Characterization of methicillin-resistant Staphylococcus aureus from healthy carrier chickens

Abstract

Methicillin-resistant Staphylococcus aureus (MRSA) has long been recognized as an important pathogen in human medicine leading to hospital and community-acquired infections. However, it is now also considered a growing problem in veterinary medicine, although causing little or no disease. Although MRSA has already been detected in livestock including poultry, little is known about the epidemiology of MRSA in broiler and layer chickens. We therefore investigated 372 poultry farms in Belgium. We also compared the isolation method recommended by the European Food Safety Authority using two enrichment steps with an isolation method using only one enrichment step. Isolated MRSA was characterized by means of antimicrobial resistance profiling, spa typing, multi-locus sequence typing, and SCCmec typing. MRSA prevalence was 0.8% using the double broth enrichment method, while using the single broth enrichment method it was 1.8%. Five MRSA strains belonged to the livestock-associated (LA) MRSA ST398 (four with spa type t011 and one with t899), and three to the hospital-acquired MRSA ST239 spa type t037. The ST239 strains carried SCCmec type III while those belonging to ST398 carried SCCmec type IV or V. All isolates showed additional resistance to erythromycin and tetracycline apart from the expected resistance to cefoxitin and penicillin. All strains were susceptible to linezolid, mupirocin and vancomycin. In conclusion, a higher sensitivity for the isolation of LA-MRSA was obtained using only one enrichment step. While the typical LA-MRSA ST398 was present at low prevalence in poultry, human-associated strains have also been found.

Introduction

Methicillin-resistant Staphylococcus aureus (MRSA) strains are an important cause of hospital-acquired and community-acquired infections worldwide (Stewart & Holt, 1963; von Eiff et al., 2001; Kluytmans-VandenBergh & Kluytmans, 2006). However, MRSA strains are no longer confined to healthcare settings, and are also a growing problem in veterinary medicine. In livestock, MRSA was first reported in a case of bovine mastitis (Devriese et al., 1972). It has been shown that this clone was human associated. In animals, MRSA infections were mainly of human origin until 2005, when a high prevalence of a specific clone of MRSA was reported in pigs in the Netherlands. This clone was later named “livestock associated MRSA” (LA-MRSA) and corresponded to the clonal complex (CC) 398 (Voss et al., 2005). Ever since, this clone has been found in many animal species, including poultry, all over the world (Persoons et al., 2009; Van den Eede et al., 2009; Fessler et al., 2010; Graveland et al., 2010; Graveland et al. 2011; Crombé et al., 2012).

In poultry, a first report on MRSA came from South Korea in 2003 (Lee, 2003). The strains were both human and animal associated. Other studies demonstrated MRSA in raw chicken meat (Kitai et al., 2005; Lee, 2006; Dohoo et al., 2009) and broilers at slaughterhouses (Mulders et al., 2010). LA-MRSA isolates in poultry have previously been found in the Netherlands (Leenders et al., 2007; Geenen et al., 2012) and Belgium (Nemati et al., 2008; Persoons et al., 2009; Vanderhaeghen et al., 2010a; Pletincks et al., 2011). However, these studies were rather limited and different isolation methods were used. Hence, a detailed understanding of the epidemiology of MRSA in poultry has so far been lacking. We therefore conducted a national survey on the prevalence of MRSA in both layers and broilers. Since isolation methods are of importance in prevalence estimations, we compared the method proposed by the European Food Safety Authority (Anonymous, 2009) with a less laborious modified method, which will allow international comparisons.

Materials and Methods

Sample origin and isolation methods

A total of 372 farms, of which 92 were raising broilers and 280 were egg producing, were sampled in 2011 all over Belgium. In each farm, 20 chickens were swabbed in the nostrils. Swabs were pooled at the laboratory and two different isolation methods were used. In the first isolation method, proposed by the European Food Safety Authority, pooled swabs were inoculated in Mueller–Hinton broth (Becton Dickinson, Franklin Lakes, New Jersey, USA) supplemented with NaCl (6.5%) at 37°C for 20 to 24 h. One millilitre of this broth was added to Tryptic Soy Broth (Becton Dickinson, Franklin Lakes, New Jersey, USA) supplemented with cefoxitin (3.5 mg/l) and aztreonam (75 mg/l) and was incubated at 37°C overnight. Ten microlitres of this broth were then plated on MRSA-ID (bioMérieux, Marcy-l'Etoile, France) and incubated for 48 h at 37°C. At both 24 and 48 h, plates were inspected and suspected colonies were purified on a Columbia sheep blood agar plate (Bio Rad Laboratories, Nazareth Eke, Belgium). These plates were incubated overnight at 37°C (Anonymous, 2007). The alternative protocol was applied to 332 farms out of the 372 sampled farms, 81 raising broilers and 251 egg producing. Both methods were applied using the same Mueller–Hinton broth in which swabs were pooled. This second isolation method differed from the above-described protocol in that the second enrichment in antibiotic-supplemented broth was omitted. For this reason, the first isolation method developed by the European Food Safety Authority will further be referenced as double broth enrichment method (DBEM) and the second one as the single broth enrichment method (SBEM).

DNA extraction, MRSA identification and characterization

DNA was extracted from all isolates as previously described (Vanderhaeghen et al., 2010b). MRSA identification was performed using a triplex polymerase chain reaction (PCR), previously published by Maes et al. (2002). This PCR allows detection of the staphylococcal-specific 16S rRNA gene, the nuc gene specific for S. aureus, and the presence of the mecA gene responsible for methicillin resistance.

All MRSA isolates were spa typed as previously described (Harmsen et al., 2003), using Ridom StaphType software (www.ridom.de/staphtype). CC398 PCR was performed on all MRSA following the protocol described by Stegger et al. (2011), which allows the rapid identification of S. aureus ST398. MRSA isolates that were negative in the CC398 were subjected to multi-locus sequence typing (Enright et al., 2000). Sequences of internal fragments were then compared with the international database (http://saureus.mlst.net) to obtain the sequence type (ST). SCCmec typing of all MRSA was performed using the two multiplex PCRs to type the mec complex and the ccr complex as described by Kondo et al. (2007). Appropriate control strains were used.

Determination of antimicrobial resistance

Antimicrobial resistance was determined using the micro-broth dilution method (Sensititre; Trek Diagnostic Systems, Magellan Biosciences, Tampa, Florida, USA) following the manufacturer's instructions and using the European Committee on Antimicrobial Susceptibility Testing breakpoints. Data from the Committee's MIC distribution website was last accessed on 30 November 2012 (http://www.eucast.org). The antibiotics tested were those included in the European Committee on Antimicrobial Susceptibility Testing custom panel plate for Staphylococcus. The minimum inhibitory concentration (MIC) was defined as the lowest concentration at which no visible growth could be detected.

Statistical analysis

Cohen's kappa coefficient was calculated and interpreted according to Landis & Koch (1977) in order to compare the two isolation methods. This analysis was carried out on those 332 farms for which both isolation methods were available (Table 1). The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), likelihood ratio positive (LR+) and negative (LR–) of both methods were also calculated using the previously described formulae (Dohoo et al., 2009) and Win Episcope 2.0 (Clive, Edinburgh, United Kingdom). Cohen's kappa, Pearson chi-square and Fisher's exact tests were computed using IBM SPSS Statistics® version 20.0 (IBM, Armonk, New York, USA).

Table 1.  Comparison of MRSA isolates detected using DBEM or SBEM.

	DBEM
SBEM	Number of positive results	Number of negative results	Total number
Number of positive results	1	5	6
Number of negative results	2	324	326
Total number	3	329	332

Results

Methicillin-resistant Staphylococcus aureus detection

Out of 372 farms tested, three were positive for MRSA (0.8%, 95% confidence interval [CI] = 0.76 to 0.84) using the DBEM (Table 2). All of these positive farms raised broiler chickens, leading to a total prevalence of 3.3% (95% CI = 2.3 to 4.2) for broiler farms (Table 3). Conversely, six farms were positive (1.8%, 95% CI = 1.68 to 1.92) using the SBEM (Table 2). Two positive farms were raising layers and four were raising broilers. Prevalence with SBEM (Table 3) can thus be split into broiler prevalence (4.8%, 95% CI = 3.7 to 5.9) and layer prevalence (0.8%, 95% CI = 0.75 to 0.84). Interestingly, only one farm was found positive by both methods. Thus two farms that detected positive with the DBEM were not detected with the other method, and five farms that detected positive with the SBEM were not detected with the other method (Table 1). A total of eight farms (2.4%, 95% CI = 1.92 to 2.86), broilers and layers included, were therefore found to be MRSA-positive using both methods (Table 2).

Table 2.  Total number and prevalence of MRSA-positive farms using DBEM or SBEM.

Isolation method^a	Number of positive farms	Prevalence (%)	Farm type
DBEM	3	0.8 (n = 372)	Broilers
SBEM	6	1.8 (n = 332)	2 layers and 4 broilers
DBEM and SBEM^b	8	2.4 (n = 332)	Layers and broilers
n, total number of broiler or layer farms. ^aPrevalence is computed out of 372 farms for DBEM, and out of 332 farms for SBEM. ^bDBEM and SBEM is the comparison of both methods used in parallel.

Table 3.  MRSA prevalence in different farms for DBEM and SBEM.

Isolation method^a	Broilers (%)	Layers (%)
DBEM	3.3 (n = 92)	0.0 (n = 280)
SBEM	4.8 (n = 81)	0.8 (n = 251)
n, total number of broiler or layer farms. ^aPrevalence is computed out of 92 broilers and 280 layers for DBEM, and computed out of 81 broilers and 251 layers for SBEM.

Cohen's kappa coefficient was 0.21, indicating a fair agreement between the two methods, and Fisher's exact test showed no significant difference between the two methods (P>0.05). Based on the assumption that all farms that tested positive in at least one test are true positive farms (n = 8), the sensitivity of the DBEM and the SBEM method is 0.375 (95% CI = 3.95 to 71.0) and 0.75 (95% CI 45.0 to 100.0), respectively (Table 4). Both PPVs are equal to one since there are no false positives. NPVs of the DBEM and SBEM methods are 0.985 (95% CI = 97.2 to 99.8) and 0.994 (95% CI = 98.5 to 100.0), respectively. The LR+ for DBEM and SBEM methods was 25 and 125, respectively, whereas the LR– for DBEM and SBEM methods was 0.635 and 0.252, respectively. The specificity, PPV and NPV for both methods were not significantly different (P>0.05). However, there was a significant difference in MRSA prevalence between broiler and layer farms (Fischer exact P<0.05).

Table 4.  Comparison of sensitivity, specificity, positive and negative predictive values for the isolation methods used in this study.

	DBEM (95% CI)	SBEM (95% CI)
Sensitivity	0.375 (3.9 to 71.0)	0.75 (45.0 to 100.0)
Specificity	1	1
PPV	1	1
NPV	0.985 (97.2 to 99.8)	0.994 (98.5 to 100.0)

Methicillin-resistant Staphylococcus aureus characterization

Among the MRSA isolates recovered, three different spa types were detected; four strains belonged to t011, three to t037 and one to t899. The SBEM detected two t037, two t011, and one t899, while the DBEM detected one t037 and one t011. However, one t011 was detected by both methods. All t011 and the single t899 MRSA were isolated from broilers, but t037 strains were isolated from layers and broilers (Table 5). All t011 and t899 isolates were ST398, while the three MRSA type t037 strains belonged to ST239. These three strains carried SCCmec type III (3A), while the ST398 strains carried SCCmec IV (2B) or SCCmec V (5C2) cassettes.

Table 5.  Genotyping of MRSA.

Strain	spa type	Sequence type	ccr complex	mec	SCCmec type	Origin	Isolation method
M72	t037	ST239	A3/B3	A	III	Layer	SBEM
M86	t037	ST239	A3/B3	A	III	Layer	SBEM
M118b	t899	ST398	A2/B2	B	IV	Broiler	SBEM
M213	t011	ST398	A2/B2	B	IV	Broiler	SBEM + DBEM
M282	t037	ST239	A3/B3	B	III	Broiler	DBEM
M286	t011	ST398	C2	C	V	Broiler	DBEM
M363	t011	ST398	C2	C	V	Broiler	SBEM
M371	t011	ST398	A2/B2	B	IV	Broiler	SBEM

All ST239 strains showed the same resistance profile and were resistant to cefoxitin, penicillin, erythromycin, tetracycline, chloramphenicol, kanamycin, rifampicin, sulfamethoxazole, and streptomycin. None were resistant to ciprofloxacin, clindamycin, fusidic acid, gentamicin, quinupristin/dalfopristin, tiamulin, and trimethoprim. All ST398 strains were resistant to cefoxitin, penicillin, erythromycin, tetracycline, clindamycin, and trimethoprim. Four ST398 isolates showed resistance to gentamicin, kanamycin, ciprofloxacin, and sulfamethoxazole. Two ST398 isolates were resistant to streptomycin, chloramphenicol, fusidic acid, and tiamulin, and one was resistant to rifampicin and quinupristin/dalfopristin (Table 6).

Table 6.  Antimicrobial resistance of all MRSA strains isolated in this study.

Strain	CHL	CIP	CLI	ERY	FOX	FUS	GEN	KAN	LZD	MUP	PEN	RIF	SMX	STR	SYN	TET	TIA	TMP	VAN
M72	64^a	≤0.25	≤0.12	>8	>16	≤0.5	≤1	>64	2	≤0.5	>2	>0.5	512	>32	≤0.5	>16	≤0.5	≤2	≤1
M86	64	≤0.25	≤0.12	>8	>16	≤0.5	≤1	>64	≤1	1	>2	>0.5	512	>32	≤0.5	>16	≤0.5	≤2	≤1
M282	64	≤0.25	≤0.12	>8	>16	≤0.5	≤1	>64	≤1	≤0.5	>2	>0.5	>512	>32	≤0.5	>16	≤0.5	≤2	≤1
M118	≤4	2	>4	4	16	1	8	64	2	≤0.5	>2	0.03	≤64	8	1	>16	>4	>32	≤1
M213	64	0.5	>4	>8	>16	2	>16	>64	≤1	≤0.5	>2	>0.5	256	>32	≤0.5	>16	1	>32	≤1
M286	16	2	>4	>8	16	≤0.5	≤1	≤4	2	≤0.5	>2	≤0.016	≤64	8	1	>16	≤0.5	>32	≤1
M363	32	2	>4	>8	16	≤0.5	>16	>64	≤1	≤0.5	>2	≤0.016	256	>32	4	>16	>4	>32	≤1
M371	8	>8	>4	>8	16	≤0.5	>16	64	2	≤0.5	>2	≤0.016	≤64	8	1	>16	2	>32	≤1
CHL, chloramphenicol; CIP, ciprofloxacin; CLI, clindamycin; ERY, erythromycin; FOX, cefoxitin; FUS, fusidic acid; GEN, gentamicin; KAN, kanamycin; LZD, linezolid; MUP, mupirocin; PEN, penicillin; RIF, rifampicin; SMX, sulfamethoxazole; STR, streptomycin; SYN, quinupristin/dalfopristin; TET, tetracyclin; TIA, tiamulin; TMP, trimethoprim; VAN, vancomycin. ^aResults in bold indicate resistance breakpoints.

Discussion

In this study, we investigated 372 farms raising broilers or layer chickens in order to determine the MRSA prevalence. The prevalence was low, ranging from 0.8% (95% CI = 0.76 to 0.84) to 1.8% (95% CI = 1.68 to 1.92) depending on the isolation method used. In our study we showed that MRSA prevalence was significantly higher in broiler farms compared with layer farms. This may explain the low overall prevalence, since 75% of sampled farms raised layers. Although the inclusion of a broth supplemented with aztreonam and cefoxitin has previously been considered as improving MRSA recovery (Böcher et al., 2010), the comparison between both isolation methods showed that using this broth may be too selective for MRSA detection in a low-prevalence population. Indeed, only three farms were detected as positive with the DBEM method while the SBEM method detected six farms. However, using the SBEM, more non-S. aureus staphylococci were isolated (data not shown). Since these isolates were very similar to S. aureus on MRSA identification, this could lead to extended laboratory work. Although the differences between the two methods are not statistically significant, the most sensitive isolation method is preferred since it is important to detect as many positive farms as possible in a low-prevalence environment to avoid the further spread to other farms. The same swabs were used for both isolation methods and the isolation steps were performed in parallel by the same technicians. Therefore, it is unlikely that the differences between methods were due to sampling or other accidental influences.

An interesting finding in this study is the presence of the non-ST398. Indeed, next to the classical MRSA ST398 spa type t899 and t011, three hospital-acquired MRSA ST239 spa type t037 with SCCmec type III were isolated. ST239 clones are disseminated worldwide and are among the oldest MRSA clones found in Europe (Monecke et al., 2011). They account for 90% of the hospital-acquired MRSA in Asia and have also been detected in South America and are nowadays mainly circulating in Eastern Europe (Yamamoto et al., 2012). Interestingly, ST239 shows geographic variations in terms of the spa type and the t037 found in this study is thought to be the ancestral ST239 spa type (Harris et al., 2010). This spa type has been reported recently in different countries such as Malaysia and Russia (Neela et al., 2010; Yamamoto et al., 2012). Furthermore, while no MRSA were detected previously in layers (Persoons et al., 2009), MRSA ST239 was isolated both from broiler and layer farms.

All strains showed resistance to at least seven different antimicrobials and to a maximum of 14 out of 19 antimicrobials tested. As expected, all strains were resistant to penicillin and cefoxitin. All strains were also resistant to erythromycin and tetracycline. None was resistant to linezolid, mupirocin and vancomycin. In the recent study performed among poultry in Belgium by Persoons et al. (2009) all strains were susceptible to chloramphenicol, ciprofloxacin, quinupristin/dalfopistin, and rifampicin. In contrast, in our study, four (80%) ST398 strains were resistant to ciprofloxacin, two (40%) were resistant to chloramphenicol, and one to rifampicin and quinupristin/dalfopristin. Interestingly, the three ST239 shared the same resistance pattern, showing susceptibility to gentamicin, clindamycin and ciprofloxacin and resistance to chloramphenicol and rifampicin. These strains seem different from those isolated in Asia, where this clone is usually resistant to gentamicin, clindamycin and ciprofloxacin, and only in few cases resistant to rifampicin and susceptible to chloramphenicol (Kim et al., 2006).

Since MRSA ST239 spa type 037 is a hospital-acquired strain that is, to our knowledge, not reported in livestock, researchers and technicians that had worked in the laboratory during the surveillance were controlled in order to check their MRSA status. All were negative for MRSA. No information could be obtained about the MRSA status of the field workers or farmers. Furthermore, this spa type has not been recovered during the previous surveillance in hospitals in Belgium (Stien Vandendriessche, personal communication). This is, to our knowledge, the first report of ST239 spa type t037 in Belgium.

In conclusion, the MRSA prevalence in broiler farms was 3.3% with DBEM and 4.8% with SBEM, which is significantly higher than that in layer farms. Nevertheless, the overall prevalence is low. Since broiler chickens have a higher prevalence than layers it is important to take this into account for proper prevalence determination. Prevalence should then be seen as a function of the sampling and isolation methods. The common LA-MRSA ST398 has been detected but we found for the first time hospital-acquired MRSA ST239 spa type t037, which is not common in livestock or in the hospital according to the recent surveys conducted in Belgium. As yet the cause and origin of this clone in poultry is still unknown.

Acknowledgements

This research was funded by the Federal Agency for the Safety of the Food Chain and the EMIDA ERA-Net Project “Methicillin-resistant Staphylococcus aureus Lineages in Primary Productions: Multi-host Pathogen, Spill-over and Spill-back between Animals and Humans?” (project acronym LA-MRSA). Dr M.A. Argudín is supported by a research grant from the Fundación Alfonso Martín Escudero.

The authors thank Andy Lucchina and Déborah Petrone for technical assistance. They acknowledge Dr Florence Crombé, Dr Wannes Vanderhaeghen and Dr Yves Van der Stede for their critical review of the manuscript.