Identification of Multiple Variant Extensively Drug-Resistant Typhoid Infections across Pakistan

ABSTRACT.

Typhoid fever, caused by Salmonella enterica serovar Typhi (S. Typhi), is a life-threatening bacterial infection. Recently, an outbreak of a new sublineage of extensively drug resistant (XDR) S. Typhi emerged in Pakistan in the province of Sindh. This sublineage had both a composite multidrug resistance transposon integrated on the chromosome and an acquired IncY plasmid carrying the extended spectrum beta-lactamase, blaCTX-M-15, which conferred resistance to third-generation cephalosporins. We observed previously that XDR typhoid had spread beyond the originating southern Sindh Province. Thus, we sought to determine the genetic diversity of 58 ceftriaxone-resistant S. Typhi clinical isolates by whole genome sequencing collected across Pakistan from November 2018 to December 2020 to provide insights into the molecular epidemiology of the evolving outbreak. We identify multiple novel genomic integrations of the extended spectrum beta-lactamase gene into the chromosome in S. Typhi, revealing the existence of various XDR typhoid variants circulating in the country. Notably, the integration of the IncY plasmid bearing antibiotic resistance genes may allow for subsequent plasmid acquisition by these variants, potentially leading to further plasmid-borne multidrug resistance. Our results can inform containment initiatives, help track associated outcomes and international spread, and help determine how widespread the risk is.

INTRODUCTION

Typhoid fever, caused by Salmonella enterica serovar Typhi (S. Typhi), is a life-threatening bacterial infection that presents with numerous symptoms, including a prolonged high fever and gastrointestinal issues. Recent and continued outbreaks of multidrug resistant (MDR) and extensively drug resistant (XDR) typhoid demonstrate the difficulty in containing this infectious disease.^1–3 According to the WHO, MDR S. Typhi are defined as resistant to the first-line recommended drugs for treatment, which include chloramphenicol, ampicillin, and trimethoprim–sulfamethoxazole.⁴ XDR S. Typhi are defined as resistant to all recommended antibiotics for typhoid fever, which include third-generation cephalosporins and fluoroquinolones.⁵ The current burden of MDR and XDR typhoid is largely in low- and middle-income countries, with isolated cases resulting from travel reported.^3,4,6

Haplotype 58 (lineage 4.3.1) is the major causative driver of MDR S. Typhi in Asia and Africa.^3,7 Multidrug resistance in the H58 lineage is largely associated with an MDR composite transposon element (Figure 1). This Tn2670-like complex is comprised of Tn6029, which encodes bla_TEM-1, sul2, strA (aph[3″]-Ib), strB (aph[6]-Id), and a partial IncQ1 replicon; inserted into Tn2670, which is made up of Tn21 carrying a class I integron that encodes dfrA and sul1, which is inserted within Tn9, which encodes catA1.^3,8 Integration of the MDR composite transposon (Figure 1) on the chromosome at different IS1 sites distinguishes the sublineage 4.3.1.1.⁹ Recently, a new variant sublineage of an XDR extended spectrum beta-lactamase (ESBL)–producing S. Typhi (4.3.1.1.P1) emerged in Pakistan.^1,9 This sublineage has the MDR composite transposon integrated at yidA and has also acquired an IncY plasmid (p60006) carrying the ESBL blaCTX-M-15 and quinolone resistance qnrS1 gene,¹ which together with the single chromosomal point mutation, gyrA-S83F, confer resistance to ESBLs and ciprofloxacin, respectively,¹ rendering them XDR.⁴

Figure 1.

Schematic of the composite transposon resistance region in the outbreak strain. XDR = extensively drug resistant.

The XDR typhoid outbreak in Pakistan was first observed in the Sindh Province in 2016, with infections clustered around sewage lines in the city of Hyderabad.¹⁰ Previously, we observed² an increase in XDR typhoid cases in the Punjab Province of Pakistan, with children bearing the greatest burden. This indicated that XDR typhoid had spread beyond the southern Sindh Province. Indeed, further studies¹¹ have also reported the incidence of XDR typhoid outside of Sindh. Although the original outbreak strain contained blaCTX-M-15 on the IncY plasmid, a recent article⁶ showed that chromosomal integrations in XDR typhoid isolates from people with recent travel to Pakistan. Thus, after observing incidence of XDR typhoid across Pakistan, we sought to determine the genetic diversity of ceftriaxone-resistant S. Typhi clinical isolates collected across Pakistan by whole-genome sequencing (WGS). These data would provide insights into the molecular epidemiology of the outbreak and can inform containment initiatives, help model evolution of the outbreak, and help determine how widespread the risk is.

MATERIALS AND METHODS

Isolation and identification of isolates.

Collection, isolation, and identification of isolates were performed at the section of Microbiology, Shaukat Khanum Memorial Cancer Hospital and Research Center (SKMCH&RC), Lahore, Pakistan. Blood culture samples collected via SKMCH&RC’s laboratory network were cultured using the BacT/Alert blood culture system (bioMerieux, France). Positive signaled bottles were subcultured onto chocolate, blood, and MacConkey agar plates. Colonies suggestive of salmonellae were identified based on colony morphology, serotyping, and biochemical tests using the API 20E (bioMerieux, France). Salmonella isolates were tested for antibiotic susceptibility using the Kirby–Bauer disk diffusion method on Muller–Hinton agar with standard antimicrobial disks. Salmonella isolates were tested for ampicillin, ceftriaxone, cefixime, chloramphenicol, ciprofloxacin, imipenem, meropenem, and azithromycin susceptibility in accordance with Clinical Laboratory Standards Institute 2020 guidelines.¹² Isolates identified as ceftriaxone-resistant S. Typhi between November 2018 and December 2020 were archived at –80°C. During this period, 58 strains out of a total 1,370 ceftriaxone-resistant S. Typhi positive blood cultures (just > 4% of total) were then selected randomly from these archived strains. This sample size represents a 95% chance of detecting a variant at 5% frequency in a specific population.¹³ The isolates were recultured on blood agar and DNA was extracted for sequencing from single colonies.

Genomic DNA extraction, WGS assembly, mapping, and resistance gene identification.

To extract genomic DNA, Salmonella isolates were grown on chocolate agar plates and phenol–chloroform–isoamyl alcohol DNA extraction was performed as described previously.¹⁴ Extracted genomic DNA was prepared for sequencing using the NEXTERA XT DNA Library Preparation Kits (Illumina, San Diego, CA). Sequencing (151-bp paired-end reads) was performed on Illumina MiSeq at the Broad Institute (Cambridge, MA). Raw BAM files were trimmed using the BBDuk trimmer in Geneious version 2021.2.2 (Dotmatics, Boston, MA), with default parameters and a minimum Q value of 20. Trimmed sequences were exported to paired FASTQ files for further analysis. Paired reads were next assembled using SPAdes version 3.15.2 using default settings and existing trim regions in Geneious version 2021.2.2 (Dotmatics, Boston, MA), and the outputted scaffolds were assembled to the outbreak reference genome (NZ_ LT882486) and reference plasmid (LT906492.1) using the Geneious Mapper using default settings (Medium Sensitivity/Fast) and existing trim regions. Transposon events were annotated as described previously.^2,3 Genotyping was performed using the Genotyphi scheme.⁷ For reads where there was ambiguity resulting from repeated regions in the building of scaffolds, raw reads were mapped onto the genome to assess coverage.

To determine the presence of known resistance genes, scaffolds were analyzed for acquired resistance determinants described in the outbreak strain (blaCTX-M-15, blaTEM-1, aac[6′]-laa, aph[3″]-lb [strA], aph[6]-ld[strB], qnrS1, dfrA7, sul1, sul2, catA1, gyrA S83F), as well as any other known determinants or chromosomal point mutations using the ResFinder 4.1 database (90% identity, 60% cutoff), using the online CGEinterface¹⁵. Plasmid replicons were identified using the PlasmidFinder 2.1 (90% identity, 60% cutoff) online interface.¹⁶

To determine the location of chromosomal integrations and genomic rearrangements (deletions) for isolates that lacked the IncY or partial IncQ1 replicon on the chromosome or lacked any expected resistance genes, the alignments of composite resistance island and plasmids were probed manually for partially aligned scaffolds and deletions (no assembled reads). The partially aligned scaffold and scaffolds containing the resistance gene were searched for insertion junctions (partial alignment to plasmid and partial alignment to chromosome). Alignments were performed with Mauve in Geneious version 2021.2.2. Gene Graphics was used for creating figures.¹⁷

Phylogenetic analysis.

CSI Phylogeny-1.4 was used to determine high-quality (HQ) single nucleotide polymorphism (SNP) differences¹⁸ using a set of predefined filtering parameters noted here and described in Kaas et al.¹⁸ Assembled scaffolds for all 58 isolates were uploaded to the online interface, along with the outbreak strain as the reference (which was included in the final phylogeny) and submitted with the preset parameters (minimum depth at SNP positions, 10×; minimal relative depth at SNP positions, 10%; minimum distance between SNPs, 10 bp; minimum SNP quality, 30; minimum read mapping quality, 25; and minimum z-score, 1.96. The resulting SNP pair count matrix was extracted and the resulting FASTA file containing the SNP alignment was used to construct a phylogenetic tree using RaxML version 8.2.11 (nucleotide model, GTR GAMMA; algorithm, rapid hill-climbing; number of starting trees or bootstrap replicates, 100; parsimony random seed, 1, start with complete random tree).

RESULTS

Antimicrobial susceptibility profiles.

Fifty-eight isolates of ceftriaxone-resistant S. Typhi collected from across Pakistan from February 8, 2018 to December 10, 2020 (Supplemental Figure 1) were randomly chosen for WGS. Full susceptibility data of sequenced isolates is provided in Supplemental Table 1. All isolates were resistant to ampicillin and ceftriaxone (Figure 2A). Fifty-four of 58 were resistant to ciprofloxacin; 4 of 58 were intermediate to ciprofloxacin. Almost all were resistant to chloramphenicol (57 of 58) (Figure 2A). All isolates were susceptible to imipenem, meropenem, and azithromycin (Figure 2A). Based on WGS, all were ESBL-producing (either chromosomally or on the plasmid) as a result of the presence of blaCTX-M-15. The age of patients ranged from 1 to 45 years old, with the majority being younger than 15 years (42 of 58) (Figure 2B).

Figure 2.

(A) Antibiotic resistance characteristics of isolates. (B) Age distribution of Salmonella enterica serovar Typhi cases isolated. Y = years.

Genomic architecture of drug resistance.

Reads were assembled and mapped to the previous Sindh outbreak strain (NZ_LT882486) and IncY plasmid (LT906492.1, p60006).¹ All location numbers for the chromosome or plasmid correspond to these two sequences, respectively. No sequenced isolates had any resistance genes or known resistant point mutations outside of what was reported previously in the outbreak strain. All contained the GyrA S83F mutation.

There was limited genetic diversity among the isolates, with only two isolates (28163, 16067) containing more than 10 HQ SNP differences from the outbreak strain. All isolates belonged to type 4.3.1 and either had the IncY plasmid (p60006) or integrated a portion of the p60006 plasmid into the chromosome. The phylogenetic tree of the isolates (RaxML) with the outbreak strain based on HQ SNP differences shows the close relationships between the isolates (Figure 3).

Figure 3.

Phylogenetic tree of extended spectrum beta-lactamase–producing isolates in this study with the outbreak reference strain (NZ_LT882486) from Klemm et al.¹ The different genomic architecture is described fully in the text and the date of collection is annotated for each isolate in the tree. All isolates were from Punjab, unless noted otherwise. Briefly, the label consists of isolate no.-genomic architecture/integration no.-date of collection-location (if not from Punjab). Genomic architecture A contains all previously identified resistance genes and plasmid replicons of the outbreak strain. Genomic architecture B contains an ∼7.8-kb deletion of the composite multidrug resistant transposon compared with the outbreak strain. A.1 and B.1 contain a variation of plasmid p6006 (p60006v1), which contains a small deletion. There are five different integrations of blaCTX-M-15 on the chromosome, which are illustrated in Figure 5.

Twenty-five of 58 isolates had all the previously identified resistance genes and plasmid replicons of the outbreak strain (indicated as architecture A in Figures 3 and 4A). These were all resistant to ceftriaxone and ciprofloxacin, as expected. Two common changes among isolates were an ∼7.8-kb deletion (4,599,301–4,607,157) of the majority of Tn6029 of the composite transposon (Figure 1), which contained the partial IncQ1 replicon and straA, straB, bla-TEM-1, and sul2 (Figure 5A, indicated as architecture B in Figures 3 and 4A), and a 331-bp deletion (65,039–65,369) of p60006, which we will refer to as p60006v1 (A.1 and B.1, based on respective genomic architectures). Of the 25 isolates that had all outbreak resistance genes and plasmid replicon present, nine had p60006v1. Thirteen of 58 isolates had the outbreak plasmid (2 of 13 were p60006v1) and the Tn6029 deletion (Figure 4A). All isolates (with genomic architecture A and B) described earlier were from Punjab (Figure 4B). From the literature¹¹ (independent of our sequenced isolates), one of five previously sequenced XDR typhoid isolates from Punjab in 2018 lacked a similar ∼7.8-kb internal region containing the partial IncQ1 replicon and strA, strB, bla-TEM-1, and sul2. Two previously sequenced isolates from Nair et al.⁶ also reported an ∼7-kb deletion of the composite MDR transposon in this region. One isolate from our study had p60006 and the whole composite transposon missing (Figure 4A). This isolate was sensitive to chloramphenicol, as expected because of the loss of the catA1 gene Supplemental Table 1).

Figure 4.

(A) Distribution of the genomic architecture variations among all Salmonella enterica serovar Typhi isolates (N = 58). (B) Distribution of all S. enterica serovar Typhi isolates by genomic architecture and location. KPK = Khyber Pakhtunkhwa; MDR = multidrug resistant.

Figure 5.

(A) Schematic of common deletion of Tn6029 in multiple isolates. (B–F) Schematic of gene architecture of chromosomal drug resistance integrations. Insertion sequences from p6006 are illustrated in the shaded box. ATPase = adenosine triphosphatase; mRNA = messenger RNA; XDR = extensively drug resistant.

Of the isolates missing the IncY plasmid replicon, but still containing the blaCTX-M-15 gene, there were five integration events, one of which was more frequent than the others (Figure 4A). The integration events, along with location and collection date analyses, are described next and illustrated in Figure 5.

Integration 1.

Thirteen of 58 isolates integrated ∼44 kb containing bla-TEM-1, bla-CTX-M-15, and qnrS1 from p6006v1 into the chromosome between Tn21 (at IS-6 element sequence 4,600,044) and before the other IS-6 element (4,606,772). Alternatively, only bla-CTX-M-15 and qnrS1 were moved from the plasmid into the chromosome between position 4,600,044 and just after IS-6 element (before 4,605,328) (Figure 5B, Supplemental Table 1). Based on the resulting sequences, a previous insertion event and subsequent deletion of Tn6029 containing sul2, strA, and strB occurred. Alternatively, deletion during recombination may have occurred. These isolates were collected between January 31, 2020 and December 9, 2020, and 7 of 13 of isolates were from Quetta, 5 of 13 were from Punjab, and 1 of 13 were from Khyber Pakhtunkhwa (KPK) (Figure 4B). As expected, these were all resistant to ceftriaxone and ciprofloxacin Supplemental Table 1).

Integration 2.

Three of 58 isolates contained a 4,228-bp insertion of ISECp1-blaCTX-M-15-tnp (transposase) from p60006 (69,598–65,370) in gutQ (3,563,153) (Figure 5C, Supplemental Table 1), in addition to the Tn6029 deletion (Figure 5A). These were collected from August 7, 2019 to February 2, 2020, and one of three of the isolates are from Quetta; two of three were from Punjab (Figure 4B). As expected, because they have blaCTX-M-15, but lack qnrS1, all three isolates are resistant to ceftriaxone but are intermediate to ciprofloxacin Supplemental Table 1).

Integration 3.

One isolate had a 3,051-bp insertion of ISECpl-blaCTX-M15-tnp (65,369–68,419) from p60006 just before ompF (before 2,730,382) (Figure 5D, Supplemental Table 1). This was collected on February 28, 2020 in KPK (Figure 4B). Because this isolate has blaCTX-M-15, it is resistant to ceftriaxone; however, it is intermediate to ciprofloxacin, which corresponds to the lack of the qnrS1 gene on the integrated fragment Supplemental Table 1).

Integration 4.

One isolate had a 10,726-bp insertion of blaCTX-M-15 and qnrS1 from p60006 in fadJ (before 1,301,557) (Figure 5E, Supplemental Table 1), in addition to the Tn6029 deletion (Figure 5A). This was collected August 11, 2019, from Punjab (Figure 4B). This isolate is resistant to ceftriaxone and ciprofloxacin Supplemental Table 1).

Integration 5.

One isolate had an ∼71-kb (71,261-bp) insertion from p6006 containing blaCTX-M-15 and qnrS1 between Tn21 (at yidA – 4,591,208) and bla-TEM-1 on the chromosome (4,605,328 – NZ_LT882486), resulting in the deletion of sul2, strA, and strB (Figure 5F, Supplemental Table 1). This was collected February 2, 2020, from Quetta (Figure 4B). This isolate is resistant to ceftriaxone and ciprofloxacin Supplemental Table 1).

DISCUSSION

We describe XDR typhoid that has established outside the initial outbreak center in Pakistan. All the isolates collected were of the H58 lineage and appear related to the outbreak strain. However, on the molecular level, we find that the MDR composite transposon is very susceptible to genomic rearrangement and integration events, resulting from the abundance of insertion and transposon element sequences. We found that about one third of all isolates sampled (∼33%) had integrated blaCTX-M-15 into the chromosome, in five different integration events. This is compared with recent work⁶ that found ∼19% of isolates (N = 69) with integration of blaCTX-M-15 on the chromosome in three separate locations, with the integration site near gutQ being similar to one of our identified integration sites. Aside from integration of the plasmid into the chromosome, a large deletion of Tn6029 within the composite MDR transposon was common. The increased spread of isolates with the resistance genes integrated on the chromosome may be indicative of increased fitness and energy conservation compared with plasmid maintenance¹⁹; however, this is beyond the scope of this work. Integration of the ESBL on the chromosome and loss of plasmid, as observed in ∼33% of sequenced isolates (Figure 5), also enable acquisition of another plasmid containing drug resistance or virulence genes. In addition, newly acquired transposon elements and insertion sequences on the chromosome afford additional opportunities for genomic rearrangements and acquisition of additional resistance or virulence genes from plasmids. As seen with COVID-19, it is essential to be proactive about emerging threats. Our results reveal there are many circulating typhoid variants in Pakistan that have lost the IncY plasmid after integrating the ESBL into the chromosome. These could acquire new plasmids to make them resistant to the few antibiotics to which it is susceptible (pan-resistant). In addition, because the transposon region is highly amenable to insertions—as demonstrated by the identification of multiple variants—the more time it circulates, the more time it has to evolve new variants with additional resistant genes. Thus, approaches and initiatives to reduce spread and burden (i.e., vaccine campaigns, targeted surveillance to monitor prevalence and prevalence rates) need to be a top priority.

Location analyses reveals clustering of isolates with the same integration events in different regions. Notably, all isolates collected from outside of Punjab, in Quetta and KPK, had integrated blaCTX-M-15 into the chromosome. Furthermore, the spread of isolates with integration 1 to Quetta, Punjab, and KPK suggests its emerging dominance. Integration 1 was also larger (∼44 kb) than the previously largest reported⁶ integration of ∼25 kb of the plasmid. Although the IncY outbreak plasmid contains genes for different virulence and defense systems, interestingly, like the previously reported⁶ ∼25-kb integration element, the portion of the IncY plasmid integrated in integration 1 carries genes associated with a RelE/ParE family messenger RNA toxin–antitoxin system,²⁰ which may be important for stabilization of the resistance genes in the chromosome. Integrated sequences also contains genes involved in the bacteriophage exclusion system,²¹ which is involved in phage resistance. Integration 5, which is even larger (∼71 kb), in addition to all the aforementioned systems, contains genes associated with a type IV secretion system, involved in virulence.²² The relevance of these other systems to pathogenicity and patient outcomes remains an open question.

From previous analysis, nearly all H58 isolates carry a copy of insertion sequence IS1 between the chromosomal genes STY3618 and STY3619, near cyaA.³ During our sequence analyses, we found that these IS1 sequences are also present in all ceftriaxone-resistant isolates collected. Although there is partial annotation on the reference genome, we find the corresponding sequences within the unmapped scaffolds. Thus, the abundance of IS1 sites and multiple repeats resulting from transposon elements require extra scrutiny for sequences analysis, and also provide opportunities for further genomic rearrangements.

Overall, we show the continued spread of XDR typhoid and, based on WGS, observe that there are now multiple XDR isolates circulating within Pakistan, with resistance resulting from chromosomal integration of the resistance plasmid continuing to increase. Given the continued observed spread of XDR typhoid, and the associated genomic diversity, it is critical to take steps to lower the burden of S. Typhi infections, by improving vaccination rates and quality of drinking water, as well as by promoting good hygiene practices. Similar to our previous findings², we see that children have the greatest burden of typhoid fever (Figure 1). In 2018, the WHO prequalified the new conjugate typhoid vaccine (Typbar TCV, Bharat Biotech) and recommended use in children 6 months old in countries endemic with typhoid.²³ Stewardship to ensure availability, access, and trust of the vaccine should thus be a priority.

Identifying XDR typhoid genomic sequences allows better monitoring of how infections are introduced and spread both locally and globally (i.e., one can determine the sequence of XDR typhoid isolated abroad and see whether it matches variants in a specific region). Knowledge about whether a specific case is a novel variant or travel-related is important information for containment initiatives. Furthermore, as with different COVID-19 variants, sequencing information can help determine whether a specific variant is associated with more dangerous outcomes.²⁴ Understanding the genomic architecture and prevalence of variants will play a critical role in building predictive models of the XDR typhoid outbreak.²⁵

Note: Supplemental figure and tables appear at www.ajtmh.org.