Molecular characterization of Carbapenem-resistant Escherichia coli isolates from sewage at Mulago National Referral Hospital, Kampala: a cross-sectional study

Escherichia coli (E. coli) is one of the most frequent causes of fatal bacterial infections affecting both humans and animals. The resistance to Carbapenems is mainly associated with enzyme-mediated resistance mechanism, through the acquisition of Carbapenemase genes. In Uganda, no studies have been done to detect presence of Carbapenem-resistant E. coli in sewage. We therefore carried out a study to characterize Carbapenem-resistant E. coli from sewage from Mulago National Referral Hospital. In this cross-sectional study, a total of 104, sewage samples were aseptically collected, cultured on MacConkey agar supplemented with Meropenem 1 µg/ml with other standard microbiology methods to screen for Carbapenem-resistant E. coli (CREC). Antimicrobial susceptibility testing was performed on the CREC, using Imipenem (10 mg/disc) and Meropenem (10 mg/disc), Carbapenem drugs readily available on market. Multiplex PCR was performed on selected Carbapenem-resistant and susceptible isolates to detect Carbapenemase genes. Later the isolates were pathotyped for virulence genes that included pathogenicity islands (PAIs) and phylogenetic markers. The results showed that the Carbapenem-resistant E. coli isolates were more resistant to Meropenem (64%) than Imipenem (60%). KPC gene was the most predominant (75%), followed by NDM gene (30%) while no OXA-48, IMP-1, and IMP-2 genes were detected. Pathotyping of virulence genes showed presence of eae gene, as the most predominant (40%), followed by elt gene (25%) and negative for stx and aggR genes. For PAI markers, only the PAI IV536 gene was detected at 10%. Then, pathotyping of the phylogenetic markers was present in 85% of the typed isolates with yjaA gene the most abundant (60%) while both chuA and TSPE4.C2 were detected in 5% of the isolates. Both pathogenic and non-pathogenic Carbapenem-resistant E. coli strains are present in the sewage of Mulago National Referral Hospital in Uganda.

Accumulation of multidrug resistance traits may correspond to the ultimate pan drug resistance of Enterobacteriaceae (Basak et al. 2016;Poirel et al. 2016) as an advent of untreatable infections.Beta-lactams are so far the most used antibiotics around the world and these include the Penicillins, Cephalosporins, Monobactams, and Carbapenems.They are distinct due to the common beta-lactam ring and act similarly by binding to and inactivating the Penicillin-binding proteins (PBPs), which are required in the formation of the bacterial cell wall (Khalil et al. 2017;Kong et al. 2010).Among the beta-lactams, Gram-positive and Gram-negative bacteria are most susceptible to Carbapenems; hence, it is a broad-spectrum antibiotic with a unique molecular structure that combine a Carbapenem and the Beta-lactam ring.The Carbapenems are regarded as the most reliable last-line treatment for bacterial infections; hence, the emergency and rapid spread of Carbapenem resistance among Gram negative bacteria is a major problem (Basak et al. 2016;Meletis 2016).Resistance to Carbapenem among Gramnegative bacteria in general can be acquired through several mechanisms: generation of new extended-spectrum beta-lactamases (ESBL) by amino acid substitution from available plasmid-mediated beta-lactamases, acquisition of genes encoding ESBL from the environmental bacteria, dissemination of plasmid mediated Carbapenemases, increased expression of chromosome-encoded beta-lactamases genes, and or ability to develop biofilm (Thakur et al. 2016).Infections due to these resistant strains are associated with higher morbidity and mortality rates.Therefore, rapid detection of these resistant E. coli strains is crucial for appropriate antimicrobial therapy and infection control measures (Birgy et al. 2012), especially with regard to the more recent characterization of Carbapenem-resistant Enterobacteriaceae from patients admitted at Mulago National Referral Hospital, Uganda (Okoche et al. 2015).
It was reported that Carbapenem-resistance prevalence is comparatively high in isolates obtained from the hospital rendering the hospital environment a potential source of infection for patients and health workers (Kateete et al. 2016).Therefore, the beta-lactam antibiotics which were the main treatment of Enterobacteriaceae infections, nonetheless have been rendered ineffective; for example, the percentage of E. coli ESBL producers have gradually increased from 18% in 2010 to 24% in 2014 and reached 31.7% in 2017 (January-June) as reported by (Eltai et al. 2018).Beta (β)-lactamases are divided into four functional groups: Penicillinases, ESBLs, Carbapenemases, and AmpC-type Cephalosporinases (Bush and Jacoby 2010).ESBLs are divided into three groups according to the encoding of TEM, SHV and CTX-M genes (Eltai et al. 2018;Lahlaoui et al. 2014).CTX-M enzymes are the most common and are further classified into five major phylogenetic groups based on gene sequences namely, CTX-M-1, CTX-M-2, CTX-M-8, CTX-M-9, and CTX-M-25 (Lahlaoui et al. 2014).ESBL-producing E. coli are the predominant organisms in human infections, and they pose significant threat to human health (Bakshi et al. 2013).Since the Carbapenem-resistant Enterobacteriaceae are resistant to all beta-lactams, they were also found to be resistant to Ciprofloxacin, Gentamycine, Co-trimoxazole, and Tetracycline (Cusack et al. 2019).Thus, it is important to monitor the use of Carbapenem combinations in treatment of infections and institute active surveillance to identify Carbapenem-resistant E. coli pathotypes.Then, promote control practices that restrict its further dissemination.This study, therefore, was aimed at characterizing the Carbapenem-resistant E. coli, isolated from sewage within Mulago National Referral Hospital environment.

Study design and study area
This was designed as a cross-sectional study to identify carbapenem-resistant E. coli isolated from different sewage sites within Mulago National Referral Hospital that serves as a general hospital for Kampala metropolitan, receives nationwide patients with different health complications and severities requiring antibiotic treatment as well as those presenting Carbapenem resistance.Six manholes were selected for the sampling and these included microbiology lab manhole, 1C ward manhole, Causality area manhole, hallway manhole, the hospital main manhole and mortuary manhole.Both effluent and swab-samples were collected on a 2-weekly basis from the different manholes and later transported under cool temperatures to the microbiology laboratory located at College of Veterinary Medicine, Animal Resources and Biosecurity (COVAB) for culture and isolation.After which the E. coli isolates were tested for susceptibility to two selected Carbapenems (Meropenem and Imipenem) that were readily available on the Ugandan market.Later on, selected E. coli isolates, were transferred to the Molecular laboratory at COVAB, where their DNA was extracted and amplified to detect for the presence of Carbapenemase genes; bla IMP-1 , bla IMP-2 , bla OXA-48 , bla VIM , bla KPC , and bla NDM .Following Carbapenemase detection, pathotyping of E. coli was performed to detect for the presence of virulent genes: eae, stx, est, elt paH, and aggR; the presence of PAI markers: PAI I 536 , PAI II 536 , PAI IV 536 , PAI I J96 , PAI II J96 , PAI I CFTO73 , and PAI II CFT073 ; as well as the presence of chuA, yjaA, and the DNA fragment TSPE4 C2.

Sample size calculation
Sample size was calculated using the Cochran formula to obtain an ideal sample size given a desired level of precision, desired confidence level, and the estimated proportion of the attribute present in the sample population.

Sampling and sample collection
Sewage samples were collected purposively within and around Mulago National Referral Hospital; on a twoweekly basis, collecting from all the sites respective to the initial collection day.Both effluent and swab samples were collected from six manholes which include microbiology lab manhole, 1C ward manhole, causality area manhole, hallway manhole, hospital main manhole, and mortuary manhole.To collect effluent samples from the different manholes, sterile falcon tubes were filled allowing a small headspace of 1 ml as a provision for mixing the sample and recapped immediately.The samples were labeled with the correct identification, placed in a biohazard bag, and then placed into a cool box with ice packs.For collection of the swabs from the different manholes, a sterile swab was used to rub and roll firmly several times across the sampling area.Each swab was afterwards placed into a falcon tube containing Stuart's transport media (HiMedia, Uk) and capped tightly then labelled with its respective identification, placed in a biohazard bag, and then put in a cool box with ice packs.Samples were transported in a cool box with ice to maintain the integrity of the samples under a cold chain.

Culture and isolation of E. coli
The samples were inoculated immediately on the pre-set sterile MacConkey agar (Condalab, Spain) plates supplemented with Meropenem 1 µg/ml (Oxoid ™ , UK) used as the selective media for screening Carbapenem-resistant E. coli (Kumar et al., 2015).Each effluent sample was first vortexed and using a sterile wire loop, a loopful of sample was obtained and inoculated onto an agar plate containing the MacConkey supplemented with Meropenem 1 µg/ml.For the swab samples, a primary inoculum was created on the agar plate and used a sterile wire loop to streak the sample onto the plate.After an anerobic incubation at 44 °C for 18 to 24 h, the culture was evaluated for growth.All morphologically distinct colonies were subcultured on standard MacConkey agar plate, and isolates were identified by conventional biochemical tests.Quality control procedures were done using E. coli BAA 1706 as the positive control and E. coli ATCC 25922 as the negative control.

Confirmatory tests of E. coli
Gram staining Gram staining was used to distinguish between gram negative and gram-positive bacteria based on their different cell wall constituents.A drop of normal saline was placed on a glass slide and a loop full of well-isolated bacteria colony added.A smear was made gently and then air-dried.The dried smear was fixed by gently passing the slide on the flame at least three times immediately.The fixed smear was flooded with crystal violet stain for about 1 min, then washed slowly under running water and again flooded with Lugol's iodine for about 1 min, after which it was washed with tap water.Then, the smear was decolorized with 50% acetonealcohol slowly until the purple color stopped running.The slide was rinsed with tap water and then flooded with 2% Safranin as a counter satin for 10-15 s.The slide was rinsed with slowly running tap water and allowed to air dry on a draining rack before examination.During examination, a drop of immersion oil was added on to the smear and examined under the light microscope with the 100 × objective to visualize the morphology of the bacteria.The gram-negative bacteria stain pink and gram positive bacteria stain purple (Bartholomew and Mittwer 1952).E. coli is a gram-negative bacteria hence stained pink.

Biochemical confirmatory tests
After phenotypic colony identification and cell identification of isolates by microscopic visualization through gram staining, conventional biochemical tests were performed to determine the isolated bacteria classification according to Gram-negative identification protocol (Cowan 1993).The IMViC biochemical tests (Benathen 1992), these include Indole test, Methyl red test, Voges Proskauer test, and citrate utilization test as well as Urease production test and Triple Sugar Iron (TSI) (Zinnah et al. 2007) were preformed to identify E. coli.

Indole test
The indole test used to detect the ability of bacteria to split amino acid tryptophan to form indole. Sterilized tubes containing 4 ml of SIM broth media (Condalab, Spain) were prepared.From the pure colonies of the test organism, each tube was inoculated aseptically and incubated at 37 °C for 24-28 h in ambient air.Then, 0.5 ml of Kovac's reagent added to the broth culture and later observed for the presence or absence of a ring.The presence of a ring indicated positive results and absence indicated negative results (Miller and Wright 1982).E. coli is indole positive and formed a ring on addition of the Kovac's reagent.

Methyl red test
The methyl red test used to detect the ability of an organism to produce stable acids as end products from supplied glucose.Using a sterile loop, the pure colonies of the organism were inoculated into fresh sterile MR-VP broth medium (Condalab, Spain).Another tube contained the medium and the control.Both tubes inoculated at 37 °C for 2-5 days.After incubation, 5 drops of methyl red reagent were added to both and observed for color change from yellow to red.A positive reaction is when the culture medium turns red and remains yellow for the negative reactions.E. coli is methyl red positive hence medium turned red.
Voges-Proskauer test Voges-Proskauer test used to determine if an organism produces acetyl-methyl carbinol from glucose fermentation.Using a sterile wire loop, pure colonies were inoculated into a fresh sterile MR-VP broth medium (Condalab, Spain).With the other tube containing the control and medium.Both tubes were incubated at 37 °C for 24-48 h.After incubation, at least 3 drops of Barrit's reagent A and 3 drops of Barrit's reagent B were added to both the tubes and well mixed.After 15-20 min, a red color change forms for a positive result and no color change for the negative result.E. coli is Voges-Proskauer negative, showed no color change.
Urease test Urease hydrolyses urea to yield two molecules of ammonia and one of CO 2 .This reaction can be detected by increasing the medium pH caused by ammonia production.Urease-positive species vary in amount of enzyme produced; hence, the designations of positive, weakly positive, and negative.A heavy inoculum from an 18 to 24-h pure culture was streaked on the entire slant surface.The tubes were then incubated with loosened caps at 37 °C, and color change in the slant was observed from 6 h, 24 h, and everyday up to 6 days.Urease production was indicated by bright color while negative test was indicated by no color change.E. coli is urease negative, showed no color change.
TSI The triple sugar iron test employing Triple Sugar Iron agar is designed to differentiate among organisms based on the differences in carbohydrate fermentation patterns and hydrogen sulphide production.A sterile wire inoculation needle was used to pick a well-isolated colony by just tapping.TSI was then inoculated first by stabbing through the center of the medium to the bottom of tube and then streaked the surface of the agar slant.The tube was loosely capped, and the incubated at 37 °C in ambient air for 18 to 24 h and then examined the color changes in the butt and slant and gas production (displacement of the media).E. coli showed a yellow butt and slant with or without displacement of the media.

Citrate utilization test
The citrate utilization test was used in identification of bacteria that use citrate as the main source of carbon.During the test, bijou bottles were used to prepare slopes of Simmons citrate agar (Condalab, Spain).Using a sterile wire loop, the slope was streaked with a well isolated colony from a pure culture of the test organism and the butt was stabbed with a straight sterile wire loop then cultured at 37 °C for 24 h.The positive test indicated by a bright blue color while negative test showed no color change.Klebsiella pneumoniae was used as a positive control; the test used bromothymol blue indictor, for the blue color (Elazhary et al. 1973).E. coli is citrate-negative and produced no change in color.
Antibiotic susceptibility testing Susceptibility testing was done using Kirby Bauer disc diffusion, on Muller Hinton agar (BiolabZrt Budapest, Hungary), and interpreted according to Clinical and Laboratory Standards Institute (CLSI) guidelines 2018 (Weinstein et al., 2018).Two antibiotic discs that were readily available on market, Imipenem (10 µg) and Meropenem (10 µg) (Oxoid ™ , UK), were used for testing.Three colonies of test isolate were emulsified into sterile saline and the resulting suspension adjusted to turbidity of 0.5 McFarland and then used a sterile swab to inoculate the suspension on the Muller Hinton agar plates to form a lawn, plates impregnated with the two antibiotic discs and incubated aerobically at 37 °C for 24 h.After which, the diameters of zone of inhibition were measured and interpreted according to the recommended criteria by the CLSI guidelines 2018 (Weinstein et al., 2018).Quality control procedures were done using E. coli BAA 1706 as the positive control and E. coli ATCC 25922 as the negative control.

Characterization for the different E. coli pathotypes
DNA extraction DNA extraction was done using the QIAamp DNA Mini Kit (Qiagen, Germany) following the manufacturer's instructions.Briefly, 1 ml of bacteria suspension was transferred in an Eppendorf tube and later centrifuged at 4500 rpm for 5 min at 4 °C.The supernatant was then discarded, and the pellet obtained was resuspended in 200 µl buffer AL and 20 µl Proteinase K.The suspension was thoroughly mixed by vortexing at least 3 times for 5 s each time to obtain a homogenous mixture.Thereafter, the homogenous mixture was incubated at 56 °C in a water bath for 1 h.Following incubation, 200 µl of ethanol (96-100%) was added to the mixture and vortexed to obtain a homogenous mixture.Afterwards, the solution was carefully transferred into a mini column which was correctly capped and then centrifuged at 8000 rpm for 1 min.Following centrifugation, the filtrate was discarded and 500 µl of buffer AW1 was added to the mini column.The cap was then closed and centrifuged at 8000 rpm for 1 min.After which, the cap was carefully opened and 500 µl of buffer AW2 was added to the mini column.The cap was then closed and centrifuged at 14,000 rpm for 3 min.Thereafter, the mini column was transferred into a clean 1.5 ml Eppendorf tube and the collection tube containing the filtrate was then discarded.Afterwards, 60 µl of buffer AE was added to the mini column and the cap was closed.The mini column was then centrifuged at 8000 rpm for 1 min.After which, the mini column was discarded and the Eppendorf tube closed.The DNA was later stored at −20 °C.
Detection of Carbapenemase genes by multiplex PCR Carbapenemase genes were detected using specific primers targeting: bla IMP-1 , bla IMP-2 , bla OXA-48 , bla VIM , bla KPC , and bla NDM genes (Table 1).The PCR reactions were performed in a total of 12.5 µl, "containing 1 × HS Taq PCR Master Mix" (Biolab, New England) and 6 oligonucleotide primer pairs (Eurofins, USA) each at 0.8 µM.One microliter of genomic DNA was added to each PCR reaction tube as template.The thermo cycling (Applied Table 1 Gene-specific primer pairs targeting the different Carbapenemase genes and their corresponding sizes in base pairs biosystems, USA) temperature and time profile was 95 °C for 5 min (initial denaturation) followed by 35 cycles of 95 °C for 30 s (denaturation), 56 °C for 30 s (annealing), 72 °C for 1 min (extension), and a final extension of 72 °C for 10 min.A negative control, double distilled (without template DNA) was included in the PCR amplification.
Ten microliters of each amplicon were "mixed with 2 µl of a 6X loading dye" (Biolabs, New England) and subjected to electrophoresis on a 1.5% agarose gel containing ethidium bromide.The amplified products were visualized using an ultra violet transilluminator (Waghtech international, UK), and the band sizes were estimated by comparison with a 100 bp standard DNA marker (Hyper-Ladder IV [Bioline], Germany).

Pathotyping of E. coli
Detection of virulent genes Carbapenem-resistant isolates were pathotyped using specific primers to detect the presence of the virulent genes: eae, stx, est, elt, paH, and aggR as shown in Table 2.The PCR reactions were performed in a total volume of 12.5 µl, "containing 1 × HS Taq PCR master mix" (Biolab, New England) and 6 oligonucleotide primer pairs (Eurofins, USA) each at 0.8 µM.
One microliter of genomic DNA was added to each PCR reaction as a template.The thermocycling (Applied biosystems, USA) temperature and time profile was 95 °C for 5 min (initial denaturation) followed by 30 cycles of 95 °C for 30 s (denaturation), 50 °C for 30 s (annealing), 72 °C for 1 min (extension), and a final extension of 72 °C for 30 min.A negative control, double distilled water (without DNA template added) was included in the PCR amplification.Ten microliters of each amplicon "mixed with 2 µl of a 6X loading dye" (Biolabs, New England) and were subjected to electrophoresis on a 1.5% agarose gel containing ethidium bromide.The amplified products were visualized using an ultra violet transilluminator (Waghtech international, UK), and the band sizes were estimated by comparison with a 1000 bp standard DNA marker (HyperLadder IV [Bioline], Germany).

Detection of PAI markers
The presence of sequences associated with seven different PAI markers; PAI I 536 , PAI II 536 , PAI IV 536 , PAI I J96 , PAI II J96 , PAI I CFT073 , and PAI II CFT073 as shown in Table 3.The PCR reactions were performed in a total volume of 12.5 µl, "containing 1 × HS Taq PCR master mix" (Biolabs, New England) and seven oligonucleotide primer pairs (Eurofins, USA) each at 0.8 µM.One microliter of genomic DNA was added to each PCR reaction as a template.The thermocycling (Applied biosystems, USA) temperature and time profile was 94 °C for 5 min (initial denaturation) followed by 30 cycles of 94 °C for 1 min (denaturation), 55 °C for 1 min (annealing), 72 °C for 1 min (extension), and a final extension of 72 °C for 10 min.A negative control, double distilled water (without DNA template added) was included in the PCR amplification.Ten microliters of each amplicon "mixed with 2 µl of a 6X loading dye" (Biolabs, New England) and were subjected to electrophoresis on a 1.5% agarose gel containing ethidium bromide.The amplified products were visualized using an ultra violet transilluminator (Waghtech international, UK), and the band sizes were estimated by comparison with both 100 bp standard DNA marker (HyperLadder IV [Bioline], Germany) and Bench top 1 kb DNA ladder (Promega, UK).
Phylogenetic classification Showed that the E. coli strains belonged to four groups (A, B1, B2, or D) based on the presence of the chuA and yjaA genes and the DNA ACA GAA TCG TCA GCA TCA GC fragment (TSPE4.C2) using primers in Table 4.The PCR reactions were performed in a total of 12.5 µl, "containing 1 × HS Taq PCR Master Mix" (Biolab, New England) and three oligonucleotide primer pairs (Eurofins, US) each at 0.8 µM.One microliter of genomic DNA was added to each PCR reaction tube as template.The thermo cycling (Applied biosystems, USA) temperature and time profile was 94 °C for 4 min (initial denaturation) followed by 30 cycles of 94 °C for 5 s (denaturation), 54 °C for 10 s (annealing), 72 °C for 30 s (extension), and a final extension of 72 °C for 5 min.A negative control, double distilled (without template DNA) was included in the PCR amplification.Ten microliters of each amplicon were subjected to electrophoresis on a 2% agarose gel containing ethidium bromide.The amplified products were visualized using an ultra violet transilluminator (Waghtech international, UK), and the band sizes were estimated by comparison with a 1000 bp standard DNA marker (HyperLadder IV [Bioline], Germany).

Data analysis
Data analysis was done using Microsoft Excel and then Statistical Package for the Social Sciences, SPSS version 21.Descriptive statistics namely frequency, percentage distributions, and means were used to present patterns in the data.Genotypes were analyzed using a chi-square or Fisher's exact test.A P value of ≤ 0.05 was considered as evidence of significant statistical difference.

Carbapenem-resistant E. coli isolation
A total of 104 samples were collected from six manholes at Mulago National Referral Hospital every 2 weeks for up to 6 weeks.Of the 104 samples, 32 (30.8%) were collected from the microbiology laboratory manhole, 26 (25%) were collected from manholes around 1C ward, 16 (15.4%)were collected from the hallway manhole, 8 (7.7%) were collected from Causality area manhole, 10 (9.6%) were collected from the Hospital Main Manhole,  and 12 (11.5%)were collected from the mortuary manhole.Over the entire sampling period, a total of 25 (24%) were detected as being Carbapenem-resistant E. coli positive.Of the 25 Carbapenem-resistant E. coli strains, the majority were from the manholes around 1C ward, 13 (52%) followed by the hallway manhole, 9 (36%) and the mortuary manhole, 3 (12%) as shown in Table 5 and while the microbiology laboratory, causality area, and hospital main manholes lacked resistant E. coli strains.

Antibiotic susceptibility
Antibiotic susceptibility testing for 25 Carbapenemresistant E. coli isolates was done using two Carbapenems (Imipenem and Meropenem) as they were readily available on the Ugandan market.The overall resistance to both Carbapenem drugs was with 16 (64%) resistant to Meropenem and 15 (60%) resistant to Imipenem according to CLSI interpretation.Furthermore, 15 (60%) isolates showed co-resistance to both Imipenem and Meropenem (Fig. 1).On comparison of Imipenem resistance between the different manholes, results showed that majority of resistant isolates were from the manholes around 1C ward, 8 (53.33%) followed by the hallway manhole, 5 (33.3%) and mortuary, 2 (13.3%).While comparison of Meropenem resistance between the different manholes, results showed that the majority of resistant isolates were from the manholes around 1C ward, 9 (56.3%)followed by the hallway manhole, 5 (31.3%) and mortuary manhole, 2 (12.5%).The Carbapenem susceptibility of E. coli strains from the different manholes are shown in Table 6.

Detection of Carbapenem-resistant genes
In order to determine the mechanisms of Carbapenem resistance, five Carbapenemase genes were typed for the 20 amplified isolates from which 16 of these isolates were resistant and four isolates, susceptible to Carbapenem.
With regard to the manhole sites, 1C ward manholes had 9 (60%) isolates with bla KPC gene, 3 (50%) isolates with bla NDM gene, and 1 (100%) isolate with bla VIM gene, hallway manhole had 4 (26.7%)isolates with bla KPC gene and 2 (33.3%) isolates with bla NDM gene while the mortuary manhole had 2 (13.3%) isolates with bla KPC gene and 1 (16.7%)isolate with bla NDM gene as shown in Table 7.All the susceptible isolates in this study did not harbor Carbapenemase genes.No significant differences were noted when Carbapenemase genes were compared between Imipenem and Meropenem-resistant isolates.

Discussion
E. coli is a bacterial pathogen emerging as one of the most frequent cause of fatal bacterial infections affecting both humans and animals (Rojas-Lopez et al. 2018) with a rapidly growing multi-drug resistance and has been reported to cause a mortality of up to 50% of infected patients (Okoche et al. 2015).The current global emerging resistance of E. coli to Carbapenems, a class of antibiotic often considered as a last resort drug used in the management of multidrug-resistant gram-negative bacilli (Bharadwaj et al. 2018;Garcia 2013), is very worrying.E. coli has become intrinsically resistant to beta-lactamases due to the emergence of organisms carrying extended spectrum beta-lactamases (ESBLs) and plasmid mediated AmpC beta-lactamases (Nair and Vaz 2013).Carbapenem resistance is attributed to three main mechanisms: porin-mediated resistance, which reduces the uptake of Carbapenems, efflux pumps that expel the Carbapenems outside the cell and enzyme-mediated resistance effected through the acquisition of Carbapenemase genes (Codjoe and Donkor 2017;Elshamy and Aboshanab 2020;Nordmann et al. 2011).Therefore, the overall objective of this study was to characterize the Carbapenem-resistant E. coli from sewage using molecular techniques.
The prevalence of Carbapenem-resistant E. coli (CREC) in this study was 24%.This finding is comparable with other studies in Uganda and Africa that showed a 22.4% CREC prevalence isolated from clinical samples in Uganda (Okoche et al. 2015), 27.1% CREC prevalence isolated from the blood, urine, and wounds in Egypt (Kotb et al. 2020), and elsewhere in China indicating a 4/17 (23.5%) prevalence of CREC in effluent samples (Zhang et al. 2020).The high prevalence of 13/25 (52%) CREC from the 1C ward manhole site could be due to the many patients admitted with different health complications and severities requiring antibiotic treatment including those presenting Carbapenem resistance.Moreover, there is also a similar report showing presence of antibiotic resistance in fecal and body fluids from both healthy and non-healthy human beings (Gwenzi 2020).While the low prevalence of 3/25 (12%) CREC from the mortuary manhole site could be due to persistence of antibioticresistant bacteria and their resistant genes even after one's death (Gwenzi 2020) that run off with the waste water used in cleaning of the corpses.
In the present study, we found that 60% Imipenemresistant isolates and 64% Meropenem-resistant isolates were positive for bla KPC , bla NDM , and bla VIM Carbapenemase genes.Out of the 16 Carbapenem-resistant E. coli isolates, 15 (93.8%) were identified with bla KPC (most abundant) gene, 6 (37.5%) had bla NDM gene, and 1 (6.3%) had bla VIM gene.The results were in agreement with another study by Okoche et al. 2015, in Uganda, in which bla VIM , bla KPC , and bla NDM genes were obtained, although bla VIM (10.7%) was the most prevalent gene, followed by bla OXA-48 (9.7%), bla IMP (6.1%), bla KPC (5.1%), and bla NDM-1 (2.6%) (Okoche et al. 2015).In contrast, our study findings were not in agreement with a study conducted in Khartoum-Sudan by Mahmoud et al. 2020, which showed that of the 86% of the Imipenem-resistant E. coli isolates, 15.5% harbored the bla OXA-48 , as the most abundant gene and no bla KPC was detected (Mahmoud et al. 2020).Furthermore, a study conducted in the USA by Hoelle et al. 2019, which showed that about 55% of E. coli isolates were positive for bla VIM gene and 1% were positive for IMP gene (Hoelle et al. 2019).The variations may be due to the difference in the samples collected, site of collection, and their large sample sizes.On the other hand, the CREC isolates were negative for bla OXA-48 , bla IMP-1 , and bla IMP-2 genes as well as all susceptible isolates in this study did not harbor Carbapenemase genes.Failure to detect bla OXA-48 type producers could be due to their point mutant analogs with ESBLs, resulting into the most difficult Carbapenemase producers to be identified (Nordmann et al. 2011;Queenan and Bush 2007).
Resistance to Carbapenems may be due to intrinsic and or acquired resistance mechanisms for both commensals and pathogenic bacteria (Codjoe and Donkor 2017).
Out of the twenty amplified samples, 16 CREC isolates and four susceptible isolates were pathotyped for six virulence genes, seven PAI markers, and three phylogenetic markers.Among the six virulence genes, only four genes were identified: Eae 8 (50%) as the most prevalent gene, followed by elt 5 (31.3%), est 1(6.3%), and paH with 1 (6.3%).These results deduce presence of enteropathogenic E. coli (EPEC) strains as most abundant pathogenic E. coli with eae genes, followed by enterotoxigenic E. coli (ETEC) strains with elt and est genes and then enteroinvasive E. coli (EIEC) with the paH gene.The four identified virulence genes were obtained from among the CREC isolates and were all negative for stx and aggR genes.It was not surprising that all susceptible isolates did not harbor any virulence genes and hence regarded as non-pathogenic E. coli strains.
Furthermore, E. coli associated infections outside the gastrointestinal tract for example meningitis, urinary tract infections (UTI), septicemia, and pneumonia, are caused by extra-intestinal pathogenic E. coli (Kaper et al. 2004;Russo and Johnson 2000).ExPEC pathogenicity is determined by the existence of virulence genes found either on the plasmids or chromosomes.If located on the chromosomes, the virulence factors are normally situated in specific regions termed as pathogenicity islands (PAI) (Johnson and Stell 2000;Sabaté et al. 2006;Sadat et al. 2022;Tangi et al. 2015).In this study, a multiplex PCR was employed to detect the pathogenicity islands.Of the seven PAI markers, only PAI IV 536 (2/10%) sequence was identified, from 1 (6.3%)CREC isolate and 1 (25%) susceptible isolate.PAI I J96 , PAI II J96 , PAI I CFT073 , PAI II CFT073 , PAI I 536 , and PAI II 536 genes were found negative from the isolates.Results show presence of PAI IV 536 belonging to the uropathogenic E. coli (UPEC) pathogenicity islands (da Silva et al. 2017;Dobrindt et al. 2002) identified in the CREC isolate.This is in agreement with Ssekatawa et al. (2021b) who reported PAI IV 536 as the most predominant pathogenicity island typed in E. coli archived isolates obtained from the Microbiology Laboratory of Mulago National Referral Hospital (Ssekatawa et al. 2021a).Several studies have found PAI IV536 to be of significantly high prevalence and hence named it high pathogenicity island (HPI) (Middendorf et al. 2001;Tangi et al. 2015).While the presence of the PAI IV 536 gene in the susceptible isolate maybe explained as a result of horizontal gene transfer mechanisms mediated by mobile genetic elements (da Silva et al. 2017;Dobrindt et al. 2002).Of the three phylogenetic markers pathotyped, yjaA 15 (60%) was the most abundant gene, followed by TSPE4.C2 and chuA genes with 1 (4%) prevalence.We have also reported that yjaA gene had a 14 (93.3%)prevalence in CREC isolates and 1 (25%) prevalence in susceptible isolates while TSPE4.C2 and chuA had a 1 (25%) prevalence in susceptible isolates, both of which were not harbored in any of the CREC isolates.The presence of yjaA genes, belonging to phylogenetic group B2 is conclusive for virulent extra-intestinal pathogenic E. coli (ExPEC) (Clermont et al. 2000), identified in the CREC isolates.
In this study, we detected the Carbapenemase gene in E. coli with phenotypical resistance to two carbapenem drugs, Meropenem and Imipenem, but only screened for the Carbapenemases commonly produced (KPC, NDM, VIM, OXA-48, IMP-1, and IMP-2) as the resistance mechanisms using Carbapenemase primers.This might be due to the limited number of genes targeted in our study as well as to other mechanisms of resistance such as efflux pumps and porin loss/mutations (Codjoe and Donkor 2017).
An increase in the prevalence of antibiotic resistant pathogenic and commensal E. coli strains has been reported worldwide.Antibiotic resistance in Enterobacteriaceae is mainly due to the expression of beta-lactamase enzymes that breakdown the peptide bond of the beta-lactam ring rendering them inactive.Carbapenemases are the most clinically significant beta-lactamases because they can hydrolyze all beta-lactam antibiotics.Thus, Carbapenemase expressing Enterobacteriaceae are resistant to all beta-lactam antibiotics including Carbapenems yet Carbapenems are the most suitable antibiotics for the treatment gram-negative bacterial infections when other drugs have failed (Nordmann et al. 2012;Ssekatawa et al. 2021b).Therefore, this study evaluated the coexistence of Carbapenem resistance and virulence genetic factors among E. coli phylogroups.We observed that all the 16 (100%) and 1 (5%) CREC isolates had DEC virulence genes and PAI IV 536 , while 93.3% (14/16) of the CREC isolates belonged to phylogenetic group B2 that mainly contains virulent (ExPEC) (Clermont et al. 2000).Similar results were documented by Ssekatawa et al. (2021a, b).
Co-existence of virulence genes, Carbapenem resistance genes, PAI, and yjaA markers that signifies phylogenetic groups B2 in environmental E. coli isolates should be considered as an immense threat as they can lead to severe hard to treat hospital-acquire infections.
Limitations to our study include the few samples collected that did not represent a large population of the E. coli present in hospital sewage, the use of few primers that did not target all known Carbapenemase genes as well as the few pathotyping primers that did not target all known virulence genes due to the limited research budget.Despite the limitations, the study provided the prevalence of CREC in hospital sewage harboring virulence genes and the magnitude of the problem.Finally, the presence of CREC in hospital sewage calls for further investigations to identify all the possible CREC pathotypes that might pose a health-risk to the community living near the hospital sewage drainage system.Furthermore, PAI IV 536 mainly harbors the Yersiniabactin siderophore iron-uptake.However, we never attempted to genotype the Yersiniabactin siderophore iron uptake system genes in PAI IV 536.

Conclusion
In this study, the prevalence of CREC isolated from sewage manholes within Mulago National Referral Hospital, Kampala, was 24%.The prevalence of the investigated virulence genes showed abundancy of various E. coli pathovars such as EPEC, ExPEC, ETEC, and EIEC in the sewage hence a potential major reservoir of pathogenic bacteria harboring virulence genes.Therefore, the presence of pathogenic and non-pathogenic Carbapenemresistant E. coli in the sewage is a threat to management of bacterial infections in clinical settings in Uganda.Highlighting the importance of surveillance of hospital sewage treatment before its release into the common waste and close monitoring to eliminate potential health risks for humans and animals.

Fig. 2
Fig. 2 Prevalence of the various Carbapenemase genes obtained

Table 2
(Dias et al. 2012)imer sequences for molecular characterization of E. coli adopted from(Dias et al. 2012)

Table 3
Primers for detection of PAI markers

Table 4
Primers for phylogenetic classification of E. coli R: CGC GCC AAC AAA GTA TTA CG

Table 5
Number of confirmed Carbapenem-resistant E. coli strains

Table 6
Carbapenem susceptibility of E. coli strains from the different manholes

Table 7
Number of Carbapenem-resistant E. coli isolates positive for Carbapenemase genes

Table 8
Number of E. coli isolates identified with virulence genes

Table 9
Number of E. coli isolates identified with PAI and phylogenetic markers