Bacteria of the genus Bordetella are of primary importance in human and veterinary medicine because of their ability to colonize the respiratory tract, causing a wide range of pulmonary and bronchial infections. The common human- and animal-adapted pathogens B. pertussis, B. parapertussis, and B. bronchiseptica are known as the “classical” Bordetella species. B. pertussis is a strictly human pathogen, but B. parapertussis consists of two lineages, one infecting humans and the other infecting sheep (Mattoo and Cherry, 2005). In contrast to these examples of adaptation to a single host, B. bronchiseptica colonizes a variety of animals and even humans (Register et al., 2015), resulting in a broad array of respiratory diseases, from asymptomatic colonization to lethal pneumonia (Goodnow, 1980). Phylogenetic analyses (Musser et al., 1986; Diavatopoulos et al., 2005) and genome comparisons (Parkhill et al., 2003) have revealed that B. pertussis and B. parapertussis represent human-adapted forms of B. bronchiseptica that have evolved independently from a B. bronchiseptica-like ancestor. The genus also contains a number of additional, more recently described species. For example, B. avium (Kersters et al., 1984) causes respiratory disease in birds. B. hinzii (Vandamme et al., 1995) is generally regarded as a non-pathogenic colonizer of the respiratory tract of poultry but some strains were shown to cause disease in turkeys when experimentally inoculated (Register and Kunkle, 2009). Meanwhile, the closely related species B. pseudohinzii colonizes laboratory mice (Ivanov et al., 2015, 2016). B. holmesii (Weyant et al., 1995) causes pertussis-like disease and septicemia in humans (Shepard et al., 2004), and B. bronchialis, B. flabilis, and B. sputigena (Vandamme et al., 2015) were also isolated from human respiratory specimens. In contrast to other bordetellae, B. trematum (Vandamme et al., 1996) and B. ansorpii (Ko et al., 2005) are not associated with respiratory problems but were isolated from human wound infection.

B. petrii, a species originally isolated from a dechlorinating bioreactor enriched by river sediment, represents the first described environmental species within the Bordetella genus (von Wintzingerode et al., 2001). B. petrii strains were also found in marine sponges (Wang et al., 2007), grass root consortia (Wang et al., 2007), and in other environmental samples as members of microbial communities involved in the degradation of aromatic hydrocarbons, such as benzenes (Bianchi et al., 2005; Wang et al., 2007). In apparent conflict with its environmental source, the B. petrii genome contains genes that allow for the synthesis and secretion of factors specifically associated with the virulence of the pathogenic Bordetella sp., including the BvgAS master regulon and filamentous hemagglutinin (Gross et al., 2008). In addition to these environmental sources, B. petrii was also isolated from immunocompromised patients with ear infection, cystic fibrosis and chronic pulmonary disease (Fry et al., 2005; Biederman et al., 2015; Nagata et al., 2015), suggesting broad adaptability of this bacterial species to both environmental conditions and as an opportunistic pathogen of humans and possibly other animals.

Other Bordetella species have been obtained from environments not associated with animal hosts. Ten different bacterial strains were cultured from cotton swabs taken from the plaster wall surface of 1300-year-old mural paintings inside the stone chamber of the Takamatsuzuka Tomb, an ancient circular burial mound in Japan. Taxonomic classification of these isolates revealed three novel species that were then named B. muralis, B. tumulicola, and B. tumbae (Tazato et al., 2015). Isolation of the bacteria from the paintings, but not from the surrounding stone walls, suggested that these species might be involved in the observed biodeterioration of the colorful paintings (Kigawa et al., 2013).

According to their 16S rRNA gene sequences, other environmental bacteria from soil also belong to the genus Bordetella. Interestingly, the majority of those samples originated from contaminated sites such as soil polluted with chlorinated benzenes (Wang et al., 2007), from arctic soils contaminated with polycyclic aromatic hydrocarbons such as oil, diesel fuel or tar (Eriksson et al., 2003), from the sediment of a municipal wastewater plant (Nisola et al., 2010) and from arsenic polluted soils (Cavalca et al., 2010; Bachate et al., 2012). All these observations suggest that members of the Bordetella genus may have the potential for biodegradation of a great variety of organic compounds.

Although these anecdotal findings suggest that members of the Bordetella genus may be found in nature, there is currently no systematic analysis of the occurrence of Bordetella outside human or animal hosts, and the potential impact of environmental isolates on human and animal health is uncertain. Environmental niches of pathogenic Bordetella species have been proposed but not identified. Yet, the ability of Bordetella to survive and persist outside mammalian hosts would allow for its greater dissemination and persistence, and could contribute to a wide distribution of infections and disease, often without an obvious infectious source.

Here, we search the NCBI nucleotide database for 16S ribosomal RNA gene sequences of Bordetella-like microorganisms from various environments and compare them to those of the described species, including the classical bordetellae, to determine their phylogenetic relatedness. We identified 10 clades of related strains, all of which contained samples isolated from environmental sources, though only four clades also contained sequences from animal-associated species. Sequences from environmental samples had a significantly higher genetic diversity and were located closer to the root of the phylogenetic tree than those from animal-associated isolates, suggesting an environmental origin of the genus Bordetella. In addition, we show that the animal-adapted pathogens B. bronchiseptica and B. hinzii grow efficiently in soil extract, indicating that diverse pathogenic bordetellae may have retained the ability to proliferate in the environment.

Bacteria of the genus Bordetella occupy remarkably diverse ecological niches, ranging from soil, water, and plants, to the respiratory tracts of a wide variety of animals including humans. Several environmental Bordetella strains were isolated from soils polluted with oil and oil derivatives (Table 2), including halogenated polycyclic hydrocarbons (Eriksson et al., 2003; Bianchi et al., 2005; Wang et al., 2007). Other strains were found in garden soil, compost, and various sediments suggesting these organisms are quite adaptable to diverse sites. The only sequenced and analyzed genome of an environmental isolate, B. petrii strain DSMZ 12804, revealed a possible genomic basis for substantial metabolic versatility (Gross et al., 2008). The genome encodes multiple auxiliary pathways for the utilization of a variety of nutrients, including pectate, numerous sugar derivatives from degraded plant products and various aromatic compounds. Five of the eight genomic islands that have been identified in this genome contain genes coding for enzymes for the metabolism of aromatic compounds, particularly clusters of genes encoding enzymes of the chlorocatechol pathway, including gene clusters that show high similarity to genes in a 1,2,4-trichlorobenzene-degrading Pseudomonas strain (Gross et al., 2008). The presence of multiple chlorocatechol gene clusters in addition to several different central pathways for aromatic metabolism may provide a competitive advantage for growth in contaminated environments.

Another striking feature of environmental Bordetella isolates is their resistance to heavy metals (Cavalca et al., 2010). Ten out of 52 soil samples (Table 2) were isolated from iron mines (e.g., B. sp. d16, B. sp. f17), from uranium mines (B. sp. FB-8, B. sp. A2–436), or from soil polluted with arsenic (e.g., B. sp. As3–3). Such remarkable metal tolerance is most likely conferred by heavy metal resistance systems. Indeed, the genome of B. petrii strain DSMZ 12804 contains several heavy metal resistance operons on a genomic island absent from the genomes of other sequenced bordetellae, whereas other strains contain different islands of genes. Ultimately, the presence of multiple heavy metal resistance systems may allow environmental Bordetella isolates to thrive in metal rich environments.

Most plant-associated Bordetella strains were recovered from roots (B. sp. R8–804, B. sp. R8–551, B. sp. Juv992) and the rhizosphere at the plant-soil interface (B. sp. CCBAU 10842). Thus, these isolates may in fact represent soil samples or, alternatively, may be involved in interactions with plants at the plant-soil interface. The resemblance between plant responses to bacterial virulence factors and the responses of mammalian immune cells (Berg et al., 2005) serve as evidence that bacteria-plant interactions may have paved the way for bacterial adaptation to animals. In this regard, plant-root isolates B. sp. R8–804 and B. sp. R8–551 from plant roots are closely related to bird pathogens, B. hinzii and B. avium, supporting the view that plants cells could serve as a “training ground” for environmental strains that eventually gain the ability to colonize animal hosts (Berg et al., 2005).

In addition to these plant root isolates, several other environmental isolates were also found to be very closely related to animal-associated pathogens (Figure 1). Interestingly, those strains were isolated from very diverse sources, namely (polluted) soil in India (B. sp. BAB-4396, B. sp. IITR02), industrial waste water (B. sp. MT-E1), and oil-based metal-working emulsion in Germany (B. sp. 13.1 KSS), as well as from river biofilms in China (B. sp. HF38 and B. sp. HF72). The two isolates from a river biofilm in China are of particular interest. The 16S rRNA sequence of one of those (strain HF72) showed 99.56% sequence similarity to that of the human pathogen B. trematum (6 SNPs). According to 16S rRNA gene sequence, the other isolate (B. sp. HF38) is even more closely related (99.78%, three SNPs) to the animal pathogen B. bronchiseptica strain RB50 and the human pathogen B. parapertussis strain 12822, which share an identical sequence in this gene. By this measure, isolate HF38 is as closely related to B. bronchiseptica strain RB50 and B. parapertussis strain 12822 as it is to B. pertussis. This exceptionally close phylogenetic relatedness makes several evolutionary scenarios conceivable. First, isolate B. sp. HF38 may be an environmental, non-pathogenic strain closely related to the animal/human pathogens among the classical bordetellae. Second, this isolate might be a descendant or relative of an ancestor of the classical bordetellae which later became pathogenic after acquisition of several virulence-associated factors, such as pertussis toxin, adenylate cyclase toxin, and dermonecrotic toxin. Third, this isolate may in fact represent a B. bronchiseptica or B. parapertussis strain that naturally survives and/or grows within an environmental reservoir. Although the classical bordetellae have not yet been isolated from outside a mammalian host, our results suggest that animal-pathogenic Bordetella species retain the ability to grow in soil as an environmental niche. This implies that B. bronchiseptica and other species might be found (at least transiently) in soil, for example at farms with suitable animal hosts such as cattle, pig, sheep and horse, or near dog kennels. Interestingly, even fastidious B. pertussis bacteria remained able to be cultured for up to 5 days when spread onto various hospital-setting surfaces such as fabrics, plastics, glass, and paper, and also in several infant foods (Ocklitz and Milleck, 1967). Fourth, B. sp. HF38 as well as other isolates from water and soil may be protected internally by a non-vertebrate host. For example, amoebae are known to host bacteria such as Legionella pneumophila (Molmeret et al., 2005), and amoeba-grown L. pneumophila exhibited radically increased resistance to harsh environmental conditions such as fluctuations in temperature, osmolarity, acidity, as well as to biocides that may facilitate bacterial survival and persistence in the environment (Barker et al., 1995; Abu Kwaik et al., 1997, 1998; Winiecka-Krusnell and Linder, 1999). Amoebae are ubiquitously found in most environments, and shared habitats between amoeba and Bordetella could be an important factor for the persistence of the bacteria. Indeed, our group has shown that the animal-adapted B. bronchiseptica is able to survive and multiply intracellularly in the trophozoites and sori of the amoeba Dictyostelium discoideum before being disseminated with the amoeba spores to novel geographical locations (Bendor et al., in revision). Thus, in addition to our recent data demonstrating that B. bronchiseptica can circulate and efficiently transmit amongst mammals, these data demonstrate that this species can also grow and disseminate efficiently in association with amoebae. These independent but interconnected Bordetella lifecycles allow for disease propagation, transmission, and re-emergence in the absence of an infected animal host.

Strains included in this study were identified as Bordetella spp. based on their 16S rRNA gene sequence. Currently, there are no data available regarding potential pathogenicity of these species. Whole genome sequencing will provide valuable insights into the evolution and ecology of environmental vs. animal-pathogenic bordetellae. Of special interest are environmental isolates closely related to animal pathogens, particularly isolate B. sp. HF38, and analysis of their genomes will reveal whether they are non-pathogenic relatives of known animal pathogens or if they in fact represent environmental reservoirs of B. bronchiseptica or B. parapertussis.

Finally, the majority of environmental B. sp. were recovered from soil samples indicating that soil could be the most frequent natural habitat of bordetellae. Indeed, sequences identified from soil samples were found in 8 of 10 sequence clusters, including samples from compost in cluster X at the root of the tree (Figure 1). The sequence of branching events within the phylogenetic tree, the significantly higher sequence diversity in samples from soil and water than in those from animals, as well as the preserved ability of animal pathogens to grow in soil, suggest an environmental, likely soil-based, origin of the genus Bordetella. Thus, similar to bacteria of the closely related genus Achromobacter, which are of environmental origin but also contain opportunistic pathogens (Li et al., 2013), Bordetella appears to be a bacterium of environmental origin that adapted and became pathogenic via the acquisition of factors mediating specific interactions with animal hosts.

IHS, BL, and ETH conceived and designed the experiments. IHS and BL performed the experiments and analyzed the data. IHS, BL and ETH wrote the paper.