Genomic insights into antibiotic-resistant non-typhoidal Salmonella isolates from outpatients in Minhang District in Shanghai

Categories: Disease & Virus

June 13, 2025

Dataset

A total of 458 Salmonella isolates were isolated from outpatients in the Minhang District of Shanghai, China, from 2012 to 2021 (Supplementary Fig. 1a). The emergence of these isolates was mostly concentrated in June to October (Supplementary Fig. 1b). The isolates were mainly from young people aged 21–40 years (50.44%) (Supplementary Fig. 1c), and the ratio of females (54.80%) to males (45.20%) was 1.2:1 (Supplementary Fig. 1d). A total of 40 Salmonella serovars were identified using both traditional serology and WGS prediction. S. Enteritidis was predominant (39.08%) (Supplementary Fig. 1e), accounting for more than 30% each year on average (Supplementary Fig. 1f) and most from ages 21 to 40 years (Supplementary Fig 2), and was followed by S. Typhimurium (15.07%, 69/458), S. I 1,4,[5],12:i:- (8.08%, 37/458), S. London (4.37%, 20/458), S. Thompson (3.71%, 17/458) and S. Infantis (3.49%, 16/458) (Supplementary Fig. 1e).

A total of 53 STs were identified. ST11 (38.65%) and ST19 (12.88%) account for 50%, followed by ST34 (9.83%, 45/458), ST155 (4.37%, 20/458), and ST26 (3.71%, 17/458) (Supplementary Fig. 1g). Similar temporal changes were observed between dominant serovars and STs (Supplementary Fig. 1f, h). Furthermore, five STs were detected in 2012, and at least two previously undefined STs were detected each year thereafter (Supplementary Table 1). The vast majority (98.88%) of S. Enteritidis belong to ST11, 85.51% of S. Typhimurium belong to ST19, and all S. I 1,4,[5],12:i:- belong to ST34. The MST based on cgMLST displayed the temporal distribution of different STs (Supplementary Fig. 3).

AMR characteristics of Salmonella in Minhang District in Shanghai, China

AST results indicated that streptomycin (62.66%) showed the highest AMR rate, followed by nalidixic acid (55.24%), ampicillin (53.93%), ampicillin/sulbactam (42.58%), tetracycline (34.28%), polymyxin E (28.38%), sulfamethoxazole (17.25%), imipenem (14.19%), chloramphenicol (13.54%), and cefazolin (12.88%) (Supplementary Table 2). Interestingly, 458 isolates were 100% sensitive to cefotaxime/avibactam (Supplementary Table 2). Among the classes of antibiotics, aminoglycoside had the highest resistance level (64.19%), followed by beta-lactam (61.35%), fluoroquinolone (59.17%), tetracycline (34.28%), polymyxin E (28.38%), sulfonamide (17.25%), chloramphenicol (13.54%) and macrolide (4.15%) (Fig. 1a). Differences in the distribution of AMR rates of different classes of antibiotic in different serovars was observed (Fig. 1b). Compared to other serovars, S. Enteritidis showed higher AMR rate to nalidixic acid, S. I 1,4,[5],12:i:- showed higher AMR rate to cefepime and S. Thompson showed higher AMR rate to cefotaxime, ceftazidime, norfloxacin, and levofloxacin (Fig. 1c).

**Fig. 1: Phenotypic distribution of different antibiotics in clinical testing.**

From the perspective of the time dimension, the resistance rate of fluoroquinolones did not show an obvious increase trend, while beta-lactams showed an obvious increase trend, and after 2014, the resistance rates of both were above 40% (Fig. 1d, e). Since 2018, the MICs of beta-lactam agents meropenem, imipenem, and ceftazidime clavulanic acid showed a downward trend, while the MICs of fluoroquinolone agents norfloxacin, ciprofloxacin, and levofloxacin did not change significantly from year to year (Supplementary Fig. 4a, b). An obvious fact was observed that ciprofloxacin, norfloxacin, levofloxacin, cefotaxime/clavulanic acid, and ceftazidime/clavulanic acid had significantly higher MICs against S. Thompson than other serovars (Fig. 1d, e). The MICs of cefepime to S. I 1,4,[5],12:i:- isolates also exhibited higher levels than other serovars (Fig. 1e). On the other hand, the MICs of polymyxin E showed a downward trend after 2017 and was significantly higher in S. Enteritidis compared to other serovars. The MICs of azithromycin showed a downward trend after 2018 and were significantly higher in S. London and S. Thompson compared to other serovars (Supplementary Fig. 5). It’s worth noting that a total of 259 (56.55%) isolates showed MDR patterns based on the results of AST, with the highest showed resistance to 7 classes of antibiotic agents (Supplementary Table 3). Among MDR isolates, the most common (18.92%) MDR pattern was resistant to four drugs: aminoglycoside, beta-lactam, fluoroquinolone, and polymyxin E. Among the 6 dominant serovars and STs, S. I 1,4,[5],12:i:- and ST34 had the highest MDR rate, at 91.89% and 93.33%, respectively.

The draft genomes of 458 S. enterica were examined to elucidate the AMR profiles. The results were not exactly the same as AST (Supplementary Table 4). Aminoglycosides and fluoroquinolones had the highest resistance rates, at 100% and 86.46% (Supplementary Fig. 6a), respectively, and both predicted results were higher than AST. This may be due to the presence of unexpressed antibiotic resistance genes (ARGs) in the results of WGS analysis. For beta-lactams (51.75%), tetracyclines (33.62%), and macrolides (3.71%), the predicted results of these three showed similar to AST. In addition, the resistance rate of sulfonamides displayed an increasing trend, and was above 40% after 2014 (Supplementary Fig. 6b). The AMR rates of beta-lactams, sulfonamides, and tetracyclines in S. I 1,4,[5],12:i:- displayed higher levels, while the AMR rates of rifampicin and trimethoprim in S. London displayed higher levels (Supplementary Fig. 6c).

Distribution and dynamics of ARGs in Minhang District in Shanghai, China

Among the predicted 61 ARGs, the top 20 are AAC(6′)-Iy (64.63%), bla_TEM-1 (43.45%), APH(3″)-Ib (40.61%), APH(6)-Id (40.61%), sul2 (40.61%), AAC(6′)-Iaa (34.93%), tet(A) (23.58%), floR (11.35%), sul1 (9.83%), qnrS1 (9.17%), tet(B) (8.95%), dfrA12 (8.52%), sul3 (7.42%), ANT(3″)-IIa (7.21%), cmlA1 (6.55%), aadA2 (5.90%), arr-3 (5.68%), APH(3′)-Ia (5.46%), and AAC(3)-IV (4.15%) (Supplementary Table 5). For fluoroquinolones, a total of 82.53% of isolates acquired chromosome mutations in the gyrA (D87Y, D87G, D87N, S83F, and S83Y), gyrB (E466D) and parC (T57S and S80I) (Supplementary Table 6) and 9 plasmid-mediated quinolone resistance (PMQR) genes were detected, including qnrS1 (9.17%), oqxA (2.18%), oqxB (2.18%), qnrB17 (1.75%), qepA2 (1.31%), qnrB4 (1.09%), qnrB5 (0.66%), qnrS2 (0.44%), and qnrA1 (0.22%) (Supplementary Table 5). For beta-lactams, 10 ARGs were screened, including bla_TEM-1 (43.45%), bla_CTX-M-55 (3.28%), bla_OXA-1 (2.40%), bla_CMY-59 (1.97%), bla_TEM-141 (1.75%), bla_CTX-M-14 (1.31%), bla_DHA-1 (1.09%), bla_CARB-3 (0.66%), bla_CTX-M-9 (0.66%), and bla_OXA-10 (0.66%) (Supplementary Table 5). For fosfomycin-RGs, fosA7 and fosA3 were detected in 14 and 7 isolates, respectively. For macrolides, 17 isolates were resistant to azithromycin, of which 2 had chromosome mutations in AcrB (R717L) and 15 had plasmid-mediated resistance gene mph(A) (Supplementary Tables 5 and 6). We also identified qacH gene that was associated with resistance to disinfecting agents and antiseptics in 33 isolates. Importantly, based on WGS analysis, more than 50% (58.95%) of the isolates carried at least 3 ARG types belonging to different classes of antibiotics, and exhibiting MDR (Fig. 2a). Of particular concern is the 100% MDR rate observed in S. I 1,4,[5],12:i:- and the MDR rate of S. Enteritidis also up to about 75% (Fig. 2b). Equally concerned, among 6 dominant STs, the MDR rata of ST34 also reached 100%. More importantly, we have noticed that the MDR rate is increasing year by year (Fig. 2c).

**Fig. 2: Distribution characteristics of multidrug resistance (MDR) predicted.**

We then selected dominant ARGs from each class of antibiotics for the temporal variation study. It can be surveyed that the proportion of some ARGs shows an increasing trend, such as APH(3″)-Ib, APH(6)-Id, bla_TEM-1, sul2, tet(A), and gyrA p.D87Y (Fig. 3a). The distribution of ARGs also varies among different serovars. AAC(6′)-Iaa was only present in S. I 1,4,[5],12:i:-, S. London, and S. Typhimurium, while AAC(6′)-Iy was observed in S. Enteritidis, S. Infantis, and S. Thompson, exclusivity. Compared with other serovars, bla_CTX-M-55, sul2, and tet(B) were more common in S. I 1,4,[5],12:i:-, qnrS1 was more common in S. Tompson, arr-3 and dfrA27 were more common in S. London and gyrA p.D87Y was more common in S. Enteritidis (Fig. 3b). Moreover, fosA3, bla_OXA-1, and catB3 were mainly found in S. I 1,4,[5],12:i:-, fosA7 were mainly identified in S. Derby, cmlA1 were mainly detected in S. Typhimurium, and mph(A) were mainly observed in S. Tompson and S. London (Supplementary Fig. 7).

**Fig. 3: Distribution characteristics of antibiotic resistance genes (ARGs).**

In order to explore the correlation between ARGs, we constructed the ARGs co-occurrence network, which was selected from the dominant ARGs in each class of antibiotic (Fig. 4a). For AAC(6′)-Iy, the weight values between AAC(6′)-Iy and bla_TEM-1, sul2, and APH(3″)-Ib are the highest, at 260, 242, and 236, respectively, indicating the strongest correlation. For bla_CTX-M-55, the weight values between bla_CTX-M-55 and bla_TEM-141, qnrS1, bla_TEM-1, floR, sul2, and tet(A) are the highest, at 16, 14, 12, 12, 12, and 12, respectively, indicating the strongest correlation. For mph(A), the weight values between mph(A) and sul1, arr-3, sul2, and tet(A) are the highest, at 30, 20, 20, and 20, respectively, indicating the strongest correlation. The coexistence of ARGs belonging to different antibiotic classes also explains, to some extent, the emergence of MDR.

**Fig. 4: Co-occurrence networks of ARGs, plasmids, and MGEs.**

Characteristics of MGEs and VFs

Among 458 Salmonella genomes, the plasmid replicons detection rate was 83.41% (382/458), with 34 plasmid replicons (Supplementary Table 7). The dominant plasmid replicons are IncFIB(S) (n = 224) and IncFII(S) (n = 224), followed by IncX1 (n = 128), Col(pHAD28) (n = 56), Col440I (n = 43), ColRNAI (n = 35), IncQ1 (n = 32), IncHI2A (n = 26), IncHI2 (n = 26), IncC (n = 14), IncI1-I(Alpha) (n = 11), Col156 (n = 10). Over time, IncFIB(S), IncFII(S), IncX1, and Col440I showed an increasing trend (Supplementary Fig. 8a). In terms of serovar distribution, IncFIB(S), IncFII(S), and IncX1 were mainly distributed in S. Enteritidis, followed by S. Typhimurium. Col(pHAD28) was mainly distributed in S. London. IncC was mainly distributed in S. Thompson. IncHI2, IncHI2A, and IncQ1 were mainly distributed in S. I 1,4,[5],12:i:- (Supplementary Fig. 8b).

For other MGEs, a total of 3089 MGEs were identified, belonging to 82 types, divided into 4 classes (Supplementary Table 8). The majority of MGEs are insertion sequences (ISs, n = 2007), followed by MITEEc1 (n = 925), unit transposons (Tns, n = 156), and miniature inverted repeats transposable elements (MITESen1, n = 1). The number of isolates carrying IS26, ISEcl10, ISKpn2, ISSen7, and Tn2 presented a growing trend (Supplementary Fig. 9a). There are also differences in the distribution of MGEs among serovars (Supplementary Fig. 9b). ISEch12 was principally detected in S. Infantis. ISEcl10 and ISSen7 were principally detected in S. Enteritidis and S. Typhimurium, and ISKpn2 and Tn2 were principally detected in S. Enteritidis. ISSty2 was principally detected in S. Tompson and S. London.

In order to explore the correlation between ARGs and MGEs, correlation networks were constructed (Fig. 4). We found that several ARGs are closely associated with MGEs. The weight values between IS26 and bla_OXA-1, qnrS1, fosA3, and floR are higher than other MGEs, with weight values of 9, 29, 5, and 34, respectively, indicating that IS26 is more conducive to the dissemination of the four ARGs compared to other MGEs (Fig. 4b–e). The correlation between plasmids and ARGs was also discrepant. For beta-lactam resistance genes, the weight values between bla_OXA-1 and IncHI2, bla_CTM-55 and IncFIB(S), and bla_OXA-1 and IncHI2A are higher than others, with weight values of 9, 8, and 8, respectively (Fig. 4f). For fluoroquinolone resistance genes, the weight values between qnrS1 and Col4401, Col(pHAD28) and IncC are higher than others, with weight values of 16, 12, and 11, respectively (Fig. 4g). For fosfomycin resistance genes, the weight values between fosA3 and IncHI2 and IncHI2A are higher than others, with weight values of both 4 (Fig. 4h). For amphenicol resistance genes, the weight values between floR and IncHI2A and IncHI2 are higher than others, with weight values of 20 and 19, respectively (Fig. 4i).

A total of 129 VFs were detected and each strain carried a large number of virulence genes (ranging from 94 to 117). Most of the VFs were present in all isolates (Supplementary Table 9). We randomly selected 45 types for analysis (Supplementary Table 10). Most of them belong to Salmonella Pathogenic Islands (SPIs)-encoded type III protein secretion system (T3SS) or fimbriae-related genes. No significant changes were observed in VFs (Supplementary Fig. 10a). However, differences were observed in different serovars (Supplementary Fig. 10b). The astA, cdtB, faeCDE, and gyvA genes were mainly found in S. London and the pefAD and rck genes were mainly found in S. Enteritidis and S. Typhimurium. The entA, lpfB, pipB2, and steB genes were observed to be more prevalent in people over 50 years old (Supplementary Fig. 10c).

A total of 15 SPIs were detected, one of which was “not named”. Among them, the detection rates of SPI-1, SPI-2, SPI-3, and SPI-9 were 100%. No significant changes were observed in SPIs (Supplementary Fig. 11a). The distribution of SPIs varied in serovars (Supplementary Fig. 11b). SPI-4 existed in S. Typhimurium, S. I 1,4,[5],12:i:-, and S. London. SPI-10 was only existed in S. Enteritidis. We also detected SGI-1, which existed in S. Infantis and S. London. SPI-8 was more common in people under 50 years old, and SPI-4, SPI-10, SPI-13, and SPI-14 were more common in people over 50 years old (Supplementary Fig. 11c).

Phylogenetic analysis confirms the presence of outbreaks caused by S. Typhimurium and S. Enteritidis isolates

We selected 106 S. Typhimurium (and monophasic, called S. I 1,4,[5],12:i:-) and 179 S. Enteritidis for phylogenetic analysis, respectively. S. Typhimurium had two evolutionary branches, with ST19 and ST34 being the dominant clones (Fig. 5a). ST34 carried more ARGs than ST19 and the MGEs carried by both also differ. IncFIB(S) and IncFII(S) were only detected in ST19, while IncHI2 and most of IncQ1 and IncHI2A were detected in ST34. In our early passive surveillance, we recorded 12 suspect outbreak events of S. Typhimurium. According to the previous report⁴⁶, an outbreak of foodborne disease is defined as the occurrence of two or more cases of similar diseases caused by ingestion of ordinary food. In addition, isolates that cause diseases need to meet the requirement of having 2-32 SNP differences⁴⁷. Finally, we confirmed 8 outbreaks caused by S. Typhimurium. In addition, for S. Typhimurium and S. I 1,4,[5],12:i:-, 19 isolates with rough colonies and 87 with smooth colonies were confirmed. Compared with the smooth S. Typhimurium, all rough S. Typhimurium belonged to ST34, carried IncHI2 and IncHI2A, and showed a higher ARGs coverage rate, such as bla_CTX-M-55, dfrA12, and floR. Notably, three outbreaks of rough S. Typhimurium isolates were observed, and the strains in the seventh outbreak all carried bla_CTX-M-14. It can be also observed that all rough S. Typhimurium isolates carried IncHI2 and IncHI2A, which may lead to the emergence of multiple resistance genes.

**Fig. 5: Phylogenetic analysis of S. Typhimurium and S. Enteritidis.**

For S. Enteritidis, the phylogenetic tree showed two branches, and most (98.9%, 177/179) of isolates belonged to ST11 (Fig. 5b). Eighteen outbreaks occurred from 2012 to 2020, including one caused by ST183 and 16 by ST11. Only IncFII(S) was detected in ST183, while more IncFII(S), IncFIB(S), and IncX1 were detected in ST11. No ARGs were detected in ST183 isolates. Eight isolates of ST11 carried bla_CTX-M-55. The gene dfrA17 was not observed in the outbreak strains in ST11. APH(3″)-Ib, bla_TEM-1, sul2, and tet(A) were more prevalent in ST11 isolates. In addition, the strains that emerged in the thirteenth outbreak all carried bla_CTX-M-55.