by the polymerase chain reaction. Disruption of the rne gene, which encodes ribonuclease E, was performed with primers that eliminate the C-terminal region (VIEGAS et al. 2007). For the construction of transcriptional and translational lac fusions in the Salmonella chromosome, FRT sites generated by excision of Kmr cassettes (DATSENKO and WANNER 2000) were used to integrate either plasmid pCE37 or pCE40 (ELLERMEIER et al. 2002). Unless specified otherwise, all lac fusions used in this study are translational. Addition of 3xFLAG and HA epitope tags to protein-coding DNA sequences was carried out using plasmids pSUB11 (Kmr, 3xFLAG) and pSU314 (Cmr, HA) as templates (UZZAU et al. 2001). Transductional crosses using phage P22 HT 105/1 int201 [(SCHMIEGER 1972) and G. Roberts, unpublished] were used for strain construction operations involving chromosomal markers. The transduction protocol was described elsewhere (GARZON et al. 1995). To obtain phagefree isolates, transductants were purified by streaking on green plates. Phage sensitivity was tested by cross-streaking with the clear-plaque mutant P22 H5. Re-construction of chromosomal duplications by P22 HT transduction was performed as previously described (CAMACHO and CASADESUS 2001). Table 1 16 Construction of strain SV5828: Strain SV5298 was transduced with a Tn10dTc pool prepared as previously described (CANO et al. 2002). Transductants were selected on LB plates supplemented with tetracycline and X-gal. Independent Lac+ transductants were sought and purified on green plates. Individual isolates were then patched on LB with X-gal and LB with X-gal and tetracycline. An isolate which was Lac+ in LB + X-gal + tetracycline and Lac– in LB + X-gal was used as donor in a P22 HT transductional cross to introduce the Tn10dTc insertion in a wild type background. A transductant of this kind was propagated as SV5828. Two-strand DNA sequencing of the Tn10dTc element of SV5828 revealed that insertion had occurred in a GGG/GCT motif upstream of hilD, with the tetA promoter pointing out towards hilD. The insertion had thus generated a conditional, tetracycline-dependent hilD allele. Additional details about this allele are given in Figures S1 and S2. Protein extracts and Western blot analysis. Total protein extracts were prepared from bacterial cultures grown at 37°C in LB medium until stationary phase (final O.D.600 ~2.5). Bacterial cells contained in 0.2 ml of culture were collected by centrifugation (16,000 g, 2 min, 4°C) and suspended in 50 ml of Laemmli sample buffer [1.3% SDS, 10% (v/v) glycerol, 50 mM Tris-HCl, 1.8% -mercaptoethanol, 0.02% bromophenol blue, pH 6.8]. Proteins were resolved by Tris-Tricine-PAGE, using 12% gels. Conditions for protein transfer have been described elsewere (JAKOMIN et al. 2008). Primary antibodies were anti-Flag M2 monoclonal antibody (1:5,000, Sigma Chemical Co, St. Louis, MO), anti-HA HA.11 monoclonal antibody (1:1,000, Covance, Princeton, NJ), and anti-GroEL polyclonal antibody (1:20,000, Sigma). Goat anti-mouse horseradish peroxidase-conjugated antibody (1:5,000, BioRad, Hercules, CA) or Goat anti-rabbit horseradish peroxidase-conjugated antibody (1:20,000, Santa Cruz Biotechnology, Heildelberg, Germany) were used as secondary antibodies. Proteins recognized by the antibodies were visualized by chemoluminescence using the luciferinluminol reagents. Quantitative reverse transcriptase PCR (quantitative RT-PCR) and calculation of relative expression levels. RNA was extracted from S. enterica stationary phase cultures (O.D.600 ~2.5) using the SV total RNA isolation system (Promega Co., Madison, WI) as described at http://www.ifr.ac.uk/safety/microarrays/protocols.html. The quantity and quality 17 of the extracted RNA were determined using a ND-1000 spectrophotometer (NanoDrop Technologies, Wilmington, DE). To diminish genomic DNA contamination, the preparation was treated with DNase I (Turbo DNA free, Applied Biosystems/Ambion, Austin, TX). An aliquot of 0.6 g of DNase I-treated RNA was used for cDNA synthesis using the High- Capacity cDNA Archive Kit (Applied Biosystems, Foster City, CA). Quantitative RT-PCR reactions were performed in an Applied Biosystems 7500 Fast Real-time PCR System. Each reaction was carried out in a total volume of 25 µl on a 96-well optical reaction plate (Applied Biosystems) containing 12.5 µl Power SYBR Green PCR Master Mix (Applied Biosystems), 11.5 µl cDNA (1/10 dilution), and two gene-specific primers at a final concentration of 0.2 mM each. Real-time cycling conditions were as follows: (i) 95°C for 10 min; (ii) 40 cycles at 95°C for 15 s, 60°C for 1 min. No-template control was included for each primer set. Melting curve analysis verified that each reaction contained a single PCR product. Gene expression levels were normalized to transcripts of ompA or gmk, two housekeeping genes that served as internal controls. Gene-specific primers, designed with PRIMER3 software (http://primer3.sourceforge.net), are listed in Table S1. Analysis of hilD mRNA decay. Use of quantitative RT-PCR to monitor mRNA decay has been previously described (BAKER et al. 2007). An overnight LB culture of the strain under study was diluted 50 fold, and incubated at 37oC with shaking until an OD600 around 2.5. Transcription initiation was stopped by adding 500 mg/ml rifampicin, and shaking vigorously during 10 s. Cultures were kept at 37oC. Aliquots were extracted at 1 min intervals and treated with a phenol (5%)-ethanol (95%) mixture. Each aliquot was immediately immersed in liquid N2 and kept frozen until RNA extraction. RNA was extracted using the standard protocol described above. Four independent qRT-PCR reactions, all using primers for the 5' region of hilD mRNA, were used. -galactosidase assays: Levels of -galactosidase activity were assayed using the CHCl3- sodium dodecyl sulfate permeabilization procedure (MILLER 1972). 18 RESULTS Levels of the SPI-1 transcription factors HilA, HilC, HilD, and InvF in Dam+ and Dam– hosts. We examined the effect of Dam methylation on the levels of the main SPI-1 regulatory proteins: HilA, HilC, HilD, and InvF. For this purpose, we used HilA, HilC and InvF protein variants tagged with the 3xFLAG epitope, and a HilD variant tagged with the HA epitope. Western blot analysis in extracts from isogenic Dam+ and Dam– strains indicated that all four regulators were less abundant in Dam– hosts (Figure 2). This observation confirmed that SPI- 1 expression is entirely under Dam methylation control as previously proposed (BALBONTIN et al. 2006), but did not provide any hint about the target(s) of Dam-dependent regulation. In silico examination of GATC site distribution in or near the hilA, hilC, hilD, and invF genes was likewise uninformative (data not shown). Figure 2 Dam-dependent regulation of SPI-1 is transmitted via HilD. In an attempt to identify the SPI-1 regulator(s), if any, involved in transmission of Dam-dependent control to SPI-1, we examined the involvement of the SPI-1 "general" transcription factors HilA, HilC, and HilD, and the sip-specific transcription factor InvF (DARWIN and MILLER 1999; EICHELBERG and GALAN 1999). RtsA, a general SPI-1 transcription factor encoded outside SPI-1 (ELLERMEIER and SLAUCH 2003), was also included in the survey. SPI-1 expression was monitored in a set of mutants, each lacking one SPI-1 transcription factor. Epistasis analysis took advantage of two well known traits of SPI-1 expression. One is regulatory redundancy by certain transcription factors (e. g., HilC and HilD) (ALTIER 2005; JONES 2005). The other is that lack of a single transcription factor does not completely abolish expression in certain transcriptional units (ELLERMEIER et al. 2005). Expression of SPI-1 transcriptional units was monitored by measuring b-galactosidase activities of lac fusions in representative genes. Only those regulators that are known to control a specific SPI-1 transcriptional unit were included in the analysis. For instance, expression of hilC in the absence of HilA was not tested because hilC is not regulated by hilA (LOSTROH et al. 2000; RAKEMAN et al. 1999). In turn, expression of the hilA in the absence of InvF was omitted because InvF is downstream from HilA in the 19 SPI-1 regulatory cascade (EICHELBERG et al. 1999) (Figure 1). The results of these surveys are shown in Figure 3, and can be summarized as follows: (i) Dam-dependent regulation of hilA was not abolished in the absence of HilC. No information was obtained, however, on the potential involvement of HilD on Dam-dependent hilA regulation, since a hilD mutation completely abolished expression of the hilA::lac fusion (Figure 3). In an analogous fashion, Dam-dependent regulation of invF was still observed in HilA–, HilC–, and RtsA– backgrounds, and no information was obtained in a HilD– background (Figure 3). Similar observations were made for sipB, which remained under Dam methylation control in HilA–, HilC–, RtsA– , and InvF– backgrounds. As above, absence of sipB expression in both HilD– Dam+ and HilD– Dam– hosts prevented any conclusion about Dam methylation dependence (Figure 3). However, these experiments provided evidence that none of the HilA, HilC, RtsA, and InvF transcription factors is involved in Dam-dependent control of SPI-1. (ii) Expression of a hilC::lac fusion was not completely abolished in a HilD– background (Figure 3), and similar levels of b-galactosidase activity were detected in cultures of HilD– Dam+ and HilD– Dam– hosts. Similar results were obtained for an invH::lac fusion, which remained under Dam methylation control in HilA– and HilC– hosts, but not in a HilD– background (Figure 3). The epistatic effect of a hilD mutation over a dam mutation thus provided evidence that Dam-dependent regulation of SPI-1 requires a functional hilD gene. Figure 3 Dam methylation regulates the level of hilD mRNA. In an attempt to confirm that Dam methylation regulates hilD expression, the activity of a hilD::lac transcriptional fusion was monitored in Dam+ and Dam– hosts. To our suprise, no difference was found (Figure 4). However, these experiments left one possibility open. Transcription of hilD is under the control of an autogenous, positive feedback loop by the HilD product (ELLERMEIER et al. 2005; ELLERMEIER and SLAUCH 2008). Hence, use of a hilD::lac fusion might prevent the observation of differences, if any, between Dam+ and Dam– hosts, simply because the hilD::lac strain is HilD–. To circumvent this potential problem, the hilD::lac fusion was 20 transduced to isogenic Dam+ and Dam– strains carrying a chromosomal duplication that includes SPI-1 (CAMACHO and CASADESUS 2001). b-galactosidase activities were then monitored in Dam+ HilD+/hilD::lac and Dam– HilD+/hilD::lac merodiploids. No difference was found (Figure 4), thus ruling out the possibility that similar levels of hilD expression in Dam+ and Dam– hosts resulted from disruption of the HilD feedback loop. Evidence that transcription of the hilD gene is not under Dam methylation control (Figure 4) was in stark contrast with Western blot experiments showing different levels of HilD protein in Dam+ and Dam– hosts (Figure 2). Analysis of hilD mRNA content in Dam+ and Dam– hosts (ATCC 14028 and SV5264, respectively) was performed by quantitative reverse transcriptase PCR, using primer pairs complementary to both the 5' and the 3' regions of hilD. A lower level of hilD mRNA was found in the Dam– background (Figure 4). Hence, decreased levels of both hilD mRNA and HilD protein were found in Salmonella Dam– hosts (Figs. 2 and 4), even though a hilD::lac transcriptional fusion did not show Dam-dependent control (Figure 4). Figure 4 Expression of hilD from an heterologous promoter is Dam-dependent. The failure of a hilD::lac transcriptional fusion to show Dam-dependent regulation admits a number of explanations, artefactual or not. Hence, we considered the possibility that hilD regulation by Dam methylation might be in fact transcriptional. If such was the case, we reasoned, Damdependent hilD regulation should not be longer observed when hilD expression was driven from an heterologous promoter. In contrast, Dam dependence in a hilD gene driven from an heterologous promoter would provide evidence for postranscriptional control. On these grounds, we examined whether hilD expression remained Dam-dependent in strain SV5828. This strain, whose construction is described in Materials and Methods, carries a conditional hilD mutation that renders the strain HilD– in the absence of tetracycline, and HilD+ in the presence of either tetracycline or autoclaved chlortetracycline. Using this strain and its isogenic Dam– derivative SV5829, we compared hilD mRNA levels in Dam+ and Dam– hosts in the presence and in the absence of tetracycline. Expression of hilD was Dam-dependent in the presence of tetracycline (Figure 5), thus indicating that a hilD transcript driven by the tetA 21 promoter remained under Dam methylation control like wild type hilD mRNA. As a validation for this conclusion, we observed that expression of invF::lac and sipB::lac fusions remained under Dam methylation control when hilD expression was tetracycline-dependent (Figure 5). These results supported the view that Dam methylation might not regulate hilD transcription but hilD mRNA stability. This possibility was puzzling, because Dam methylation is a DNA modification function, not known to interact with nucleic acid molecules other than double-stranded DNA (MARINUS 1996; WION and CASADESUS 2006). Figure 5 Dam methylation regulates hilD mRNA stability. To compare hilD mRNA stability in Dam+ and Dam– hosts, stationary cultures (O.D.600 = 2.5) were treated with rifampicin to stop transcription. RNA samples were extracted at 1 min intervals, and subjected to quantitative RT-PCR primed by two oligonucleotides of the 5' region of hilD. In all RNA preparations, hilD mRNA was found to decay in a linear manner from 1 min to 4 min after rifampicin addition, and a substantial difference in the decay rate was observed between the RNA preparations from a Dam+ strain and those from a Dam– mutant (Figure 6). The half lives of hilD mRNA were calculated as 67 s in a Dam+ host, and 47 s in a Dam– host. These experiments provided direct evidence that hilD mRNA is less stable in the absence of Dam methylation. Because increased turnover of RNA is not a trait of Salmonella Dam– mutants (BALBONTIN et al. 2006), we interpret that hilD mRNA may undergo different postranscriptional regulation in Dam+ and Dam– hosts. Figure 6 Lack of Hfq enhances hilD mRNA instability in Salmonella Dam– mutants. The evidence that hilD mRNA undergoes postranscriptional control led us to test the involvement of Hfq, an RNA chaperone that is known to interact with multiple RNA molecules including hilD mRNA (SITTKA et al. 2008). To investigate whether lack of Hfq affected hilD mRNA stability, analysis of hilD mRNA content was performed in isogenic Dam+ Hfq+, Dam– Hfq+, Dam+ Hfq–, and Dam– Hfq– isogenic strains. Oligonucleotides complementary to both the 5' and the 3' regions of hilD were used to prime quantitative RT-PCR. In a Dam– background, 22 the hilD mRNA level decreased 2.5 fold in the presence of Hfq and >10 fold in the absence of Hfq (Figure 7). Hence, lack of Hfq enhances the hilD mRNA instability caused by a dam mutation. A recent study has suggested that binding of Hfq to the AU-rich hilD mRNA might be peculiar, in the sense that Hfq might not bind one or more specific RNA regions but the entire mRNA molecule (SITTKA et al. 2008). This binding pattern might contribute to the Hfq protective effect. Figure 7 Lack of Hfq enhances the SPI-1 expression defect of Salmonella Dam– mutants. The effect of an hfq null mutation on Dam-dependent SPI-1 expression was examined in five SPI- 1 genes, selected on the basis of their strong HilD dependence. b-galactosidase activities were measured in Dam+ Hfq+, Dam– Hfq+, Dam+ Hfq–, and Dam– Hfq– isogenic strains carrying hilA::lac, sicA::lac, invF::lac, sipB::lac, and sipC::lac fusions. Raw data are shown in Table S2. Figure 8 is an elaboration of Table S2 data which outlines the differences between Dam– Hfq+ and Dam– Hfq– mutants. Because lac fusions in individual SPI-1 genes have disparate bgalactosidase activities, the activity of each fusion has been normalized to 100 in the Dam+ background. Lack of Hfq caused a decrease in SPI-1 expression (Table S2), as previously described (SITTKA et al. 2007). For the purpose of our study, however, the noteworthy result was that an hfq mutation enhanced the SPI-1 expression defect of Dam– mutants (Figure 7). Dam-dependent expression of SPI-1 was also affected by an hfq mutation when hilD was expressed from an heterologous promoter. In the experiments summarized in Figure S3, we compared the expression of lac fusions in two SPI-1 genes, invF and sipB, in isogenic Hfq+ Dam+, Hfq+ Dam–, Hfq– Dam+ and Hfq– Dam– hosts, all expressing hilD under the control of the tetA promoter. Lack of Hfq enhanced the SPI-1 expression defect of Salmonella Dam– mutants (Figure S3). Hence, an hfq mutation enhances the hilD mRNA instability associated to lack of Dam methylation, irrespective of the promoter that drives hilD expression. Lack of degradosome components polyribonucleotide phosphorylase and ribonuclease E suppresses hilD mRNA instability in Salmonella Dam– mutants. If lack of Dam methylation decreases hilD mRNA stability, we reasoned, mutations that reduce RNA 23 SV5384 (hilC’-lacZ+)(Hyb) This study SV5385 dam-231 (hilC’-lacZ+)(Hyb) This study SV5386 hilD (hilC’-lacZ+)(Hyb) This study SV5387 dam-231 hilD (hilC’-lacZ+)(Hyb) This study SV5399 hilD (hilA’-lacZ+)(Hyb) This study SV5400 dam-231 hilD (hilA’-lacZ+)(Hyb) This study SV5401 hilC (hilA’-lacZ+)(Hyb) This study SV5402 dam-231 hilC (hilA’-lacZ+)(Hyb) This study SV5403 hilA (invF’-lacZ+)(Hyb) This study SV5404 dam-231 hilA (invF’-lacZ+)(Hyb) This study SV5405 hilC (invF’-lacZ+)(Hyb) This study SV5406 dam-231 hilC (invF’-lacZ+)(Hyb) This study SV5407 hilD (invF’-lacZ+)(Hyb) This study SV5408 dam-231 hilD (invF’-lacZ+)(Hyb) This study SV5415 hilD (invH’-lacZ+)(Hyb) This study SV5416 dam-231 hilD (invH’-lacZ+)(Hyb) This study SV5417 hilC (invH’-lacZ+)(Hyb) This study SV5418 dam-231 hilC (invH’-lacZ+)(Hyb) This study SV5419 hilA (invH’-lacZ+)(Hyb) This study SV5420 dam-231 hilA (invH’-lacZ+)(Hyb) This study SV5455 hilC::3xFLAG This study SV5456 hilA::3xFLAG This study SV5457 invF::3xFLAG This study SV5540 rtsA (sipB’-lacZ+)(Hyb) This study SV5541 dam-231 rtsA (sipB’-lacZ+)(Hyb) This study SV5542 rtsA (invF’-lacZ+)(Hyb) This study SV5543 dam-231 rtsA (invF’-lacZ+)(Hyb) This study SV5592 DUP[(purG)*MudP*(argG)] (hilD-lacZ) This study SV5594 DUP[(purG)*MudP*(argG)] hilD (hilD-lacZ) This study 30 SV5596 dam-231 DUP[(purG)*MudP*(argG)] (hilD-lacZ) This study SV5598 dam-231 DUP[(purG)*MudP*(argG)] hilD (hilD-lacZ) This study SV5624 hilD::HA This study SV5625 dam-231 hilD::HA This study SV5646 ∆hfq::cat M. Jakomin SV5826 PtetA-hilD (sipB’-lacZ+)(Hyb) This study SV5827 dam-231 PtetA-hilD (sipB’-lacZ+)(Hyb) This study SV5828 PtetA-hilD This study SV5829 dam-231 PtetA-hilD This study SV5847 dam-231 ∆hfq::cat This study SV5848 ∆hfq::cat (hilA’-lacZ+)(Hyb) This study SV5849 dam-231 ∆hfq::cat (hilA’-lacZ+)(Hyb) This study SV5850 ∆hfq::cat (invF’-lacZ+)(Hyb) This study SV5851 dam-231 ∆hfq::cat (invF’-lacZ+)(Hyb) This study SV5852 ∆hfq::cat (sipB’-lacZ+)(Hyb) This study SV5853 dam-231 ∆hfq::cat (sipB’-lacZ+)(Hyb) This study SV5854 ∆hfq::cat (sipC’-lacZ+)(Hyb) This study SV5855 dam-231 ∆hfq::cat (sipC’-lacZ+)(Hyb) This study SV5856 ∆hfq::cat (sicA’-lacZ+)(Hyb) This study SV5857 dam-231 ∆hfq::cat (sicA’-lacZ+)(Hyb) This study SV5873 dam-231 hilC::3xFLAG This study SV5874 dam-231 hilA::3xFLAG This study SV5875 dam-231 invF::3xFLAG This study SV5876 ∆hfq::cat PtetA-hilD (sipB’-lacZ+)(Hyb) This study SV5877 dam-231 ∆hfq::cat PtetA-hilD (sipB’-lacZ+)(Hyb) This study SV5878 ∆hfq::cat PtetA-hilD (invF’-lacZ+)(Hyb) This study SV5879 dam-231 ∆hfq::cat PtetA-hilD (invF’-lacZ+)(Hyb) This study SV5961 ∆rne::cat This study SV5962 dam-231 ∆rne::cat This study SV5963 ∆pnp::cat This study 31 SV5964 dam-231 ∆pnp::cat This study SV5965 ∆rne::cat (hilA’-lacZ+)(Hyb) This study SV5966 dam-231 ∆rne::cat (hilA’-lacZ+)(Hyb) This study SV5967 ∆rne::cat (sicA’-lacZ+)(Hyb) This study SV5968 dam-231 ∆rne::cat (sicA’-lacZ+)(Hyb) This study SV5969 ∆rne::cat (invF’-lacZ+)(Hyb) This study SV5970 dam-231 ∆rne::cat (invF’-lacZ+)(Hyb) This study SV5971 ∆rne::cat (sipB’-lacZ+)(Hyb) This study SV5972 dam-231 ∆rne::cat (sipB’-lacZ+)(Hyb) This study SV5973 ∆rne::cat (sipC’-lacZ+)(Hyb) This study SV5974 dam-231 ∆rne::cat (sipC’-lacZ+)(Hyb) This study SV5975 ∆pnp::cat (hilA’-lacZ+)(Hyb) This study SV5976 dam-231 ∆pnp::cat (hilA’-lacZ+)(Hyb) This study SV5977 ∆pnp::cat (sicA’-lacZ+)(Hyb) This study SV5978 dam-231 ∆pnp::cat (sicA’-lacZ+)(Hyb) This study SV5979 ∆pnp::cat (invF’-lacZ+)(Hyb) This study SV5980 dam-231 ∆pnp::cat (invF’-lacZ+)(Hyb) This study SV5981 ∆pnp::cat (sipB’-lacZ+)(Hyb) This study SV5982 dam-231 ∆pnp::cat (sipB’-lacZ+)(Hyb) This study SV5983 ∆pnp::cat (sipC’-lacZ+)(Hyb) This study SV5984 dam-231 ∆pnp::cat (sipC’-lacZ+)(Hyb) This study 32 LEGENDS TO FIGURES FIGURE 1.– Diagram showing the transcriptional units of Salmonella enterica SPI-1 and the regulatory circuits under the control of transcription factors HilA, HilD, HilC, RtsA, and InvF (adapted from ALTIER 2005; ELLERMEIER and SLAUCH 2003; JONES 2005; and LOSTROH and LEE 2001). FIGURE 2.– Levels of HilA, HilC, HilD, and InvF in protein extracts from Dam+ and Dam– isogenic strains. Epitope-tagged proteins were detected by Western blotting with either anti- FLAG or anti-HA commercial antibodies, as appropriate. The charge control was GroEL in all cases. Strains were SV5456 (hilA::3xFLAG), SV5874 (hilA::3xFLAG Dam–), SV5455 (hilC::3xFLAG), SV5873 (hilC::3xFLAG Dam–), SV5624 (hilD::HA), SV5625 (hilD::HA Dam–), SV5457 (invF::3xFLAG), and SV5875 (invF::3xFLAG Dam–). FIGURE 3.– -galactosidase activities of hilA::lac, invF::lac, sipB::lac, hilC::lac, and invH::lac fusions in the presence and in the absence of individual transcription factors involved in SPI-1 control. Black histograms represent -galactosidase activities measured in a Dam+ background. White histograms represent -galactosidase activities measured in a Dam– background. Strains were SV5284 (hilA::lac), SV5285 (hilA::lac Dam–), SV5401 (hilA::lac HilC–), SV5402 (hilA::lac HilC– Dam–), SV5399 (hilA::lac HilD–), SV5400 (hilA::lac HilD– Dam–), SV5297 (invF::lac), SV5298 (invF::lac Dam–), SV5403 (invF::lac HilA–), SV5404 (invF::lac HilA– Dam–), SV5405 (invF::lac HilC–), SV5406 (invF::lac HilC– Dam–), SV5407 (invF::lac HilD–), SV5408 (invF::lac HilD– Dam–), SV5542 (invF::lac RtsA– ), SV5543 (invF::lac RtsA– Dam–), SV5382 (sipB::lac), SV5383 (sipB::lac Dam–), SV5316 (sipB::lac HilA–), SV5308 (sipB::lac HilA– Dam–), SV5318 (sipB::lac HilC–), SV5310 (sipB::lac HilC– Dam–), SV5320 (sipB::lac HilD–), SV5312 (sipB::lac HilD– Dam–), SV5540 (sipB::lac RtsA–), SV5541 (sipB::lac RtsA– Dam–), SV5322 (sipB::lac InvF–), SV5314 (sipB::lac InvF– Dam–), SV5384 (hilC::lac), SV5385 (hilC::lac Dam–), SV5386 (hilC::lac HilD–), SV5387 (hilC::lac HilD– Dam–), SV5301 (invH::lac), SV5302 (invH::lac Dam–), SV5419 (invH::lac HilA–), SV5420 (invH::lac HilA– Dam–), SV5417 (invH::lac HilC–), SV5418 (invH::lac HilC– Dam–), SV5415 (invH::lac HilD–), and SV5416 (invH::lac HilD– Dam–). Data are averages and standard deviations from 3 experiments. 33 FIGURE 4.– A. -galactosidase activity of a hilD::lac transcriptional fusion in Dam+ (SV5286) and Dam– (SV5288) isogenic hosts. Data are averages and standard deviations from 3 experiments. B. -galactosidase activity of the same hilD::lac transcriptional fusion in Dam+ HilD+ (SV5592), Dam+ HilD– (SV5594), Dam– HilD+ (SV5596), and Dam– HilD– (SV5598) isogenic merodiploids (averages of 3 experiments). C. Relative amounts of hilD mRNA in Dam+ (ATCC 14028) and Dam– (SV5264) strains, normalized to ompA mRNA. Two primer pairs, complementary to 5' and 3' hilD regions, were used. Histograms represent the averages from 3 independent experiments. FIGURE 5.– A. Relative amounts of hilD mRNA in Dam+ (black histograms) and Dam– (white histograms) isogenic strains expressing hilD from an heterologous, tetracyclinedependent promoter. Levels of hilD mRNA were normalized to ompA mRNA, as above. Strains were SV5828 (PtetA-hilD), and SV5829 (dam PtetA-hilD). Data are averages and standard deviations from 3 independent experiments. B. Transcription levels of two SPI-1 genes under HilD control (invF and sipB) in Dam+ (black histograms) and Dam– (white histograms) strains that express hilD from an heterologous, tetracycline-dependent promoter. Strains were SV5297 (invF::lac), SV5298 (invF::lac Dam–), SV5335 (PtetA-hilD invF::lac), SV5336 (PtetA-hilD invF::lac Dam–), SV5382 (sipB::lac), SV5383 (sipB::lac Dam–), SV5826 (PtetA-hilD sipB::lac), and SV5827 (PtetA-hilD sipB::lac Dam–). Data are averages and standard deviations from 3 independent experiments. FIGURE 6.– Stability of hilD mRNA in Dam+ (ATCC 14028) and Dam– (SV5264) isogenic hosts. Values are averages from 4 independent qRT-PCR reactions. Error bars are not shown because the standard deviations were extremely small. FIGURE 7.– A. Enhancement of hilD mRNA instability in the absence of Hfq. Black histograms are for Dam+ strains, and white histograms for their Dam– derivatives. RNA levels were normalized to either ompA mRNA or gmk mRNA. Strains were ATCC 14208 (wild type), SV5264 (Dam–), SV5646 (Hfq–), and SV5847 (Hfq– Dam–). Values are averages and standard deviations from 3 independent experiments. B. Enhancement of the SPI-1 expression defect of S. enterica Dam– mutants by hfq null mutations. Black histograms are for Dam+ 34 strains, and white histograms for their Dam– derivatives. To facilitate visual perception of differences, the -galactosidase activities of individual lac fusions in Dam+ hosts have been normalized to 100. Strains were as follows: SV5284 (hilA::lac), SV5285 (hilA::lac Dam–), SV5278 (sicA::lac), SV5279 (sicA::lac Dam–), SV5297 (invF::lac), SV5298 (invF::lac Dam–), SV5382 (sipB::lac), SV5383 (sipB::lac Dam–), SV5293 (sipC::lac), SV5294 (sipC::lac Dam– ), SV5848 (hilA::lac Hfq–), SV5849 (hilA::lac Hfq– Dam–), SV5856 (sicA::lac Hfq–), SV5857 (sicA::lac Hfq– Dam–), SV5850 (invF::lac Hfq–), SV5851 (invF::lac Hfq– Dam–), SV5852 (sipB::lac Hfq–), (SV5853 (sipB::lac Hfq– Dam–), (SV5854 (sipC::lac Hfq–), and SV5855 (sipC::lac Hfq– Dam–). Data are averages and standard deviations from 3 experiments. FIGURE 8.– A. Suppression of hilD mRNA instability in the absence of degradosome components ribonuclease E (Rne) and polynucleotide phosphorylase (Pnp). Black histograms are for Dam+ strains, and white histograms for their Dam– derivatives. RNA levels were normalized to either ompA mRNA or gmk mRNA. Strains were ATCC 14028 (wild type), SV5264 (Dam–), SV5961 (Rne–), SV5962 (Rne– Dam–), SV5963 (Pnp–), and SV5964 (Pnp– Dam–). Values are averages and standard deviations from 3 independent experiments. B. Suppression of the SPI-1 expression defect of S. enterica Dam– mutants by rne and pnp mutations. Black histograms are for Dam+ strains, and white histograms for their Dam– derivatives. To facilitate visual perception of differences, the -galactosidase activities of lac fusions in individual SPI-1 genes in Dam+ hosts have been normalized to 100. Strains were as follows: SV5284 (hilA::lac), SV5285 (hilA::lac Dam–), SV5278 (sicA::lac), SV5279 (sicA::lac Dam–), SV5297 (invF::lac), SV5298 (invF::lac Dam–), SV5382 (sipB::lac), SV5383 (sipB::lac Dam–), SV5293 (sipC::lac), SV5294 (sipC::lac Dam–), SV5965 (hilA::lac Rne–), SV5966 (hilA::lac Rne– Dam–), SV5967 (sicA::lac Rne–), SV5968 (sicA::lac Rne– Dam–), SV5969 (invF::lac Rne–), SV5970 (invF::lac Rne– Dam–), SV5971 (sipB::lac Rne–), SV5972 (sipB::lac Rne– Dam–), SV5973 (sipC::lac Rne–), SV5974 (sipC::lac Rne– Dam–), SV5975 (hilA::lac Pnp–), SV5976 (hilA::lac Pnp– Dam–), SV5977 (sicA::lac Pnp–), SV5978 (sicA::lac Pnp– Dam–), SV5979 (invF::lac Pnp–), (SV5980 (invF::lac Pnp– Dam–), SV5981 (sipB::lac Pnp–), SV5982 (sipB::lac Pnp– Dam–), SV5983 (sipC::lac Pnp–), and SV5984 (sipC::lac Pnp– Dam–). Data are averages and standard deviations from 3 experiments. 35 LITERATURE CITED AKBAR, S., L. M. SCHECHTER, C. P. LOSTROH and C. A. LEE, 2003 AraC/XylS family members, HilD and HilC, directly activate virulence gene expression independently of HilA in Salmonella typhimurium. Mol. Microbiol. 47: 715-728. ALTIER, C., 2005 Genetic and environmental control of Salmonella invasion. J. Microbiol. 43: 85-92. ALTIER, C., M. SUYEMOTO and S. D. LAWHON, 2000 Regulation of Salmonella enterica serovar Typhimurium invasion genes by csrA. Infect. Immun. 68: 6790-6797. BAJAJ, V., C. HWANG and C. A. LEE, 1995 HilA is a novel OmpR/ToxR family member that activates the expression of Salmonella typhimurium invasion genes. Mol. Microbiol. 18: 715-727. BAJAJ, V., R. L. LUCAS, C. HWANG and C. A. LEE, 1996 Co-ordinate regulation of Salmonella typhimurium invasion genes by environmental and regulatory factors is mediated by control of hilA expression. Mol. Microbiol. 22: 703-714. BAKER, C. S., L. A. EORY, H. YAKHNIN, J. MERCANTE, T. ROMEO et al., 2007 CsrA inhibits translation initiation of Escherichia coli hfq by binding to a single site overlapping the Shine-Dalgarno sequence. J. Bacteriol. 189: 5472-5481. BALBONTIN, R., G. ROWLEY, M. G. PUCCIARELLI, J. LOPEZ-GARRIDO, Y. WORMSTONE et al., 2006 DNA adenine methylation regulates virulence gene expression in Salmonella enterica serovar Typhimurium. J. Bacteriol. 188: 8160-8168. BAXTER, M. A., T. F. FAHLEN, R. L. WILSON and B. D. JONES, 2003 HilE interacts with HilD and negatively regulates hilA transcription and expression of the Salmonella enterica serovar Typhimurium invasive phenotype. Infect. Immun. 71: 1295-1305. BELL, D. C., and C. G. CUPPLES, 2001 Very-short-patch repair in Escherichia coli requires the dam adenine methylase. J. Bacteriol. 183: 3631-3635. BLYN, L. B., B. A. BRAATEN, C. A. WHITE-ZIEGLER, D. H. ROLFSON and D. A. LOW, 1989 Phase-variation of pyelonephritis-associated pili in Escherichia coli: evidence for transcriptional regulation. EMBO J 8: 613-620. BODDICKER, J. D., and B. D. JONES, 2004 Lon protease activity causes down-regulation of Salmonella pathogenicity island 1 invasion gene expression after infection of epithelial cells. Infect. Immun. 72: 2002-2013. 36 CAMACHO, E. M., and J. CASADESUS, 2001 Genetic mapping by duplication segregation in Salmonella enterica. Genetics 157: 491-502. CAMACHO, E. M., and J. CASADESUS, 2002 Conjugal transfer of the virulence plasmid of Salmonella enterica is regulated by the leucine-responsive regulatory protein and DNA adenine methylation. Mol. Microbiol. 44: 1589-1598. CAMPELLONE, K. G., A. J. ROE, A. LOBNER-OLESEN, K. C. MURPHY, L. MAGOUN et al., 2007 Increased adherence and actin pedestal formation by dam-deficient enterohaemorrhagic Escherichia coli O157:H7. Mol. Microbiol. 63: 1468-1481. CANO, D. A., G. DOMINGUEZ-BERNAL, A. TIERREZ, F. GARCIA-DEL PORTILLO and J. CASADESUS, 2002 Regulation of capsule synthesis and cell motility in Salmonella enterica by the essential gene igaA. Genetics 162: 1513-1523. CARPOUSIS, A. J., 2002 The Escherichia coli RNA degradosome: structure, function and relationship in other ribonucleolytic multienzyme complexes. Biochem. Soc. Trans. 30: 150-155. CASADESUS, J., and D. LOW, 2006 Epigenetic gene regulation in the bacterial world. Microbiol. Mol. Biol. Rev. 70: 830-856. CHAN, R. K., D. BOTSTEIN, T. WATANABE and Y. OGATA, 1972 Specialized transduction of tetracycline by phage P22 in Salmonella typhimurium. II. Properties of a high frequency transducing lysate. Virology 50: 883-898. DARWIN, K. H., and V. L. MILLER, 1999 InvF is required for expression of genes encoding proteins secreted by the SPI1 type III secretion apparatus in Salmonella typhimurium. J. Bacteriol. 181: 4949-4954. DATSENKO, K. A., and B. L. WANNER, 2000 One-step inactivation of chromosomal genes in Escherichia coli K-12 using PCR products. Proc. Natl. Acad. Sci. USA 90: 6640- 6645. EICHELBERG, K., and J. E. GALAN, 1999 Differential regulation of Salmonella typhimurium type III secreted proteins by pathogenicity island 1 (SPI-1)-encoded transcriptional activators InvF and hilA. Infect. Immun. 67: 4099-4105. EICHELBERG, K., W. D. HARDT and J. E. GALAN, 1999 Characterization of SprA, an AraC- like transcriptional regulator encoded within the Salmonella typhimurium pathogenicity island 1. Mol. Microbiol. 33: 139-152. 37 ELLERMEIER, C. D., J. R. ELLERMEIER and J. M. SLAUCH, 2005 HilD, HilC and RtsA constitute a feed forward loop that controls expression of the SPI1 type three secretion system regulator hilA in Salmonella enterica serovar Typhimurium. Mol. Microbiol. 57: 691-705. ELLERMEIER, C. D., A. JANAKIRAM and J. M. SLAUCH, 2002 Construction of targeted single copy lac fusions using lambda Red and FLP-mediated site-specific recombination in bacteria. Gene 290: 153-161. ELLERMEIER, C. D., and J. M. SLAUCH, 2003 RtsA and RtsB coordinately regulate expression of the invasion and flagellar genes in Salmonella enterica serovar Typhimurium. J. Bacteriol. 185: 5096-5108. ELLERMEIER, J. R., and J. M. SLAUCH, 2007 Adaptation to the host environment: regulation of the SPI1 type III secretion system in Salmonella enterica serovar Typhimurium. Curr. Opin. Microbiol. 10: 24-29. ELLERMEIER, J. R., and J. M. SLAUCH, 2008 Fur regulates expression of the Salmonella pathogenicity island 1 type III secretion system through HilD. J. Bacteriol. 190: 476- 486. FAHLEN, T. F., N. MATHUR and B. D. JONES, 2000 Identification and characterization of mutants with increased expression of hilA, the invasion gene transcriptional activator of Salmonella typhimurium. FEMS Immunol. Med. Microbiol. 28: 25-35. FALKER, S., J. SCHILLING, M. A. SCHMIDT and G. HEUSIPP, 2007 Overproduction of DNA adenine methyltransferase alters motility, invasion, and the lipopolysaccharide O- antigen composition of Yersinia enterocolitica. Infect. Immun. 75: 4990-4997. FOLICHON, M., V. ARLUISON, O. PELLEGRINI, E. HUNTZINGER, P. REGNIER et al., 2003 The poly(A) binding protein Hfq protects RNA from RNase E and exoribonucleolytic degradation. Nucleic Acids Res. 31: 7302-7310. FORTUNE, D. R., M. SUYEMOTO and C. ALTIER, 2006 Identification of CsrC and characterization of its role in epithelial cell invasion in Salmonella enterica serovar Typhimurium. Infect. Immun. 74: 331-339. GALAN, J. E., and R. CURTISS, 3RD, 1989 Cloning and molecular characterization of genes whose products allow Salmonella typhimurium to penetrate tissue culture cells. Proc Natl Acad Sci U S A 86: 6383-6387. 38 GARCIA-DEL PORTILLO, F., M. G. PUCCIARELLI and J. CASADESUS, 1999 DNA adenine methylase mutants of Salmonella typhimurium show defects in protein secretion, cell invasion, and M cell cytotoxicity. Proc. Natl. Acad. Sci. U S A 96: 11578-11583. GARZON, A., D. A. CANO and J. CASADESUS, 1995 Role of Erf recombinase in P22-mediated plasmid transduction. Genetics 140: 427-434. HAAGMANS, W., and M. VAN DER WOUDE, 2000 Phase variation of Ag43 in Escherichia coli: Dam-dependent methylation abrogates OxyR binding and OxyR-mediated repression of transcription. Mol. Microbiol. 35: 877-887. HEITHOFF, D. M., R. L. SINSHEIMER, D. A. LOW and M. J. MAHAN, 1999 An essential role for DNA adenine methylation in bacterial virulence. Science 284: 967-970. HEUSIPP, G., S. FÄLKER and M. A. SCHMIDT, 2007 DNA adenine methylation and bacterial pathogenesis. Int. J. Med. Microbiol. 297: 1-7. JAKOMIN, M., D. CHESSA, A. J. BAUMLER and J. CASADESUS, 2008 Regulation of the Salmonella enterica std fimbrial operon by DNA adenine methylation, SeqA and HdfR. J. Bacteriol. 190: 7106-7413. JONES, B. D., 2005 Salmonella invasion gene regulation: a story of environmental awareness. J. Microbiol. 43: 110-117. KAGE, H., A. TAKAYA, M. OHYA and T. YAMAMOTO, 2008 Coordinated regulation of expression of Salmonella pathogenicity island 1 and flagellar type III secretion systems by ATP-dependent ClpXP protease. J. Bacteriol. 190: 2470-2478. KÜCHERER, C., H. LOTHER, K. KÖLLING, M. A. SCHAUZU and W. MESSER, 1986 Regulation of transcription of the chromosomal dnaA gene of Escherichia coli. Mol. Gen. Genet. 205: 115-121. LEE, C. A., B. D. JONES and S. FALKOW, 1992 Identification of a Salmonella typhimurium invasion locus by selection for hyperinvasive mutants. Proc. Natl. Acad. Sci, U S A 89: 1847-1851. LØBNER-OLESEN, A., O. SKOVGAARD and M. G. MARINUS, 2005 Dam methylation: coordinating cellular processes. Curr. Opin. Microbiol. 8: 154-160. LOSTROH, C. P., V. BAJAJ and C. A. LEE, 2000 The cis requirements for transcriptional activation by HilA, a virulence determinant encoded on SPI-1. Mol. Microbiol. 37: 300-315. 39 1 2 3 0 200 400 600 800 Control +Tc invF::lacZ 30000 40000 sipB::lacZ Relative RNA content β-galactosidase activity A B Figure 5 0Control +Tc 0 10000 20000 Control +Tc 46 0,1 1 hilD mRNA level wt dam Figure 6 0,01 12345 Relative hilD Time (min) dam 47 20 40 60 80 100 120 140 0 20 40 60 80 100 120 140 5’ region 3’ region 20 40 60 80 100 120 140 20 40 60 80 100 120 140 5’ region 3’ region Relative RNA content Relative RNA content Hfq+Hfq  A Figure 7 0 20 40 60 80 100 120 hilA sicA invF sipB sipC 0 20 40 60 80 100 120 hilA sicA invF sipB sipC 0 0 00 BHfq+ β-galactosidase activity β-galactosidase activity Hfq  48 0 20 40 60 80 100 120 hilA sicA invF sipB sipC 0 30 60 90 120 150 180 0 30 60 90 120 150 180 5’ region 3’ region Relative RNA content β-galactosidase activity 80 100 120 150 180 120 150 180 5’ region 3’ region Relative RNA content galactosidase activity AB Control Rne Control Rne Figure 8 0 20 40 60 80 hilA sicA invF sipB sipC 0 30 60 90 120 0 30 60 90 120 Relative RNA content βgalactosidase activity 0 20 40 60 80 100 120 hilA sicA invF sipB sipC 0 30 60 90 120 150 180 0 30 60 90 120 150 180 5’ region 3’ region Relative RNA content β-galactosidase activity PnpPnp 49 Table S1. Oligonucleotides used in this study (5' 3') Oligonucleotide Sequence hilCUP agggcatattgatttttcttcactggaagtttcctatgacattccggggatccgtcgacc hilCDO attgtacgcataaagctaagcggtgtaatcttaaaatgccgtgtaggctggagctgcttc hilDUP aaatgtaacctttgtaagtaatagtcatcagcgtcctgccattccggggatccgtcgacc hilDDO ttcattcttgccgataagtagatgtcgctaaagctggtacgtgtaggctggagctgcttc hilAUP atccgagagtctgcattactctatcgtgaagggattatcgattccggggatccgtcgacc hilADO gcttcgccgtgggcaaccagcactaacggtaataatcccggtgtaggctggagctgcttc invFUP aggattagtggacacgacatatgctgaatccgataaatggattccggggatccgtcgacc invFDO aaatgtgaaggcgatgagtaaccatgattaacggctaattgtgtaggctggagctgcttc sipBUP cctcgctgaggcggcttttgaaggcgttcgtaagaacacgattccggggatccgtcgacc sipBDO cgcgaagcatccgcattttgctgtaccgcagaagacatgggtgtaggctggagctgcttc sipCUP tagcagcagtaaagtcagtgacctggggttgagtcctacaattccggggatccgtcgacc sipCDO tcctgaatcaggctggtcgatttacgtgaactttcacggggtgtaggctggagctgcttc sicAUP ggaaatgatttgggatgccgttagtgaaggcgccacgctaattccggggatccgtcgacc sicADO tccttttcttgttcactgtgctgctctgtctccgccgtttgtgtaggctggagctgcttc invHUP tcctgtctttttactgatcggctgtgctcaggtgcccctcattccggggatccgtcgacc invHDO gcttgcagtctttcatgggcagcaagtaacgtctgatatagtgtaggctggagctgcttc rtsAUP aaatttactgcagtccgtactcatcaagctcaccacgggtattccggggatccgtcgacc rtsADO ttaacatattgatgacgagaggaagataaaaacgctaaaagtgtaggctggagctgcttc hilD-HAUP taaaactacgccatcgacattcataaaaatggcgaaccattatccgtatgatgttcctga hilD-HADO ttaataaaaatctttacttaagtgacagatacaaaaaatgcatatgaatatcctccttag hilC-3xFLAGUP taagattacaccgcttagctttatgcgtacaatgaaccatgactacaaagaccatgacgg hilC-3xFLAGDO taacgcaaacagatagtaacgtttaaaataatttcacaaacatatgaatatcctccttag hilA-3xFLAGUP caaaagatggaaacaggatccccgcttgattaaattacgggactacaaagaccatgacgg 50 hilA-3xFLAGDO acgatgataaaaaaataatgcatatctcctctctcagattcatatgaatatcctccttag invF-3xFLAGUP gccgcggaaattatcaaatattattcaattggcagacaaagactacaaagaccatgacgg invF3xFLAGDO gcggcacatgccagcactctggccaaaagaatatgtgtctcatatgaatatcctccttag RT-hilD5’-UP agtttgctttcggagcggta RT-hilD5’-DO agcaccaacatcccaggttc RT-hilD3’-UP agcttacggatgttgccgatc RT-hilD3’-DO gcctgattcattcttgccgata RT-ompA-UP tgtaagcgtcagaaccgatacg RT-ompA-DO gagcaacctggatccgaaag RT-gmk-UP ttggcagggaggcgttt RT-gmk-DO gcgcgaagtgccgtagtaat hilC-E1 acgaaatgaacgcgcgttgg hilC-E2 tcactggtgtagcgatactg hilD-E1 agaccattgccaacacacgc hilD-E2 gcgtgttaatgcgcagtctg hilA-E1 tactcaacatggacggctcc hilA-E2 aagccagcaatcagcccatg invF-E1 accagtatcaggagacctgg invF-E2 tgtaaccagaacaagcgcgg sipB-E1 gcgttggtctatctggaggc sipB-E2 tttatgcgcgactctggcgc sipC-E1 gcttcgcaatccgttagcgc sipC-E2 atagcagcgagtgcggatgc sicA-E1 tgttcactaaccaccgtcgg sicA-E2 gctttcgttgccaccacatc invH-E1 gtcagataacgttctgacgg invH-E2 gatgagttcagccaacggtg rtsA-E1 gttgtatgcctttcctggcc 51 rtsA-E2 tccagagttgccttgcctac rneUP gaaacgaaaaccgtcgaaacagccgcgccgaaagcggaagcatatgaatatcctccttag rneDO aaaagccgacctggcggtcggctttgtatcagcatttacatgtaggctggagctgcttcg pnpUP gcgcgtcaggccactgccgctgttatggtaagcatggatgcatatgaatatcctccttag pnpDO agccgcaggttgagactgctcggttgcttctttaatgctctgtaggctggagctgcttcg rne-E1 gacattcgctatgccagatg rne-E2 tcataaacgcctggagtgac pnp-E1 cttccgttgcagaggttcgc pnp-E2 tcaacaaggcgtccagccag RT-hfq-UP cgatttctactgttgtcccgtc RT-hfq-DO ccgtgatggtagttattgctgg RT-rne-UP aagagacaaaagcggaagcg RT-rne-DO acttttccaccacctgggc RT-pnp-UP tcccggttaaggttctggaa RT-pnp-DO caggttgagactgctcggttg 52 Genetic cross-talk between Std fimbriae, SPI-1 and flagellum in Salmonella enterica serovar Typhimurium Javier López-Garrido, Elena Puerta-Fernández and Josep Casadesús 53 ABSTRACT Invasion of intestinal epithelial cells is a critical step in Salmonella infection and requires the expression of genes located in the Salmonella pathogenicity island 1 (SPI- 1). We previously reported that methylation of adenines in the DNA (Dam methylation) is necessary to sustain a high level of SPI-1 expression. Dam methylation controls the expression of the SPI-1 transcriptional activator HilD at postranscriptional level, suggesting that the regulation is indirect. A genetic screen using a multicopy plasmid library of Salmonella genome has shown that std fimbrial operon is the link between Dam methylation and SPI-1. We have characterized 3 new ORFs belonging to std operon (stdD, stdE, and stdF), and have shown that all of them are upregulated in Damˉ background. Deletion of stdE or stdF suppresses SPI-1 repression in dam mutants, and their overproduction in Dam+ hosts leads to SPI-1 repression. Overexpression of StdE and StdF fail to regulate a hilD::lac transcriptional fusion, but reduce the level of hilD mRNA, suggesting that they control hilD expression at postranscriptional level. In addition to repress SPI, ectopic expression of StdE and StdF inhibits motility and represses flagellar gene expression. The regulatory corss-talk mediated by StdE and StdF may contribute to coordinate Std fimbriae-dependent adhesion, invasion and motility in vivo. 54 INTRODUCTION Salmonella enterica is a Gram-negative bacterium that can cause gastric and systemic diseases is a variety of animal hosts. Salmonella is a typical foodborne pathogen, and infection usually starts by the ingestion of contaminated food or water. Salmonella has the ability to penetrate inside epithelial cells in the small intestine, in a process known as invasion. After invasion, the infection can remain localized in the intestine, usually producing self-limiting gastroenteritis. Depending on the specific strain-host combination, in some cases Salmonella can cross the intestinal epithelial barrier, and disseminates inside the host producing a systemic life-threatening infection, such as typhoid fever in humans. It has been estimated that 94 million of cases of gastroenteritis due to Salmonella species and 21 million of cases of typhoid fever occurs around the world every year, resulting in 155,000 and 200,000 deads respectively (Majowicz et al., 2010; Crumpet al., 2004). Salmonella and E. coli are close relatives, and it has been estimated that both species diverged from 120 to 160 million years ago (Ochman and Wilson, 1987). The evolution of Salmonella pathogenicity has involved the sequential acquisition of genetic elements, each one contributing to different aspects of Salmonella virulence (Ochman and Groisman, 1997; Kelly et al., 2009). Amongst those elements are the Salmonella pathogenicity islands (SPIs), which are clusters of virulence genes in Salmonella chromosome. More than 10 SPIs has been described so far (Hensel, 2004), but some of them are serotype-specific. Since those regions are absent in E. coli chromosome, many of them have a different G-C content than the average Salmonella chromosome, and some are flanked by insertion sequences, it is thought that SPIs have been acquired by horizontal gene transfer (Kelly et al., 2009; Prowollik and McClelland, 2003). A key point of Salmonella pathogenesis success is the coordinated expression of virulence genes. That is achieved because SPI gene expression is integrated into preexisting regulatory networks, what generates a cross-talk between the core genome and horizontally-acquired elements (Ochman and Groisman, 1997). One of the better characterized SPIs is the Salmonella pathogenicity island 1 (SPI-1), necessary for invasion of epithelial cells in the animal intestine. SPI-1 encodes a whole type 3 secretion system (TTSS), and some effector proteins that are translocated into the eukaryotic cell cytoplasm (Lostroh and Lee, 2001; Darwin and Miller, 1999). SPI-1 55 RESULTS Genetic screens for regulators of hilD expression using a plasmid library of Salmonella genome We reported previously that SPI-1 regulation by Dam methylation is transmitted via HilD (Lopez-Garrido et al., 2010). However, several lines of evidence suggest that the regulation is indirect: (i) Dam methylation regulates hilD expression at postranscriptional level, while direct regulation is usually transcriptional; (ii) hilD promoter and regulatory regions lack GATC sites; and (iii) although hilD coding sequence contains 3 GATC sites, site directed mutagenesis has demonstrated that they are dispensable for regulation by Dam methylation (not shown). We made two alternative hypotheses to explain Dam-dependent expression of hilD: (i) Dam+ hosts produce a factor necessary for sustaining a high level of hilD expression; (ii) Damˉ host produce a factor able to repress hilD expression at postranscriptional level. Those hypothesis entail that the expression of the factor itself must be Dam-dependent. In addition, its ectopic expression would lead to SPI-1 repression in Dam+ background, or SPI-1 induction in Damˉ background, depending on the case. Thus, we decided to perform a genetic screen for SPI-1 regulators in Dam+ and Damˉ backgrounds, using a pBR328-based multicopy plasmid library of Salmonella genome. As reporter, we used hilD::lac930 fusion. That fusion has lacZ is inserted right after hilD stop codon, and its expression is Dam-dependent. (Figure S1). Dam+ and Damˉ isogenic strains carrying the hilD::lac930 fusion were transduced with 9 pools of plasmid library, each containing around 1000 independent plasmids. Chloramphenicol resistant transductants were selected on LB plates with X-gal. We looked for colonies with reduced βgalactosidase activity (white colonies) in Dam+ background, and colonies with increased β-galactosidase activity (intense blue) in Damˉ background. The cloned fragments were sequenced using specific primers flanking the insertion point (Table S1). Then we checked if the expression of any of the genes contained in the cloned fragments was differentially regulated by Dam methylation, comparing the sequencing results with transcriptomic data from Dam+ and Damˉ isogenic backgrounds (Balbontin et al. 2006). In Damˉ background, we selected 12 independent candidates with increased βgalactosidase activity. Sequencing of the candidates revealed that all of them carried the same fragment cloned in pBR328 (Figure S2B). That fragment contained the gene encoding RtsA protein, known to activate hilD transcription (Ellermeier et al., 2005), 62 amongst other genes (Figure S2B). Thus, we concluded that increased hilD::lac930 expression was due to overproduction of RtsA. However, neither rtsA nor the rest of the genes contained in the plasmid were regulated by Dam methylation, according to transcriptomic data. In Dam+ background, five different plasmids reduced hilD::lac930 β-galactosidase activity (Figure S2A). One of them contained a fragment with the std fimbrial gene cluster, amongst other genes (Figure S2A). According to transcriptomic data, std mRNA is more that 100-folds increased in Damˉ background (Balbontin et al. 2006). We speculated that overexpression of std gene cluster could be the cause of SPI- 1 repression in dam mutants. All the genes in std gene cluster are overexpressed in dam mutants Transcriptomic analyses have shown that stdA, stdB, stdC, and the uncharacterized ORFs STM3026, and STM3025 are repressed by Dam methylation. Dam-dependent expression of stdA, stdB, and stdC has been confirmed by independent methods (Balbontin et al., 2006; Jakomin et al., 2008). However, regulation by Dam methylation of STM3026 and STM3025 has not been further analyzed. DNA sequence indicates the existence of an additional uncharacterized ORF, designed STM3025.1N, in the intergenic region between STM3026 and STM3025. We studied Dam-dependent expression of STM3026, STM3025.1N, and STM3025 by two independent methods: (i) analysis of β-galactosidase activity of STM3026::lac, STM3025.1N::lac, and STM3025::lac translational fusions in Dam+ and Damˉ backgrounds (Figure 2A); and (ii) determination of STM3026, STM3025.1N, and STM3025 protein levels in protein extracts from Dam+ and Damˉ hosts, using protein variants tagged with the 3xFLAG epitope (Figure 2B). Both, β-galactosidase assay and Western blot analysis show that the 3 genes are overexpressed in Damˉ background, thus confirming the transcriptomic data. In addition, detection of the proteins by western blot demonstrates that the 3 ORFs are translated. We renamed STM3026, STM3025.1N and STM3025 as stdD, stdE, and stdF respectively in Salmonella enterica serovar Typhimurium. 63 stdA, stdB, stdC, stdD, stdE, and stdF constitute a polycistronic operon stdA expression is driven by a promoter whose transcription is Dam-dependent (Jakomin et al., 2008; Jakomin et al., in preparation). The fact that expression all the genes in std cluster is induced in a dam mutant provides evidence that they constitute a polycistronic operon transcribed from stdA promoter. In order to confirm that, we experimentally analyzed cotranscription of contiguous ORFs by retroranscription and PCR. Total RNA was extracted from a dam mutant. Traces of DNA were removed by a treatment with Turbo DNase (Ambion). The RNA sample was split in two fractions: one fraction was retrotranscribed to cDNA using random primers; the other fraction suffered the same treatment, but water was added instead of retrotranscriptase. Then, we performed PCR with couples of primers specific for contiguous ORFs (Figure 3A, Table S1), in the presence of different templates: Salmonella genomic DNA as positive control, non-retrotranscribed RNA as negative control, and cDNA as query. The PCR product were resolved in a 2 % agarose with 0.5 µg/ml ethydium bromide, and visualized under UV light. As shown in Figure 3B, a PCR product of the expected size was obtained using either genomic DNA or cDNA. No band was observed when RNA was used as template. That indicates that std gene cluster constitute a polycistronic operon, coordinately transcribed from the promoter identified upstream stdA (Jakomin et al., 2008; Figure 3A). However, internal promoters may also exist. stdE and stdF are the molecular link between Dam methylation and SPI-1 As discussed above, we considered the possibility that overexpression of std operon was the cause of SPI-1 repression in dam mutants. If such were the case, we reasoned, SPI-1 repression in Damˉ background would be suppressed by deletion of std operon. We examined the expression of invF::lac and sipB::lac fusions in isogenic Dam+ and Damˉ strains with either an intact std operon or a deletion covering the whole operon. As shown in Figure 4A, β-galactosidase activities of invF::lac and sipB::lac fusions were reduced more than two folds in Damˉ background in the strain with a functional std operon (Control). However, in a strain lacking the std operon, both fusions displayed similar β-galactosidase activities in Dam+ and Damˉ backgrounds. That result suggests that one or more proteins encoded in std operon are involved in the transmission of Dam-dependent regulation to SPI-1. In order to identify such protein(s), Dam- 64 dependent regulation of an invF::lac fusion was monitored in a set of mutants carrying in frame deletions in individual std genes (Figure 4B). invF::lac expression remains Dam-dependent in strains lacking stdA, stdB, stdC, and stdD, suggesting that those genes are not directly required for Dam-dependent control of SPI-1. However, repression of invF::lac expression in Damˉ background is suppressed in strains lacking either stdE or stdF. That indicates that the product of both genes are necessary for SPI-1 repression in dam mutants. StdE and StdF independently repress SPI-1 expression A conceivable model SPI-1 for regulation by Dam methylation is that, in Damˉ background, std operon is overexpressed and stdE and stdF gene products repress SPI- 1. If that were the case, overexpression of stdE and stdF would repress SPI-1 expression in Dam+ background as well. To test that hypothesis, we placed PLtetO promoter (Lutz and Bujard, 1997) upstream stdE and stdF, to get a constitutive Dam-independent expression of those genes. To avoid possible interferences, stdA native promoter and all the upstream genes in the operon were removed. We made two basic constructions: PLtetO-stdEF in which PLtetO was placed upstream stdE in the chromosome and, in consequence, both stdE and stdF were constitutively expressed; and PLtetO-stdF in which PLtetO was inserted right upstream stdF, thus expressing constitutively only that gene. As controls, we used the same strains carrying in frame deletions in stdE, stdF, or both, depending on the case (Figure 5A). To check if the constructions were working, we analyzed the levels of StdE and StdF in protein extracts from PLtetO-stdEF and PLtetO- stdF strains by Western blot, using protein variants tagged with the 3xFLAG epitope (Figure S3). The level of StdE was around 160 and 40 folds higher in PLtetO-stdEF extracts compared to wild type and dam mutants extracts respectively. The level of StdF in PLtetO-stdEF and PLtetO-stdF extracts was around 16 folds higher that in wild type extracts, but similar to that of dam mutants. Thus, PLtetO constructs overexpress both, stdE and stdF, but the overexpression is higher for stdE than for stdF (Figure S3). We examined the expression the SPI-1 genes hilA, invF, and sipB in strains carrying PLtetO- stdE, PLtetO-stdF, and their respective controls. We analyzed the expression of the selected genes by two independent methods: (i) using lac fusions and measuring βgalactosidase activity; and (ii) determining protein levels by Western blot, using protein 65 variants tagged with the 3xFLAG epitope. Equivalent results were obtained by both methods for the 3 genes studied (Figure 5B). The results can be summarized as follows: expression hilA, invF, and sipB are strongly repressed when PLtetO is inserted upstream stdE. The repression is partially relieved when stdE is deleted. However, deletion of stdF alone does not relieve the repression. Deletion of both genes completely restores SPI-1 expression to wild type levels, suggesting that SPI-1 repression is due to stdE and stdA expression and not to polar effects. Insertion of PLtetO upstream stdF considerably represses the expression of the selected genes, but less than insertion upstream stdE. Deletion of stdF completely suppresses such repression. Those results provide evidence that both, StdE and StdF, can repress SPI-1 expression. In addition, both proteins can repress SPI-1 independently, since each of them can do it in the absence of the other. However, the impact of StdE seems to be quantitatively bigger than StdF, but it may be due to its higher overexpression (Figure S3) StdE and StdF regulate hilD expression at postranscriptional level We previously reported that Dam methylation was necessary to sustain high levels of the SPI-1 transcriptional activator HilD (Lopez-Garrido and Casadesus, 2010). In addition, regulation of hilD by Dam methylation was not transcriptional, but postranscriptional. Taking that into account, we would expect that StdE and StdF repressed hilD expression at postranscriptional level. We examined the expression of hilD using hilD::lac1 transcriptional, in which lacZ was inserted right in hilD transcription start site. We have determined that hilD::lac1 fusion reflects hilD regulation at transcriptional level, since it is activated in the presence of a multicopy plasmid encoding RtsA (Figure S1), a known transcriptional activator of hilD (Ellermeier et al., Figure S4). Expression of hilD::Lac1 fusion was determined in wild type (Control), PLtetO-stdEF, PLtetO-stdEF ∆stdEF, PLtetO-stdF, and PLtetO-stdF ∆stdF backgrounds (Figure 5A, Figure6A). β-galactosidase activities were similar in all the strains, suggesting that StdE and StdF do not regulate hilD transcription initiation. In order to study if StdE and StdF repressed hilD expression at postranscriptional level, we analyzed hilD mRNA levels by Northern blot in the following backgrounds: wild type (Control), PLtetO-stdEF, PLtetO-stdEF ∆stdE, PLtetO-stdEF ∆stdF, PLtetO-stdEF ∆stdEF, PLtetO-stdF, and PLtetO-stdF ∆stdF. As shown in Figure 6B, the level of hilD mRNA is 66 reduced around 4 folds in PLtetO-stdEF strain compared to wild type. Deletion of stdE partially recovers hilD mRNA level, and simultaneous deletion of stdE and stdF completely restores hilD mRNA to wild type level. The amount of hilD mRNA is reduced around two folds in PLtetO-stdF background, and such reduction is completely abolished by stdF deletion. Taking together, those results support the idea that StdE and StdF repress hilD expression at postranscriptional level. Inhibition of motility by StdE and StdF Inhibition of motility by proteins encoded in fimbrial operons has been reported in different bacterial species. (Lin et al., 2001; Simms and Mobley, 2008). We wandered if StdE and StdF were also able to inhibit motility. We examined motility of strain carrying the constructions PLtetO-stdEF, PLtetO-stdF and their respective deletion controls in motility agar plates (Figure 7A). Bacteria simultaneously expressing stdE and stdF are non-motile. A partial recovery of motility is observed when either stdE or stdF is deleted, and deletion of both genes completely restores motility to wild type levels. Motility is also reduced when PLtetO is placed right upstream stdF, and deletion of stdF completely suppresses such reduction. We considered the possibility that StdE and StdF were somehow repressing flagellar gene expression. We analyzed the expression of an flgK::lac fusion in strains carrying either PLtetO-stdEF or PLtetO-stdF, and their respective deletion controls (Figure 7B). The pattern of flgK expression correlates with the defects observed in motility, what suggests that StdE and StdF inhibit motility by repressing flagellar gene expression. 67 DISCUSSION We have characterized three new ORFs in the std gene cluster: STM3026, STM3025.1N, and STM3025, renamed stdD, stdE, and stdF respectively. Western blot analyses have demonstrated that those genes are expressed in dam mutants (Figure 2B). stdD encodes a predicted outer membrane protein, while StdE and StdF are predicted cytoplasmic proteins. Our results indicate that StdE and StdF repress SPI-1 and flagellar gene expression, suggesting the existence of a regulatory cross-talk that might coordinate Std fimbriae production, invasion and motility. We have provided evidence that stdA, stdB, stdC, stdD, stdE, and stdF constitute a polycistronic operon: (i) expression of all those genes is activated in Damˉ background (Figure 2); and (ii) retrotranscription and PCR show that they are cotranscribed (Figure 3). std transcription is driven by a promoter located upstream stdA (Jakomin et al., 2008). Transcription from PstdA is activated by direct binding of HdfR protein to a regulatory region upstream the promoter. However, methylation of two GATC sites in the regulatory region prevents binding of HdfR, thus repressing std expression (Jakomin et al., 2008; Jakomin et al., in preparation). It is likely that all std genes are coordinately regulated by Dam methylation due to a common transcription from PstdA. However, internal promoters may also exist. Salmonella enterica dam mutants are attenuated in the mouse model and present a plethora of virulence-related defects both at the intestinal stage of the infection and during systemic infection (Marinus and Casadesus, 2009). We previously reported that SPI-1 expression was repressed in Damˉ background (Balbontin et al., 2006; Lopez- Garrido et al., 2010). Dam methylation activates SPI-1 by controlling hilD expression at postranscriptional level (Lopez-Garrido and Casadesus, 2010), what together with the absence of GATC sites in hilD regulatory regions, suggests that the regulation is indirect. Our genetic screens and subsequent experiments have identified std fimbrial operon as the link between Dam methylation and SPI-1: (i) a multicopy plasmid containing the whole std operon represses hilD expression; (ii) std genes are upregulated in Damˉ background (Figure 2; Balbontin et al., 2006; Jakomin et al., 2008); and (iii) SPI-1 regulation by Dam methylation is completely suppressed in a strain lacking the whole std operon (Figure 4A). Altogether, those results suggest that overexpression of std in dam mutants leads to SPI-1 repression. It has been previously 68 shown that the extreme attenuation of Salmonella enterica serovar Typhimurium dam mutants upon oral infection (Garcia-del Portillo et al., 1999; Heithoff et al., 1999) is partially suppressed by deletion of std, suggesting that overexpression of Std fimbriae is detrimental for Salmonella virulence (Jakomin et al., 2008). The regulatory link between std and SPI-1 provide evidence that the detrimental effect of std overexpression for Salmonella virulence may be due to SPI-1 repression. Epistasis analysis indicates that Dam-dependent control of SPI-1 requires the last two genes of std operon, StdE and StdF. That is further supported by the following observations: (i) constitutive expression of stdE and stdF in Dam+ background represses SPI-1 expression (Figure 5); (ii) StdE and StdF are overproduced in Damˉ background (Figure 2); (iii) Dam methylation, StdE, and StdF regulate SPI-1 expression through HilD; and (iv) as happens in the case of Dam methylation, StdE and StdF does not regulate hilD transcription, but controls the level of hilD mRNA (Figure 6). A conceivable model to explain regulation by Dam methylation of SPI-1 is depicted in Figure 8: in Dam+ background, GATC sites of PstdA regulatory region are methylated, preventing binding of HdfR and activation of std transcription. In the absence of Dam methylation, HdfR activates transcription from PstdA and all the proteins encoded in the operon are overproduced. Then, StdE and StdF repress hilD expression at posttranscriptional level, and as a consequence, the whole SPI-1 is downregulated. When constitutively expressed, StdE and StdF can regulate hilD expression independently, since each one can do that in the absence of the other (Figure 5). However, deletion of any of them suppresses SPI-1 repression in dam mutants, suggesting that both are necessary for repression. It may be possible that both trigger the same regulatory pathway, but they can do that independently when overproduced. StdE shares around 40 % and 50 % identity with the transcriptional activators GrlA and CaiF from E. coli and Enterobacter cloacae respectively. Interestingly, StdF is similar to an uncharacterized protein encoded just downstream CaiF in Enterobacter cloacae chromosome, that is part of a hypothetical fimbrial gene cluster which genetic organization resembles that of std operon, suggesting that they may have a common origin. StdF is also 27 % similar to the SPI-1 encoded protein SprB from Salmonella. SprB is a transcriptional regulator able to bind to hilD and siiA promoters and repress and activate their expression respectively (Saini and Rao, 2010). Even though StdE and StdF are similar to known transcriptional regulators, they do not regulate hilD at 69 transcriptional level, but at postranscriptional level. Thus, either they have acquired the ability to control gene expression at postranscriptional level in Salmonella enterica, or they regulate transcription of a postranscriptional regulator of hilD. In addition to repress SPI-1 expression, StdE and StdF can also inhibit motility. Expression of an flgK::lac fusion is repressed by constitutive expression of stdE and stdF, suggesting that motility inhibition is the consequence of flagellar gene repression. Overexpression of either stdE or stdF produces a mild motility inhibition (Figure 7). However, simultaneous expression of both genes has a synergistic effect and completely inhibiting motility and flhK expression (Figure 7). Thus, it may be possible that StdE and StdF regulate flagellar gene expression through the same pathway. Several studies have reported coordinated expression of fimbrial, flagellar and invasion genes: in Proteus mirabilis, the protein encoded by the last gene in the mrp fimbrial operon, MrpJ, inhibits motility when the fimbrial operon is expressed (Lin et al., 2001). Similarly, PapX, the product of the last gene in the pap fimbrial operon of uropathogenic Escherichia coli, represses the expression of the flagellar mater regulator FlhDC by direct binding to its promoter region (Simms and Mobley, 2008). One case particularly interesting is the coordinated expression of type I fimbriae, flagellum and invasion genes mediated by FimY and FimZ. Those two proteins are encoded in independent transcriptional units next to the type I fimbrial operon fim. FimY and FimZ are essential for fim operon transcription (Yeh at al., 1995; Tynker and Clegg, 2000). In addition, FimZ represses SPI-1expression by activating transcription of the gene encoding the SPI-1 negative regulator HilE (Baxter and jones, 2005; Saini et al., 2009), and inhibits motility by repressing flhDC expression (Clegg and Hughes, 2002). That situation resembles that of StdE and StdF, suggesting that coordinated expression of fimbrial, flagellar and invasion genes is important for Salmonella virulence and persistence in the intestine (Saini et al., 2010) A tempting speculation derived from the above results is that invasion and motility would be inhibited when std operon was expressed. std is not expressed under laboratory growth conditions in wild type Salmonella (Humphries et al., 2003; Humphries et al., 2005; Jakomin et al., 2008). However, several lines of evidence suggest that Std fimbriae is produced in the animal intestine: (i) mice infected with serovar Typhimutium seroconvert to StdA, the major fimbrial component of Std 70 fimbriae (Humphries et al., 2005); and (ii) std deletion reduces the ability of Salmonella to colonize and persist in the cecum of infected mice, while producing no defect in colonizing the small intestine. (Weening et al., 2005). According to that, it has been reported that Std fimbriae bind α(1,2)fucose residues, which are abundant in the cecal mucosa (Chessa et al., 2008). Salmonella invasion takes place preferentially in the ileum, while in the cecum invasion is inhibited. std expression in the cecum might contribute to inhibition of invasion. In addition, fimbriated bacteria would inhibit motility and live attached to cecal mucosa, what could help to the persistence of Salmonella in the host intestine. The genome of Salmonella has evolved by the acquisition of genetic modules that provided new abilities to interact with eukaryotic cells and exploit different niches (Ochman and Groisman, 1997; Prowllik and McLelland, 2003). A critical point of that modular evolution is to get a coordinate expression of the different genetic modules. In some cases, the modules themselves carry regulatory genes of its own expression, which serve as connection with the core genome (Ochman and Groisman, 1997). In addition, there are some examples of cross-talk between genetic modules independently acquired: the SPI-1 encoded regulator HilD can activate SPI-2 expression during late stationary growth phase (Bustamante et al., 2008); expression of SPI-4 genes is activated by the SPI-1-encoded SprB transcriptional regulator (Saini and Rao, 2010); HilE, a SPI-1 negative regulator, is encoded in a region of Salmonella chromosome that has been proposed to be a pathogenicity island (Baxter et al., 2003); SPI-1 and SPI-2- encoded transcriptional regulators control the expression of effector proteins located outside those islands (Darwin and Miller, 2001; Knodler et al., 2002), and some are located in horizontally-acquired DNA fragments (Hardt et al., 1998; Wood et al., 1998). std genes are well conserved amongst Salmonella serovars, but are absent in closely related species (Prowllik and McCLelland, 2003), suggesting that the cluster has been acquired by horizontal gene transfer. Thus, the connection between std and SPI-1 provides an additional example of cross talk between horyzontally-aquired genes. 71 HilE hilE FimYZ Mlc Fur Lon CsrA BarA/ SirA csrB csrC hilD FliZ Type-3 Secretion System Effector proteins hilA invF Lrp Dam PhoP/ PhoQ EnvZ/ OmpR HilC HilD HilA InvF PhoB/ PhoR Figure 1 78 100100 STM3025.1N- 3xFLAG STM3025- 3xFLAG Dam+Dam_Dam+Dam_ Tagged protein Loading control (GroEL) Quantification 125 STM3026- 3xFLAG Dam+Dam_ 0 20 40 60 80 100 120 STM3026::lac STM3025.1N::lac STM3025::lac Relative β-galactosidase activity A B Figure 2 79 RNA (Dam-) Non retrotranscribed Retrotranscribed Genomic DNA 1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 stdA stdB stdC sdtD stdE stdF STM3025.1N 1 2 3 4 5 STM3025STM3026 PstdA Figure 3 A B 80 0 2000 4000 6000 8000 10000 sipB::lac 0 100 200 300 invF::lac Control ∆std β-galactosidase activity β-galactosidase activity ∆stdControl 0 100 200 300 β-galactosidaseactivity Control ∆stdA ∆stdB ∆stdC invF::lac ∆stdD ∆stdE ∆stdF Figure 4 A B 81 0 100 200 300 400 500 β-galactosidase activity HilA-3xFLAG Loading control (GroEL) Quantification 100 36.5 135.9 111.13.7 3.4 12.5 Control PLtetO-stdEF PLtetO-stdF StdE StdF + + _ + + _ _ _ _ _ +_ hilA::lac 0 100 200 300 400 invF::lac InvF-3xFLAG Loading control (GroEL) Quantification 100 Control PLtetO-stdEF PLtetO-stdF StdE StdF + + _ + + _ _ _ _ _ +_ 38,1 105,9 80,00,7 3,4 12,0 β-galactosidase activity 0 2000 4000 6000 8000 10000 12000 SipB-3xFLAG Loading control (GroEL) Quantification 100 30.6 117.7 137.42.7 8.2 28.4 Control PLtetO-stdEF PLtetO-stdF StdE StdF + + _ + + _ _ _ _ _ +_ sipB::lac β-galactosidase activity stdE stdF PLtetO PLtetO-stdEF PLtetO-stdEF ∆stdE PLtetO stdF ∆ stdE PLtetO ∆ PLtetO ∆ stdF PLtetO ∆ PLtetO PLtetO-stdEF ∆stdF PLtetO-stdEF ∆stdEF PLtetO-stdF PLtetO-stdF ∆stdF Figure 5 A B 82 100 26.9 54.7 30.1 115.7 59.6 135.0 PLtetO-stdEF PLtetO-stdF + + _ + _StdE StdF + + _ _ _ _ _ Control hilD mRNA Loading control (rnpB) Quantification 0 20 40 60 80 100 120 140 160 Control PLtetO-stdEF PLtetO-stdF + + _ _ _ _ + _StdE StdF β-galactosidase activity hilD::lac1 Figure 6 A B 83 wt 1. PLtetO-stdEF 2. PLtetO-stdEF∆stdE 3. PLtetO-stdEF∆stdF 4. PLtetO-stdEF ∆stdEF 1 23 4 1. PLtetO-stdF 2. PLtetO-stdF ∆stdF wt 1 2 0 500 1000 1500 2000 2500 3000 3500 + + + + + _ _ _ _ _ _ _ StdE StdF Control PLtetO-stdEF PLtetO-stdF β-galactosidase activity flgK::lac Figure 7 A B 84 std operon PstdA CH3CH3 std operon PstdA HdfR hilD StdE StdF HilD Dam+background Dam_background Figure 8 SPI-1 85 Table S1. Oligonucleotides used in this study (5' 3') Oligonucleotide Sequence stdDUP acgcaggggcgacatcatgacagaatggatttttaatctgattccggggatccgtcgacc stdDDO gattagttataggtaacagtaacgggtattgcagcagaaagtgtaggctggagctgcttc stdD-E1 tgcagatgaatcgctacacc stdD-E1 ttccccgataactcagtcag stdEUP ccagttatggagaggttttatgtgccctgataatacacacattccggggatccgtcgacc stdEDO ttaccgacccggcgttttgataccagcggcggtccggcttgtgtaggctggagctgcttc stdE-E1 tgctgcaatacccgttactg stdE-E2 caggctgcctgtatgcg stdFUP ggtccggagatttatgccgggctgcaactgtgaaaccgcaattccggggatccgtcgacc stdFDO tgtcagtgtttctggatagggtcgccggaggcgggttattgtgtaggctggagctgcttc stdF-E1 ggtcggtaatggtgacagg stdF-E2 gaaaggccatacattcagcg stdF-3xFLAGUP ttgtaaatcactgcagcgaaccccgtttatcgctccgcaagactacaaagaccatgacgg stdF-3xFLAGDO cggcgtgtcagtgtttctggatagggtcgccggaccgccccatatgaatatcctccttag stdE-3xFLAGUP gaagatcctgaaaaagaaggaggatgaggatgacggaacagactacaaagaccatgacgg PLtetO-std UP tacattaaaaagtatttctttgatgattattcttaaattaaggcttacccgtcttactgtc PLtetO-stdEF-DO ttcagggcacataaaacctctccataactgggtaaatgatgtgctcagtatctctatcactgatag PLtetO-stdF-DO cccgcatttctgttactgcacagccggttccacagttcatgtgctcagtatctctatcactgatag stdA-FOR atagccctgacagatgccg stdB-REV ggcctgcgacttcaggac stdB-FOR ctacctgacaggtctcagc stdC-REV gggtccggtcaacattgac stdC-FOR tgcagatgaatcgctacacc stdD-REV cctagctcaaccgcatacac stdD-FOR ctattacaggacgtgtcacc stdE-REV catcatggttggtctgtccg stdE-FOR ggtcggtaatggtgacagg stdF-REV gtttccgacgtaattgctgc 86 stdAUP taaaattcttttcactggtaccatcaccaactcaccctgtattccggggatccgtcgacc stdADO ctgtcgttatttaccgcgtgaaatcacaggtatttcaggggtgtaggctggagctgcttc stdAUP2 taaaattcttttcactggtaccatcaccaactcaccctgtcatatgaatatcctccttag stdA-E1 ggaaagttcaggtgcttcg stdA-E2 gctttcgtgttgtcgtcc stdBUP gcgccatgccatgaaaattacccggcttgccattctgattattccggggatccgtcgacc stdBDO tttagacctgctctgtgacagggatatttttatctgcagggtgtaggctggagctgcttc stdB-E1 acgacagggagaagccg stdB-E2 cgcatccatgataatacgg stdCUP cacagagcaggtctaaaggagaaaacaggtgaaaaacagcattccggggatccgtcgacc stdCDO aatccattctgtcatgatgtcgcccctgcgtccttaaacggtgtaggctggagctgcttc stdC-E1 cggaaacggtgacactcag stdC-E2 tgaccggaagtgactgcac hilDriboprobeUP atggaaaatgtaacctttgtaag hilDriboprobeDO gtttttttaatacgactcactatagggaggtatatcgaaatccatgtggc rnpBriboprobeUP rnpBriboprobeDO pBR328-Fw actgtccgaccgctttgg pBR328-Rv gccagcaaccgcacctg lacZ hilDUP930 aactacgccatcgacattcataaaaatggcgaaccattaacatatgaatatcctccttag hilDUP1 agagcatttacaactcagattttttcagtaggataccagtcatatgaatatcctccttag hilDDO2 gcaaatagttctcagagggaacggatgatgtataaatatggtgtaggctggagctgcttc hilD-E1 agaccattgccaacacacgc hilD-E2’ atcatcctcaggctggctcc 87 INTRODUCTION Salmonella enterica is a gram negative bacterium able to produce gastric and systemic infections in a variety of animal hosts, including humans. An important trait of Salmonella virulence is its ability to penetrate inside non-phagocytic epithelial cells in the animal small intestine. Such process, known as invasion, requires the translocation of bacterial proteins directly from bacterial cytoplasm to epithelial cell cytoplasm through a type-3 secretion system (TTSS) (Darwin and Miller, 1999; Lostroh and Lee, 2001). Those proteins, called effectors, interact with specific targets inside intestinal epithelial cells, triggering a cascade of molecular events that culminates with Salmonella invasion (Darwin and Miller, 1999). Some effector proteins and all the components of TTSS are encoded in a 40-kb region of Salmonella chromosome, known as Salmonella pathogenicity island 1 (SPI-1) (Altier, 2005; Jones, 2005; Lostroh and lee, 2001). Expression of SPI-1 genes is coordinately regulated by several environmental and molecular factors through a regulatory network made of SPI-1-encoded transcriptional activators: HilA, HilC, HilD, and InvF (Altier, 2005; Ellermeier and Slauch, 2007; Jones, 2005). InvF activates transcription of genes that encode effector proteins (Darwin and Miller, 1999; Eichelberg and Galan, 1999). HilA directly activates transcription of genes encoding TTSS components, and indirectly transcription of genes encoding effector proteins by activating transcription of the gene for InvF (Bajaj et al., 1996). HilC and HilD redundantly activate transcription of hilA and invF (Akbar et al., 2003; Olekhnovich and Kadner, 2006; Rakeman et al., 1999; Schechter and Lee, 2001). In addition, HilD can activate hilC transcription and its own transcription, by direct binding to both promoters (Ellermeier et al., 2005; Olekhnovich and Kadner, 2002). That regulatory network serves as link between SPI-1 expression and general regulatory systems encoded elsewhere in the chromosome (Altier, 2005; Ellermeier and Slauch, 2007; Jones, 2005). Most of the regulatory systems known to control SPI-1 primarily target HilD expression, and then regulation is transmitted to the rest of SPI-1 genes (Ellermeier and Slauch, 2007). Surprisingly, those regulatory systems seem to control hilD expression at postranscriptional or postranslational level, rather than at the level of transcription initiation (Ellermeier and Slauch, 2007). In such context, it has been shown that HilD 94 protein is the target for several regulatory systems: HilD is degraded by the ATP- dependent Lon protease (Takaya et al., 2005); HilE, a negative SPI-1 regulator (Fahlen et al., 2000), physically interacts with HilD (Baxter and Jones, 2003), likely interfering with its function; it has been recently proposed that FliZ activates SPI-1 expression by somehow controlling HilD activity (Chubitz et al., 2010). Regulation of hilD expression at mRNA level has also been proposed: overproduction of the RNA binding protein CsrA represses SPI-1 expression (Altier et al., 2000; Martinez et al., 2011), and it has been shown that CsrA binds to a region in hilD mRNA that overlaps with the SD sequence, likely preventing translation and accelerating mRNA decay (Martinez et al., 2011); DNA adenine (Dam) methylation contribute to sustain high levels of SPI-1 expression by decreasing hilD mRNA turnover (López-Garrido and Casadesus, 2010). Hence, postranscriptional control of hilD expression seems to be a key event for SPI-1 regulation. Here we report that hilD mRNA possesses a long 3’unstranslated region (3’UTR) of 310 nucleotides. Long 3’UTR are common in eukaryotes, where they can play regulatory roles, but are not well characterized in prokaryotes. Our results suggest that hilD 3’UTR is a target for hilD mRNA degradation, and is necessary for hilD and SPI-1 regulation by the RNA binding protein Hfq. Thus, hilD 3’UTR may be involved in the postranscriptional integration of signals for SPI-1 regulation. 95 MATHERIALS AND METHODS Bacterial strains, bacteriophages, and standard strain construction: All the Salmonella enterica strains listed in Table 1 belong to serovar Typhimurium, and derive from the mouse virulent strain ATCC 14028. For simplicity, Salmonella enterica serovar Typhimurium is often abbreviated as S. enterica. Targeted gene disruption was achieved using plasmid pKD4 or pKD13 (DATSENKO and WANNER 2000). Antibiotic resistance cassettes introduced during strain construction were excised by recombination with plasmid pCP20 (DATSENKO and WANNER 2000). The oligonucleotides used for disruption (labeled "UP" and "DO") are listed in Table S1, together with the oligonucleotides (labeled "E") used for allele verification by the polymerase chain reaction. For the construction of lac fusions in the Salmonella chromosome, FRT sites generated by excision of Kmr cassettes (DATSENKO and WANNER 2000) were used to integrate either plasmid pCE37 or pCE40 (ELLERMEIER et al. 2002). Unless otherwise specified, all lac fusions used in this study are translational. Addition of 3xFLAG epitope tag to protein-coding DNA sequences was carried out using plasmids pSUB11 (Kmr, 3xFLAG) as template (UZZAU et al. 2001). Transductional crosses using phage P22 HT 105/1 int201 [(SCHMIEGER 1972) and G. Roberts, unpublished] were used for strain construction operations involving chromosomal markers. The transduction protocol was described elsewhere (GARZON et al. 1995). To obtain phage-free isolates, transductants were purified by streaking on green plates. Phage sensitivity was tested by cross-streaking with the clear-plaque mutant P22 H5. Growth conditions Luria-Bertani (LB) broth was used as standard liquid medium. Solid media were prepared by the addition of 1.5 % agar. For determination of expression of SPI-1 genes by β-galactosidase assay, Western blot, or Northern blot, saturated cultures were diluted 1:50 in LB and incubated at 37 ºC with shaking (200 rpm). Samples were taken when the cultures had reached stationary phase (O.D. 2-2.5). When required, Km (50 µg/ml) or Cm (20 µg/ml) were added to the culture medium. Green plates were prepared according to Chan and co-workers (CHAN et al. 1972), except that methyl blue (Sigma Chemical Co, St. Louis, MO) substituted for aniline blue. 96 Construction of relevant strains For construction of hilD ∆3’UTR allele, Kmr gene was amplified from pKD13 with the primers JVO5462 and JVO5463 and inserted in Salmonella chromosome by λRed recombinase-dependent recombination, deleting a 231 nt fragment of hilD 3’UTR starting from the first nucleotide after hilD stop codon. Kmr gene was healed by recombination of flanking FRT sequences mediated by pCP20-encoded FLP recombinase, leaving a scar of 82 nt. As result, a shorter 3’ UTR of 162 nt was produced, keeping the native hilD Rho-independent transcriptional terminator. Expression of hilD from a heterologous promoter was achieved replacing its native promoter by PLtetO promoter (Lutz and Bujard, 1997). A fragment containing the cat gene and PLtetO promoter was amplified by PCR using pXG1 as template (Urban and Vogel, 2007). The primers were labelled PLtetO-hilD UP and PLtetO-hilD DO (Table S1). The PCR product was treated with DpnI to remove template traces. The construction was inserted in the chromosome by λRed recombinase-mediated recombination (Datsenko and Wanner, 2000) and Cmr colonies were selected. Insertion of the construction was verified by PCR, using a couple of primers specific for cat gene and the target gene (Table S1). RNA extraction procedures 2 ml of S. enterica cells reaching stationary phase were taken by centrifugation, and the pellet resuspended in 100 ul of a lysozyme solution (3 mg/ml in water). Cells lysis was facilitated by three consecutive freeze-thaw cycles. After lysis, RNA was extracted using 1ml of Trizol reagent (Invitrogen), according to manufacter’s instructions. Finally, total RNA was resuspended in 30 ul of RNase-free water for subsequent uses. Quality and quantity of the obtained RNA was determined using a Nanodrop instrument. 3’RACE 3’RACE was performed as described by Argaman and collaborators (2001). The hilD- specific primer used for PCR was JVO5536. Specific RCR products were cloned in pTOPO vector, and 4 independent clones were sequenced using external primers. 97 Northern blots in polyacrylamide gels 10 µg of total RNA resuspended in 2X RPA loading buffer (98 % Formamid, 1 mM EDTA, 0.1 % Xylene cyanole, 0.1 % Bromphenol blue) were loaded in a 4.5 % polyacrylamide 7 M urea gel, prepared in TBE, and solidified with 0.08 % (w/V) APS and 0.106 % (V/V) TEMED. Electrophoresis was carried out at room temperature in TBE buffer, at 300 V for 4 hours. RNA was transferred onto Hybond-XL membranes (Amersham) by wet electrophoresis in TBE buffer (50 V, 1 h, 4 ºC), and crosslinked with UV light. hilD mRNA was detected using a 32P-labelled riboprobe specific for the first 300 nt of hilD coding sequence. Northern blot in agarose gels For northern blot analysis, 10 µg of total RNA was loaded per lane and electrophoresed in denaturing 1% agarose formaldehyde gels. Transfer and fixation to Hybond-N+ membranes (GE Healthcare) were performed by vacuum using 0.05M NaOH. Filters were then hybridized using an internally labelled ([α-32P]UTP) riboprobe specific for the first 300 nts of the hilD coding sequence. Hybridization was carried out at 65°C. As a control of RNA loading and transfer efficiency, the filters were hybridized with a riboprobe of the RNase P RNA gene (rnpB). Images of radioactive filters were obtained with a Fuji, and quantification was performed using the Multy Gauge software. β-galactosidase assays Levels of β-galactosidase activity were assayed using the CHCl3-sodium dodecyl sulfate permeabilization procedure (MILLER 1972). Unless otherwise indicated, the results shown are the average and standard deviation of 3 independent experiments. Protein extracts and Western blot analysis Total protein extracts were prepared from bacterial cultures grown at 37°C in LB until stationary phase (final O.D.600 ~2.5). Bacterial cells taken according to 1 O.D.600 were collected by centrifugation (16,000 g, 2 min, 4°C) and suspended in 100 μl of Laemmli sample buffer [1.3% SDS, 10% (v/v) glycerol, 50 mM Tris-HCl, 1.8% βmercaptoethanol, 0.02% bromophenol blue, pH 6.8]. Proteins were resolved by Tris- 98 Tricine-PAGE, using 12% gels. Conditions for protein transfer have been described elsewhere (JAKOMIN et al. 2008). Optimal dilutions of primary antibodies were as follows: anti-Flag M2 monoclonal antibody (1:5,000, Sigma Chemical Co, St. Louis, MO), and anti-GroEL polyclonal antibody (1:20,000, Sigma). Goat anti-mouse horseradish peroxidase-conjugated antibody (1:5,000, BioRad, Hercules, CA) or Goat anti-rabbit horseradish peroxidase conjugated antibody (1:20,000, Santa Cruz Biotechnology, Heildelberg, Germany) were used as secondary antibodies. Proteins recognized by the antibodies were visualized by chemoluminescence using the luciferinluminol reagents. 99 RESULTS hilD mRNA has a long 3’UTR hilD transcription start point has been identified previously (Olekhnovich and Kadner, 2002), and leaves a 5’UTR of 35 nt in hilD mRNA. As postranscriptional control of hilD expression seems to be a key event in SPI-1 regulation, we decided to determine hilD transcription termination point. Sequence analysis with RNAfold software predicts the existence of a secondary structure similar to a Rho-independent transcriptional terminator around 300 nt downstream hilD stop codon, consisting on a stem of 12 nt with a 1-nt bulge in its 3’ side, and a hairpin loop of 4 nt, followed by a U-rich region (Figure 1A). That fits with the proposed structure for a Rho independent transcriptional terminator (Carafa et al., 1990; Lesnik et al., 2001). We experimentally determined hilD transcription stop point by 3’RACE. As shown in Figure 1B, hilD transcription stops right at the end of the U-rich region downstream the predicted steam-loop, suggesting that such structure constitute a functional Rho-independent transcriptional terminator. According to that, of hilD mRNA should be 1275 nt length. In order to estimate the size of hilD mRNA, we detected hilD mRNA by Northern blot in a 4.5 % polyacrylamide gel, using a P32-labelled riboprobe specific for hilD. Comparison of hilD size mRNA with pUC8 DNA marker (Fermentas) shows that full-length hilD RNA runs slower than the higher band in the marker (1118 nt), supporting the expected size of hilD mRNA (Figure 1C). Hence, hilD mRNA molecule includes a 5’UTR of 35 nt, a coding sequence of 930 nt, and an unusually long 3’UTR of 310 nt (Figure 1D). Deletion of hilD 3’UTR results in increased levels of hilD mRNA Although it is known that 3’UTRs are important for postranscriptional regulation in eukaryotes (Grzybowska et al., 2001), long 3’UTRs are poorly studied in prokaryotes. We considered the possibility that hilD 3’UTR was playing a role in postranscriptional control of hilD expression. Thus, we constructed a strain carrying a modified hilD 3’UTR (hilD ∆3’UTR): a 231-nt fragment of hilD 3’UTR starting from the first nucleotide after hilD stop codon was deleted, leaving the Rho-independent transcriptional terminator intact. The deleted fragment was exchanged by the 82-nt pKD4 scar, resulting in a shorter 3’UTR with a different sequence, keeping the native 100 hilD Rho-independent transcriptional terminator (Figure 2A). We detected hilD mRNA in RNA extracts from wild type and hilD∆3’UTR strains by Northern blotting in agarose gels, using a P32-labelled riboprobe specific for the first 300 nt of hilD coding sequence. As shown in Figure 2B, hilD ∆3’UTR mRNA is shorter than hilD native mRNA, thereby confirming the existence of a long 3’UTR in hilD mRNA. In addition, the level of hilD mRNA is around 11 folds higher in the absence of its 3’UTR than in wild type strain, suggesting that the presence of the 3’UTR somehow represses hilD expression. Since hilD 3’UTR is part of hilD mRNA, it is conceivable to speculate that it may control hilD expression at postranscriptional level. In order to confirm that, we examined the level of hilD mRNA in isogenic hilD 3’UTR+ and hilD 3’UTRˉ strains in which hilD was transcribed from PLtetO promoter (Figure 2C). The absence of a native 3’UTR increases hilD mRNA level 3.5 folds even if hilD is transcribed from a heterologous promoter, indicating that 3’UTR influences hilD expression at postranscriptional level. Note that hilD overexpression upon 3’UTR deletion is higher when hilD is transcribed from its own promoter than when transcription is driven by PLtetO, suggesting that autoactivation of hilD transcription may amplify hilD overexpression in the absence of its 3’UTR. SPI-1 is overexpressed in hilD ∆3’UTR hosts The above results show that hilD mRNA is overproduced in the absence of its native 3’UTR. We wandered if such overproduction was functional, and triggered overexpression of the whole SPI-1. We examined the expression of 4 SPI-1 genes (invF, invH, sipB, and sipC) in isogenic hilD 3’UTR+ and hilD 3’UTRˉ strains by two independent methods: (i) measurement of β-galactosidase activity of invF::lac, invH::lac, sipB::lac, and sipB::lac fusions (Figure 2D); and (ii) determination of InvF, InvH, SipB, and SipC levels by Western blotting, using protein variants tagged with the 3xFLAG epitope (Figure 2E). β-galactosidase activities are around 3-4 folds higher in hilD 3’UTRˉ than hilD 3’UTR+ hosts, for the 4 genes analyzed. Similarly, InvF, InvH, SipB and SipC levels are also from 2.6 to 9.6 folds higher in hilD 3’UTRˉ strains, compared to hilD 3’UTR+ background. Hence, hilD mRNA overproduction observed in the absence of its native 3’UTR correlates with SPI-1 overexpression, suggesting that hilD 3’UTR may be a target to control SPI-1 expression. 101 Inactivation of RNA-degradosome components ribonuclease E and polynucleotide phosphorylase suppresses hilD overexpression in the absence of its 3’UTR In eukaryotes, mRNA degradation usually depends on their long 3’UTRs (Barreau et al., 2005; Beelman and Parker, 1995). We hypothesized that the long 3’UTR of hilD mRNA might be a target for mRNA degradation. If such were the case, we reasoned, inactivation of factors involved in RNA turnover might suppress the differences observed in the level of hilD mRNA with and without its native 3’UTR. We constructed deletion mutants in genes encoding the endonucleases ribonuclease E (RNase E) and ribonuclease G (RNase G), the exonuclease polynucleotide phosphorylase (Pnp), and the poly(A) polymerase I (PapI). Since RNase E is essential in Salmonella, we only removed the C-terminal region of the protein (Viegas et al., 2007), involved in the assembly of RNA degradosome (Carpousis, 2002). hilD and hilD ∆3’UTR mRNA levels were determined in strains lacking RNase E, RNase G, Pnp, or PapI (Figure 3). hilD ∆3’UTR mRNA levels are higher than hilD mRNA levels in control, RNase Gˉ and PapIˉ backgrounds, suggesting that hilD 3’UTR is not involved in hilD mRNA degradation by RNase G or polyadenilation-dependent pathways. However, similar levels of hilD and hilD ∆3’UTR mRNAs are detected in strains lacking either RNase E or Pnp. Those two proteins are component of the RNA degradosome (Carposius, 2002), suggesting that hilD 3’UTR may be a target for mRNA degradation by RNA degradosome. Hfq regulates hilD expression at posttranscriptional level We considered the possibility that hilD 3’UTR was involved on SPI-1 regulation by some of the signals that control SPI-1 expression. In that sense, we looked at the RNA binding protein Hfq. It has been reported that SPI-1 expression is repressed in mutants lacking Hfq (Sittka et al., 2007). In addition, Hfq CoIP experiments have determined that Hfq binds to hilD mRNA (Sittka et al., 2008). In order to determine if hilD expression was regulated by Hfq, we examined the level of hilD mRNA in Hfq+ and Hfqˉ backgrounds by Northern blotting (Figure 4A). Lack of Hfq reduces the level of hilD mRNA almost 5 folds. However, the expression of a hilD::lac transcriptional 102 fusion is similar in Hfq+ and Hfqˉ backgrounds (Figure 4B), suggesting that the regulation is postranscriptional. That is further supported by the observation that the level of hilD mRNA is still Hfq-dependent when hilD is transcribed from a heterologous promoter (Figure 4C). Those results support the idea that Hfq may control hilD expression by direct binding to hilD mRNA molecule. Deletion of hilD 3’UTR suppresses regulation of hilD expression by Hfq Interestingly, RNA fragments belonging to hilD 3’UTR were enriched after Hfq CoIP, suggesting that Hfq may directly bind that region (Sittka et al., 2008). That opens the interesting possibility that hilD 3’UTR is required for the regulation of hilD expression by Hfq. To test that hypothesis, we analyzed the levels of hilD mRNA and hilD ∆3’UTR mRNA in Hfq+ and Hfqˉ backgrounds. As shown in Figure 4D, the level of hilD mRNA is 5-fold reduced in Hfqˉ compared to Hfq+ background. However, similar levels of hilD and hilD ∆3’UTR mRNAs are detected in Hfq+ and Hfqˉ backgrounds, indicating that the 3’UTR of hilD mRNA is important for regulation of hilD expression by Hfq. SPI-1 regulation by Hfq is transmitted via HilD It has been previously reported that SPI-1 expression is repressed in hfq mutants (Sittka et al., 2007), and that correlates with reduced levels of hilD mRNA in Hfqˉ background (Figure 4A, C). Since HilD is one of the main transcriptional activators of SPI-1 expression, it is conceivable that Hfq regulates SPI-1 expression through HilD. To study that possibility, we examined regulation by Hfq of selected SPI-1 genes in a set of mutants lacking the main SPI-1-encoded transcriptional activators (HilA, HilC, or HilD). SPI-1 expression was monitored using lac fusions in representative genes. βgalactosidase activity of each fusion was determined in Hfq+ and Hfqˉ strains, containing all SPI-1-encoded transcriptional activators (Control), or lacking individual activators. As shown in Figure 5A, β-galactosidase activities of hilA::lac, invF::lac, sipB::lac and invH::lac fusions are reduced in strains lacking HilC, but their expression is still Hfq-dependent. Similarly, deletion of hilA reduces expression of invF::lac and sipB::lac fusions, but Hfq-dependent regulation is still observed. Those results indicate 103 SV6797 ∆hfq::cat ∆hilD (invF’-lacZ+)(Hyb) This study SV5852 ∆hfq::cat (sipB’-lacZ+)(Hyb) This study SV5316 ∆hilA (sipB’-lacZ+)(Hyb) This study SV6798 ∆hfq::cat ∆hilA (sipB’-lacZ+)(Hyb) This study SV5318 ∆hilC (sipB’-lacZ+)(Hyb) This study SV6799 ∆hfq::cat ∆hilC (sipB’-lacZ+)(Hyb) This study SV5312 ∆hilD (sipB’-lacZ+)(Hyb) This study SV6800 ∆hfq::cat ∆hilD (sipB’-lacZ+)(Hyb) This study SV6801 ∆hfq::cat (invH’-lacZ+)(Hyb) This study SV5419 ∆hilA (invH’-lacZ+)(Hyb) This study SV6802 ∆hfq::cat ∆hilA (invH’-lacZ+)(Hyb) This study SV5417 ∆hilC (invH’-lacZ+)(Hyb) This study SV6803 ∆hfq::cat ∆hilC (invH’-lacZ+)(Hyb) This study SV5415 ∆hilD (invH’-lacZ+)(Hyb) This study SV6804 ∆hfq::cat ∆hilD (invH’-lacZ+)(Hyb) This study SV5384 (hilC’-lacZ+)(Hyb) This study SV6805 ∆hfq::cat (hilC’-lacZ+)(Hyb) This study SV5386 ∆hilD (hilC’-lacZ+)(Hyb) This study SV6806 ∆hfq::cat ∆hilD (hilC’-lacZ+)(Hyb) This study SV5854 ∆hfq::cat (sipC’-lacZ+)(Hyb) This study SV6641 ∆hfq::cat hilD ∆3’UTR (invF’-lacZ+) This study SV6647 ∆hfq::cat hilD ∆3’UTR (sipB’-lacZ+) This study SV6652 ∆hfq::cat hilD ∆3’UTR (sipC’-lacZ+) This study SV6643 ∆hfq::cat invF::3xFLAG This study SV6644 ∆hfq::cat hilD ∆3’UTR invF::3xFLAG This study SV6649 ∆hfq::cat sipB::3xFLAG This study SV6650 ∆hfq::cat hilD ∆3’UTR sipB::3xFLAG This study SV6654 ∆hfq::cat sipC::3xFLAG This study SV6655 ∆hfq::cat hilD ∆3’UTR sipC::3xFLAG This study SV5298 ∆dam-231 (invF’-lacZ+)(Hyb) This study SV6228 hilD ∆3’UTR ∆dam-231 (invF’-lacZ+)(Hyb) This study 110 UUC C G G C U U U A A C C C U G U G GA U U A A G G C C G G C A U U U U A U U5’- -3’ UAAcauuuuuuguaucugucacuuaaguaaagauuuu uauuaaaauuguaauaauuuaaaauucagacugcgca uuaacacgcucuaucaggaugggaggcuauucaauau cauuguucuguccggaagacagcuuauacugauaucu augguaauuuaaaguaaggcugauuauauaacacgau uuuugugaacuugucaucgcuaugaugacugguaaaa cgauauugccuuauucacagcguaagaauucguccag augacacuaucuccuuccggcuuuaacccuguggauu aaggccggcauuuuauu Transcription stop point 1118 nt 881 nt 692 nt 501 nt 489 nt 404 nt 5’UTR (35 nt) 3’UTR (310 nt)Coding sequence (930 nt) Figure 1 AB CD 111 5’UTR (35 nt) 3’UTR (310 nt)Coding sequence( 930 nt) 5’UTR (35 nt)∆3’UTR (162 nt)Coding sequence( 930 nt) hilD mRNA hilD ∆3’UTR mRNA 1 11,2 ∆3’UTRWT Loading control (rnpB) Quantification 1 2,6 InvF-3xFLAG ∆3’UTRWT 1 9,6 InvH-3xFLAG ∆3’UTRWT 1 4,0 SipB-3xFLAG ∆3’UTRWT 1 2,9 SipC-3xFLAG ∆3’UTRWT Loading control (GroEL) Tagged protein Quantification 0 100 200 300 400 Relative β-galactosidase activity invF::lac invH::lac sipB:lac sipC:lac ∆3’UTRWT PLtetO-hilD Loading control (rnpB) Quantification Figure 2 A BC D E 112 1 9.8 1 0.81 1.5 Loading control (rnpB) 3’UTRˉ3’UTR+ Control Pnp_ RNase E_ 3’UTRˉ3’UTR+3’UTRˉ3’UTR+3’UTRˉ3’UTR+3’UTRˉ3’UTR+ PapI_ RNase G_ 1 17.1 1 15.0 Figure 3 113 1 13.2 0.2 13.6 WT hfq∆3’UTR ∆3’UTR hfq Loading control (rnpB) 0 50 100 150 200 β-galactosidase activity WT T hfq PhilD WT hfq PLtetO Figure 4 A B CD hilD::lac Quantification 114 0 500 1000 1500 2000 β-galactosidase activity 0 100 200 300 400 0 100 200 300 400 0 5000 10000 15000 0 100 200 300 400 β-galactosidase activity β-galactosidase activity β-galactosidase activity β-galactosidase activity ∆hilA ∆hilC ∆hilD ∆hilA ∆hilC ∆hilD ∆hilC ∆hilD ∆hilA ∆hilC ∆hilD ∆hilD Control Control Control Control Control hilA::lac invF::lac sipB::lac hilC::lac invH::lac 0 200 400 600 800 0 10000 20000 30000 40000 0 10000 20000 30000 40000 100 2.0 222.5 189.9 Loading control (GroEL) Tagged protein WT hfq ∆3’UTR ∆3’UTR hfq 100 3.8 279.7 408.0 Loading control (GroEL) Tagged protein WT hfq ∆3’UTR ∆3’UTR hfq 100 22.8 260.6 215.2 Loading control (GroEL) Tagged protein WT hfq ∆3’UTR ∆3’UTR hfq SipC-3xFLAG SipB-3xFLAG InvF-3xFLAG β-galactosidase activity β-galactosidase activity β-galactosidase activity Figure 5 A B C Quantification Quantification Quantification 3’UTR+3’UTR_ 3’UTR+3’UTR_ 3’UTR+3’UTR_ invF::lac sipB::lac sipC::lac 115 Table S1. Oligonucleotides used in this study (5' 3') Oligonucleotide Sequence hilDriboprobeUP atggaaaatgtaacctttgtaag hilDriboprobeDO gtttttttaatacgactcactatagggaggtatatcgaaatccatgtggc rnpBriboprobeUP gaagaagtgaaactgaccgataagc rnpBriboprobeDO taatacgactcactataggccgaagctgaccagacagtcg hilD-E2’ atcatcctcaggctggctcc RT-hilD3’-UP agcttacggatgttgccgatc JVO5462 aactacgccatcgacattcataaaaatggcgaaccattaaattccggggatccgtcgacc JVO5463 ggagatagtgtcatctggacgaattcttacgctgtgaatagtgtaggctggagctgcttc rngUP accgaattcgcgtcgactcacgtctgacctatgagtcgctcatatgaatatcctccttag rngDO cggctgatggccactgtaatgctccagcttactggtcatttgtaggctggagctgcttcg rng-E1 ggtgcatacggtattcaggg rng-E2 tcgtcgagattacggtgtcc pcnBUP gcccgtctgccgtaatgcgaaagacatagagtaagttgatcatatgaatatcctccttag pcnBDO tgggatgttccatcagtttccaggcgcgtttgccctgacgtgtaggctggagctgcttcg pcnB-E1 cgctgagctatgattagccg pcnB-E2 tccggtctaatgacgcaagc 116 0 100 200 300 400 500 600 β-galactosidase activity invF::lac 3’UTR+3’UTR_ 117 Inhibition of Salmonella enterica invasion by L-arabinose Javier Lopez-Garrido, Elena Puerta-Fernández, and Josep Casadesus 118 ABSTRACT Invasion of intestinal epithelial cells is a critical step in Salmonella infection. Invasion requires the expression of genes located the Salmonella Pathogenicity Island 1 (SPI-1). SPI-1 expression is controlled by environmental factors allowing high expression of SPI-1 in the ileum, the portion of the small intestine where invasion takes place. Several evidences suggest that the pentose L-arabinose is an specific signal for SPI-1 repression: (i) the expression of SPI-1 genes is repressed in the presence of L-arabinose, even at low concentration; (ii) that effect is independent of L-arabinose metabolism, since it is observed in a mutant lacking L-arabinose isomerase, the first enzyme involved in L-arabinose degradation; and (iii) addition of other pentoses such as D- arabinose and D-xylose does not repress SPI-1 expression. Deletion of the gene encoding the L-arabinose permease, araE, suppresses SPI-1 repression by L-arabinose, suggesting that L-arabinose needs to be inside the cell to repress SPI-1. Once inside the cell, L-arabinose represses SPI-1 independently of the L-arabinose-responsive regulator AraC. Epistasis analysis shows that SPI-1 repression by L-arabinose depends on the SPI-1-encoded transcriptional activator HilD, and our results suggest that L-arabinose controls HilD activity. L-arabinose is produced by plants and is poorly absorbed in the digestive tract of mammals and chicks. The presence of sugars that can be metabolized by Salmonella enterica in the small intestine might be a signal for inhibition of invasion. L-arabinose might also play a signaling role during the recently discovered colonization of plants by Salmonella. 119 RESULTS SPI-1 expression is downregulated in the presence of L-arabinose pBAD vectors are a set of plasmids which contain the PBAD promoter of the L-arabinose operon and its regulatory gene, araC (Guzman et al., 1995). In the presence of L- arabinose, transcription from PBAD promoter is turned on, and in its absence transcription levels are very low. That provides a system that allows conditional expression of genes cloned under the control of PBAD promoter, dependent on L- arabinose (Guzman et al., 1995). We considered the possibility of using the pBAD vectors for studying how overproduction of different proteins affected SPI-1 expression. Since expression from PBAD promoter depends on L-arabinose, we decided to examine whether SPI-1 expression was affected by the presence of L-arabinose in a strain without pBAD as a control. For this purpose, we studied the expression level of an invF::lac fusion in LB and in LB plus 0.2 % of L-arabinose. As shown in Figure 1 A (left panel), invF::lac expression is reduced around 4 folds in LB plus L-arabinose compared to LB. That observation suggests that the presence of L-arabinose in the culture medium somehow represses the expression of SPI-1 genes. SPI-1 repression in the presence of L-arabinose is independent of L-arabinose catabolism Luria-Bertani (LB) broth contains a low concentration of sugars fermentable by Salmonella. The carbon sources for Salmonella in LB are catabolizable amino acids, not sugars (Sezonov et al., 2007). In such conditions, bacterial metabolism undergoes gluconeogenesis, and culture-medium pH is slightly alkalinized during growth. However, if we add a catabolizable sugar to LB exogenously, Salmonella will use it as a carbon source, and bacterial metabolism will undergo glycolysis. As a consequence, pyruvic acid will be produced and culture medium will be acidified (Figure S2). Salmonella enterica is able to regulate SPI-1 expression in response to pH. It has been reported that SPI-1 expression is almost 10 folds higher at pH 8 than at pH 6 (Bajaj et al., 1996). Thus, we considered the possibility that repression of SPI-1 expression in the presence of L-arabinose could be an indirect effect due to the acidification of the culture medium. We measured pH of LB and LB plus 0.2 % of L-arabinose before inoculation 126 of bacteria (before growth) and when bacterial population had reached an OD600 of 2.5 (after growth) (Figure 1 B). Before growth, both LB-pH and the pH of LB plus L- arabinose were near neutrality (7.04 and 7.08 respectively). However, after bacterial growth pH of LB was slightly alkaline (7.56) while pH of LB plus L-arabinose was 4.97. That suggests that catabolism of L-arabinose by Salmonella acidifies the culture medium. Hence, SPI-1 repression in the presence of L-arabinose could be a consequence of culture medium acidification. To address that hypothesis, we used a mutant with a deletion in the araA gene which encodes the L-arabinose isomerase, the first enzyme involved in L-arabinose catabolism (Englesberg, 1961). That mutant is unable to metabolize L-arabinose and cannot grow in minimal medium with L- arabinose as sole carbon source (Figure S3). In addition, culture-medium pH is not acidified after growth in the presence of 0.2 % of L-arabinose (pH of 7.56 compared to 7.49 in LB without L-arabinose) (Figure 1 B, Figure S3). Thus, if SPI-1 repression in the presence of L-arabinose was due to culture medium acidification, it would not be observed in the araA mutant. However, to our surprise, an invF::lac fusion is repressed more than 10 folds in the presence of 0.2 % of L-arabinose (Figure 1 A, right panel). That suggests that is L-arabinose itself rather than culture medium acidification what is responsible for SPI-1 repression. Indeed, repression is bigger in the araA mutant than in wild type background (10 folds compared to 4 folds) what could reflect a reduction in L-arabinose concentration due to its utilization as carbon source by AraA+ bacteria. Taken together, those results open the possibility that L-arabinose is a signal for SPI-1 repression. In order to avoid indirect effects due to culture medium acidification, the following experiments were done in AraAˉ, unless otherwise indicated. SPI-1 in general is repressed by L-arabinose The above results show that expression of an invF::lac fusion is downregulated in the presence of L-arabinose. We wanted to know if such downregulation affected just to a subset of SPI-1 genes, or if the whole island was repressed by L-arabinose. To address that question, we analyzed the expression of different SPI-1 genes (Figure 2). We selected genes encoded in separated transcriptional units, and representing different functional categories. The selected genes were hilA, encoding a key transcriptional 127 activator of SPI-1 expression; invF, encoding a transcriptional activator of a subset of SPI-1 genes; sipB, encoding an effector protein; and prgH, encoding a component of the type-three secretion system. We analyzed the expression of those genes in LB and in LB plus 0.2% of L-arabinose by two independent methods: (i) by measuring βgalactosidase activity of lac fusions in selected genes (Figure 2 A); and (ii) by determining protein levels by western blot using protein versions tagged with 3xFLAG epitope (Figure 2 B). In order to avoid acidification of the culture medium due to L- arabinose catabolism, the experiments were carried out in AraAˉ background. Figure 2 shows that both, β-galactosidase activity and protein level are reduced in the presence of L-arabinose in the four genes studied (from 3 to almost 20 folds, depending on the gene and the method used for determination of expression). That suggests that L-arabinose represses SPI-1 in general, affecting to different transcriptional units and functional categories. It has been previously shown that HilA plays a central role in the co-ordinate regulation of invasion genes by environmental and regulatory conditions (Bajaj et al 1996). hilA is strongly repressed in the presence of L-arabinose (more than 10 folds in β-galactosidase assays and almost 20 folds in western blot assays). For that reason, in the following experiments we will use hilA as a reporter of SPI-1 expression. Repression of SPI-1 expression by L-arabinose is dose-dependent So far we have been using a concentration of L-arabinose of 0.2 % (equivalent to 13.3 mM). That concentration has been traditionally used for induction of PBAD expression (Guzman et al, 1995), but lower concentrations also work. If L-arabinose is a signal for SPI-1 repression it could have a repressive effect at concentrations lower than 0.2 %. A hilA::lac fusion and a HilA-3xFLAG protein version were used to determine SPI-1 repression with increasing concentrations of L-arabinose by β-galactosidase assay and western blot respectively (Figure 2 C; Figure 2 D). As above, the experiments were carried out in AraAˉ background to avoid culture-medium acidification. A gradual decrease of hilA expression as L-arabinose concentration increases is observed both by β-galactosidase assay (Figure 2 C) and by western blot analysis (Figure 2 D). Note that at the lower concentration of L-arabinose assayed (0.01 %, equivalent to 66.7 µM) hilA 128 expression is still significantly reduced (more that 3 folds in β-galactosidase assay and more than 2 folds in western blot), further supporting the idea that L-arabinose is a signal for SPI-1 repression. Effect of other pentoses on SPI-1 expression L-arabinose is a pentose metabolizable by Salmonella (Gutnick et al., 1969). We wandered if other pentoses were able to repress SPI-1 expression or if the effect was specific of L-arabinose. Thus, we studied the effect of two additional pentoses: D- xylose and D-arabinose. (i) D-xylose is the most abundant pentose in nature, and can be used by Salmonella as sole carbon source (Gutnick et al., 1969). The first enzyme involved in D-xylose catabolism is D-xylose isomerase, encoded by the xylA gene (Shamanna and Sanderson, 1979). To examine the effect of D-xylose on SPI-1 expression, we constructed a strain with a deletion in the xylA gene. The xylA mutant is unable to use D-xylose as sole carbon source, avoiding the change of culture-medium pH due to D-xylose catabolism (Figure S4). We monitored the expression of a hilA::lac fusion on a XylAˉ background, grown on LB, or LB plus 0.2 % or 1 % of D-xylose. As shown in Figure 3 B, the presence of D-xylose in the culture medium does not have a significant effect on hilA::lac expression. However, the same fusion is strongly repressed in the presence of either 0.2 % or 1 % of L-arabinose (Figure 3 A). (ii) D-arabinose is the D isomer of arabinose and it cannot be used by Salmonella as sole carbon source (Gutnick et al., 1969). However, some Salmonella mutants display a D-Ara+ phenotype (Old and Morlock, 1977). In such mutants, D-arabinose induces the synthesis of two enzymes usually involved in L-fucose metabolism: L-fucose isomerase and L-fuculokinase. Those enzymes have a bifunctional activity, and can convert D- arabinose into D-ribulose I phosphate. Those observations imply that, unless it cannot be metabolized by wild type Salmonella, D-arabinose may enter wild type Salmonella cells. Figure 3 C shows that hilA::lac expression is similar in LB and LB plus 0.2 % or 1 % of D-arabinose, suggesting that it is not involved in SPI-1 regulation. Thus, neither D-xylose nor D-arabinose is able to repress SPI-1 expression, as happens with L-arabinose. That suggests that the effect of L-arabinose is specific. 129 Intracellular L-arabinose is responsible for SPI-1 repression In an attempt to understand the mechanism of SPI-1 repression by L-arabinose, we started by analyzing if L-arabinose could exert is repressive effect from outside the cell, or if it needed the cell. In Salmonella, L-arabinose enters the cell through a specific permease encoded by the araE gene (Lee et al., 1981; Lee et al., 1982). We constructed an araE mutant. That mutant cannot grow in minimal medium plus 0.1 % of L- arabinose as sole carbon source, grows slowly in the presence of 0.2 %, and grows as wild type strain in the presence of 1 % of L-arabinose (Figure S3). That suggests that AraE permease is essential for transport of L-arabinose when the extracellular concentration is low. However, when we raise the extracellular concentration, L- arabinose may enter by other routes. With that background, we made the following reasoning: if L-arabinose needed to be inside the cell to repress SPI-1, the repression would be suppressed in an araE mutant at low L-arabinose concentrations. However, as we increased the extracellular concentration of L-arabinose, it would enter the cell through AraE-independent pathways, and SPI-1 would be repressed. If, on the contrary, extracellular L-arabinose could repress SPI-1 expression, the repression would be similar in AraE+ and AraEˉ backgrounds at low and high L-arabinose concentrations. We examined the expression of a hilA::lac fusion in isogenic araA and araA araE backgrounds. The strains were grown in LB and LB plus 0.2; 1; 2; or 3 % of L- arabinose (Figure 4 A, two first columns). araE mutation almost completely suppresses hilA repression by L-arabinose at 0.2 %. However, in the presence of higher concentrations, hilA is significantly repressed. Nevertheless, repression is higher in AraE+ than in AraEˉ background in all L-arabinose concentrations assayed. To be sure that the repression observed at high L-arabinose concentrations was not due to the increase in osmolarity, we examined hilA::lac expression in the presence of up to 3 % of sucrose. hilA::lac expression was not repressed under any of the sucrose concentrations assayed (Figure S5) Those results indicate that L-arabinose needs to be inside the cell in order to repress SPI-1 expression. SPI-1 repression by L-arabinose is AraC-independent L-arabinose binds to AraC in bacterial cytoplasm. That binding modifies AraC structure, which can then activate the expression of genes involved in L-arabinose 130 catabolism, including araE (Scheilf 2010). The evidence that L-arabinose needs to be inside the cell to repress SPI-1 expression opens an interesting possibility: it might be possible that activation of AraC by L-arabinose in bacterial cytoplasm was directly or indirectly responsible for SPI-1 repression. To study that hypothesis, we analyzed repression by L-arabinose of a hilA::lac fusion in isogenic araA and araA araC backgrounds, in the presence of increasing concentrations of L-arabinose. Mutation of araC suppresses hilA repression by 0.2 % L-arabinose (Figure 4 A, compare first and third column). However, hilA expression is increasingly repressed as L-arabinose concentration increases. The repression pattern is very similar to the pattern observed in the araE mutant (Figure 4 A, second column). As AraC activates araE transcription, the suppression observed in AraCˉ background could be due to a decrease of intracellular L-arabinose due to lack of AraE permease. To circumvent that problem we did two different approximations: (i) If suppression of SPI-1 regulation by L-arabinose in AraCˉ background was due to lack of AraE permease, mutation of araE would be epistatic over araC mutation. Thus, the double mutant araE araC would display the same phenotype than the araE single mutant. However, if the effect of araC mutation was independent of AraE permease, araE and araC mutations would have additive effects. As shown in Figure 4 A araE araC double mutant display the same phenotype than araE single mutant. (ii) We constructed a strain that express araE constitutively, independently of AraC. To achieve that we placed PLtetO promoter (Lutz and Bujard, 1997) upstream araE gene, replacing its native promoter. That strain grows similarly to wild type in minimal medium plus 0.1 % of L-arabinose, suggesting that AraE permease is being produced (Figure S6). When araE is expressed constitutively, mutation of araC no longer suppresses hilA repression by L-arabinose (figure 4 B). The above results suggests that, once inside the cell, L-arabinose represses SPI-1 expression through an AraC-independent pathway 131 Evidence for SPI-1 repression at low intracellular concentration of L-arabinose We have previously shown that a concentration of L-arabinose of 0.01 % in the culture medium was sufficient to significantly repress hilA expression. However, the fact that the production of AraE permease depends on the presence of L-arabinose produces a curious effect: cells that randomly transport L-arabinose will produce more AraE permease that will transport of more L-arabinose inside the cell. However, if L- arabinose does not enter the cell, production of AraE permease will not be induced and L-arabinose will not be transported. Thus, in the presence of L-arabinose there are two subpopulations in a Salmonella culture: one producing AraE permease and with a high level of intracellular L-arabinose, and another with low levels of AraE permease and intracellular L-arabinose (Figure 5 A). What changes when bacteria are cultured with different concentrations of L-arabinose is not the intracellular concentration of L- arabinose of single cells, but the proportion of cells of each subpopulation (Siegele and Hu, 1997; Figure 5 A). Hence, it is conceivable that SPI-1 repression by L-arabinose actually requires high concentrations intracellular L-arabinose. To circumvent that potential problem, we used the strain that expresses constitutively araE, independently of the presence of L-arabinose (described above). It has been reported that constitutive expression of araE avoid the formation of subpopulations in the presence of L- arabinose, and intracellular L-arabinose concentration increases gradually with extracellular concentration (Khlebinkov et al., 2001; Figure 5 A). We examined the expression of a hilA::lac fusion in isogenic araA and araA PLtetO-araE backgrounds, cultured in the presence of increasing concentrations of L-arabinose (Figure 5 B). Both strains shows a repression of more than two folds at the minimal L-arabinose concentration assayed (0.005 %, equivalent to 33 µM), and the repression is bigger as the concentration of L-arabinose increases. That suggests that low intracellular concentrations of L-arabinose are sufficient to significantly repress SPI-1 expression. L-arabinose-dependent expression of SPI-1 is transmitted via HilD SPI-1 expression is controlled by a regulatory network of SPI-1-encoded transcriptional activators. On top of the network are the transcriptional activators HilA, HilC and HilD (Figure S1). We have shown that hilA expression is regulated by L-arabinose. However, that regulation might not by direct, and could be transmitted via HilC and/or 132 HilD. Mutation of hilC reduces hilA expression 2-3 folds (Lopez-Garrido and Casadesus, 2010), while addition of 0.2 % of L-arabinose to the culture medium represses hilA more than 10 folds. That makes unlikely that HilC is responsible for transmission of L-arabinose regulation. However, mutation of hilD reduces hilA expression more than 100 folds (Lopez-Garrido and Casadesus, 2010). Thus, we examined the possibility that L-arabinose regulated SPI-1 expression via HilD. To do that, we analyzed the regulation by L-arabinose of three HilD-activated genes (hilC, rtsA, and invH) using lac fusions, in AraAˉ HilD+ and AraAˉ HilDˉ backgrounds (Figure 6) Because lac fusions in those genes have disparate β-galactosidase activities, the activity of each fusion was normalized to 100 in the HilD+ background growing LB. Expression of all of them is reduced in the presence of L-arabinose in HilD+. In the absence of HilD, expression of those genes is reduced but not completely abolished, and similar levels of β-galactosidase activity are detected in LB and LB plus L-arabinose (Figure 6). The lost of L-arabinose repression in a hilD mutant provides evidence that L-arabinose-dependent regulation of SPI-1 requires a functional hilD gene. L-arabinose regulates hilD expression The evidence that L-arabinose regulates SPI-1 expression through HilD suggests that hilD expression itself may be controlled by L-arabinose. We monitored the expression of two different hilD::lac fusion in LB and LB plus 0.2 % L-arabinose: (i) hilD::lac1, a transcriptional fusion in which lacZ has been inserted right in the transcription start point; and (ii) hilD::lac477, a translational fusion inserted at position 477 of hilD coding sequence (Figure 7 A). To our surprise, similar β-galactosidase levels were observed in the absence and in the presence of L-arabinose in both fusions. That suggests that L-arabinose does not regulate neither hilD transcription nor translation initiation. We then analyzed the level of hilD mRNA by northern blot (Figure 7 B). The amount hilD mRNA is reduced in the presence of L-arabinose. That indicates that L- arabinose regulates hilD expression. In addition, the regulation might be at posttranscriptional level. 133 L-arabinose regulates HilD at protein level The discrepancy between hilD::lac fusions and hilD mRNA level in L-arabinose- dependent regulation made us consider the following possibility: the hilD::lac fusions used above generated hilD null mutations, while the northern blot was carried out in HilD+ background. It has been described that hilD is under the control of an autogenous transcriptional activation (Ellermeier et al., 2005). If L-arabinose were somehow impairing the ability of HilD to activate gene transcription, the reduction in hilD mRNA level observed in the presence of L-arabinose might indeed have a transcriptional origin, due to lack of HilD autoactivation. If such were the case, hilD transcription would be L-arabinose-dependent in a strain carrying a functional HilD protein. We constructed a hilD::lac transcriptional fusion right after hilD stop codon (hilD::lac930). That fusion leaves an intact hilD coding sequence, thus allowing autoactivation of hilD transcription. As shown in Figure 8A, β-galactosidase activity of the hilD::lac930 fusion is reduced almost 5 folds in the presence of 0.2 % L-arabinose. To determine if such regulation was in fact due to HilD autoregulation, we expressed hilD from a heterologous promoter. We inserted PLtetO promoter upstream hilD, deleting its native promoter. We designed the construction to keep the same transcription start site than the native promoter. β-galactosidase activity of hilD::lac930 fusion transcribed from PLtetO is similar in LB and LB plus 0.2 % L-arabinose (Figure 8A). Furthermore, the level of hilD mRNA is not reduced in the presence of L-arabinose when hilD is transcribed from PLteto (Figure 8B). Taken together, those results suggest that the primary target of L- arabinose is HilD protein. We hypothesized that L-arabinose might interfere with HilD function, what could indirectly affects transcription of hilD itself and the rest of SPI-1 genes. To further analyze that possibility, we followed two different strategies: (i) we examined the regulation by L-arabinose of hilA::lac and invF::lac fusions when hilD was expressed from PLtetO. Expression of both fusions is reduced around four folds in the presence of 0.2% L-arabinose (Figure 8C). Note that the regulation is smaller than when hilD is expressed from its native promoter, suggesting that autoactivation of hilD may serve to amplify regulatory inputs; (ii) we analyzed regulation by L-arabinose of rtsA gene. It is encoded outside SPI-1 but is directly activated by HilD. That provides the possibility to study rtsA regulation in the absence of SPI-1, expressing hilD ectopically. For that purpose, we cloned hilD in a plasmid called pXG-10, under the control of PLtetO promoter. We examined regulation by L-arabinose of an rtsA::lac 134 fusion in three different backgrounds: araA with pXG-10 empty vector, araA ∆SPI-1 with pXG-10 empty vector, and araA ∆SPI-1 with pXG10-hilD vector. As shown in Figure 8D, rtsA::lac is regulated by L-arabinose is a strain with SPI-1. Deletion of SPI- 1 decreases β-galactosidase activity, and suppresses regulation by L-arabinose. However, ectopic expression of hilD from a plasmid restores rtsA::lac expression and regulation by L-arabinose in the absence of SPI-1. That supports that L-arabinose affects HilD protein, and confirms that HilD alone is sufficient for transmission of L- arabinose-dependent regulation. Translocation of SPI-1 effectors is reduced in the presence of L-arabinose To determine if SPI-1 repression by L-arabinose was relevant for Salmonella interactions with host cells, we analyzed translocation of SPI-1 effectors into eukaryotic HeLa cells (Figure 9). For that purpose, we used a fusion of the SPI-1 effector encoding gene sipA with the cya gene of Bordetella pertusis. Bordetella adenylate cyclase requires Calmodulin in order to synthesize cAMP. Thus, cAMP will be produced only if the SipA-Cya fusion protein is translocated into eukaryotic cells. The sipA::cya fusion was introduced in isogenic araA, araA araE, and araA prgH backgrounds, and the strains were grown in LB and LB plus 0.2% L-arabinose before mixing with HeLa cells. Translocation was estimated by measuring the amount of cAMP. Translocation is strongly reduced in the presence of L-arabinose in araA background (Figure 9). However, in araA araE background, addition of L-arabinose does not reduce translocation, reflecting that L-arabinose needs to be inside the cell to repress SPI-1 expression (Figure 9). araA prgH background was used as negative control, since prgH encodes an essential component of SPI-1 type-three secretion system. Those results confirm that SPI-1 repression caused by L-arabinose impairs Salmonella interaction with host cells. 135 Strain designation Genotype or description Reference or source 14028 Wild type ATCC SV5999 araA This study SV5297 (invF’-lacZ+)(Hyb) This study SV6000 araA (invF’-lacZ+)(Hyb) This study SV6205 araA (hilA’-lacZ+)(Hyb) This study SV6206 araA (sipB’-lacZ+)(Hyb) This study SV6207 araA (prgH’-lacZ+)(Hyb) This study SV6209 araA invF::3xFLAG This study SV6208 araA hilA::3xFLAG This study SV6210 araA sipB::3xFLAG This study SV6211 araA prgH::3xFLAG This study SV5284 (hilA’-lacZ+)(Hyb) This study SV6218 xylA (hilA’-lacZ+)(Hyb) This study SV6213 araA araE (hilA’-lacZ+)(Hyb) This study SV6212 araBAD araC (hilA’-lacZ+)(Hyb) This study SV6807 araBAD araE araC (hilA’-lacZ+)(Hyb) This study SV6244 araA PLtetO-araE (hilA’-lacZ+)(Hyb) This study SV6245 araBAD araC PLtetO-araE (hilA’-lacZ+)(Hyb) This study SV6219 araA (hilC’-lacZ+)(Hyb) This study SV6222 araA hilD (hilC’-lacZ+)(Hyb) This study SV6220 araA (rtsA’-lacZ+)(Hyb) This study SV6223 araA hilD (rtsA’-lacZ+)(Hyb) This study SV6221 araA (invH’-lacZ+)(Hyb) This study SV6224 araA hilD (invH’-lacZ+)(Hyb) This study SV6417 araA (hilD-lacZ1) This study SV6419 araA (hilD’-lacZ+477)(Hyb) This study SV6421 araA (hilD-lacZ930) This study SV5808 araA PLtetO-hilD (hilD-lacZ930) This study SV5809 araA PLtetO-hilD This study SV6657 araA PLtetO-hilD (hilA’-lacZ+)(Hyb) This study SV6658 araA PLtetO-hilD (invF’-lacZ+)(Hyb) This study SV6659 araA (rtsA’-lacZ+)(Hyb)/pXG10 This study SV6660 araA spi-1 (rtsA’-lacZ+)(Hyb)/pXG10 This study SV6661 araA spi-1 (rtsA’-lacZ+)(Hyb)/pXG10-hilD This study SV6199 araE This study SV6197 araC This study SV6243 PLtetO-araE This study SV6201 xylA This study SV6423 araA (sipA’-cya+)(Hyb) This study SV6424 araA araE (sipA’-cya+)(Hyb) This study SV6425 araA prgH (sipA’-cya+)(Hyb) This study 142 0 100 200 300 400 invF::lac 0 100 200 300 400 ∆araA invF::lac pH aftergrowth 7,56 4,97 7,49 7,59 β-galactosidaseactivity β-galactosidaseactivity Figure 1. pH beforegrowth 7,04 7,08 7,04 7,08 LB LB + L-araLB LB + L-ara 143 0 20 40 60 80 100 120 hilA::lac invF::lac sipB::lac prgH::lac HilA - InvF - SipB - PrgH - Relative β-galactosidase activity Fig 2. A B HilA - 3xFLAG InvF - 3xFLAG SipB - 3xFLAG PrgH - 3xFLAG Tagged protein Lopding control (GroEL) LB LB+L-ara LB LB+L-ara LB LB+L-ara LB LB+L-ara Quantification 100 100 100 1005.1 19.3 12.0 8.0 144 β-galactosidase activity hilA::lac C 0 20 0.01 0.02 0.05 0.1 0.2 L - arabinose 0 0.01 0.02 0.05 0.1 0.2 Tagged protein Lopding control (GroEL) Quantification 100 46.2 35.7 15.0 10.7 6.9 HilA-3xFLAG D L-arabinose concentration (%) 0 0.01 0.02 0.05 0.1 0.2 L - arabinose concentration (%) 0 145 100 200 300 400 500 100 200 300 400 500 100 200 300 400 500 galactosidase activity galactosidase activity galactosidase activity L-arabinose D-xylose D-arabinose ∆araA hilA::lac ∆xylA hilA::lac hilA::lac A B C Fig. 3 0 100 0 100 0 100 βgalactosidase βgalactosidase βgalactosidase 0 0.2 1 0 0.2 1 0 0.2 1 Pentose concentration (%) 146 0 100 200 300 400 500 ∆araA hilA::lac β-galactosidase activity Fig 4. A B ∆ araA hilA :: lac LB LB + 0.2 % L-ara LB + 1 % L-ara LB + 2 % L-ara LB + 3 % L-ara Control araE araC araE araC 0 100 200 300 400 500 β-galactosidase activity B ∆ araA hilA :: lac LB LB + 0.2 % L-ara 147 Fig 5. A 0 100 200 300 400 500 β-galactosidase activity ∆araA hilA::lac 0 100 200 300 400 500 β-galactosidase activity ∆araA PLtetO-araE hilA::lac LB LB + 0.005 % L-ara LB + 0.01 % L-ara LB + 0.02 % L-ara LB + 0.05 % L-ara LB + 0.1 % L-ara LB + 0.2 % L-ara B 148 40 60 80 100 120 Relative β-galactosidase activity Fig. 6 0 20 hilC::lac rtsA::lac invH::lac Relative HilD + -+ + - - 149 0 50 100 150 200 250 0 20 40 60 80 hilD::lac1 hilD::lac477 hilD mRNA LB LB + L-ara β-galactosidase activity β-galactosidase activity Fig. 7 A B 150 LB LB + L-ara LB LB + L-ara 0 100 200 300 400 500 600 0 100 200 300 400 500 600 0 100 200 300 400 500 600 β-galactosidase activity β-galactosidase activity hilA::lac invF::lac 0 5000 10000 15000 20000 β-galactosidase activity β-galactosidase activity Control PLtetO-hilD Control PLtetO-hilD Control PLtetO-hilD Control PLtetO-hilD Control/ pXG10 ∆SPI-1/ pXG10 ∆SPI-1/ pXG10-hilD A B C D rtsA::lac Fig. 8 151