Multidrug-resistant Klebsiella pneumoniae: a retrospective study in Manaus, Brazil.

Klebsiella pneumoniae is an opportunistic pathogen that can cause several infections, mainly in hospitalised or immunocompromised individuals. The spread of K. pneumoniae emerging virulent and multidrug-resistant clones is a worldwide concern and its identification is crucial to control these strains especially in hospitals. This article reports data related to multi-resistant K. pneumoniae strains, isolated from inpatients in the city of Manaus, Brazil, harbouring virulence and antimicrobial-resistance genes, including high-risk international clones belonging to clonal group (CG) 258. Twenty-one strains isolated from different patients admitted to four hospitals in the city of Manaus, located in the state of Amazonas, Northern Brazil (Amazon Rainforest region) were evaluated. The majority of strains (61.9% n = 13) were classified as multidrug-resistant (MDR), and five strains (23.8%) as extensively drug-resistant (XDR). Several virulence and antimicrobial-resistance genes were found among the strains and eight strains (38.1%) presented the hyper-mucoviscous phenotype. MLST analysis demonstrated a great diversity of STs among the strains, totaling 12 different STs (ST11, ST23, ST198, ST277, ST307, ST340, ST378, ST462, ST502, ST3991, ST3993 and ST5209). Three of these (ST11, ST23 and ST340) belong to CG258.

Introduction

Klebsiella pneumoniae is a member of the Enterobacteriaceae family and an opportunistic pathogen that could cause several infections (pneumonia, urinary tract infections, bacteremia, among others), mainly in hospitalised or immunocompromised individuals (Bengoeche et al. 2019).

The appearance and spread of hyper-virulent strains have increased the number of people susceptible to infections; also, strains of K. pneumoniae have become increasingly resistant to antibiotics, making antibiotic therapy more challenging. Several new antimicrobial-resistance genes were discovered in K. pneumoniae before spreading to other pathogens; blaKPC, blaOXA-48-like and blaNDM-1 are examples (Wyres et al. 2018). Molecular epidemiology analyses allow us to determine the global spread of high-risk clones, thus, providing the necessary data to develop strategies to limit the spread of clinically dangerous strains (Mathers et al. 2015).

The purpose of this study was to determine the pathogenic potential and the antimicrobial-resistance profiles of K. pneumoniae strains isolated from different patients admitted to four hospitals in the city of Manaus, located in the state of Amazonas, Northern Brazil. Besides, to analyse the genetic diversity and epidemiological relationship of the strains, Manaus is a Brazilian municipality, Amazonas capital, and the leading financial centre in the North of the country. It is the most populous city in the entire Brazilian Amazon, located in the centre of the world's largest tropical forest (IBGE 2019).

Recently, the city suffered a collapse in the public health system due to the COVID-19 pandemic (Ferrante et al. 2020). Therefore, studies even before the pandemic period aiming to elucidate and prevent the spread of hyper-virulent and multidrug-resistant bacteria in hospitals in this region are critical (Ribas et al. 2020).

Materials and methods

Bacterial strains

In this study, 21 K. pneumoniae strains isolated from different patients admitted to four hospitals in the city of Manaus, located in the state of Amazonas, Northern Brazil, were evaluated. The strains were randomly selected from November 2014 to May 2016, from different sources, including urine, blood, tracheal secretion, wound secretion, catheter tip, rectal swab (Fig. 1). Bacterial species were initially identified in the hospitals where they were isolated by the automated system VITEK 2 (Biomérieux). Subsequently, to guarantee the correct identification of the species, all strains were molecularly identified by 16S rRNA sequencing using the primers fD1 (5′-AGAGTTTGATCCTGGCTCAG-3′) and rP2 (5′-ACGGCTACCTTGTTACGACTT-3′), according to Weisburg et al. (1991).

Fig. 1

Phenotypic and molecular characterisation of 21 Klebsiella pneumoniae isolates from clinical sources in Manaus, Brazil. Antimicrobials tested: amoxicillin–clavulanate (AMC), amikacin (AMI), ampicillin–sulbactam (APS), aztreonam (ATM), ceftazidime (CAZ), cefaclor (CFC), cefixime (CFM), cefoxitin (CFO), cefazolin (CFZ), ciprofloxacin (CIP), chloramphenicol (CLO), colistin (COL), cefepime (CPM), ceftriaxone (CRO), cefuroxime (CRX), ceftaroline (CTL), cefotetan (CTT), cefotaxime (CTX), doripenem (DOR), doxycycline (DOX), ertapenem (ERT), streptomycin (EST), fosfomicin (FOS), gentamicin (GEN), imipenem (IPM), levofloxacin (LEV), lomefloxacin (LMX), minocycline (MIN), meropenem (MPM), nalidixic acid (NAL), netilmicin (NET), nitrofurantoin (NIT), norfloxacin (NOR), ofloxacin (OFX), piperacillin–tazobactam (PIT), sulphonamide (SUL), trimethoprim–sulfamethoxazole (SUT), ticarcillin–clavulanate (TAC), tetracycline (TET), tobramycin (TOB), trimethoprim (TRI). aColored square: non-susceptible strains; blank squares: susceptible strains. No color was associated to polymyxins, since all strains analysed was susceptible to colistin. bXDR extensively drug-resistant, MDR multidrug-resistant, NC not classified. cFilled circles: detected; unfilled circles: not detected

Hypermucoviscosity test

For the detection of hyper-mucoviscous phenotype (HMV), all strains were inoculated on Mueller–Hinton agar (Oxoid), for approximately 18 h at 37 ºC. Following bacterial growth, using the bacteriological loop, an isolated colony was touched and raised vertically. The formation of a dense string ≥ 5 mm was considered a positive HMV phenotype, according to Wiskur et al. (2008).

Antimicrobial susceptibility test

Antimicrobial susceptibility tests were performed by disc diffusion with 40 different antibiotics on Mueller–Hinton agar (Oxoid), according to Clinical Laboratory Standards Institute (CLSI 2020) recommendations for Enterobacterales. Susceptibility to colistin was determined by microdilution, according to CLSI 2020. The antibiotics tested are described in Fig. 1. Escherichia coli ATCC 25922 and Pseudomonas aeruginosa ATCC 27853 were used as controls in this experiment.

Each strain was considered susceptible or non-susceptible (either intermediate or resistant) to each antibiotic tested. Based on the susceptibility profile, the strains were classified as multidrug-resistant (MDR), extensively drug-resistant (XDR) or pandrug-resistant (PDR), according to Magiorakos et al. (2012).

Detection of virulence and β-lactamase encoding genes

All strains were subjected to Polymerase Chain Reaction (PCR) assays to detect 10 virulence genes (rmpA, kfu, allS, fimH, mrkD, ycfM, entB, iutA, K2, magA) and 15 β- lactamase encoding genes, including carbapenemases (blaGES, blaIMP, blaKPC, blaNDM, blaOXA-1-like, blaOXA-48-like, blaVIM, blaSPM and blaGIM), extended-spectrum β-lactamases—ESBLs (blaCTX-M-Gp1, blaCTX-M-Gp2, blaCTX-M-Gp9, blaCMY-2, blaVEB e blaBEL) and also colistin-resistance gene (mcr-1). The protocols used to investigate the virulence and β-lactamase encoding genes were performed as described by Nakamura-Silva et al. (2021b).

To confirm the detected genes' identity, an amplicon from each gene was randomly selected for DNA sequencing (ABI 3500xL Genetic Analyzer; Applied Biosystems, Foster City, CA). The obtained sequences were compared with those available in GenBank using the BLAST algorithm (http://blast.ncbi.nlm.nih.gov/Blast.cgi).

Multilocus sequence typing (MLST)

To determine sequence types and evaluate the clonal relationships, all strains were submitted to the MLST technique using the primers and PCR conditions for protocol 2 in the MLST database for K. pneumoniae (http://www.pasteur.fr/recherche/genopole/PF8/mlst/Kpneumoniae.html).

Whole-genome sequencing (WGS) and bioinformatic analysis

Aiming a more detailed analysis to elucidate the resistome, viruloma and plasmidoma, ten strains (KpAm03, KpAm04, KpAm05, KpAm06, KpAm07, KpAm09, KpAm12, KpAm14, KpAm17 and KpAm21) were subjected to whole-genome sequencing (WGS), which were selected based on the antimicrobial-resistance classification, presence of hyper-mucoviscosity and virulence genes.

Total genomic DNA was extracted using a PureLinkTM Quick Gel Extraction Kit (Life Technologies, CA). Genomic DNA quality and quantity were assessed using a Qubit1 2.0 fluorometer (Life Technologies) and sequenced via 2 × 150-bp paired-end library on a NextSeq550 or MiSeq platform (Illumina Inc., San Diego, CA), using 250-bp paired-end (PE) reads (Table supplementary 1). Reads with a PHRED quality score below 20 were discarded, and adapters were trimmed using TrimGalore v0.6.5 (https://github.com/FelixKrueger/TrimGalore). The reads were de novo assembled using Unicycler v0.4.0 (Wick et al. 2017) and annotated using the Prokaryotic Genome Annotation Pipeline v.3.2 (PGAP) NCBI.

Antimicrobial-resistance (AMR) genes were identified using ABRicate v1.0.1 (https://github.com/tseemann/abricate) with ResFinder v4.1 database (Bortolaia et al. 2020) and a database for heavy metals and biocides-resistance genes. ABRicate was also used with PlasmidFinder v2.1 database (Carattoli et al. 2014) to detect plasmid replicons among the assemblies contigs. MLST, virulence factors, capsule synthesis (K-locus), lipopolysaccharide (O-locus), and mutations on porins and quinolone resistance determining regions were identified using Kleborate v2.0.1 (Lam et al. 2021).

CSI Phylogeny (Kaas et al. 2014) was used to generate an approximately maximum-likelihood phylogenetic tree the assemblies, using K. pneumoniae strain HS11286 chromosome (RefSeq accession number NC_016845.1) as reference. The tree was rooted at midpoint and annotated with data from ABRicate and Kleborate, as well as metadata, using iTOL v6 (Letunic et al. 2021). iTOL was also used with a dummy tree to generate heatmaps with antibiogram and PCR results for all the 21 strains.

Results and discussion

Infections caused by K. pneumoniae with multiple antibiotic-resistance genes, associated with several virulence factors are increasingly reported in hospitalized patients in Brazil and in different parts of the world. These infections have been a global concern, as the therapeutic limitations associated with the pathogenicity of many strains with MDR and XDR phenotypes are related with a large number of morbidity and mortality (Gonçalves et al., 2017; Campos et al., 2018; Lam et al., 2021; Nakamura-Silva et al. 2021b). Therefore, this study aimed to evaluate 21 K. pneumoniae strains isolated from different patients admitted to four hospitals in Manaus, located in the state of Amazonas, northern Brazil.

All strains were subjected to antimicrobial susceptibility test, with a total of forty-one antibiotics of different classes (Fig. 1), most strains were non-susceptible to the class folate pathway antagonists, sulphonamide (SUL) (95.2% n = 20); trimethoprim-sulfamethoxazole (SUT) (85.7%) and trimethoprim (TRI) (85.7%); as well nitrofurantoin (NIT) and cefuroxime (CRX) (both 90.5%); a high rate of non-susceptibility was also demonstrated for β-lactam combination agents, ticarcillin–clavulanate (TAC) and ampicillin–sulbactam (APS) (both 80.9%). Non-susceptibility rates above 70% were also observed for the antibiotics: streptomycin (EST), tobramycin (TOB), cefaclor (CFC) and ceftaroline (CTL). The results of non-susceptibility of each antibiotic are shown in Fig. 1. Similar results were described by Pereira et al. (2019) who evaluated E. coli (362 strains) and Klebsiella spp. (73 strains) isolated from urinary tract infections in southern Brazil, which demonstrated high percentage of resistance in 91.3% (n = 21/23) antimicrobials analysed: ampicillin (AMP), ampicillin–sulbactam (APS), piperacillin–tazobactam (PIT), amoxicillin–clavulanate (AMC), cefazolin (CFZ), cefotaxime (CTX), ceftazidime (CAZ), ceftriaxone (CRO), cefuroxime (CRX), cefepime (CPM), cefoxitin (CFO), aztreonam (ATM), ertapenem (ERT), meropenem (MPM), imipenem (IPM), amikacin (AMI), tobramycin (TOB), nitrofurantoin (NIT), sulfamethoxazole (SUT), tetracycline (TET), ciprofloxacin (CIP), levofloxacin (LEV). The exceptions are APS and IPM.

The majority of strains (61.9% n = 13) were classified as MDR and five strains (23.8%) were classified as XDR (Fig. 1). These high rates of MDR and XDR phenotypes have been reported in other studies carried out in Brazil. Ferreira et al. (2019) investigated the resistance profile of 25 K. pneumoniae from the Tocantins, northern Brazil, finding a high incidence of MDR among them (84%). Gonçalves et al. (2017) studied 26 K. pneumoniae strains from patients admitted in a tertiary hospital in Londrina, southern Brazil, being 53.8% classified as MDR and 26.9% as XDR and 11.5% PDR.

PCR analyses detected in eleven strains (52.4%) presence of the blaOXA-1- genes, ten with blaCTX-M-Gp1 genes (47.6%) and seven strains harboring blaKPC (33.3%). Accession numbers MT330306, MT330308 and MT330310. The XDR KpAm06 and KpAm08 encoded all the three of these β-lactamases genes, another ten strains (KpAm01, KpAm02, KpAm03, KpAm07, KpAm09, KpAm13, KpAm14, KpAm17, KpAm19 and KpAm21) harboured two of these three β-lactamases. The same genes have been reported in K. pneumoniae from similar studies in southern, southeastern and northern Brazil (Gonçalves et al. 2017; Azevedo et al. 2019; Ferreira et al. 2019).

Among the virulence genes, the fimH gene was detected in majority strains (95.2% n = 20), the second-most prevalent gene was ycfM (90.5% n = 19), followed by entB (85.7% n = 18) and mrkD (80.9% n = 17) (Fig. 1). Accession numbers of all detected virulence genes were MT330312, MT330313, MT330314, MT330316, MT330317, MT330319, MT330320, MT330322 and MT330324. Remya et al. (2019) studied virulence factors in 370 K. pneumoniae clinical strains and showed that entB was present in 90% of the strains, fimH 89.1%, ybtS 44.3%, kfu 27.8%, rmpA 5.1%, K2 2%, allS 1% and magA 0.2%. The large detection of entB and fimH; reasonable ybtS and; low rmpA and magA are similar to our results. Fimbrial adhesins (fimH, mrkD), siderophores (entB, mainly) and lipopolysaccharides (ycfM) were the most common genes in this study and similar results were found by Candan et al. (2015), with 74% ycfM and 65% of mrkD, fimH and entB, which studied 15 K. pneumoniae clinical strains isolated from different sources. The high prevalence of fimH, mrkD and entB genes found in this study was already expected, other studies have shown a high correlation of these genes in K. pneumoniae strains (Azevedo et al. 2019; Ferreira et al. 2019; Kus et al. 2017). The strain KpAm11 presented nine of the virulence genes studied, only K2 was not found in this strain (Fig. 1). Eight strains (38.1%) presented the HMV phenotype (Fig. 1), the rmpA2 (mucoid phenotype A2 regulator) and magA genes (mucoviscosity-associated A gene) were associated with HMV phenotype, however, subsequent studies showed that the magA is associated to specific capsular serotype K1 (wzy K1) (Catalán-Nájera et al. 2017). Among the eight strains positive for this phenotypic in our study, only KpAm11and KpAm24 presented magA and rmpA genes. Similar results were found by Lee et al. (2010) which found three (8%) among 35 strains that presented the phenotype without the presence of at least one of the three genes, magA, rmpA and rmpA2. Other studies also showed equivalents results, 10.3% (n = 6/58) HMV K. pneumoniae from Taiwan did not have the magA and rmpA genes (Yu et al. 2006) and among the results of Mohammed et al. (2018), one out of six clinical strains demonstrated the characteristic in question. Moreover, our findings corroborate the results recently described by our research group (Nakamura-Silva et al. 2021a) and also by Garza-Ramos (2015) who did not find these genes in K. variicola HMV strains. These results indicate that other genes may be involved in this phenotype in different species of Klebsiella which need further investigation. Walker et al. (2020) suggest that the HMV phenotype is probably due to factors not exclusive to the capsule. These authors suggest that it is necessary to investigate and clarify the link between the HMV phenotype and the capsule.

MLST analysis demonstrated a great diversity of STs among the strains, totaling 12 different STs (ST11, ST23, ST198, ST277, ST307, ST340, ST378, ST462, ST502, ST3991, ST3993 and ST5209). Three of these strains (ST11, ST23 and ST340) belong to the clonal group (CG) 258, which has been globally described as a KPC spreader and is often related to carbapenemase production. Therefore, strains with STs grouped into this CG are classified as international high-risk clones (Munoz-Price et al. 2013; Gonçalves et al. 2017; Azevedo et al. 2019). In fact, among the eight strains harbouring blaKPC gene in this study, seven belong to CG258. K. pneumoniae strains belonging to CG258 harbouring virulence and resistance genes have been described causing both hospital and community infections (Azevedo et al. 2019). ST11 was found in three strains in the study (KpAm01, KpAm02 and KpAm13), all classified as XDR, this ST is widespread in Brazil and is internationally considered a high-risk clone. The ST340 was found in three strains (KpAm06, KpAm08 and KpAm09) with XDR profile in the first two and MDR in the last, and is also known to be widespread in Brazil (Gonçalves et al. 2017) (Fig. 2).

Fig. 2

Phylogenetic relations, resistome, virulome and plasmidome of 10 Klebsiella pneumoniae isolates from clinical sources in Manaus, Brazil. aOneBR ID refers to the ID of the isolates in the One Health Brazilian Resistance Integrated Genomic database (OneBR). bMutations in quinolone resistance determining regions. cQAC quaternary ammonium compounds

In a study of the genomic population of K. pneumoniae, Wyres et al. (2020) describes the six hyper-virulent global problem clones, the ST23 found in our strains belongs to this group. The KpAm11 and KpAm24 (ST23), both isolated from tracheal secretion, have the HMV phenotype harboring several virulence genes and wzi K1, indicating that these strains are hyper-virulent; several other studies describe ST23 worldwide as a hyper-virulent clone (Cheng et al. 2015; Mukherjee et al. 2020; Pereira et al. 2017). Moreover, the ST23 was previously described in Brazil in a colistin-resistant K. pneumoniae harboring blaKPC and several other resistance genes (Boszczowski et al. 2019), Coutinho et al. (2014) reported an invasive liver abscess syndrome caused by K. pneumoniae clone ST23 and also in non-human primates (Alouatta clamitans) with hyper-virulent and hyper-mucoid characteristics (Anzai et al. 2017). The ST378 and ST462 are double-locus variants of the ST23 and appear in few studies, composing the bacterial collection among several STs (Lin et al. 2016; Zhong et al. 2014; Yan et al. 2015; Zhan et al. 2021; Saxenborn et al. 2021). KpAm16 presented ST307 and Wyres et al. (2019) emphasizes the need for more attentive epidemiological surveillance of this clone, due to its rapid and emerging global spread. This ST was also reported in other studies in southeastern Brazil as described by Dropa et al. (2016), that found K. pneumoniae harboring CTX-M-15 from wastewater, and Sartori et al. (2019) related MDR CTX-M-15 causing urinary tract infection in a dog. ST277 is a double-locus variant ST258, and was previously reported by Chmelnitsky et al. (2013) in two K. pneumoniae ST277 carrying blaKPC gene and the isolate KpAm03 presented the same characteristic. The ST198 and ST502 grouped into smaller CGs, the first one was found in the strains KpAm14, KpAm17 and KpAm21 and was previously described in Brazil in K. pneumoniae producing CTX-M-15 isolated from commercial lettuce (Lopes et al. 2017).

The ICEKp is a mobile genetic element associated with invasive infection in K. pneumoniae (Lam et al. 2018), where ICEKp12 with ybt lineage 60 were identified in the WGS analyses in the three ST198 strains. The ST502 was found in three MDR strains (KpAm12, KpAm18 and KpAm19), this ST has been described to cause infections in humans, including community infections in Brazil (Azevedo et al. 2019). Three new STs were described in this study, ST3991 (KpAm22) due to a new allele number combination, ST3993 (KpAm10), due to the new infB (177), mdh (298) and phoE (417) alleles and ST5209 (KpAm05), due the new tonB (705) allele. Through data analysis by the curators of the MLST database, the KpAm10 strain had its species re-identified as Klebsiella quasipneumoniae subsp. quasipneumoniae and was deposited in the database with this identification. Accession numbers of all WGS strains are listed in Supplementary Table 1.

Conclusion

The results presented in this study showed that more than 80% of the studied strains isolated from inpatients in the city of Manaus, located in the state of Amazonas in Brazil were resistant to several antimicrobials. The majority of strains were classified as MDR and five strains as XDR. Moreover, several virulence and antimicrobial-resistance genes were found and eight strains presented the hyper-mucoviscous phenotype. MLST analysis demonstrated a great diversity among the strains, totaling 12 different STs including high-risk international clones belonging to CG258.

In 2020, Manaus suffered a collapse in the public health system due to the COVID-19 pandemic and a large number of patients from all over the region were admitted to the city hospitals undergoing massive antimicrobial therapies. Although the number of strains evaluated in this study is relatively small and they were isolated before the pandemic period, they come from different hospitals suggesting a worrying scenario. Therefore, after the COVID-19 pandemic emergency, further studies aiming to elucidate and prevent the spread of hyper-virulent and multidrug-resistant bacteria in the hospitals of this region will be extremely important.

Data

This Whole Genome Shotgun project has been deposited at DDBJ/ENA/GenBank, and genomic information of K. pneumoniae strains are available on the OneBR platform http://onehealthbr.com/ (Table supplementary 1).

Below is the link to the electronic supplementary material.

Supplementary file1 (DOCX 492 KB)