In Children’s Hospital of Zhejiang University School of Medicine from July 2019 to July 2020, all patients for whom clinical specimen were collected and found to host Salmonella were included in this study. For the routine work in the department of clinical laboratory, briefly, stool samples were inoculated onto blood agar (BIOIVD, Zhengzhou, China) and SS agar (Comagal, Shanghai, China) and cultured at 35°C for 18h. Blood samples were cultured with a Bactec FX400 System (BD, NJ, United States), and the other types of samples were incubated on blood agar (BIOIVD, Zhengzhou, China). Then, the correct clones were picked and streaked to purity for the tests. Patients’ information, including age, sex, and infection site, was obtained from the digital record system, and their names were masked. The age of children was calculated in decimal years by considering a month approximately 0.083years, and a day approximately 0.0027years. This study was approved by the Ethics Committee of Children’s Hospital of Zhejiang University School of Medicine (2021-IRB-031).

All the isolates were identified by matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) mass spectrometry systems (Bruker Daltonics, Bremen, Germany). The serogrouping was conducted by a seroagglutination test (Tianrun, Ningbo, China), and serogroups A, B, C1, C2-C3, D1, and E1 were determined. These serogroups were also identified as O:2, O:4, O:7, O:8, O:9, and O:3, 10, respectively.

Antimicrobial agent susceptibility was tested for all collected isolates, including against ampicillin, ceftriaxone, chloramphenicol, ciprofloxacin, levofloxacin, meropenem, and trimethoprim/sulfamethoxazole. Escherichia coli ATCC 25922 was used as the quality control. The agar dilution method was used following the guidelines of the Clinical & Laboratory Standards Institute (CLSI), and the breakpoint was interpreted according to the CLSI guidelines (M100 29th edition). The AST of colistin was determined by the micro broth dilution method, and a minimal inhibitory concentration (MIC)>2mg/L was considered resistant following European Committee on Antimicrobial Susceptibility Testing (AST; EUCAST) breakpoint interpretation.

Ceftriaxone-resistant Salmonella were selected for whole-genome sequencing, and genomic DNA was extracted with a QIAamp DNA Mini Kit (QIAGEN, Hilden, Germany). The Illumina HiSeq X-Ten platform (Illumina, San Diego, CA, United States) was chosen for whole-genome sequencing with a 150bp paired-end strategy, as previously described (Shi et al., 2020). Then, the reads were assembled by Shovill,¹ and unqualified contigs (coverage <10 or length<200bp) were eliminated from subsequent analysis.

Detection of the multilocus sequence typing (MLST) and the Inc-type plasmid of the strain were screened and determined by using the MLST 2.0 server and Plasmid Finder 1.3 at the Center for Genomic Epidemiology,² respectively (Larsen et al., 2012; Carattoli et al., 2014), and the molecular serovar was identified by SISTR (Yoshida et al., 2016) on Pathogenwatch.³ Monophasic Salmonella Typhimurium (1, 4, [5], 12: i: -) was classified as Salmonella Typhimurium (1, 4, [5], 12: i: 2).

For ceftriaxone-resistant isolates, antimicrobial resistance genes (ARGs) were identified with whole genome sequences (WGSs) by ABRicate with parameter identity >90% and minimum length>60%,⁴ and the National Center for Biotechnology Information (NCBI) bacterial antimicrobial resistance reference gene database was set as the reference (Feldgarden et al., 2019). The bla_TEM, bla_CTX-M-1 group, bla_CTX-M-9 group, and bla_CMY-2 among the ceftriaxone susceptible isolates were screened by polymerase chain reaction (PCR) and Sanger sequencing, and the primers used in this study are shown in Supplementary Table S1.

The contigs that contained ESBL genes were extracted and annotated by Prokka (Seemann, 2014), and particularly by ISfinder for identifying mobile genetic elements (MGEs; Siguier et al., 2006).⁵ The transposons or translocatable units that were responsible for ESBL gene transfer were ascertained and their genetic structure was visualized.

All 53 ceftriaxone-resistant isolates were imported for phylogenetic relationship analysis based on their single-nucleotide polymorphisms (SNPs). The genome alignment was established by Snippy with default parameters,⁶ and one internal strain, SM-1 (accession number: SAMN20422865), was used as a reference, and the SNPs numbers between each isolate were calculated by Snp-dists.⁷ Fasttree (version 2.1.11) and generalized time-reversible model was used in phylogeny calculation, and the tree was illustrated and annotated by Evolview (Subramanian et al., 2019).

The comparison of antimicrobial resistance rates was performed by Fisher’s exact, and p<0.05 was considered as statistically significant.

The WGS of ceftriaxone-resistant Salmonella spp. were deposited in the NCBI database under BioProject accession number PRJNA749852.

During the period of the current study, a total of nonduplicated 205 Salmonella isolates were obtained. The isolates from feces accounted for the vast majority (195/205), and the remaining 10 isolates were from several types of sterile body fluids. Six out of 10 were from blood samples in the gastroenterology department, respiratory department, neonatal department, and pediatric intensive care unit (PICU). One strain (SM-201) was isolated from pleural fluid and the other (SM-290) was isolated from abdominal pus (general surgery department). There were two isolates from subperiosteal pus (SM-103 and SM-301, orthopedics department). A total of 86.8% (178/205) of Salmonella isolates were from children less than 4years of age in this study, and the median age of these children was 1.25 (IQR: 0.79 to 2.29), suggesting that younger children were the most potentially susceptible population.

The seroagglutination test and AST were performed to explore the distribution of Salmonella serogroups and antimicrobial susceptibility in a tertiary children’s hospital (Table 1). Fifty-three ceftriaxone-resistant isolates were found in these collected Salmonella isolates and resistance rate reached 25.9%, and the other antimicrobial resistance rates for chloramphenicol, ciprofloxacin, levofloxacin, ampicillin, and trimethoprim/sulfamethoxazole were 44.4% (91/205), 23.4% (48/205), 11.7% (24/205), 75.6% (155/205), and 37.1% (76/205), respectively. The multidrug resistance (MDR) rate in the current study was 37.6% (77/205), and there was no meropenem-resistant Salmonella. Interestingly, the resistance rates for ceftriaxone, ciprofloxacin, and levofloxacin were significantly higher in younger children (age<1) (Supplementary Figure S1), and 94.3% (50/53) of ceftriaxone-resistant Salmonella isolates were distributed in children less than 4years of age. Furthermore, the resistance rates for ciprofloxacin and levofloxacin in the ceftriaxone-resistant group were significantly higher than those in the ceftriaxone-susceptible group (p=0.0006 and<0.0001, respectively). For instance, the ciprofloxacin resistance rate was 41.5% (22/53) in the ceftriaxone-resistant group, compared with 17.1% (26/152) in the ceftriaxone-susceptible group; and the difference resistance rates were 34.0% (18/53, ceftriaxone resistant) vs. 3.9% (6/152, ceftriaxone susceptible) for levofloxacin.

Serogroup determination by seroagglutination testing showed that serogroup B was the most prevalent type (137/205) among these Salmonella isolates, followed by serogroup D1 (21/205), serogroup C1 (19/205), serogroup E1 (12/205), and serogroup C2-C3 (10/205). There were six isolates that could not be classified into a specific serogroup by the seroagglutination testing. One of these isolates, SM-233, was resistant to ceftriaxone, and further whole genome sequencing showed that SM-233 belonged to serovar Kedougou, which is serogroup G Salmonella (Rodriguez-Urrego et al., 2010). Serogroup B Salmonella isolates also showed the most prevalent ceftriaxone resistance.

Whole-genome sequencing data of ceftriaxone-resistant Salmonella isolates were utilized to further clarify their molecular characteristics and phylogenetic relationship. All the ceftriaxone-resistant Salmonella isolates were from stools except one from blood. The MLST results showed that ST34 was the predominant sequence type (33/53, 62.3%), and the others were diverse, including one ST11, one ST17, two ST19, one ST40, two ST99, four ST198, one ST2529, one ST321, one ST358, one ST359, one ST469, one ST1543, and two ST684. Furthermore, one novel ST (ST8839) was found with the MLST profile (aroC 844, dnaN 71, hemD 43, hisD 12, purE 190, sucA 20, thrA 18), and it was nearest to ST808, with the point mutation C102T in aroC.

Among these isolates, serovar Typhimurium and its monophasic variant were predominant, and 37 serovar Typhimurium and its monophasic variant isolates corresponded to more than one sequence type, which included ST19, ST34 and ST99. ST358 and ST2529 belonged to serovar Goldcoast. In the other serovars, there was a one-to-one relationship between the serovar and the MLST, for instance, serovar Kentucky corresponded to ST198, and serovar Uganda corresponded to ST684.

The phylogenetic tree showed that serovars, such as Enteritidis, Muenster, Rissen, Virchow, Indiana, Kedougou, and Kottbus, were differently positioned on the phylogenetic tree and unrelated (Figure 1). The results showed that the phylogenetic relationship among the same ST isolates was indistinguishable, such as among the ST684 serovar Uganda or among the ST198 serovar Kentucky. The mean number of SNPs between each pair of ST34 isolates was 52.5±22.7, indicating that the relationships within the predominant ST strains were relatively close. There were 23 indistinguishable pairs of isolates in ST34 that has less than 5 SNPs. SM-90 (ST358) isolated from stool and SM-18 (ST2529) isolated from blood were nearly unrelated, although both of them belonged to the serovar Goldcoast.

To understand the ceftriaxone resistance mechanisms of these Salmonella spp. isolates, the antimicrobial resistance genes were analyzed. Almost all ceftriaxone resistant isolates were CTX-M-type ESBL producers, except one producer of CMY-2 β-lactamase, which is an AmpC cephalosporinase. In contrast, no CTX-M ESBLs or AmpC cephalosporinase genes were detected among the ceftriaxone-susceptible isolates, suggesting that those enzymes were possibly responsible for ceftriaxone resistance. Among the CTX-M-producing isolates, bla_CTX-M-55 (33/53) was the most prevalent, followed by bla_CTX-M-65 (12/53), bla_CTX-M-14 (4/53), bla_CTX-M-9 (2/53), bla_CTX-M-64 (1/53), and bla_CTX-M-130 (1/53). In particular, SM-55 harbored two different types of ESBLs, bla_CTX-M-14 and bla_CTX-M-55. TEM β-lactamases were mainly distributed (40/53) in ceftriaxone resistant isolates; however, the bla_TEM-1 gene was still detected in 62.5% (95/152) of ceftriaxone susceptible isolates, showing its limited effect on ceftriaxone resistance. Among the ceftriaxone-resistant isolates, 38 isolates harbored bla_TEM-1, and two isolates (SM-20 and SM-22) had bla_TEM-215, which were both ST684. Four isolates were bla_OXA-1 positive; eight isolates were bla_OXA-10 positive, and there were three isolates contained bla_LAP-2 (Figure 2).

Genetic environment analyses showed that the ISEcp1-ESBL-IS903B structure was the most prevalent ESBL gene transfer. The CTX-M-1 group included CTX-M-55 and CTX-M-64, and the former had three different environments: ISEcp1-bla_CTX-M-55-IS903B, ISEcp1-bla_CTX-M-55-bla_TEM-1-IS903B, and IS1F-bla_CTX-M-55. Similarly, the gene structure of the latter was ISKpn26-ISEcp1-bla_CTX-M-64-IS26. The CTX-M-9 group contained CTX-M-9, CTX-M-14, CTX-M-65, and CTX-M-130. The genetic structures of bla_CTX-M-14, bla_CTX-M-65, and bla_CTX-M-130 were similar, which were mediated by ISEcp1 and IS903B, but bla_CTX-M-9 was located on the chromosome, which was still mediated by ISEcp1. The gene upstream of CMY-2 was ISEcp1, which was located on the IncI1-I (Alpha) type plasmid (Figure 2).

Interestingly, SM-66 and SM-212 both harbored the mcr-1 gene, which mediates resistance against colistin, the last-resort therapy choice for multidrug-resistant Gram-negative bacterial infection. The colistin MIC of isolate SM-212 (accession number: SAMN20422907) reached 8mg/L; however, the colistin MIC of isolate SM-66 (accession number: SAMN20422894) was just 0.5mg/L, remaining susceptible. WGS analysis showed that the deletion mutation (GTGGCGAGTGTTG) of mcr-1 in SM-66 was observed between nucleic acid positions 55 and 67. The plasmid type of these two mcr-1-harboring isolates was IncHI2, which is common plasmid groups carrying multidrug resistance determinants.