Comparative Genome Analysis of Extended-Spectrum-β-Lactamase-Producing Escherichia coli Sequence Type 131 Strains from Nepal and Japan

The global spread of ESBL-E. coli has been driven in large part by pandemic sequence type 131 (ST131). A recent study suggested that, within E. coli ST131, certain sublineages have disseminated worldwide with little association with their geographical origin, highlighting the complexity of the epidemiology of this pandemic clone. ST131 bacteria have also been classified into four virotypes based on the distribution of certain virulence genes. Information on virotype distribution in Asian ST131 strains is limited. We conducted whole-genome sequencing of ESBL-E. coli ST131 strains collected in Nepal and Japan, two Asian countries with a high and low prevalence of ESBL-E. coli, respectively. We systematically compared these ST131 genomes with those reported from other regions to gain insights into the molecular epidemiology of their spread and found the distinct phylogenetic characteristics of the spread of ESBL-E. coli ST131 in these two geographical areas of Asia.

Nepal and India. The ST131 prevalence of 14% found among ESBL-E. coli isolates from Japan is also similar to that reported in a previous study where 21% of 130 ESBL-E. coli isolates collected between 2002 and 2003 from Japan were ST131 (7). However, the prevalence was lower than that reported in a more recent study conducted in Japan, in which 37% of 581 ESBL-E. coli isolates collected from 2001 to 2010 were identified as ST131 (8). This discrepancy might be attributed to the differences in the year of collection, geographical locations in the country, and methods used to identify ST131 (PCR-based screening versus WGS).
fimH alleles and H30Rx sublineages in E. coli ST131. fimH alleles and H30Rx sublineages in E. coli ST131 were determined and compared among various geographical regions (Table 1). fimH30, especially fimH30Rx comprised the majority of the fimH alleles in various geographical regions, followed by fimH41 and fimH22. fimH27 was observed only in E. coli ST131 from Africa. We found that all 54 ESBL-E. coli isolates from Nepal belonged to the H30Rx sublineage. There is little data on the prevalence of the H30Rx sublineage in South Asia. A previous study that included only a few ST131 strains from India revealed that all of them belonged to clade C (i.e., H30) and produced CTX-M-15 (3). Our result is in accordance with the high prevalence of CTX-M-15producing ESBL-E. coli reported among travelers returning from the Indian subcontinent (6). However, the limited isolate collection period in one facility might have affected the clonal distribution of ESBL-E. coli ST131 isolates from Nepal in our study. H30R (non-Rx) were more prevalent among ESBL-E. coli ST131 isolates from Japan compared to isolates from Nepal, although more than 60% of ESBL-E. coli ST131 isolates from Japan still belonged to H30Rx. A previous report from Japan suggested a higher prevalence of H30R (non-Rx) and H22 among ESBL-E. coli ST131 isolates collected from 2001 to 2012 in 10 Japanese acute-care hospitals located in Kyoto, Shiga area, which is more than 460 km (286 miles) away from the hospital in this study (9). The relatively high prevalence of H30R (non-Rx) and H22 observed in this study in the United States and Europe, respectively, were consistent with previous reports (3,10). Antimicrobial susceptibility and antibiotic resistance genes. Antimicrobial susceptibility data were available for 54 and 11 ESBL-E. coli ST131 isolates from Nepal and Japan, respectively (Fig. 1). The susceptibilities of the 105 ESBL-E. coli isolates (including 54 ESBL-E. coli ST131 isolates described in this study) from Nepal were described elsewhere (5). The susceptibilities to trimethoprim-sulfamethoxazole, gentamicin, and levofloxacin were not significantly different in isolates from Nepal and Japan. All ESBL-E. coli ST131 isolates from Nepal and Japan were susceptible to cefmetazole and fosfomycin.
The frequency of resistance genes among the E. coli ST131 isolates is summarized in Table 2. More than 94% (n ϭ 51) and 66% (n ϭ 8) of ESBL-E. coli ST131 isolates were positive for bla CTX-M-15 in Nepal and Japan, respectively. Overall, 50 (70%) of the 71 bla CTX-M-15 -positive isolates in the entire cohort were also positive for bla OXA-1 and   aac(6=)-Ib-cr (encoding an aminoglycoside/fluoroquinolone acetyltransferase); of these 71 isolates, 38 isolates were obtained from Nepal, 5 isolates from Japan, and 6 isolates from Europe. The aforementioned study from Japan (9) revealed higher prevalence of bla CTX-M-14 and bla CTX-M-27 in addition to bla CTX-M-15 among ESBL-E. coli isolates collected from another area in Japan. A study from Korea on E. coli isolates collected from 2012 to 2013 suggested that the majority of H30Rx isolates harbored bla CTX-M-15 , whereas about half of the H30 non-Rx isolates harbored bla CTX-M-14 or bla CTX-M-27 (11). Geographical differences and different times of these studies may account for these discrepancies.
Virotypes and virulence-associated genes. The distribution of virotypes and virulence-associated genes among the E. coli ST131 strains is summarized in Table 3. The predominant virotypes differed across the geographical regions. The majority of E. coli ST131 isolates from the United States and half of those from Japan belonged to virotype C, whereas almost half of the isolates from Nepal belonged to virotypes other than A, B, C, D, and E. Two-thirds of E. coli ST131 isolates from Europe belonged to either virotype C or D. The virulence-associated genes such as iha, sat, fyuA, traT, ompT, and malT were highly prevalent among isolates from most geographical regions. The papGII gene was frequently observed only among isolates from Nepal and Tanzania, and hlyA and cnf1 were common only in E. coli ST131 isolates obtained from Africa. Most of the virotypes obtained in our study belonged to virotype C (Table 1). Our finding that virotype C is the most prevalent virotype among E. coli ST131 isolates is consistent with previous reports on virotype distribution in ESBL-E. coli ST131 (4, 12). As previously suggested (4,12), our results also suggested the association between virotype C and H30Rx sublineage. In our study, more isolates were identified as "other" virotype than previous studies (4,12), which might be due to the different methodology used (e.g., WGS versus PCR). The majority of the ESBL-E. coli ST131 isolates (n ϭ 28 [52%]) from Nepal were positive for papGII and sat but were negative for hlyA, and thus, were not categorized into any of the previously described virotypes (A to E) (13).
Plasmid replicon types. The distribution of plasmid replicon types was similar across geographical regions except that FIA was prevalent in Nepal and the United States and considerably less prevalent in Europe; other geographical regions showed intermediate prevalence (Table 4). FII was commonly found among ESBL-E. coli ST131 isolates from Nepal, whereas it was considerably less prevalent in Japan. IncP was detected in 40% of E. coli ST131 isolates from Africa, but it was very rarely found or not found in isolates from other areas.
The association of bla CTX-M- 15 and IncF replicon has been reported previously (14), and the H30Rx sublineage was created by introduction of IncF (15). We previously found that a plasmid resembling pEC958 and harboring FIA and FII (16) was present in approximately 80% of the ESBL-E. coli ST131 isolates from Nepal (5). The high prevalence of IncF (IncFIA, FIB, and FII) observed among ESBL-E. coli ST131 isolates from Nepal is consistent with these previous reports and is reasonable considering the high prevalence of H30Rx among ESBL-E. coli ST131 isolates.

Phylogenetic analysis of E. coli ST131 in different geographical regions.
In the phylogenetic analysis based on WGS (Fig. 1), the majority of ESBL-E. coli ST131 isolates from Nepal clustered together, whereas those from Japan were more diverse. Thus, ESBL-E. coli ST131 may have been introduced into Japan more sporadically over time. Strict requirements for antibiotic prescription in Japan might have resulted in relatively low selective pressure, along with more advanced medical and social infrastructure. Carriers of ESBL-E. coli are usually placed under contact precaution in health  care facilities in Japan, including the hospital in this study, to minimize transmission. These factors might explain at least in part the differences in the prevalence and clonality of ST131 between these two Asian countries. In contrast, prevalent clonal spread of ESBL-E. coli appears to have occurred in Nepal, where poor infection control practices and sanitation might facilitate dissemination of antimicrobial-resistant strains.  In Nepal, antibiotics can be purchased in the community at general retail stores and pharmacies. According to recent reports, 80% of the drugs are purchased outside the government-supplied health system (17,18). In addition, inappropriate prescription occurs in up to 40% of patients (17,18). E. coli ST131 isolates from the United States are distributed across the phylogeny, even those isolates that were collected in the same year. The wide diversity of population in the United States might explain in part this phylogenetic characteristic. E. coli ST131 isolates from Europe (including three E. coli ST131 isolates from Lebanon) could be roughly divided into four clusters, with one major cluster consisting mainly of There are several limitations to this study. Since ESBL-E. coli ST131 isolates from Nepal and Japan were collected during a relatively short period of time, epidemiologically related isolates may have been included in the analysis, and the trends across years could not be elucidated. However, as shown in Fig. 1, the distribution of ESBL-E. coli ST131 isolates in each region suggests that the contribution of such clonal isolates was unlikely to have been remarkable. Due to the small number of isolates from Japan,