Salmonella enterica is a Gram negative facultative intracellular bacterial pathogen that is capable of causing disease in a variety of hosts. Salmonella enterica consists of more than 2600 serovars of which S. enterica subspecies enterica is comprised of serovars like S. Typhi that are host-specific, while others like S. Typhimurium are generalists (Porwollik et al., 2004; Gal-Mor, 2019). Typhoid fever and invasive disease are typical symptoms that are associated with human infections caused by S. Typhi (Dougan and Baker, 2014). Non-typhoidal Salmonella (NTS) species mostly cause self-limiting gastrointestinal illness which may lead to hospitalization (requiring antibiotic treatment) and death. However, in sub-Saharan Africa, NTS species cause invasive disease (iNTS) that manifests as bacteremia in 8–45% of all community-acquired blood stream infections resulting in an overall case fatality rate of 20.6% (Haselbeck et al., 2017; Uche et al., 2017). The main source of NTS infections are the consumption of contaminated poultry products, though recently many outbreaks have been linked to fresh produce (Hanning et al., 2009). The global burden of NTS is estimated at 93.8 million cases of human infections resulting in 155,000 deaths annually (Ao et al., 2015). According to the CDC, NTS species are the second leading cause of foodborne illness in the U.S. which is estimated at 1.2 million annual cases of Salmonellosis. This results in 19,336 hospitalizations, 17,000 quality adjusted life years lost and USD $3.3 billion in total medical expenditures and lost productivity annually (Cummings et al., 2012). In addition, S. Enteritidis and S. Typhimurium are the most common serotypes associated with human infection in Africa, Asia, Europe, Latin America and the Caribbean, and in North America (Galanis et al., 2006). However, in Saudi Arabia very little is known about the prevalence of Salmonella serovars or their virulence properties associated with human infection. This information is vital for determining the source of these infections in order to develop intervention strategies aimed at reducing the levels of Salmonella species across the country (Gast, 2007). Hence, NTS human infections translate into a significant impact on both the healthcare system and the economy of a given country (Ghunaim and Desin, 2015).

During the process of infection, Salmonella uses two specialized nanomachines known as Type 3 Secretion Systems (T3SS) that are encoded by Salmonella Pathogenicity Island-1 (SPI-1) (Collazo and Galán, 1997) and SPI-2 (Hensel, 2000), respectively, to inject virulence factors directly into host cells. Traditionally, the SPI-1 T3SS has been associated with bacterial invasion of the host intestinal epithelial cell surface, while the SPI-2 T3SS has been linked to intracellular survival and maintenance of the Salmonella Containing Vacuole (SCV) (Galán, 2001). However, there has been recent evidence suggesting that there is a certain level of cross-talk between the two T3SS, indicating that the two systems are closely linked in their virulence functions (Brown et al., 2005; Coburn et al., 2005; Coombes et al., 2005). Many in-vitro and in-vivo studies have shown that the two Salmonella T3SS are important virulence factors as the deletion of these systems results in attenuation (Zhang et al., 2002; Hapfelmeier et al., 2004; Raffatellu et al., 2005; Desin et al., 2009; Wisner et al., 2010; Buckner et al., 2011). In human infections, the roles of SPI-1 and SPI-2 have not been defined as clearly, though a study in China has indicated that clinical isolates from a food-borne disease outbreak of S. Senftenberg lacking SPI-1 were still able to cause gastroenteritis (Hu et al., 2008). Several genes that encode for T3SS components which consist of structural or secreted effectors including invA (SPI-1 structure), sipA (SPI-1 effector), sopE2 (SPI-1 effector), sifA (SPI-2 effector), and ssaR (SPI-2 structure) have been used as markers to test for the presence of these pathogenicity islands (Hu et al., 2008).

The emergence of antibiotic resistant bacteria worldwide is a significant public health concern resulting in 23,000 deaths per year in the U.S. at a cost of over $20 billion (Centers for Disease Control and Prevention, 2013). Moreover, drug-resistant NTS have been classified as a serious threat by the CDC with reports of increasing resistance to ceftriaxone and ciprofloxacin which are used as first-line antibiotic treatment. Alternative antimicrobials such as aminoglycosides, folic acid pathway inhibitors or the carbapenems are used for treatment of resistant NTS infections (Frye and Jackson, 2013). Therefore, from a patient safety perspective, it is critical to monitor antibiotic resistance patterns associated with human Salmonella infections. The objective of this study was to determine the prevalence of Salmonella serovars associated with human infections in Saudi Arabia using the rapid CTS system and to determine if the isolates contain SPI-1 and SPI-2 as well as characterizing the associated antimicrobial resistance.

NTS species are a public health concern as they are prevalent worldwide causing a significant number of human infections (Gal-Mor, 2019). Rapid identification of Salmonella species associated with human infections is a critical component of a Salmonella control program since this allows for epidemiological investigations that can assist in determining the source of the infections (Ferrato et al., 2017). In addition, the presence of important virulence factors such as SPI-1 and SPI-2 enhance our understanding of the virulence potential of these strains (Dos Santos et al., 2019). Similarly, characterization of antimicrobial resistance associated with these strains can further assist in reducing the spread of antibiotic resistance mechanisms within Salmonella species (Parisi et al., 2018). The aim of the present study was to serotype Salmonella species associated with human infections in Saudi Arabia using a rapid method (CTS system) and to characterize the clinical isolates by testing for the presence of SPI-1 and SPI-2 virulence determinants as well as determining antimicrobial resistance patterns.

We chose to use the CTS system since it is a rapid technique that can be used for serotyping Salmonella species with a turnaround time of 8 hours. This system has been shown to be highly effective and accurate in determining Salmonella serotypes (Wattiau et al., 2008a; Ferrato et al., 2017). The CTS system was able to identify 96.5% of the Salmonella clinical isolates used in the current study. This establishes a fast and reliable method for the identification of Salmonella species associated with human infections in Saudi Arabia and in the region. The remaining isolates were identified using MLST which has been used successfully for the serotyping of Salmonella species (Achtman et al., 2012). The results from MLST imply that some of the serotypes not identified by the CTS were rare serotypes like S. Obogu and S. Telhashomer that are not present in the CTS database. In addition, ST3774 and the strain that was not identified by MLST could possibly be novel serotypes that are associated with human infection in Saudi Arabia. Our results indicate that the predominant serovars associated with human infections in Riyadh are S. Enteritidis and S. Typhimurium, followed by S. Livingstone, S. Kentucky and S. Poona, respectively. This is the first report in the region which demonstrates that besides S. Enteritidis and S. Typhimurium being the leading cause of human infections, serotypes S. Livingstone, S. Kentucky, and S. Poona are also significant. Moreover, the data also show that 36 unique Salmonella serotypes were circulating among the human population in Riyadh (Saudi Arabia). The fact that S. Enteritidis and S. Typhimurium are the most common serotypes is in agreement with surveillance data from the U.S. reported by FoodNET (Marder Mph et al., 2018) and the European Union (Gal-Mor, 2019). These results are also supported by a preliminary study performed in Saudi Arabia where the authors examined 33 isolates which included Typhoidal Salmonella species (El-Tayeb et al., 2017). Our results differ with respect to the third, fourth and fifth most common serovars since FoodNET has indicated that they are S. Newport, S. Javiana and the monophasic variant of S. Typhimurium (I 4, [5],12:i:-) (Marder Mph et al., 2018), in decreasing order, while in the European Union they are monophasic S. Typhimurium, S. Infantis and S. Derby (Gal-Mor, 2019). The differences seen in the list of prevalent serovars may arise due to differences in food consumption habits, geographical locations and climate which may give certain serovars advantages over others. As well, a larger sample size of Salmonella strains in Saudi Arabia may reveal a different distribution of serotypes across the country.

The main Salmonella serovars identified in this study, S. Enteritidis and S. Typhimurium, were also isolated from non-fecal samples such as the blood. This may suggest that these strains, especially S. Enteritidis, are more invasive and have the potential to cause systemic infection compared to the other serotypes. However, this conclusion has to be confirmed with a larger sample size that includes regions from across the country as it is possible that this observation is only being seen since the aforementioned serotypes are more frequently isolated. As well, we are unable to determine if any of the isolates caused iNTS at this time as there is no data available on the presence of these strains in Saudi Arabia or in the region.

Traditionally, the prevailing view was that SPI-1 was mainly associated with intestinal invasion of epithelial cells, while SPI-2 was involved in systemic infection (Hansen-Wester and Hensel, 2001). The observation that there is a certain level of cross talk between the two secretion systems has led to an understanding that both SPI-1 and SPI-2 are important virulence factors during the process of infection (Moest and Méresse, 2013). However, the majority of the data has been derived from cell culture models or animal models of infection. Relatively, the role of SPI-1 and SPI-2 in human infections is still not clearly defined. The present study was an attempt to determine if SPI-1 and SPI-2 are present in clinical isolates associated with human disease in Saudi Arabia. Our data clearly show that SPI-1 was present in the majority of the Salmonella strains and that strains missing SPI-1 genes only had a single missing gene like sopE2. The absence of the aforementioned SPI-1 genes indicates that these strains are capable of forming a fully functional SPI-1 T3SS with a minimal effect, if any, on virulence since the effector proteins encoded by these genes are known to have important but redundant roles (Zhang et al., 2018). On the other hand, the SPI-2 gene sifA was not detected in several isolates, while the ssaR gene (SPI-2) was present in all the samples. This demonstrates that an intact SPI-2 T3SS was likely present in the sifA deficient isolates, since this gene is located outside SPI-2 and plays a role in the formation of Salmonella induced filaments and maintenance of the Salmonella containing vacuole (Stein et al., 1996). However, sifA mutants are known to be highly attenuated in systemic infection and intracellular replication in mice (Beuzón et al., 2000; Rajashekar et al., 2014), thereby suggesting that isolates with these genotypic characteristics might have a reduced systemic infection potential in humans, though the exact role of SPI-2 in human infection remains to be elucidated. Our SPI-1 results are not in agreement with a study in China which indicated that SPI-1 was not required for human gastroenteritis since two S. Senftenberg strains isolated from a food-borne disease outbreak were SPI-1 deficient (Hu et al., 2008). However, our finding that SPI-2 genes, with the exception of sifA, were present in all the test strains imply that SPI-2 is associated with human disease and are in accordance with the findings of the former study (Hu et al., 2008). Another noteworthy observation was that 97% (28/29) of the sifA deficient strains were only isolated from stool samples (data not shown), implying that SPI-2 is important in human infection. However, a detailed analysis of the exact role of the individual SPI-1 and SPI-2 effectors during human enterocolitis remains to be elucidated.

Antibiotic resistance has been reported in Salmonella species since the 1990's (Threlfall et al., 2000) and it appears that resistance is on a rise (Frye and Jackson, 2013). Our findings indicate that antibiotic resistance among the strains tested were the highest to nitrofurantoin, followed by ampicillin, cefalotin, tigecycline, ciprofloxacin, and trimethoprim/sulfamethoxazole. The rate of resistance to ampicillin is in agreement with a previous report from the U.S. where resistance was 14% among 23,585 human Salmonella isolates (Frye and Jackson, 2013). However, Ciprofloxacin resistance in our study (8%) was significantly higher and in contrast to the findings of this report (0.11%), but in agreement with data from countries in the Arab League (Moghnieh et al., 2018). The U.S. study did not assess resistance to Nitrofurantoin, Cefalotin or Tigecycline making the comparison with our data difficult (Frye and Jackson, 2013). Nevertheless, resistance to other antibiotics in our study like Amikacin, Gentamicin, Amoxicillin/Clavulanic acid and Ceftriaxone were similar to those reported. Additionally, our data also suggest that S. Agona, S. Give, S. Havana, S. Kentucky, and S. Minnesota are MDR strains, which is in line with similar observations by others for serotypes like S. Agona (Kuang et al., 2014; Irfan et al., 2015), S. Kentucky (Le Hello et al., 2013; Seiffert et al., 2014), and S. Minnesota (Campos et al., 2018), while it is rare for S. Give and S. Havana (Bekal et al., 2013).

Our observation that 70% of our test strains conferring β-lactam resistance contained the bla_TEM−1 gene while very few strains contained bla_PSE−1/bla_CARB−2 and bla_OXA−1 is also supported by a report from the UK where the authors examined 3,491 Salmonella isolates (Neuert et al., 2018). These results are partially in agreement with a study performed in Shanghai where β-lactam resistance was distributed between the bla_TEM and bla_OXA genes (Wang et al., 2017). However, our results are in contrast to data from France where the bla_PSE−1 gene was predominantly associated with β-lactam resistance whereas the bla_TEM gene was detected at low levels (Weill et al., 2006). The latter two studies may not be an accurate comparison with our work since they focused on S. Typhimurium. The presence of CTX-M-type extended spectrum β-lactamases was detected in only 2 isolates (S. Give and S. Kentucky) in our study and is similar to what has been previously published (Wang et al., 2017; Neuert et al., 2018). Our finding that the MDR carbapenem S. Kentucky strain contained the bla_OXA−48 gene is a significant finding and the first report of its kind in Saudi Arabia. The association of this strain with human infections is concerning due to the fact that an increase in the number of such strains can lead to new challenges faced by physicians who encounter patients infected with this strain. Similar findings associated with S. Kentucky are rare, but have been reported previously in other countries (Le Hello et al., 2013; Seiffert et al., 2014), implying that this serotype may be an emerging MDR pathogen and may require increased surveillance in order to minimize outbreaks related to such strains. The association of mutations with the gyrA and parC genes when associated with ciprofloxacin resistance is evident from our data since an MIC >4 μg/ml resulted in two point mutations in the gyrA gene (codons 83 and 87) and parC gene (codons 57 and 80) as shown elsewhere (Weill et al., 2006; Wang et al., 2017; Neuert et al., 2018).

Taken together, we have established that the CTS system can be used to rapidly serotype Salmonella species in Riyadh. We have demonstrated that S. Enteritidis and S. Typhimurium are the most prevalent serovars associated with human infection in Riyadh (Saudi Arabia) and that S. Livingstone, S. Kentucky and S. Poona are also significant among a list of 36 serovars. Our data further suggest that important virulence factors like SPI-1 and SPI-2 were present in these isolates and strains lacking sifA were not associated with systemic infection. However, extensive work is still required to determine the roles of these T3SS in human enterocolitis and systemic infection. We also report the finding of a unique MDR resistant S. Kentucky isolate conferring carbapenem resistance. Further, our results indicate that there is a rising trend of ciprofloxacin and nitrofurantoin resistance among Salmonella species in the region.

The raw data supporting the conclusions of this manuscript will be made available by the authors, without undue reservation, to any qualified researcher.

TD, AM, SA, HB designed the study. TD, AM, and SA assisted in sample collection. MA did the majority of the experimental work, while MAZ was instrumental in MLST PCR and confirming carbapenem resistance. TD analyzed all the serotyping data. MFA analyzed all the MLST Data and the sequencing results. MD and TD led the AMR work and analysis. TD and WK led the detection of SPI-1 and SPI-2 and analyzed the results. TD wrote the manuscript with major contributions from MA, MD, and WK. AM, SA, and HB critically read the manuscript with important feedback and input. All the authors contributed to, read and approved the final manuscript.