Phage-Mediated Immune Evasion and Transmission of Livestock-Associated Methicillin-Resistant Staphylococcus aureus in Humans

These bacteria are capable of adapting to humans, leading to increased spread into the community and healthcare settings.

L ivestock-associated methicillin-resistant Staphylococcus aureus (LA-MRSA) clonal complex (CC) 398 is a major cause of zoonotic disease in Denmark and several other countries in Europe that have industrial pig production (1)(2)(3). In Denmark, the prevalence of LA-MRSA CC398 in pig farms increased from 3.5% in 2008 to ≈90% in 2018, when LA-MRSA CC398 accounted for 21% of all human MRSA infections (3). Most LA-MRSA CC398 infections occur in young and otherwise healthy livestock workers and their household contacts (3)(4)(5). Although LA-MRSA CC398 seems to be poorly adapted for human-to-human transmission (6), it is nonetheless able to spread to and cause serious illness and even death in elderly and immunocompromised persons in community and healthcare settings (3)(4)(5).
S. aureus uses a diverse range of immune-evasive strategies to maintain a lifelong relationship with the human host, many of which are encoded on phages and other mobile genetic elements (7). Of note, human S. aureus clones harbor a genetic element, the immune evasion cluster (IEC), on a 44-kb ΦSa3int prophage that is stably integrated into the hlb gene on the bacterial chromosome (8,9). The IEC element encodes >1 immune modulators, including staphylococcal complement inhibitor, chemotaxis inhibitory protein of staphylococci, staphylokinase, staphylococcal enterotoxin A, and staphylococcal enterotoxin P, which interact specifically with components of the human innate immune system (7). LA-MRSA CC398 isolates are descendants of a human variant of S. aureus CC398 but have lost the ΦSa3int prophage and the associated IEC element in connection with the host switch event (10); this change provides a potential explanation for the observed relatively low human-to-human transmissibility of LA-MRSA CC398 (6). However, some studies have shown that LA-MRSA CC398 might be capable of readapting to the human host through acquisition of phage-encoded immune modulators. For example, a Denmark study showed that 6% of human LA-MRSA CC398 isolates collected during 2004-2011 harbored the IEC element, whereas a more recent study revealed that 40% of LA-MRSA CC398 isolates from pigs in Denmark produce another phage-encoded immune modulator enzyme known as TarP, which enables S. aureus to subvert antibody-mediated immune recognition by altering a dominant cell surface epitope known as wall teichoic acids (WTA) (3,11).
These earlier findings raise important questions about the source and dynamics of phages encoding IEC and TarP in LA-MRSA CC398 and their role in host adaptation. In this study, we sequenced and compared a collection of epidemiologically wellcharacterized LA-MRSA CC398 isolates from humans and pigs in North Denmark Region to determine their population structure and the contribution of IEC and TarP to household transmission. We also used national surveillance data to further investigate whether IEC plays a role during spread of LA-MRSA CC398 into the community and healthcare settings.
Data used in this study were collected as part of the national MRSA surveillance program, as approved by the Danish Data Protection Agency (protocol no. 2001-14-0021). The National Committee on Health Research Ethics waived the need for approval and informed consent because data and biologic material were fully anonymized and collected in compliance with national legislation on statutory notification of MRSA in humans.

Study Population
North Denmark Region (≈7,900 km 2 ) is a semirural area in northwest Denmark that had a population of ≈580,000 persons and ≈3 million pigs in 2011 (http://www.statbank.dk). We identified the study population using surveillance data collected by the regional infection control staff in North Denmark Region; it included all patients colonized or infected with LA-MRSA CC398 during 2004-2011, as well as all their household contacts who tested negative during the same period. We interviewed all of these persons to obtain relevant information, including sex, age, livestock contact, residential address, and workplace, and assigned each to 1 of 3 categories: livestock-exposed persons (direct contact); household contacts of livestock-exposed persons (indirect contact); and persons not connected to livestock production (no contact). We defined livestockexposed persons as primary cases in their households, and the index person as the primary case in households with no connection to livestock production.

Study Isolates and Data Aggregation
The study isolates comprised 96 human LA-MRSA CC398 isolates from North Denmark Region collected by regional infection location of disease onset (e.g., HO, HACO, or CO); and presence or absence of the IEC element in the corresponding LA-MRSA CC398 isolate. We calculated the excess number of clinical cases due to increased spread of IEC-positive isolates into a given patient group of interest as the total number of cases in the patient group of interest multiplied by the difference between the proportion of IEC-positive isolates in the patient group of interest (the sink) and the patient group with direct contact to livestock (the source).

Statistical Analysis
We used Fisher exact test to analyze categorical data and Student t test to analyze continuous data (Graph-Pad Prism version 5; GraphPad, https://www. graphpad.com). We reported prevalence differences between different groups as prevalence ratios (PRs) and 95% CIs. The significance level was set at α = 0.05.

Study Population
A total of 96 patients were colonized or infected with LA-MRSA CC398 in North Denmark Region during 2004-2011, including 67 primary cases and 29 secondary cases from 65 households. A total of 71 household contacts tested negative. The 67 primary cases comprised 57 persons with direct animal contact, 2 with indirect animal contact, and 8 with no animal contact. Those with direct animal contact included 44 pig farm employees from 42 households and 23 animal farms (2 households each contained 2 pig farm employees), 3 mink farm employees from 3 households and animal farms, 1 cattle farm employee, 1 turkey farm employee, 1 pig veterinarian, 3 lorry drivers transporting pigs from 3 households, 2 pig abattoir workers from 2 households, and 2 craftsmen working in pig stables from 2 households. Persons with indirect contact were from 2 households (a wife and a child of pig farm employees who were never tested), and those with no contact were from 8 households.

Distribution of IEC and tarP in LA-MRSA CC398
Most of the human isolates collected from persons living in the same household clustered together, with an average pairwise single-nucleotide polymorphism (SNP) distance of 5.9 (range 0-18 SNPs) and were genotypically homogeneous with respect to presence of specific IEC-harboring and tarP-harboring prophages (Appendix 2 Figure). Furthermore, human isolates from different households connected to the same animal farm also tended to cluster together but were genotypically more diverse than isolates from the same household (19. We identified IEC in 20 isolates, which clustered within a closely related clade (Appendix 2 Figure). The IEC-harboring ΦSa3int prophages could be divided into 6 variants (I-VI) and 5 phylogenetic clusters (A-E) on the basis of their phylogenetic relationship, IEC type, and chromosomal integration site (Table 1; Figure 1). Each of the 6 ΦSa3int variants was unique to isolates from a single household. The 6 households comprised 4 pig farm employees (ΦSa3int-I-IV), a pig veterinarian (ΦSa3int-V), and a mink farm employee (ΦSa3int-VI), and their household contacts ( Table  1). We also detected ΦSa3int-VI in 3 pig isolates that were closely related to the isolates from the mink farm employee's household; the prophage was furthermore highly similar to ΦSa3int-V present in the isolate from the pig veterinarian but integrated into a different part of the chromosome (Table 1).
We identified the tarP gene in 45 isolates, which were more widely distributed across the phylogeny than the IEC-positive isolates (Appendix 2 Figure). Analysis of the genetically linked int genes showed that tarP was carried on 5 different prophages (ΦSa1int, ΦSa3int, ΦSa7int, ΦSa9int, and a prophage with an untypeable int gene hereafter referred to as ΦUT1). BLASTN searches (https://blast.ncbi.nlm. nih.gov/Blast.cgi) showed that the int gene in ΦUT1 was 100% identical to the int gene in the bovine S. aureus phage ΦDW2 (GenBank accession no. KJ140076). We found ΦtarP-Sa9int exclusively in pig isolates (n = 15), whereas the other tarP-harboring prophages were identified in 25 human isolates from 16 households and in 5 pig isolates.

Role of IEC and tarP during Household Transmission of LA-MRSA CC398 in North Denmark Region
Household transmission of LA-MRSA CC398 was based on detection of secondary cases of colonization or infection with isolates that were closely related and genotypically indistinguishable from the isolate of the primary case. Households consisting of 1 person (n = 26) and households with >2 persons reporting direct animal contact (n = 2) were excluded from both analyses. In addition, 1 household with a mixed population of IEC-positive and IEC-negative isolates was excluded from the corresponding analysis.
If IEC-positive isolates become fixed (i.e., self-sustainable) in the general population, we expect their proportion to increase over time in patients with CO and HO/HACO infections, compared with patients with direct contact to livestock. However  panel A), the prevalence ratio of IEC-positive isolates among patients with CO and HO/HACO infections either decreased or remained relatively stable over the years (Figure 2, panel B).

Discussion
Successful spread of S. aureus in humans depends upon the bacterium's ability to survive and multiply in newly colonized persons. For example, S. aureus must compete with other bacteria and avoid the innate and adaptive immune defenses of the skin and nasal environments. In this study, we have shown that acquisition of IEC, but not tarP, is associated with increased household transmission of LA-MRSA CC398 and excess spread into the community and healthcare settings. The findings should be interpreted with some caution because of the small number of households and cases analyzed, and more follow-up studies should be done to further evaluate the relative contribution of IEC and other risk factors, such as household size (13), to the spread of LA-MRSA CC398 in humans. Animal and human studies have shown that IEC is ubiquitous in human S. aureus clones, whereas it is consistently absent in livestock-associated S. aureus clones (9,10). Staphylococcal complement inhibitor and other IEC-encoded immune modulators have human-specific activities toward central components of the innate immune response, such as neutrophils and complements, and are produced in both healthy carriers and patients with S. aureus infection (14)(15)(16). Our findings suggest that IEC promotes survival of S. aureus in the human host, which in turn is expected to increase the likelihood of human-to-human transmission.
Nonetheless, the disease burden of IEC-positive LA-MRSA CC398 isolates in the community and healthcare settings remains relatively low in Denmark. We saw no suggestion that they will become fixed in the general population; van Alen et al. reached the same conclusion after finding a low, albeit slightly increasing, prevalence of IEC among LA-MRSA CC398 isolates from hospital patients during 2000-2015, ranging from 1.1% during 2000-2006 to 3.9% during 2007-2015 (17). Research indicates several possible reasons why IEC-positive LA-MRSA CC398 isolates have not become self-sustainable in humans. First, our findings showed that the 20 IEC-positive isolates from North Denmark Region clustered within a closely related clade, indicating that not all LA-MRSA CC398 isolates have the same ability to acquire ΦSa3int phages. Second, a recent study (18) has shown that LA-MRSA CC398 carries substitutions at the usual attachment site (attB) located within the hlb gene, which interferes with phage integration and might explain why ΦSa3int phages often integrate into other genomic regions than  the hlb gene in LA-MRSA CC398 through recombination between the phage attachment site (attP) and alternative attB sites, as demonstrated in our study and others (17)(18)(19). Third, it is possible that LA-MRSA CC398 has undergone other genetic changes, in addition to losing the IEC-harboring ΦSa3int prophage during the human-to-animal host switch event, that are beneficial in the livestock reservoir but detrimental in the human host. For example, LA-MRSA CC398 isolates from Denmark have acquired several mobile genetic elements encoding resistance to a wide range of the most frequently used antimicrobial drugs in pigs, including β-lactams, aminoglycosides, macrolides, tetracyclines, and zinc (12), which are likely to exert a fitness cost outside the livestock reservoir (20). This possibility is supported by our recent finding that LA-MRSA CC398 isolates from hospital patients carry far fewer antimicrobial resistance genes of veterinary importance than LA-MRSA CC398 isolates from pigs (21). The widespread distribution of IEC among human S. aureus clones, the phylogenetic clustering of IEC-harboring ΦSa3int prophages in LA-MRSA CC398 isolates from the same household, and the low prevalence of IEC in LA-MRSA CC398 isolates from pigs suggest that acquisition of IEC by LA-MRSA CC398 mainly occurs through transfer of ΦSa3int phages from human S. aureus donors circulating in households. In support of this view, we found little evidence for phage transfer or transmission of IECpositive isolates between pigs and humans. An exception was a possible transmission chain involving 3 IEC-positive isolates from pigs and 4 isolates from a mink farm employee and his household members and a possible phage transfer event between this cluster of isolates and an isolate from a pig veterinarian. Of note, the IEC-harboring ΦSa3int prophage found in the 3 pig isolates also encoded TarP, thus raising the possibility that IEC can be passively maintained in the pig population by co-selection for other traits.
TarP-mediated protection against anti-WTA antibodies did not seem to influence household transmission of LA-MRSA CC398 in our study despite the fact that anti-staphylococcal antibodies are present at high levels in serum and nasal secretions of both persistent S. aureus carriers and noncarriers (15,16). Instead, there is evidence that the much more widely distributed staphylococcal protein A (SpA), which is produced by all known human and livestock-associated S. aureus clones, is sufficient for escaping the adaptive immune response. SpA contains several immunoglobulin-binding domains capable of binding both the Fcγ of IgG antibodies and the Fab of V H 3-idiotype antibodies, thereby limiting opsonophagocytosis and broad-spectrum antibody responses to other secreted and surface-bound antigens during S. aureus colonization and infection (22)(23)(24)(25)(26)(27)(28)(29)(30).
In summary, our study suggests that acquisition of IEC, but not tarP, is associated with increased household transmission of LA-MRSA CC398 and spillover into the community and healthcare settings, which might also explain why IEC is widespread among human S. aureus clones. Despite these findings, the attributable disease burden remains relatively low in Denmark, and we found no evidence to suggest that they have become self-sustainable in the general population. However, the dynamic nature of S. aureus genome evolution and host adaptability, as documented here and elsewhere, underscores the need for continued surveillance at the human-animal interface to detect evolutionary as well as epidemiologic changes that affect public health.