The need to ensure the safety of powdered infant formula (PIF) led the FAO/WHO to establish the microbiological or epidemiological relationship of microbial agents found in PIF with infant infection. They identified three categories of microorganisms based on evidence of a causal relationship between the presence of these microorganisms and the disease they cause. The first category of microorganisms with clear causality were identified as Cronobacter spp. and Salmonella enterica. The second consists of microorganisms for which causality is possible but has not yet been demonstrated. These were mainly from the Enterobacteriaceae family, but also included Acinetobacter. Finally, the third involves microorganisms for which causality is less likely or has not yet been shown and has not been identified in PIF. Based on this the FAO/WHO recommended the absence of Cronobacter spp. and Salmonella in PIF for target age less than 6 months (FAO/WHO, 2004, 2006; Forsythe, 2018).

Cronobacter is a genus of bacterial pathogens consisting of seven species: C. sakazakii, C. malonaticus, C. universalis, C. turicensis, C. muytjensii, C. dublinensis, and C. condimenti (Iversen et al., 2008; Joseph et al., 2012; Stephan et al., 2014). The species with the greatest clinical significance are C. sakazakii and C. malonaticus and have been reported in cases and outbreaks associated with PIF in infants (Forsythe, 2018; Parra-Flores et al., 2021a). At present in the United States, three cases of illness caused by PIF contaminated with C. sakazakii, resulting in one fatality, and one case of Salmonella Newport are being investigated; this has prompted an international voluntary recall of these PIFs by the manufacture (U.S FDA, 2022). The severity of the clinical condition has been associated with the presence of virulence factors encoded on plasmids (Shi et al., 2018; Aly et al., 2019), adherence and invasion traits (Cruz et al., 2011; Parra-Flores et al., 2018a; Holý et al., 2021), and various other genes such as aut, cpA, fliC, hly, ompA, sip, plas, and inv (Cruz et al., 2011; Franco et al., 2011; Aldubyan et al., 2017; Holý et al., 2019). Other factors are the use of sialic acid as a carbon source, capsule composition and the presence of its capsule, and endotoxin production (Ogrodzki and Forsythe, 2015). Another important aspect is the resistance to β-lactam antibiotics such as cephalothin, cefotaxime, ceftazidime, and ampicillin in addition to the presence of resistance genes such as marA, glpT, ampH, blaCSA, and mcr (Flores et al., 2011; Lee et al., 2012; Fei et al., 2017; Holý et al., 2021).

Salmonella enterica is a gram-negative, rod-shaped, facultative anaerobic genus. More than 2,600 serotypes belonging to S. enterica have been described worldwide que incluyen, which can cause diseases in humans and animals (Mezal et al., 2014). Salmonella is the most widely studied microbial pathogen, and can be isolated from a variety of foods, including PIF associated with disease outbreaks in infants (Angulo et al., 2008; Carrasco et al., 2012; Jourdan-da Silva et al., 2018). Gastroenteric Salmonella infections usual develop as self-limiting gastroenteritis, and antibiotic treatment is necessary only in severe cases more often associated with immunocompromised patients or those at the extremes of age such as infants (de Toro et al., 2014). Therefore, the emergence of strains that are resistant to β-lactams and cephalosporins is a relevant public health and food safety problem (Güerri et al., 2004; de Toro et al., 2011; Wang et al., 2019). In addition, Salmonella exhibits virulence factors that play a decisive role in systemic infections, such as pathogenicity islands (PAIs), invasion and adherence genes, and enterotoxin coding (Murugkar et al., 2003; Nayak et al., 2004; Huehn et al., 2010; Thung et al., 2018).

Salmonella and Cronobacter species are known persist in low-moisture foods such as milk powder and powdered infant formula for up to 2 years (Caubilla-Barron and Forsythe, 2007). Consequently, outbreaks due to the consumption of contaminated products have been reported (Forsythe, 2018; Jones et al., 2019). Cronobacter can survive spray-drying and persist in the manufacturing environment as biofilms. Genotyping has shown the persistence of specific Salmonella and Cronobacter strains within production facilities for many years (Craven et al., 2010; Jones et al., 2019). Contamination may occur post-pasteurization due to the addition of contaminated ingredients (FAO/WHO, 2008).

Whole-genome sequencing (WGS) has facilitated the in-depth study of pathogenic organisms by generating extensive information that helps to determine relationships and taxonomic differences between them (Leopold et al., 2014). It is not only used for isolate identification, but also extensive profiling and genotyping; such as conventional 7-loci multilocus sequence typing (MLST), core genome MLST (cgMLST) and/or single nucleotide polymorphism (SNP) analysis, molecular serotyping, CRISPR-Cas array profiling, and detection of genes associated with antibiotic resistance and virulence. Consequently, more precise epidemiological links can be established (Marraffini, 2013; Leopold et al., 2014; Uelze et al., 2020). Therefore, the analysis of the complete genomes and their comparison enables a more complete analysis of the pathogenesis process of C. sakazakii (Lehner et al., 2018).

In 2017, a recall of powdered formula samples contaminated with Cronobacter occurred in Chile (Parra-Flores et al., 2018b). This situation led to the incorporation of microbial criteria (n = 30; c = 0) for Cronobacter spp. in PIF intended for consumption by infants aged less than 12 months into the Chilean Food Sanitary Regulations (RSA; Parra-Flores et al., 2018b), given that the microbiological criteria in the RSA for Salmonella was already defined (n = 10; c = 0). This study considers the safety of PIF from 2018 to 2020 with the objective of performing a genomic characterization of five Cronobacter spp. and six Salmonella strains isolated from PIF sold in Chile. These PIF had been and manufactured in Chile and Mexico.

WGS enables the molecular typing of bacterial strains on a routine basis for use in epidemiology and real-time infection control. Its usefulness has also been demonstrated in identifying antimicrobial resistance markers, virulence, and the genetic prediction of antibiotic susceptibility test results (Lepuschitz et al., 2020).

In the present study, we identified and characterized C. sakazakii ST1 and ST31, C. malonaticus ST60, and S. Typhimurium ST19 strains from commercially available PIF. The three strains of the pathovar C. sakazakii ST1 were isolated from different batches of PIF which came same manufacturer. This may reflect either the widespread occurrence of this sequence type in PIF manufacturing environment or that a common ingredient was contaminated with the same strain (Sonbol et al., 2013). Cronobacter sakazakii ST1 has frequently been found in PIF commercialized in different countries, in the PIF processing environment, and in invasive clinical cases such as fatal meningitis and septicemia (Joseph and Forsythe, 2012; Sonbol et al., 2013; Fei et al., 2015; Csorba et al., 2021; Holý et al., 2021; Parra-Flores et al., 2021b). A survey of Cronobacter in the Americas showed the majority of reports isolations were from North America (57.4%, n = 465) and Brazil (42.6%, n = 465). There were a total of 75 sequence types, with the most frequently reported being the C. sakazakii pathovars ST4 (CC4) and ST1 (CC1; Costa et al., 2021).

The cgMLST scheme analysis clustered the three C. sakazakii ST1 strains closely to the C. sakazakii ST1 strains isolated from the food alert in Chile on 2017 with one to three alleles differences (Parra-Flores et al., 2018b). When analyzing the isolated strains of human infections in a European multicenter study using cgMLST, eight C. sakazakii ST1 isolates were found among all the C. sakazakii strains. Of these eight ST1 isolates, two strains were analyzed from an outbreak affecting two newborns suffering from necrotizing enterocolitis in Austria in 2009, which differed by only one allele. In addition, three ST1 isolates from Austria and one ST1 from Denmark differed by 203 alleles with the ATCC BAA-894 strain, which was isolated from PIF associated with a fatal case of an infant in the United States in 2001. Cronobacter sakazakii ST31 has also been isolated from clinical cases with fatal outcomes; however, it has been less prevalent in PIF and the environment (Sonbol et al., 2013; Fei et al., 2015; Ogrodzki and Forsythe, 2015; Lepuschitz et al., 2019). In this context, C. sakazakii ST1, ST8, and especially ST4 are the STs with the highest risk of causing disease in infants (Joseph and Forsythe, 2011; Forsythe, 2018; Lachowska et al., 2021). Cronobacter malonaticus ST60 has also been isolated in powdered milk, food, and the environment, but with less significant clinical cases than C. malonaticus ST7 and C. sakazakii ST1 (Forsythe, 2018; Costa et al., 2021).

Salmonella enterica is the most frequently identified cause of food poisoning in the European Union; serotype Typhimurium ST19 is most often associated with disease and death (Carroll et al., 2017; de Frutos et al., 2018) and commonly identified in human clinical samples, animals, food, and the environment (Panzenhagen et al., 2018; Monte et al., 2020). The cgMLST scheme analysis revealed a cluster of six S. Typhimurium ST19 with one to two allele differences. In this context, the present study is the first to identify S. Typhimurium ST19 in PIF, whereas recent reports of Salmonella outbreaks in PIF have involved S. Agona (Brouard et al., 2007; Jourdan-da Silva et al., 2018). We also presumptively identified two Salmonella monophasic variant strains. Monophasic Salmonella has been identified in different human cases in the United States, Spain, Brazil, and Thailand and characterization of these strains revealed resistance to multiple antibiotics (Mossong et al., 2007).

All five C. sakazakii strains analyzed in the present study were resistant to cephalothin and only two strains to ampicillin. The resistance of C. sakazakii to cephalothin and ampicillin has been reported in several studies, also suggesting an almost intrinsic resistance of the Cronobacter spp. genus to cephalothin (Kim et al., 2008; Molloy et al., 2009; Flores et al., 2011; Chon et al., 2012). Parra-Flores et al. (2020) found that 100% of the C. sakazakii strains isolated in powdered infant formula distributed in Latin America were resistant to cefotaxime and ampicillin, 60% to cefepime, 40% to amikacin, and 20% to cephalothin. Furthermore, one of the strains was resistant to six of the 12 evaluated antibiotics and another strain was resistant to five antibiotics. Multiple drug resistance (MDR) is a cause for concern; in a case of neonatal meningitis caused by C. sakazakii in China, the isolated strains were resistant to eight antibiotics (Zeng et al., 2018).

For the antibiotic resistance genes, all C. sakazakii strains and the C. malonaticus strain exhibited the same efflux, antibiotic inactivation, and antibiotic target alteration genes that confer antibiotic resistance to β-lactams, fluoroquinolones, aminoglycosides, imidazoles, and disinfectants such as triclosan. The marA gene, whose transcription function regulates multidrug efflux, modulates membrane permeability and activates the transcription of the AcrAB-TolC efflux pump that plays an important role in antibiotic resistance (Wang X. et al., 2021). Several authors have detected the msbA, emrR, H-NS, emrB, marA, CRP, PBP3, H-NS, and msrB genes that are associated with antibiotic resistance by using efflux pumps, regulatory systems, and antibiotic target protection genes (Aly et al., 2019; Lepuschitz et al., 2019; Parra-Flores et al., 2021b). All strains exhibited the glpT gene that encodes fosfomycin resistance. This is relevant because fosfomycin is considered effective in patients with MDR bacterial infections (Falagas et al., 2019). We also identified the mcr-9.1 gene that confers resistance to colistin and the bla_CSA and bla_CMA genes that confer resistance to cephalothin. The mcr-9.1 gene can generate resistance to colistin in several enteropathogens and can silently circulate undetected unless induced by colistin (Carroll et al., 2019; Kieffer et al., 2019). The presence of mobile genes resistant to colistin (mcr) is a worldwide public concern because colistin is considered as a last resort to treat infections caused by multidrug-resistant Enterobacteriaceae (Borowiak et al., 2020). The bla_CSA and bla_CMA genes were first described in 2014 and are associated with β-lactamase class C resistance; they are not inducible and are regarded as cephalosporinases (Müller et al., 2014). Jang et al. (2020) found variants of class C bla resistance genes identified as CSA-2 or CSA-1 and CMA in all the analyzed strains. Holý et al. (2021) encountered bla_CSA genes in all the C. sakazakii strains isolated in powdered milk produced in the Czech Republic between 2010 and 2014.

In the present study, all six S. Typhimurium ST19 strains were resistant to oxacillin, five to ampicillin, four to cephalothin, and one to gentamicin. S. Typhimurium ST19 has shown extensive resistance to a variety of critically important antimicrobials (Monte et al., 2020). Jain et al. (2018) reported that all the evaluated S. Typhimurium ST19 strains were resistant to 7 of the 15 tested antibiotics and encountered only strains that were susceptible to ampicillin and gentamicin; this contrasts with our study in which strains were resistant to these two antibiotics.

The Salmonella strains carried many genes such as aac(6′)-Iaa that encodes resistance to aminoglycosides (e.g., gentamicin), which have been found in multidrug-resistant S. Typhimurium strains and caused a foodborne outbreak at a banquet in China in 2017 (Xiang et al., 2020). Wei et al. (2019) reported the presence of β-lactamase ampC genes the same as in our study; in addition to finding β-lactamase ampC-1, we found that it co-harbored with β-lactamase ampH. The β-lactamase ampC is chromosomally encoded in several gram-negative bacteria, including Enterobacter spp., Citrobacter freundii, or Serratia marcescens. Furthermore, plasmid-encoded ampC genes can be horizontally transferred to other Enterobacteriaceae without the presence of chromosomally encoded ampC, such as Klebsiella and Salmonella, resulting in a highly effective and dynamic dissemination mechanism (Ingram et al., 2011). We also found the uhpT and glpT genes that encode fosfomycin resistance. The UhpT and GlpT transporters facilitate fosfomycin incorporation into bacterial cells (Silver, 2017). The reduced expression or introduction of glpT or uhpT mutations and the efflux pump can decrease fosfomycin uptake and thus lower antibiotic susceptibility (Garallah and Al-Jubori, 2020). Several studies have now shown that the overuse of antibiotics in the food chain and the presence of several antibiotic resistance operons promote multiresistance to these drugs (Ferri et al., 2017).

Among the 30 virulence genes detected in C. sakazakii, we identified C. sakazakii ompA and ompX, which encode proteins for basolateral adherence in cell lines and a possible involvement in the blood–brain barrier penetration (Mange et al., 2006; Kim et al., 2010). The Cpa protein is related to C. sakazakii serum resistance and invasion. Recent studies have suggested that the cpa locus could be considered specific for C. sakazakii and C. universalis but is absent in C. malonaticus (Franco et al., 2011). Jang et al. (2020) noted that highly virulent C. sakazakii ST8 clinical strains that carry the pESA3 plasmid do not possess the cpa gene; this indicates that the disease can be associated with other virulence factors. Hemolysins (Hly) are outer membrane proteins occurring in various pathogens belonging to the Enterobacteriaceae family that have hemolytic ability, such as E. coli, Klebsiella, and some gram-positive pathogens (Mare et al., 2020; Mazzantini et al., 2020). Other important genes of the C. sakazakii strains isolated from PIF are nanAKT that encode for exogenous sialic acid utilization. Sialic acid is naturally present in breast milk and supplemented in PIF because of its association with brain development (Forsythe, 2018). Sialic acid also regulates the expression of enzymes, such as sialidase and adhesins, or inhibits transcription factors of the fimB gene that mediates epithelial cell adherence and invasion. Escherichia coli K1 use this compound to modify their cell surface (Severi et al., 2007; Sohanpal et al., 2007).

In our study, we found the fic toxin encoding gene and the relB gene that encodes for the relE antitoxin. The fic gene encodes for the toxin component of the toxin-antitoxin bicistronic operon. The toxin-antitoxin (TA) systems are small genetic elements found in plasmids, phage genomes, and in the chromosomes of different bacterial species. The TA genes play a fundamental role in the physiology of bacterial stress response, such as in stabilizing horizontally acquired mobile genetic elements and participating in a persistence phenotype in some species, including E. coli and Salmonella (Deter et al., 2017; Walling and Butler, 2019). Finkelstein et al. (2019) noted in preliminary studies with C. sakazakii isolates that 2 typical toxin genes, fic and hipA, followed the evolutionary lines of the species and that C. sakazakii ST1 strains were the only strains containing the 22 TA homologs.

In the Salmonella strains, 110 of the 121 detected virulence genes were similar in all the strains, including the invA, ssaAR ssrAB, sipAC, sopBE, spvBC, and rck genes. Genes such as invA, sipA, sopB, and sopE are associated with adherence in epithelial cells and phagocyte invasion. Salmonella pathogenicity islands (SPI) are virulence gene clusters acquired by horizontal transfer that promote virulence in Salmonella spp. The SPI-2 gene encodes genes such as ssaR and ssrA that promote survival and replication within the macrophages and their dissemination in the host (Barilli et al., 2018). The spvC gene enables the survival and rapid growth of Salmonella in the host (Hu et al., 2021), while the spvB gene is known as an intracellular toxin that encodes an enzyme that ADP-ribosylates actin and destabilizes the cytoskeleton of eukaryotic cells (Lesnick et al., 2001).

The rck gene is defined as an invasin that generates a colonization niche or as a cyclomodulin with genotoxic activity (Mambu et al., 2020). To date, 24 SPIs have been identified in Salmonella of which SPI-1 and SPI-2 are the most important for virulence traits. SPI-1 is involved in the entire Salmonella infection process, including invasion, macrophage proliferation, and host responses. Both SPI-1 and SPI-2 encode different secretion systems, such as T3SS, that transport effector proteins to the host cells (Cheng et al., 2019; Lou et al., 2019).

Only the C. sakazakii ST1 strains carried the COL(pHHAD28) plasmids, which are associated with antibiotic resistance genes; these have previously been detected in C. sakazakii strains isolated from dairy products in Chile (Ramsamy et al., 2020; Khezri et al., 2021; Parra-Flores et al., 2021b). The C. malonaticus strain harbored a plasmid homologous to pESA3 called IncFIB (pCTU1), which encodes iron acquisition virulence genes necessary for pathogen survival, but not an external protease known as plasminogen activator (cpa) that enhances pathogen propagation in the host (Franco et al., 2011). Meanwhile, all the Salmonella strains showed IncFII(S) plasmid incompatibility, which is associated with several antibiotic resistance genes, such as aac(6′)-Iaa and mcr-9, and β-lactamase, such as ampC (Cha et al., 2020; Richter et al., 2020).

Bacterial CRISPR-Cas systems are considered as mechanisms of acquired immunity because they provide them with the ability to avoid bacteriophage infection and the acquisition of mobile genetic material from plasmids. This immunity is due to the information stored in the matrices of these systems, specifically in the spacer sequences. It was possible to determine that repeated sequences and cas genes in Cronobacter spp. were associated with two types of systems previously identified as I-E and I-F (Ogrodzki and Forsythe, 2016; Zeng et al., 2017), including cases in which the same strain showed both systems. The opposite was observed in Salmonella spp. genomes that were characterized only by the presence of type I-E (Louwen et al., 2014). Although both genera carry systems of the same type, there are differences between the associated cas gene operons because the Salmonella spp. systems show a larger number of genes. However, these systems are characterized by the presence of the cas1 and cas2 genes, which are involved in integrating and processing the information that is integrated into functional CRISPR arrays. Salmonella enterica and E. coli are closely related and harbor the IE-type CRISPR-Cas system in a similar manner (Makarova et al., 2015). The array sizes of the Cronobacter spp. genomes are smaller. However, when showing more than two arrays, they tend to have a larger number of spacer, suggesting that they have been exposed to a larger number of invasive elements than S. Typhimurium. Although it was not possible to determine that all the spacers of the systems are similar to bacteriophage sequences, there are cases in which they are specific to these genera. Of particular interest, some of the S. Typhimurium spacer were associated with CrAssphages and other phages associated with the human intestinal microbiota (Guerin et al., 2018). It is known that Salmonella is recognized as an intestinal pathogen; therefore, this is an ecosystem where bacteria can acquire external genetic material and integrate it into their genome by various horizontal transfer mechanisms and by other means such as CRISPR-Cas systems (Louwen et al., 2014). The cas1 and cas2 genes are indispensable for integrating and processing the information acquired by the bacteria; in their absence, the system loses the ability to acquire information (Parra-Flores et al., 2021b). The search for prophages in the studied genomes showed that they may have few intact phages. Cronobacter spp. tend to show a larger number of intact and incomplete prophages, which are not only characteristic of this genus but also of Salmonella. Therefore, our data suggest that these genomes carry the necessary information to prevent these bacteriophages from infecting both Cronobacter spp. and Salmonella strains. In addition, this information can be useful in the future when using gene therapy as a therapeutic option for infections caused by these pathogens that are difficult to treat (Gordillo Altamirano and Barr, 2019).

In the present study, WGS allowed us to determine multiple virulence and antibiotic resistance genes in bacterial pathogens isolated from PIF intended for consumption by infants aged less than 6 months and distributed throughout Latin America. In our opinion, the identification of C. sakazakii and S. Typhimurium in PIFs not only violates current health regulations but also endanger the health of infants consuming these products. It is therefore necessary that health authorities conduct more preventive control activities related to these products and carry out campaigns that emphasize the use of rehydration water at 70°C for infant formula. This recommendation by the World Health Organization highlights that the 70°C temperature has a proven effect in significantly decreasing the risk of disease by pathogens in reconstituted PIF (FAO/WHO, 2004, 2006). Control of Salmonella and Cronobacter during the production of milk powder and PIF is through microbiological sampling according to Codex Alimentarius Commission (2007) of finished product, as well as ingredients and intermediate products (International Standards for Organization (ISO), 2017; Podolak and Black, 2017). The use of zoning production facilities to focus environmental sampling according to risk is used (Cordier, 2008; FDA, 2019).

C. sakazakii, C. malonaticus, and S. Typhimurium strains isolated from PIF exhibit antibiotic resistance and various virulence genes and resistance to β-lactam antibiotics. Continuous monitoring of these powdered infant formulas is necessary due to the risk associated with pathogen contamination of the product and consumption by the immunologically vulnerable child population.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found in the article/Supplementary Material.

JP-F, OH, SA, SL, AC-C, JM-R, AC, and SF conceived the experiments and prepared the manuscript. JP-F, AC-F, PC-S, AC-C, AP, SL, and WR conducted the laboratory work. JP-F, OH, SA, AC-C, JX-C, JM-R, AC, WR, and SF drafted the manuscript. All authors contributed to the article and approved the submitted version.

This research was funded by the Research Directorate of the Universidad del Bío-Bío, Projects 191520 4/R and GI 195420/EF.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.