Diversity of CTX-M Extended-Spectrum β-Lactamases in Escherichia coli Isolates from Retail Raw Ground Beef: First Report of CTX-M-24 and CTX-M-32 in Algeria

Introduction

Antibiotic resistance in bacteria has reached alarming levels worldwide in both hospitals and the community, resulting in frequent treatment failures. This is inherent in indiscriminate use of antibiotics in human and veterinary medicine and in agriculture, which provides selection pressure and in turn promotes horizontal genetic exchange and recombination events resulting in the emergence and spread of antibiotic resistance mechanisms. Extended-Spectrum β-Lactamase (ESBL) production is the main mechanism for β-lactam resistance in Enterobacteriaceae.^1,2 These plasmid-mediated serine enzymes inactivate all penicillins and cephalosporins and aztreonam, but not cephamycins and carbapenems, and are inhibited by clavulanic acid. ³

Originally, ESBLs were derived by point mutations from TEM-1, TEM-2, and SHV-1 penicillinases; thereafter, subsequent to the introduction of extended-spectrum cephalosporins (ESC) in the 1980s, new ESBLs have emerged, of which CTX-M is now the most prevalent and widely distributed in the world. ⁴ ESBL CTX-M would have as origin chromosomal β-lactamases of various species of Kluyvera and are divided into six phyla (CTX-M-1, CTX-M-2, CTX-M-8, CTX-M-9, CTX-M-25, and CTX-M-45) on the basis of their amino acid sequences.

Dominant variants of CTX-M differ between countries, with the exception of the common worldwide alleles CTX-M-15 and CTX-M-14. ⁴ The association of bla_CTX-M genes with mobile genetic elements, epidemic plasmids, and successful clones ensured them a rapid and wide dissemination that changed the epidemiology of antibiotic resistance worldwide.^1,2,4 Furthermore, their association with resistance to other classes of antibiotics (fluoroquinolones, aminoglycosides, and co-trimoxazole), due to common genetic platforms or resistance mechanisms, significantly restricts therapeutic possibilities.^2,5 Among ESBL-producing Enterobacteriaceae, in particular CTX-M, Escherichia coli is the most prevalent in the world, followed by Klebsiella pneumoniae. ⁴ Agriculture animals, especially poultry and cattle, are recognized as important vectors of ESBL carriers E. coli, in particular CTX-M,^6–8 and contaminated meat is thought to be a possible vehicle, among others, for their transmission to humans through consumption or handling. ⁷

Many studies have been conducted worldwide on the occurrence and epidemiology of acquired resistance to ESCs in food-producing animals and food, while very little is known about the situation in North Africa, especially Algeria, where there are no data on this issue. The objective of this study was to investigate prevalence and molecular features of ESC resistance in E. coli isolates recovered from ground beef samples collected from butcher shops in different districts of Algiers.

Materials and Methods

Results

Of 371 ground beef samples, 102 (27.5%) were found to contain cefotaxime-resistant E. coli isolates; phylogenetic analysis showed their distribution into A (n = 25, 24.5%), B1 (n = 62, 60.8%), and D (n = 15, 14.7%) phylogroups. In particular, phylogroup B2 was not present in this collection. Besides resistance to cefotaxime, antibiotic susceptibility testing showed a marked resistance to tetracycline (92.2%), cefpirome (85.3%), nalidixic acid (82.4%), ofloxacin (80.4%), pefloxacin (72.6%), and ciprofloxacin (70.6%), followed by SXT (53.9%), kanamycin (30.4%), amoxicillin/clavulanic acid (27.5%), and ceftazidime (21.6%), while low resistance rates were observed for cefoxitin (3.9%) and gentamicin (2.9%). All isolates were susceptible to imipenem, amikacin, and colistin (Table 1).

Agar dilution MICs ranged from 64 to >512 mg/L for cefotaxime, 16 to 128 mg/L for ceftazidime, 2 to >64 mg/L for ciprofloxacin, ≥1216/64 mg/L for sulfamethoxazole–trimethoprim, 64–256 mg/L for gentamicin, and 256–>512 mg/L for tetracycline. The combinations of resistance phenotypes allowed to distinguish 46 antibiotic resistance patterns, including 3–11 antibiotics, and 88.2% of isolates had a MDR phenotype, they resist to at least three antibiotic classes (Table 2).

All isolates were positive for the DDST presumptive of ESBL production. PCR detection showed bla_CTX-M genes in all isolates, 74 (72.5%) of CTX-M-1 group and 28 (27.5%) of CTX-M-9 group. A rate of 47.3% of CTX-M-1 group–producing isolates belonged to B1 phylogroup, 32.4% to A, and 20.3% to D. For CTX-M-9 group–producing isolates, 96.4% belonged to B1 and 3.6% to A.

Sequencing analysis revealed six different bla_CTX-M alleles as follows: bla_CTX-M-1 (n = 47, 46%), bla_CTX-M-15 (n = 24, 23.5%), bla_CTX-M-3 (n = 2, 2%), and bla_CTX-M-32 (n = 1, 1%) among the 74 CTX-M-1 group genes and bla_CTX-M-14 (n = 27, 26.5%) and bla_CTX-M-24 (n = 1, 1%) among the 28 CTX-M-9 group genes (Tables 2 and 3). The distribution of bla_CTX-M genes according to phylogroups showed a perfect positive association (100%) between bla_CTX-M-14 and the phylogroup B1 (p < 0.0001) and a negative association between bla_CTX-M-15 and the phylogroups A (p = 0.002) and B1 (p = 0.0005) (Table 3).

The bla_TEM-1 gene resulting in production of narrow spectrum β-lactamase was detected in 39 isolates (38.2%) in association with all CTX-M variants (Tables 2 and 3).

The ISEcp1 insertion sequence was found upstream of all bla_CTX-M genes with a spacer region of 48 bp (W sequence) for bla_CTX-M-15 (n = 24) and bla_CTX-M-1 (n = 26), 45 bp for bla_CTX-M-3 (n = 2) similar to W sequence with only one nucleotide substitution, and 79 bp to 82 bp (sequences X + W) for bla_CTX-M-1 (n = 20) and bla_CTX-M-32 (n = 1). In two isolates harboring bla_CTX-M-1 and bla_CTX-M-32, ISEcp1B was found disrupted by an ISKpn26 insertion sequence (IS5 family).

Conjugation assays performed on all isolates allowed the transfer of bla_CTX-M-1 (n = 20), bla_CTX-M-15 (n = 4), bla_CTX-M-3 (n = 2), and bla_CTX-M-14 (n = 1) genes to E. coli BM21. Plasmid replicon analysis showed the association of bla_CTX-M-1 with IncI1 plasmids of about 128 kb, bla_CTX-M-15 with IncI1 and IncK+B/O plasmids of about 128 kb and 80 kb, bla_CTX-M-3 with IncK plasmids of about 95 kb, and bla_CTX-M-14 with IncF1B or IncK plasmids of about 150 and 48 kb, respectively (Table 4).

The screening for PMQR determinants of all isolates showed the presence of qnrS1, qnrB, and aac(6′)-Ib-cr in 16 (15.7%), 1 (1%), and 2 (2%) isolates, respectively (Tables 2 and 3), while qnrA, qnrC, qnrD, qepA, oqxA, and oqxB were absent. It should be noted the presence of qnrS1 (n = 6) and qnrB (n = 1) genes in quinolone susceptible isolates. qnrS1 gene was cotransferred with bla_CTX-M-15 from one isolate (Table 4). Sequence analysis of the QRDR of gyrA and parC genes showed mutations in 71/72 ciprofloxacin-resistant isolates (98.6%), while no mutations were observed in pefloxacin- and/or ofloxacin-resistant isolates. The amino acid substitution pattern “GyrA: S83L+D87N, ParC: S80I” (46.5%) and the single substitution “ParC: S80I” (42.3%) were predominant, followed by the mutation patterns “GyrA: S83L+D87Y, ParC: S80I” (2.8%), “GyrA: S83L+D87N, ParC: S80I+E84G” (2.8%), “GyrA: S83A+D87Y, ParC: S80I” (1.4%), “GyrA: S83L+D87N, ParC: E84K” (1.4%), “GyrA: S83I, ParC: S80R” (1.4%), and the single mutation “ParC: E84K” (1.4%) (Table 2).

Among the 94 tetracycline-resistant strains, tet genes were detected in 90 isolates (95.7%), with a prevalence of 81.1% (n = 73) for tetA, 8.9% (n = 8) for tetB, and 10% (n = 9) for tetA+tetB (Tables 2 and 3). tetA gene was cotransferred with bla_CTX-M-1 and bla_CTX-M-15 from 14 and 2 isolates, respectively (Table 4).

The sul genes were present in 53 of 55 SXT-resistant isolates (96.4%), the rates of sul1, sul2, sul3, sul1+sul2, sul2+sul3, and sul1+sul3 were of 9.4% (n = 5), 39.6% (n = 21), 18.9% (n = 10), 20.8% (n = 11), 7.5% (n = 4), and 3.8% (n = 2), respectively (Tables 2 and 3). sul2 gene was cotransferred with bla_CTX-M-15 and bla_CTX-M-1 from six and one isolates, respectively, and sul3 with bla_CTX-M-15 from one isolate (Table 4).

In total, 47 of 55 (85.5%) SXT-resistant isolates harbored dfr genes. The gene clusters dfrA1, dfrA5, dfrA7, dfrA8, and dfrA12 were detected in 14.9% (n = 7), 14.9% (n = 7), 17% (n = 8), 4.3% (n = 2), and 27.7% (n = 13) of isolates, respectively. The concomitant presence of dfrA5+dfrA12, dfrA1+dfrA5, dfrA7+dfrA12, dfrA5+dfrA7, and dfrA1+ dfrA5+ dfrA7 was observed in 6.4% (n = 3), 4.3% (n = 2), 4.3% (n = 2), 4.3% (n = 2), and 2.1% (n = 1) of isolates, respectively (Tables 2 and 3). The association sul+dfr was found in 83.6% of SXT-resistant isolates. dfrA7 gene cluster was cotransferred with bla_CTX-M-1 from three isolates and dfrA5 with bla_CTX-M-15 or bla_CTX-M-1 from one isolate (Table 4).

Class 1 and class 2 integrons were detected in 69 (67.6%) and 10 (9.8%) isolates, respectively, while class three integrons were absent. Fifty-one (73.9%) of class 1 integrons lacked the 3′-conserved sequence (3′-CS) comprising the qacEΔ1-sul1 genes. Class 1 integrons were cotransferred with bla_CTX-M-15 and with bla_CTX-M-1 from two and three isolates, respectively (Tables 2 and 3).

The clonal relationship between 94 isolates investigated by PFGE revealed 72 DNA profiles. Similarities between profiles were observed within 33 isolates mostly originating from different districts, they are divided into 10 clusters of 2 isolates and two clusters of 6 isolates (PFGE profile P15) and 7 isolates (PFGE profile P27) (Table 2). The latter harbored CTX-M-14 and belonged to phylogroup B1. As regards resistance profiles, the six-isolate cluster had a same profile, while the seven-isolate cluster presented differences in their resistance profiles (Table 2).

Discussion

In this study, we investigated the prevalence and molecular features of resistance to ESC in E. coli recovered from ground beef samples at retail in Algeria. A total of 102/371 (27.5%) ground beef samples were found to contain ESBL-producing cefotaxime-resistant E. coli. The occurrence of ESBL-carrying E. coli in beef varies among countries, and comparisons are difficult because of the variety of methods used in studies; rates of 59%, 26%, 7%, 0%, and 0% were reported in studies from Spain, Tunisia, Turkey, Switzerland, and Czech Republic, respectively.^10,24–27 The contamination of meat with ESBL-producing E. coli may have resulted from their fecal carriage by animals exposed to antibiotics for prophylaxy and growth promotion.^28,29

Phylogenetic grouping of isolates showed that B1 was the main phylogroup followed by phylogroup A and to a lesser extent by D. This finding matches with what was observed in CTX-M–producing E. coli from livestock (including cattle).^8,30 Thus, even though other hypotheses may include cross-contamination of human origin through the handling of the beef samples before getting at retail, those phylogroups, and particularly the absence of phylogroup B2 which is prevalent in humans, rather argue for an animal origin.

Besides beta-lactam resistance, high prevalence resistance toward fluoroquinolone, SXT, and tetracycline and to a lesser extent to kanamycin was observed, and 88.2% of the isolates were MDR phenotype; this may have resulted from co-selection with ESBL due to associated chromosomally- or plasmid-encoded resistance mechanisms.^5,8,24

All ESBLs detected in this study were members of CTX-M family, in accordance with the worldwide prevalence of CTX-Ms in Enterobacteriaceae, particularly E. coli. ⁴ Among the CTX-M variants identified in our study, CTX-M-1 (46%), CTX-M-14 (24.5%), and CTX-M-15 (23.5%) were largely predominant; this is consistent with the frequent detection of CTX-M-1 and CTX-M-14 variants in food-producing animals and foods.^{6,7,10,24,25,30} Of note, cattle are the animal group, compared to others, where CTX-M-15 has been recognized at a significant prevalence.^8,31,32 Therefore, it may not be surprising to find CTX-M-15 producers in those meat samples as they originate from beef, and this also constitutes an argument that this colonization may rather originate from the animal sector. The most CTX-M variants detected up until now in Algeria were of CTX-M-1 group; these are CTX-M-3 and CTX-M-28 in clinical isolates,^15,33 CTX-M-15 in clinical and chicken isolates,^15,34 and CTX-M-1 in chicken isolates. ³⁵ The second group described was CTX-M-9 through only the CTX-M-14 variant. ³⁶ CTX-M-24 of CTX-M-9 group and CTX-M-32 of CTX-M-1 group found in our isolates are therefore reported for the first time in Algeria. These findings are indicative of the potential role of cattle and beef meat in the dissemination of CTX-M ESBLs, particularly those of CTX-M-9 group, which up until now are rare in clinical setting in Algeria. The association of bla_TEM-1 with bla_CTX-M is very common; they often coexist on the same plasmids. ¹⁵

The described associations between CTX-M variants and phylogroups vary according to studies, depending on strains and their origin^8,30,37; it was reported that the emergence of a type of ESBL results from its interaction with the genetic background of the strain and the surrounding selection pressure. ³⁸ In this study, CTX-M-14 was 100% associated with phylogroup B1; this genotype seems well established in beef; it certainly contributes in the spread of resistance through food chain.

All bla_CTX-M genes were found flanked upstream by an ISEcp1B insertion sequence known to be involved in the expression and mobility of these genes ²¹ ; this association at high rate is well documented. ⁴ The sequences of spacer regions between ISEcp1B and bla_CTX-M of group 1 previously found in clinical and environmental Algerian strains were W and V + W^15,20; this is the first time that X + W and the 45-bp sequences were described in Algeria, while they were previously reported in other countries.^10,39,40 This finding suggests a foreign origin of at least some of our strains, possibly carried by imported animals. In two isolates harboring bla_CTX-M-1 or bla_CTX-M-32, ISKpn26, first described in Klebsiella pneumoniae (GenBank: NC_016845), was localized inside ISEcp1B; such a genetic environment was already reported in E. coli (GenBank: AB976580).

The conjugation transfer of bla_CTX-M genes was positive for only 26.5% of our isolates, when plasmid horizontal transfer is known as an important contributor in the epidemiology of these genes^1,2,4; in this regard, transfer rates of 86% and 100% of bla_CTX-M gene were reported.^41,42 Our result could be explained by a chromosomal localization of bla_CTX-M genes ⁴ or a very low transfer frequency of plasmids.

Plasmid replicon analysis showed that all transferred bla_CTX-M-1 were associated with IncI1 plasmid group, which is among the main groups involved in the spread of bla_CTX-M.^1,4,6 Many studies reported a close association between CTX-M-1 and IncI1 plasmid types in human, animal, food, and environmental isolates, and the emergence of plasmid IncI1/bla_CTX-M-1 was first observed in animals and subsequently in humans.^1,6,7,42,43 Our results support the role of IncI1 plasmids in the dissemination of bla_CTX-M-1 and the food chain in their transmission to humans. IncI1 plasmids are considered to be epidemic resistance plasmids; their successful dissemination is probably related, among others, to plasmid-intrinsic factors such as addiction systems (e.g., pndAC and relBE systems) and colonization factors (e.g., type IV pili).^1,44 bla_CTX-M-15 was found associated with IncI1 and IncK+B/O plasmids; the association CTX-M-15-IncI1 was already described, especially in an outbreak of E. coli O104:H4 in Northern Europe, ⁶ whereas linkage of CTX-M-15 with a multireplicon plasmid IncK-B/O is reported for the first time in this study. The multireplicon plasmids have advantage of being stable even in the presence of a second plasmid carrying a homologous replicon. bla_CTX-M-14 was found associated either with IncF1B or IncK plasmids; a close association between bla_CTX-M-14 and IncK plasmids was previously documented in different studies; indeed, this plasmid type was described as being behind the diffusion of CTX-M-14 among animals and humans from different European countries.^1,6,7,37 bla_CTX-M-3, described as dominantly harbored by IncL/M and IncI1, is reported for the first time on an IncK plasmid in this study.

A rate of 80.4% of the isolates was resistant to fluoroquinolones, the association between ESBLs, especially CTX-M, and fluoroquinolone resistance was commonly reported.^2,7 A rate of 18.6% of isolates harbored PMQR determinants with a predominance of qnrS1; this is consistent with previous reports. ⁴⁵ Some strains harboring qnrB or qnrS1 were susceptible to quinolones, this is in accordance with the very low level resistance (below the resistance threshold, according to EUCAST-CASFM) conferred by these genes; however, PMQR determinants can facilitate the emergence of high-level resistant mutants. ⁴⁶ Furthermore, animal experiments showed that qnr genes can reduce bactericidal activity of ciprofloxacin like a gyrA mutation. ⁴⁷

Almost all ciprofloxacin-resistant isolates (98.6%) had mutations in gyrA and/or parC genes, all substitutions detected were already described in E. coli, ⁴⁸ and the prevalent mutation profiles were the pattern “GyrA: S83L+D87N, ParC: S80I” followed by the single substitution ParC: S80I. “GyrA: S83L+D87N, ParC: S80I” was previously reported among the predominant profiles in clinical fluoroquinolone-resistant isolates in Algeria ⁴⁹ and worldwide.^50,51 The abundance of this genotype may be related to elimination of fitness cost associated with S83L+D87N by additional ParC S80I substitution. ⁵² High levels of resistance up to 64 mg/L were observed with the single ParC S80I substitution; the same finding was reported. ⁵³ However, previous studies showed that this mutation alone is not associated with any increase in the MIC of ciprofloxacin ⁵² ; thus, the resistance levels observed are probably related to other mechanisms such as decreased expression of outer membrane porins or overexpression of multidrug efflux pumps.

Tetracycline is a broad-spectrum antibiotic used with other first-generation tetracyclines, such as chlortetracycline and oxytetracycline as animal growth promoter. Second-generation tetracyclines, such as minocycline and doxycycline, are commonly used in human and veterinary medicine. Almost all of our tetracycline-resistant isolates harbored tetA and/or tetB genes with a clear predominance of tetA gene, in agreement with the fact that generally efflux mechanisms are most prominent, and the tetA gene is commonly encountered in E. coli isolated from human, animals, and animal-derived foods in many countries.^54–59 tetB gene was found at low prevalence in our strains; however, it confers a high level of resistance, in particular to minocycline and doxycycline. ⁵⁸

Commonly, sulfonamides and trimethoprim resistance in E. coli arise from the acquisition of sul and dfr genes, encoding drug-resistant variants of dihydropteroate synthases and dihydrofolate reductases. ⁶⁰ Consistent with our results, sul2 gene was reported as the most prevalent in E. coli from human, animals, and animal-derived foods.^{29,54,55,57,59} E. coli carrying sul2 gene was described as able to colonize both animals and humans, and animal strains may be implicated in human infections such as UTI and septicemia. ⁶¹ sul3 gene, reported as rare, was detected at a substantial rate in our isolates, almost equivalent to that of sul1 gene usually reported as most frequent after sul2.^54,57,59 Concerning trimethoprim resistance genes, clusters dfrA1, dfrA5, dfrA7, dfrA8, and dfrA12 detected in our study are among the most reported.^10,29,59,62

Integrons constitute a mechanism for acquisition and expression of resistance genes; they play an important role in the emergence and spread of MDR. ⁶³ Over half of our isolates harbored class 1 integrons, while class 2 integrons were detected in only about 10%; this is in accordance with the predominance of class 1 integrons in E. coli from food animals and foods.^9,64,65 The use of antimicrobials, such as tetracycline and sulfonamides, as growth promoters for farm animals as beef cattle, may promote the presence and maintenance of class 1 integrons in E. coli. ²⁹ The majority of detected integrons (73.9%) lacked the 3′-conserved sequence comprising the qacEΔ1-sul1 genes; this truncated structure was already described, it generally contains a sul3 at the 3′-end and is linked to various insertion sequences probably responsible for this rearrangement. ⁶⁶

The PFGE analysis revealed a genetic diversity within the 94 typed isolates; however, clonality was observed among some isolates of which two clusters of six and seven CTX-M-14–producing isolates were unrelated to districts; these findings are indicative of a possible clonal spread of CTX-M-14 producers. The difference in resistance genetic content of some clonal isolates may be related to changes due to selective pressure or genetic exchange.

Conclusions

To the best of our knowledge, this is the first report to study bacterial antibiotic resistance in fresh retail beef meat in Algeria; it highlighted the relatively massive presence of E. coli isolate producers of various CTX-M ESBLs. The CTX-M variants, CTX-M-24 and CTX-M-32, are reported for the first time in Algeria. Almost all of the isolates were MDR, with resistance to major antibiotics used in human medicine such as broad spectrum beta-lactams, fluoroquinolones, aminoglycosides, and trimethoprim–sulfamethoxazole. Evidence of both clonal and horizontal dissemination of resistance genes was also demonstrated. These findings are indicative of possible failures in the management of the use of antibiotics in the food sector and/or lack of hygiene during the food process and at retail. From a one health perspective, this is of considerable importance in view of the possibility of antimicrobial resistance transfer between animals and humans.