Gene expression analysis of Salmonella enterica SPI in macrophages indicates differences between serovars that induce systemic disease from those normally causing enteritis.

Global gene expression of the invasive Salmonella serovars S. Enteritidis and S. Typhimurium, and the less-invasive S. Infantis and S. Hadar was studied during infection of a chicken macrophage cell line. Major functional gene groups responsible for intracellular physiological changes were regulated similarly in all four serovars. However, SPI1 and SPI4 genes of S. Enteritidis and S. Typhimurium were strongly repressed in the macrophages whereas S. Infantis, S. Hadar and other similar serovars maintained up-regulation of these gene sets. This phenomenon may explain some of the biological differences between invasive and non-invasive Salmonella serovars.

Introduction

Salmonella serovars are an important cause of human and animal disease ranging from enteritis to typhoid-like infections associated with high mortality. In countries where human waste does not mix with drinking water, most human Salmonella infections arise as a result of food poisoning. Poultry meat is implicated as possibly the major source of human Salmonella infections (European Food Safety Authority, 2013). Salmonella enterica serovars Enteritidis and S. Typhimurium are the serovars most frequently isolated from humans and poultry with S. Infantis and S. Hadar of increasing significance as Typhimurium and Enteritidis are brought under control in poultry (European Union, 2003; European Food Safety Authority, 2013; Hauser et al., 2012).

Salmonella organisms usually infect by the oral route. The pathogen penetrates the intestinal mucosal epithelium and survives in phagocytic cells (Sterzenbach et al., 2013). Systemic disease requires translocation to organs rich in macrophage–monocyte series cells, such as the spleen and liver, where they survive and multiply. Virulence requires Salmonella pathogenicity islands (SPIs) (Sterzenbach et al., 2013). In Salmonella enterica SPI1 and SPI2 (Marcus et al., 2000), are essential for invasion of epithelial cells and an ensuing enteritis and systemic disease, respectively (García-del Portillo, 2001; Groisman and Ochman, 1997). During epithelial invasion the majority of SPI1 genes are up-regulated while SPI2 genes are down-regulated; this pattern is reversed inside the macrophage phagosome (Eriksson et al., 2003; Hautefort et al., 2008).

Although S. Enteritidis and S. Typhimurium cause severe enteritis and systemic disease in several host species, S. Infantis or S. Hadar are usually limited to the gastrointestinal tract with mild clinical symptoms (Barrow et al., 1994). Since virulence determinants are highly conserved among these serovars (Huehn et al., 2010), these differences are difficult to explain at the molecular level. For a better understanding of this, gene expression in these serovars was studied in cells in vitro by using microarrays.

Materials and methods

Bacterial strains and culture

Salmonella Enteritidis (SE) P125109, S. Typhimurium (ST) 4/74 (SL1344), S. Infantis (SI) 1326/28 and S. Hadar (SH) 18 (http://www.sanger.ac.uk/Projects/Salmonella) were used. The SE strain is highly invasive for newly-hatched birds and colonises the intestine (Barrow, 1991). The ST is virulent for mice, pigs and poultry (Aabo et al., 2002). The SI and SH strains colonise the chicken intestine but are poorly invasive in vivo although fully invasive in vitro (Barrow et al., 1994; Berndt et al., 2007) as has been found with additional strains of these serovars (Barrow, unpublished).

Cell culture and infection model

Avian macrophage-like HD11 cells (Beug et al., 1979) were used. Culture and infection protocols were as described previously (Setta et al., 2012). Sampling points were at 4 and 8 h post infection when the cells were lysed for bacterial culture or RNA extraction. Bacterial RNA was also harvested from bacteria which had been cultured in the cell culture medium for 2 h after a 20-fold dilution from the overnight nutrient broth culture to produce bacterial numbers similar to those applied to the cell monolayers. This acted as control RNA (late log-phase in vitro culture).

RNA extraction and processing

The RNA extraction protocol was that of Eriksson et al. (2003). The quality and host RNA contamination of bacterial RNA isolated from HD11 was checked by a 2100 Bioanalyzer (Agilent). The RNA obtained was first amplified using the MessageAmp™ II-Bacteria Kit (Ambion) according to the manufacturer's instructions, resulting in aminoallyl-UTP labelled amplified RNA (aRNA). Labelling with Cy3 or Cy5 (Amersham) was done following manufacturer's instructions. Cy3 was used for in vitro control RNA, with Cy5 for bacterial RNA extracted from macrophages.

Microarray design and data analysis

At the time of the experiments fully annotated genome sequences were only available for S. Enteritidis P125109. The sequence of S. Enteritidis P125109 and the un-annotated genome sequences of S. Typhimurium SL1344, S. Infantis 1326/28 and S. Hadar 18 (N. Thomson, pers. comm.) formed the basis of our custom microarray design. All the ORFs predicted were regarded as potentially transcribed genes for probe design. Designing microarray probes was done with the Agilent eArray system (https://earray.chem.agilent.com/earray/) with the following settings during the microarray probe design: Tm (70 °C) matching methodology, 60-mer probe length, 3 probes/target. The protocol is described in detail at http://www.ebi.ac.uk/arrayexpress/. Data analysis was done using GeneSpring GX 10.0 (Agilent).

Quantitative real time-PCR

This was done according to the method of Setta et al. (2012).

Results

Bacterial kinetics during intracellular infection

With an initial bacterial count of (6–8) × 108 CFU/ml for infection, the viable counts inside the HD11 macrophages were between 106 and 107 CFU/ml in all serovars 4 h after infection and did not drop below 106 even after 8 h. Bacterial counts started to decrease to 104 by 24 h (data not shown). The HD11 cells also began to deteriorate around 12 h p.i. with a rough cell surface and membrane disorganisation. RNA was sampled at 4 and 8 h p.i. and each experiment repeated three times.

Microarray analysis of gene expression

We focused on genes whose expression might be affected in the Salmonella-containing vacuoles (SCV), i.e. genes coding for membrane proteins/surface antigens, virulence determinants and sensing and responding to environmental changes.

The following functional groups showed significant changes in all serovars in comparison with control bacteria grown in cell culture medium: (i) oxidative stress response genes including the non-specific export systems responsible for efflux of harmful agents present in the SCV, (ii) surface-protein synthesis genes, (iii) ion transport systems importing inorganic solutes (all this data not shown), and (iv) virulence genes. Summary tables containing the normalised GE data of these functional groups can be found in the Supplemental material (Tables S1 and S2). Increased-expression of oxidative stress response elements (soxRS and katE), membrane stress and efflux systems (pspABCDE and marRAB) and SOS response genes (sulA and recA) was observed across the four strains representing these four serovars. Small changes in expression of genes associated with uptake/metabolism of iron were found with equal numbers of iro, ent and sit genes either very slightly down-regulated or up-regulated (<5-fold).

The majority of surface antigens, including OMPs, LPS, flagellar components and certain fimbrial proteins, were significantly down-regulated.

The most characteristic changes in ion transport involved the remarkable (in some cases several hundred fold) over-expression of magnesium (mgtA, mgtBC), manganese (mntH) and phosphate (pst, phoBR) uptake systems (supplementary tables).

Validation of microarray data

Seven up- and down-regulated genes (invJ, ssaM, mgtC, invJ, pspA, fliM and flgF) belonging to different functional groups, were selected for quantitative real-time PCR analysis. We used rimM, coding for a 16S rRNA processing protein, as a reference; its expression was unchanged for all four serovars. Each analysis was carried out in triplicate. The results correlated well with the level of gene expression changes with Spearman correlation coefficient (ρ) of between 0.857 and 0.884 (ST, 8 h and 4 h, respectively), 0.928 (SE, both time points) and 0.964 (SH and SI both time points) (data not shown).

Expression of SPI genes

Most SPI genes were also uniformly regulated across the four serovars. Those responsible for intra-macrophage survival (SPI2 and SPI3) were highly expressed (Fig. 1). For clarity the data for S. Enteritidis and S. Infantis only are shown.

Fig. 1

The majority of SPI2 genes responsible for intra-macrophage survival were uniformly up-regulated in S. Enteritidis and S. Infantis.

Most of the SPI1 and SPI4 genes, required for epithelial invasion but dispensable for macrophage intracellular survival, were down-regulated in SE and ST, compared to the control RNA pool representing late log-phase in vitro cultured bacteria (Fig. 2). All SI SPI genes showed the opposite expression pattern of that seen in SE and ST. Of 35 SPI1 genes 27 were significantly (>1.5-fold) up-regulated in SI, as opposed to the 25 and 23 down-regulated genes out of the 35 in SE and ST, respectively. Also, five out of six SPI4 genes were up-regulated in SI, contrary to the results of the other three serovars. The expression levels in SH were intermediate between those of SI and SE and ST.

Fig. 2

Contrasting expression profiles of invasion related virulence genes in S. Enteritidis and S. Infantis inside macrophages. SPI1 and SPI4 genes were expected to be only activated during epithelial invasion and strongly repressed during intramacrophage survival. Surprisingly, all these invasion-related genes showed the opposite expression pattern in S. Infantis of what was observed in the invasive serovars.

Q-RT-PCR analysis was carried out using SPI1 specific primers on cell cultures of individual strains of additional non-invasive serovars, S. Anatum (serogroup E1), S. Kedougou (serogroup G), S. Montevideo (serogroup C1) and one additional strain of S. Infantis and two of S. Hadar not used for the first microarray experiments. In this case the macrophage experiments involved sampling at 4 h only with each done in triplicate. The expression results obtained from HD11-cultured bacteria indicated unambiguous up-regulation of orgB and spaR (both SPI1) and siiB (SPI4) genes (2.2–6.8-fold, 2.4–12.5-fold and 1.5–3.6-fold, respectively) in all the strains tested. In ST and SE the changes in levels of expression of these genes were −1 to −2-fold.

Discussion

This study was carried out to determine whether the study using S. Typhimurium in mouse macrophages (Eriksson et al., 2003) was applicable to serovars associated with poultry if avian macrophage-like cells were used. The gene expression patterns for normal physiological processes were similar for both host species despite the fact that HD11 cells differ in a number of properties from monocyte-derived macrophages.

The transcriptome data of the strains representing the four Salmonella serovars analysed in this study generally were highly concordant and are expected responses to the nutritional and other environmental conditions within the SCV indicating that these factors in the mouse and chicken are similar (Eriksson et al., 2003; Hautefort et al., 2008). Expected results include increases in the oxidative stress response elements (soxRS and katE), membrane stress and efflux systems (pspABCDE and marRAB) and SOS response genes (sulA and recA) (Eriksson et al., 2003; Lynch and Lin, 1996; Walker, 1996), and over-expression of magnesium (mgtA and mgtBC), manganese (mntH) and phosphate (pst and phoBR) uptake systems (Eriksson et al., 2003; Smith et al., 1998). One exception was iron uptake. Our data suggested iron shortage inside HD11 vacuoles while iron is thought to be easily accessible in the mouse. The J774-A.1 cells they used originated from an Nramp1− (BALB/c) mouse, potentially defective in intracellular iron translocation, which may explain the differences observed.

The expression pattern of invasion virulence genes in SI and SH inside macrophages was unexpected. Most SPI1 and SPI4 genes are associated with epithelial attachment and invasion (Morgan et al., 2004; Gerlach et al., 2007) and are activated during the first phase of infection in the intestine (Bajaj et al., 1996). Similar patterns in other less-invasive, serovars S. Anatum, S. Kedougou and S. Montevideo and an additional strain of SI and two of SH were also seen.

The expression pattern of virulence genes required for invasion in the non-invasive serovars SI and SH inside macrophages suggests the hypothesis that the reduced virulence of less-invasive food poisoning serovars, which cause enteritis but rarely systemic infection, is linked with the inability to down-regulate SPI1 inside macrophages. These bacteria might be less efficient in the SCV due to (i) the metabolic costs of unnecessary gene expression and possibly because (ii) the loose regulation of SPI1 genes inside phagocytes might trigger a more effective innate host response against Salmonella. It has been demonstrated that the SPI1 effector, SopB, stimulates nitric oxide production (Drecktrah et al., 2005), and that SI triggers a stronger cytokine response and inflammatory response in alveolar macrophages than does ST (Volf et al., 2010).

Most invasion genes encoded by SPIs in S. Infantis and other less virulent serovars from a variety of serogroups showed the opposite expression pattern to that observed in the invasive serovars Typhimurium and Enteritidis, i.e. SPI1 and SPI4 genes generally remained up-regulated during avian macrophage infection. The expression pattern of the same genes in S. Hadar was intermediate.

How these attributes are regulated and how they arose, since they are found in several different serovars, would be interesting to study further. Construction of ST and SE strains unable to down-regulate SPI1 genes would be an interesting follow up experiment. Regulation of both is complex and SPI1 and SPI4 genes are known to be jointly regulated by HilA master regulator of SPI1 (Saini and Rao, 2010). Genetic variation between serovars is also known to contribute to differences in the virulence phenotype (Suez et al., 2013). It is tempting to speculate that these genetic differences might also contribute to the unusual expression pattern of these non-invasive serovars. We suggest that the reduced virulence of the non-invasive food poisoning serovars is linked to the absence of or inability to down-regulate SPI1 inside macrophages.