Healthcare Facilities as Potential Reservoirs of Antimicrobial Resistant Klebsiella pneumoniae: An Emerging Concern to Public Health in Bangladesh

The emergence of virulent extended spectrum β-lactamase producing Klebsiella pneumoniae (ESBL-KP) including carbapenem-resistant Klebsiella pneumoniae (CRKP) in hospital-acquired infections has resulted in significant morbidity and mortality worldwide. We investigated the antibiotic resistance and virulence factors associated with ESBL-KP and CRKP in tertiary care hospitals in Bangladesh and explored their ability to form biofilm. A total of 67 ESBL-KP were isolated from 285 Klebsiella pneumoniae isolates from environmental and patient samples from January 2019 to April 2019. For ESBL-KP isolates, molecular typing was carried out using enterobacterial repetitive intergenic consensus polymerase chain reaction (ERIC-PCR), antibiotic susceptibility testing, PCR for virulence and drug-resistant genes, and biofilm assays were also performed. All 67 isolates were multidrug-resistant (MDR) to different antibiotics at high levels and 42 isolates were also carbapenem-resistant. The most common β-lactam resistance gene was blaCTX-M-1 (91%), followed by blaTEM (76.1%), blaSHV (68.7%), blaOXA-1 (29.9%), blaGES (14.9%), blaCTX-M-9 (11.9%), and blaCTX-M-2 (4.5%). The carbapenemase genes blaKPC (55.2%), blaIMP (28.4%), blaVIM (14.9%), blaNDM-1 (13.4%), and blaOXA-48 (10.4%) and virulence-associated genes such as fimH (71.6%), ugeF (58.2%), wabG (56.7%), ureA (47.8%) and kfuBC (28.4%) were also detected. About 96.2% of the environmental and 100% of the patient isolates were able to form biofilms. ERIC-PCR-based genotyping and hierarchical clustering of K. pneumoniae isolates revealed an association between environmental and patient samples, indicating clonal association with possible transmission of antimicrobial resistance genes. Our findings can help in improving patient care and infection control, and the development of public health policies related to hospital-acquired infections.


Introduction
Klebsiella pneumoniae is linked to community and hospital-acquired infections and shows high resistance to antibiotics. The present disease load in Bangladesh is unknown; however, it is higher in susceptible individuals, including newborns and the elderly. The the various origin, and it is considered one of the most efficient, comparatively easy, and cost-effective techniques [28,32]. This also provides an excellent basis for understanding the molecular epidemiology of K. pneumoniae, particularly in resource-limited countries.
In this investigation, we looked for virulent genes in K. pneumoniae that were ESBL positive (ESBL-KP) and resistant to carbapenem antibiotics (CRKP) in both patients and hospital surrounding (environmental) samples in three Bangladeshi tertiary care hospitals. These ESBL-KP and CRKP isolates were also analyzed for their biofilm-forming properties and their association between antibiotic resistance and biofilm-forming capacity.
In our study, we observed 23.5% (67/285) ESBL-KP isolates in all three studied hospitals. Among the 67 ESBL-KP isolates, 52 were isolated from the surrounding environment and 15 were isolated from patients as carriage samples. The distribution of CRKP in our studied hospitals was diverse. Using microbiological culture 90% (9/10) of the isolates from Rangpur medical college hospital; 85.7% (18/21) of the isolates from Rajshahi medical college hospital and 41.7% (15/36) of the isolates from Faridpur medical college hospital were resistant to carbapenems. The distribution of ESBL-KP and CRKP among different sampling points of various hospitals was illustrated in Table 1.

Figure 1.
Antibiotic resistance patterns of the K. pneumoniae isolates. All the isolates were tested against 19 antibiotics from 14 different therapeutic classes. All the isolates were classified as multi Figure 1. Antibiotic resistance patterns of the K. pneumoniae isolates. All the isolates were tested against 19 antibiotics from 14 different therapeutic classes. All the isolates were classified as multi drug-resistant MDR and extensively drug-resistant (XDR) but none of them as pan drugresistant (PDR).

Detection of Virulence and Resistance Genes
The distribution of five virulence genes is shown in Figure 2. Of the 67 isolates, 62 harbored at least one virulence gene, whereas five isolates carried all five tested virulent genes. The virulence and virulence-associated genes fimH, wabG, ugeF, ureA, and kfubc were commonly present in both environmental and patient isolates and 71.6% (48/67) fimH, 56  PCR was carried out for all the K. pneumoniae isolates to detect ESBL and carbapenemase genes. Three or four broad-spectrum β-lactamase genes (blaTEM, blaSHV, blaOXA-1, blaCTX-M-1, blaCTX-M-2 or blaCTX-M-9) were frequently found in the isolates. Among the β-lactamase genes, the most frequent one encountered was blaCTX-M-1, which was found in 91% (61/67) of the study isolates, followed by blaTEM in 76.1% (51/67), blaSHV in 68.7% (46/67)     including carbapenemase (blaVIM, blaIMP, and blaKPC and blaOXA-48) genes among the K. pneumoniae isolates. Three or four broad-spectrum β-lactamase genes were frequently found in the isolates but the distribution of carbapenemase genes varied among patient and environmental isolates.  The carbapenemase genes indicated their distribution varied among patient and environmental isolates. blaKPC-, blaIMP-, blaNDM-1-, and blaOXA-48-like genes were more prevalent in patient isolates than the environmental isolates ( Figure 5). Surprisingly, no patient isolates possessed carbapenemase gene blaVIM, present in 19.2% (10/52) of the environmental isolates. Table S2 has a detailed description of the β-lactamases including carbapenemases identified in the 67 KP isolates.

Biofilm Forming Ability of the ESBL-KP and CRKP Isolates
The ability to produce biofilms was temperature-dependent for both the environmental and patient isolates. After 48 h of incubation, the biofilm production increased at 25 °C and reduced at 37 °C. Environmental isolates produced less biofilm than the patient isolates at both temperatures. Half of the environmental isolates showed moderate biofilm production at 25 °C, whereas it was 66.67% for patient isolates. The number of isolates for different categories of biofilm producers can be seen in Figure 5. The carbapenemase genes indicated their distribution varied among patient and environmental isolates. bla KPC -, bla IMP -, bla NDM-1 -, and bla OXA-48 -like genes were more prevalent in patient isolates than the environmental isolates ( Figure 5). Surprisingly, no patient isolates possessed carbapenemase gene bla VIM , present in 19.2% (10/52) of the environmental isolates. Table S2 has a detailed description of the β-lactamases including carbapenemases identified in the 67 KP isolates.

Molecular Typing of the Isolates
Enterobacterial Repetitive Intergenic Consensus Polymerase Chain Reaction (ERIC-PCR) was used to explore the relationship between the carbapenemase and ESBL-producing pathogenic K. pneumoniae. Figure 6 depicts the genetic profiling of isolates using ERIC-

Biofilm Forming Ability of the ESBL-KP and CRKP Isolates
The ability to produce biofilms was temperature-dependent for both the environmental and patient isolates. After 48 h of incubation, the biofilm production increased at 25 • C and reduced at 37 • C. Environmental isolates produced less biofilm than the patient isolates at both temperatures. Half of the environmental isolates showed moderate biofilm production at 25 • C, whereas it was 66.67% for patient isolates. The number of isolates for different categories of biofilm producers can be seen in Figure 5.

Molecular Typing of the Isolates
Enterobacterial Repetitive Intergenic Consensus Polymerase Chain Reaction (ERIC-PCR) was used to explore the relationship between the carbapenemase and ESBL-producing pathogenic K. pneumoniae. Figure 6 depicts the genetic profiling of isolates using ERIC-PCR fingerprinting. According to the dendrogram, ERIC analyses revealed 43 distinct patterns (P1-P43) of K. pneumoniae isolates with similarity >80%. The isolates produced 4-19 amplicons ranging from~140 to~1300 bp, where~300,~410 and~680 bp were common among the isolates. We found that the maximum number of four isolates belonged to both P19 and P43 patterns, suggesting their genetic similarity, and a single isolate in each of 28 different patterns. The isolates (KP-119 and KP-120) obtained from the pediatric floor in the same hospital showed the highest similarity and identical antibiotic resistance gene profiles, suggesting that those isolates are from the same clonal lineage (Figures 4 and 6). Likewise, two isolates (KP-256 and KP-257) obtained from the nasal-throat of the neonatal ward of the same hospital also showed similar genetic profiles.

Associations between Sample Source and Phenotypic and Genotypic Traits
Hierarchical clustering (along with heatmap) (Figure 4), correlation matrix analysis (Figure 7), and principal component analysis ( Figure 8A,B) were used to determine associations between the phenotypic and genotypic traits and the source of the isolates. The correlation coefficient and p-values obtained from correlation analysis showed very few positive relationships with the presence of β-lactamase genes and resistance to β-lactams. However, resistance to cefepime significantly correlates with the presence of the bla TEM and bla CTX-M-1 genes ( Figure 8, p < 0.05, r = 0.31 and 0.25, respectively). In the case of the carbapenem group of antibiotics, resistance to imipenem and meropenem were significantly correlated with the presence of carbapenemase genes-bla KPC , bla IMP , bla NDM-1 , and bla OXA-48 like genes ( Figure 8, p < 0.05).
Significant positive correlations of phenotypic antibiotic resistance indicated that cooccurrence of resistance was prevalent ( Figure 8, p < 0.05). For instance, resistance to the fluoroquinolone group antibiotic, ciprofloxacin, was positively correlated with chloramphenicol, amikacin, fosfomycin, azithromycin, nitrofurantoin, nalidixic acid, and tigecycline ( Figure 8, p < 0.05). Similarly, resistance to fosfomycin was positively correlated with resistance to the combination drug sulfamethoxazole-trimethoprim, amikacin, azithromycin, nitrofurantoin, nalidixic acid, ciprofloxacin and tigecycline ( Figure 8, p < 0.05). Alongside the co-occurrence of resistance, we also found the coexistence of virulence genes and antibiotic resistance in both ESBL-KP and CRKP isolates. The presence of virulence genes was also examined where a significant positive correlation between type 1 fimbriae adhesive subunit FimH and the resistance to cefepime (Figure 8, p < 0.05, r = 0.28) was observed. Likewise, the presence of ureA gene was positively correlated with the resistance to amikacin, and fosfomycin ( Figure 8, p < 0.05). Among the virulence genes, the presence of wabG gene was positively correlated with the resistance to the highest numbers of antibiotics, namely tigecycline, nalidixic acid, ciprofloxacin, sulfamethoxazole-trimethoprim and fosfomycin ( Figure 8, p < 0.05). . ERIC-PCR molecular fingerprint profiles of K. pneumoniae isolates from patients and hospital environment along with their representative ID, geological location, sampling point, and location. In the dendrogram, 43 distinct patterns (P1-P43) of K. pneumoniae isolates with similarity > 80% were observed. The isolates produced 4-19 amplicons ranging from ~140 to ~1300 bp, where ~300, ~410, and ~680 bp were common among the isolates. . ERIC-PCR molecular fingerprint profiles of K. pneumoniae isolates from patients and hospital environment along with their representative ID, geological location, sampling point, and location. In the dendrogram, 43 distinct patterns (P1-P43) of K. pneumoniae isolates with similarity > 80% were observed. The isolates produced 4-19 amplicons ranging from~140 to~1300 bp, where~300,~410, and~680 bp were common among the isolates. correlation coefficient and p-values obtained from correlation analysis showed very few positive relationships with the presence of β-lactamase genes and resistance to β-lactams. However, resistance to cefepime significantly correlates with the presence of the blaTEM and blaCTX-M-1 genes (Figure 8, p < 0.05, r = 0.31 and 0.25, respectively). In the case of the carbapenem group of antibiotics, resistance to imipenem and meropenem were significantly correlated with the presence of carbapenemase genes-blaKPC, blaIMP, blaNDM-1, and blaOXA-48 like genes (Figure 8, p < 0.05).  Significant positive correlations of phenotypic antibiotic resistance indicated that cooccurrence of resistance was prevalent (Figure 8, p < 0.05). For instance, resistance to the fluoroquinolone group antibiotic, ciprofloxacin, was positively correlated with chloramphenicol, amikacin, fosfomycin, azithromycin, nitrofurantoin, nalidixic acid, and tigecy- There was no significant correlation between antimicrobial resistance and biofilm formation in ESBL-KP and CRKP isolates. However, we observed an association between biofilm formation and resistance to some specific antibiotics. All the isolates resistant to imipenem, meropenem, and amikacin showed the ability to form biofilm at different levels. In imipenem-resistant isolates, 56.5%, 39.1%, and 4.3% were categorized as weak, moderate, and strong biofilm formers, respectively. A similar percentage of biofilm formers was also found for meropenem-resistant isolates ( Table 2). For amikacin, 66.7% of the resistant isolates were moderate biofilm formers and the rest were weak biofilm formers. In the same way, higher percentages of tetracycline-, ciprofloxacin-, and fosfomycin-resistant isolates showed weak to moderate biofilm-forming ability. Similar results were also observed in the case of other antibiotics including the β-lactams (Table 2). The principal component analyses showed similar relations of β-lactam antibiotic cefepime with the β-lactamase genes bla TEM and association of imipenem with the bla OXA-48like carbapenemase gene ( Figure 8A). Moreover, the isolates that showed resistance to cefepime were also resistant to fosfomycin, azithromycin, aztreonam, and cefixime. Isolates that harbored bla CTX-M-2 and bla CTX-M-9 β-lactamase genes did not show resistance to cefepime. The presence of β-lactamase genes of bla TEM and bla CTX-M-1 were also visualized to be in association with ciprofloxacin and nalidixic acid resistance ( Figure 8A).
Hierarchical clustering analysis allowed the segregation of the 67 K. pneumoniae isolates according to their phenotypic (antibiotic resistance profiles) and genotypic (resistance gene profiles) traits ( Figure 4). Interestingly, a separation of K. pneumoniae isolated from environmental sources (Figure 4, Cluster B) and patient sources ( Figure 5, Cluster D) was observed. Hierarchical clustering segregated the isolates into five main clusters. Cluster A included 9 K. pneumoniae isolates from the environment samples (eight of them from the floor and the remaining one from the bed pillow) and four from patient origins (all of them from the nasal-throat). Most of the isolates of cluster A were from Rangpur and Faridpur hospitals. Three isolates of Rangpur hospital (KP-118, KP-119, and KP-120) also clustered together in ERIC PCR. Of interest in this cluster, an environmental isolate (KP-176, floor) clustered together with three patient isolates (KP-257, KP-255, and KP-256; nasal-throat) of the same hospital and the clustering was visible in PCA as well ( Figure 8B). KP-257 and KP-256 clustered together in ERIC PCR also. Similarly, the clustering of patient isolate KP-126 with the environmental isolates KP-118, KP-120, KP-119, and KP-122 of Rajshahi hospital was also reflected in PCA ( Figure 8B). Here, one of the large clusters, cluster B, consisted of 16 isolates from the environment (11 from the bed pillow and 5 from the floor) and two isolates from patient sources (nasal-throat). Among them, 12 isolates from Rajshahi hospital were clustered together. Corresponding to cluster A, here patient isolate KP-59 clusters together with the environment isolates (KP-30, KP-31, KP-32, KP-33, KP-34, KP-35, and KP-37) of the same hospital in PCA ( Figure 8B). Cluster C in Figure 4 was relatively smaller and confined, with seven environmental isolates (four from the floor and three from the caregiver's hand) and two patient isolates (nasal-throat) mostly from Faridpur hospital. Another large cluster (Figure 4, Cluster D) contained 12 K. pneumoniae isolates from the environment (seven from the floor, three from the bed pillow, and two from the caregiver's hand) and six isolates from patient samples (five from the nasal-throat and the remaining from pus sample). In this cluster, isolates from Faridpur and Rajshahi hospitals were clustered together. Remarkably, a patient isolate (KP-184) and an environmental isolate from a caregiver's hand (KP-154) from the same hospital were clustered together, which was also observed in PCA. While being from different sources in the same hospital, these two isolates demonstrated the same phenotype and genotype. Likewise, to cluster A and B in Figure 4, patient isolate KP-48 clustered together with the environment isolates (KP-92, KP-93, KP-94, and KP-96) of the same hospital which is also reflected in PCA. The remaining smaller cluster (Figure 4, Cluster E) had eight isolates from the environment (four from the floor, three from the bed pillow, and one from the caregiver's hand) and one from the patient origin (nasal-throat) and all of them from the same hospital. Notably, in cluster E, the clustering of patient isolate KP-260 and environmental isolate KP-248 was also reproduced in PCA ( Figure 8B).

Discussion
K. pneumoniae is responsible for various multidrug-resistant infections, and any transmission between the environment and patients has yet to be proven [34,35] in Low to Middle Income Countries (LMICs). The current study assessed the molecular characterization of antibiotic resistance genes, biofilm-forming capacity, and virulence factors associated with ESBL-KP and CRKP isolated from hospitalized patients and hospital environments. In a recent study in Bangladesh, 30% of the 3G cephalosporin-resistant K. pneumoniae present in clinical samples among hospitalized patients was identified as carbapenemresistant [36]. Similar results were also reported in other studies for CRKP and other important hospital-acquired pathogens [37,38], which are in agreement with our findings.
The inappropriate use of antimicrobials in hospitals leads to the high prevalence of resistance in K. pneumoniae [39]. Multidrug-resistant, ESBL-KP are major causes of hospitalacquired infections which are difficult to treat, particularly in LMICs [34,35]. All the K. pneumoniae isolates in our study showed resistance to ceftriaxone, cefuroxime, and cefotaxime. Cephalosporins and penicillins are widely used antibiotics in Bangladesh [40] and all of our tested isolates were resistant to these antibiotics. A high prevalence of resistance was also observed against cefixime, aztreonam, cefepime, ceftazidime, trimethoprimsulfamethoxazole, azithromycin, tetracycline, nitrofurantoin, and nalidixic acid. These are in line with the prevalence of resistance observed globally among hospital-acquired infections in Nepal, China, and Pakistan [41][42][43]. In our study, all the ESBL-KP isolates were MDR showed high resistance to the major antibiotics either alone or in combination, such as β-lactams (including 3G and 4G cephalosporins, monobactams, carbapenems), aminoglycosides, quinolones, and glycylcycline. There is limited information available for such a high frequency of MDR K. pneumoniae from patient samples and hospital environmental samples in Bangladesh, although a high incidence of MDR K. pneumoniae has been observed before from clinical samples around the world [34,41]. In the current study, 30.8% and 23.1% of the environmental isolates were imipenem and meropenem resistant respectively, for patient isolates, the resistance patterns peaked at 46.7%, which is an alarming rise in the carbapenem resistance scenario among clinical isolates of Bangladesh compared to the past few years [44][45][46]. The evading mechanisms of K. pneumoniae against carbapenem, are very crucial to investigate. In the present study, we examined the prevalence of ESBL genes (bla CTX-M-1 , bla CTX-M-2 and bla CTX-M-9 , bla TEM , bla SHV , bla GES , and bla OXA-1 ) and carbapenemase genes (bla IMP , bla KPC , bla VIM , bla NDM-1 , and bla OXA-48 like genes) in K. pneumoniae from patient and environmental isolates. Among the ESBL-genes, bla CTX-M , bla TEM , and bla SHV are the primary genes associated with most of the K. pneumoniae infections [47,48], which corresponds to the findings of our research. In most countries, the bla CTX-M type of ESBL is the most prevalent, surpassing that of bla SHV and bla TEM ESBLs [49,50]. We also found bla CTX-M-1 to be the most frequent gene in the ESBL-positive K. pneumoniae isolates, followed by bla TEM and bla SHV , which confer resistance to fluoroquinolone, aminoglycoside, and β-lactams [51]. Our result is a bit higher than the previous study with bla CTX-M (51.4%), bla TEM (49.4%), and bla SHV (26.8%) in K. pneumoniae originated from clinical isolates in Bangladesh [45]. Carbapenems are the most commonly used antibiotics to treat severe infections caused by Enterobacterales that produce ESBLs [51,52]. The prevalence of bla KPC in our study was 55.2%; with 88% of the CRKP isolates containing this gene, which is higher compared to the rates previously described in clinical samples from Bangladesh [44,53]. We also found that 14.9% of the isolates contained the bla GES gene class, which behaves like carbapenemase or those that hydrolyze the extended-spectrum beta-lactam class of antibiotics. As we did not sequence these genes, we were unable to precisely identify the blaGES variants, some of which behave like carbapenemase (GES-2, -4, -5, -6) or those that hydrolyze the extended-spectrum beta-lactam class of antibiotics such as cephamycins and cephalosporins (GES-1, -10, -11, -12, -13) (Lopes et. al, unpublished data). Apart from the class A carbapenemases, KPC and GES, class B metallo-β-lactamases (MBLs) such as IMP, VIM, NDM, and class D carbapenem-hydrolyzing oxacillinase-48 (OXA-48) have also been the most often discovered carbapenemases in K. pneumoniae worldwide [20]. We examined a number of carbapenemase genes in our study and found a diverse prevalence of genes IMP, VIM, NDM-1, and OXA-48 [44], which is the first description of these genes found in hospital environmental samples in Bangladesh.
The findings of the virulence genes suggest that the fimH, ureA, ugeF, kfuBC, and wabG genes were frequently disseminated among patient and hospital environmental isolates [54,55]. The presence of these genes in hospital-associated environments could indicate that these isolates are pathogenic and provide a possible hospital-acquired risk to patients. The gene fimH encodes adhesins of type 1 fimbriae, which is responsible for attachment to the extracellular matrix, promotes colonization and biofilm formation, invasion, and pathogenicity [54][55][56], upon testing we found that 71.6% of the MDR K. pneumoniae contained this gene, indicating their potential ability to form biofilms. In addition to fimH, the gene responsible for the production of lipopolysaccharide (LPS), wabG, and ugeF genes were also present in more than half (56.7% and 58.2%) of the isolates, respectively [57,58]. The importance of these genes has been studied as their presence in K. pneumoniae indicates the strains are able to cause diseases such as urinary tract infection, pneumonia, and sepsis [59,60]. In our study, nearly half (47.8%) of strains contained the ureA gene, which helps in gastrointestinal (GI) tract colonization and has been associated with K. pneumoniae hospital-acquired infections [20]. We also found the kfuBC gene in 28.4% of the studied isolates, which has previously been found to be more abundant among invasive clinical strains such as those that arise from liver abscesses and are responsible for meningitis or endophthalmitis [4,61], explains the lower prevalence among environmental and patient strains.
Klebsiella spp. is known to form biofilm on biotic (e.g., epithelial surfaces) and abiotic surfaces, that is, on plastic, used in indwelling medical devices such as catheters and endotracheal tubing [62]. In the current study, we found that patient isolates were more likely to be weak to moderate biofilm formers than the environmental isolates, previously clinical samples have been described as producing high percentages of biofilm [63,64]. In our study, biofilm formation increased at 25 • C for both patient and environmental isolates. We postulate that this may be due to stress caused by the suboptimal temperature [64].
Interestingly, though no clear associations were observed between source types (patient or environmental) and antibiotic resistance or gene profile-K. pneumoniae from patient origins showed more resistance to β-lactams including carbapenems, quinolones, macrolides, and nitrofurantoin compared to environmental samples. The coexistence of virulence and antibiotic-resistant genes as revealed from our correlation analysis is in agreement with previous studies [65,66]. We found positive associations between biofilm production and specific antibiotics resistance, including imipenem, meropenem, amikacin, ciprofloxacin, tetracycline, and fosfomycin. Similar findings were also reported previously with associations between resistance to piperacillin-tazobactam and colistin correlating with biofilm formation in K. pneumoniae [67].
The association between ERIC-PCR genotype and antibiotic resistance of K. pneumoniae has been described previously from hospital isolates [34]. Because of variances in nucleotide sequences, pathogenic K. pneumoniae is very diverse [68]. For instance, isolates screened from the same sampling point and hospital, KP-118, KP-119, and KP-120, grouped in cluster A of hierarchical clustering, also represented similar genotypic patterns in ERIC. Similarly, KP-255-257; KP-31, KP-33, KP-35, KP-37, KP-92, and KP-94 also followed the trend with similar antibiotic-resistant determinants and genotypic patterns from the same hospital.
One of the limitations of the study was that we did not have the clinical information of the patients. Clinical information would have shed some light to understand the actual scenario of K. pneumoniae infections in the healthcare facilities of Bangladesh. Additional genes examining the efflux pump mediated resistant genes and capsule and type III encoding virulence genes would also have provided more discrimination and newer methods such as whole genome sequencing would have provided more insight into resistance and pathogenicity at the genome level.

Sample Collection and Isolation of K. pneumoniae
Over the course of four months, January 2019 to April 2019, samples were collected from three tertiary care medical college hospitals in Bangladesh: Faridpur, Rajshahi, and Rangpur. The geographical locations of these hospitals are shown in Figure 9. Samples were taken from several wards within these hospitals, including adult male and female medicine, surgery, pediatrics, and neonatal wards. All swabs were transported to the Laboratory of Environmental Health, International Centre for Diarrhoeal Disease Research, Bangladesh (icddr,b), Mohakhali, Dhaka for all laboratory work. A convenient sampling procedure was performed on admitted patients and their immediate surroundings in the hospital to investigate the presence of antibiotic-resistant K. pneumoniae. The Samples were taken from several wards within these hospitals, including adult male and female medicine, surgery, pediatrics, and neonatal wards. All swabs were transported to the Laboratory of Environmental Health, International Centre for Diarrhoeal Disease Research, Bangladesh (icddr,b), Mohakhali, Dhaka for all laboratory work. A convenient sampling procedure was performed on admitted patients and their immediate surroundings in the hospital to investigate the presence of antibiotic-resistant K. pneumoniae. The patient samples, including pus and nasal-throat swabs, were collected from convenient patients in the specified wards of the hospitals, together with the swabs from corresponding surfaces of the patient's bed pillow, bedside rails, caregivers' hands, and the floor (per 100 cm 2 ) of that ward.
A sterile cotton swab was saturated with physiological saline, dabbed over the surface, and then transported to the icddr,b laboratory within 8-12 hours of collection in phosphatebuffered saline (PBS) while maintaining the cold chain (4-10 • C). A vortex mixer was used to homogenize the PBS containing swabs. The homogenized liquid was then spread onto a MacConkey Agar (BD-Difco, New Jersey, USA) plate and incubated at 37 • C for 24 h. Gram staining and indole testing were performed on suspected K. pneumoniae colonies on MacConkey agar, and the API 20E system was used to confirm the results (bioMerieux, Marcy-l'Étoile, France). Positive isolates were chosen for PCR using species-specific 16S rRNA primers [70]. Cultures were stocked in 30% glycerol-Lysogeny Broth (BD-Difco, New Jersey, USA) and stored at −80 • C for further investigation.

Molecular Typing of K. pneumoniae
ERIC-PCR was performed for molecular typing using the primer ERIC2 (5 -AAGTAA GTGACTGGGGTGAGCG-3 ). 2 µL of DNA template were mixed with 12.5 µL of DreamTaq TM Green PCR Master mix (Thermo Scientific, Vilnius, Lithuania), 1 µL of primer (10 pmol), and an adequate volume of sterile nuclease-free water to make a 25 µL reaction mixture. The PCR was carried out in a Bio-Rad T100™ Thermal Cycler (Bio-Rad, Hercules, CA, USA), with the following protocol: initial denaturation at 94 • C for 15 min, followed by 40 cycles of amplification. Denaturation at 95 • C for 1 min, primer annealing at 37 • C for 1 min, and primer extension at 72 • C for 1 min were used in each cycle. After the amplification cycles, samples were kept at 72 • C for 10 min to allow partially produced DNA to extend [78]. Then, 2% w/v agarose gel electrophoresis was used to visualize ERIC fragments, and the data were evaluated using GelJ v.2.0 software (Open source Java software) [83,85]. Invitrogen 1 kb plus ladder (Thermo Fisher Scientific, Waltham, MA, USA) was used to normalize the fragment patterns in GelJ v.2.0 software [27].

Biofilm Formation
Biofilm formation was detected using a quantitative adhesion method [86,87]. Each isolate was grown overnight in LB (Luria-Bertani) at 37 • C. Following that, 2 µL of bacterial suspension was injected into a sterile 96-well microtiter plate (Corning Life Sciences, Kennebunk, ME, USA) containing 198 µL of LB. In each run, LB broth was added as a negative control. It was then incubated for 48 h, statically at 25 • C and 37 • C. Each well was gently rinsed three times with 200 µL PBS to avoid disturbing the biofilm, then dried inverted at room temperature. The resultant biofilm mass was dyed with 50 µL of 0.1% crystal violet for 15 min. Again, the microtiter plate was cleansed three times with 200 µL of PBS and dried. Finally, the stained biofilm mass was solubilized by dissolving the wells in 200 µL of 5% isopropanol. A microplate reader set to 590 nm was used to measure optical density (OD). The mean OD was calculated after three separate experiments with triplicate of each isolate and negative control. The average OD of negative controls + 3 × SD (standard deviation) of negative controls was used to designate the OD cut-off (ODc). Isolate OD ≤ ODc was classified as a non-biofilm producer. On the other hand, the isolate was grouped as biofilm producers, with weak biofilm producers if ODc < OD ≤ 2 × ODc; moderate biofilm producers if 2 × ODc < OD ≤ 4 × ODc; and strong biofilm producers if OD > 4 × ODc [86,87].

Statistical Analysis
For statistical assessment, antibiotic resistance summaries along with harboring/lack of virulent and resistance genes were transformed to binary coding. Antibiotic sensitivity and resistance were denoted as 0 and 1, respectively. The harboring and lack of a given gene (for example, bla TEM ) were likewise represented as 1 and 0, respectively. The open statistical program R (V: 4.1.1) was used to conduct the statistical analysis [7]. The 'corr' function was used to generate correlations for binary variables, and the 'corr.test' function was used to determine the significance using Pearson correlations and the (r) Phi coefficient. The 'corrplot' tool was used to illustrate significant correlations [36] and a heatmap, as well as hierarchical clustering, was prepared [88]. Ward's approach was used to create distance matrices for hierarchical clustering [89]. For principal component analysis, the R packages 'FactoMineR' and 'factoextra' were utilized (PCA) [30,31].

Conclusions
Our study points out some worrisome findings in which K. pneumoniae present throughout the hospital environment is resistant to frequently used antibiotics to treat infections as well as last-resort medications for life-threatening illnesses. Additionally, our findings showed the presence of genetic determinants responsible for virulence and antibiotic resistance among the isolates, which might be transferable via conjugation and transposable genetic elements within the hospital environment. This is the first study assessing the presence of virulent and multidrug-resistant K. pneumoniae among both environmental and carriage samples in the hospital settings of Bangladesh. Most of the studied isolates produced biofilms when exposed to laboratory-induced stress, a feature that would be important in hospital-acquired infections originating from hospital surroundings and medical devices. The outcomes of this study also indicate an association between the presence of virulence genes, antibiotic resistance, and biofilm-forming capacity in K. pneumoniae. As a result, our findings should serve as a warning about the necessity for MDR K. pneumoniae prevention and control in Bangladeshi hospitals, and the need for rigorous infection prevention and control practices.