Molecular Epidemiology of Extended-Spectrum Beta-Lactamase and AmpC Producing Enterobacteriaceae among Sepsis Patients in Ethiopia: A Prospective Multicenter Study

Extended-spectrum beta-lactamases (ESBLs) and AmpC producing Enterobacteriaceae are public health threats. This study aims to characterize ESBL and AmpC producing Enterobacteriaceae isolated from sepsis patients. A multicenter study was conducted at four hospitals located in central (Tikur Anbessa and Yekatit 12), southern (Hawassa) and northern (Dessie) parts of Ethiopia. Blood culture was performed among 1416 sepsis patients. Enterobacteriaceae (n = 301) were confirmed using MALDI-TOF and subjected for whole genome sequencing using the Illumina (HiSeq 2500) system. The overall genotypic frequencies of ESBL and AmpC producing Enterobacteriaceae were 75.5% and 14%, respectively. The detection of ESBL producing Enterobacteriaceae at Hawassa, Yekatit 12, Tikur Anbessa and Dessie was 95%, 90%, 82% and 55.8%, respectively. The detection frequency of blaCTX-M, blaTEM and blaSHV genes was 73%, 63% and 33%, respectively. The most frequently detected ESBL gene was blaCTX-M-15 (70.4%). The common AmpC genes were blaACT (n = 22) and blaCMY (n = 13). Of Enterobacteriaceae that harbored AmpC (n = 42), 71% were ESBL co-producers. Both blaTEM-1B (61.5%) and blaSHV-187 (27.6%) were the most frequently detected variants of blaTEM and blaSHV, respectively. The molecular epidemiology of ESBL producing Enterobacteriaceae showed high frequencies and several variants of ESBL and AmpC genes. Good antimicrobial stewardship and standard bacteriological laboratory services are necessary for the effective treatment of ESBL producing Enterobacteriaceae.


Introduction
Globally, Enterobacteriaceae that harbor extended-spectrum beta-lactamase (ESBL) genes are spreading and causing serious infections, such as sepsis [1]. Sepsis is a lifethreatening condition resulting from a dysregulated immune response to the infection, which ultimately results in organ dysfunction [2,3]. Antibiotic options for the management of septic patients caused by ESBL producing Enterobacteriaceae (ESBL-pE) is narrow, which can lead to longer hospital stays, increased hospital costs and increased mortality [4].
ESBL-pE has become a global health problem [5] because ESBL can make a diverse range of β-lactam antibiotics ineffective, including penicillins, cephalosporins and monobactams [6]. Enterobacteriaceae acquire and disseminate these ESBL-encoding genes horizontally, mainly through plasmids [7]. In addition to Escherichia coli, various species of Klebsiella, Enterobacter, Serratia and Salmonella are major ESBL producers in the family Enterobacteriaceae [8][9][10].
Generally, all beta-lactamase variants are classified into four classes, A (serine penicillinases), B (metallo-beta-lactamases), C (cephalosporinases) and D (oxacillinases), which give resistance to penicillins, most β-lactams, cephalosporins and cloxacillin, respectively [11]. Some key families of ESBL-encoding genes, such as bla CTX-M , bla TEM and bla SHV are grouped in class A [8,12,13]. The bla CTX-M family, that spread rapidly during the last decade, originated from environmental bacteria and its variants have developed due to point mutations [14]. All bla CTX-M variants are ESBL genes [8,13] in which bla CTX-M-15 is presently the most prevalent [10]. Other ESBL genes include bla OXA , bla VEB , bla VER and bla GES [7,8,13]. The co-presence of ESBL genes in the same species is a significant threat of transfer to other species [11].
Enterobacteriaceae that produce AmpC enzymes can also hydrolyze β-lactams and inhibitor-β-lactam combinations [15]. AmpC may be encoded in the chromosomes or plasmids of most members of Enterobacteriaceae [16]. Around the globe, several AmpC genes are detected in Enterobacteriaceae and of these genes the bla CMY and bla DHA families are the most common [17][18][19][20].
The World Health Organization (WHO) has listed ESBL producing Enterobacteriaceae as critical priority pathogens [21] and the magnitude of such strains is increasing worldwide [6,22,23]. However, there is scarcity of data related to the genetic epidemiology of ESBL and AmpC producing Enterobacteriaceae in sub-Saharan countries, including Ethiopia. It is a crucial time to determine the genetic epidemiology of ESBL and AmpC producing Enterobacteriaceae at a larger scale in order to guide future antimicrobial resistance control programs. Hence, this study aims to determine the molecular epidemiology of ESBL and AmpC producing Enterobacteriaceae among patients investigated for sepsis at four Ethiopian teaching/referral hospitals, which are located in the central, southern and northern parts of the country. These hospitals are serving millions of people in the surrounding catchment area and people who are referred to these hospitals.

Sociodemographic Characteristics
In the present study, a total of 1416 patients investigated for sepsis from four different hospitals were enrolled. The number of patients from Tikur Anbessa Specialized Hospital (TASH) was 501, and the numbers from Yekatit 12 Specialized Hospital Medical College (Y12HMC), Dessie Referral Hospital (DRH) and Hawassa University Comprehensive Specialized Hospital (HUCSH) were 298, 301 and 316, respectively. The male participants were 55.3% while the females were 44.7%. The patients' ages ranged from half a day to 90 years with a mean age of 8.85 years (Table 1).

Detection of bla CTX-M , bla TEM and bla SHV Genes
Among all Enterobacteriaceae, bla CTX-M , bla TEM and bla SHV were detected in 73%, 63% and 33% of the samples at least once ( Figure 1). Among the Enterobacteriaceae isolated at each hospital, the detection of bla CTX-M at HUCSH, Y12HMC, TASH and DRH was 95%, 88%, 78% and 54%, respectively. At DRH, bla TEM (59%) was detected at a higher frequency while bla SHV (10%) detection was low ( Figure 1) At HUCSH, bla TEM was detected at 82% while bla SHV detection was relatively low (35%). At TASH, bla TEM was more frequent than bla SHV . At Y12HMC, while the detection of bla CTX-M was the highest with 88%, bla TEM and bla SHV detection was the same at 75%. Differences in the detection of bla CTX-M , bla TEM and bla SHV showed statistically significant associations per study site (p < 0.001).

Detection of blaCTX-M, blaTEM and blaSHV Genes
Among all Enterobacteriaceae, blaCTX-M, blaTEM and blaSHV were detected in 73%, 63% and 33% of the samples at least once ( Figure 1). Among the Enterobacteriaceae isolated at each hospital, the detection of blaCTX-M at HUCSH, Y12HMC, TASH and DRH was 95%, 88%, 78% and 54%, respectively. At DRH, blaTEM (59%) was detected at a higher frequency while blaSHV (10%) detection was low ( Figure 1) At HUCSH, blaTEM was detected at 82% while blaSHV detection was relatively low (35%). At TASH, blaTEM was more frequent than blaSHV. At Y12HMC, while the detection of blaCTX-M was the highest with 88%, blaTEM and blaSHV detection was the same at 75%. Differences in the detection of blaCTX-M, blaTEM and blaSHV showed statistically significant associations per study site (p < 0.001). K. pneumoniae was the most frequent isolate and harbored high frequencies of blaCTX-M (95%), blaSHV (93%) and blaTEM (78%) genes ( Figure 2). Similarly, most Enterobacteriaceae were found to have blaCTX-M and blaTEM gene families with different detection rates between the strains. The detection of blaSHV in K. variicola (1%) and E. coli (2%) was very low, while there was no detection of blaSHV among E. cloacae (n = 21) and P. dispersa (n = 20). No blaCTX- K. pneumoniae was the most frequent isolate and harbored high frequencies of bla CTX-M (95%), bla SHV (93%) and bla TEM (78%) genes ( Figure 2). Similarly, most Enterobacteriaceae were found to have bla CTX-M and bla TEM gene families with different detection rates between the strains. The detection of bla SHV in K. variicola (1%) and E. coli (2%) was very low, while there was no detection of bla SHV among E. cloacae (n = 21) and P. dispersa (n = 20). No bla CTX-M , bla TEM and bla SHV genes were detected in the rare isolates of A. xylosoxidans, E. bugandensis, K. cowanii, L. amnigena and S. dysenteriae ( Figure 2).

Molecular Epidemiology of ESBL Producing Enterobacteriaceae
Of all the Enterobacteriaceae subjected to WGS, 75.5% encoded at least one ESBL gene. At least one ESBL gene was detected among 95% of Klebsiella pneumoniae, 68% of K. variicola, 53% of E. coli and 43% of E. cloacae ( Figure 3). The frequencies of ESBL producing Enterobacteriaceae detected at HUCSH, Y12HMC, TASH and DRH were 95%, 90%, 82% and 55.8%, respectively (p < 0.001) ( Table 3). Of the Enterobacteriaceae obtained from the neonatal intensive care units (n = 189) and pediatrics wards (n = 68), 73.5% and 87% had at least one ESBL gene, respectively (Table 3). Among the patients who showed blood culture positivity for Enterobacteriaceae, the possible risk factors for the increased ESBL producing Enterobacteriaceae were assessed. However, the multivariate analysis did not show any statistically significant association between the independent variables and higher frequencies of ESBL producing Enterobacteriaceae (Table 3).

Molecular Epidemiology of ESBL Producing Enterobacteriaceae
Of all the Enterobacteriaceae subjected to WGS, 75.5% encoded at least one ESBL gene. At least one ESBL gene was detected among 95% of Klebsiella pneumoniae, 68% of K. variicola, 53% of E. coli and 43% of E. cloacae ( Figure 3). The frequencies of ESBL producing Enterobacteriaceae detected at HUCSH, Y12HMC, TASH and DRH were 95%, 90%, 82% and 55.8%, respectively (p < 0.001) ( Table 3). Of the Enterobacteriaceae obtained from the neonatal intensive care units (n = 189) and pediatrics wards (n = 68), 73.5% and 87% had at least one ESBL gene, respectively (Table 3). Among the patients who showed blood culture positivity for Enterobacteriaceae, the possible risk factors for the increased ESBL producing Enterobacteriaceae were assessed. However, the multivariate analysis did not show any statistically significant association between the independent variables and higher frequencies of ESBL producing Enterobacteriaceae (Table 3).

ESBL Genes
Several variants of blaCTX-M that are ESBL were detected among the Enterobacteriaceae sequenced from all the study sites ( Table 4). The most frequent gene was blaCTX-M-15 with an overall detection rate of 70.4%. The frequency of blaCTX-M-15 at DRH, TASH, HUCSH and Y12HMC was 50.4%, 73.6%, 93% and 87.5%, respectively. In addition to blaCTX-M-15, two

ESBL Genes
Several variants of bla CTX-M that are ESBL were detected among the Enterobacteriaceae sequenced from all the study sites ( Table 4). The most frequent gene was bla CTX-M-15 with an overall detection rate of 70.4%. The frequency of bla CTX-M-15 at DRH, TASH, HUCSH and Y12HMC was 50.4%, 73.6%, 93% and 87.5%, respectively. In addition to bla CTX-M-15 , two other bla CTX-M variants were found at DRH only. K. pneumoniae (92%) was the most common bla CTX-M-15 producer ( Figure 4). The detection of bla CTX-M-15 from K. variicola, E. coli, P. dispersa and K. oxytoca was 65%, 51%, 90% and 69%, respectively ( Figure 4). bla SHV-106 was another ESBL gene detected at 3.7% frequency with a majority from DRH and TASH. bla TEM-207 , bla SRT-1 , bla GES-11 and bla CMY-42 were the other rare ESBL genes detected. K. oxytoca was found to carry several variants of the intrinsic bla OXY gene, also classified as ESBL (Table 4).

Co-Occurrence of Multiple ESBL Genes
In several Enterobacteriaceae, either the co-occurrence of multiple ESBL genes or one ESBL gene with other bla TEM and bla SHV variants was detected ( Table 6). Enterobacteriaceae that carried the bla CTX-M-15 gene were found to harbor several other ESBL genes or non-ESBL variants of bla TEM and bla SHV . The co-occurrence of bla CTX-M-15 and bla TEM-1B (n = 166) was the most frequent gene combination followed by bla CTX-M-15 and bla SHV-187 . While the 3 combinations of bla CTX-M-15 * bla TEM-1B * bla SHV-187 were detected among 56 Enterobacteriaceae, the 4 bla CTX-M-15 * bla TEM-1B * bla SHV-187 * bla SHV-106 gene combination was detected from 9 Enterobacteriaceae. A combination of five multiple genes (bla CTX-M-15 * bla TEM-1B * bla SHV-187 * bla TEM-207 * bla SHV-106 ) was detected from one Enterobacteriaceae (Table 6). Table 6. Co-occurrence of multiple ESBL genes and ESBL genes with other common non-ESBL variants of bla TEM and bla SHV .

Discussion
The current study is the first report showing the molecular epidemiology of ESBL producing Enterobacteriaceae among patients investigated for sepsis at four referral hospitals located in the northern, central and southern parts of Ethiopia. Of all the Enterobacteriaceae subjected to WGS, 75.5% encoded at least one ESBL gene. Similar to our study, the previous reports from different African countries also showed a high occurrence of ESBL [24][25][26]. The high frequency of Enterobacteriaceae producing ESBL enzymes among sepsis patients in Ethiopia also showed similarities with studies conducted in other continents [4,[27][28][29]. The spread of bacteria producing ESBL enzymes among sepsis patients is worrying because of the invasive nature of the disease, coupled with the limited antibiotics options left for its management. More importantly, the high level of ESBL producing strains among sepsis causing Enterobacteriaceae in low-income countries is a major public health problem, due to the limited laboratory services and therapeutic options available. ESBL are often encoded by plasmids that are able to transfer horizontally, which contributes to the rapid spread of ESBL-pE. Additionally, as documented for developed countries, the ESBL genes detected in food, animals and environment, human migration and limited access to sanitations could also contribute to the rise of ESBL genes globally [30]; however, this needs further investigation for the case of Ethiopia.
In a comparison between the hospitals, very high ESBL-pE levels were detected in the three hospitals, HUCSH (95%), Y12HMC (90%) and TASH (82%), which are located in the southern and central parts of Ethiopia. In contrast, the detection of ESBL enzymes at DRH (55.8%), which is located in the northern part of Ethiopia, was comparatively low. However, the lower level of ESBL detection at DRH was still higher compared to studies conducted in South Africa [31], Mozambique [20] and South and South East Asia [32]. While it is known that the spread of ESBL genes varies geographically [33], the factors that contribute to its high occurrence in some hospitals and a lower detection rate in the other hospital within a single country requires investigation so that effective control mechanisms can be designed.
In this study, bla CTX-M (73%) was the most frequently detected ESBL family across the four referral hospitals. This finding was similar to a multicenter study performed in neighboring Sudan that reported 78% of bla CTX-M [25]; however, a very low detection rate was reported in the U.S.A. [34]. Bla CTX-M-15 (70.4%) was the most abundant ESBL gene detected. This finding showed similarities with different studies across the globe [1,14,20,27,29,32]. Bla CTX-M-15 was very abundant at TASH (73.6%) and HUCSH (93%), while it was detected at a comparatively lower rate at DRH (50.4%). While bla CTX-M-15 was disseminated in all hospitals, bla CTX-M-3 and bla CTX-M-9 were detected only at DRH (northern). The detection of high levels of bla CTX-M at TASH (central) may possibly be explained because the hospital is the main destination of patients referred from all over the country, and these patients can carry such strains to the hospital.
In this study, bla TEM (73%) was also detected in abundance, which showed agreements with other studies [12,33]. Out of these, only bla TEM-207 were recorded as an ESBL gene, though its detection rate was rare and restricted to only two hospitals (TASH and DRH). This rare detection of bla TEM ESBL variants showed similarities with a study conducted in Mozambique that reported a single detection of the bla TEM gene [20]. Of several variants of bla TEM detected, bla TEM-1B (61.5%) was the most frequent. Other variants of bla TEM were either broad-spectrum beta-lactamases or inhibitor-resistant beta-lactamases. Even though most bla TEM beta-lactamases were not ESBL, it is worrying that the majority of bla TEM variants were co-detected with ESBL genes, such as bla CTX-M variants and other ESBL genes.
Similarly, the detection of bla SHV (33%) showed similarities with a study conducted in Sudan that reported 28% of the bla SHV genes [25]. The ESBL variants of bla SHV detected in this study were bla SHV-12 , bla SHV-65 and bla SHV-106 . The bla SHV-106 variant was detected at DRH (northern) and TASH (central), but not at Y12HMC. A similar bla SHV-106 detection rate was reported in Portugal [35]. However, a very minimal bla SHV occurrence was reported in China that investigated 499 E.coli [36]. bla SHV-187 (27.6%), was the most frequently detected variant of bla SHV and is a broad-spectrum beta-lactamase. A majority of the bla SHV genes were co-detected with bla CTX-M and other ESBL genes, which worsens the problem.
In addition, different intrinsic bla OXY variants encoded with K. oxytoca that are ESBL were detected even though most were found at TASH and DRH, but rarely at HUCSH and Y12HMC. After the first detection of bla OXY in Spain in 2009 [37], several variants were detected in this study showing how fast the variants are spreading throughout the world [38,39]. In addition to bla GES-11 , the rarely detected ESBL genes that are grouped in class C and has having ESBL activities were bla SRT-1 and bla CMY-42 , and this showed agreements with other studies [40,41].
Multiple ESBL genes were detected in several cases and most ESBL-encoding strains were harboring other beta-lactamase genes. The bla CTX-M-15 gene co-occurred with several other ESBL genes and the non-ESBL variants of bla TEM and bla SHV . The co-occurrence of bla CTX-M-15 and bla TEM-1B was the most frequently detected gene combination, followed by the bla CTX-M and bla SHV-187 combination. The findings of this co-occurrence of multiple ESBL genes were in agreement with the other studies [20,25]. These findings showed the successful spread of Enterobacteriaceae harboring multiple ESBL genes in the study settings, which is very alarming.
In the current study, 14% of Enterobacteriaceae harbored at least one AmpC gene while seven strains had multiple AmpC genes. However, a higher AmpC detection was reported in Cairo, Egypt [16]. Several variants of AmpC were detected, with bla ACT (n = 25) as the most common variant, which was in accordance with another study [19]. bla ACT-5 , bla ACT-7 , bla ACT-16 , bla ACT-14 and bla ACT-15 were the different variants of bla ACT detected in this study. Another commonly detected AmpC gene was bla CMY and its different variants bla CMY-6 , bla CMY-148 , bla CMY-2 and bla CMY-42 . The detection of bla CMY was in agreement with the other studies [19,20]. The other rarely detected AmpC genes detected were bla DHA-1 , bla CMH-3 , bla SRT-1 , bla SRT-2 and bla ADC-25 . A majority of AmpC producing Enterobacteriaceae (71%) were also ESBL producers that could easily facilitate the rapid spread of both ESBL and AmpC genes, as described in a review article from Africa [24].
At least one ESBL gene was detected among 95% of K. pneumoniae, 68% of K. variicola, 53% E. coli and 43% of E. cloacae. These findings were in line with other studies that recognized these organisms as the main ESBL carriers [9,10,42]. K. pneumoniae was the most frequent isolate that harbored high frequencies of ESBL bla CTX-M-15 (92%), and also showed high frequencies of bla SHV-187 (78%) and bla TEM-1B (74%). A similar high level of ESBL genes encoded by K. pneumoniae were reported in Tanzania [43]; however, a very low detection rate was reported in Germany [10] and the U.S.A. [34]. In addition to the ESBL genes, 9% of K. pneumoniae were harboring different AmpC genes showing that these strains are the main ESBL carriers in all hospitals. The detection of ESBL genes in sepsis isolates is very important to guarantee that appropriate antibiotic therapy is prescribed when an ESBL producing strain is identified, and to limit unnecessary antibiotic prescriptions when ESBL genes are not identified [34].

Study Design and Study Sites
Among the patients investigated for sepsis, a multicenter prospective cross-sectional study was conducted between October 2019 and September 2020 at four selected hospitals located in the central, southern and northern parts of Ethiopia ( Figure 5). University and referral hospitals that had established microbiology laboratories or a link with a nearby government regional microbiology laboratory were selected. These were the Tikur Anbessa Specialized Hospital (TASH) and the Yekatit 12 Specialized Hospital Medical College (Y12HMC) in the central, Hawassa University Comprehensive Specialized Hospital (HUCSH) in the southern and Dessie Referral Hospital (DRH) in the northern parts of Ethiopia. Figure 5. The four Ethiopian referral hospitals selected for this study and from where Enterobacteriaceae were isolated. TASH, located in Addis Ababa, is the teaching hospital of Addis Ababa University and the largest referral and oldest hospital in the country. It provides tertiary level referral diagnoses and treatment for patients referred from all over the country with over 800 beds. DRH is one of the largest public hospitals in the northern part of Ethiopia, located in Dessie. It is a referral hospital with 560 beds, providing services for the surrounding areas and residents of the neighboring regions. Y12HMC is located in Addis Ababa and provides health care services to Addis Ababa residents, referral cases from health centers in Addis Ababa and its bordering regions. The hospital has over 300 beds and serves more than 5 million people in its catchment area. HUCSH, located in Hawassa city, is one of the largest health facilities in the southern part of the country and provides teaching, public health services and research activities with over 400 beds.

Blood Culture and Enterobacteriaceae Isolation and Identification
All patients investigated for sepsis who sought medical service at the study sites were included. The attending physician's decision was used to identify the eligible patients as sepsis cases and their socio-demographic and possible risk factors data were gathered. All age groups were included, but patients who had been on antibiotic treatment within the preceding ten days were excluded from the study. From all the study sites, a total of 1416 clinically diagnosed cases of sepsis from different wards were enrolled in the study. A single blood culture bottle system was processed from all the patients, and bacterial identification was performed in accordance with a standardized laboratory protocol. At each study site, Enterobacteriaceae were characterized by their colony characteristics, Gramstaining and conventional biochemical tests using triple sugar iron, indole, urea, citrate, lysine decarboxylase, motility and malonate. All strains were stored at −70 °C or −16 °C and transported to the Armauer Hansen Research Institute, and later brought to Sweden for further characterization. All the Enterobacteriaceae were re-identified and confirmed using MALDI-TOF MS at the Clinical Microbiology Department of Uppsala University Hospital, Uppsala, Sweden, and the Karolinska Institute, Stockholm, Sweden. Each laboratory test was processed in accordance with the established protocols and carefully recorded. Each MALDI-TOF run included quality control using E. coli ATCC 25922. A total of 301 Enterobacteriaceae isolated from all the study sites was subjected for whole genome Figure 5. The four Ethiopian referral hospitals selected for this study and from where Enterobacteriaceae were isolated. TASH, located in Addis Ababa, is the teaching hospital of Addis Ababa University and the largest referral and oldest hospital in the country. It provides tertiary level referral diagnoses and treatment for patients referred from all over the country with over 800 beds. DRH is one of the largest public hospitals in the northern part of Ethiopia, located in Dessie. It is a referral hospital with 560 beds, providing services for the surrounding areas and residents of the neighboring regions. Y12HMC is located in Addis Ababa and provides health care services to Addis Ababa residents, referral cases from health centers in Addis Ababa and its bordering regions. The hospital has over 300 beds and serves more than 5 million people in its catchment area. HUCSH, located in Hawassa city, is one of the largest health facilities in the southern part of the country and provides teaching, public health services and research activities with over 400 beds.

Blood Culture and Enterobacteriaceae Isolation and Identification
All patients investigated for sepsis who sought medical service at the study sites were included. The attending physician's decision was used to identify the eligible patients as sepsis cases and their socio-demographic and possible risk factors data were gathered. All age groups were included, but patients who had been on antibiotic treatment within the preceding ten days were excluded from the study. From all the study sites, a total of 1416 clinically diagnosed cases of sepsis from different wards were enrolled in the study. A single blood culture bottle system was processed from all the patients, and bacterial identification was performed in accordance with a standardized laboratory protocol. At each study site, Enterobacteriaceae were characterized by their colony characteristics, Gramstaining and conventional biochemical tests using triple sugar iron, indole, urea, citrate, lysine decarboxylase, motility and malonate. All strains were stored at −70 • C or −16 • C and transported to the Armauer Hansen Research Institute, and later brought to Sweden for further characterization. All the Enterobacteriaceae were re-identified and confirmed using MALDI-TOF MS at the Clinical Microbiology Department of Uppsala University Hospital, Uppsala, Sweden, and the Karolinska Institute, Stockholm, Sweden. Each laboratory test was processed in accordance with the established protocols and carefully recorded. Each MALDI-TOF run included quality control using E. coli ATCC 25922. A total of 301 Enterobacteriaceae isolated from all the study sites was subjected for whole genome sequencing (WGS) for the current analysis. The number of Enterobacteriaceae subjected to WGS from DRH, TASH, HUCSH and Y12HMC were 113, 91, 57 and 40, respectively.

DNA Extraction, Whole Genome Sequencing (WGS) and the Identification of Resistance Genes
From all the Enterobacteriaceae, DNA was extracted manually using a QIAamp DNA Mini Kit (QIAGEN, Hilden, Germany), according to the manufacturer's instructions. DNA extractions were performed by taking 2-5 pure colonies that grew on cystine lactose electrolyte deficient agar at 37 • C for 24 h aerobically. After the extraction, the DNA concentrations were measured with QubitTM3.0 (Thermo scientific, Waltham, MA, USA). All the extracted DNA samples were kept at −20 • C until they were submitted for whole genome sequence determination.
All the Enterobacteriaceae were subjected to WGS at the Science for Life Laboratory, Solna, Sweden. From each DNA sample, 20 µL was transferred into a 96-well WGS plate. Sequencing libraries were generated using Nextera XT (Illumina kits) and short-read sequencing was run on Illumina (HiSeq 2500) systems with a 150 bp insert size paired end sequencing protocol at the Science for Life Laboratory. SPAdes (version 3.9) was used for the genome assembly. With the assembled genomes, the acquired antimicrobial resistance genes were identified using the ResFinder 4.1 web tool at the Center for Genomic Epidemiology http://www.genomicepidemiology.org/ (accessed on 16 August 2021) using a threshold of 90% and 60% coverage. Each WGS run included quality control.

Statistical Analysis
The data was prepared using a Microsoft Office Excel sheet and was imported to SPSS version 27 for analysis. The frequencies of the resistance genes, ESBL and AmpC producers and the co-occurrence of multiple ESBL genes and other variables were calculated. The binary logistic regression analysis was used to observe the associations of the sociodemographic and possible risk factors of ESBL producing Enterobacteriaceae. A p-value < 0.05 was considered as statistically significant.

Conclusions
To our knowledge, this is the first multicenter study that reported the molecular epidemiology of ESBL and AmpC producing Enterobacteriaceae among sepsis patients at four Ethiopian hospitals located in the northern, central and southern parts of the country. A very high genotypic frequency of ESBL producing Enterobacteriaceae among sepsis patients was detected. Several variants of ESBL genes were detected and the most frequent ESBL gene was bla CTX-M- 15 . In addition to the ESBL genes, diverse variants of bla TEM and bla SHV beta-lactamases were detected, where bla TEM-1B and bla SHV-187 were the most frequently detected variants in their respective families. Multiple combinations of ESBL genes were detected and most ESBL genes were concurrently detected with the bla TEM-1B and bla SHV-187 beta-lactamases. Different AmpC genes were detected in some Enterobacteriaceae where variants of bla ACT and bla CMY were commonly detected. The current findings strongly suggest the urgent need for high standard bacteriological laboratory services to guide the antibiotic treatment of sepsis and other life-threatening infections due to ESBL and AmpC producing Enterobacteriaceae. It is also helpful for effective antimicrobial stewardship so that infection control programs can be improved.  Informed Consent Statement: Written informed consent was obtained from all patients involved in the study.

Data Availability Statement:
The data presented in this study are available on request from the corresponding author. The data are not publicly available due to privacy restrictions.