Plague is an often fatal, primarily flea-borne zoonotic disease caused by the bacterium Yersinia pestis and is characterized by quiescent periods punctuated by rapidly spreading epizootics (Barnes, 1982). Although maintenance of plague bacteria in nature is poorly understood, it is largely believed that Y. pestis circulates in enzootic cycles involving transmission between moderately resistant rodents and their fleas (Eisen & Gage, 2009; Gage & Kosoy, 2005; Politzer, 1954). The ‘blocked flea’ model, described by Bacot & Martin (1914), has been recognized as the primary mechanism of flea-borne transmission in the western literature for nearly a century (Eisen et al., 2009). Under this scenario, Y. pestis adheres to and forms an obstruction in the flea proventriculus, the spine-filled valve connecting the oesophagus to the midgut; aggregation of bacteria in the proventriculus prevents its normal function and complete blockage prevents blood from reaching the midgut (Bacot & Martin 1914). More recently, the molecular mechanisms facilitating blocked flea transmission and survival of the plague bacterium in the flea midgut have been identified. Specifically, the gene encoding Yersinia murine toxin (ymt) and the genes in the haemin storage locus (hms) have been shown to be important in facilitating colonization (i.e. survival and reproduction) of the midgut and proventricular blockage, respectively, within the flea (Hinnebusch et al., 1996, 1998, 2000, 2002; Jones et al., 1999; Kirillina et al., 2004; Perry et al., 1990; Rudolph et al., 1999).

The blocked flea model is likely to be important for the long-term maintenance of Y. pestis; however, mathematical modelling has suggested that it is unlikely that transmission by blocked fleas could completely explain the rapid rates of transmission typically observed during epidemics. Due to the long extrinsic incubation period, the low transmission efficiency, and the short infectious window from when a flea becomes blocked and infectious until it dies of starvation, blocked flea transmission could not drive epidemic or epizootic transmission unless the number of fleas per host is very high and the flea species involved frequently becomes blocked within relatively short periods of time (Drancourt et al., 2006; Eisen et al., 2006; Lorange et al., 2005; Webb et al., 2006). Recent work on what has been termed ‘early-phase transmission’ (EPT), has shown that transmission occurs efficiently during the first 1–4 days post-infection (p.i.) of the flea, a period prior to when block formation has been demonstrated to occur (Eisen et al., 2006, 2007a, b, 2008a, b; Wilder et al., 2008a).

Although EPT provides an ecological mechanism for the rapid spread of plague bacteria during epidemics and epizootics (Eisen et al., 2006), the bacteria–vector interactions that result in EPT are not well understood. Potential mechanisms of EPT include: (i) regurgitation of Y. pestis from the flea’s midgut through the proventriculus, (ii) reflux of Y. pestis from the foregut or internal mouthparts, (iii) mechanical transmission of Y. pestis from contaminated external mouthparts, or (iv) excretion of Y. pestis contaminated faeces (for a review, see Hinnebusch, 2012). Here, we evaluated whether Ymt, which promotes colonization of the midgut, a prerequisite for transmission via regurgitation from the midgut, was necessary for EPT by Oropsylla montana (Siphonaptera: Ceratophyllidae), an important bridging vector in North America, and Xenopsylla cheopis (Siphonaptera: Pulicidae), an important bridging vector worldwide. Both species are known to be efficient vectors during the EPT period (Eisen et al., 2007b; Vetter et al., 2010). Specifically, we infected O. montana and X. cheopis with a mutant strain of Y. pestis containing a non-functional ymt, which rendered this strain incapable of long-term colonization (≥ 1 week p.i.) of the flea midgut (Hinnebusch et al., 2002). We showed that EPT did not depend on the presence of Ymt, a protein that is important for colonization of the midgut, but not the proventriculus.

EPT was documented for all three Y. pestis strains and both flea species. The bacterial concentrations contained in the feeders for each artificial infection feed differed between flea species, but not amongst flea species (Fig. 1).

In contrast to the blocked flea model of transmission for which the physiological and molecular mechanisms are well understood, the interactions between the flea and Y. pestis that result in EPT remain elusive. However, the genetic mechanisms driving blocked flea transmission seem to be distinct from those involved in EPT. Under the blocked flea model, certain genes are required for blockage of the proventriculus (e.g. hmsHFRS). Both ymt and hms are known to be important in facilitating colonization (i.e. survival and reproduction) of the midgut and proventricular blockage, respectively, within the flea (Hinnebusch et al., 1996, 1998, 2000, 2002; Jones et al., 1999; Kirillina et al., 2004; Perry et al., 1990; Rudolph et al., 1999). Our study, combined with a prior study that evaluated the role of biofilm formation in EPT (Vetter et al., 2010), has shown that neither Ymt nor Δ hmsR or Δ hmsT are required for this mode of transmission.

Infection prevalence reported here for X. cheopis fleas infected with ymtH188N differs from that reported by Hinnebusch et al. (2002). Previous work has shown that at 24 h p.i., the ymtH188N mutant displayed an aberrant spheroplast form in the flea midgut that was assumed to have preceded elimination from the flea, as the mutant was not detected in the midgut at 7 days p.i. (Hinnebusch et al., 2002). Although it is unlikely (Hinnebusch et al., 2002) ymtH188N may survive long enough (>3 days p.i.) in the flea midgut to promote EPT, it is possible that the bacteria are occupying other regions of the flea that would leave Y. pestis more accessible for transmission (e.g. the proventriculus). We did not perform midgut dissections to determine the distribution of Y. pestis within each flea.

Consistent with previous early-phase studies, our results show no correlation between bacterial loads in the flea and transmission success (Ajl et al., 1955; Eisen et al., 2006, 2007a, b, 2008a, b Vetter et al., 2010; Wilder et al., 2008a, b). Our results support previous assertions that midgut colonization may not be necessary for EPT and that location of the bacteria within the flea is more important in predicting transmission outcomes than the flea’s bacterial load (Eisen et al., 2007a; Eisen & Gage, 2009). Specifically, the probability of transmission likely declines as bacteria move further posterior in the digestive tract during the early phase (Eisen et al., 2007a).

One might ask why bacteria in the midgut would be less transmissible than those in the foregut or mouthparts. During the acquisition of a blood meal, blood is drawn up from the host by two pumps powered by muscles attached to the cibarial and pharyngeal regions of the foregut, respectively, in the head of the flea. The cibarial and pharyngeal pumps contract one after the other to force blood backwards through the oesophagus, into the proventriculus and finally into the midgut; at the same time, a salivary pump forces secretions from the salivary glands out through the hypopharynx (Furman & Catts, 1982). When the flea is feeding, the proventriculus opens and closes in concert with the cibarial and pharyngeal pumps in the head, maintaining the blood flow generally from anterior to posterior within the flea. Between blood meals, the proventriculus remains tightly closed, maintaining separation between the midgut and oesophagus. As a result, bacteria in the midgut might be less likely to be transmitted than bacteria found further anterior in the flea. However, because the proventriculus is constantly opening and closing during feeding, it is reasonable to presume that some midgut contents could escape and be regurgitated. If regurgitation from the midgut is a primary source of EPT, it would be expected that fleas infected with the ymt mutant strain would have exhibited lower transmission rates than the two WT strains tested here, which is contrary to what we observed in our study. If the proventriculus is being passively colonized during the initial ingestion of a blood meal and bacteria are being transmitted from the proventriculus during EPT, we would expect to observe no difference amongst strains regarding EPT efficiency. However, the likelihood of proventricular colonization differs between vector species and should be accounted for. O. montana are rarely observed to form a proventricular blockage and Y. pestis is rarely obvious in the proventriculus for the first few days following infection, whereas X. cheopis are often colonized by Y. pestis in both the proventriculus and midgut, with colonies of the bacterium being visible in the proventriculus soon after initial infection (Burroughs, 1947; Engelthaler et al., 2000; Eskey, 1938). These well-known differences in blockage formation suggest for unknown reasons that the proventriculus of O. montana is less conducive to Y. pestis establishment than the proventriculus of X. cheopis, which is readily colonized and supports rapid growth of the plague bacterium in the days immediately following infection. This difference in colonization capacity, combined with the fact that ymt-mutant-infected O. montana received a lower concentration of Y. pestis than WT-infected fleas in the infectious meal, could explain why ymtH188N-infected O. montana had significantly fewer Y. pestis per flea compared with WT-infected O. montana, whilst we observed no difference in bacterial loads amongst the same strains within infected X. cheopis.

Compared with the need for midgut colonization, the extent to which the proventriculus may play a role in EPT is less clear; although it is apparent that the capacity for aggregation and biofilm formation is not required for EPT to occur (Vetter et al., 2010). Y. pestis strains incapable of or with diminished abilities for making biofilm were transmitted as efficiently as WT strains during the early-phase period (Vetter et al., 2010). However, we cannot rule out the possibility that, during the early phase, plague bacteria may become temporarily lodged within the spines of the proventriculus, and are then sloughed off and regurgitated during subsequent feeds. Hinnebusch et al. (2002) showed that a few X. cheopis challenged with ymtH188N had chronic infections at day 28 in the proventriculus and that a few (0.5 %) of these fleas became blocked.

Although the principal mechanism of EPT remains undefined, several additional potential mechanisms can and have been ruled out, whilst others await further investigation. Areas in the anterior portion of the flea, including the anterior pharynx and the posterior mouthparts, which are protected within the head capsule of the flea, may harbour viable Y. pestis for longer periods of time (Plague Commission, 1907; Hinnebusch, 2012). Residual blood has been observed within the pharynx and between the flea mouthparts after feeding. When a flea feeds, the mouthparts come together to form a feeding canal through which blood passes into the flea, as well as a salivary canal, which carries saliva out of the flea; transmission could occur if infected residual blood contaminates the grooves of the salivary canal and is transmitted to the bite site (Hinnebusch, 2012). Further, if midgut colonization is not essential for EPT, it is unlikely that faecal contamination is a primary mechanism of EPT. It should be noted, however, that we cannot rule out faecal contamination as a mode of transmission in our work or in the study by Vetter et al. (2010). Although Y. pestis KIM5+ strains ymtH188N and Δ hmsR (Vetter et al., 2010) are unable to become established within the flea midgut and proventriculus, respectively, some small possibility remains that viable Y. pestis might have been excreted in the faeces of potentially infected fleas resulting in EPT. The use of SKH-1 mice in this study, which are hairless, would have optimized the likelihood of this form of transmission compared with similar experiments done with a host covered by a pelt that might have reduced contact between the skin and flea faeces. Such results could also argue against the suggestion that EPT occurs as a result of contamination of blood feeding sites with infectious flea faeces.

We conclude that the pathogen–vector interactions that result in EPT are distinctly different from those described for the blocked flea transmission model and do not require the presence of Ymt. Although the mechanism of EPT remains undefined, our study suggests that future efforts should focus on early colonization events of the proventriculus or perhaps other portions of the flea foregut.