Flumequine, a fluoroquinolone in disguise

Abstract

Fluoroquinolone resistance in E. coli isolates from livestock in Europe remains high despite EMA restrictions on fluoroquinolone use in animals. However, flumequine, a quinolone not classified as a fluoroquinolone by various regulatory bodies, is still used in livestock in the Netherlands, Belgium, Greece and France. We investigated whether flumequine selects for the same resistance mechanisms in E. coli. Resistant and non-resistant E. coli isolates were obtained from caecal fermentation assays and broilers exposed to concentrations of flumequine and enrofloxacin. Flumequine usage leads to an approximately 3-fold increase in resistant E. coli in the caecal fermentation, similar to enrofloxacin. In vitro exposure to both flumequine and enrofloxacin revealed the same amino acid substitutions (S83L, D87G) in GyrA. Additionally, the same resistance-causing substitutions were found in phenotypically resistant E. coli isolates from broilers treated with either enrofloxacin or flumequine. Flumequine induces similar resistance mechanisms as enrofloxacin, warranting equivalent restrictions on its use.

Introduction

Antimicrobial compounds are arguably one of the most powerful discovered drugs in the history of human medicine^1,2. However, their success also seems to be their downfall since antimicrobial resistance (AMR) is a rising threat for the human and animal health^3–5. The usage of antimicrobials in livestock is contributing to the selection for antimicrobial resistant bacteria, as antimicrobials are often used metaphylactically, where the whole herd or flock is treated when only one or a few animals show illness^6,7. This large-scale use of antimicrobials in livestock can also select for resistant genes which can subsequently be transferred to other pathogens via horizontal gene transfer⁸. The transmission of AMR and resistant bacteria via different pathways into the environment or to humans is a cause of concern^6,9. Therefore, it is crucial to have guidelines for AMR usage in husbandry to reduce the spread of AMR bacteria.

The European Medicine Agency (EMA) has classified antimicrobial compounds in different categories from A (avoid) to D (prudence) to decrease usage, and parallel with that selection for AMR bacteria, of medically important antimicrobials for the human health¹⁰. Although the EMA has classified all quinolones and fluoroquinolones among restricted usage (since they are medically important for the human health) the report of European Surveillance of Veterinary Antimicrobial Consumption (ESVAC) in 2022 shows significant consumption of quinolones in Greece, The Netherlands, Belgium and France¹¹. A reason for this could be caused by divergent categorisation by the national commissions responsible for the antimicrobial usage in the veterinary practice. As a consequence, quinolones can be prescribed for livestock instead of individual animals since it is considered as a 2^nd choice antimicrobial compound for husbandry usage in for example the Netherlands¹².

Quinolones are a widely used class of antimicrobials of which the first compound was nalidixic acid, synthesised by Lesher et al. ¹³. However, nalidixic acid was not used extensively because of the limited antimicrobial activity¹⁴. Due to chemical modifications of the structure of quinolones, new drugs were created with greater potency, broader spectra of activity, improved pharmacokinetics and lower frequency of development of resistance^14,15. One of these chemical modifications led to the development of a class now known as fluoroquinolones. The resistance to (fluoro)quinolones can be related to three different mechanisms: (i) chromosomal single nucleotide polymorphisms (SNPs) in the quinolone resistance determining region (QRDR) resulting in amino acid changes, genes in this area include gyrA, gyrB, parE and parC. (ii) plasmid-mediated quinolone resistance (PMQR) genes encoded by qnr genes. (iii) upregulation of efflux pumps which can reduce drug accumulation^16–18.

All fluoroquinolones have the same mechanisms of action regardless of whether they are used in human or veterinary medicine; they target DNA gyrase or the topoisomerase VI inhibiting the unwinding of the supercoiled DNA which results in impaired DNA replication. Furthermore, it may results in breaks in de DNA strands which are lethal if unrepaired^19,20. Even though none of the fluoroquinolones licensed for use in humans are approved for use in animals, there is still the possibility of increased fluoroquinolone resistance in human pathogens due to use of a veterinary approved quinolone. In the Netherlands there is an increasing trend in fluoroquinolone resistance in animals observed by the Monitoring of AMR and Antibiotic Usage in Animals in the Netherlands (MARAN). This has been observed in Salmonella, Campylobacter and E. coli until 2022, although fluoroquinolone resistant E. coli remains remarkably high²¹. This is surprising since fluoroquinolones can only be administered to individual animals when there is no other alternative for individual usage. One explanation could be the use of flumequine. Flumequine, although possessing a fluorine atom at the R6 position, is often classified as a (other)quinolone, the reason is not entirely clear. It lacks additions such as a piperazine ring at the R7 position and/or the addition of a cyclopropyl group to the R1 position²² (Supplementary Fig. 1). It is classified as a quinolone by the Netherlands Veterinary Medicines Institute, and as ‘other quinolone’ by the European Medicines Agency (EMA) in their ESVAC project¹¹.

Considering the potential selection by flumequine for fluoroquinolone resistance, the main objective in this study was to investigate if flumequine has the same selective properties as enrofloxacin in E.coli. To establish this, we first conducted in vitro work which consisted of a direct and a long-term exposure study of E. coli to concentrations of flumequine and enrofloxacin. Subsequently we studied the selection of fluoroquinolone resistant E. coli in a setting closer to reality by performing caecal fermentation assays treated with flumequine and enrofloxacin. Lastly, we sequenced E. coli isolates which we acquired from broilers treated with enrofloxacin or flumequine and investigated SNP occurrence in the QRDR region resulting in an amino acid substitution and phenotypic resistance. These experiments are covering in vitro and in vivo situations to investigate if the quinolone flumequine has the same selective properties as the fluoroquinolone enrofloxacin.

Results

gyrA mutations in E. coli after direct exposure to ECOFF concentrations

At first, we conducted a direct exposure on selection plates with fluoroquinolone and enrofloxacin to study if we could obtain resistant isolates and determine co-resistance (Table 1). The isolates were collected as single colonies from selection plates, and we further characterised the strains by phenotypic resistance, MIC levels, and amino acid substitutions in GyrA. After sequencing, we found the amino acid change S83L is mostly occurring after inoculation with enrofloxacin while the D87G and the S83L amino acid substitutions were observed after inoculation with flumequine. Furthermore, we found that an amino acid change of S83L or D87G results in growth on selection plates with enrofloxacin 0.125 mg/L and flumequine 2 mg/L. Contrasting is the observation that the MIC for both enrofloxacin and flumequine is higher in the isolates with the S83L amino acid substitution compared to the isolates harbouring the D87G amino acid change.

Table 1. Characterisation of isolates from direct exposure

Strain	Selection plate	GyrA Amino acid change	Growth on enrofloxacin	Growth on flumequine	Enrofloxacin (mg/L)		Flumequine (mg/L)		SNP difference	Sequence type (ST)
Repeat					1	2	1	2
37	-	-	No	No	0	0	0	0.25	0	5334
37.1	enrofloxacin	S83L	Yes	Yes	0.125	0.125	4.0	4.0	75	5334
37.2	enrofloxacin	S83L	Yes	Yes	0.125	0.125	4.0	4.0	111	5334
37.3	enrofloxacin	S83L	Yes	Yes	0.125	0.125	4.0	4.0	58	5334 (SLV)
37.5	flumequine	S83L	Yes	Yes	0.25	0.25	4.0	4.0	76	5334
37.6	flumequine	S83L	Yes	Yes	0.125	0.25	4.0	4.0	135	5334
88	-	-	No	No	0	0	1.0	0.25	0	40
88.1	flumequine	D87G	Yes	Yes	0.06	0.06	2.0	2.0	85	40
88.2	flumequine	D87G	Yes	Yes	0.06	0.06	4.0	2.0	7	40
88.3	flumequine	D87G	Yes	Yes	0.06	0.06	4.0	4.0	73	40
88.4	flumequine	D87G	Yes	Yes	0.06	0.06	2.0	2.0	95	40
88.5	enrofloxacin	S83L	Yes	Yes	0.25	0.25	4.0	4.0	15	40

In the columns the amino acid substitutions in the GyrA protein, phenotypic resistance, MICs of flumequine and enrofloxacin are displayed. The last column shows the SNP difference between the parental strain and the obtained isolates.

SLV single locus variant.

De novo mutations in gyrA gene in E. coli after stepwise exposure

After the direct exposure on selection plates, we performed a stepwise increasing exposure to susceptible E. coli isolates to observe if induction of de novo mutations over time in the gyrA gene was possible. We started with 24 strains, however after MinION nanopore sequencing we had to exclude 14 strains due to different sequence type at T = 9. In Fig. 1 we can observe the increased resistant strains during the stepwise treatment of enrofloxacin and flumequine, 100% resistance is reached for both flumequine and enrofloxacin. Interestingly, 100% flumequine resistance is reached before 100% enrofloxacin resistance is occurring in the enrofloxacin treatment. This phenomenon is also observed when treating with flumequine when 100% resistance towards enrofloxacin is reached before 100% flumequine resistance. Nevertheless, we can clearly observe cross-resistance between enrofloxacin and flumequine while treated with one of these antimicrobials. Furthermore, the low concentrations of either enrofloxacin or flumequine can already induce a phenotypic resistance at ECOFF concentrations.

Fig. 1 [Images not available. See PDF.]

Percentage of surviving E. coli isolates after stepwise exposure per timepoint the maximum could be 10 isolates resistant (n = 10).

A represents the curve after the enrofloxacin treatment and (B) after the flumequine treatment. r = correlations coefficient (Kendalls correlation test). P = the p value which shows that the null hypothesis is rejected.

After MinION nanopore sequencing, we determined the sequence type of the E. coli strains at T = 9 in combination with amino acid substitutions in the QRDR region. We observed (Table 2) that the sequence type of the strains at T = 9 was identical at T = 0 of the experiment and in combination with one or more amino acid substitutions occurring at T = 9 shows the occurrence of de novo mutations. For some strains a sequence type was not determined but with a spanning tree we could observe clustering of the different strains which allowed us to see their phylogenetic relatedness (Supplementary Fig. 2). In Table 2 we can also observe that some strains did not show any point mutation which should enhance their resistance towards enrofloxacin or flumequine. Nevertheless, these strains showed resistance as we can observe in Fig. 1.

Table 2. Point mutation in QRDR and sequence type (ST) of the E. coli strains at the start of the long-term exposure T = 0 and after treatment with flumequine or enrofloxacin at T = 9

	T = 0		T = 9 flumequine		T = 9 enrofloxacin
Strain	Amino acid substitution	ST	Amino acid substitution	ST	Amino acid substitution	ST
WGS-ecoli-sample-38	-	5334	GyrA A119Ea	5334	Unknown	5334
WGS-ecoli-sample-46	-	349	GyrA D87H	349	GyrA D87G	349
WGS-ecoli-sample-47	-	8070	GyrA D87Y	8070	GryA D678Ea	8070
WGS-ecoli-sample-49	-	58	GyrA D87G	58(SLV)	GyrA D87G	58
WGS-ecoli-sample-64	-	4642	GyrA D87G	4642	GyrA D87G	4642
WGS-ecoli-sample-70	-	2307	GyrA D87G	2307	GyrA D87G	2307(DLV)
WGS-ecoli-sample-88	-	40	GyrA L447Ma	40	GyrA D87G	40
WGS-ecoli-sample-107	-	101	GyrA E153Ga	101	GyrA D87G	101
WGS-ecoli-sample-113	-	720	ParC S57T	720(SLV)	GyrA D87Y, ParC S57T	720
WGS-ecoli-sample-99 (Control)	GyrA S83L	752	GyrA S83L	752	GyrA S83L	752

SLV single locus variant, DLV double locus variant.

^aamino acid substitutions which were not present in the Resfinder database.

Quantifying fluoroquinolone resistant E. coli proportions from caecal fermentations

After conducting in vitro experiments which showed that both de novo mutations and selection for existing fluoroquinolone resistant E. coli was possible, we aimed our focus to a more natural setting by performing caecal fermentation with fresh caeca material. The caecal fermentation assays provided a clear result in terms of selective advantages for resistant strains in a more complex setting which harbours natural E. coli strains, Fig. 2. The highest concentrations of flumequine 2 mg/L and enrofloxacin 0.125 mg/L showed an increase in the resistant E. coli strains. More interestingly, the proportions in both caecal fermentation assays, treated with flumequine or enrofloxacin, showed similar counts for resistance against flumequine and enrofloxacin. The treatment with enrofloxacin 0.125 mg/L showed an almost 3-fold increase in the number of resistant isolates for flumequine or enrofloxacin compared to the control group at 6 or 30 h (**** p ≤ 0.0001). The same result we observed in the caecal fermentation treated with flumequine 2 mg/L, a 3-fold increase in number of resistant isolates in flumequine and enrofloxacin compared to the control group 6 or 30 h (**** p ≤ 0.0001).

Fig. 2 [Images not available. See PDF.]

Resistant E. coli colonies isolated from the caecal fermentations.

Graph (A) Caecal fermentation treated with enrofloxacin. Graph (B) Caecal fermentation with flumequine. Error bars are representing standard errors of the mean. **** p ≤ 0.0001.

Fluoroquinolone resistant E. coli after therapeutic exposure to flumequine

Cloaca swabs from broilers, before and after treatment with therapeutical concentrations of flumequine, were screened for enrofloxacin and flumequine resistant E. coli, as is shown in Fig. 3. The control group showed no remarkable differences of E. coli isolates growing on selection plates with flumequine or enrofloxacin between T = 0 and T = 1, no significance. On the other hand, in the flumequine treatment group we observed almost a 4-fold increase of ECOFF (flumequine **** p ≤ 0.0001 and enrofloxacin **** p ≤ 0.0001) resistant E. coli isolates at T = 1 compared to T = 0. We even observed a small increase in clinical breakpoint resistant E. coli isolates at T = 1 after treatment, for either flumequine and enrofloxacin. Although this increase seems not as substantial as the increase at ECOFF concentrations, it is still around 3.5-fold more compared to T = 0 (flumequine **** p ≤ 0.0001 and enrofloxacin **** p ≤ 0.0001). In comparison with therapeutic concentrations of enrofloxacin we observed similar cross-resistance with flumequine (Supplementary Fig. 3).

Fig. 3 [Images not available. See PDF.]

Percentage resistant E. coli colonies after exposure of clinical concentrations of flumequine.

T = 0 is before initiating the treatment and T = 1 directly after termination of the treatment. A represents the control group which was not treated with flumequine and (B) the group that was treated with flumequine. On the x-axis the concentrations in the control plates are displayed. ECOFF concentration (2 mg/L flumequine and 0.125 mg/L enrofloxacin) and clinical breakpoint concentrations (8 mg/L flumequine and 2 mg/L enrofloxacin). Error bars are representing standard errors of the mean. **** p ≤ 0.0001.

Furthermore, we sequenced with 32 susceptible isolates from the treated group at T = 0 and 16 E. coli isolates which have shown phenotypic growth on clinical breakpoint concentrations of enrofloxacin after treatment, shown in Table 3. We observed that the isolates from the clinical breakpoint selection plates had three amino acid substitutions in the QRDR (S83L, D87N, S80I) and mostly had the same sequence type. We compared the sequence type of resistant E. coli after treatment with the sequence types of the isolates sequenced at T = 0 but we were not able to find a susceptible E. coli strain with the same sequence type. We could therefore not specifically observe inducing of the amino acid change as a consequence of flumequine treatment. However, we did observe selection for enrofloxacin clinical breakpoint resistant E. coli after treatment with clinical concentrations of flumequine.

Table 3. Characteristics of E. coli isolates before and after flumequine treatment with therapeutical concentrations

Before treatment (T=0)	Sequence type	Amino acid substitution in QRDR	Frequency	After treatment	Sequence type	Amino acid substitution in QRDR	frequency
E.coli susceptible	10	-	2	E. coli clinical breakpoint resistant	1485	GyrA S83L, D87N, ParC S80I	15
	43	-	4		-	GyrA S83L, D87N, ParC S80I	2
	57	-	1
	135	-	3
	154	-	3
	371	-	8
	1146	-	1
	1170	-	2
	3209	-	2
	-	ParE I355T	3

In the columns the sequence type, amino acid change and the frequency of the isolates are displayed.

Discussion

Flumequine is selecting for fluoroquinolone resistant bacteria. Flumequine specifically selects for bacteria which acquired a mutation resulting in an amino acid substitution in the QRDR region, mostly occurring in the GyrA or ParC proteins to be precise. This mutation can be pre-existing or can occur de novo during treatment. The usage of flumequine for treating livestock in several European countries occurs for example in the Netherlands and Greece (≥0.75 mg per population correction unit) according the ESVAC¹¹; its potentially increases fluoroquinolone resistant bacteria within livestock which is not only a risk for treatment failure of animals but also for the possible transfer of these AMR resistant bacteria to humans. Vanni et al. ²³ already reported cross resistance of fluoroquinolones in poultry by analysing E. coli isolates obtained from non-treated poultry²³. This stresses the importance of cross resistance as fluoroquinolones are a critical class of antimicrobials for treating infections in human medicine²⁴. Fluoroquinolones such as ciprofloxacin, levofloxacin or norfloxacin will no longer be effective due to a similar or identical resistance mechanism as is induced or selected by using flumequine²⁵.

In this research we mainly focused on the ECOFF concentrations to distinguish the E. coli isolates based on their MIC compared to wildtype (lower MIC than ECOFF) and non-wildtype (higher or comparable MIC than ECOFF). This concentration is used mainly for surveillance purposes²⁶. If growth was observed, the E. coli isolate was scored as resistant. One can argue that this is not clinically relevant. However, if the E. coli isolate can grow on ECOFF concentrations it is reasonable to expect that mutations are acquired resulting in phenotypic resistance.

We encountered enrofloxacin clinical breakpoint resistant E. coli isolates after treatment of broilers with clinical dosages of flumequine. This underlines that flumequine can select for fluoroquinolone clinical breakpoint resistant bacteria when used for farm animals. We did not demonstrate that flumequine used in the farm environment can induce amino acid substitutions in the QRDR leading to fluoroquinolone resistance, likely because of the high diversity of strains in a broiler flock and the difficulty of obtaining the same clone before and after treatment. However, we did prove that a single amino acid change in E. coli can be induced by the usage of flumequine and these isolates are able to withstand ECOFF concentrations of either flumequine or enrofloxacin. In terms of clinical breakpoint resistance mostly two or multiple amino acid substitutions in the QRDR are necessary^18,27. This means that the mutation frequency towards clinical resistance in the QRDR region is much lower if there is already one amino acid substitution present in the QRDR. If the mutation frequency is 10⁻⁶ per cell division it means that two amino acid substitutions in QRDR region may occur in 10⁻¹² cell divisions and when one substitution is already present the potential to get clinically resistant E. coli is 10⁻⁶ cell divisions less, hugely increasing the probability for clinical breakpoint resistant E. coli^28,29.

We used the indicator organism E. coli in this research to determine whether flumequine had similar inducing and selecting properties as enrofloxacin. E. coli is widely used as an indicator organism for resistance levels as a consequence of the ease it can be propagated in the lab and studied due to its well-known genomics³⁰. Still E. coli is just a single bacterial species in the very extensive microbial species composition, for example 15% in the faeces of healthy broilers³¹. Therefore, it might be interesting to study the amino acid substitution occurrence of flumequine in other species and also observe if this is also enhancing and spreading fluoroquinolone resistance. For example, in the Netherlands fluoroquinolone resistance in Campylobacter is increasing²¹, which might be due to flumequine usage.

The ESVAC has classified quinolones in two subgroups; fluoroquinolones and other quinolones. Among other quinolones we encounter flumequine, oxolinic acid and cinoxacin. This divergent classification might be arbitrary if we consider the outcome of our research, in which we clearly show that flumequine has selective properties which enhance fluoroquinolone resistance. As a result, it would be interesting to study the effect of oxolinic acid or cinoxacin on the selection for amino acid substitutions in the QRDR of proteins encoded by chromosomal gyrA and parC genes. Especially since oxolinic acid is used in aquaculture where it can be a potential factor to increase fluoroquinolone resistance^32–34. This is observed in a study conducted in Chile where cross resistance towards enrofloxacin, oxolonic acid and flumequine in Piscirickettsia salmonis isolates from diseased fish was encountered³⁵. In addition to this, a study conducted by Ham et al. ³⁶ showed that oxolinic acid used in apple and pear orchards can also select for amino acid substitutions in the QRDR of Erwinia amylovora³⁶. This is implying that oxolinic acid, described as another quinolone, has the same effect as flumequine. Interestingly,oxolinic acid is a non-fluorinated quinolone in contrast to enrofloxacin and flumequine.

In the stepwise exposure experiment we observed some peculiar results which were unexpected. First, we observed amino acid substitutions in GyrA in the E. coli isolates at T = 9, after treatment with either flumequine or enrofloxacin. However, some amino acid substitutions were not included in the Resfinder database and were recently discovered as substitutions in the QRDR inducing resistance³⁷. Among these substitutions we found the following; GyrA A119E, GyrA D678E, GyrA L447M and GyrA E153G^38–40. Furthermore, we encountered one strain, WGS-sample-38, which showed phenotypic resistance after treatment with enrofloxacin, but we did not detect a corresponding amino acid substitution known for enrofloxacin resistance. This could be attribute to a unknow amino acid change to induce fluoroquinolone resistance or that efflux pumps were upregulated during the experiment⁴¹. Secondly, there was a rapid increase in the percentage of resistant E. coli at ECOFF concentrations even when exposed to very low concentration of enrofloxacin or flumequine. This underlines that low concentrations of (fluoro)quinolone can already select for resistance which illustrate the effect of residual antimicrobial concentrations^9,42. Finally, we observed in susceptible E. coli isolates at T = 0 from the field samples, that some E. coli harboured an amino acid substitution in the ParE protein that also belongs to the QRDR. Although this feels counterintuitive since the E. coli isolates were scored susceptible, but in the literature, it is shown that a single amino acid substitution in the ParE protein does not result in growth on ECOFF concentrations⁴³, however this could lead to higher resistance when combined with other amino acid substitutions.

Given the selection of flumequine among the E. coli isolates in this study, authorities regulating or setting the prescription of antimicrobials for livestock should abolish the differentiation between fluoroquinolone and ‘other quinolones’. Especially since flumequine can create a reservoir of fluoroquinolone resistant bacteria which can transfer to humans or the environment⁴⁴. Institutions as EMA already classified all quinolones among restricted usage which is according to our results appropriate to reduce AMR^10,11,45. However, national advisory groups such as the Dutch WVAB allow flumequine as a 2nd choice antimicrobial (can be used after 1st choice) instead of a 3rd choice antimicrobial (restricted and only for individual animals)⁴⁶. This classification might explain the relatively high fluoroquinolone resistance in commensal E. coli in The Netherlands in broilers²¹. Moreover, the European Centre for Disease Prevention and Control (ECDC) reported in 2023 that fluoroquinolone resistance in E. coli isolates from humans is above 25% in 17 of the 45 countries⁴⁷. This emphasises that fluoroquinolone resistance remains a multi sectoral problem.

To conclude, flumequine which shows the same selective properties as enrofloxacin, should be considered as fluoroquinolones in the classification for the usage in livestock.

Methods

Obtaining fluoroquinolone resistant E. coli by direct exposure

In this research epidemiology cut off values (ECOFF) concentrations are used to distinguish bacterial isolates between wildtype and non-wildtype (further referred to as resistant) based on their MIC to detect phenotypically acquired resistance. The protocol on how to select resistant E. coli using ECOFF concentrations of fluoroquinolones was derived from refs. ^48,49. The parental strains 37 and 88 (Supplementary Table 1) obtained from fermentation assays from ref. ⁵¹, inoculated on blood agar plates and incubated overnight at 37 °C. One colony was taken from each plate and transferred to 10 ml LB broth and incubated overnight at 37 °C while shaking. Afterwards the cell cultures were centrifuged, and the pellet was suspended in 1 ml sterile LB broth to acquire an inoculum of ~10⁹–10¹⁰ cfu/mL. Cell suspensions (100 µl) were inoculated on MacConkey agar plates containing the ECOFF concentrations of enrofloxacin (0.125 mg/L) or flumequine (2.0 mg/L), overnight at 37°C. Single colonies were taken from the selection plates and from both the enrofloxacin and flumequine resistant colonies. Afterwards the selected colonies were reinoculated on the opposite selection plate to determine co-resistance.

The MIC of the strains of interest were determined by standard broth microdilution adapted from EUCAST ESCMID (2003).

Long-term exposure to fluoroquinolones with non-fluoroquinolone resistant E. coli strains

The long-term exposure experiment was derived and modified from ref. ⁵⁰. For this experiment we selected 24 E.coli strains which all have been isolated from broilers and were susceptible for flumequine and enrofloxacin, except one positive control strain (supplementary Table 2 (only the surviving strains)). Strains were stored at the −80 °C and thereafter first inoculated on blood agar plates and incubated overnight at 37 °C. Subsequently, per strain a single colony was selected and transferred to a well containing 1 ml LB medium in a 12-well plate (in total two 12-well plate were used), referred to as T = 0. The plates were incubated overnight at 37 °C, the next day 100 µl per strain was transferred to a well in a 96-wells plate. Additionally, with a stamp the strains were transferred to squared MacConkey plates containing flumequine (2 mg/L), enrofloxacin (0.125 mg/L) and a control plate with no antimicrobials. This was done to determine the susceptibility of the 23 selected strain and the growth of the control strain. Next, 100 µl of the T = 0 12-well plates were transferred to a new 12-well plate containing LB medium with a concentration of 0.0125 mg/L enrofloxacin (two 12-well plates) and 0.2 mg/L flumequine (also two 12-well plates), referred to as T = 1. Subsequently the plates were incubated overnight at 37 °C and the next day the strains were transferred to squared McConkey selections plates with ECOFF concentrations of flumequine and enrofloxacin. This procedure was repeated for 10 days and each day the concentration increased by 0.0125 mg/L for enrofloxacin and 0.2 mg/L for flumequine. Every day the proportion of non-wildtype (referred to as resistance) at ECOFF concentrations was determined. After 10 days the strains were exposed to clinical breakpoint concentrations of enrofloxacin (2 mg/L) and flumequine (8 mg/L) and analysed for survival and resistance.

Caecal fermentation to determine co-resistance of E. coli from caecum faeces

The protocol for the fermentation studies was completely adopted from ref. ⁵¹. Only alteration or addition was the use of flumequine (2 mg/L, 0,2 mg/L and 0.02 mg/L). The fermentation was performed at 41 °C, shaking at 300 rpm in an anaerobic hood to mimic the conditions in the intestine. Measurements were taken after 6- and 30-h of incubation. At the beginning of the experiment, a baseline measurement was conducted. Directly after removing the Eppendorf tubes from the anaerobic hood 10 µl of each fermentation was inoculated five times, on five individual MacConkey plates. After overnight incubation at 37 °C ~20 single colonies per MacConkey plate were picked with a pipette tip and transferred to a single well in a 96-wells plate that contained 100 µl LB medium per well. The 96 colonies per treatment were transferred to squared MacConkey plates containing ECOFF concentrations of enrofloxacin (0.125 mg/L), flumequine (2 mg/L) and a control plate without antimicrobials. Next, resistant colonies were scored after overnight incubation at 37 °C. An E. coli colony was scored resistant if it was able to grow on the plate with ECOFF concentration of the specific antimicrobial. The caecum material was obtained from broilers from commercial farms in the Netherlands which encountered serious lameness. The caeca were removed from the euthanized broilers which were used for educational purposes at the faculty of veterinary medicine (registration number: AVD10800202115056). The broilers aged between 15 and 40 days, due to different sample moments as this experiment was repeated three times and for every experiment caecum material of two different broilers was used.

Resistance characterisation of E. coli isolates from poultry farms treated with clinical concentrations of flumequine

Samples, in terms of cloaca swabs, were taken from commercial broiler farms when a treatment with flumequine needed to be performed according to a veterinarian. The number of broilers was calculated with an R package CRITSize which make sample size estimation in group randomised trials (zenodo repository https://doi.org/10.5281/zenodo.12155715). In total 16 broilers were swabbed in the poultry house before treatment and 16 in the untreated poultry house as a baseline measurement. After the treatment again 16 broilers were swabbed in the untreated and treated poultry house. All the 16 broilers were randomly selected per timepoint and treatment. The cloaca swabs were transferred to the lab for further processing at the sampling day. In the lab the swabs were first inoculated on MacConkey plates and incubated overnight at 37 °C. Afterwards two colonies (per swab) from each MacConkey plate were transferred to blood agar plates and incubated at 37 °C overnight. Subsequently, the colonies, 32 in total per timepoint and treatment, were transferred to separate wells in a 96-wells plate which enabled transferring the colonies to squared MacConkey plates by a stamp. The squared MacConkey plates contained ECOFF concentrations (flumequine 2 mg/L, enrofloxacin 0.125 mg/L), clinical breakpoint concentrations (flumequine 8 mg /L, enfrofloxacin 2 mg/L) or no antimicrobials (control plate). After overnight incubation the proportion of resistant E. coli could be calculated with comparing the selection plates to the control plate.

Sequencing of E. coli isolates

Several experiments were followed up by sequencing for SNPs to determine amino acid substitutions and sequence type determination. For that reason, DNA extraction of E. coli was preformed using the NGS DNeasy UltraClean Microbials kits from Qiagen. Additionally, the concentrations of all the DNA isolates were checked using Thermo Fishers’ Qubit Fluorometer to guarantee high enough concentration for sequencing.

MinION sequencing

Sequencing was done following the Oxford Nanopore Rapid barcoding Kit (SQK-RBK096) and sequenced in R9.4.1 flow cells (FLO-MIN106) using the MinION Mk1B (ONT) device. ONT raw reads were subjected to base-calling using MinKNOW (v4.5.3) with the Super Accurate model. Afterwards, the reads were trimmed with Filtlong (v0.2.1) (https://github.com/rrwick/Filtlong) and the quality was assessed with FastQC (v0.11.4) (https://github.com/s-andrews/FastQC). Reads were assembled to contigs with Flye (v2.9)⁵². Assemblies were polished using Medaka (v1.4.3) (https://github.com/nanoporetech/medaka) and Homopolish (v0.3.4)⁵³. The quality of all sequences was checked with Checkm (v1.1.3), and only genomes with a contamination threshold of <5% and completeness threshold of >95% were included in the analysis⁵⁴. ResFinder (v4.0) was used to identify the amino acid changes and MLST was determined using mlst^55–57.

Illumina sequencing

Illumina sequencing was performed using Illumina Nextseq 500 (Useq, Utrecht sequencing facility) with a maximum read length of 2 × 150 bp. Libraries were prepared with Illumina Nextera XT DNA Library Preparation Kit according to the manufacturers protocol⁵⁸. Read processing and assembly was performed as described for Illumina reads in ref. ⁵¹.

All sequences can be found in the SRA.

Statistical analysis

We used R (v4.1.0) to perform statistical analysis. For the survival analysis of the E. coli isolates we used Kendall’s correlation test to observe if resistance towards enrofloxacin and flumequine is positively correlated after treatment with flumequine or enrofloxacin. Moreover for the fermentation studies we used Poisson regression model with experiment as an random effect combined with selection. Lastly for the difference in resistant E. coli isolates after therapeutic usage of flumequine we used a logistic regression model in which we included swabbing timepoints as an random effect as it was not possible to swab the same broilers at each timepoint due to the commercial setting. Residuals were tested beforehand for normality Post hoc Tukey test was used to determine significant differences between groups. R scripts are available at Zenodo repository https://doi.org/10.5281/zenodo.1215571).

Acknowledgements

This work was supported by the Netherlands Centre for One Health (NCOH). We also would like to thank Jeroen Leus for helping collect the field samples. Furthermore, we would like to thank the suppliers of the caecal material which have been used in this study.

Author Contributions

A.S., J.W., E.F. and A.Z. designed the study. A.S., L.K., R.C. performed the experiment and data collection. A.S., J.W., A.Z. and E.F. analysed the sequencing data and performed statistical analyses and data visualisation. All authors contributed to data interpretation. A.S. prepared the manuscript with input and contributions from all authors. All authors have read and agreed to the published version of the manuscript.

Data availability

Sequence data for this article can be found in the SRA under accession PRJEB75157.

Code availability

Codes and statistical analysis including the data are available at https://doi.org/10.5281/zenodo.12155715.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s44259-024-00044-5.

Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. Katz, L; Baltz, RH. Natural product discovery: past, present, and future. J. Ind. Microbiol. Biotechnol.; 2016; 43, pp. 155-176.[COI: 1:CAS:528:DC%2BC28XjtFKhug%3D%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/26739136][DOI: https://dx.doi.org/10.1007/s10295-015-1723-5]

2. Hutchings, M; Truman, A; Wilkinson, B. Antibiotics: past, present and future. Curr. Opin. Microbiol; 2019; 51, pp. 72-80.[COI: 1:CAS:528:DC%2BC1MXitV2msbjP] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31733401][DOI: https://dx.doi.org/10.1016/j.mib.2019.10.008]

3. Schäberle, TF; Hack, IM. Overcoming the current deadlock in antibiotic research. Trends Microbiol.; 2014; 22, pp. 165-167. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24698433][DOI: https://dx.doi.org/10.1016/j.tim.2013.12.007]

4. Velazquez-Meza, ME; Galarde-López, M; Carrillo-Quiróz, B; Alpuche-Aranda, CM. Antimicrobial resistance: one health approach. Vet. World; 2022; 15, pp. 743-749.[COI: 1:CAS:528:DC%2BB38XhvVOkurbL] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35497962][DOI: https://dx.doi.org/10.14202/vetworld.2022.743-749]

5. Prestinaci, F; Pezzotti, P; Pantosti, A. Antimicrobial resistance: a global multifaceted phenomenon. Pathog. Glob. Health; 2015; 109, pp. 309-318. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/26343252][DOI: https://dx.doi.org/10.1179/2047773215Y.0000000030]

6. Woolhouse, M; Ward, M; van Bunnik, B; Farrar, J. Antimicrobial resistance in humans, livestock and the wider environment. Philos. Trans. R. Soc. B Biol. Sci.; 2015; 370, 20140083. [DOI: https://dx.doi.org/10.1098/rstb.2014.0083]

7. Holmes, AH et al. Understanding the mechanisms and drivers of antimicrobial resistance. Lancet; 2016; 387, pp. 176-187.[COI: 1:CAS:528:DC%2BC2MXhvVOqtr3F] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/26603922][DOI: https://dx.doi.org/10.1016/S0140-6736(15)00473-0]

8. Munk, P et al. Abundance and diversity of the faecal resistome in slaughter pigs and broilers in nine European countries. Nat. Microbiol.; 2018; 3, pp. 898-908.[COI: 1:CAS:528:DC%2BC1cXhtlOis73N] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30038308][DOI: https://dx.doi.org/10.1038/s41564-018-0192-9]

9. Andersson, DI; Hughes, D. Evolution of antibiotic resistance at non-lethal drug concentrations. Drug Resist. Updat.; 2012; 15, pp. 162-172.[COI: 1:CAS:528:DC%2BC38XptVOrtLc%3D] [DOI: https://dx.doi.org/10.1016/j.drup.2012.03.005] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/22516308]

10. https://www.ema.europa.eu/en/news/categorisation-antibiotics-used-animals-promotes-responsible-use-protect-public-animal-health.

11. European Medicines Agency. European Surveillance of Veterinary Antimicrobial Consumption, 2022. ‘Sales of veterinary antimicrobial agents in 31 European countries in 2022’ (EMA/299538/2023) (2022).

12. SDa. Het gebruik van antibiotica bij landbouwhuisdieren in 2022. 2023.

13. Lesher G. Y. et al. 1,8-Naphthyridine derivatives a new class of chemotherapeutic agents. UTC. Available at: https://pubs.acs.org/sharingguidelines (1962).

14. Kidwai, M; Misra, P; Kumar, R. The fluorinated quinolones. Curr. Pharm. Des.; 1998; 4, pp. 101-118.[COI: 1:CAS:528:DyaK1cXisFOrs74%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10197035][DOI: https://dx.doi.org/10.2174/138161280402221007112335]

15. Andersson, MI; MacGowan, AP. Development of the quinolones. J. Antimicrobial Chemother.; 2003; 51, pp. 1-11.[COI: 1:CAS:528:DC%2BD3sXktlOqurc%3D] [DOI: https://dx.doi.org/10.1093/jac/dkg212]

16. Jacoby G. A. Mechanisms of resistance to quinolones. Available at: https://academic.oup.com/cid/article/41/Supplement_2/S120/307501 (2005).

17. Fàbrega, A; Madurga, S; Giralt, E; Vila, J. Mechanism of action of and resistance to quinolones. Micro. Biotechnol.; 2009; 2, pp. 40-61. [DOI: https://dx.doi.org/10.1111/j.1751-7915.2008.00063.x]

18. Hayer, SS et al. Genetic determinants of resistance to extended-spectrum cephalosporin and fluoroquinolone in escherichia coli isolated from diseased pigs in the United States. mSphere; 2020; 5, pp. 10-128. [DOI: https://dx.doi.org/10.1128/mSphere.00990-20]

19. Drlica, K. Mechanism of fluoroquinolone action. Curr. Opin. Microbiol; 1999; 2, pp. 504-508.[COI: 1:CAS:528:DyaK1MXntlCku78%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10508721][DOI: https://dx.doi.org/10.1016/S1369-5274(99)00008-9]

20. Drlica, K et al. Quinolones: action and resistance updated. Curr. Top. Med Chem.; 2009; 9, pp. 981-998.[COI: 1:CAS:528:DC%2BD1MXhsFWisb%2FF] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19747119][DOI: https://dx.doi.org/10.2174/156802609789630947]

21. MARAN. Monitoring of antimicrobial resistance and antibiotic usage in animals in the Netherlands in 2022. 2023.

22. Pham, TDM; Ziora, ZM; Blaskovich, MAT. Quinolone antibiotics. Medchemcomm; 2019; 10, pp. 1719-1739.[COI: 1:CAS:528:DC%2BC1MXht1yitL3J] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31803393][DOI: https://dx.doi.org/10.1039/C9MD00120D]

23. Vanni, M et al. Fluoroquinolone resistance and molecular characterization of gyrA and parC quinolone resistance-determining regions in Escherichia coli isolated from poultry. Poult. Sci.; 2014; 93, pp. 856-863.[COI: 1:CAS:528:DC%2BC2sXhsFChsL3L] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24706962][DOI: https://dx.doi.org/10.3382/ps.2013-03627]

24. WHO. Antimicrobial Resistance: Global Report on Surveillance. https://apps.who.int/iris/handle/10665/112642 (2014).

25. LiverTox: Clinical and Research Information on Drug-Induced Liver Injury [Internet]. Bethesda (MD): National Institute of Diabetes and Digestive and Kidney Diseases. Fluoroquinolones. https://www.ncbi.nlm.nih.gov/books/NBK547840/ (2020).

26. Kahlmeter, G; Turnidge, J. How to: ECOFFs—the why, the how, and the don’ts of EUCAST epidemiological cutoff values. Clin. Microbiol. Infect.; 2022; 28, pp. 952-954. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35218980][DOI: https://dx.doi.org/10.1016/j.cmi.2022.02.024]

27. Khan, DD et al. A mechanism-based pharmacokinetic/pharmacodynamic model allows prediction of antibiotic killing from MIC values for WT and mutants. J. Antimicrobial Chemother.; 2015; 70, pp. 3051-3060.[COI: 1:CAS:528:DC%2BC28Xot1Wqs78%3D] [DOI: https://dx.doi.org/10.1093/jac/dkv233]

28. Davies, J; Davies, D. Origins and evolution of antibiotic resistance. Microbiol. Mol. Biol. Rev.; 2010; 74, pp. 417-433.[COI: 1:CAS:528:DC%2BC3cXhsVSqurnO] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20805405][DOI: https://dx.doi.org/10.1128/MMBR.00016-10]

29. Coculescu, B-I. Antimicrobial resistance induced by genetic changes. J. Med Life; 2009; 2, pp. 114-123. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20108530]

30. Anjum, MF et al. The potential of using E. coli as an indicator for the surveillance of antimicrobial resistance (AMR) in the environment. Curr. Opin. Microbiol; 2021; 64, pp. 152-158.[COI: 1:CAS:528:DC%2BB38Xjs1ynt78%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34739920][DOI: https://dx.doi.org/10.1016/j.mib.2021.09.011]

31. Abd El Tawab, A; Abd El Aal, S; mazied, E; EL. Morsy, D. Prevalence of E. coli in broiler chickens in winter and summer seasons by application of PCR with its antibiogram pattern. Benha Vet. Med J.; 2015; 29, pp. 119-128. [DOI: https://dx.doi.org/10.21608/bvmj.2015.31683]

32. Vanamala P., Sindhura P., Sultana U., Vasavilatha T., Gul M. Z. Common bacterial pathogens in fish: an overview. In Bacterial Fish Diseases 279–306 (Elsevier, 2022).

33. Rigos, G; Nengas, I; Alexis, M; Troisi, GM. Potential drug (oxytetracycline and oxolinic acid) pollution from Mediterranean sparid fish farms. Aquat. Toxicol.; 2004; 69, pp. 281-288.[COI: 1:CAS:528:DC%2BD2cXlvFylu70%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/15276333][DOI: https://dx.doi.org/10.1016/j.aquatox.2004.05.009]

34. Chong RS-M. Vibriosis. in Aquaculture Pathophysiology 447–464 (Elsevier, 2022).

35. Henríquez, P; Bohle, H; Bustamante, F; Bustos, P; Mancilla, M. Polymorphism in gyrA is associated to quinolones resistance in Chilean Piscirickettsia salmonis field isolates. J. Fish. Dis.; 2015; 38, pp. 415-418. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24917068][DOI: https://dx.doi.org/10.1111/jfd.12255]

36. Ham, H; Oh, G-R; Park, DS; Lee, YH. Survey of oxolinic acid-resistant Erwinia amylovora in Korean apple and pear orchards, and the fitness impact of constructed mutants. Plant Pathol. J.; 2022; 38, pp. 482-489.[COI: 1:CAS:528:DC%2BB3sXlvF2jtrg%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/36221920][DOI: https://dx.doi.org/10.5423/PPJ.OA.04.2022.0059]

37. https://bitbucket.org/genomicepidemiology/resfinder_db.

38. Kong, M; Liu, C; Xu, Y; Wang, J; Jin, D. Concordance between genotypic and phenotypic drug-resistant profiles of Shigella isolates from Taiyuan City, Shanxi Province, China, 2005 to 2016. Microbiol. Spectr; 2023; 11, e0011923. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37249442][DOI: https://dx.doi.org/10.1128/spectrum.00119-23]

39. Mbanga, J et al. Genomic insights of multidrug-resistant Escherichia coli from wastewater sources and their association with clinical pathogens in South Africa. Front. Vet. Sci; 2021; 8, 636715. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33718473][DOI: https://dx.doi.org/10.3389/fvets.2021.636715]

40. Ayandiran, TO; Falgenhauer, L; Schmiedel, J; Chakraborty, T; Ayeni, FA. High resistance to tetracycline and ciprofloxacin in bacteria isolated from poultry farms in Ibadan, Nigeria. J. Infect. Dev. Ctries.; 2018; 12, pp. 462-470.[COI: 1:CAS:528:DC%2BC1MXhsFGlt73F] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31940298][DOI: https://dx.doi.org/10.3855/jidc.9862]

41. Kern, WV; Oethinger, M; Jellen-Ritter, AS; Levy, SB. Non-target gene Mutations in the development of fluoroquinolone resistance in Escherichia coli. Antimicrob. Agents Chemother.; 2000; 44, pp. 814-820.[COI: 1:CAS:528:DC%2BD3cXitFOltLg%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/10722475][DOI: https://dx.doi.org/10.1128/AAC.44.4.814-820.2000]

42. Gullberg E. et al. Selection of resistant bacteria at very low antibiotic concentrations. PLoS Pathog.7, e1002158 (2011).

43. Massé, J et al. Description of antimicrobial-resistant Escherichia coli and their dissemination mechanisms on dairy farms. Vet. Sci.; 2023; 10, 242. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37104397]

44. Booton, RD et al. One health drivers of antibacterial resistance: quantifying the relative impacts of human, animal and environmental use and transmission. One Health; 2021; 12, [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33644290][DOI: https://dx.doi.org/10.1016/j.onehlt.2021.100220]

45. Schmerold, I; van Geijlswijk, I; Gehring, R. European regulations on the use of antibiotics in veterinary medicine. Eur. J. Pharm. Sci.; 2023; 189,[COI: 1:CAS:528:DC%2BB3sXhs1GhsLnP] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37220817][DOI: https://dx.doi.org/10.1016/j.ejps.2023.106473]

46. Dutch Work Group Veterinary Antimicrobial Policy. WVAB-richtlijn classificatie van veterinaire antimicrobiële middelen. Available at: www.cbg-meb.nl (2015).

47. Stockholm: European Centre for Disease Prevention and Control and World Health Organization; Antimicrobial resistance surveillance in Europe 2023–2021 data. (2023).

48. Pasquali, F; Manfreda, G. Mutant prevention concentration of ciprofloxacin and enrofloxacin against Escherichia coli, Salmonella Typhimurium and Pseudomonas aeruginosa. Vet. Microbiol; 2007; 119, pp. 304-310.[COI: 1:CAS:528:DC%2BD2sXis1elug%3D%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/16987619][DOI: https://dx.doi.org/10.1016/j.vetmic.2006.08.018]

49. Cesaro, A et al. Low selection of topoisomerase mutants from strains of Escherichia coli harbouring plasmid-borne qnr genes. J. Antimicrobial Chemother.; 2008; 61, pp. 1007-1015.[COI: 1:CAS:528:DC%2BD1cXksF2lsLk%3D] [DOI: https://dx.doi.org/10.1093/jac/dkn077]

50. Lenski, RE; Rose, MR; Simpson, SC; Tadler, SC. Long-term experimental evolution in Escherichia coli. I. Adaptation and divergence during. Am. Nat.; 1991; 138, pp. 1315-1341. [DOI: https://dx.doi.org/10.1086/285289]

51. Swinkels A. F. et al. Defining minimal selective concentrations of amoxicillin, doxycycline and enrofloxacin in broiler-derived cecal fermentations by phenotype, microbiome and resistome. bioRxiv 2023: 2023.11.21.568155. Available at: http://biorxiv.org/content/early/2023/11/22/2023.11.21.568155.abstract.

52. Kolmogorov, M; Yuan, J; Lin, Y; Pevzner, PA. Assembly of long, error-prone reads using repeat graphs. Nat. Biotechnol.; 2019; 37, pp. 540-546.[COI: 1:CAS:528:DC%2BC1MXosV2qsrs%3D] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30936562][DOI: https://dx.doi.org/10.1038/s41587-019-0072-8]

53. Huang, Y-T; Liu, P-Y; Shih, P-W. Homopolish: a method for the removal of systematic errors in nanopore sequencing by homologous polishing. Genome Biol.; 2021; 22,[COI: 1:CAS:528:DC%2BB3MXht1KrtrvF] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33789731][DOI: https://dx.doi.org/10.1186/s13059-021-02282-6]

54. Parks, DH; Imelfort, M; Skennerton, CT; Hugenholtz, P; Tyson, GW. CheckM: assessing the quality of microbial genomes recovered from isolates, single cells, and metagenomes. Genome Res.; 2015; 25, pp. 1043-1055.[COI: 1:CAS:528:DC%2BC2MXht1SltbfE] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/25977477][DOI: https://dx.doi.org/10.1101/gr.186072.114]

55. Camacho, C et al. BLAST+: architecture and applications. BMC Bioinform.; 2009; 10, [DOI: https://dx.doi.org/10.1186/1471-2105-10-421]

56. Bortolaia, V et al. ResFinder 4.0 for predictions of phenotypes from genotypes. J. Antimicrobial Chemother.; 2020; 75, pp. 3491-3500.[COI: 1:CAS:528:DC%2BB3cXisV2rtbrP] [DOI: https://dx.doi.org/10.1093/jac/dkaa345]

57. Jolley, KA; Bray, JE; Maiden, MCJ. Open-access bacterial population genomics: BIGSdb software, the PubMLST.org website and their applications. Wellcome Open Res.; 2018; 3, 124. [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30345391][DOI: https://dx.doi.org/10.12688/wellcomeopenres.14826.1]

58. Nextera XT DNA Library Prep Kit. accessed 8 December 2022. Available at: https://emea.support.illumina.com/sequencing/sequencing_kits/nextera_xt_dna_kit/documentation.html (2022).