1. IntroductionC. burnetii, a Gram-negative obligate intracellular bacterium, is the causative agent of Q fever, a widely distributed zoonosis caused by inhalation of as few as 1 to 10 C. burnetii particles [1]. The most common sources for transmission to humans are birth products of domestic small ruminants, as huge bacterial numbers (109 bacteria/g tissue) are excreted during parturition [2]. Seroprevalence of C. burnetii in herds of goats and sheep is reportedly very high, i.e., 50 to 60% [3,4,5,6].Even though cattle are not a recognized source of human Q fever in Europe, C. burnetii seroprevalence in cattle has been found to be up to 79% at herd level in Western Europe [5,7], with values increasing with the age of animals [8]. C. burnetii infections were reported to be associated with placentitis in cases of bovine abortion and detection of Coxiella-like organisms within trophoblasts [9]. In infertile dairy cattle with chronic endometritis, C. burnetii was detected intralesionally and intracytoplasmically in macrophages in the endometrium [10]. C. burnetii shedding scarcely and sporadically occurs in bovine feces, whereas 50% of cows were found to shed the agents by vaginal mucus intermittently or sporadically [11]. Asymptomatic, chronically infected cattle primarily shed C. burnetii in milk [12,13], and in higher numbers as compared to small ruminants [14]. Almost 40% of cows were detected as milk shedders with persistent and sporadic shedding patterns in one study [11]. Significantly higher estimated titers of C. burnetii were observed in cows with persistent shedding patterns, suggesting the existence of heavy-shedder cows [11]. C. burnetii shedding in milk was associated with chronic subclinical mastitis as measured by milk somatic cell counts [15]. These findings imply that C. burnetii exerts a tropism for the mammary glands in dairy cattle. The target cell in the mammary tissue has not been resolved in situ, but in vitro experiments have revealed that C. burnetii replicates in bovine mammary gland epithelial cells and even more efficiently than in epithelial cells from the placenta, lungs or intestine [16].Studies from the US, Spain and Germany have shown that >94%, 72% and 63%, respectively, of tested bulk milk tank samples were PCR-positive for C. burnetii [17,18,19]. When testing shop and farm retail dairy products from Latvia, 26.67% of unpasteurized Latvian cow milk samples were PCR-positive, whereas 76.47% of pasteurized equivalents and 63.13% of fermented milk products were PCR-positive [20]. High C. burnetii numbers in bovine milk are of particular concern from a One Health aspect. Shedding in milk is considered to be associated with rapid dispersal of bacteria within a herd, e.g., by transmission from dams to calves via contaminated milk and colostrum [14,21,22]. Proposed by several authors as sources of human infection [23,24,25], the contribution of milk ingestion, mainly drinking unpasteurized milk, to Q fever infection in humans is difficult to establish [26]. Modern pasteurization protocols for high-temperature, short-time pasteurization of milk are effective in inactivating C. burnetii [27], reassuring the high safety level of off-the-shelf dairy products. However, in light of the marketing of dairy products produced from raw milk and increasing consumer preference for purchasing unpasteurized milk on the farm, the basis for the actual discrepancy between high C. burnetii prevalence in bovine bulk tank milk and the apparent low risk for humans must urgently be resolved.Underlying reasons might be either genotypic or phenotypic peculiarities of the C. burnetii strains infecting cattle and colonizing the mammary gland, or host factors shaping the local environment for C. burnetii replication and release from host cells, or combinations thereof. Despite the high C. burnetii prevalence, genetic diversity of C. burnetii within a herd is rather low [28], and a single genotype of C. burnetii (ST20) is commonly found in US bovine milk samples [29,30], which suggests the existence of bovine udder-specific, low-virulence strains. However, ST20 is closely related to strain MST20, which was implicated in a recent outbreak in the Netherlands in 2011, where more than 4000 human cases were reported and approximately 52,000 ruminants were culled to control the outbreak [31]. Three different genotypes (MST20, MST33 and G) were isolated from humans and ruminants during that outbreak [32]. Strains isolated from humans can also be found in ruminants, even though the disease does not seem to be the same [32], whereas the US ST20 strain has a reduced ability to cause diseases in human or animals [29]. in vitro analyses with bovine and human macrophages indicated that C. burnetii replication is primarily determined by genotype; strains of MLVA genotype IV, such as Nine Mile Phase I (NMI) or the human strain Henzerling, replicate with a very high efficiency in cells of both species [33]. The public health risk in Belgium was linked to specific genomic groups, which were mostly found in small ruminant strains [34]. In France, MLVA clusters were found to be significantly associated with ruminant species, with all the cattle genotypes belonging to a “cattle-specific” cluster, whereas small ruminant genotypes were essentially grouped into two other clusters [35]. The results obtained by a functional assay deploying human peripheral blood mononuclear cells in turn suggested that cytokine responses are dependent on host origin rather than MLVA genotype [36].

Aiming at identifying markers for host adaptation and virulence of C. burnetii strains, we used the bovine mammary gland epithelial cell line PS as an in vitro infection model for the characterization of bacterial replication and host cell response after infection with 15 different C. burnetii isolates, which represented MLVA groups I to IV and were isolated from different hosts and organs, including three bovine milk isolates.

4. DiscussionA recent bibliometric analysis [43] revealed that an article on a comparison of C. burnetii shedding in the milk of dairy bovine, caprine and ovine herds [14] was the most highly cited in the current literature on C. burnetii. The study demonstrated that the bacterium is mainly excreted through the milk of infected cattle and goats, while in sheep it is mainly excreted through feces and vaginal excretions [14]. Indeed, the findings may explain the higher association of human outbreaks with sheep as compared to cattle and goats. However, the events leading to or preventing transmission of C. burnetii from cattle to humans have not been adequately investigated. Preceding in vitro studies with a bovine mammary gland epithelial cell line had demonstrated that prototypic C. burnetii strains NMI and II replicated in these cells [16]. We therefore studied replication and host cell response to a selection of C. burnetii strains of different genotypes and from different sources, including three bovine milk isolates, in the PS cell model.The different C. burnetii strains all replicated in bovine mammary gland epithelial cells with high efficiency. The possibility of genuine replication of C. burnetii in bovine mammary gland tissue is consistent with reports that the bulk tank milk of dairy herds in the US and Europe contains substantial numbers of C. burnetii, e.g., more than 102 bacteria/mL [18,19,44,45]. The differences in replication efficiencies that we observed in the present study were minor, even though strains of the MLVA genotype II, on average, replicated significantly better than genotype I strains. This deviates from findings with bovine monocyte-derived macrophages (MDMs), in which replication rates varied mainly between MLVA genotypes III and IV [33]. While human MDMs supported the replication of different C. burnetii strains more equally, bovine MDMs particularly supported the propagation of strains NMI, Henzerling and Scurry. The strain Dugway, as well as isolates from ruminants, rather poorly replicated in the cells. This pattern is reminiscent of the pattern observed in this study for bovine mammary gland epithelial cells, as the growth curves of the latter strains were characterized by a prominent delay in replication, which started only after 3 d p.i. However, this pattern was shown by all more recent C. burnetii isolates, and the apparent genotype restriction of C. burnetii bovine milk isolates does not appear, therefore, to result from the selective propagation of certain strains by tissue-specific cells. This study also revealed that, on average, bovine and ovine isolates replicated better in PS cells than caprine isolates, which correlates with the primary C. burnetii shedding route of the animals [14]. We had hypothesized earlier that bovine macrophages serve in bacterial transport from the entry site in the body, which might be the airways or the intestinal mucosa, to target cells for bacterial replication at exit sites, such as epithelial cells, in different organs [16]. Bovine MDMs are less supportive of C. burnetii replication than PS cells by approximately one order of magnitude [16,33]. Nevertheless, it is tempting to speculate that selection for bacterial host adaptation and virulence starts early in the infection process, i.e., inside the initial C. burnetii target cells, such as lung macrophages and dendritic cells, which was not mimicked by the model applied herein. Co-culturing models or inoculation of mammary gland epithelial cell cultures with infected macrophages may be needed to properly represent all aspects of this process.Another set of determinants to be considered that are instrumental for limiting C. burnetii replication and subsequent shedding in milk are the local inflammatory responses after some rounds of replication in tissue-specific cells. Indeed, C. burnetii shedding was linked to subclinical mastitis, with elevated levels of somatic cell counts indicative of some inflammation upon infection of the organ [15]. Epithelial cells constitute the first line of defense against microbial pathogens, act as a barrier for bacterial recognition and use their immunoregulatory function to alert the immune system to reduce immersive pathogens [46]. Interactions with pathogens usually induce host responses with an up-regulation of different inflammatory factors, e.g., cytokines, chemokines or cell-associated surface markers. Bovine MDMs respond to C. burnetii infection with an early (3 h p.i.) increase in the transcription of pro-inflammatory cytokine genes, such as those encoding for IL-1β, IL-12 and TNF-α [40]. In contrast, bovine mammary gland epithelial cells showed no immune response after infection with the reference strains NMI and NMII during the first 24 h p.i. [16]. Similar results were obtained herein after infection with different C. burnetii strains. Attachment of C. burnetii to surface PAMP receptors, such as TLR2 [47], seemingly fails to induce pro-inflammatory factors in udder cells. Moreover, some strains suppressed inflammatory gene expression at 7 d p.i., i.e., when cells were analyzed after the onset of C. burnetii replication. Re-stimulation studies with E. coli LPS and C. burnetii-infected PS cells revealed that the lack of inflammatory cellular response results from a refractory state rather than from lack of recognition of pathogen-associated molecule patterns (PAMPs). Although the early induction of IL-1β in bovine MDMs by the strain NMI is independent of a functional bacterial metabolism [40], the refractory state of infected PS cells was found to be an actively controlled process of C. burnetii, since cells exposed to heat-inactivated strain NMI suspensions reacted like uninfected controls. Previous reports show that Coxiella inhibit the activation of apoptosis in host cells to promote cell viability [48]. Three anti-apoptotic effector proteins (AnkG, CaeA and CaeB) have been identified [49,50,51] and are transported to the cytosol via a type IV secretion system [52]. Specific virulence factors may also be deployed by C. burnetii to manipulate the host response to create a safe replicative niche in bovine udder cells. Cross-signaling to inflammatory response genes as an explanation of the anti-inflammatory effects of C. burnetii in bovine cells deserves further investigation.Of note, a single isolate, Cb30/14, obtained from sheep, induced a significant increase in cytokine gene transcription, e.g., IL-1β and TNF-α. Increased concentrations of pro-inflammatory cytokines were found in milk from cows with coliform mastitis [53]. Mastitis in dairy herds of cattle [15] is associated with shedding of C. burnetii in milk. Secretion of pro-inflammatory cytokines stimulates the migration of somatic cells and neutrophils to the udders [53]. In bovine MDMs, infection with isolate Cb30/14 resulted in a mitigated general cellular response but a selective increase in IL-1β and IFN-γ expression [33]. IFN-γ promotes an up-regulation of major histocompatibility complex (MHC) expression in epithelial cells [54]. In Chlamydia-infected epithelial cell lines, increased MHCI expression promotes the degradation of the pathogen by activation of CD8+ cytotoxic T-cells [55]. Earlier findings by our group, however, called into question whether the increased amounts of mRNA after C. burnetii infection of bovine cells are translated into proteins. During the early phase of infection of bovine MDMs, C. burnetii induces a marked increase in IL-1β mRNA but fails to increase mature bioactive IL-1β [40]. Different from the situation in bovine MDMs, infected epithelial cells respond to Cb30/14 infection rather late, i.e., days after the invasion process, and therefore presumably do not engage cell surface receptors. Activation in C. burnetii-infected mammary gland epithelial cells probably occurs instead via receptors located inside parasitophorous vacuoles, such as Toll-like receptors (TLRs) 3, 7, 8 and 9. Endosomal TLR3 plays a critical role in host immune response in Chlamydia-infected epithelial cells [56]. Although all other C. burnetii strains acted differently but were similar to each other, the exceptional findings for strain Cb30/14 imply that some differences in host cell–pathogen interactions do occur at the main exit site for C. burnetii in cattle.The genotypes of the selected C. burnetii isolates have been defined only at the level of MLVA genotype using seven previously described microsatellites [38,39] for the purpose of this study. A harmonized reference method for the molecular characterization of C. burnetii has yet to be designed. The most commonly adopted methods to define phylogeny are multi-loci variable-number tandem repeat analysis (MLVA) and multispacer sequence typing (MST), with MLVA being more discriminatory than MST [34]. When attempting to dissolve the phylogeography of human and animal strains by genetic fingerprinting in Belgium, the high discriminative power conferred by the MLVA method performed with 13 markers allowed the definition of three MLVA clusters divided into 23 subclusters, which was crucial for in-depth genetic analysis [34]. Deploying more markers than were used in our study may also help to better link genotypes to phenotypes, even though more isolates would have to be included to represent each subcluster in sufficient numbers to confirm correlations.The virulence of C. burnetii differs between strains, apparently because of isolate-specific genes, pseudogenes [57] or plasmid types [58]. In a rodent model of acute Q fever infection, the virulence of C. burnetii was associated with bacterial genotype. Strains with the same genotype cause the same pathology in guinea pigs and the same cytokine secretion in response to C. burnetii infection in a mouse model [59]. To date, more than 130 putative effector proteins have been identified [60], some of which have anti-apoptotic activity [50,51,52], whereas the majority still await functional characterization [49]. In a selection of different C. burnetii strains, pseudogenes were found that could be attributed to pathotype-specific virulence of Coxiella [57]. Moreover, C. burnetii isolates from raw bovine milk, uterus swab samples from dairy cattle with reproductive disorders, aborted bovine fetus samples and mammary gland samples from healthy dairy cattle had various degrees of pathogenicity for guinea pigs, even though the protein and lipopolysaccharide (LPS) profiles of these strains were similar to those of the reference strain of phase I [61]. A bovine C. burnetii isolate was found to multiply faster than goat isolates in a bovine macrophage cell line, pointing to a preferential specificity of this strain for homologous host cells [62]. Identification of particular genetic determinants encoding for host adaptation and virulence was beyond the scope of this study. The findings obtained herein add important aspects to our understanding of the interactions of C. burnetii with bovine hosts, in that it is evident that beyond host of isolation and (MLVA) genotype, the type of the incremented host cell is another variable to be considered in assessing C. burnetii isolates as to their level of host adaptation and virulence. Moreover, preliminary data from our laboratory indicate that cells from cattle and small ruminants interact differentially with C. burnetii organisms (data not shown).There are indications that C. burnetii isolates possess some degree of adaptation to the host species they mostly circulate in and only occasionally become transferred to an aberrant host, e.g., humans [34,36]. We therefore hypothesized that bovine C. burnetii isolates from milk are more adapted to their host organs than other bovine isolates. Our results show that milk isolates replicated with nearly the same efficiency in bovine udder epithelial cells than bovine isolates with other origins. Consequently, the distribution of C. burnetii inside the host is isolate-independent, and organotropism cannot be confirmed. Some genetic diversity of C. burnetii isolates was also described in different Spanish bulk tank samples [28]. Separated isolates are closely related to isolates from human outbreaks and persist inside a herd over a long time [28]. Thus, the milk of infected animals represents a risk factor for the spread of C. burnetii infection within a flock, apparently independently of C. burnetii isolate genotype.