Cloacal Lactobacillus isolates from broilers show high prevalence of resistance towards macrolide and lincosamide antibiotics

Abstract

Eighty-seven Lactobacillus strains isolated from cloacal swabs of broiler chickens derived from 20 different farms in Belgium were identified to species level and tested for susceptibility to macrolide and lincosamide antibiotics. Five different Lactobacillus species were identified as being predominantly present in the cloacae of broilers: Lactobacillus crispatus, Lactobacillus salivarius subsp. salivarius, Lactobacillus amylovorus, Lactobacillus gallinarum and Lactobacillus reuteri. Acquired resistance prevalence to macrolides and lincosamides was very high in the investigated lactobacilli: 89% of the strains were resistant to either or both lincosamide and macrolide class antibiotics. The vast majority of these resistant strains (96%) displayed constitutive resistance. More than one-half of the macrolide and/or lincosamide resistant strains carried an erm(B), erm(C), mef(A), lnu(A) gene or a combination of these genes.

Les Souches de Lactobacillus isolées de poulets de chair montrent des prévalences élevées de résistance aux macrolides et lincosamides

Quatre-vingt-sept souches de Lactobacillus isolées d'écouvillons cloacaux de poulets de chair appartenant à 20 élevages différents en Belgique ont été identifiées au niveau espèce et ont été testées pour leur sensibilité aux macrolides et lincosamides. Cinq espèces différentes de Lactobacillus ont été identifiées comme étant principalement présentes dans le cloaque des poulets de chair : L. crispatus, L. salivarius subsp. salivarius, L. amylovorus, L. gallinarum et L. reuteri. La prévalence de la résistance acquise aux macrolides et lincosamides a été très élevée parmi les lactobacilles investigués : 89 pourcent des souches étaient résistantes aux lincosamides et/ou macrolides. La grande majorité de ces souches résistantes (96%) ont montré une résistance constitutive. Plus de la moitié des souches résistantes aux macrolides et/ou aux licosamides portait un des gènes erm(B), erm(C), mef(A), lnu(A), ou une combinaison de ces gènes.

Lactobacillus-Isolate aus der Kloake von Broilern weisen eine hohe Resistenzrate gegenüber Makrolid- und Lincosamidantibiotika auf

87 Lactobacillus-Stämme, die aus Kloakenabstrichen von Broilern aus 20 verschiedenen Farmen in Belgien isoliert worden waren, wurden bis zur Speziesebene identifiziert und auf ihre Empfindlichkeit gegenüber Makrolid- und Lincosamidantibiotika getestet. Es wurden fünf verschiedene Lactobacillus-Spezies als vorwiegende Lactobazillen in den Kloaken von Broilern identifiziert: L. crispatus, L. salivarius subsp. salivarius, L. amylovorus, L. gallinarum und L. reuteri. Bei den untersuchten Lactobazilli war die erworbene Resistenzrate gegenüber Makroliden und Lincosamiden sehr hoch: 89 % der Stämme waren entweder gegen eine oder gegen beide Antibiotikagruppen resistent. Die überwiegende Mehrzahl dieser resistenten Stämme (96 %) wiesen eine konstitutive Resistenz auf. Mehr als die Hälfte der Makrolid und/oder Lincosamid resistenten Stämme besaß ein erm(B)-, erm(C)-, mef(A)-, lnu(A)-Gen oder eine Kombination dieser Gene.

Aislamientos cloacales de Lactobacillus procedentes de pollos de engorde muestran una elevada resistencia frente a antibióticos macrólidos y lincosaminas

Se identificaron hasta el nivel de especie ochenta y siete cepas de Lactobacillus aisladas de hisopos cloacales de pollos de engorde procedentes de 20 granjas distintas de Bélgica y se evaluó su susceptibilidad frente a antibióticos macrólidos y lincosaminas. Cinco especies distintas de Lactobacillus fueron identificadas como predominantes en la cloaca de los pollos de engorde: L. crispatus, L salivarius subsp. salivarius, L. amylovorus, L. gallinarum and L. reuteri. La prevalencia de resistencias adquiridas a macrólidos y lincosaminas en los lactobacilos estudiados fue muy elevada: el 89 por ciento de las cepas fueron resistentes a alguna de las dos o a ambas clases de antibióticos macrólidos y lincosaminas. La gran mayoría de estas cepas resistentes (96%) mostraron resistencias constitutivas. Más de la mitad de las cepas resistentes a los macrólidos y/o lincosaminas eran portadoras de un gen erm(B), erm(C), mef(A), lnu(A) o de una combinación de estos genes.

Introduction

Lactobacilli constitute an important part of the gastro-intestinal microbiota of humans and animals (Garriga et al., 1998; Lu et al., 2003). In general, these bacteria are considered apathogenic and are attributed beneficial properties when administered as probiotic feed additives. Nevertheless, severe infections with Lactobacillus rhamnosus, Lactobacillus paracasei subsp. paracasei, Lactobacillus crispatus and Lactobacillus casei have been reported in immunocompromised human patients (Parola et al., 1998; Felten et al., 1999).

The increased use of antimicrobials has caused an increase of antibiotic resistance occurring not only in potentially pathogenic bacteria but also in apathogenic microbiotal constituents. In this respect, gastro-intestinal commensal bacteria might be considered to be a pool of antimicrobial resistance genes (Catry et al., 2003). In-vivo transfer of genes encoding for erythromycin resistance between Enterococcus faecium strains in the broiler gastro-intestinal tract has been proven to occur (Netherwood et al., 1999). In vitro, transfer of resistance genes from lactobacilli to Lactococcus lactis subsp. lactis and Enterococcus faecalis, a potential pathogen in man and poultry, has already been demonstrated (Gevers et al., 2003). Thus, the presence of resistance genes in lactobacilli might pose a threat to successful treatment of pathogens in poultry and man, and this hypothesis needs to be further elucidated.

Although not the primary antibiotics of choice, macrolide, lincosamide and streptogramin (MLS) antibiotics may be used as an alternative therapy for treatment of enterococcal infections (Murray, 1990; Portillo et al., 2000; Molander & Dahlén, 2003). In Belgium, several antimicrobial preparations registered for poultry contain macrolide or lincosamide antibiotics (Anonymous, 2003) and are often used in poultry medicine for treatment of bacterial diseases. However, the number of macrolide-resistant enterococci of human and poultry origin has increased significantly in the past decades, jeopardizing successful treatment of clinical infections (Murray, 1990; De Leener, 2005).

Although structurally unrelated, macrolides and lincosamides have a similar mode of action (Roberts et al., 1999). Different mechanisms of antibiotic resistance against macrolides and lincosamides have been described with modification of the antibiotic's target, efflux of the antibiotic from the bacterial cell and enzymatic inactivation of the antibiotic as the major modes of action (Roberts et al., 1999). Target modification results in macrolide–lincosamide–streptogramin B (MLS_B) resistance and finds its origin in a mutation in the 23 S rRNA gene or is encoded by erythromycin-resistant-methylase (erm) genes. The erm genes encode adenine-N6-methyltransferases, which dimethylate the 23S rRNA. Different classes of erm genes have been described (Roberts et al., 1999) and expression of those genes may be induced or constitutive.

Efflux encoding genes may belong to the families of the major facilitators (e.g. mef(A) and lmr(A)) or the ATP-binding transporters (e.g. car(A), msr(A), ole(B), srm(B)). Enzymatic inactivation of the antibiotic compound may be achieved through genes encoding esterases (ere(A) and ere(B)), transferases (lnu(A) and lnu(B)) or phosphorylases (mph(A) and mph(B)). Genes encoding efflux or inactivation of the antibiotic compound will mostly confer resistance to only one or two of the antibiotic classes of the MLS_B complex (Roberts et al., 1999).

In the current literature no recent data are available on the occurrence of macrolide and lincosamide resistance among cloacal Lactobacillus isolates from poultry or on the eventual presence of inducible MLS_B resistance phenotypes in these isolates. So far, only erm(B) and erm(GT) have been described in lactobacilli: Lin & Chung (1999) found erm(B)-carrying plasmids in Lactobacillus reuteri from poultry origin; Stroman et al. (2003) found erm(B) in L. crispatus isolated from suckling piglets, and Martel et al. (2003) found erm(B) in Lactobacillus animalis, Lactobacillus johnsonii, L. reuteri and Lactobacillus salivarius originating from piglets. erm(GT), an erm gene similar to erm(C), was retrieved from lactobacilli by Tannock et al. (1994).

The aim of this study was to determine the presence of phenotypic and genotypic macrolide and lincosamide resistance characteristics in cloacal Lactobacillus isolates from poultry as an aid in assessing the potential risk of poultry lactobacilli acting as a pool of resistance genes.

Materials and Methods

Isolation and identification of lactobacilli

Cloacal swabs were taken from 40 4-week-old broilers on Belgium farms. Twenty farms were selected at random and two broilers per farm were sampled. Swabs were inoculated on Man, Rogosa, Sharpe (MRS) agar (LAB M, Bury, UK) and incubated anaerobically (GasPakPlus, BBL, Cockeysville, USA) at 37°C for 48 h. Different colony types were selected on the basis of several macroscopic characteristics: colour, opacity, roughness of the surface and edges and microscopic appearance of the bacterial cells following Gram-staining. Up to five colonies were selected per plate.

Selected bacterial colonies were further identified using API 50 CH (BioMerieux, Marcy l'Etoile, France), tDNA-PCR (Baele et al., 2002) and ¹⁶S rDNA-sequencing (De Graef et al., 2003). Isolates from an individual bird that belonged to the same Lactobacillus species and had identical API patterns were included only once in the final results.

Agar dilution tests

Minimal inhibitory concentration (MIC) values were determined for erythromycin, tylosin and lincomycin (LAB M) on MRS agar plates containing two-fold serial antibiotic dilutions ranging from 64 to 0.06 µg/ml. The antibiotics were dissolved in appropriate solvents and then further diluted in distilled water as outlined in the National Committee for Clinical Laboratory Standards guidelines (NCCLS, 2002). The reference strain used for determination of MIC values was Enterococcus faecalis ATCC 29212.

The bacteria under test were first grown in MRS broth (37°C, 24 h) and then diluted in phosphate-buffered saline to a turbidity of 0.5 on the McFarland scale. These suspensions were diluted 1/10 in phosphate-buffered saline and approximately 1×10⁵ colony-forming units were inoculated onto the plates with a multipoint inoculator (Denley Mast, Sussex, UK). Plates were incubated anaerobically (GasPakPlus, BBL) at 37°C for 24 h. The MIC was defined as the lowest concentration producing no visible growth. Antimicrobial-free agar plates were included as a control for normal growth. Erythromycin was tested at pH 6.4 (normal pH of MRS agar) and pH 7.2 (adjusted with NaOH).

Disk diffusion tests

Antimicrobial susceptibility phenotypes were determined by disk diffusion on MRS agar using erythromycin, tylosin, lincomycin and clindamycin (Rosco Diagnostics, Taastrup, Denmark). For detection of induced resistance disks of erythromycin, tylosin, lincomycin and clindamycin were placed 20 mm apart on MRS agar, which had been inoculated with a swab dipped into a bacterial suspension with a turbidity equivalent to that of 0.5 on the McFarland standard. After 18 h of incubation at 37°C the plates were examined for blunting of a zone of inhibition around a disk near to another disk indicating inducible resistance.

Resistance genotype

All Lactobacillus isolates were tested for the presence of erm(A), erm(B), erm(C), mef(A) and lnu(A) genes. Primer sequences, amplification protocols and amplicon sizes for erm(A), erm(B), erm(C), mef(A) and lnu(A) were derived from previous publications (Sutcliffe et al., 1996; Lina et al., 1999). DNA preparation and gel electrophoresis were performed as previously described by Martel et al. (2001). Reference strains used as positive controls were Staphylococcus aureus RN1389 (erm(A)), Streptococcus pyogenes STP016 (erm(B) and mef(A)), Staphylococcus aureus RN4220 (erm(C)) and Staphylococcus haemolyticus BM4610 (lnu(A)) (Horinouchi & Weisblum, 1982; Abraham & Rood, 1985; Murphy, 1985; Kimpe et al., 2003).

Results

Lactobacillus isolates

Eighty-eight Lactobacillus isolates were obtained from the 40 broilers. Based on API 50 CH and tDNA-PCR results, 31 isolated were identified as L. salivarius subsp. salivarius, eight as L. reuteri and 49 isolates as belonging to either Lactobacillus amylovorus, L. crispatus or Lactobacillus gallinarum. Further identification of this last group was performed with ¹⁶S rDNA sequencing and resulted in the identification of eight L. amylovorus, 10 L. crispatus and 31 L. gallinarum strains.

Agar dilution tests

No differences were observed for erythromycin sensitivity testing at pH 6.4 or 7.2. The distribution of MIC values for erythromycin, tylosin and lincomycin is presented in Table 1.

Table 1.  Distribution of MIC values of erythromycin, tylosin and lincomycin for Lactobacillus isolates obtained from broilers

	MIC (µg/ml)
Lactobacillus sp.	≤0.06	0.12	0.25	0.5	1	2	4	8	16	32	64	>64	Number of isolates tested	Percentage resistance
L. crispatus			2	1							1^a	6^a	10	70
L. salivarius subsp. salivarius					1	2	2	1	2^a	1^a	4^a	18^a	31	81
L. reuteri					1							7^a	8	88
L. amylovorus												8^a	8	100
L. gallinarum			8	2		1					4^a	16^a	31	65
L. crispatus	1	2										7^a	10	70
L. salivarius subsp. salivarius				3	3					1^a	13^a	11^a	31	81
L. reuteri				1								7^a	8	88
L. amylovorus												8^a	8	100
L. gallinarum	3	5	1	1								21^a	31	68
L. crispatus							1		1			8^a	10	80
L. salivarius subsp. salivarius								2	2	3	4^a	20^a	31	77
L. reuteri												8^a	8	100
L. amylovorus									1		1^a	6^a	8	88
L. gallinarum						1	1		2^a	2^a	1^a	24^a	31	94
Data are number of isolates.
^aIsolates are considered to have acquired resistance.

Interpretation of MIC data was based on the microbiological criteria for resistance because breakpoints for testing lactobacilli are not given in the NCCLS guidelines or other scientific literature. In many cases the MICs showed a bimodal distribution. Isolates in the higher range of the MICs of these antibiotics were considered to have acquired resistance. An extended frequency distribution range of MICs of tylosin and lincomycin was seen for L. salivarius subsp. salivarius isolates. The division between susceptible and resistant isolates was less clear here, but those with MICs ≥16 µg/ml for tylosin and ≥64 µg/ml for lincomycin were considered to have acquired resistance because a correlation could be made with carriage of resistance genes. MICs of tylosin and erythromycin for all L. amylovorus isolates and MICs of lincomycin for all L. reuteri isolates were >64 µg/ml, indicating that they were resistant against these antimicrobial agents.

Based on these criteria, the percentage of resistant isolates for the different antimicrobial agents tested is presented in Table 1. Resistance to both macrolides and lincosamides was found in 76.1% of the investigated isolates, 11.4% was resistant to lincosamides only, 1.1% to macrolides only and 10.4% was sensitive to both.

Disk diffusion tests

Three out of the 88 Lactobacillus isolates tested were found to have inducible macrolide–lincosamide resistance. In all three cases, the inducing antibiotic was erythromycin. One L. reuteri isolate had inducible tylosin and lincomycin resistance, one L. salivarius subsp. salivarius isolate had inducible tylosin and clindamycin resistance, and in one L. gallinarum isolate inducible lincomycin and clindamycin resistance was manifested.

Resistance genotype

Resistance genotypes of the different Lactobacillus isolates and their relationship with resistance phenotypes are presented in Table 2, which shows that carriage of the erm(B) gene always concurred with phenotypic resistance against macrolides and lincosamides. Of two isolates possessing the erm(C) gene, one possessed a macrolide and lincosamide-resistant phenotype while the other, a L. salivarius subsp. salivarius isolate, was found to be phenotypically sensitive to macrolides as well as lincosamides. None of the macrolide-sensitive and lincomycin-resistant isolates possessed a lnu(A) gene. Two L. reuteri isolates that possesed a lnu(A) gene had phenotypic macrolide and lincomycin resistance. In one of these isolates this phenotype was linked to the presence of an erm(B) gene. One L. salivarius subsp. salivarius isolate carried the mef(A) gene as well as the erm(B) gene and displayed phenotypic macrolide and lincomycin resistance.

Table 2.  Resistance phenotypes and genotypes among Lactobacillus isolates obtained from broilers

Lactobacillus species	Total number of isolates	Resistance phenotype^a/resistance genotype^b	Number of isolates
L. crispatus	10	T + E+L/erm(B)	3
		T + E+L/–	4
		L/–	1
		–/–	2
L. salivarius subsp. salivaruis	31	T + E+L/erm(B)	15
		T + E+L/erm(B) + mef(A)	1
		T + E+L/–	8
		T + E/–	1
		–/erm(C)	1
		–/–	5
L. reuteri	8	T + E+L/erm(B)	5
		T + E+L/erm(B) + Inu(A)	1
		T + E+L/Inu(A)	1
		L/–	1
L. amylovorus	8	T + E+L/erm(B)	5
		T + E+L/–	3
L. gallinarum	31	T + E+L/erm(B)	11
		T + E+L/erm(C)	1
		T + E+L/–	9
		L/–	8
		–/–	2
^aT, E or L, resistance to tylosin (T), erthromycin (E) or lincomycin (L) only; T + E, T + L, E + L, resistance to both tylosin and erthromycin, tylosin and lincomycin, or erthromycin and lincomycin; T + E+L, resistance to tylosin, erthromycin and lincomycin; –, susceptible to all antibiotics tested.
^b erm(B), erm(C), Inu(A) or mef(A), strains tested positive for the presence of erm(B), erm(C), Inu(A) or mef(A) genes; erm(B) + Inu(A), erm(B) + mef(A), strains tested positive for the presence of both erm(B) and Inu(A) or for erm(B) and mef(A) genes; –, strains tested negative for the presence of erm(B), erm(C), Inu(A) and mef(A) genes.

Discussion

In order to obtain a broad idea of the resistance situation of Lactobacillus species on broiler farms, and since prevalence was expected to be high, we elected to study a few isolates from each of many different farms, rather than many isolates from a few farms. It was for this reason that 20 Belgian farms were selected and two broilers per farm were sampled.

Five different Lactobacillus species were identified as being predominantly present in broiler cloacae: L. crispatus, L salivarius subsp. salivarius, L. amylovorus, L. gallinarum and L. reuteri. Lan et al. (2002) also found L. crispatus, L. salivarius and L. reuteri among the most important lactobacilli present in the caeca of 12-week-old White Leghorn laying hens. They also found Lactobacillus acidophilus to be an important caecum floral constituent, whereas in the present study L. amylovorus and L. gallinarum were the additional species frequently encountered in 4-week-old broilers. However Lan et al. (2002) did not cultivate their bacteria but performed DNA sequencing on amplification products obtained from DNA extracted directly from chicken caeca.

Differences in composition of the Lactobacillus flora might also be due to genetic factors (Tannock et al., 1994), age and intestinal location (Lu et al., 2003), ecological factors such as diet and lifestyle (Tannock et al., 1994) and even the type of housing (De Waard et al., 2002).

In our study a clear bimodal distribution of MIC ranges was observed for most antimicrobials tested, and for these interpretation of MIC values with respect to the prevalence of acquired resistance was straightforward. An extended frequency distribution range of MICs of tylosin and lincomycin was seen for L. salivarius subsp. salivarius with no clear bimodal distribution, making interpretation more difficult. The critical concentration differentiating between susceptibility and resistance was taken somewhat arbitrarily since MIC breakpoints for testing lactobacilli are not given in NCCLS guidelines or in other guidelines. In previous studies, Dutta & Devriese (1981) considered Lactobacillus strains from poultry with an MIC ≥32 µg/ml for tylosin or lincomycin to be resistant. In the present study, isolates with MICs ≥16 µg/ml for tylosin and ≥64 µg/ml for lincomycin were considered to have acquired resistance because they carried the corresponding resistance genes and these genes were not detected in isolates with lower MICs.

One L. salivarius subsp. salivarius strain was shown to carry an erm(C) gene without displaying phenotypic macrolide or lincosamide resistance. This lack of correlation between resistance phenotype and genotype may be due to defective expression of the resistance gene and has been described earlier (Martineau et al., 2000; Martel et al., 2003).

MICs of tylosin and erythromycin for all L. amylovorus strains and MICs of lincomycin for all L. reuteri strains were ≥64 µg/ml but the number of isolates tested was too few to decide whether this should be considered as acquired or intrinsic resistance. In any case, for interpretation of MIC values, identification of the Lactobacillus strains to species level is essential since sensitivity to antimicrobial agents may vary markedly between species belonging to the same genus.

In the present study, 78 and 87% of isolates of poultry lactobacilli were found to have resistance to the macrolide and lincosamide antibiotics, respectively. The vast majority of resistant strains (96%) displayed constitutive resistance, which is in accordance with general findings on MLS_B resistance by Roberts et al. (1999). The authors are not aware of any specific data in the literature on the occurrence of inducible MLS_B resistance phenotypes in lactobacilli. In a study performed by Dutta & Devriese (1981), the percentages of poultry Lactobacillus isolates with acquired resistance to macrolides and lincosamides were 46% with both antibiotics, thus indicating that resistance against macrolides and lincosamides has increased over the past 20 years.

There is a substantial discrepancy between our results and those from studies on erythromycin resistance in lactobacilli of human origin, which report very low percentages of resistance among isolates (Felten et al., 1999; Mandar et al., 2001). The cause of this discrepancy might be a difference in the species tested or a difference in test methods but it could also be due to the use of antimicrobial drugs in the broiler industry. In Belgium, eight out of the 21 antimicrobial products registered for oral use in poultry contain macrolide or lincosamide antibiotics (Anonymous, 2003) and are often used in poultry medicine. Furthermore, the instructions with some of these products encourage prophylactic use from hatching until the age of 3 to 5 days and for 1 day at the age of 4 weeks. However, at present no specific numeric data on antibiotic use in the Belgian poultry industry are available (BAPCOC, 2004).

The results of this study confirm that erm(B) is the most frequently encountered erm gene in lactobacilli isolated from poultry (Lin & Chung, 1999). This gene is also widely spread in other Gram-positive poultry-derived bacterial species, such as enterococci and streptococci (Aarestrup et al., 2000; Khan et al., 2002; Simjee et al., 2002).

This appears to be the first report of erm(C), mef(A) and lnu(A) genes in lactobacilli, although these genes have been found previously in other Gram-positive poultry-derived bacterial species.

The erm(A) and erm(C) genes were predominantly found in staphylococci strains derived from poultry and associated products (litter and meat) (Aarestrup et al., 2000; Khan et al., 2002; Simjee et al., 2002) while the lnu(A) gene was demonstrated in a Clostridium perfringens isolate from poultry (Martel et al., 2004).

Our study shows that lactobacilli may contain a pool of resistance genes in the chicken gut. Since these genes might be transferred to potentially pathogenic bacteria and agents of zoonotic importance, their presence in poultry lactobacilli might constitute a possible public health hazard. Further studies are needed to elucidate this possibility.