Molecular characterization and distribution of cephalosporin resistance determinants in Escherichia coli and Klebsiella pneumoniae isolated from patients attending Kampala International University Teaching Hospital in Bushenyi, Western Uganda

ABSTRACT Cephalosporins are the first-line therapy antibiotics used in the treatment of gram-negative bacterial infections. However, high prevalence of cephalosporins resistance in Klebsiella pneumoniae and Escherichia coli has been reported worldwide. Studies conducted in Uganda reported high incidences of cephalosporin resistance (CR). Successive studies at Mulago National Referral Hospital indicated a decline in the resistance levels pointing to the need for regular antibiotic resistance surveillance. Therefore, this study carried out molecular characterization of CR determinants in E. coli and K. pneumoniae isolated from patients attending Kampala International University Teaching Hospital (KIU-TH). A retrospective study using E. coli and K. pnuemoniae samples previously obtained from surgical wounds and urinary tract infections among patients treated at KIU-TH between September 2016 and August 2018 was conducted. Biochemical assays were used to confirm the identity of the samples. Combined disc and boronic acid assays were used to determine the cephalosporine resistance profile of the isolates. Multiplex PCR amplification was used to characterize the extended spectrum beta-lactmase (ESBL) encoding genes. The study revealed that E. coli (130/81.2%) isolates were more predominant than K. pneumoniae (30/18.8%) among the archived samples. K. pneumoniae showed the highest phenotypic resistance with a mean prevalence of 90.6% but comparable to that of E. coli (89.3%). Of the 160 isolates screened, 105 (65.6%) were ESBL producers. Multiplex PCR revealed that the most predominant ESBL encoding gene was bla SHV at a prevalence of 42.0%, followed by bla TEM at 27.3%, bla CTX-M at 22.4% and bla CTX-M-15 at 8.4%. The incidence of phenotypic resistance and distribution of ESBL genes were significantly higher in patients of Ishaka division. Our study reports a high prevalence of cephalosporin-resistant E. coli and K. pnuemoniae isolated from patients attending KIU-TH and highlights the need for routine screening of antimicrobial resistance in health-care facilities so as to guide clinicians on the rational prescription of antibiotics. List of abbreviations: ATCC: American Type Culture Collection; ESBL: extended spectrum beta-lactmases; AmpC: aminopenicillin cephalosporinase; CTX-M: cephotaxime (M-first detected in Munich) hydrolyzing capabilities; CTX-M-U: cephotaxime hydrolyzing capabilities gene Universal primer; SHV: sulfhydryl variables (variant-2); CMY-2: cephamycins (variant-2); TEM: temoneira; ACT-1: AmpC type (variant-1); Bla: Beta lactam; DNA: deoxyribonucleic acid; PCR: Polymerase chain reaction; MNRH: Mulago National Referral Hospital; MRRH: Mbarara Regional Referral Hospital; KRRH: Kabale Regional Referral Hospital (KRRH); KIU-TH: Kampala International University Teaching Hospital; CDL: Central Diagnostic Laboratory; CoVAB: College of Veterinary Medicine, Animal Resources and Biosecurity; CR: cephalosporin resistance


Introduction
Cephalosporins form a relatively large group of antibiotics with a β-lactam ring. This class of antibiotics is bactericidal against gram-negative bacteria. Cephalosporins are classified into first, second, third, fourth, and fifth generations [1]. The third generation cephalosporins are the most widely used in the treatment of gram-negative bacterial infections. They include cefotaxime, ceftriaxone, ceftazidime, cefixime, ceftizoxime, and cefoperazone [2]. Cephalosporin resistance (CR) in Enterobacteriaceae, most importantly, Escherichia coli and Klebsiella pneumoniae, which are the leading causes of hospital-and community-acquired bacterial infections, has been reported worldwide [3][4][5]. Antibiotic resistance to beta-lactam antibiotics arises through a number of mechanisms, which include: (a) inactivation of antibiotics through hydrolyzing the beta-lactam ring by bacterial enzymes encoded by resistance genes located on chromosomes and plasmids, (b) modification and downregulation of porin expression that leads to the decrease in permeability of the outer membrane to cephalosporins; (c) flushing out of cephalosporins by efflux pumps; and (d) acquisition of a new penicillinbinding protein (PBP) or modification of the PBP, thereby reducing the affinity of PBP component to cephalosporins [6,7]. The mode of resistance exhibited by most gram-negative bacteria to cephalosporins is mainly attributed to bacterial enzymes called betalactamases, which are encoded by different genes that include TEM, SHV, CTX, ACT-1, CMY-2, CMY-4, and DHA-1 [6][7][8].
This variation in the CR prevalence and resistance genes reported in Uganda point to the need for antibiotic resistance surveillance at several localities. This is because drug resistance within the pathogens is an evolving process that requires routine surveillance and monitoring in order to provide physicians with the current status of antibiotic resistance for rational prescription of drugs [11]. Furthermore, hospital-based antibiotics resistance stewardship dictates antibiotics prescription in a given hospital and is determined by the prevailing resistance to a given class of antibiotics [11,12]. Thus, this study focused on prevalence and genotypic characterization of CR genes in E. coli and K. pneumoniae isolated from patients attending Kampala International University Teaching Hospital (KIU-TH) in Uganda.

Study design and bacterial samples
We carried out a retrospective study using E. coli and K. pneumoniae samples previously obtained from surgical wounds and UTIs among patients of Ishaka, Bushenyi, and Nyakabirizi divisions treated at KIU-TH between September 2016 and August 2018.
Archived resistant E. coli and K. pneumoniae isolates were collected and recovered from the Microbiology Laboratory, KIU-TH, and transferred to the Central Diagnostic Laboratory, Makerere University, for identity confirmation by biochemical assays. CR profile was determined using the disc diffusion method and, thereafter, the characterization of CR genes using multiplex PCR.

Inclusion and exclusion criteria
Archived samples previously isolated from urine and surgical wounds of in-and out-patients at KIU-TH were included. A total of 180 samples were included. Samples that were not isolated from patients of Bushenyi-Ishaka Municipality (Ishaka, Bushenyi, and Nyakabirizi divisions) and those that could not be confirmed as E. coli or K. pnuemoniae or those where no bacterial growth was observed were excluded from subsequent analysis.

Detection of ESBLs and aminopenicillin cephalosporinases (AmpCs)
To distinguish between ESBLs and AmpCs producers, the phenotypic confirmatory disc diffusion test was performed as follows. The Mueller Hinton Agar (MHA) was inoculated with standard inoculums (0.5 McFarland) of the test isolate. Antibiotic discs containing a mixture of ceftazidime and clavulanic acid at a concentration of 30 µg/10ug and ceftazidime alone was used to taste for the level of resistance. A larger inhibitory zone diameter (>5 mm) in the presence of clavulanic acid than with ceftazidime alone indicated the presence of ESBLs [18]. The phenotypic detection of AmpC β-lactamases was achieved by using boronic acid disc test. Discs containing boronic acid were prepared as follows. One hundred twenty milligrams of phenyl boronic acid (benzeneboronic acid) were dissolved in 3 mL of dimethyl sulfoxide. Then, 3 mL of sterile distilled water was added to this solution. Twenty microliters of this solution was dispensed onto the discs containing 30 µg of cefotetan discs and allowed to dry for 30 minutes and used immediately. The boronic acid test was performed by inoculating MHA with standard inoculums of E. coli and K. pneumoniae followed by placing a disc containing 30 µg of cefotetan and a disc containing 30 µg of cefotetan and 400 µg of boronic acid solution onto the agar. The inoculated plates were incubated overnight at 35°C. Isolates that demonstrated zones of inhibition around the disc containing cefotetan and boronic acid ≥5 mm were considered AmpC producers. Isolates that were positive for phenotypic confirmatory disc diffusion test and boronic acid test were considered to express both ESBLs and AmpCs [18][19][20].

Molecular detection of ESBLs encoding genes
Genomic DNA was extracted using the Qiagen DNA extraction kit following the manufacturer's protocol. Multiplex PCR was performed as previously described by Voets et al. [21] and Dallenne et al. [22], with modifications to enable the concurrent amplification of all the target gene variants using bla TEM , bla SHV , bla CTX-M , and bla CTX-M-15 general primers (Table 1). Briefly, a final PCR volume of 40 µL containing 5 µL of 0.5 µg DNA template, 1 µL of 1 U Taq polymerase, 1 µl of 200 µM dNTPs, 8 µL of all primer combinations (2 μl/primer pair), and 25 µL PCR buffer (2 mM MgCl 2 ,100 mM Tris-HCl,500 mM KCl pH 8.3) was constituted. The PCR program consisted of an initial denaturation at 94°C for 5 minutes followed by 35 cycles at 94°C for 1 minute, 60°C for 30 seconds, 72°C for 1 minute, and a final extension at 72°C for 10 minutes. PCR products were run on a 1.5% agarose gel, stained using ethidium bromide, visualized and imaged under UV light using Image analyzer. The size of bands was used to identify the gene responsible for resistance (Table 1) [23]. K. pneumoniae 7881, K. pneumoniae ATCC 700603, and E. coli ATCC 35218 containing bla SHV , bla CTX-M , and bla TEM genes were used as positive controls, whereas E. coli ATCC 25922 was used as the negative control.

Data analysis
Data analysis was done using OriginPro version 8.5. Comparisons of CR between E. coli and K. pneumoniae, distribution of phenotypic resistance, and genetic determinants of resistance among communities (division of the patients) were performed using one-way analysis of variance (ANOVA) followed by Tukey's multiple comparisons test. A P-value of ≤0.05 indicated significant statistical variance.

CR profile
Of the 180 isolates obtained from the archives, 130 (81.2%) were confirmed to be E. coli, 30 (18.8%) K. pneumoniae, and 13 (7.2%) isolates were neither E. coli nor K. pneumoniae, while 7 (3.9%) isolates exhibited no growth. Bacterial isolates were predominantly obtained from urine (96/60%), Tables 2 and 3. All the 160 isolates were at least resistant to one of the antibiotics. Resistance to Cefotaxime was the highest but not significantly different from the resistance registered by ceftriaxone, ceftazidime and cefixime. K. pneumoniae and E. coli exhibited statistically similar phenotypic resistance, Table 2.

Prevalence of ESBLs and AmpCs producers
Phenotypic confirmatory disc diffusion and boronic acid disc assays revealed that out of 160 isolates screened, 105 (65.6%) were ESBLs producers where 56.2% (69 E. coli and 21 K. pneumoniae) were positive for ESBLs, while 9.4% (12 E. coli and 3 K. pneumoniae) harbored both ESBLs and AmpCs. None of the isolates was positive for only AmpCs while 55 (34.4%) were negative for both ESBLs and AmpCs.

Distribution of ESBLs encoding genes
The 105 isolates that stood to be ESBL producers were typed by multiplex PCR and recorded a total of 143 genes ( Figure 1).  bla TEM genes coexisting. However, the mean occurrence of ESBL encoding genes was substantially higher in E. coli (Table 4; Figure 2).

Comparison of phenotypic resistance prevalence among the divisions
One-way analysis of variance using Tukey's multiple comparisons test package was conducted to compare the mean phenotypic resistance prevalence among the different divisions of Bushenyi-Ishaka Municipality. The multiple comparison analysis registered P values >0.05 indicating statistically similar levels of phenotypic resistance among the different divisions ( Figure 3).

Distribution of cephalosporin genetic resistance determinants among the divisions
ESBLs encoding genes were highly distributed among the Ishaka division population with a prevalence of 15.5% followed by Bushenyi division (7.3%) and Nyakabirizi division (2.3%). Comparison of the distribution of ESBLs encoding genes among the different divisions located in Bushenyi-Ishake Municipality was performed in one-way ANOVA. Tukey's multiple comparison analysis computed a P value >0.05 between Bushenyi and Nyakabirizi divisions indicating no statistical difference in the prevalence of ESBLs encoding genes among the population, whereas P values <0.05 were obtained between Ishaka and Bushenyi divisions, Ishaka and Nyakabirizi division signifying statistically higher prevalence of the ESBL genes among the Ishaka population. The distribution of the ESBL genes is illustrated by Figure 4.

Discussion
Phenotypic and molecular characterization revealed that 105 (65.6%) of the isolates expressed ESBLs. The phenotypic resistance in this study is comparable to that (62.0%) reported at MNRH in 2015 [25] but lower than resistance frequencies of 81.4% and 89% reported at MNRH in 2013 [9] and Kabale Regional Referral Hospital (KRRH) [26], respectively. Contrary to this, lower ESBLs phenotypic (29.9% and 13.4%) and genotypic (51%) frequencies were reported by studies carried out at MRRH [10] and MNRH in 2017 [27]. Nosocomial outbreaks in hospital settings with limited therapeutic options have been attributed to AmpC β-lactamase producing bacteria [28]. In this study, combined disc diffusion and boronic acid assays revealed that only 15 isolates co-expressed ESBLs  and AmpCs. Thus, a low prevalence (9.4%) of AmpC β-lactamase producers was registered by this study. These findings corroborate well with other studies that reported relatively low prevalence (0.5%-30%) of coproduction of ESBLs and AmpCs [29][30][31][32] E. coli (81.2%) isolates were more predominant than K. pneumoniae (18.8%) among the archived samples. This is in agreement with other studies that attempted to establish the prevalence of ESBLs producing Enterobacteriaceae in Uganda [9,10,25,27]. The mean prevalence of CR was higher in K. pneumoniae (90.6% ± 9.07) but not significantly different from that registered by E. coli isolates (89.3% ± 5.51). This is in contrast with two studies where the prevalence of CR reported was 52% and 44% at KRRH [26] and 72.7% and 58.1% at MNRH in 2015 [25] but corroborates with other studies that reported CR incidence as 79.2% and 92.3% at MNRH in 2013 [9] and 85% and 88% globally [33] in E. coli and K. pneumoniae, respectively. However, studies by Moses et al. [10] at MRRH in 2013 and Ampaire et al. [27] at MNRH in 2017 reported E. coli as the most predominant ESBL producer. Antimicrobial resistance (AMR) stewardship strategies put forward to combat antibiotic resistance can only be assessed by routine quantification of AMR prevalence. Successive studies were carried out to determine the prevalence of ESBLs producers at MNRH. Indeed, the results are promising as the prevalence of ESBL producers reduced from >75% [9] to 13.4% [27] reported in 2013 and 2017, respectively.
For molecular characterization of ESBL genes, bla SHV primer that amplified two genes, probably bla SHV-1-like and bla SHV-2-like , with different amplicon sizes was used. The commonest ESBL gene identified was bla SHV (SHV-1 and SHV-2) (42%) followed by bla TEM (27.3%), bla CTX-M (22.4%), and bla CTX-M-15 (8.4%). However, these findings are not in agreement with the ESBL genes prevalence registered at MRRH where the most prevalent ESBL gene was CTX-M (70%), SHV (34%), and TEM (0.8%) [10] but comparable to ESBL gene prevalence reported in other parts of the world. For example, studies conducted in Australia, Belgium, Turkey, South Africa, and USA reported bla SHV (67.1%) as the most prevalent gene, trailed by CTX-M-type(23.3%) and TEM-type (16.4%) [34], and this incidence is similar in many other countries [35][36][37]. Furthermore, 35 out of the 105 ESBLs positive isolates expressed more than one ESBL gene. This collaborates with Moses et al.'s [10] study at MRRH where 87 of the 213 isolates harbored more than one ESBL gene and other studies done outside Uganda. For instance, in Turkey, 68 isolates had SHV/TEM coexisting, 61 had TEM/CTX-M, and 54 had SHV/CTX-M genes [38]. Moreover, seven (21.87%) isolates containing TEM/SHV were reported in Iran [39]. It is worth noting that this study registered a rare occurrence of three genes: bla SHV-1 /bla CTX-M /bla TEM coexisting in three of the E. coli isolates. Existence of three ESBL genes in one isolate is in line with Mohsen et al.'s [40] study who reported the cooccurrence of bla CTX-M /bla TEM /bla SHV in 14 isolates.
Furthermore, 55 out of 160 isolates contained neither ESBLs nor AmpCs but exhibited excellent CR. This resistance may be attributed to the presence of resistance intrinsic factors, such as downregulation and mutation of the outer membrane proteins, through which antibiotics diffuse [41][42][43][44]. High prevalence of carbapenem resistance has been reported in Uganda [45][46][47]. Acquisition of carbapenemase encoding genes by Enterobacteriaceae confer resistance to all beta lactam antibiotics [41,45]. Thus, the most likely source of resistance in the 55 isolates is probably the presence of carbapenem resistance genes.
Among the three divisions, Ishaka has the highest population density (891.0/km 2 ), followed by Bushenyi division also known as Central division (490.4/km 2 ), and Nyakabirizi division (248.1/km 2 ) [48]. Statistical comparison of phenotypic resistance prevalence and distribution of cephalosporinresistance determinants within the divisions of Bushenyi-Ishaka Municipality revealed higher phenotypic resistance and distribution of ESBLs encoding genes in the population of Ishaka division. The results corroborate well with Bruinsma et al.'s [49] findings. While studying the impact of population density on antibiotic resistance, Bruinsma et al. [49] observed that population density is a vital aspect that influences the development of antibiotic resistance. Similar findings were reported by Honda et al. [50].

Conclusions
High prevalence of CR in E. coli and K. pneumoniae isolated from patients treated at KIU-TH was revealed by this study. K. pneumoniae was encountered as the most prevalent resistant isolate and blaSHV as the most predominant gene. Thus, the third-generation cephalosporin endorsed for treatment of Enterobacteriaceae infections were not effective in vitro. Indeed, this study accentuates routine screening of AMR in different healthcare facilities. This will guide clinicians on the rational prescription of antibiotics so as to control the emergence and transmission of resistant strains. This study has provided an insight into the possible presence of determinants of CR other than ESBL and AmpCs; thus, further investigations are required to determine the other factors responsible for CR among Enterobacteriaceae isolated from patients attending KIU-TH.

Limitations
This study did not look at carbapenem resistance, yet carbapenemase production confers resistance to all beta-lactams. Phenotypic resistance in isolates that neither harboured ESBLs nor AmpCs may be due to the presence of carbapenemase genes. Primers used to detect ESBLs encoding genes in this study targeted most gene variants. Thus, sequencing of the amplicons should be done to distinguish between the several gene variants.

Authors' contributions
This work was carried out in collaboration between all authors. Eddie Wampande and Kenneth Ssekatawa conceptualized and designed the format for this study. Kenneth Ssekatawa and Herbert Mbyemeire carried out all the laboratory experiments. Kenneth Ssekatawa conducted data analysis. All authors drafted and managed manuscript revisions. All authors read and approved the final manuscript.