Molecular and Genetic Mechanisms That Mediate Transmission of Yersinia pestis by Fleas

The ability to cause plague in mammals represents only half of the life history of Yersinia pestis. It is also able to colonize and produce a transmissible infection in the digestive tract of the flea, its insect host. Parallel to studies of the molecular mechanisms by which Y. pestis is able to overcome the immune response of its mammalian hosts, disseminate, and produce septicemia, studies of Y. pestis–flea interactions have led to the identification and characterization of important factors that lead to transmission by flea bite. Y. pestis adapts to the unique conditions in the flea gut by altering its metabolic physiology in ways that promote biofilm development, a common strategy by which bacteria cope with a nutrient-limited environment. Biofilm localization to the flea foregut disrupts normal fluid dynamics of blood feeding, resulting in regurgitative transmission. Many of the important genes, regulatory pathways, and molecules required for this process have been identified and are reviewed here.


Introduction
Yersinia pestis is a newly minted arthropod-borne pathogen, having adopted this lifestyle only within the last 3000 to 6000 years [1][2][3]. Due to its notorious virulence to humans and public health significance, most research on this pathogen justifiably has focused on the genetic and molecular mechanisms responsible for its pathogenicity in mammalian hosts. Its insect host, the flea, was proven to be the essential vector for Y. pestis transmission over 100 years ago; however, it has received sporadic and comparatively little attention over the years. Yet half of the life cycle and biology of Y. pestis can be said to occur in the insect host. From this standpoint, the term vector seems somewhat dismissive, as it may connote a simple conveyance from one mammalian host to the next. However, the ability to produce a transmissible infection in the flea is as critical to Y. pestis as its ability to produce a transmissible infection in the mammal. The two in fact are interrelated: reliance on the blood-feeding flea for transmission naturally selected for Y. pestis strains that produce a high-density bacteremia and, therefore, high virulence, in the mammal [4]. Thus, the study of the genetic and molecular mechanisms responsible for the ability of Y. pestis to infect and be transmitted by fleas is of interest from evolutionary, ecological, and general biology perspectives. It also has practical applications. As described below, Y. pestis develops a biochemically complex life stage in the flea that includes both bacterial and flea-derived components. It is in this unique form that the bacteria enter the mammal during a flea bite and first encounter the mammalian innate immune response.

The Insect Host Environment
Just as in the mammalian host, Y. pestis must sense specific environmental conditions and immune responses and adapt accordingly in order to successfully infect its insect host. The situation is less complicated in the insect, however, as Y. pestis remains confined to the within the first few hours, when many of the bacteria coalesce into large aggregates. These multicellular aggregates often extend from the proventricular valve into the midgut ( Figure 1B,C). Fleas so colonized are able to transmit Y. pestis when they next feed again, a phenomenon referred to as early-phase transmission or mass transmission. In this case, the bacterial mass in the proventriculus is sufficient to partially and temporarily interfere with blood flow into the midgut, resulting in some backflow or regurgitation of blood containing dislodged bacteria into the bite site [17][18][19]. The obstruction is transient, however, as the bulk of the bacterial mass is readily flushed out of the proventriculus back into the midgut en masse by the incoming flow of blood. Following feeding, one part of the aggregate in the midgut can sometimes be seen to maintain the bulbous shape of the proventricular valve, ostensibly because it had become molded into a "cast" in its previous location [19]. This early-phase transmission was the first to be described and is relatively inefficient, as it requires several infected fleas feeding on an individual naïve host for successful transmission [20,21]. A second, related phase of transmission was subsequently described by Bacot and Martin, who observed that, with time, the bacterial aggregates could become more or less permanently entrenched in the proventriculus, with a high degree of resistance to being flushed back into the midgut during a blood meal [22]. In this case, blood feeding is partially or completely blocked by a bacterial mass that can completely fill the proventriculus ( Figure 1E,F), and the ensuing hydrodynamic forces generated by continuous futile feeding act to reflux blood mixed with bacteria back into the bite site. In contrast to early-phase transmission, a single blocked flea has high transmission potential, particularly since it will make repeated feeding attempts over a period of a few days until it succumbs to starvation.

Y. pestis Transmission Factors
A major pursuit of microbial pathogenesis research has been the discovery of virulence factors, the biomolecules that are responsible for disease in vertebrates. The analogous biomolecules that are required to produce a transmissible infection in the invertebrate vector host have been termed transmission factors. Just as they do in the mammal, arthropodborne pathogens must adapt to environmental conditions and antibacterial factors in their invertebrate hosts. They also often develop a particular phenotypic stage in a specific location in the arthropod that promotes their transmission. A shift from 37 • C to ambient, flea temperature by itself markedly affects Y. pestis gene expression in vitro [23,24]. Prominent virulence factors such as the F1 capsule and pH 6 antigens, iron acquisition operons, and RovA are downregulated [25]. Transcriptomic analyses of Y. pestis in mammalian host tissue and the flea digestive tract, as well as analyses of the flea transcriptomic response to oral infection with Y. pestis, have pointed out genes and gene regulatory systems that are induced in mammal or flea [10,13,[25][26][27][28][29]. These have helped to lay the groundwork for much current research to identify and characterize the specific molecular interactions of Y. pestis with its insect host that lead to a transmissible infection.

Biomolecules Required to Survive and Grow in the Flea Digestive Tract: Resistance to the Flea Antibacterial Response
After entering the flea, Y. pestis experiences a rapid decrease in temperature and other changes in environment as the flea immediately begins to digest its blood meal. Digestion of host blood by fleas appears to generate a potent antibacterial product that is cytolytic to Y. pestis and other Gram-negative bacteria. A key biomolecule produced by Y. pestis is Ymt, a phospholipase D enzyme that acts to protect against the cytolytic agent [30] ( Table 1). The ymt gene was acquired by Y. pestis after it diverged from Y. pseudotuberculosis by lateral gene transfer. Interestingly, the most closely related homologs of Ymt in GenBank have been found in other insect-associated Enterobacteriaceae, particularly Arsenopholus and Photorhabdus species. The Y. pestis Ymt protein is most similar to one from an Arsenopholus symbiont of an Asian insect pest of rice, consistent with phylogenetic evidence that Y. pestis first emerged in Asia.

Y. pestis Transmission Factors
A major pursuit of microbial pathogenesis research has been the discovery of virulence factors, the biomolecules that are responsible for disease in vertebrates. The analogous biomolecules that are required to produce a transmissible infection in the invertebrate vector host have been termed transmission factors. Just as they do in the mammal,

Name Molecular Function Role in the Flea Reference
Ymt phospholipase D protection against toxic byproduct of blood digestion [30] RovM transcriptional regulator (i) metabolic adaptation to the flea gut; (ii) upregulates hmsCDE, hmsT; increases intracellular c-di-GMP and biofilm production [31,32] OxyR transcriptional regulator protection against reactive oxygen species generated in the flea [13,33,34] GalU, ArnB lipid A modification enzymes protection against antimicrobial peptides of flea immune response [35] HmsHFRS glycolsyl transferase, polysaccharide deacetylase, transport porin synthesis of extracellular β-1,6-GlcNAc polysaccharide biofilm matrix component [36][37][38]  The flea immune response to oral infection is analogous to the mammalian innate immune response, and the expression of several Y. pestis factors that protect against such a response is upregulated in the flea [10,27]. The PhoPQ two-component system (2CS) senses the decrease in pH experienced in the flea gut, which mediates lipid A modifications that protect against antimicrobial peptides [45]. Likely in response to the flea's Duox-ROS system, the OxyR regulatory system protects against oxidative stress encountered in the flea, as a Y. pestis oxyR mutant is deficient for colonization [13]. On the other side, bacterial infection is potentially harmful to the health of the flea. It would be counterproductive if Y. pestis caused significant morbidity and mortality to its flea host before it was transmitted. Midgut infection with Y. pestis does not cause much morbidity, and fleas can remain infected and viable for months [59] (Figure 2). An important factor in this regard was the loss of the urease component gene ureD during the evolution of Y. pestis [60] ( Table 2). The strong urease activity of the Y. pseudotuberculosis progenitor generates acute toxicity and high mortality of fleas soon after oral infection. Restoration of ureD in Y. pestis results in the same effect. Of course, once proventricular blockage develops, fleas starve to death within a few days. However, as noted previously, starvation actually enhances transmission because hungry and dehydrated blocked fleas make continued, persistent feeding attempts, each one with the possibility of regurgitative transmission.

Biomolecules Required to Survive and Grow in the Flea Digestive Tract: The Bacterial Autoaggregation Phenotype
The antibacterial effector pathways and molecules are important for resistance to the flea immune response and survival in the flea gut, but Y. pestis is also subject to clearance from the flea simply by elimination in flea feces. This is potentially a significant risk because Y. pestis does not adhere to flea digestive tract epithelia or proventricular spines. Y. pestis counteracts this by coalescing into large aggregates within a few hours after being ingested that localize and become enmeshed in the proventricular spines [17,19]. In addition to serving as a means to prevent elimination, aggregating bacteria are likely shielded from environmental stresses and antibacterial host responses. However, bacteria in the center of an aggregate may, for the same reason, be shielded from nutrients. Most examples of bacterial autoaggregation involve outer membrane proteins or adhesins [66], but to date no such involvement has been documented for Y. pestis in the flea. In fact, several adhesins of Y. pseudotuberculosis are absent or are pseudogenes in Y. pestis, including those encoding for InvA (invasin), YadA, YadE, Ifp/InvB, and Type IV pili [67,68]. Presumably, adhesion to and invasion of the flea digestive tract lining are detrimental because both would preclude regurgitative transmission. Bacterial fimbriae are also common adhesin molecules, and the Y. pestis genome contains nine gene clusters that encode intact chaperone/usher pathway fimbriae, including the Psa (pH 6 antigen) and Caf (F1 capsule) virulence factors. We have made deletions in six of these, but none of these individual mutants showed any deficiency in ability to colonize or block fleas [47], nor are any of the six essential for autoaggregation or biofilm formation in vitro [69]. The Y. pestis Ail outer surface protein can mediate autoaggregation in vitro, but an Ail mutant has a normal phenotype in the flea [70,71]. Exopolysaccharides can also be autoagglutinins, and the Hms polysaccharide is critical to the cohesiveness of Y. pestis aggregates in the flea gut; however, the large, dense bacterial aggregates form before the Hms exopolysaccharide is produced [17,19].
It is possible that redundant known or unknown Y. pestis autoagglutinins function to cause autoaggregation in the flea gut, but it is also possible that autoaggregation is passive, not mediated by bacterial factors but by physical conditions in the midgut. For example, simply transferring Y. pestis from pH 7 to pH 5.5 LB medium results in rapid autoaggregation [28]. Thus, the mechanism could involve surface electrostatic effects, such as hydrophobic interactions in an aqueous solution, or depletion interaction aggregation, a type of mechanically driven phase separation [66,72,73]. A non-bacterial substance pre- Figure 2. Digestive tract from a flea dissected one month after infection with Y. pestis expressing green fluorescent protein, photographed using bright field (A) and fluorescence (B) microscopy. This flea appeared healthy and took a normal blood meal even though its midgut was partially occupied by a large mass of Y. pestis biofilm.

Biomolecules Required to Survive and Grow in the Flea Digestive Tract: The Bacterial Autoaggregation Phenotype
The antibacterial effector pathways and molecules are important for resistance to the flea immune response and survival in the flea gut, but Y. pestis is also subject to clearance from the flea simply by elimination in flea feces. This is potentially a significant risk because Y. pestis does not adhere to flea digestive tract epithelia or proventricular spines. Y. pestis counteracts this by coalescing into large aggregates within a few hours after being ingested that localize and become enmeshed in the proventricular spines [17,19]. In addition to serving as a means to prevent elimination, aggregating bacteria are likely shielded from environmental stresses and antibacterial host responses. However, bacteria in the center of an aggregate may, for the same reason, be shielded from nutrients. Most examples of bacterial autoaggregation involve outer membrane proteins or adhesins [66], but to date no such involvement has been documented for Y. pestis in the flea. In fact, several adhesins of Y. pseudotuberculosis are absent or are pseudogenes in Y. pestis, including those encoding for InvA (invasin), YadA, YadE, Ifp/InvB, and Type IV pili [67,68]. Presumably, adhesion to and invasion of the flea digestive tract lining are detrimental because both would preclude regurgitative transmission. Bacterial fimbriae are also common adhesin molecules, and the Y. pestis genome contains nine gene clusters that encode intact chaperone/usher pathway fimbriae, including the Psa (pH 6 antigen) and Caf (F1 capsule) virulence factors. We have made deletions in six of these, but none of these individual mutants showed any deficiency in ability to colonize or block fleas [47], nor are any of the six essential for autoaggregation or biofilm formation in vitro [69]. The Y. pestis Ail outer surface protein can mediate autoaggregation in vitro, but an Ail mutant has a normal phenotype in the flea [70,71]. Exopolysaccharides can also be autoagglutinins, and the Hms polysaccharide is critical to the cohesiveness of Y. pestis aggregates in the flea gut; however, the large, dense bacterial aggregates form before the Hms exopolysaccharide is produced [17,19].
It is possible that redundant known or unknown Y. pestis autoagglutinins function to cause autoaggregation in the flea gut, but it is also possible that autoaggregation is passive, not mediated by bacterial factors but by physical conditions in the midgut. For example, simply transferring Y. pestis from pH 7 to pH 5.5 LB medium results in rapid autoaggregation [28]. Thus, the mechanism could involve surface electrostatic effects, such as hydrophobic interactions in an aqueous solution, or depletion interaction aggregation, a type of mechanically driven phase separation [66,72,73]. A non-bacterial substance present in the flea gut could also favor autoaggregation. For example, the bacterial aggregates appear to be surrounded by a brown, viscous, lipid-or waxy-appearing material, which may derive from blood lipid, protein and adsorbed hemin (Figures 1 and 2).

Biomolecules Required to Survive and Grow in the Flea Digestive Tract: The Response to Nutrient Limitation and Other Stresses
Several lines of evidence indicate that the flea digestive tract is a physically stressful environment to Gram-negative bacteria. A transcriptional response is induced during infection of the flea gut that acts to protect Y. pestis to acid, osmotic, and other cell envelope stresses [25]. The PhoPQ gene regulatory system appears to be primarily induced by acid stress in the flea rather than by low Mg 2+ ion concentration and modifies the bacterial outer membrane to confer resistance to insect antimicrobial peptides and presumably other cell envelope stressors [45]. If the PhoPQ system is nonfunctional, several other general and acid stress response genes are upregulated in the flea [28]. As the flea digests and absorbs blood meal elements, the osmolarity of the midgut contents increases [8], and several Y. pestis genes with osmoadaptive and osmoprotective functions are upregulated [25,27,29,74]. The stress-sensing alternative sigma factor, RpoE, is also upregulated in the flea [25].
Y. pestis is able to rapidly grow to large numbers in the flea gut after ingestion, often up to 10 6 CFU per infected flea. Nevertheless, the midgut contents are apparently nutrientrestricted in some ways and require a substantial shift in bacterial metabolism. The gene expression pattern in the flea indicates that Y. pestis uses alternate carbon sources in the flea, particularly amino acids, rather than primary carbon sources, such as glucose [25,27], an adaptation that requires the cAMP receptor protein (CRP), the regulator of alternate carbon source metabolism and other cellular processes [48]. The gene for RovM, a transcriptional regulator of Y. pseudotuberculosis and Y. pestis that orchestrates the adaptive response to nutrient limitation in vitro, is highly upregulated in the flea [25,27]. A Y. pestis rovM mutant has a growth deficiency in the flea [31], another indication that the midgut is nutrient-limited.
Thus, although the oral ID 50 dose required for Y. pestis to successfully produce a chronic infection in its flea host is high (a bacteremia level of~10 7 CFU/mL; [4,75]), and the flea gut a rather hostile environment for Gram-negative bacteria, Y. pestis is able to adapt, survive, and grow in that environment. Its numbers increase during the first week after an infectious blood meal and then plateau. However, in addition to intrinsic bacterial factors, it is important to note that characteristics of host blood can influence the infection and transmission efficiency of Y. pestis in fleas. For example, host bloods with a relatively insoluble hemoglobin molecule, such as rat blood, engender a more cohesive foregut infection that results in increased early-phase transmission efficiency [18]. Uncharacterized differences in flea species physiology also can influence vector competence-the infection rate of Oropsylla montana, a North American ground squirrel flea, is lower than that of Xenopsylla cheopis following an infectious mouse blood meal [76].

Biomolecules Required for Mature Biofilm Formation in the Flea Digestive Tract
A nearly universal adaptive response of bacteria to a stressful, growth-restricted environment is the development of a multicellular biofilm, which provides a protected and homeostatic niche that maximizes survival and persistence [77]. The Y. pestis life stage in its insect host is an example of this strategy. The initial autoaggregation and localization to the proventriculus can be seen as the first steps in this process. A mature biofilm, however, depends on a specific Y. pestis exopolysaccharide that is an important stabilizing component of the extracellular polymeric substance (EPS), or matrix, of the biofilm. The synthesis and export of this poly-β-1,6-N-acetyl-D-glucosamine exopolysaccharide is mediated by the Y. pestis hmsHFRS genes [36,64]. The Hms expolysaccharide is foundational for biofilm development in vitro and in the flea, and it is required for proventricular blockage in the flea [37,38].
Although the stresses, particularly nutrient limitation, that induce biofilm formation are common to many bacteria, the regulatory pathways and developmental programs are diverse and complex among different bacteria. For Y. pestis, two major regulatory systems that are induced in the flea (RovM and PhoPQ) have also been found to positively regulate biofilm development [32,78]. PhoPQ as well as another 2CS, OmpR-EnvZ, modify the bacterial outer surface to counteract environmental stresses, and these alterations are also important for biofilm formation. Y. pestis PhoP and OmpR-EnvZ mutants are able to infect fleas but are deficient in proventricular blocking ability [8,45], most likely due to lack of appropriate outer membrane modifications. For example, certain genes that affect LPS structure, such as gmhA, yrbH, and waaA, are required for a normal rate of flea blockage [51,56]. Ribose-P-isomerase, RpiA, is also required for optimal proventricular blockage [19]. The mechanism behind this is unknown, but also appears to be due an effect of RpiA on the Y. pestis outer membrane or biofilm EPS. From these results, it can be seen that Y. pestis outer surface characteristics, in addition to the Hms exopolysaccharide, are required for normal biofilm formation in the flea.
Another global regulator, CRP-cAMP, is critically involved in both metabolic adaptation to and biofilm development in the flea. CRP controls the carbon catabolite repression response by upregulating alternate carbon source metabolic pathways when glucose and other preferential carbohydrate sources are limited, which is the case in the flea midgut. The cAMP-CRP system appears to be a critical link between alternative carbon metabolism and biofilm development, as Y. pestis lacking CRP is deficient in biofilm formation in vitro [48,49]. Furthermore, although this mutant strain is able to survive in and produce a chronic infection in fleas, it never produces proventricular blockage [47].
Another layer of regulation of biofilm development occurs via small RNA molecules (sRNA). Two sRNAs, named HmsA and HmsB, have been implicated in activating biofilm development in vitro by upregulating the expression of the HmsCDE, HmsT, and HmsHFR genes [54,55]. A role for sRNA in biofilm development in the flea is also indicated by the fact that the Hfq RNA chaperone protein is required for proventricular blockage [52], although this may be due to a stabilizing effect that Hfq has on hmsP mRNA [53]. Another clue that sRNAs are involved in the regulation of biofilm development is that CsrA, the sRNA-binding protein component of the carbon storage system CsrABC, is required for normal Hms-exopolysaccharide production and biofilm development in vitro [48].
These complex and varied regulatory systems that upregulate biofilm formation in the flea function together to induce the Hms-exopolysaccharide, and other factors, essential for a cohesive EPS. In Y. pestis and many other Gram-negatives, the bacterial second messenger c-di-GMP is a critical molecule that induces a shift from planktonic to biofilm growth [79,80]. Intracellular levels of c-di-GMP in Y. pestis are determined by the relative activities of two diguanylate cyclase (DGC) enzymes, HmsT and Hms D, and one phosphodiesterase (PDE) enzyme, HmsP, that, respectively, synthesize and degrade c-di-GMP [39,40]. HmsD is the most active in the flea and is encoded in an operon that includes two flanking genes, hmsC and hmsE, which act to regulate hmsD expression [40][41][42][43]. These and other factors that act to promote Hms-exopolysaccharide and biofilm production are summarized in Table 1. Their complex interrelated regulatory patterns have been reviewed recently [32,78].

Biomolecules Left behind during the Evolution to Flea-Borne Transmission
The evolutionary change of Y. pestis to an arthropod-borne pathogen can be seen, in large part, as a fine-tuning of the preexisting biofilm development pathways of its Y. pseudotuberculosis ancestor to upregulate biofilm growth in, and transmissibility from, the flea gut. Y. pseudotuberculosis is able to form Hms-dependent biofilm in certain environmental conditions but does not do so in the flea [81]. In addition to gene gain (e.g., ymt), gene loss was important to this evolutionary process. For example, the Rcs multicomponent regulatory system inhibits biofilm development in Yersinia and other bacteria, and this system is highly expressed in the flea [25,62,82,83]. One of its components (rcsA), however, is a pseudogene in Y. pestis, thereby alleviating this negative regulation and increasing biofilm production in the flea [84]. In addition, two c-di-GMP degrading PDEs of Y. pseudotuberculosis were eliminated by gene loss in Y. pestis [39,40]. The importance of these genetic changes was demonstrated by introducing them into a Y. pseudotuberculosis strain, which then became capable of forming proventricular-blocking biofilm in fleas [61]. Gene losses that were important to adapting to the flea-borne route of transmission are summarized in Table 2.

Future Directions
Over 100 years after the seminal report of Bacot and Martin describing the proventricular blockage phenomenon and its importance for transmission, many molecular and genetic details have recently been discovered about how Y. pestis produces a transmissible infection in fleas. However, many missing pieces remain to be discovered. Although they both lead to c-di-GMP and Hms-exopolysaccharide production, the regulatory pathways differ in some ways between Y. pestis and Y. pseudotuberculosis that have yet to be explained. For example, RovM reportedly upregulates biofilm development in Y. pestis but represses it in Y. pseudotuberculosis by negatively regulating hmsHFRS [31,32,85]. In addition, biofilm formation in vitro and in the flea depends upon the expression of a Type 6 secretion system (T6SS) in Y. pseudotuberculosis but not in Y. pestis [29].
The effector mechanisms of the c-di-GMP signaling network have not yet been characterized. In other bacteria, these typically involve c-di-GMP-binding proteins and RNA molecules that interact with downstream targets to upregulate biofilm development [80]. Other unknowns include the molecular makeup of the Y. pestis biofilm EPS in the flea, much of which forms around Y. pestis in the flea before production of the Hms-exopolysaccharide [17,19]. The physiological state of Y. pestis during chronic infection of the flea in the biofilm state and the mechanism of bacterial release from the biofilm, an important element of transmission, have also not been investigated. Finally, although much can be learned from in vitro models of biofilm development, it will be essential to test the effect of implicated factors in the flea model, which cannot be duplicated by any in vitro conditions. Much work remains to be done to comparatively evaluate the vector competence of the many flea species implicated as vectors in the many ecological settings in which Y. pestis circulates. The biochemical characteristics of different host bloods and differences in flea-feeding behaviors, flea digestive tract physiology and proventricular anatomy all likely contribute to reported vector competence differences [17,18], but the mechanisms in large part have yet to be characterized.