Typical Y. pestis strains form red colonies (pigmented or Pgm⁺) after growth at ≤34°C on media containing Congo red, but white colonies (Pgm^–) may be isolated at a frequency of 10⁻⁴ [4]. Most spontaneous Pgm^– mutants result from the deletion of a 102-kb chromosomal region termed the pgm locus [5], [6]. This locus includes the haemin storage operon (hmsHFRS) which is essential for the pigmentation phenotype and for the production of a biofilm in the flea gut that can block normal blood feeding; the blockage of the flea's digestive tract is considered to be an important process for flea-borne transmission [7]. The pgm locus also contains the Yersinia high-pathogenicity island (HPI), which carries among other genes the irp1-irp2-ybtU-ybtT-ybtE, the ybtP-ybtQ-ybtX-ybtS and the psn loci that encode the yersiniabactin (Ybt) siderophore-based iron acquisition and transport system. The irp genes encode the high molecular weight proteins (HMWP) 1 and 2 which act in concert with YbtU, YbtE, YbtS and probably YbtT to synthesize the Ybt siderophore [8]. Ybt is secreted, acquires iron from transferrin and lactoferrin in host tissues, then is transported back into Y. pestis by the TonB-dependent outer membrane receptor Psn and the inner membrane ABC-transporter YbtP-YbtQ. A critical role of the Ybt system in bubonic plague is indicated by the fact that Ybt^– Y. pestis strains are essentially avirulent by the subcutaneous inoculation route that mimics the flea bite, although these strains retain complete or nearly complete virulence when inoculated intravenously [5], [9], [10], [11]. Presumably, Ybt is required to provide sufficient iron at the peripheral injection site, in the draining lymphatic system, and/or in the lymph nodes, suggesting that Ybt would be an essential virulence factor for flea-borne bubonic plague.

A flea-borne transmission model was used to determine Y. pestis infectivity after challenge by flea bite [2]. Xenopsylla cheopis rat fleas were allowed to feed on heparanized mouse blood containing wild-type Y. pestis or an isogenic irp2 mutant, using an artificial feeding device, and maintained as previously described [7]. Beginning 13 days after their infectious bloodmeal (the time required for Y. pestis to block X. cheopis), 49 to 115 fleas were applied to a restrained mouse and allowed to feed for 60 min. Immediately after the challenge, the fleas were examined individually under a dissecting microscope to determine how many infective (blocked) fleas had bitten each mouse [7]. Mice that did not develop any symptoms within 5 days following a challenge were re-challenged until the cumulative number of bites from blocked fleas was high enough to consider that a successful transmission occurred [2], [14]. Mice received one to six different sequential challenges. Challenged animals were observed at least three times daily for three weeks and euthanized upon signs of terminal plague (evidence of lethargy, hunched posture, and reluctance to respond to external stimuli) [15]. Bacterial load in the spleen and blood was determined by colony-forming unit (CFU) count. Hematoxylin and eosin (H&E) and immunohistochemical (IHC) staining to detect Y. pestis [15] was performed on formalin-fixed inguinal lymph node sections.

Six of the nine sick mice infected by fleabite with the wild-type strain were diagnosed with typical bubonic plague, characterized by severe lymphadenitis with destruction of the lymph node architecture and the presence of numerous bacteria admixed with cellular debris (Fig. 2A to 2C). The other three mice did not have obvious lymphadenitis, but the spleen and blood of all nine mice were heavily colonized (8.7±1.2 and 6.9±0.9 log₁₀ CFU per ml of blood and spleen respectively). From these results one can infer that six mice had bubonic plague followed by sepsis and 3 mice developed primary septicemic plague (Table 1). In contrast, disease outcome in mice bitten by fleas infected with the irp2 mutant was significantly different (P=0.0015 by Fisher's exact test). Neither of the two mice that developed terminal disease after being bitten by fleas infected with the irp2 mutant had typical bubonic plague (Figure 2). Histologic analyses of the lymph nodes proximal to the flea bite site of these mice did not reveal any bacteria (Fig. 2F and 2I), but the mouse that developed terminal plague at 6 days had evidence of lymphadenopathy (Figure 2D and 2E). Lymph node pathology was localized; however, many immature lymphocytes and macrophages containing ingested apoptotic lymphocytes were present throughout the entire lymph node. The etiology of this lymphadenitis is uncertain. It could have resulted from hematogenous spread subsequent to primary septicemic plague; alternatively, the lymph node may have been initially colonized and the bacteria disseminated to the blood before being eliminated from the node, in an atypical form of bubonic plague. Regardless of lymph node histopathology, both mice had a high bacterial load in the blood and the spleen (Table 1) (8.3 and 8.6 log₁₀ CFU in the spleen and 3.8 and 7.6 log₁₀ CFU per ml of blood). Hence, one mouse infected with the irp2 mutant had primary septicemic plague, and the other had septicemic plague associated with a mild lymphadenitis. For mice challenged with either wild-type or Δirp2 Y. pestis, disease outcome (bubonic plague, primary septicemic plague, or no disease) did not correlate with the cumulative number of challenges or infective flea bites (P>0.05 by Fisher's exact test).

The present results with a Y. pestis irp2 mutant mirror our previous results with a Y. pestis plasminogen activator (pla) mutant [2]. Both mutants are avirulent by the ID route and fully virulent by the IV route, but cause plague at low incidence following fleabite, despite the fact that their LD50 by the ID route (>10⁵ CFU) is several orders of magnitude higher than the number of CFU transmitted by blocked fleas (median <100 CFU) [2], [17], [18]. We previously proposed that direct injection of bacteria into a dermal blood vessel during the flea bite can lead to primary septicemic plague, with no prior bubonic stage [2]. The data herein provides independent support for this model.