Expression of the Fis protein is sustained in late-exponential- and stationary-phase cultures of Salmonella enterica serovar Typhimurium grown in the absence of aeration


*E-mail; Tel. (+353) 1896 2013; Fax (+353) 1679 9294.


The classic expression pattern of the Fis global regulatory protein during batch culture consists of a high peak in the early logarithmic phase of growth, followed by a sharp decrease through mid-exponential growth phase until Fis is almost undetectable at the end of the exponential phase. We discovered that this pattern is contingent on the growth regime. In Salmonella enterica serovar Typhimurium cultures grown in non-aerated SPI1-inducing conditions, Fis can be detected readily in stationary phase. On the other hand, cultures grown with standard aeration showed the classic Fis expression pattern. Sustained Fis expression in non-aerated cultures was also detected in some Escherichia coli strains, but not in others. This novel pattern of Fis expression was independent of sequence differences in the fis promoter regions of Salmonella and E. coli. Instead, a clear negative correlation between the expression of the Fis protein and of the stress-and-stationary-phase sigma factor RpoS was observed in a variety of strains. An rpoS mutant displayed elevated levels of Fis and had a higher frequency of epithelial cell invasion under these growth conditions. We discuss a model whereby Fis and RpoS levels vary in response to environmental signals allowing the expression and repression of SPI1 invasion genes.


The factor for inversion stimulation (Fis) was named for its role in facilitating the bacterial and bacteriophage DNA inversion-based genetic switches that are catalysed by site-specific recombinases of the serine-invertase protein family (Kahmann et al., 1985; Johnson et al., 1986). Fis is also an important regulator of transcription that can act positively or negatively, depending on the location of its binding site with respect to the target promoter (Xu and Johnson, 1995; Schneider et al., 2003; Opel et al., 2004). Its own gene, fis, is among those that it represses (Pratt et al., 1997), while it acts to upregulate many of the genes that encode components of the translation machinery of the cell (Bosch et al., 1990; Nilsson et al., 1992; Ross et al., 1990; Ball et al., 1992; Ninnemann et al., 1992; Walker et al., 1999; Schneider et al., 2003; Paul et al., 2004). In many cases, the Fis protein cooperates with DNA supercoiling to modulate promoter function (McLeod et al., 2002; Rochman et al., 2002; 2004; Travers and Muskhelishvili, 2005; Ó Cróinín et al., 2006). It has a complex relationship with DNA supercoiling due to its ability to influence transcription of the genes coding for DNA gyrase (Schneider et al., 1999; Keane and Dorman, 2003) and DNA topoisomerase I (Weinstein-Fischer et al., 2000; Weinstein-Fischer and Altuvia, 2007), the enzymes that respectively create and relax negative supercoiling in DNA (Wang, 1996; Dorman, 2006). The fis promoter is itself sensitive to DNA supercoiling (Schneider et al., 2000). Fis can act to preserve microdomains of supercoiling due to its preference for binding to DNA of intermediate superhelical density, and this allows it to sustain transcription at certain promoters that respond to changes in local DNA supercoiling levels (Rochman et al., 2002; 2004). The positive effect of Fis on promoter function is not restricted to its influence on DNA topology; it can also act as a conventional transcription factor by recruiting RNA polymerase and assisting it in the formation of open transcription complexes (Bokal et al., 1997; McLeod et al., 2002). Fis has also been shown to redistribute DNA twist by suppression of single-stranded bubbles in DNA at sites prone to supercoiling-induced duplex destabilization. Here the displaced twist can act at promoters to drive the isomerization of closed to open transcription complexes (Opel et al., 2004).

The Fis protein is an 11.2 kDa DNA binding and bending protein that forms a homodimer. The classic description of Fis expression involves a burst of expression at the onset of exponential growth following inoculation of fresh growth medium (Osuna et al., 1995). The level of Fis protein then declines thereafter until it is almost undetectable by the end of the exponential phase of growth. This expression pattern reflects the activity of the fis promoter which is repressed by the Fis protein and requires negative DNA supercoiling for optimal activity (Schneider et al., 2000). This latter feature requires the presence of a G+C-rich discriminator sequence between the Pribnow box and the transcription start site. These DNA sequences are found in stringently regulated promoters and are difficult to melt due to the high level of hydrogen bonding between the bases (Travers and Muskhelishvili, 2005).

The DNA sequences that are bound by Fis show weak homology to a degenerate consensus (Hengen et al., 1997). Data from DNA microarray and from ChIP-on-chip experiments indicate that Fis can influence large numbers of promoters in Salmonella enterica and Escherichia coli (Kelly et al., 2004; Grainger et al., 2006). Many of the genes that are regulated positively by Fis in Gram-negative bacterial pathogens contribute to their virulence, and Fis has been identified as a virulence gene regulator in enteropathogenic E. coli, enteroaggregative E. coli, S. Typhimurium and Shigella flexneri (Falconi et al., 2001; Goldberg et al., 2001; Sheikh et al., 2001; Wilson et al., 2001; Schechter et al., 2003; Yoon et al., 2003; Kelly et al., 2004; Prosseda et al., 2004). In the case of S. Typhimurium, Fis cooperates with DNA supercoiling to modulate virulence gene transcription in bacteria living in macrophage (Ó Cróinín et al., 2006). The involvement of Fis in the control of virulence genes points to an interesting paradox. Many of them are expressed best in bacteria that are growing slowly, a situation where the classic view of Fis expression dictates that this protein should be almost undetectable. In this study, we re-examined the pattern of expression of the Fis protein in the context of those growth conditions that have been described as optimal for the expression of the S. Typhimurium virulence genes in the SPI1 pathogenicity island. These involve overnight growth of the culture without aeration in the complex growth medium Luria–Bertani (LB). SPI1 genes are subject to multiple regulatory influences and represent an interesting case in the evolution of bacterial gene regulatory circuits. This is because the SPI1 island is composed of A+T-rich DNA that distinguishes it from the DNA of the ancestral chromosome. It is thought that such islands were acquired from an unidentified source by lateral genetic transfer, and that their arrival conferred upon Salmonella the ability to invade host cells (Groisman and Ochman, 1997; Hacker and Kaper, 1999; Ochman et al., 2000). The nucleoid-associated protein H-NS plays an important role in repressing transcription of genes in SPI1, and other horizontally acquired blocks of DNA, through its preference for binding to regions of A+T-rich DNA displaying planar curvature (Rimsky et al., 2001; Lucchini et al., 2006; Navarre et al., 2006; Oshima et al., 2006; Bouffartigues et al., 2007; Dorman, 2007; Doyle et al., 2007). It has been speculated that this negative influence may assist in the integration of the genes into the new bacterial cell following acquisition by minimizing their impact on competitive fitness through inappropriate expression of their gene products. This attractive hypothesis creates the need for antagonistic influences to overcome H-NS-mediated repression so that the new genes can be expressed to the benefit of the bacterium (Dorman, 2007). Several regulatory proteins encoded by the ancestral chromosome influence SPI1 gene expression positively (Lee et al., 2000; Feng et al., 2003; 2004; Kato and Groisman, 2004; Bijlsma and Groisman, 2005; Linehan et al., 2005; Navarre et al., 2005; Shin and Groisman, 2005; Mangan et al., 2006), as do transcription factors encoded by regulatory genes located within the island (Rhen and Dorman, 2005). Fis is one of the chromosomally encoded proteins that upregulates SPI1 gene transcription (Wilson et al., 2001; Schechter et al., 2003; Kelly et al., 2004). Its role here is consistent with its known ability to oppose the repressive activity of the H-NS protein. Here we report that, contrary to predictions based on the classic pattern of Fis protein expression, Fis is expressed at elevated levels in S. Typhimurium bacteria grown to stationary phase under non-aerated, SPI1-inducing conditions. In addition, we implicate the alternate stress response sigma factor RpoS as playing a role in repressing Fis expression, and show that induction of Fis in the absence of aeration is facilitated by a reduction of RpoS levels under these conditions. Lastly, we demonstrate a correlation between the balance of expression of these two global regulators and the ability of Salmonella to invade epithelial cells.


Invasion of epithelial cells and Fis expression during growth with or without aeration

The genes in the SPI1 pathogenicity island of S. Typhimurium are known to be Fis-dependent for expression (Wilson et al., 2001; Schechter et al., 2003; Kelly et al., 2004), and previous data indicated that these genes are induced when bacteria are grown to late exponential growth phase without aeration (Lee and Falkow, 1990; Song et al., 2004). Furthermore, SPI1 gene expression continues to depend on Fis even in late exponential phase with aeration (Kelly et al., 2004). This represents a point in the growth curve where Fis is poorly detected in E. coli (Ball et al., 1992; Appleman et al., 1998; Keane and Dorman, 2003), although approximately 10 000 Fis dimers were found to be present in an LT2 culture grown under aerated conditions to late logarithmic stage (Osuna et al., 1995). We hypothesized that the induction of SPI1 genes under non-aerated conditions could be due to the presence of the Fis protein. Wild-type and fis knockout mutant strains of S. Typhimurium SL1344 and E. coli CSH50 were grown for 1 h or overnight with or without aeration and tested for the presence of Fis protein by Western immunoblotting (Fig. 1A). The Fis protein was readily detectable in S. Typhimurium SL1344 cultures at 1 h whether these had been grown with or without aeration. The protein was not detectable in cultures grown overnight with aeration, in agreement with previous findings. In contrast, Fis protein was readily detected in stationary-phase cultures that had grown under non-aerated conditions (Fig. 1A, lane 5). Densitometric analysis showed that Fis was present here at ∼50% of the level seen in either the aerated or the non-aerated exponential-phase cultures (Fig. 1A, compare lane 5 with lanes 3 and 7). This presence of Fis in cultures grown without aeration was shown to correlate with an increase in invasion of epithelial cells, confirming that these were conditions under which SPI1 genes were induced (Fig. 1B). No such non-aerated-growth-dependent increase in invasion was observed in the fis knockout mutant.

Figure 1.

The effect of aeration on the expression of Fis in S. Typhimurium SL1344 and E. coli CSH50 (A). Wild-type (WT) and fis mutant strains were grown under aerating conditions to stationary phase (lanes 1 and 2) and exponential phase (lanes 3 and 4), and compared with bacteria grown without aeration at stationary phase (lanes 5 and 6) and exponential phase (lanes 7 and 8). Stationary-phase samples were taken after 24 h, whereas exponential-phase samples were taken after 1 h. The position of the Fis protein band is indicated. A cross-reacting protein band detected by the anti-Fis antibody in S. Typhimurium is indicated by the arrow labelled ‘X’. The identity of the Fis protein band was confirmed (B) by comparing fis mutant and WT samples grown to exponential phase (lanes 1 and 2) with a well containing purified Fis protein (lane 3). Fis expression in stationary-phase cultures was shown to correlate with a high level of epithelial cell invasion by S. Typhimurium (C). Filled bars represent invasion by bacteria grown without aeration, and open bars are data for bacteria grown with aeration. Results are presented for WT SL1344 and for its fis knockout mutant derivative. Error bars represent standard deviations.

The pattern of Fis expression in E. coli strain CSH50 was similar to that in S. Typhimurium SL1344 for the aerated cultures (Fig. 1A). However, CSH50 cultures did not express detectable levels of Fis protein when grown to stationary phase without aeration, a result that was in marked contrast to that obtained with SL1344. The results obtained with E. coli strain CSH50 were consistent with the classic growth phase-dependent pattern of Fis protein expression. In contrast, the data obtained with the S. Typhimurium SL1344 culture suggested that this classic expression pattern was contingent on aerated growth conditions, at least in Salmonella.

Transcription of the fis gene in strains SL1344 and CSH50

Fis protein expression is controlled at the level of fis transcription (Osuna et al., 1995; Walker et al., 1999) and by a post-transcriptional mechanism (Owens et al., 2004). To allow a comparison of fis promoter activity and Fis protein expression under the growth conditions used in this study, the promoters of the fis genes from E. coli strain CSH50 and S. Typhimurium strain SL1344 were coupled individually to a gfp reporter gene. This allowed fis–gfp transcription to be monitored as a function of growth rate and aeration by determining the levels of Gfp fluorescence based on flow cytometry. It also allowed each promoter to be studied in cells of the other bacterial species.

The SL1344 fis promoter in the SL1344 background displayed the expected classic pattern of fis transcription under aerated growth conditions, with a burst of expression in the first few hours followed by a swift reduction in transcription. However, without aeration a different pattern was observed (Fig. 2A). After the initial burst of expression, transcription levels dropped but then began to rise again, giving about a two-fold higher level of fluorescence at 24 h than that observed under aerated conditions (Fig. 2A). This difference in transcription levels at 24 h between aerated and non-aerated cultures mirrored that observed for Fis protein levels in the Western blots. The CSH50 fis promoter in the CSH50 background displayed only the classic pattern of fis gene expression with and without aeration, with an initial burst in early exponential growth phase that declined thereafter (Fig. 2B). These results agreed with the data from the Fis Western blots and suggested that E. coli strain CSH50 was capable of supporting only the classic pattern of Fis expression with or without aeration.

Figure 2.

Transcription of a fis–gfp fusion as a function of growth in S. Typhimurium strain SL1344 and E. coli strain CSH50. Transcription from the SL1344 fis promoter (A and C) and the CSH50 fis promoter (B and D) was compared in SL1344 (A and D) and in CSH50 (B and C). Transcription of an rpsM–gfp fusion was also measured in S. Typhimurium strain SL1344 (E). Gfp levels were measured in cultures grown with (◆) or without (▪) aeration, and the insets show the growth curves of each strain growing with and without aeration with the OD600 (y-axis) plotted against time (x-axis). All cultures were inoculated 1/100 from an overnight culture grown under aerating conditions to ensure a starting culture which was negative for fis transcription. Both fis promoters displayed an initial burst of fis transcription in each strain background. However, elevated expression of fis in stationary phase in the absence of aeration was observed only in the S. Typhimurium background (A and D). Each datum point is the average of three independent measurements, and the error bars represent standard deviations.

It was necessary to investigate whether this difference in regulation of transcription of fis during growth without aeration was due to sequence differences in the promoter regions of the fis genes from E. coli and Salmonella. The fis gene is transcribed from a single promoter, and the sequences of the fis promoters from E. coli and S. Typhimurium are identical over a region extending from the translation start codon to a position lying 14 base pairs upstream of the −35 box (Mallik et al., 2004). Upstream from this point the sequences show an increasing level of divergence. We cloned the promoter sequences on DNA fragments from both SL1344 and CSH50, which extended from position −310 to +287 with respect to the transcription start site (+1) that had been defined previously (Mallik et al., 2004). The activity of the Salmonella fis promoter was tested in E. coli strain CSH50, and the reciprocal experiment was carried out in which the activity of the E. coli fis promoter was monitored in S. Typhimurium strain SL1344.

The Salmonella fis promoter in E. coli revealed no significant difference in its transcription profile whether grown in the presence or absence of aeration at 24 h (Fig. 2C), despite having shown induction of fis transcription when grown without aeration in stationary phase when in its native SL1344 background (Fig. 2A). However, when the activity of the E. coli fis promoter was measured in Salmonella, it displayed a pattern that was reminiscent of that seen with the Salmonella fis promoter in the SL1344 background: fis–gfp transcription was sustained in the non-aerated culture at the onset of stationary phase (Fig. 2D). These results suggest that the difference in expression of the Fis protein with and without aeration depended on the type of bacterial cell and was independent of any features associated with the fis gene promoters of E. coli CSH50 or S. Typhimurium SL1344. To confirm that these results were specific for the fis promoter and not a more general observation of transcription in the Salmonella background, the transcription of the rpsM gene was monitored in the SL1344 background and was found to be unaffected by the degree of aeration in stationary phase (Fig. 2E). Given that different expression patterns for fis had been detected in S. Typhimurium and E. coli, it was important to determine whether these were bacterial species-specific or merely strain-specific differences.

Transcription of fis in other bacterial strains

The transcription profile of the Salmonella fis promoter was investigated in S. Typhimurium strain LT2. Here the non-aerated phenotype was evident, with an approximately three-fold difference between aerated and non-aerated cultures at 24 h (Fig. 3A) and at an even higher level than that observed in strain SL1344 (Fig. 2A). When the transcription of fis was measured in the E. coli strain MC4100, an approximately threefold difference was also observed between bacteria grown with and without aeration after 9 h (Fig. 3B), although transcription levels were significantly less than those observed in strain LT2. These results demonstrated that this pattern of fis expression without aeration was strain-specific and not species-specific. What strain differences might explain these effects?

Figure 3.

Transcription of a fis–gfp fusion in the absence of aeration in other S. Typhimurium and E. coli strains and its correlation with RpoS sigma factor expression. Transcription from the SL1344 fis promoter in S. Typhimurium strain LT2 (A) and the SL1344 fis promoter in E. coli strain MC4100 (B) was measured as a function of growth. Gfp levels were measured from cultures grown with (◆) or without (▪) aeration. All cultures were inoculated 1/100 from an overnight culture grown under aerating conditions to ensure a starting culture which was negative for fis transcription. Both strains displayed elevated expression of fis–gfp in the absence of aeration compared with bacteria grown with aeration. Each datum point is the average of three independent measurements, and the error bars represent standard deviations. Western blotting was used to test the following strains for RpoS protein expression after 24 h growth under low aeration (C): SL1344 (lane 1), SL1344 fis (lane 2), LT2 (lane 3), CSH50 (lane 4), CSH50 fis (lane 5) and MC4100 (lane 6).

RpoS is the stress and stationary-phase sigma subunit of RNA polymerase (Hengge-Aronis, 2002), and S. Typhimurium strain LT2 is known to express less RpoS than other Salmonella strains due to a mutation that alters codon usage and results in less efficient translation of the rpoS message (Lee et al., 1995; Swords et al., 1997; Wilmes-Riesenberg et al., 1997). As fis transcription was higher without aeration in LT2 than SL1344, we wished to investigate the possibility of a correlation between RpoS expression and fis transcription in the absence of aeration in both the Salmonella and E. coli strains. Western blotting of bacteria grown to stationary phase without aeration revealed an anti-correlation between RpoS levels, and transcription of fis. E. coli strain CSH50 had the highest levels of RpoS protein under these conditions (Fig. 3C) and showed no expression of Fis at the onset of stationary phase (Fig. 1). S. Typhimurium strain SL1344 and E. coli strain MC4100 showed comparable levels of RpoS expression, which were significantly less than those observed in CSH50 (Fig. 3C), and both strains showed comparable levels of fis transcription without aeration (compare Figs. 2A and 3B). Furthermore, S. Typhimurium strain LT2 showed only trace amounts of RpoS protein expression (Fig. 3C) and gave the highest level of fis transcription under these conditions (Fig. 3A). These results revealed a negative correlation between expression of RpoS and Fis under non-aerated growth conditions in stationary-phase cultures.

RpoS and Fis expression during growth with or without aeration

Western blotting was used to test the effect of the aeration regime on the expression of RpoS protein in S. Typhimurium strain SL1344 and E. coli strain CSH50 (Fig. 4A). Again an anti-correlation was observed between the expression of RpoS and Fis. In strain SL1344, RpoS levels were ∼5-fold higher with aeration than without (Fig. 4A, lanes 1 and 2). This was the opposite of the Fis expression pattern in SL1344 with and without aeration (Fig. 1A). In contrast, strain CSH50, which produced no detectable Fis protein in late exponential growth phase regardless of aeration (Fig. 1A), showed only a ∼2-fold decrease in RpoS levels in response to the presence or absence of aeration (Fig. 4A, lanes 3 and 4).

Figure 4.

The interplay between aeration and RpoS and Fis expression.
A. Comparison of RpoS expression with and without aeration in stationary phase (24 h) post inoculation (A): SL1344 RpoS expression with (1st lane) and without (2nd lane) aeration; CSH50 RpoS expression with (3rd lane) and without (4th lane) aeration.
B and C. Expression patterns of the RpoS and Fis proteins, respectively, in SL1344 and in its fis and rpoS knockout mutant derivates grown with or without aeration for 24 h: SL1344 with aeration (1st lanes); SL1344 without aeration (2nd lanes), SL1344 fis with aeration (3rd lanes) SL1344 fis without aeration (4th lanes), SL1344 rpoS with aeration (5th lanes), and SL1344 rpoS without aeration (6th lanes).

Given that a previous study (Hirsch and Elliott, 2005) had reported that the Fis protein can act as a repressor of RpoS transcription during exponential growth, we decided to investigate whether RpoS in turn represses Fis expression under stationary-phase conditions and whether this contributes to the observed anti-correlation between Fis and RpoS expression. To investigate this, we examined the levels of RpoS and Fis with and without aeration in wild-type SL1344 and in its fis and rpoS knockout mutant derivatives (Fig. 4B). As expected, wild-type SL1344 displayed stationary-phase expression of Fis without aeration but not with aeration. In contrast, RpoS levels were ∼5-fold higher with aeration than without (Fig. 4B, lanes 1 and 2). The fis mutant expressed RpoS to a ∼10-fold higher level with aeration than without (Fig. 4B, lanes 3 and 4). The rpoS mutant revealed a higher level of Fis expression without aeration than that observed in the wild-type strain (Fig. 4C, lanes 2 and 6). Furthermore, Fis could clearly be detected when the rpoS mutant was grown under aerated conditions, whereas the protein was absent in wild-type cells grown under the same conditions (Fig. 4C, lanes 1 and 5). Although these levels were lower than those observed in the rpoS mutant under low aeration, they were comparable to those observed in wild-type cells grown under low aeration. These results showed that the absence of RpoS in the absence or presence of aeration correlated with greater levels of Fis protein. This was consistent with a role for RpoS in repressing Fis expression. However, other factors are clearly involved because the absence of RpoS in the aerated culture did not lead to the same levels of Fis expression as that observed without aeration: the non-aerated Sl1344 rpoS culture expressed at least twice as much Fis protein as its aerated counterpart (Fig. 4C, lanes 5 and 6).

RpoS and repression of fis transcription

To investigate whether RpoS affected the level of transcription of fis, the activity of the fis promoter in the SL1344 rpoS mutant was monitored during the growth of the culture grown with and without aeration (Fig. 5A and B). When compared with the SL1344, wild-type strain higher levels of fis transcription were clearly observed in the rpoS mutant throughout the growth curve. However, a clear increase in fis transcription levels under low-aeration conditions in stationary phase was observed in both the wild-type and the rpoS mutant strains but absent in both strains when grown under aerating conditions. This result suggested that although the absence of RpoS leads to higher levels of fis transcription than those observed in the wild type, other factors play a role in the specific increase in transcription of fis that is seen in the absence of aeration. To ensure that the fis–gfp recombinant plasmid reproduced faithfully the response of the fis promoter in the rpoS knockout strain, the experiment was repeated without the reporter fusion, monitoring fis mRNA expression directly from the bacterial chromosome by reverse transcription polymerase chain reaction (RT-PCR). The level of fis mRNA was ∼2-fold higher in the rpoS mutant than in the wild-type strain grown under non-aerated conditions, whereas no such difference was observed in the case of the control transcript rpsM (Fig. 5C).

Figure 5.

The effect of RpoS on the transcription of a fis–gfp fusion in bacteria grown with or without aeration. The activity of the SL1344 fis promoter in S. Typhimurium SL1344 (◆) and SL1344 rpoS (▪) was measured as a function of growth using a fis–gfp transcriptional fusion. All cultures were inoculated 1/100 from an overnight starter culture grown with aeration to ensure a starting culture which was negative for fis transcription. Gfp levels were measured in cultures grown with (A) or without (B) aeration. Each datum point is the average of three independent measurements, and the error bars represent standard deviations. A similar increase in transcription was seen when fis expression from the chromsome was monitored by RT-PCR from RNA isolated at 9 h growth (C); no difference was detected in the transcription of the control gene rpsM (which encodes a ribosomal protein).
D. Transcription of the rpoS gene from the pBAD promoter in plasmid pBADrpoS was induced using increasing concentrations of arabinose or repressed using glucose (0.2% w/v). RpoS protein expression was monitored by Western blotting in cultures grown with or without aeration containing no arabinose (1st lanes), 0.2% glucose (2nd lanes), 0.02% arabinose (3rd lanes), 0.2% arabinose (4th lanes) or 0.5% arabinose (5th lanes) for 24 h. RpoS expression was only evident in the culture grown with aeration in the presence of 0.5% arabinose. In contrast, Fis protein expression from the pBAD promoter on plasmid pBADfis could be induced in cultures grown with or without aeration.

In an attempt to examine the link between RpoS and fis gene expression in more detail, the open reading frame of the rpoS gene was cloned into the plasmid pBAD24 and transformed into the SL1344 rpoS mutant. This placed the expression of rpoS under the control of an arabinose-inducible promoter to allow complementation of the rpoS mutation. However, although RpoS protein could clearly be expressed when the bacteria were grown with aeration, no RpoS protein could be detected when they were grown in its absence (Fig. 5D). It was important to rule out the possibility that the lack of RpoS did not reflect an inability to induce rpoS transcription with arabinose in pBADrpoS under non-aerated growth conditions. With this in mind, the rpoS open reading frame was replaced with that from the fis gene, creating pBADfis, which was transformed into strain SL1344 fis. Fis protein could be detected from pBAD24fis following arabinose induction in bacteria grown in either the presence or absence of aeration, showing that the arabinose-dependent regulatory circuit operated under both sets of growth conditions (Fig. 5D). Expression of the fis gene was more easily induced under low-aeration conditions, confirming that other factors play a role in this low-aeration induction of the Fis protein. These results suggested that RpoS was post-transcriptionally repressed under non-aerated growth conditions.

RpoS, Fis and invasion of epithelial cells

Both the growth of Salmonella under non-aerated conditions and Fis protein expression are associated with invasion of epithelial cells (Lee and Falkow, 1990; Wilson et al., 2001). Therefore, we postulated that an rpoS mutant displaying elevated levels of Fis expression under these conditions should be more adept at invasion of epithelial cells. Invasion assays in both CACO2 and CHO cells confirmed that a fis mutant showed consistently less invasiveness when grown under these conditions than the wild-type (Fig. 6). This result correlates well with previously published data (Wilson et al., 2001). However, an rpoS mutant was shown to be more invasive than the wild-type strain (Fig. 6). In addition, a fis rpoS double mutant had a similar invasive phenotype to that of the fis mutant. These results suggest that the increased levels of Fis protein expression observed in an rpoS mutant grown without aeration is associated with an increased ability to invade epithelial cells.

Figure 6.

The effect of RpoS and Fis on SPI1 invasion of epithelial cells. SL1344 and fis, rpoS and fis rpoS mutant derivates were grown under non-aerated (SPI1 inducing) conditions and used to invade CACO2 (A) or CHO (B) epithelial cells. Error bars represent standard deviation.


The proteins encoded by the SPI1 pathogenicity island play a critical role in expression of the invasive phenotype of S. Typhimurium (Mills et al., 1995; Collazo and Galán, 1997; Hardt et al., 1998; Hansen-Wester and Hensel, 2001). Several independent investigations have established the importance of non-aeration as a prerequisite for the induction of the transcription of the virulence genes in SPI1, leading to speculation that the microaerophilic conditions of the mammalian gut may represent an important environmental signal for their upregulation (Galán and Curtiss, 1990; Lee and Falkow, 1990; Lostroh and Lee, 2001). These ‘SPI1-inducing conditions’ also include a role for growth phase: bacteria approaching the late exponential phase of growth in non-aerated culture conditions show optimal invasiveness of epithelial cells growing in vitro (Lee and Falkow, 1990). At first glance, this observation is difficult to reconcile with the finding that optimal expression of the genes in the SPI1 island requires the Fis DNA binding protein, because this is expressed best in the early stages of logarithmic growth and becomes almost undetectable in cultures approaching the end of exponential growth (Ball et al., 1992; Nilsson et al., 1992; Ninnemann et al., 1992; Keane and Dorman, 2003; Mallik et al., 2004). In this study, we demonstrate that there is no conflict between the expression pattern of Fis and the dependence of the SPI1 genes on this protein, because we have discovered that Fis expression is sustained into stationary phase in non-aerated cultures (Fig. 1).

The influence of low oxygen tension on expression of the Fis protein was shown to be a manifestation of its impact on the activity of the fis gene promoter. This promoter is subject to complex regulation, responding to the stringent response, growth phase and DNA supercoiling (Ball et al., 1992; Nilsson et al., 1992; Ninnemann et al., 1992; Mallik et al., 2004). These features are common to the fis promoters of both E. coli and S. Typhimurium (Keane and Dorman, 2003; Mallik et al., 2004). Here we show that the S. Typhimurium and the E. coli fis promoters exhibit similar expression profiles when driving transcription of a green fluorescent protein (GFP) reporter gene in S. Typhimurium or E. coli strains grown with aeration (Fig. 2). However, the fis promoters from both bacteria show a distinctly different expression profile in Salmonella to that observed in E. coli when the bacteria are grown without aeration (Fig. 2). This suggests that this difference in fis regulation observed without aeration is independent of the sequence of the fis promoter but is due instead to other regulatory factors working on the promoter. Initially this raised the tantalizing possibility that these different species have evolved distinct Fis expression profiles in order to adapt to their different ecological niches. However, in the course of these experiments we discovered that the strains used had an important influence on the pattern of fis gene expression. This suggested, at least in the case of E. coli, that this phenotype was strain- rather than species-specific.

A negative correlation between fis expression and the status of the rpoS gene was apparent in each strain we examined (Figs 3 and 4). The S. Typhimurium strain LT2 is known to be defective in rpoS gene expression (Swords et al., 1997; Wilmes-Riesenberg et al., 1997), while E. coli strain MC4100 expresses a lower level of this sigma factor than other E. coli strains, such as W3110 (Jishage and Ishihama, 1997; Jishage et al., 1996). In each case, the level of Fis protein expression seen was higher than in the S. Typhimurium strain SL1344 and E. coli strain CSH50, both of which express RpoS protein to higher levels than their counterparts. This anti-correlation in the levels of the Fis and RpoS proteins recalled an earlier observation that Fis is a repressor of rpoS gene transcription, at least in exponential-phase cultures (Hirsch and Elliott, 2005). Our data indicate that the presence of the RpoS sigma factor correlates with repression of the fis gene. How is this achieved? As it is unlikely that RpoS represses the fis promoter directly, repression may involve an indirect negative effect in which RpoS upregulates an as-yet unidentified negative regulator of Fis. Attempts at repressing fis promoter function by artificial overexpression of RpoS were unsuccessful (Fig. 5D). We were only able to detect the overexpressed sigma factor at late stages of growth in aerated cultures, conditions when the protein is naturally abundant in any case. Presumably, the post-transcriptional mechanisms responsible for regulation of RpoS protein expression were sufficiently robust to withstand our attempts at overwhelming them through ectopic transcriptional upregulation of the rpoS gene. While we have established a role for RpoS in negatively regulating fis gene expression at some level, the molecular mechanism by which this occurs is presently unknown.

It is possible that RpoS exerts a non-specific effect on fis transcription in the late exponential and the stationary phases of growth by its ability to compete with RpoD for the pool of RNA polymerase core enzyme (Farewell et al., 1998). By reducing the number of RpoD-containing holoenzymes, RpoS may downregulate RpoD-dependent promoters such as that of the fis gene. In rpoS knockout mutants or in strains that fail to achieve maximal expression of the RpoS protein, such promoters may have the potential to work at higher levels. This potential may be abrogated by other factors, such as the relaxation of DNA supercoiling that accompanies the transition into stationary phase (Dorman et al., 1988; Ó Cróinín et al., 2006) and which is known to influence the transcription of the fis gene (Schneider et al., 2000). However, growth in the absence of aeration increases the level of negative supercoiling in the DNA of E. coli and S. Typhimurium (Yamamoto and Droffner, 1985; Dorman et al., 1988; Hsieh et al., 1991a), and this would stimulate the fis promoter if RpoD were free from competition by RpoS. This stimulation would also receive a boost from the osmotic stress that is both a feature of SPI1-inducing conditions and a known enhancer of the negative supercoiling activity of DNA gyrase (Higgins et al., 1988; McClellan et al., 1990; Hsieh et al., 1991b).

It is interesting to consider the reciprocal expression patterns of the Fis and RpoS proteins in the context of the S. Typhimurium-invasive phenotype. Fis is required for optimal expression of SPI1 genes, and the non-aerated growth conditions that facilitate expression of an invasive phenotype also result in sustained Fis expression in late stages of growth. This is a period of growth when RpoS expression would normally be expected to increase, and we have confirmed that the protein is indeed present. However, an rpoS knockout mutation allows a strong increase in Fis protein levels to occur (Fig. 5), as might be expected given the anti-correlation in the expression patterns of Fis and RpoS that we have discovered. Furthermore, the loss of RpoS protein in non-aerated cultures also results in an increase in the invasiveness of S. Typhimurium (Fig. 6).

What is the role of RpoS in invasiveness? This protein is regarded as essential for the virulence of S. Typhimurium, because rpoS mutants are attenuated in virulence and have been described as candidates for live oral vaccines (Wilmes-Riesenberg et al., 1997; Song et al., 2004). Evidence that RpoS is a direct regulator of SPI1 gene transcription is lacking (Wilmes-Riesenberg et al., 1997). However, it does play a role in the upregulation of the spv virulence genes that are located on the S. Typhimurium virulence plasmid. These genes contribute to the establishment of a systemic infection and are required at the stages of infection that follow invasion of the epithelial cells lining the host gut (Rotger and Casadesus, 1999; Matsui et al., 2001). The spv genes are not regulated by Fis (Kelly et al., 2004). This indicates a succession of events in which Fis upregulates the invasion genes of the SPI1 pathogenicity island on the chromosome and then RpoS upregulates the plasmid-linked spv virulence genes to promote the later stages of the infection process. Fis is expressed in the intracellular niche, so why does it not repress transcription of the rpoS gene, preventing RpoS from activating spv transcription? This is likely to be because the repressive activity of the Fis protein at the rpoS promoter is confined to the early exponential stage of growth (Hirsch and Elliott, 2005). The reason for this differential growth phase-dependent repressive activity is unknown, but it is tempting to speculate that some property of the rpoS promoter, perhaps its topology, changes with growth phase, modulating its potential for Fis-mediated repression.

In conclusion, this study identifies a new reciprocal expression profile for the global regulators Fis and RpoS during growth in the absence of aeration. This expression is associated with a downregulation of RpoS and an upregulation of Fis expression under these conditions. Upregulation of Fis correlates with increased invasion of epithelial cells, suggesting an important role in vivo. The reciprocal modulation of RpoS and Fis expression represents an interesting trade-off between reduced expression of genes involved in the stress response in favour of genes involved in the invasion of epithelial cells leading to systemic dispersion.

Experimental procedures

Bacterial strains and growth conditions

All strains used in this study are listed in Table 1. Bacterial cultures were grown in 10 ml LB broth agitated at 200 r.p.m. in a Gallenkamp orbital incubator at 37°C under aerated (250 ml flask) or non-aerated (15 ml tube) growth conditions. Chloramphenicol, kanamycin and carbenicillin were used at final concentrations of 25, 50 and 50 μg ml−1 respectively.

Table 1.  Bacterial strains.
StrainRelevant characteristicsReference/source
E. coli
 XL-1 BluerecA1 endA1 gyrA96thi-1 hsdR17 supE44 relA1 lac[F′proAB lacIqZΔM15Tn10(TcR)]Stratagene
 CSH50FaraΔ(lac pro) strA thiCold Spring Harbor Laboratories
 CSH50fisFaraΔ(lac pro) strA thi fis::KmRKoch et al. (1988)
 MC4100F y araD139 Δ (argF-lac)U169 rpsL150 relA1 flbB5301 deoC ptsF25 rbsRCasadaban (1976)
S. Typhimurium
 LT2Wild typeB.N. Ames
 SL1344rpsL hisGHoiseth and Stocker (1981)
 SL1344 fisSL1344 fis::CmRKeane and Dorman (2003)
 SL1344 rpoSSL1344 rpoS::KmRKeane (2002)
 SL1344 fisrpoSSL1344 rpoS::KmRfis::CmRThis study

Construction of gfp transcriptional fusions

Plasmids pZepfisS and pZepfisE contained the fis promoters from S. Typhimurium strain SL1344 and E. coli strain CSH50, and were constructed as follows. The fis promoter regions were amplified from genomic DNA isolated from strain SL1344 or CSH50 using the primer pairs fisSF1 plus fisSR1 or fisEF1 plus fisER1 respectively (Table 2), and were ligated into the promoterless gfp gene in vector pZep08 (Hautefort et al., 2003). Plasmid pZeprpsM was constructed by using the primer pairs rpsMF1 and rpsMR1 (Table 2) to amplify the rpsM promoter, which was ligated into pZep08. Plasmids were introduced into strain SL1344 by electroporation (Sambrook and Russell, 2001) and transformants selected by plating on chloramphenicol. Plasmid structures were confirmed by amplification of the inserted DNA by the PCR, followed by DNA digestion using diagnostic restriction endonucleases.

Table 2.  Oligonucleotide primers.
Primer nameSequencea
  • a. 

    Restriction enzyme cleavage sites are underlined.

  • b. 

    Primer tailed with restriction site NotI.

  • c. 

    Primer tailed with restriction site XbaI.

  • d. 

    Primer tailed with restriction site SmaI.

  • e. 

    Primer tailed with restriction site HindIII.

rpsmrtf15′-GCG TTA CTT CGA TCT ACG G-3′
rpsmrtr15′-GGT ACG CTG ACC GCG AAC CG-3′

Flow cytometry

To measure GFP levels in Salmonella, cells were fixed in 2% formaldehyde in phosphate-buffered saline (PBS), then diluted to a concentration of approximately 106 bacteria per ml, and then analysed using a Beckman Coulter Epics XL flow cytometer. In total, 10 000 bacteria were assayed for each sample, and results were expressed as the mean channel fluorescence. Analyses were carried out using the Expo32 ADC program. Each experiment involved samples analysed in triplicate, and each experiment was in turn carried out a minimum of three times.

Epithelial cell invasion

The epithelial cell lines CHO and CACO2 were maintained in DMEM F12 medium supplemented with 10% fetal calf serum. Monolayers for bacterial invasion were prepared by seeding 2 × 105 cells into each well of 12-well tissue culture plates. Bacteria were added to the cultured cells at a multiplicity of infection of 100:1 and incubated for 1 h. Cells were then washed twice in PBS to remove non-adherent/non-invasive bacteria, and were re-incubated in tissue culture medium containing gentamycin (100 μg ml−1) for 1 h to kill any remaining extracellular bacteria (Marshall et al., 2000). The cells were washed again before incubation in tissue culture medium containing gentamycin (20 μg ml−1). At each time point, cells were further washed twice in PBS before being lysed in 0.5% Triton X-100 and plated for colony counting. Samples were taken at 2, 4, 6 and 8 h post infection. Invasion assays were carried out in triplicate on at least three separate occasions.

Construction of arabinose-inducible rpoS and fis genes

The complete open reading frames of the S. Typhimurium rpoS and fis genes were amplified by PCR using primers rpoSf1 and rpoSr1 and fisf1 and fisr1 respectively (Table 2). The amplicons were treated with Klenow reagent before being digested with HindIII and cloned into HindIII- and SmaI-digested plasmid pBAD24 (Guzman et al., 1995), placing the rpoS and fis open reading frames under the control of the arabinose-inducible and glucose-repressible PBAD promoter. Plasmid construction was carried out in the E. coli strain XL-1 blue, and the resulting recombinant plasmid, pBADrpoS, was introduced into S. Typhimurium strain SL1344 rpoS by electroporation (Sambrook and Russell, 2001). Plasmid pBADfis was introduced into S. Typhimurium strain SL1344 fis by the same method.

Protein purification and Western immunoblotting

Bacterial cultures normalized to equal OD600 values were lysed by boiling in final sample buffer containing 15% v/v 2-β-mercapthoethanol (Sambrook and Russell, 2001). Total proteins were separated by sodium dodecyl sulphate-polyacrylamide gel electrophoresis (16% polyacrylamide) and then transferred to nitrocellulose by electroblotting using a Trans-blot electrophoretic transfer cell (Bio-Rad). Equal loading of lanes was confirmed by densitometry of Ponceau S-stained membranes before they were incubated overnight in blocking solution (5% w/v non-fat dry milk in PBS). Membranes were probed overnight at 4°C in blocking solution containing rabbit antiserum raised against Fis (1/500 dilution) or RpoS (1/1000). The membranes were washed three times for 10 min in PBS before being incubated with blocking solution containing mouse anti-rabbit IgG (1/1000 dilution) conjugated to peroxidase for 1 h. Blots were washed as before, and antigen–antibody complexes were detected using the Supersignal West Pico chemiluminescent substrate (Pierce). Each blot was replicated a minimum of three times. Band intensities were measured by densitometry using Quantity One software (Bio-Rad). Standard deviations were ±10%.

RNA isolation and fis transcript analysis by RT-PCR

Total RNA was extracted using the TRI ReagentTM (Sigma-Aldrich) according to the manufacturer's guidelines and treated with DNase using the DNA-freeTM kit (Ambion) to ensure no DNA contamination. RT-PCR analysis was carried out using the Onestep RT-PCR kit (Qiagen). In total, 1.2 μg of DNA-free RNA was used as a template for gene-specific primer pairs fisrtf and fisrtr (fis) and rpsmrtf and rpsmrtr (rpsM) (Table 2). Reactions involved one cycle of 50°C for 30 min, 95°C for 15 min, followed by 20 cycles of 94°C for 30 s, 50°C for 30 s, and 72°C for 1 min, followed by a final extension of 72°C for 10 min. The RT-PCR products were electrophoresed through 1% (w/v) TAE agarose gels and stained with ethidium bromide, and then quantified by densitometry using Quantity-One software (Bio-Rad, Hercules, CA).


We thank M.W. Mangan and R.K. Carroll for useful discussions and R. Hengge for anti-RpoS antibody. This work was supported by a grant from Science Foundation Ireland.