ABSTRACT
Livestock-associated methicillin-resistant Staphylococcus aureus (LA-MRSA) of clonal complex (CC) 398 has become a rising issue for public health. While it is known that >80% of pig farmers are colonized with LA-MRSA, only a few studies have assessed the situation for humans with occasional livestock contact. Recently it was shown that over 75% of scientific fieldworkers visiting pigsties were temporarily carrying LA-MRSA. To find out whether they were transiently or permanently colonized, we used whole-genome sequencing (WGS) data to analyze the relatedness of isolates from these recurrently LA-MRSA-positive fieldworkers and from corresponding pigsties. Sequences were analyzed using in silico typing (spa and core genomic multilocus sequence typing [cgMLST]), and the BEAST software package was used to examine phylogeny. In total, 81 samples from three fieldworkers on eight different pigsties over a period of 2.5 years were sequenced. All isolates belonged to spa type t011, t034, or t2011, with different types found in the same fieldworker at different time points. Analysis of cgMLST revealed nine genotypic clusters, mostly correlating with the pigsty on which they were sampled. Fieldworker isolates clustered with the samples from farms that were visited on the same day. BEAST analysis corroborated the cgMLST-based clustering and suggests an origin of the lineage about 22 years ago. We conclude that nasal LA-MRSA colonization among humans with occasional livestock contact is common but most likely only temporary. Furthermore, we showed that the Western German LA-MRSA CC398 originated in the late 1990s and diversified into farm-specific genotypes, which stay relatively consistent over time.
INTRODUCTION
Livestock-associated methicillin-resistant Staphylococcus aureus (LA-MRSA) has become a rising issue for public health. The most prevalent LA-MRSA lineage is characterized by multilocus sequence typing (MLST) of clonal complex 398 (CC398), which is nowadays prevalent on more than 70% of European pig farms (1).
Humans with regular exposure to swine, such as farmers, slaughterhouse staff, and veterinarians, are frequent nasal carriers of LA-MRSA CC398, most likely due to hand-nose contact or inhalation of MRSA-contaminated dust. Carriage rates of more than 80% were found among German pig farmers, with half of them being colonized even after a period without contact to their farms (2). Since a large proportion of S. aureus infections are of endogenous origin (3), farmers bear an increased risk to develop severe infections (4).
While the prevalence of CC398 carriage or persistent colonization among pig farmers and other persons with direct livestock exposure is well investigated (4), the situation of humans with occasional or short-term livestock exposure is less clear. However, Van Cleef and colleagues (5) demonstrated that among epidemiological fieldworkers who visited pig farms in the Netherlands, 17% acquired LA-MRSA, but 94% appeared negative again when retested after 24 h. None of the fieldworkers acquired long-term colonization. In addition, Angen and colleagues (6) sampled 34 human volunteers, who had stayed for 1 h in an LA-MRSA-contaminated pig stable. They found that 94% of the volunteers acquired LA-MRSA, but 2 h after leaving the farm, the nasal MRSA count had decreased to unquantifiable levels in 95%. After 2 days, MRSA was no longer detectable in 94% of the volunteers.
Similarly, in a longitudinal study we performed on German pig farms, nasal LA-MRSA carriage was assessed among agricultural scientific fieldworkers, who repeatedly visited pigsties in order to take environmental samples and nasal swabs from pigs (7). Nasal swabs were taken from the fieldworkers before and directly after pigsty visits. Although temporal colonization rates after exposure were above 75%, all fieldworkers tested negative again prior to their next visit. These results raised the question of whether the fieldworkers were repeatedly colonized with new strains during each of their consecutive visits or whether each recurrent contact with the pigsty environment led to the temporary disclosure of a persistent clone that was suppressed under microbiological detection limits. Although it has been shown before that occasional farm visitors, who acquired LA-MRSA, test negative within 24 h, which makes a permanent colonization unlikely (5, 6), the correlation of genotypes found in the visitors’ nares and on the visited farms has so far only been done on the spa level (5). Since spa types are quite homogenous on different farms (8), this is not sufficient to prove that the detected MRSA clones in the visitors were the same as on the last-visited farms. An in-depth analysis of the genotypes present in visitors and visited farms is needed to confirm the hypothesis that LA-MRSA in humans with short-term livestock contact is only transient.
Therefore, a subset of isolates from recurrently LA-MRSA-positive fieldworkers and the corresponding visited pig farms were genotyped using whole-genome sequencing (WGS) to examine their relatedness and solve the question whether the fieldworkers were transiently or permanently colonized.
(Part of this work was presented as a poster at the 12th International Meeting on Microbial Epidemiological Markers [IMMEM XII] in Dubrovnik, Croatia.)
MATERIALS AND METHODS
Bacterial culturing.Detailed information on the analysis of nasal and environmental swabs and MRSA detection can be inferred from Müller and colleagues’ work (7). In short, swabs were transferred into 9 ml Mueller-Hinton broth plus 6.5% NaCl (Mediaproducts BV, Groningen, Netherlands) and incubated at 37°C for 18 ± 2 h. From the Mueller-Hinton broth plus 6.5% NaCl, 500 μl was added to 5 ml tryptone soya broth plus cefoxitin/aztreonam (Mediaproducts BV) and grown over 18 ± 2 h at 37°C. Subsequently 10 μl of the enriched cultures was inoculated on chromID MRSA SMART agar (bioMérieux, Marcy l’Étoile, France) and incubated at 37°C for 24 h.
Bacterial isolates.We chose three different fieldworkers (A, C, and D) from the above-mentioned study (7), whose MRSA status was recorded before and after pigsty visits over 2.5 years. MRSA isolates from nasal swabs taken after each visit as well as corresponding isolates from the pigs and environments of the respective visited farms were analyzed. In total, 81 isolates from these three workers (A, n = 12, C, n = 9; D, n = 13), from two farmers (n = 2), from 16 pigs (n = 16), and from environments of eight pig farms (n = 29) were chosen for WGS. The samples were taken on 21 different days between October 2015 and July 2018 on eight different pig farms as illustrated in Fig. 1. All pig farms were located in Westphalia, a region in Western Germany. This data set represents all available MRSA-positive samples from the three workers and at least one environmental isolate and one pig isolate per farm visit. If multiple samples from the same source are included, they represent either different spa types or different environments (i.e., air, dust, or surface). If a sample type is missing, no MRSA-positive isolate was available for this source. Detailed information on all isolates, including sample names and sample sites is provided in Table S1 in the supplemental material.
Illustration of sample scheme. Each symbol represents one isolate. Samples taken from three different fieldworkers (A, C, and D) on 21 different days are represented with boxes colored according to the visited farms. Pig, farmer, and barn symbols indicate additional samples from pig snouts, farmers, or the farm environment of the respective farms. Farm visits without available fieldworker samples are marked with an asterisk. Detailed information on all samples, including sample names, is provided in Table S1.
WGS and genotyping.The bacterial isolates had been stored at –80°C prior to our analysis. spa typing had been performed before following a previously described protocol (9). DNA extraction and sequencing were conducted on an Illumina NextSeq500 (Illumina, Inc., San Diego, CA, USA) as described before (10) with a 150-bp paired-end protocol in accordance with the manufacturer (Illumina, Inc.). The resulting reads were quality trimmed to an average base quality of 30 in a window of 20 bases and down-sampled to a coverage of 120 using Ridom SeqSphere+ software version 6.0.2 (Ridom GmbH, Münster, Germany) (11). Subsequently, the reads were mapped against NCBI reference no. NC_017333 (12) or NC_018608 (13) according to their spa types (Table S1) using Burrows-Wheeler-Aligner version 0.6.1 (14) in Ridom SeqSphere+.
For core genome MLST (cgMLST)-based genotyping, the 1,861 target genes of the S. aureus cgMLST scheme (15) were extracted from the resulting consensus sequencing. Subsequently, a minimum-spanning tree (MST) was created based on the allelic profiles of the cgMLST target genes with the parameter “ignore missing values in pairwise comparisons.” Isolates differing in ≤24 alleles were defined as a cluster following Leopold and colleagues’ definition (15). All analyses were conducted in Ridom SeqSphere+.
Phylogenetic analysis.For phylogenetic analysis, the cgMLST target sequences were extracted and concatenated for each sample. Failed (e.g., low-quality, frameshift mutation) or missing targets in one or more samples were excluded from all samples. A multiple sequence alignment (MSA) of the concatenated cgMLST sequences of all 81 samples was created with the online version of Mafft version 7 (16). A phylogenetic analysis, including ancestral dating, based on the resulting MSA was conducted using the BEAST software package version 1.10.4 (17). The analysis was run using the generalized time-reversible (GTR) substitution model with site heterogeneity and four gamma categories and a constant coalescent. A strict clock was chosen, and the prior for the clock rate was set to uniform (0 to 1, initial = 0.0001). Sampling dates were included as tip dates, and sampling sites were used as an additional trait. The chain was run for 200,000,000 steps and logged every 20,000 states. All other parameters were left at default settings. The BEAST analysis was run on the CIPRES Science gateway (https://www.phylo.org/). For quality control, three independent runs were started and checked for convergence. Moreover, a priors-only file was run to ensure the informativeness of the data. After quality control in Tracer version 1.7.1 (18), a maximum clade credibility tree (mcc) was calculated in TreeAnnotator version 1.10.4 from the BEAST package with a burn-in of 10% and median node heights. The resulting mcc was visualized in iTol version 4 (19).
Data availability.All raw reads generated were submitted to the European Nucleotide Archive (http://www.ebi.ac.uk/ena/) under accession no. PRJEB33854.
RESULTS
In order to examine potential clonality, the cgMLST genotypes of all 81 samples were determined. Figure 2A shows a minimum-spanning tree based on the allelic profiles of 1,861 cgMLST targets. Isolates differing in ≤24 targets were defined as a cluster. Isolates that share the same cluster are likely to belong to the same clonal lineage and are therefore indicative for transmission events. In Fig. 2B, the temporal distribution of farm visits and genotypic clusters is illustrated. In total, there are 10 clusters and two singletons, with 2 to 30 isolates per cluster. Allelic differences within these clusters range from 0 to 23.
Cluster analysis of 81 LA-MRSA CC398 isolates. (A) Each circle of the minimum-spanning tree represents a single genotype, i.e., an allelic profile based on up to 1,861 target genes present in the isolates with the “pairwise ignoring missing values” option turned on in the SeqSphere+ software during comparison. The circles are named with the isolate IDs, colored by farm, and the sizes are proportional to the number of isolates with an identical genotype. The number on connecting lines displays the number of differing alleles between the connected genotypes. Genotypic clusters (≤24 differing alleles) are indicated by gray shading. The three spa type groups are separated with dashed lines. Isolates designated “E” are from the environment, those with “P” are from pigs, those with “F” are from farmers, and those with “W” are from fieldworkers (WA, WC, and WD). (B) Temporal distribution of samples and their clusters. Each symbol, colored by farm, depicts one isolate. Numbers within the symbols show the genotypic cluster corresponding to panel A. Each column represents one farm visit and is numbered accordingly.
Overall, most clusters comprise environmental and pig isolates from the same farm. This means that transmissions of the same clonal lineages only occur within but not between farms. One exception is environmental sample E19 from farm VI, which clusters with isolates from farm VII. Furthermore, one isolate from farm VIII (E7b) shares the cluster with farm II isolates. Cluster 1 with isolates from farm VI showed the smallest minimum distances (≤4), indicating a very close relationship between these isolates. Isolates from farm VI were present in three different clusters, and farm VII was represented in two genotypic clusters. Farms IV, V, and VIII had one cluster plus one isolate with a different genotype each. For farms I, II, and III only one cluster was present. Overall, we found only few genotypic clusters and very few singletons for each farm, suggesting that the genotypic diversity within a farm is limited to one or few clonal lineages.
Most worker isolates also cluster with isolates from the farms visited on the respective day, indicating that a transmission from the visited farm to the worker is likely. Isolates WA7, WC7, WC9 and WD6 from fieldworkers did not cluster with the farm isolates of the respective last visit. However, except for WD6 from farm VI, which had a minimal distance of 25 alleles to next closest isolate, these were isolated on days with two consecutive visits and the genotypes fit the ones present on the farm visited first on the respective day (Fig. 1). No worker isolate with a genotype belonging to a farm that was visited more than a day ago was found. This makes a persistent carriage of acquired MRSA unlikely.
The allocation of spa types is also indicated in Fig. 2A. As expected, isolates of the same spa type were more closely related to each other than to isolates of different spa types. All isolates from farms I, III, IV, and V were t011. t2011 was found on farm VIII and the same-day sample WC7 as well as on farm II. On farms VI and VII, both t011 and t034 were detected.
To further investigate the relatedness of the isolates within and between farms, a phylogenetic analysis, including ancestral dating was done with BEAST. The analysis was based on the concatenated sequences of the 1,681 cgMLST genes that were present in all isolates. The topology of the Bayesian tree confirmed the findings from the cluster analysis (Fig. 3). It shows longer branches and thus higher relatedness of isolates within a farm than between farms. The ratio of branch lengths, which are given in years, also fits the distances shown in the MST based on allelic profiles of cgMLST target genes. Moreover, the allocation of spa types over the tree corroborated the groups described above. For t034, the variation between isolates within farms was relatively small, while they differed a lot between farms. For the t011/t2011 group, the overall variability was higher. Using the sampling dates as calibration, the median root age is 22.3 years (95% highest posterior density [HPD], 13.49 to 36.73) from the most recent sample, which means the most recent common ancestor (MRCA) of all isolates originated in the late 1990s. The MRCA of the t034 clusters on farms VI and VII emerged about 19 years ago (95% HPD, 11.37 to 31.59), while the MRCA of the relatively closely related t2011 and t011 farm VI clusters appeared around 2008 (median, 10.67; 95% HPD, 6.94 to 16.72).
Maximum clade credibility tree calculated from 9,000 sampling trees resulting from a Bayesian analysis based on 1,681 concatenated cgMLST genes. Leaves are colored according to the farms where the isolates were detected. The timeline indicates sampling and divergence times in years. spa type groups are labeled behind braces. Detailed information, including posterior probability values, is provided in Fig. S1.
DISCUSSION
In this study, we wanted to find out whether LA-MRSA carriage of humans with occasional rather than daily livestock contact is really temporary as suggested by previous studies, or whether persistent colonization also occurs. Furthermore, we examined the clonality of LA-MRSA within and between different pig farms. Our analyses showed that genotypes of isolates taken within a farm are more similar to each other than between different farms. Only 1 to 3 clusters were found per farm indicating that there are only few farm-specific clones. This assumption is supported by relatively long branches separating the different farm-groups and high posterior probability values for these clades (see Fig. S1 in the supplemental material). One exception is isolate E19 from farm VI, which clusters with farm VII samples. A possible explanation could be that there are specific clones on each farm that are predominant and therefore also mainly represented in our data set, but other genotypes are also present in lower numbers and thus only rarely sampled.
With only one exception, all isolates from fieldworkers clustered with samples from farms visited on the same day. This leads to the conclusion that indeed fieldworkers gather farm-specific clones during visits, which are quickly lost again, rather than getting permanently colonized. In some cases, two farms were visited on the same day and the clone from the first farm was detected after visiting the second farm. However, if the next visit was scheduled to a day after, like on farms IV (WC9) and V (WC5) or VI (WA12) and VII (WA13), no LA-MRSA was found. This suggests that acquired LA-MRSA is lost within 24 h, which is in concordance with previous findings (5). From the 34 isolates from fieldworkers, only isolate WD6, which had a unique genotype, did not fit into this hypothesis. This genotype could neither be linked to the recent visit nor to any of the previous farms visited. Possible explanations could be an unintentional interchange of samples or a mixed culture. Sequencing errors would also be imaginable, though previous studies found these kinds of errors extremely rare (20). Maybe this fieldworker acquired this clone in a different setting; however, it was not detected in the initial screening. Since most of the fieldworkers have an agriculture-related background, this is a probable scenario. Due to the limited data set, it could also be that the acquired genotype was also present on farm VI but less dominant so that it was not captured in the sequenced farm samples but by chance acquired by the worker.
Our findings suggest that, in contrast to pig farmers, occasional pigsty visitors do not get permanently colonized with LA-MRSA. Occasional exposure to LA-MRSA CC398 seems sufficient for a transient but not a persistent colonization. Van Cleef and colleagues (5) suggest calling these transient conditions “contamination” rather than colonization as they may result from bacteria adhering to the dust instead of epithelial cells in the nose. This also fits the findings by Angen and colleagues (6) that the degree of nasal LA-MRSA carriage correlates with the load of LA-MRSA in the air.
Farmers, other than visitors, or in our case scientific fieldworkers, usually spend time on their farms daily. Therefore, they might be exposed to LA-MRSA again before it was lost, which could lead to a potentiating effect that facilitates persistence. Furthermore, the constant exposure to a certain clone might alter the composition of microbiota in the nose, which could be beneficial for the clone to prevail. Longitudinal studies on how farm environments change the microbiome (e.g., on farmer trainees) would therefore be interesting for the future.
Besides solving the question of LA-MRSA persistence in pigsty visitors, our study provided a few additional insights. According to our BEAST analysis, the root age is about 22 years. Since the root in our data set represents the MRCA of the two most prevalent CC398 spa types, t011 and t034, this is also a hint about the general origin of LA-MRSA. Following our data, LA-MRSA CC398 evolved in the late 1990s. The first detection of LA-MRSA in humans was in 2005 (21), roughly 10 years later. Other studies found an earlier origin of the LA-MRSA CC398 clade in the 1960s (22) or early 1970s (23), while another study suggests a more recent origin around 2005 (24). The different results could be explained by epidemiological differences in the analyzed data sets such as geographical sources of isolates. Therefore, our results could indicate an origin of Western German pig-associated LA-MRSA CC398 in the late 1990s.
We aimed to investigate whether humans with occasional exposure to LA-MRSA CC398 are only temporarily colonized, as suggested by other studies, or if persistence of clones beyond the detection limit also occurs. We conclude from our data that a transient carriage of farm-specific clones is most probable. However, the study has a few limitations. Since the isolates were initially taken for other purposes, the availability of samples from different farms and days was limited and the allocation of isolates over the different sampling sites is not ideal. Moreover, from each sample only one MRSA colony was analyzed. Therefore, it cannot be ruled out that mixed populations occurred on the farms as well as in the fieldworkers’ nares. However, the absence of multiple detections of the same genotype within the fieldworkers makes a persistent clone unlikely, despite the limited data availability. For future studies, we propose longitudinal investigations of the MRSA status in humans starting regular work on pig farms, including microbiome analyses. This would help to find out more factors that might be advantageous for a permanent colonization with LA-MRSA.
ACKNOWLEDGMENTS
We thank Isabell Höfig and Ursula Keckevoet for technical assistance.
This work was supported by grants of the German Federal Ministry of Education and Research (BMBF) as part of the Research Network Zoonotic Infectious Diseases (project 1Health-PREVENT, grant no. 01Kl1727A and 01KI1727B to A.M., R.K., and M.B., respectively) and Ministry for Environment, Agriculture, Conservation and Consumer Protection of the State of North Rhine-Westphalia, Germany. Support of the DFG Research Training Group 2220 “Evolutionary Processes in Adaptation and Disease” to N.E. is gratefully acknowledged. The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication. We declare that we have no actual or potential competing financial interests.
FOOTNOTES
- Received 5 August 2019.
- Returned for modification 4 September 2019.
- Accepted 20 October 2019.
- Accepted manuscript posted online 30 October 2019.
Supplemental material is available online only.
- Copyright © 2019 American Society for Microbiology.
