Klebsiella pneumonia in Sudan: Multidrug Resistance, Polyclonal Dissemination, and Virulence

The emergence and global expansion of hyper-virulent and multidrug resistant (MDR) Klebsiella pneumoniae is an increasing healthcare threat worldwide. The epidemiology of MDR K. pneumoniae is under-characterized in many parts of the world, particularly Africa. In this study, K. pneumoniae isolates from hospitals in Khartoum, Sudan, have been whole-genome sequenced to investigate their molecular epidemiology, virulence, and resistome profiles. Eighty-six K. pneumoniae were recovered from patients in five hospitals in Khartoum between 2016 and 2020. Antimicrobial susceptibility was performed by disk-diffusion and broth microdilution. All isolates underwent whole genome sequencing using Illumina MiSeq; cgMLST was determined using Ridom SeqSphere+, and 7-loci MLST virulence genes and resistomes were identified. MDR was observed at 80%, with 35 isolates (41%) confirmed carbapenem-resistant. Thirty-seven sequence types were identified, and 14 transmission clusters (TC). Five of these TCs involved more than one hospital. Ybt9 was the most common virulence gene detected, in addition to some isolates harbouring iuc and rmp1. There is a diverse population of K. pneumoniae in Khartoum hospitals, harbouring multiple resistance genes, including genes coding for ESBLs, carbapenemases, and aminoglycoside-modifying enzymes, across multiple ST’s. The majority of isolates were singletons and transmissions were rare.


Introduction
Klebsiella pneumoniae is an important global pathogen causing a variety of infections in community and healthcare-associated infections, such as pneumonia, urinary tract infections (UTI), and bloodstream infections. It poses a serious threat to human health and is one of the six highly virulent and antibiotic resistant bacterial pathogens: Enterococcus faecium, Staphylococcus aureus, Klebsiella pneumoniae, Acinetobacter baumannii, Pseudomonas aeruginosa, and Enterobacter spp. (ESKAPE), requiring urgent global attention [1]. Multidrug resistance and carbapenem resistance in K. pneumoniae (MDR-Kp and CR-Kp) is of particular concern as treatment options are limited.
The population structure of K. pneumoniae appears to be diverse yet highly structured, with distinct clonal groups (CG) subdivided into MDR-and hypervirulent (Hv)-clones [2]. A subset of these clones contribute to global diseases and outbreaks and are referred to as 'global problem clones', involving the transfer and spread of antibiotic resistance genes and endemic plasmids in highly disseminating successful clones worldwide [3]. pneumoniae isolates were confirmed by MALDI-TOF and Kleborate, while 2 isolates (LH_F129 and LH_F72) were identified as Klebsiella quasipneumoniae subsp. quasipneumoniae and Klebsiella variicola subsp. variicola, respectively. The organisms came from a variety of samples: blood (n = 25, 29.4%), pus (n = 2, 2.4%), sputum (n = 9, 10.6%), urine (n = 37, 43.5%), and wound swabs (n = 12, 14.1%) ( Figure 1; Table 1). Most isolates came from Fedail Hospital, followed by Ribat University Hospital, both of which are two of the largest hospitals in Khartoum. Urine and blood samples were most common across all isolates.   Most isolates came from Fedail Hospital, followed by Ribat University Hospital, both of which are two of the largest hospitals in Khartoum. Urine and blood samples were most common across all isolates. Sixty-eight isolates (80%) were MDR (resistant to >3 antibiotic classes), 35 of which (51%) were carbapenem resistant (imipenem andor meropenem MIC ≥ 4 mgL). The remaining 16 isolates were susceptible to all tested antibiotics. All isolates were colistin susceptible (MIC < 2 mgL). Multiple resistance mechanisms were identified (Table 2 and Supplementary Table S1), contributing to the observed MDR phenotypes. Aminoglycoside resistance was present in 49 isolates (57%), mediated by multiple aminoglycoside-modifying enzymes. β-lactam resistance was detected in 58 isolates and was mediated by acquired β-lactamases: OXA-1 (n = 19), OXA-9 (n = 4, two of which were co-harbouring OXA-1 and TEM-1), CMY (n = 3, one of which was co-harbouring OXA-1), DHA-1 (n = 3, two of which were coharbouring TEM-1), and SCO-1 (n = 3, one of which was co-harbouring TEM-1); TEM-1 was the most prevalent β-lactamase, identified in 40 isolates. The acquired extended-spectrum β-lactamase (ESBL) CTX-M-15 was present in 64 isolates, and CTX-M-14 was present in four isolates, three of which co-harboured CTX-M-15. Seventeen isolates did not harbour any ESBLs. SHV-variants were present in all K. pneumoniae isolates.

Carb R bla
Intrinsic β-Lactamase bla Virulence Carbapenem resistance was mediated by NDM-1 (n = 23), NDM-4 (n = 1), and NDM-5 (n = 11, three of which co-harboured OXA-48); one isolate harboured OXA-48 on its own, and one isolate harboured OXA-232 (OXA-48-like enzyme). Three isolates displayed phenotypic carbapenem resistance, but no acquired carbapenemases were detected, and the observed carbapenem resistance was associated with a combination of an ESBL and modifications in the OmpK35OmpK36 (Table 2 and Table S1). LH_F164 possessed a truncated bla NDM-1 gene but also harboured OXA-1, CTX-M-15, and a nucleotide substitution (G690A) in the OmpK35 encoding gene, leading to a premature stop codon. LH_R313, on the other hand, lacked an ompK35 gene, and carbapenem resistance was mediated by a modification of OmpK36 containing an LSP insertion in the amino acid sequence at position 184.
A total of 37 different sequence types (STs) were identified in the study, highlighting the epidemiological diversity of the isolates. We observed 14 small transmission clusters (TC) by cgMLST, with isolates differing by 0-5 alleles ( Figure 2). TC-1 (ST17) comprises three isolates collected from Al-Saha Hospital (ASH) in 2019; however, they differed in their respective resistomes. Metadata relating to the location of patients or date of isolation are not available for these isolates. LH_HishK11 and LH_HishK13 are both carbapenemsusceptible, while LH_HishK12 is MDR and harbours multiple aminoglycoside-modifying enzymes (aac(3)-IIa-like;aac(6')-Ib-cr-like;aadA5) and β-lactamases: bla CTX-M-15 , bla OXA-1 , and bla NDM-1 . All other TCs comprise isolates with identical resistomes (Figure 2 and Table S1).
The most frequently isolated ST in the study was ST147 (n = 9), followed by ST20 (n = 7), ST437 (n = 7), ST101 (n = 6), ST15 (n = 6), ST307 (n = 6), ST383 (n = 5), and ST17 (n = 4), all of which are high-risk global clones (GC), defined as epidemic high-risk clones over-represented globally [2], with the remaining 29 singleton STs. Isolates within the same ST exhibited some similarities at the genomic level, but as seen in Table 2 and Table S1, some had different resistance phenotypes and genes. For example, ST147 isolates came from multiple hospitals, with isolates from FH all collected within a 1-month timeframe (July-August 2020), but they differed by ≥21 alleles and transmission was ruled out. Moreover, differences were seen in their virulence and resistome profiles, with LH_F134 harbouring no recognised virulence factors, while LH_F149, LH_F18, and LH_F97 all harboured ybt9 on a ICEKp3 structure. LH_F134, LH_F149, and LH_F18 were carbapenem resistant, mediated by bla NDM-1 ; however, LH_F97 was carbapenem susceptible. Similar data are observed in ST-147 isolates from RUH which had identical resistomes, except R344, which harboured bla NDM-5 and not bla NDM-1 and differed by >50 allelic differences per the cgMLST and was therefore not part of a single transmission cluster. The only TC in this group were two identical isolates from Al Baraha Hospital, which also had identical resistomes.
All ST15 isolates were MDR (and comprised two, TC-12 and TC-13, and two singletons), but were carbapenem-susceptible, except for LH_F164, which was carbapenemresistant (imipenem MIC 16 mgL), associated with a G690A substitution in the ompK35 gene (leading to a premature stop codon) combined with bla OXA-1 . LH_F164 was the only strain which was also positive for ybt9 (on ICEK3p) within this ST. Isolates within ST307 came from two different hospitals (RUH and FH), with 56 isolates from RUH. Three of these isolates form a transmission cluster. Interestingly, one isolate, collected 4 months later in July 2020 (LH_R275) from RUH appears to have lost the MDR and CR pheno-and geno-types.
Of the remaining 29 singleton STs, 11 of them were MDR and CR, whereas the remaining 18 were susceptible to all tested antibiotics.
Forty different K-loci were identified, as detailed in Supplementary Table S2, along with seven variants of O-loci, most common of which were O1 (n = 30) and O2 (n = 21). The O-locus variant could not be identified in 10 isolates.

Discussion
The aim of this study was to conduct a snapshot analysis of the molecular epidemiology of the K. pneumoniae population from hospitals in Khartoum, Sudan. We have included isolates from five hospitals through several sampling timeframes over the years. K. pneumoniae was implicated in multiple infections, including blood, skin and soft tissue (SST), and urine. The majority of isolates (81%) were MDR, of which more than half were also carbapenem resistant, mediated mainly by NDM-1-, NDM-5-, and OXA-48-like enzymes. Interestingly, KPC was absent from the isolate collection, despite its global prevalence in CR-Kp, particularly in ST258. Carbapenem resistance was also mediated by the presence of an ESBL and modifications in OmpK35 and OmpK36 in three isolates ( Table 2). Modifications in outer-membrane porins which restricts antibiotic entry is an important carbapenem resistance mechanism [15]. OmpK35 is usually truncated due to a mutation encoding a frame shift that results in a premature stop codon [16,17]. OmpK36 is more heterogenous, rarely truncated, and resistance mutations are common leading either to abundance of OmpK36 in the outer membrane or constriction of the pore size [15,18]. Carbapenem MIC for LH_R290 showed intermediate resistance (IMI 4 mgL) without an acquired carbapenamase; however, the isolate harbours SCO-1, a plasmid-encoded ESBL able to hydrolyse not only penicillins but also, to a lesser degree, cephalosporins and carbapenems. Since its discovery in 2007, the bla SCO-1 gene (GenBank accession no. EF063111) has been identified in Acinetobacter baumannii, Escherichia coli, Serratia marcescens, Klebsiella aerogenes, Salmonella enterica, and only four K. pneumoniae isolates [19].
The epidemiological analysis of K. pneumoniae in Sudan revealed a large diversity of 37 different STs, with 13 transmission clusters (Figure 1). Five of these clusters have been collected from different hospitals, thereby indicating intra-hospital circulation of clones. Global problem high-risk MDR clones (GC), identified by Wyres et al., 2020, have been found in this study; however, we noted that not all were MDR. We have identified multiple losses of resistance within these GCs over time. For example, isolates in ST307 at RUH collected in FebruaryMarch 2020 were carbapenem-resistant; however, isolate LH_R275 collected in July 2020 was neither MDR nor CR. Similarly, isolates within ST147: LH_F134, LH_F149, LH_F18, were CR harbouring bla NDM-1 ; however, LH_F97 (isolated from the same hospital) is carbapenem-susceptible, but had acquired the ybt9 virulence gene on ICEKp3. ST437 isolates appear to have lost bla NDM-1 (from 2018) and acquired bla NDM-5 in the isolates collected in 2020. However, we did not specifically look for plasmids in this study, and further detailed investigations of possible plasmid acquisition and loss events must be conducted to confirm the observed results. The differences in the resistomevirulomes of isolates within the same ST and transmission cluster is important to note for outbreak and epidemiological studies. The data provided by genomic analysis are an important tool where molecular epidemiology is combined with patient clinical data and the resistomes to obtain accurate information on transmission of resistant and virulent pathogens.
Yersinibectin (ybt9 and ybt10) were the most identified virulence genes in the study and were mostly associated with GCs: ST101, ST147, and ST307, in addition to their occurrence in singleton non-GCs, such as ST39, ST437, and ST882 (Table 2). ybt is usually present in 30-40% of K. pneumoniae human HAI isolates and up to 13% of communityacquired isolates [20]. Other virulence genes also identified in the study included aerobactin, iuc1, and hypermucoidy-associated rmpA genes. We have noted AMR-hypervirulence convergence events in 24 isolates, which harboured a virulence gene and were MDR: ST530, ST11, ST147, ST101, ST231, ST307, ST15, ST383, ST437, and ST45. It is important to note that we have not confirmed the virulence phenotypes of the isolates, and more research into the clinical significance of hypervirulence is required.
Antibiotics 2023, 12, 233 9 of 13 ST383 isolates were particularly unique in their resistomes and virulomes. The isolates co-harboured multiple ESBLs and carbapenemases simultaneously, CTX-M-14 and CTX-M-15, in addition to NDM-5 and OXA-48 in the same isolates and iuc-1, rmp1 on KpVP-1, and rmpA2. ST383 is a prevalent MDR clone in China, Australia, the UK, and Germany [21]. In a study from Egypt, ST383 was identified in a single isolate simultaneously encoding CTX-M-14 (on an IncLM plasmid) and CTX-M-15 (encoded on a hybrid IncHI1BIncFIB plasmid) in addition to both NDM-5 and OXA-48 [22]. Results by PlasmidFinder show the same plasmids are present in our ST383 isolates (incFIBIncHI1B and incL).This clearly indicates the endemicity of this particular clone in the region, as the data presented in the Egyptian isolate is identical to that in our ST383 cluster.
In our previous study on K. pneumonia isolated in Khartoum, 2015-2016, we performed MLST on 117 isolates. All were MDR, and 42.8% were CR-Kp [14]. A similar diversity was observed in the current study, with 52 different STs, with the most common being ST383 (n = 8), ST101 (n = 5), and ST48 (n = 2), in addition to assignment of 15 novel STs. NDM was the most common carbapenemase, in addition to VIM, which has not been detected in the current study. A study conducted in Sudan by Adam et al. from 2015Adam et al. from -2016 revealed that metallo-β-lactamases (VIM and IMP) were prevalent in K. pneumoniae (1520 isolates), and found in combination with NDM in 520 isolates, in contrast to our study where VIM and IMP were not detected and NDM was the most prevalent MBL. Comparing the data from these studies indicates the maintenance of ST101, ST383, ST219, and ST437 lineages in Khartoum hospitals over the years, and the presence of novel lineages, as reported in this study.
In comparison to the present study, we also investigated A. baumannii from several different hospitals in Khartoum during the same time period and found that the A. baumannii isolates were mostly transmissions and very few were singletons [24]. This highlights that even in an area that obviously has a high incidence of hospital-acquired infections and patient-to-patient transmissions, including between hospitals, K. pneumoniae is not easily transmitted. While most K. pneumoniae in this study were not transmissions, they do harbour similar carbapenemases, which may indicate transmission of resistance genesmobile genetic elements, which will be the focus of a follow-up study.
When compared to other African studies, the K. pneumoniae population in Southern Nigeria was found to contain four dominant lineages: ST307, ST524, ST15, and ST25, while CR remained low at 8%, with no isolate carrying a combination of carbapenemases, in contrast to our study, where carbapenem-resistance was much higher (41.6% of 84 isolates) and three strains co-harboured NDM and OXA-48 [25]. A multi-centre pilot study in Egypt revealed that ST11, ST147, ST231, ST383, and ST101 were prevalent, which is similar to our study [26]. The observed similarities and difference with the local, regional, and international studies highlights the diversity and dynamic nature of the representative local population of K. pneumoniae as well as the potential for regional dissemination of clones in different countries, and the need to support local capacity in robust epidemiological surveillance. The first WGS study on a single ST14 K. pneumoniae from Sudan was conducted in 2019 [27]. To the best of our knowledge, this is the first WGS-based study of a large collection of K. pneumoniae from Sudan.
Through this study, we aimed to generate robust epidemiological data on K. pneumoniae; albeit acknowledging some limitations. We have tried to collect as many K. pneumoniae isolates within the indicated timeframes. However, we cannot exclude sampling bias (such as missed samples or misidentification), as the study team was not directly involved in the sampling procedures. Routine sampling of infected patients is not always performed in local hospitals in Sudan, and sampling is frequently sporadic and based on individual doctors' rather than hospital policiesguidelines due to poor microbiological infrastructure in hospitals. The fragmented health service, and the large out-of-pocket contribution by individuals and households, as presented in the introduction, lead to some hospitals also outsourcing microbiology to private laboratories. This study, however, focused on data from hospital-based laboratories only. Furthermore, the study was interrupted during COVID-19 lockdowns, so isolates between April and July 2020 were not collected. Some associated demographic data were also lacking for some isolates collected in 2018-2019. The sampling strategy therefore cannot exclude missed transmissions. Nevertheless, this study provides a starting point for the understanding of the population structure and diversity of K. pneumoniae in Khartoum, Sudan. It furthermore supports national and global efforts in providing robust epidemiological information on important HAI. This study highlights the importance of using WGS in AMR surveillance. Despite the vast possibilities for implementation, major challenges relating to capacity and training, and particularly analysis, of WGS still exist in many LMICs [28].

Materials and Methods
A total of 86 non-repetitive isolates were collected within these timeframes: Twentyfour isolates in 2018-2019 from Al-Baraha Hospital (ABH), Al-Saha Hospital (ASH), and Soba University Hospital (SUH); then 60 isolates in March-September 2020 from Fedail Hospital (FH), Ribat University Hospital (RUH), and SUH. Two additional isolates (from 2016) were included from RUH. Descriptive statistics was used to summarise the details of each isolate, including date of isolation, sample type, and hospital ( Figure 1 and Table 1). The clinical microbiology laboratories in the named hospitals identify clinical specimens to the genus levels by conventional phenotypic and biochemical methods (growth and colony morphology on various media, and biochemically: indole negative, MR negative, VP positive, citrate positive, oxidase negative, and catalase positive). No specific selection criteria were implemented, as we aimed to collect and characterise any K. pneumoniae isolates identified in the hospitals during the study periods. Upon identification of K. pneumoniae by the clinical microbiology laboratories, the isolates were stored at 4 • C and collected by the study team within 48 h. It is important to note that the study team cannot guarantee any misidentification or loss of samples that were not submitted to the study.
All acquired isolates were subsequently confirmed phenotypically by conventional culture methods to exclude any contamination, then genotypically by amplification of 16S-23S rDNA internal transcribed spacer (ITS) of K. pneumoniae, as described in our preceding study [29]. Further identification confirmation was performed by MALDI-TOF prior to WGS.
Total DNA was extracted from the bacterial isolates using a MagAttract HMW DNA Kit (Qiagen, Hilden, Germany), according to the manufacturer's instructions. Sequencing libraries were prepared using the Nextera XT library prep kit (Illumina GmbH, Munich, Germany) for a 250-bp paired-end sequencing run on an Illumina MiSeq platform. De novo assembly was constructed using Velvet v1.1.04. Molecular epidemiology of the isolates was investigated by core-genome MLST (cgMLST) using Ridom ® SeqSphere + version 8.5.1 [32].
All assembled genome sequences were submitted to Genbank under BioPproject ID PRJNA912084.

Conclusions
This study aims to provide a population snapshot of K. pneumoniae in Sudan. ST101, ST147, and ST437 are epidemic and predominantly present in multiple hospitals. MDR and CR-Kp were also prevalent (80% and 50%, respectively), which is alarming. The detection of multiple virulence genes and the potential emergence of AMR-hypervirulence convergence events must also be considered for local surveillance. We conclude that a diverse population of K. pneumoniae is present in hospitals in Khartoum, Sudan. Further genomic investigations, and inclusion of more data from national and regional hospitals, would enable delivery of value on a local, national, regional, and global level in understanding the pathogen dynamics.
Supplementary Materials: The following supporting information can be downloaded at: https:// www.mdpi.com/article/10.3390/antibiotics12020233/s1. Figure S1: Minimum spanning tree (MST) of 84 K. pneumoniae by hospital information. Ridom SeqSphere+ MST for 84 samples based on 2358 columns, pairwise ignoring missing values, logarithmic scale. Cluster distance threshold: 15. Isolates grouped by colour indicate the different hospitals. Samples were collected from five different hospitals, and 37 different STs were identified, in addition to 14 transmission clusters, represented by shaded nodes and arrows. Numbers between the nodes indicate the number of allelic differences; Table S1: Sequence types, antimicrobial susceptibility and associated resistome;  Informed Consent Statement: Patient consent was waived by the Sudanese Research Ethics Committee. All patients consent that any microbiological samples collected in the hospital as part of routine clinical work may be used for research purposes. No patient identifying details were collected for this study.

Data Availability Statement:
The genome assemblies presented in this study are openly available in NCBI GenBank under BioProject ID PRJNA912084.