1 INTRODUCTION

Antimicrobial resistance (AMR) has become a major concern threatening public health (Nowakiewicz et al., 2019). For many decades, antibiotics are widely used in animal and human areas, leading to the worldwide resistance phenomenon. Indeed, bacteria have always been able to adapt by developing or acquiring mechanisms of resistance (Duval et al., 2019). In response to selective pressure and to survive antibiotic exposure, resistance occurs generally through the acquisition of genes located on mobile elements (Argudin et al., 2017; Strauss et al., 2015). Despite awareness associated with restrictive measures in animal production, the effects of the intensive use of drugs seem to persist lengthily, impacting numerous environments and ecological niches (Argudin et al., 2017; Nowakiewicz et al., 2019). Zoonotic, as well as commensal bacteria, became resistant providing them with a selective advantage (Argudin et al., 2017; Perreten et al., 2005) and sharing a potential risk of dissemination of AMR genes (De Jong et al., 2019; Nowakiewicz et al., 2019; Perreten et al., 2005).

Commensal bacteria, such as Enterococcus spp. are natural inhabitants of the gastrointestinal tract of healthy animals and humans. Enterococci can also be considered opportunistic pathogens when found related to human infections. They can persist in the environment and survive in different ecological niches such as soil, water, food items, and sewage (Nowakiewicz, 2019; Osman et al., 2019; Raza et al., 2018; Torres et al., 2018). Enterococci have emerged as one of the four most prevalent nosocomial human pathogens worldwide, especially due to their virulence, ability to form a biofilm, and intrinsic resistance (Leite-Martins et al., 2015; Raza et al., 2018). Notably, Enterococcus spp. are intrinsically resistant to a number of antimicrobials including trimethoprim-sulfamethoxazole, vancomycin (Enterococcus gallinarum, Enterococcus casseliflavus, and Enterococcus flavescens), streptogramins (Enterococcus faecalis) and exhibit low-level resistance to ß-lactams and aminoglycosides (Argudin et al., 2017; Torres et al., 2018; Zaheer et al., 2020). Since the 1980s, AMR enterococci have been the leading cause of hospital-acquired bloodstream and urinary tract infections, mainly through biofilm formation on catheters and implanted medical devices (Argudin et al., 2017; De Jong et al., 2019; Mercuro et al., 2018; Raza et al., 2018; Torres et al., 2018; Zaheer et al., 2020). Particularly, the majority of enterococcal infections are caused by E. faecalis and Enterococcus faecium, the most common species encountered in the human gut (Argudin et al., 2017; Mercuro et al., 2018; Raza et al., 2018; Torres et al., 2018). E. faecium has become a prominent cause of nosocomial infections often characterized by high-level resistance to multiple antibiotics (De Jong et al., 2019). Due to the high plasticity of enterococcal genomes, transfer and acquisition of AMR determinants in enterococci and other Gram-positive bacteria are then facilitated (Leite-Martins et al., 2015; Nowakiewicz et al., 2019).

Staphylococcus spp. are commensal bacteria widely found on the skin or mucosal surfaces of animals and humans (Alharbi, 2019; Bortolaia et al., 2016; Craft et al., 2019; Dastgheyb & Otto, 2015; Wendlandt et al., 2013). Despite preventing colonization by pathogenic bacteria (Alharbi, 2019), staphylococci are opportunistic pathogens often responsible for chronic and severe nosocomial infections (Bortolaia et al., 2016; Dastgheyb & Otto, 2015). Specifically, Staphylococcus aureus is one of the most pathogenic bacteria associated with human and animal diseases causing persisting skin and soft tissue infections (SSTIs), infectious endocarditis, septic arthritis, and osteomyelitis (Alharbi, 2019; Craft et al., 2019; Dastgheyb & Otto, 2015). The colonization of skin or mucosal surfaces by methicillin-resistant S. aureus (MRSA) and its dissemination in healthcare settings represents a global health issue (Holmes et al., 2015; Watkins et al., 2019) especially due to its capacity to acquire new AMR and to spread rapidly (Alharbi, 2019; Holmes et al., 2015). Indeed, methicillin resistance was first observed among clinical isolates before its rapid spread to the community (Turner et al., 2019). Staphylococcus spp. may exchange resistance determinants to numerous other bacteria in the same animal or human host, or between hosts by direct contact or through excretions such as sneezing, coughing, or licking (Wendlandt et al., 2013). Even if MRSA-associated infection rates have declined, these human infections are still problematic since they are characterized by broadening AMR and high rates of hospitalization and mortality (Purrello et al., 2016).

Over time, AMR has widely spread, consequently leaving healthcare personnel with last-line antibiotics (i.e., vancomycin, linezolid, and daptomycin) as the preferred means, and sometimes the only options to treat multidrug- resistant (MDR) infections (Raza et al., 2018; Sadowy, 2018; Torres et al., 2018; Zaheer et al., 2020). Resistance to last-resort drugs was already observed as well (Azhar et al., 2017; Doern et al., 2016; Holmes et al., 2015; Purrello et al., 2016) and among various bacteria, mainly reported in clinical settings (Argudin et al., 2017; Zaheer et al., 2020). Specifically, due to the alarming worldwide emergence (Osman et al., 2019; Raza et al., 2018), vancomycin-resistant enterococci (VRE) was ranked as a pathogen of high priority by the World Health Organization (WHO) (Wist et al., 2020). Due to its ability to confer high levels of vancomycin-resistance, vanA is the most reported gene among clinical VRE (Azhar et al., 2017; Turner et al., 2019; Watkins et al., 2019) and widely spread to other co-infecting bacteria such as S. aureus. Despite restricted use, resistance to linezolid has been reported in various species, strains, and settings (Sadowy, 2018). Determinants associated with linezolid resistance such as the highly mobilizable cfr coding for a ribosomal methyltransferase conferring cross-resistance to phenicols, lincosamides, oxazolidinones, pleuromutilins, and streptogramin A or the oxazolidinone-phenicol transferable resistance gene optrA have been identified in staphylococci, enterococci and other Gram-positive species isolated from both humans and animals (Azhar et al., 2017; Doern et al., 2016; Sadowy, 2018; Timmermans et al., 2022a; Torres et al., 2018). The most recently described poxtA mediating decreased susceptibility to oxazolidinones, phenicols, and tetracyclines has also been detected in different Gram-positive bacteria of animal and human origin (Brenciani et al., 2018; Ruiz-Ripa et al., 2020; Sadowy, 2018; Timmermans et al., 2022a).

In summary, AMR is not restricted to a particular species or host. It became essential to consider all bacteria as a potential pool of resistance determinants possibly transferable to other pathogenic or commensal bacteria both in animals and humans (Argudin et al., 2017; De Jong et al., 2019; Osman et al., 2019; Perreten et al., 2005; Torres et al., 2018; Zaheer et al., 2020). Therefore, it is important to identify AMR determinants spreading in humans and animals (Perreten et al., 2005; Strauss et al., 2015) and to consider them as a One Health AMR pool. The complexity of AMR and particularly the cross-resistance phenomenon, that is resistance to multiple distinct antimicrobial classes conferred by a single molecular mechanism, requires monitoring all putative main sources of AMR at the genetic level.

We describe an in-house developed array targeting major AMR genes of Gram-positive bacteria and allowing their rapid and efficient detection all at once at reduced costs. In this study, we aimed to target AMR genes commonly found and particularly shared among Enterococcus and Staphylococcus spp. The diversity of probes per antimicrobial class reflects phenotypes observed during several years of field monitoring and pointed out the most critical (e.g., glycopeptides) or emergent ones (e.g., linezolid) to investigate. The presence of these AMR genes was studied in a collection of 124 enterococci and 62 staphylococci isolated from healthy livestock animals through the official Belgian AMR monitoring during the period 2018–2020, to decipher the genetic nature of this pool of AMR and thereby provide useful data to understand how resistance can be acquired and maintained in these bacteria.

2 MATERIALS AND METHODS

3 RESULTS

In this study, a bead array based on AMR genes was developed and validated on reference strains (see Table A1) to characterize the genetic nature of resistance expressed by Gram-positive bacteria. This array was used to detect AMR genes in a selection of enterococci and staphylococci isolated from livestock animals in Belgium. A total of 186 Gram-positive bacteria including 57 E. faecalis, 67 E. faecium, and 62 MRSA were screened for the presence of the AMR genes detailed in Table 1. Altogether, 27 out of the 41 AMR probes tested positive on the 186 selected animal isolates (see Tables 3 and 4). In most cases, resistance profiles correlated with markers detected with the array. Results of this AMR gene screening are presented and discussed hereafter per antimicrobial class, followed by whole genome sequencing (WGS) analysis results conducted to further investigate linezolid and vancomycin-resistant isolates and a few isolates resistant to five or more antimicrobials.

4 DISCUSSION

Among the entire collection of 62 staphylococci and 124 enterococci investigated, a total of 91.3% (n = 718/786) resistance profiles demonstrated experimentally could be associated with the AMR gene targeted by the array, namely 89.2% (n = 305/342) in staphylococci and 93.0% (n = 413/444) in enterococci. This result supports that the most common resistance genes were targeted by our array since a genetic marker could be associated with more than 90.0% of the resistant phenotypes leaving only a few without genetic information.

This study highlighted the coexistence of both lsaA and lsaE in 20/57 (35.1%) of the E. faecalis isolates, suggesting the presence of a selection pressure since lsa genes are involved in other phenotypic resistances such as pleuromutilin (tiamulin) and lincosamide (clindamycin) resistance as reported in other studies (Feßler et al., 2018; Hollenbeck & Rice, 2012). However, antimicrobial susceptibility for these antimicrobial classes is not monitored in enterococci in official surveillance programs. In addition, lsaE was the most observed gene in MRSA with 29 isolates showing resistance to Q/D (n = 29/37; 78.4%), as well as to tiamulin (n = 29/35; 88.6%) and clindamycin (n = 29/57; 50.9%). Such pleuromutilin-lincosamide-streptogramin A resistance mediated by lsaE as found in our study among Staphylococcus spp. and Enterococcus spp. isolated from animals was reported elsewhere in animals and also in humans (Feßler et al., 2018; Schwarz et al., 2018; Wendlandt et al., 2013).

Overall, ermB was the most prevalent macrolide resistance gene in our collection (in 79.6% of all erythromycin-resistant isolates, n = 113/142) and particularly among enterococci (n = 98/106 vs. n = 15/36 staphylococci), as described in Frye & Jackson (2013). The erm genes were more frequently observed in enterococci than staphylococci (97.2% vs. 77.8%) as first described in Jensen et al. (1999). Interestingly, other studies (Jensen et al., 1999; Petinaki & Papagiannitsis, 2019) also reported ermC mostly in staphylococci and in rare cases in enterococci.

lnuB was the second most observed lincosamide resistance gene in staphylococci as found in 45.6% of isolates (n = 26), however, it has been mostly described in enterococci (Cho et al., 2020; Feßler et al., 2018) and rare cases in staphylococci (Li et al., 2013; Lozano et al., 2012; Wendlandt et al., 2013).

All tetracycline-resistant isolates of this study harbored at least one of the tested genes, namely tetL, tetM, tetK, and/or tetO, with tetM being the most prevalent one as reported elsewhere (Argudin et al., 2017; Cho et al., 2020; Frye & Jackson, 2013; Schwarz et al., 2018). So, tetL was always found in combination with other tet genes, particularly with tetM which was adjacent in 72.0% (n = 18/25) of the isolates investigated by WGS. This result suggests that tetM and tetL might be transferable together thereby spreading two different mechanisms of resistance, ribosomal protection, and proteins efflux, respectively.

In this study, at least one linezolid-resistance gene was detected in each LZD-R strain: optrA, poxtA, or a combination of both genes in 63.6%, 9.1%, and 27.3% of LZD-R enterococcal isolates, respectively. These genes were also found in LZD-susceptible isolates characterized by a MIC of 4 mg L^-1 (n = 9/15), 2 mg L^-1 (n = 5/60), or 1 mg L^-1 (n = 2/27). These results support a recent warning from EFSA suggesting that all strains displaying MICs ≥ 4 mg L^-1 for linezolid and exhibiting resistance to the other compounds typically conferred by cfr, should be screened for this gene (European Food Safety Authority & European Centre for Disease Prevention and Control, 2022, p. 116). Perhaps it should be considered to extend it to optrA and poxtA screening as well, especially as they have been mostly found in enterococci (Timmermans et al., 2022a). In line with the observation of optrA or poxtA in LZD-susceptible isolates, previous studies (Dejoies et al., 2020; Timmermans et al., 2022a) demonstrated that a 48 h-incubation conducted during susceptibility testing enhanced the detection of isolates carrying linezolid-resistance determinants suggesting that optrA and poxtA might be inducible. However, more experiments are required at this point. Few chloramphenicol susceptible isolates were also characterized by the presence of a resistance gene such as fexA (n = 4) and cat_pC194 (n = 7) as described for LZD-susceptible isolates here above. cat_pC194, cat_pC221, and cat_pC223 from the cat enzyme family as well as fex exporters are inducible (Schwarz et al., 2016), which could explain the variations observed among MICs. This highlights the limitations of AST based on phenotypic cut-offs for the screening of AMR genes, in particular for such inducible genes. In addition, future genetic investigation of chloramphenicol phenotypic resistance may partially rely on the detection of fexB, a member of the fex exporters family already reported in enterococci (Argudin et al., 2017; Schwarz et al., 2018) and found in 10 sequenced enterococci of this study. Interestingly, the concomitant presence of optrA and fexA was observed in 23 of our isolates independently of phenotypic resistance profiles. Particularly, optrA was found close to fexA in 14 of the 16 isolates investigated by WGS. The absence of IS between the two genes in these isolates suggests both genes might be transferable together.

Vancomycin resistance, although rare, was characterized by the presence of the vanA cluster in one of the two isolates, the most common gene reported in other studies (Courvalin, 2006; Ekwanzala et al., 2020; Torres et al., 2018). In the remaining vancomycin-resistant isolate (VAR-660), sequencing highlighted the presence of the rare vanL, a gene not covered by this array. The rare vanL gene cluster, first described by Boyd et al. (2008), has been so far detected on the chromosome of a single E. faecalis isolate of human origin (Ekwanzala et al., 2020) displaying low-level vancomycin resistance (Ekwanzala et al., 2020; Boyd et al., 2008) as observed here (MIC = 8 mg L⁻¹). Our study reports the presence of vanL in an E. faecalis isolate from animal origin. The origin of the vanL gene cluster and the way an animal (i.e., a pig) acquired this strain remain elusive.

Gentamicin resistance was characterized by the presence of aacA-aphD in all resistant enterococci (n = 37) and by aacA-aphD (n = 5) or aacA-aphD/aadD (n = 12) in resistant staphylococci (n = 19) of this study. In addition, resistance to gentamicin was found to be associated with kanamycin resistance in 18 of these 19 resistant staphylococci, with kanamycin-resistant isolates (n = 19) harboring aadD alone (n = 1), aacA-aphD (n = 5) or aacA-aphD/aadD (n = 12) gene combination.

In a few cases of this study (10.8% of the individual resistant phenotypes in staphylococci and 7.0% in enterococci), a genetic marker could not be associated with the resistant phenotype. In enterococci, no relevant AMR genes were found with the array to explain erythromycin (n = 3), Q/D (n = 6), chloramphenicol (n = 10), gentamicin (n = 1) and vancomycin (n = 1) phenotypic resistances (out of a total of 444 assessed phenotypes). In staphylococci, a genetic explanation was missing for clindamycin (n = 1), erythromycin (n = 6), Q/D (n = 2), tiamulin (n = 6), chloramphenicol (n = 2), gentamicin (n = 2), streptomycin (n = 7), and trimethoprim (n = 7) resistant phenotypes (out of a total of 342 assessed phenotypes). In addition, the results of this study highlighted the complexity of Q/D resistance and a lack of a complete genetic explanation in 14/90 isolates since the identified genes (except for lsaA and lsaE) have been reported to confer resistance to either streptogramin A or streptogramin B but not both. Particularly, the presence of erm (streptogramin B) alone in Q/D-resistant isolates (10/53 E. faecium, 4/37 MRSA) does not explain the observed Q/D-resistant phenotype. Similar to lsa, eatA would be an interesting target as this gene confers the same profile of cross-resistance to lincosamides, streptogramins A, and pleuromutilins (Isnard et al., 2013). The WGS investigation revealed the presence of msrC in one E. faecium (VAR-681) exhibiting resistance to Q/D, this gene being frequently reported elsewhere in this species (Frye & Jackson, 2013; Hollenbeck & Rice, 2012). And, msrC has been described in Q/D-susceptible isolates, however first reported as being specifically found in resistant E. faecium. This suggests that it might be silent or involved in resistance to other antibiotics (Smoglica et al., 2022) such as macrolides (Portillo et al., 2000; Zaheer et al., 2020). Therefore, msrA which shares significant sequence identity with msrC has been reported to confer macrolides-streptogramin B resistance in staphylococci and could be an interesting future target to investigate erythromycin and/or Q/D phenotypic resistances (Reynolds & Cove, 2005). In addition, ermF, ermY, mphB or more recently described mefD, msrF, and msrH have been found in staphylococci to confer macrolide and lincosamide resistances (Miklasinska-Majdanik, 2021; Schwarz et al., 2018; Woodford, 2005). Besides, ereA and ereB esterases have been reported to confer macrolide resistance in staphylococci isolated from animals as well (Miklasinska-Majdanik, 2021; Schwarz et al., 2018). In addition, chromosomal mutations in rplD or rplV coding for ribosomal proteins L4 and L22 leading macrolide and Q/D resistance, respectively, were also observed in Streptococcus pneumoniae and S. aureus and could be targeted in future studies (Farrell et al., 2004; Malbruny et al., 2002; Miklasinska-Majdanik, 2021). The sal gene family reported to confer lincosamide, streptogramin A and pleuromutilin resistance in staphylococci from humans or animals at various levels could be investigated in an attempt to explain the tiamulin-resistant phenotypes observed in 6/35 isolates (Mohamad et al., 2022; Schwarz et al., 2018). Pleuromutilin resistance also derives from chromosomal mutations in 23S rRNA or rplC as described in staphylococci isolated from both humans and animals (Paukner & Riedl, 2017; Van Duijkeren et al., 2014). No gene of this array was detected in 10 isolates (1 E. faecalis and 9 MRSA) exhibiting resistance to aminoglycosides, namely to gentamicin (n = 2 MRSA and 1 E. faecalis) and to streptomycin (n = 7 MRSA). In enterococci, an intrinsic low-level of aminoglycoside resistance can be occasionally observed due to the presence of the species-specific chromosomal aac(6')-Ii gene found in almost all E. faecium (Adamecz et al., 2021). Among many aminoglycosides modifying enzymes (AMEs) described in the literature, a few of them such as str could be interesting to target in the future as already described in staphylococci (Ramirez & Tolmasky, 2010; Schwarz et al., 2018). In contrast, few cases of mismatch were observed with kanamycin-susceptible isolates (n = 7 MRSA) by BMD testing carrying one gene encoding aminoglycosides resistance (aadD in this study), as already reported in other studies (Adamecz et al., 2021; Feizabadi et al., 2006; Yean et al., 2007). This could be explained if this gene is nonfunctional and could be investigated in future studies. In addition, Ida et al. (2002) have shown that rearrangements caused by the integration of insertion elements into the staphylococcal chromosome or plasmids affected the expression of adjacent genes including aminoglycoside resistance genes. Finally, 7/55 (12.7%) trimethoprim-resistant MRSA isolates were not found to be associated with the presence of a genetic marker in this study. The presence of other members of the dfr genes family (e.g., dfrF) or the presence of chromosomal mutations may explain the phenotypic resistance observed in these isolates (Woodford, 2005).

5 CONCLUSIONS

In this study, a bead array was described aiming to detect broad AMR genetic profiles of Gram-positive bacteria from healthy animals in a single experiment. A number of resistance genes circulating among Enterococcus spp. and Staphylococcus spp. were targeted by this array allowing the screening of a large number of strains in a limited time and at an affordable cost. The relatively short turnaround time (~8 h) and the use of a software source code freely available allow rapid data acquisition and analysis. In addition, the method is cost-effective with a reagent price of 18 €/sample. The designed array targeted AMR from 9 antimicrobial families, including resistance to critically important antimicrobials, linezolid, and vancomycin, which are important to monitor in both human and animal sectors. Due to the flexibility of the method, a more specific bead array (e.g., linezolid-array) could also be easily designed by removing/adding probes to respond to a particular epidemiologic situation (i.e., location) or to screen other bacterial species; and could also be extended to other research areas (i.e., virulence). Animal isolates were investigated; nevertheless, other reservoirs of AMR genes (humans, environment) could be assessed with the array. Besides AMR genes, resistance resulting from point mutations in receptors (e.g., ciprofloxacin resistance) or in specific proteins (e.g., daptomycin resistance) are targetable as well. In parallel to the official monitoring based on AST, screening with the array allowed us to rapidly determine genetic profiles circulating among livestock animals in Belgium. Isolates of this study have frequently been shown to carry two or more resistance genes conferring the same resistance phenotype, and sometimes genes from the same family, for example erm or tet genes. This accumulation of genes has been frequently observed while a single gene is sufficient to confer resistance, and may be explained by acquisition at different times and/or under different conditions, including their possible co-location on plasmids carrying multiple resistance genes even in absence of selective pressure (Schwarz et al., 2018). Many studies suggested that commensal bacteria are probable reservoirs for AMR genes and can transfer these to pathogenic organisms, for example Bacillus spp. or Salmonella spp. (Frye & Jackson, 2013; Schwarz et al., 2018). Also, while the spread of resistance genes such as ermB and tetM is limited to Gram-positive bacteria, genes such as aadE or tetL are found both in Gram-positive such as reported here but also in Gram-negative bacteria (E. coli) as reported elsewhere (Frye & Jackson, 2013; Schwarz et al., 2018). This array allowed the identification of AMR genes in resistant isolates, spotting a few isolates without genetic explanation as interesting candidates for WGS to identify new resistance mechanisms. This approach allowed us to demonstrate the presence of vanL in one E. faecalis isolate from animal origin. Our study also highlighted the presence of the lnuB gene in staphylococci, rarely reported in the literature. Finally, the missing concordance between aminoglycoside resistance genes and phenotypic resistance observed in this study illustrates the importance of still relying on the routine phenotypic susceptibility test for resistance monitoring. Oppositely, the presence of an oxazolidinone resistance gene in a susceptible isolate occurred several times and showed the importance of genotyping as a complement to phenotyping, particularly for such potentially inducible genes. Our data stress the importance of interpreting AMR with caution and the complementarity of both phenotyping and genotyping methods.

AUTHOR CONTRIBUTIONS

Carole Kowalewicz: Conceptualization (lead); formal analysis (lead); methodology (lead); writing – original draft (lead); writing – review and editing (equal). Michael Timmermans: Conceptualization (supporting); methodology (supporting); writing – review and editing (supporting). David Fretin: Writing – review and editing (supporting). Pierre Wattiau: Conceptualization (supporting); supervision (supporting); writing – review and editing (supporting). Cécile Boland: Conceptualization (supporting); formal analysis (supporting); methodology (supporting); supervision (lead); writing – original draft (supporting); writing – review and editing (equal).

CONFLICT OF INTEREST

None declared.

ETHICS STATEMENT

None required.