S. Typhi strains were isolated from the blood and stool samples of six patients residing in four different households in Mukuru, an informal settlement in Nairobi, Kenya, known for being typhoid-endemic. A typhoid fever patient, along with at least one household contact, provided stool samples for the detection of S. Typhi shedding ( Supplementary Table S1 ). The laboratory methods for isolation and identification of the isolates are described in our previous publication (Muturi et al., 2024). Briefly, blood samples were inoculated into BACTEC™ culture vials and incubated in a BACTEC blood culture system (Becton, Dickinson and Company, New Jersey, USA) before subculturing on XLD agar. Stool samples were enriched in selenite fecal (SF) broth prior to culturing on XLD agar. S. Typhi culture-positive samples were identified using biochemical tests with the Analytical Profile Index 20E (API 20E) system and confirmed by polymerase chain reaction (PCR) amplification of a 1,278-bp fragment of the VI region of the flagellin gene, using primers tviB-F (5′-TCAGCGACTTCTGTTCTATTCAAGTAAGAAAGGGGTACGG-3′) and tviB-R (5′-GCTCCTCACTGACGGACGTGCGAACGTCGTCTAGATTATG-3′). A total of 22 isolates were collected, comprising 2 from blood samples and 20 from stool samples. Nineteen of the isolates (19/22) came from four patients diagnosed with cholelithiasis. Of these patients, three exhibited typhoid symptoms and continued shedding S. Typhi post-treatment. Gallstones were confirmed via ultrasound, a standard imaging procedure for assessing gallbladder diseases. One patient with gallstones was asymptomatic but continued to shed S. Typhi and lived in the same household (household B) as an acute typhoid fever case who did not have gallstones. In household A, one contact without cholelithiasis shed S. Typhi once, whereas the index case, who had gallstones, continued shedding S. Typhi after antibiotic treatment. The index case refers to the first identified individual infected with S. Typhi and presenting typhoid fever symptoms in each household.

DNA extracted from S. Typhi strains using GenElute™ Bacterial Genomic DNA Kit (Sigma-Aldrich, Missouri, United States) was prepared for whole-genome sequencing by SeqCoast Genomics (Portsmouth, New Hampshire, United States) using an Illumina DNA Prep Tagmentation kit and unique dual indexes. Sequencing was performed on the Illumina NextSeq 2000 platform using a 300-cycle flow cell kit to produce 2 × 150-bp paired reads as previously described (Grant et al., 2023). PhiX control, 1%–2%, was spiked into the run to support optimal base calling. Read demultiplexing, read trimming, and run analytics were performed using DRAGEN v3.10.12, an on-board analysis software on the NextSeq 2000.

Quality-trimming of the reads was done using Trimmomatic (version 0.39) (Bolger et al., 2014). Read error was corrected using SPAdes (v. v3.13.1) (Bankevich et al., 2012). The reads were assembled into contigs using SPAdes (wrapped in Unicycler). Read mapping was done using Bowtie2 and SAMtools (wrapped in Unicycler, version 0.4.4) (Langmead and Salzberg, 2012; Li et al., 2009; Wick et al., 2017). Pilon (v1.24) (wrapped in Unicycler) was used in polishing of each assembly (Walker et al., 2014). Gene prediction and functional annotation were performed using BAKTA (version 1.5.1) (Schwengers et al., 2021). The annotation pipeline was as follows: prediction of protein-coding genes using Prodigal, tRNA identification using tRNAscan-SE (Chan and Lowe, 2019), tRNA and tmRNA identification using Aragorn (Laslett and Canback, 2004), prediction of rRNA sequences using Infernal and the Rfam database (Nawrocki and Eddy, 2013), CRISPR prediction using PILER-CR (Edgar, 2007), antimicrobial resistance gene identification using AMRFinderPlus (Feldgarden et al., 2021), prediction of signal peptides using DeepSig (Savojardo et al., 2018), prediction of transposases using ISFinder (Siguier et al., 2006), and computation of codon usage biases for each amino acid using the codonUsage.py script (https://github.com/Arkadiy-Garber/BagOfTricks) (Garber, 2024).

Identification and clustering of S. Typhi genotypes was performed using Pathogenwatch (https://pathogen.watch/), a web-based platform with several different components (Argimón et al., 2021; Wong et al., 2016). The platform provides compatibility with S. Typhi typing information for MLST (Sharma et al., 2016), in silico serotyping (SISTR) (Yoshida et al., 2016), and an SNP genotyping scheme (GenoTyphi) (Wong et al., 2016). S. Typhi assemblies (in fasta format) were uploaded to the platform (https://pathogen.watch/upload/fasta) for the analysis. A collection of uploaded sequences was created in Pathogenwatch to generate a phylogenetic tree. S. Typhi genotype 2.1 (GenBank Accession: SAMEA2158302) was used as the outgroup. The phylogenetic tree generated by Pathogenwatch (Newick format) was downloaded and exported to Microreact (https://microreact.org) for visualization. Associated metadata in CSV format was also exported to Microreact, and a phylogenetic tree in Newick format, reflecting branch lengths corresponding to SNP/genetic distances, was obtained.

Antimicrobial susceptibility testing was performed using the disk diffusion technique (Reller et al., 2009) for all antimicrobials commonly used in Kenya for typhoid fever treatment including ampicillin (10 µg), tetracycline (30 µg), co-trimoxazole (25 µg), chloramphenicol (30 µg), amoxicillin–clavulanate (20/10 μg), cefpodoxime (30 µg), ceftazidime (30 µg), ceftriaxone (30 µg), cefotaxime (30 µg), azithromycin (15 μg), ciprofloxacin (5 µg), nalidixic acid (10 µg), kanamycin (30 μg), and gentamicin (10 μg). The diameter of the zone of inhibition was measured after 18 h–24 h, and results were interpreted according the Clinical and Laboratory Standards Institute (CLSI), guidelines for Salmonella (Clinical and Laboratory StandardsInstitute, 2023).

NCBI Antimicrobial Resistance Gene Finder Plus (AMRFinderPlus, v3.12.8) (https://github.com/ncbi/amr/wiki) was used to identify acquired antimicrobial resistance genes and known resistance-associated point mutations in S. Typhi-assembled nucleotide sequences (Feldgarden et al., 2021).

To identify genetic variations in the longitudinal clinical isolates relative to acute isolates (S. Typhi isolated during typhoid fever diagnoses), S. Typhi strains isolated before each typhoid index case was treated with antibiotics were used as the reference genome and were compared with those isolated after treatment (follow-up isolates), and/or those isolated from household contacts using breseq v0.38.2 (https://github.com/barricklab/breseq), a computational pipeline for finding mutations relative to a reference sequence in short-read DNA (Deatherage and Barrick, 2014). The variant calling pipeline was as follows: quality-filtering of raw reads using Trimmomatic v0.39 (Bolger et al., 2014), mapping of reads against a reference genome (first strain isolated from the index case), analysis of possible mutations based on mapping data, identification of mutations, and tabular summaries of mutation profile across samples (Langmead and Salzberg, 2012).

To identify plasmids carrying acquired AMR genes in the isolated S. Typhi strains, a plasmid detection tool PLASMe (v1.1) was used (https://github.com/HubertTang/PLASMe). The tool uses the alignment component in PLASMe to identify closely related plasmids whereas diverged plasmids are predicted using order-specific Transformer models (Tang et al., 2023). Assembled S. Typhi genomes were used for this analysis.

Because of the importance of biofilm formation on gallstones in chronic carriage (Crawford et al., 2010; Gunn et al., 2014), the biofilm-forming ability of all the 22 human S. Typhi isolates was tested under gallbladder simulating conditions as previously described (González et al., 2018). Briefly, S. Typhi biofilms were grown on non-treated polystyrene 96-well plates (Corning, Kennebunkport, ME). To simulate growth conditions on gallstones, wells in two plates were precoated with cholesterol by adding a solution of 5 mg/mL in 1:1 isopropanol:ethanol and air-dried overnight (Crawford et al., 2008). A pure colony of S. Typhi on an XLD agar plate was cultured in Tryptone Soy Broth (TSB). Overnight (O/N) cultures in broth were normalized to OD₆₀₀ = 0.8 and diluted 1:2,500 in TSB or TSB containing 2.5% human bile, and 100 µL/well was dispensed into the plates. The plates were incubated at 25°C in a Fisherbrand™ nutating mixer (Thermo Fisher Scientific; Hampton, NH) at 24 rpm for 96 h. Media (TSB or TSB containing bile) were changed after every 24 h for consistent S. Typhi biofilm growth. Plates were emptied on the fourth day and washed twice before heat fixing at 60°C for 1 h. The biofilms were stained using a crystal violet solution and acetic acid (33%) used to elute crystal violet before reading the OD₅₇₀. GraphPad Prism 9.5 was used to analyze the biofilm formation results. One-way analysis of variance (ANOVA) was used to test the level of significance in biofilm formation between the different S. Typhi sub-lineages and in different conditions, i.e., biofilms in absence of cholesterol and bile, in cholesterol-coated plates in absence of bile, and in presence of cholesterol and bile. Student’s t-test was used to test the level of significance in biofilm formation in strains isolated before treatment vs. last strains shed by the patient, P-values less than 0.05 (P<0.05) were considered significant.

Given the highly structured nature of the S. Typhi population, with numerous subclades associated with specific geographical regions and antimicrobial resistance patterns, the genotypes responsible for asymptomatic carriage in the current study were identified using a combination of three methods employed by Pathogenwatch for S. Typhi typing: multi-locus sequence typing (MLST), in silico serotyping (SISTR), and the SNP genotyping scheme (GenoTyphi). Among the 22 bacterial isolates, which include two from blood samples and 20 from stool samples collected at various time points from typhoid fever patients and asymptomatic carriers of S. Typhi ( Table 1 ), all were identified as genotype 4.3.1 (S. Typhi Haplotype 58 [H58]). Of these, 11 isolates from households A and B were classified under lineage 4.3.1.1, whereas the remaining 11 isolates from households C and D fell under lineage 4.3.1.2. ( Figure 1 ). Lineage 4.3.1.2 was further divided into two sub-lineages: 4.3.1.2.EA2 (four isolates from the index case in household C) and 4.3.1.2.EA3 (seven isolates from the index case in household D). Each patient shed S. Typhi belonging to only one lineage or sub-lineage. From household A, isolates (i)–(v) were from the index case whereas isolate (vi) was from a household contact. Isolates (i) and (ii), household B, were from the index case whereas (iii), (iv), and (v) were from an asymptomatic household contact living with the index case. All household C isolates belonged to sub-lineage 4.3.1.2.EA2 whereas all household D isolates belonged to sub-lineage 4.3.1.2.EA3. Longitudinal S. Typhi isolates were collected at different time points as shown in Table 1 .

Different antimicrobial resistance patterns were observed in the isolated S. Typhi strains. The seven isolates belonging to sub-lineage 4.3.1.2.EA3 (from household D) were multidrug resistant, all expressing the following acquired antimicrobial resistance genes: sul1, dfrA7, catA1, aph(6)-Id, aph(3”)-Ib, sul2, and bla_TEM-1 , and a point mutation in the Quinolone Resistance Determining Region (QRDR) of gyrA (gyrA S83Y). Phenotypic susceptibility data showed that these seven isolates were resistant to ampicillin, chloramphenicol, trimethoprim–sulfamethoxazole, and nalidixic acid but non-susceptible to ciprofloxacin. The four sub-lineage 4.3.1.2.EA2 isolates (S. Typhi strains from Household C) had a gyrB S464F mutation in the QRDR, and all were non-susceptible to nalidixic acid and ciprofloxacin according to phenotypic susceptibility results. The third group, lineage 4.3.1.1, had six strains (household A isolates) with a gyrB S464F mutation, also demonstrating non-susceptibility to nalidixic acid and ciprofloxacin. The other five lineage 4.3.1.1 strains (household B isolates) had gyrA S83F mutations in the QRDR and showed resistance to nalidixic acid and non-susceptibility to ciprofloxacin ( Table 2 ). S. Typhi isolates carrying genes that confer resistance to ampicillin, sulfamethoxazole–trimethoprim, and chloramphenicol were classified as multidrug resistant (MDR), whereas those lacking these genes were classified as non-MDR.

Comparative genome sequencing of the S. Typhi isolates revealed several key mutations across different sub-lineages. In the sub-lineage 4.3.1.2.EA3 isolates from household D, a missense mutation in the treB gene (treB A383T) was identified in three out of seven isolates. Additionally, a second missense mutation in the tviE gene (tviE P263S), which codes for the Vi polysaccharide biosynthesis protein TviE, was observed in four out of seven isolates. Notably, two of these isolates carried both mutations. A nonsense mutation (E237*) was also found in 1/7 of the isolates, affecting a locus coding for an integrase/transposase family protein similar to DDE-type integrase/transposase/recombinase in S. Typhi CT18 (GenBank: MEM6073260.1). The first follow-up isolate from household C (sub-lineage 4.3.1.2.EA2, isolate Cii) exhibited a total of 16 mutations. This included seven silent mutations, eight missense mutations, and a deletion in the mutL gene, which encodes the DNA mismatch repair endonuclease MutL ( Table 3 and Supplementary Table S2 and Supplementary File 3 ). The second and third follow-up isolates from the same patient (Ciii and Civ) showed two silent mutations, one in the yccC gene, coding for a putative membrane protein, and another in the tehA gene, coding for a dicarboxylate transporter/tellurite-resistance protein (tehA A124A and yccC R199R mutations respectively). These isolates also had missense mutations in the amiA gene, which encodes N-acetylmuramoyl-L-alanine amidase (amiA V145A), and an additional mutation in the hpaX gene (hpaX K124E), encoding 4-hydroxyphenylacetate permease. Notably, the second follow-up isolate (Ciii) had an additional missense mutation (D78G) in a locus coding for phage baseplate assembly protein V (GenBank: MEM6147547.1). In the follow-up isolates from households A and B, which belong to sub-lineage 4.3.1.1, several mutations were detected. The first follow-up S. Typhi isolate from the index case in household A showed no mutations. However, the second follow-up sample (Aiii) had a single nucleotide polymorphism in the crl gene (crl L38P), which encodes a sigma factor-binding protein. The third and fourth follow-up isolates from the same patient (Aiv and Av) each had a silent mutation in the tnpA gene (tnpA Y41Y), which codes for IS200/IS605 family transposase. Additionally, an isolate from a household contact in the same household (Avi) had a missense mutation in the treB gene (treB A383T). From household B, S. Typhi isolated from the stool of the index case before treatment (isolate Bii) had a missense mutation in the waaK gene (waaK P167L), which encodes lipopolysaccharide N-acetylglucosaminyltransferase. This mutation was not present in the strain isolated from the patient’s blood sample collected on the same day.

S. Typhi genotype 4.3.1 lineages/sub-lineages identified in this study were found to contain plasmids. Genes conferring multidrug resistance in S. Typhi were detected in two different plasmids in sub-group 4.3.1.2.EA3 (household D) strains. The antimicrobial resistance genes sul1, dfrA, catA1, aph(6)-Id, aph(3”)-Ib, and sul2 were detected in a plasmid with sequences corresponding to S. Typhi strain 311189_252186 plasmid pHCM1, whereas bla_TEM-1 was detected in a plasmid with sequences corresponding to E. coli NES6 plasmid pN-ES-6-1 ( Table 4 ). The two plasmids were not detected in non-MDR S. Typhi Isolates.

We wanted to determine if biofilm-forming ability is correlated with the stage of typhoid fever/carriage, antimicrobial resistance, or the presence of plasmids. Biofilms were examined in gallbladder-simulating conditions. There was varying ability to form biofilms under in vitro conditions in the S. Typhi strains tested. All isolates formed weak biofilms in absence of both cholesterol and bile (OD₅₇₀ below 0.3) and significantly strong biofilms in the presence of cholesterol and 2.5% human bile ( Figure 2A ). Differences in biofilm forming ability were observed across the identified genotype 4.3.1 lineages/sub-lineages. The sub-lineage 4.3.1.2.EA2 formed the strongest biofilms (OD₅₇₀ slightly above 2.0), whereas sub-lineage 4.3.1.2.EA3 formed relatively weak biofilms (OD₅₇₀ below 0.5) even in the presence of cholesterol and human bile ( Figure 2A ). However, there was no statistical significance in biofilm forming ability in strains isolated during the symptomatic vs. asymptomatic stage in each of the three sub-groups of S. Typhi ( Figure 2C ). The weak biofilm-forming isolates from sub-lineage 4.3.1.2.EA3 had acquired antimicrobial resistance (AMR) genes encoded on plasmids. Sub-lineages 4.3.1.2.EA2 and 4.3.1.1.EA1 did not have identifiable acquired AMR genes, but isolates in these subgroups formed significantly stronger biofilms in the presence of both cholesterol and human bile ( Figure 2B ). Strains from symptomatic patients and asymptomatic carriers exhibited no statistically significant difference in biofilm-forming ability across the three S. Typhi subgroups ( Figure 2C ).