Salmonella is the most common cause of bacterial food-borne illness in the U.S. and is estimated to annually cause over 1 million cases, 19,000 hospitalizations, 350 deaths, and $2.6 billion in social costs [1, 2]. Salmonella enterica serovar Typhimurium is one the most prevalent salmonellae in humans and livestock, and many of these cases are found to be resistant to multiple antibiotics. According to the National Antimicrobial Resistance Monitoring System (NARMS), 27-83% of S. Typhimurium isolates from humans, chicken, cattle, and swine were found to be resistant to three or more classes of antibiotics [3]. A recent Salmonella Typhimurium isolate linked to an outbreak associated with ground beef was resistant to eight antibiotics: amoxicillin/clavulanic acid, ampicillin, ceftriaxone, cefoxitin, kanamycin, streptomycin, sulfisoxazole, and tetracycline [4].

Multidrug-resistant (MDR) Salmonella is associated with increased morbidity in humans and increased mortality in cattle relative to sensitive strains [5, 6]. There are several non-exclusive rationales for these clinical observations [7, 8]. One explanation is treatment failure, where the administered antibiotic is ineffective due to bacterial resistance, and therefore the infection persists and the illness progresses. Another explanation is that the normal gut flora is disrupted by an antibiotic regimen, thereby increasing the risk of an opportunistic infection by drug-resistant bacteria. Finally, there is the possibility that antibiotics can directly enhance bacterial virulence; this concept is supported by several publications reporting that certain antibiotics can alter virulence factors in some bacteria in vitro[9–12], including tetracycline in S. Typhimurium definitive phage type DT104 [13]. However, the report by Weir et al. tested a single DT104 isolate at a single tetracycline concentration during late-log growth and identified a significant change in virulence gene expression, while an earlier report by Carlson et al. evaluated over 400 DT104 isolates exposed to tetracycline that were grown to stationary phase and did not observe any isolates with a significantly increased ability to invade cells in culture [14]. Resistance to tetracycline is prominent among S. Typhimurium isolates in humans (34%), chickens (39%), cattle (59%), and swine (88%) according to a ten-year average from the National Antimicrobial Resistance Monitoring System [3, 15]; thus, our objective was to explore the relationship between gene expression and cellular invasion in response to tetracycline. We examined the effects of sub-inhibitory tetracycline concentrations on isolates of phage type DT104 and DT193 during early-log and late-log growth to determine the conditions, if any, that affect MDR Salmonella Typhimurium invasiveness after tetracycline exposure. We ascertained that an induced-invasion phenotype was a dose-dependent response due to the combination of two novel study parameters, early-log growth and DT193 isolates. We also found that expression of virulence genes can be tetracycline-induced during either early-log or late-log growth in many isolates, but this did not always correlate with increased invasiveness.

Multidrug-resistant Salmonella Typhimurium is a prevalent food safety and public health concern. Due to the fact that tetracycline resistance is frequently found in S. Typhimurium isolates from humans and livestock [3, 15], our goal was to test and characterize the conditions necessary to generate an invasive phenotype in MDR Salmonella following tetracycline exposure. Two common MDR S. Typhimurium phage types are DT104 and DT193, and these are typically resistant to three or more antibiotics, are found in humans and livestock, and have been associated with foodborne outbreaks [23–27]. DT104 and DT193 share a similar antibiotic resistance profile, but the genetics underlying their resistance phenotype differ. For instance, the majority of resistance genes in DT104 isolates reside in the Salmonella genomic island 1 on the chromosome, whereas the resistance genes of DT193 are typically encoded on plasmids. Also, DT104 isolates generally have only the tetG gene to confer resistance to tetracycline, but isolates of DT193 can have a combination of tetA-D genes and usually lack tetG[28, 29]. Since tetracycline is used therapeutically in humans and animals, and because most MDR S. Typhimurium isolates are resistant to tetracycline, our goal was to determine the effect and extent tetracycline exposure had on the invasiveness of Salmonella isolates from DT104 and DT193. We examined both cell culture invasion and virulence gene expression in vitro in response to tetracycline under a combination of three conditions: growth phase, tetracycline resistance genotype, and antibiotic concentration.

Cellular invasion is a temporally-regulated process in Salmonella that involves the activation of a sequence of genes, most importantly, hilA[21]. The hilA gene is the bottleneck in the process and its deletion prevents invasion from occurring, whereas its over-expression usually results in increased invasion [22]. The invasive response is initiated during early-log growth, and Salmonella is considered maximally invasive during the late-log growth phase [20]. We found that during early-log growth, tetracycline significantly increased cellular invasion in three isolates, while it significantly up-regulated the gene expression of virulence factors hilA, prgH, and invF in seven isolates. None of the isolates in the study had an increase in cellular invasion during late-log growth in response to tetracycline, but expression of virulence factors was up-regulated in several isolates. The increased invasiveness of the isolates during early-log was commensurate with the temporally-regulated invasive phenotype observed in each respective 0 μg/ml control isolate during late-log. Therefore, tetracycline exposure induced a shift in the invasion response to an earlier time in the growth cycle (early-log), yet tetracycline did not enhance invasion efficacy when invasion was already at its maximum potential in late-log growth. In addition, an increase in virulence gene expression did not always correlate with a reciprocal increase in invasion. The data demonstrates that the induction of invasion by tetracycline is a growth phase dependent response.

Several tetracycline concentrations were evaluated to determine if invasion induction was dependent on dose, or if the presence of tetracycline at any level would be effective. Three concentrations of tetracycline that did not inhibit growth of any of the isolates were chosen to study (1, 4, 16 μg/ml). The tetracycline-induced invasion response in the three isolates was only observed at 16 μg/ml. The induction of invasion by tetracycline is a dose dependent response.

DT104 and DT193 isolates that encode tetracycline resistance genes commonly found in S. Typhimurium (tetA, B, C, D, and G) were evaluated. The DT104 isolates all had SGI-1 and tetG, but no other tetracycline resistance genes were present. None of the DT193 isolates contained SGI-1. Of the five DT193 isolates, three had only a tetA gene, one had tetA, B, C, and D, and one had tetB, C, and D. Only the three DT193 isolates encoding a single tetracycline resistance gene, tetA, were more invasive during early-log growth at 16 μg/ml of tetracycline; the isolate that encoded tetA-D did not have this tetracycline-induced phenotype. All isolates, except the isolate encoding tetB-D (4584), had increased invasion gene expression following tetracycline exposure during early-log phase. Though a specific unknown mechanism that induces invasion in response to tetracycline may exist, it is not shared by all isolates and is independent of SGI-1. Induction of invasion due to tetracycline exposure is restricted to a subset of MDR S. Typhimurium isolates.

Previous work by Carlson et al. tested over 400 DT104 isolates that were exposed to tetracycline and grown to stationary phase, but no difference in invasion due to antibiotic treatment was observed [14]. Our data for the DT104 and DT193 isolates grown to late-log phase and then exposed to tetracycline are consistent with these results. Also, the increase in virulence gene expression during late-log growth after tetracycline exposure reported by Weir et al. [13] parallels our expression data. However, no previous study examined the effect of any antibiotic on DT193 or during early-log growth, and it was these two factors that were critical to observing the induction of the invasion phenotype due to tetracycline. The basis for the difference in response between DT193 and DT104 could be genetic content (e.g. the presence of additional virulence genes), the differential regulation of a particular response, or both.

Many studies have shown that antibiotics can directly or indirectly effect transcription and regulation of cellular processes [30–33]. In the current study, tetracycline up-regulated genes associated with virulence, but this was not always coincident with an increase in the invasive phenotype. The regulation of invasion is a complex network of interactions and responses, and it is possible that the tetracycline stimulus could affect targets downstream of hilA, invF, and prgH; such a response could up-regulate a repressor of invasion in the non-induced isolates. Genome sequencing of the isolates, plus transcriptomic analyses, will provide a more complete picture of what genes and processes are being affected by tetracycline exposure. Evaluation of other antibiotics would also discern if the response is specific to tetracycline, or if it is general to an antibiotic stress.

The response to tetracycline by some MDR S. Typhimurium isolates could provide a selective advantage to the bacteria by quickly and efficiently promoting entry into an intracellular niche within the host. Additionally, the use of efflux pumps to maintain viability in the presence of tetracycline is an active transport mechanism that requires energy to generate the proton gradient needed to drive the antiporter [34]; escaping such an environment would benefit the bacteria as fewer resources are required in the absence of the antibiotic. MDR S. Typhimurium already has a competitive advantage for survival due to its antimicrobial resistance phenotypes compared to sensitive strains, and our investigation identified an additional advantage of a tetracycline-inducible invasion phenotype that could influence colonization potential. Perhaps these factors are associated with the increased morbidity observed among MDR Salmonella patients.

We have found that tetracycline can induce invasion in MDR S. Typhimurium, and that this response is dependent on antibiotic concentration, growth phase, and isolate. It does not appear that the induction of Salmonella invasiveness is a universal phenotypic response, even though the majority of isolates had an increase in virulence gene expression; a significant increase in hilA gene expression was not an accurate indicator of increased cellular invasion. Knowledge of the parameters necessary to establish this phenotype is important to further elucidate the underlying factors associated with increased virulence of MDR Salmonella.