Yersinia pestis Intracellular Parasitism of Macrophages from Hosts Exhibiting High and Low Severity of Plague

Background Yersinia pestis causes severe disease in natural rodent hosts, but mild to inapparent disease in certain rodent predators such as dogs. Y. pestis initiates infection in susceptible hosts by parasitizing and multiplying intracellularly in local macrophages prior to systemic dissemination. Thus, we hypothesize that Y. pestis disease severity may depend on the degree to which intracellular Y. pestis overcomes the initial host macrophage imposed stress. Methodology/Principal Findings To test this hypothesis, the progression of in vitro infection by Y. pestis KIM62053.1+ of mouse splenic and RAW264.7 tissue culture macrophages and dog peripheral blood-derived and DH82 tissue culture macrophages was studied using microscopy and various parameters of infection. The study showed that during the early stage of infection, intracellular Y. pestis assumed filamentous cellular morphology with multiple copies of the genome per bacterium in both mouse and dog macrophages. Later, in mouse macrophages, the infection elicited spacious vacuolar extension of Yersinia containing vacuoles (YCV), and the filamentous Y. pestis reverted to coccobacillary morphology with genomic equivalents approximately equaling colony forming units. In contrast, Y. pestis infected dog macrophages did not show noticeable extension of YCV, and intracellular Y. pestis retained the filamentous cellular morphology for the entire experiment in DH82 cells or were killed by blood-derived macrophages. In addition, during the later stage of infection, Y. pestis infected mouse macrophages exhibited cell lysis whereas dog macrophages did not. Conclusion/Significance Overall, these results support our hypothesis that Y. pestis in mouse macrophages can overcome the initial intracellular stress necessary for subsequent systemic infection. However, in dogs, failure of Y. pestis to overcome macrophage imposed stress may result in mild or in apparent disease in dogs.


Introduction
The etiological agent of plague, the Gram-negative bacterium Yersinia pestis, causes severe disease in natural rodent hosts such as mice, ground squirrels and prairie dogs, but mild to in apparent disease in some rodent predators such as domestic dogs and coyotes [1][2][3][4]. Y. pestis is maintained in rodent populations in endemic areas as a flea transmitted disease [1,2]. Rodent predators acquire the infection by either ingestion of infected rodents or via bite of Y. pestis infected rodent fleas [5][6][7][8]. The mechanism underlying the difference in disease severity of rodents and canines to infection by Y. pestis is not understood.
The high lethality of Y. pestis infection in rodents is demonstrated by the periodic extinction of local rodent populations during seasonal plague epizootics as well as by high mortality rates in experimental infection studies in rodents. In brown Norway rats, intradermal inoculation of 5610 2 CFU per animal, to mimic the natural flea bite transmission, caused 100% mortality within 3 to 15 days depending on the site of inoculation [9]. Shortly after the intradermal injection, reddish papular eruptions occur at the site, which is followed by enlargement of local lymph nodes, septicemia and death of infected animals [9]. Similar disease progression and mortality was observed for infection by parenteral inoculation or infected flea bites in mouse models, with infected animals succumbing to the disease within 2 to 8 days post-infection depending on the inoculation dose [10,11]. Following the bite of an infected flea or experimental injection, subcutaneous Y. pestis are phagocytized by tissue neutrophils and macrophages [1,12,13]. Y. pestis are readily killed by neutrophils, but this initial neutrophilic restriction of Y. pestis is only effective for the first few hours post-infection; thereafter, expression of anti-phagocytic factor F1-antigen by Y. pestis reduces this process [1,14,15]. In contrast to killing by neutrophils, Y. pestis survives inside rodent macrophages during the early stage of infection [1]. Phagosomes containing Y. pestis mature from early endosomes to phagolysosomes, but the bacteria are able to survive and multiply, thereby allowing dissemination while evading host innate immunity [16][17][18][19][20][21][22]. While residing in phagolysosomes, Y. pestis expresses various stress response and virulence genes such as type-III secretion system, and F1-and pH 6-antigens; and modifies the phagolysosomes into spacious vacuoles to adapt for progression of the infection by systemic dissemination [23][24][25][26]. Depletion of macrophages in mice by treatment with clodronate-liposomes dimin-ished the severity of infection by Y. pestis as indicated by a marked reduction in lesions in spleens and livers of inoculated animals [27]. Overall, infection studies support the role of Y. pestis infection of host macrophages in establishing local infection and systemic dissemination of infection following introduction of Y. pestis into the host through flea bites [15].
In contrast to flea transmission in natural rodent hosts, rodent predators acquire Y. pestis primarily by ingestion of infected rodents [5,6]. Some rodent predators such as dogs and coyotes are less susceptible for developing severe disease from infection by Y. pestis, whereas other rodent predators such as black-footed ferrets and domestic cats are highly susceptible to infection by Y. pestis, developing lymphadenopathy of the lymph nodes of the head or neck and subsequent systemic dissemination similar to the disease progression observed in rodents [28,29]. In experimental infection studies in which cats ingested infected rodents, 80 to 100% of exposed cats developed clinical illness with 38 to 42% mortality [28,30]. Although domestic cats and dogs in endemic regions likely have similar rates of exposure to Y. pestis infected rodents, there are many more case reports in the literature of clinical disease in cats than in dogs, suggesting that the latter experience less severe disease [28,[30][31][32]. In one of the few case reports of plague in dogs, the clinical signs of mild fever, malaise, stomatitis, and transient submandibular lymph node swelling were observed [5]. Unlike experimentally infected cats, dogs infected experimentally with Y. pestis through either oral or subcutaneous route exhibited only mild clinical signs of short duration with no mortality [33]. The mechanism by which some species exhibit less severe disease from Y. pestis infection than others may be related to how well Y. pestis overcomes stress associated with intracellular parasitism of host macrophages. Although natural rodent hosts and canid rodent predators are infected via different routes, it appears likely that Y. pestis utilizes the same mechanism of evasion of the host innate immunity by intracellular parasitism of host macrophages and subsequent systemic dissemination in hosts which exhibit both high and low severity of disease.
During Y. pestis infection, macrophages can employ a wide spectrum of antimicrobial defenses including phagolysosomal acidification; enzymatic actions from cathepsins, lipase, nuclease and glycosidase; reactions of oxidizing agents and reactive oxygen species; and effects of cationic peptides, nitric oxide and reactive intermediates of nitric oxide. Furthermore, by the action of natural resistance-associated macrophage protein (NRAMP), many phagolysosomal metal ions such as iron, calcium, magnesium and manganese are sequestered, making the organelle very hostile for survival and growth of intracellular Y. pestis [34,35]. In order to adapt to and survive in this harsh intracellular environment, Y. pestis relies on various general stress regulators and other mechanisms to reduce the phagolysosomal antimicrobial activities [17,18,36,37].
Based on these observations, we hypothesized that differences in severity of disease from Y. pestis infection between high severity in rodents and less severe disease in dogs may be related to whether Y. pestis is able to overcome macrophage imposed stress during the intracellular parasitism phase of infection. In agreement with our hypothesis, Y. pestis in mouse macrophages exhibited morphological plasticity, survived during the entire experimental period of 27.5 h, mediated alternation in phagolysosomes, and caused cell lysis of the infected macrophages. However, in dog macrophages, intracellular Y. pestis either remained in a stressed state or were killed by infected macrophages without inflicting damage to these macrophages.

Intracellular parasitism of mouse and dog macrophages by Y. pestis
To examine the hypothesis that disease severity in different host species may be related to how well Y. pestis overcomes macrophage imposed stress, we examined the morphological changes of Y. pestis in primary macrophages derived from mouse spleen and dog peripheral blood. These macrophages were infected with Y. pestis strain KIM62053.1+, and bacterial and macrophage morphology observed by light microscopy and transmission electron microscopy (TEM) over a 27.5 h infection period for which the 0 time was the initiation of infection. In vitro Y. pestis infections of mouse macrophages have been previously reported [38], but infections of dog macrophages have not.
Mouse splenic macrophages were readily infected with Y. pestis with macrophages containing 18.0613.2 Y. pestis per infected macrophage at 2.5 h p.i. Intracellular Y. pestis were primarily coccobacilli housed in tight YCV ( Figs. 1 and 2). However, 6% of the intracellular Y. pestis exhibited filamentous bacterial morphology with a mean filament length of 6.8 mm and partial or no visible septation. At 7.5 h p.i., 82% of the mouse macrophages exhibited spacious YCVs and the number of Y. pestis per macrophage increased by 115%, but filamentous forms were not observed. At 27.5 h p.i., few splenic macrophages were observed, suggesting that macrophages lost viability between 7.5 and 27.5 h p.i. The splenic macrophages present at 27.5 h p.i. appeared larger with foamy cytoplasm, and intracellular Y. pestis had a bipolar rod appearance. Overall, microscopic detection of filamentous Y. pestis in mouse macrophages at the initial stage of infection may represent an adaptation of the bacterium to macrophage imposed stress. Subsequently by extending YCVs into spacious compartments at 7.5 h p.i., intracellular Y. pestis may be released from the macrophage imposed stress, return to normal coccobacillary bacterial morphology, and multiplied intracellularly. Y. pestis infected mouse macrophages lost viability by 27.5 h p.i., likely releasing viable and stress-adapted Y. pestis.
Dog peripheral blood derived macrophages were also readily infected with Y. pestis with macrophages containing 23.9613.9 Y. pestis per infected macrophage at 2.5 h p.i. Similar to mouse splenic macrophage infections, 5% of the intracellular Y. pestis in dog peripheral blood derived macrophages exhibited filamentous bacterial morphology with a mean filament length of 8.4 mm, and Y. pestis were contained within tight YCVs (Figs. 1 and 2). At 7.5 h p.i., the number of intact coccobacilli per macrophage increased by 38% from 2.5 h p.i., compatible with transition of filamentous Y. pestis to coccobacilli by cellular division, but unlike mouse macrophages the bacteria continued to be sequestered in tight YCVs. At 27.5 h p.i., intracellular Y. pestis appeared to be housed in YCV with double membranes in 80% of the infected macrophages; however, most of these intracellular Y. pestis were coccobacilli which exhibited irregular morphology consisting of vacuolation and uneven distribution of electron dense aggregates on TEM cross sections, compatible with killing of the bacteria by the dog macrophages. In contrast to the spatial extension of YCV in Y. pestis infected mouse macrophages, changes in dog macrophage morphology were restricted to increased foamy cytoplasm at 7.5 and 27.5 h p.i. These findings suggest that in dog macrophages, although intracellular Y. pestis exhibit similar stress morphology as in mouse splenic macrophages, Y. pestis failed to extend YCV at 7.5 h p.i., possibly facilitating subsequent killing of Y. pestis by these macrophages.    (Table 1). However, during this initial infection interval from 0 to 2.5 h, both RAW264.7 and DH82 cells exhibited a trend to lower Y. pestis CFUs but higher GEs per macrophage ( Fig. 3A and 3B), with the GE more closely reflecting the level of intracellular bacteria observed by microscopy in macrophages. Both Wright Giemsa stained light microscopic and TEM images at 2.5 h p.i., showed 7-8% of Y. pestis exhibited filamentous bacterial morphology (Figs. 4 and 5, Table 1), suggesting that the lower CFUs and accompanying higher GEs may in part reflect the filamentous growth of Y. pestis in both macrophage cell lines. This filamentous change was also evidenced under UV-fluorescent microscope in Y. pestis strain KIM6+ GFPuv infection of RAW264.7 and DH82 cells at 3 to 4 h p.i. (images not shown). The trend to lower CFUs but higher GEs may also reflect lower efficiency of microbial culture of stressed intracellular Y. pestis.
From 2.5 to 7.5 h p.i., there was a 19% increase in Y. pestis per RAW264.7 cells but a decline of 24% of Y. pestis per DH82 cells observed (Table 1). This correlated with a trend to higher CFUs with no change in GEs per macrophage for RAW264.7, but no change for either CFUs or GEs for DH82 cells (Fig. 3). The percent of filamentous Y. pestis remained about 5% in RAW264.7 cells but increased to 16% in DH82 cells (Table 1). The most remarkable change during this infection period was the increase of spacious YCV in infected RAW264.7 cells with 64% of RAW264.7 cells exhibiting spacious YCV at 7.5 h p.i. (Fig. 5, Table 1).
For RAW264.7 cells between 7.5 to 27.5 h p.i., the spacious YCV were maintained, the observed Y. pestis per RAW264.7 cell declined by 9% and filamentous Y. pestis declined from 5% to 2% (Table 1). However, the CFUs per macrophage increased significantly and became approximately equal to the GEs per macrophage, suggesting that Y. pestis which had been filamentous and/or not culturable at 7.5 h p.i. had undergone cell division to  Table 1). During this 7.5 to 27.5 h p.i. period in DH82 cells, CFUs, GEs and % filamentous Y. pestis remained unchanged, but at 27.5 h p.i., 29% of intracellular Y. pestis showed loss of viable bacterial morphology in TEM images (Fig. 5). These results suggest that Y. pestis in RAW264.7 cells experienced macrophage imposed stress as reflected by the filamentous bacterial morphology and non-culturability, but then overcame this stress coincident with vacuolar extension of YCV and return to coccobacillary morphology and culturability. But in DH82 cells, Y. pestis likely

Intracellular Y. pestis morphology in antibiotic-free media
Exposure of bacteria to gentamicin has the potential to induce aberrant bacterial morphology including filamentous forms [39]. To assess whether intracellular Y. pestis in macrophages exhibited filamentous morphology independent of gentamicin used for killing extracellular bacteria in the infection protocol, RAW264.7 and DH82 cells infected with Y. pestis were washed to remove extracellular Y. pestis rather than being treated with gentamicin. The washed infected RAW264.7 and DH82 cells were then incubated in Hank's Basic Salt Solution (HBSS) and 10% fetal bovine serum (FBS) to reduce growth of any residual extracellular Y. pestis, and the intracellular Y. pestis were observed by light microscopy for filamentous morphology. In the absence of gentamicin treatment, filamentous Y. pestis were observed at 2.5 and 5.0 h of infection in RAW264.7 and DH82 cells (Fig. 6), indicating that gentamicin was likely not responsible for the observed filamentous morphology of Y. pestis in macrophages.

Cytotoxicity of Y. pestis infected RAW264.7 and DH82 cells
As previously reported for mouse J774A.1 macrophage cell line, Y. pestis infection is associated with loss of macrophage viability [40]. Using the trypan blue dye exclusion method, RAW264.7 cells experienced an approximate 50% loss of cell viability, but only a 20% loss of cell viability was observed for Y. pestis infection of DH82 cells. To further characterize this loss of viability, lactate dehydrogenase leakage from infected macrophages was also determined. As shown in Figure 7, RAW264.7 cells infected with Y. pestis experience rapid cell lysis as indicated by specific LDH leakage during the initial 2.5 h p.i. period, and then slower, steady lysis through 27.5 h p.i. resulting in 45% lysis of infected cells as compared with uninfected control RAW264.7 cells exhibiting 0% cell lysis. In contrast, Y. pestis infection of DH82 cells exhibited cell lysis of approximately 10% for the entire infection period, which was not statistically significant from that for uninfected DH82 cells. These results show that Y. pestis infection of RAW264.7 cells causes cell lysis, which may release Y. pestis to the extracellular space in infected animals potentially initiating septicemic plague.  IntracellularYersina enterocolitica infection of mouse and dog macrophage cell lines We observed that Y. pestis overcame macrophage imposed intracellular stress in mouse but not dog macrophages. To assess whether this response was specific for Y. pestis compared to another, less pathogenic yersiniae, RAW264.7 and DH82 cells were infected with Yersinia enterocolitca. Y. pestis and Y. enterocolitca share similar genomic content and are both facultative intracellular pathogens [20,41]. As shown in Table 2, Y. enterocolitica infected both RAW264.7 and DH82 cells at similar levels of CFUs/macrophage, and this level of infection declined in both mouse and dog macrophages during the incubation period demonstrating no significant differences in intracellular parasitism of mouse versus dog macrophages in contrast to the difference in host macrophage infectivity observed for Y. pestis.

Discussion
Although Y. pestis infects a wide range of animal species, disease severity differs from species to species. Rodents are natural hosts and suffer severe disease with high mortality, but some rodent predators such as dogs and coyotes experience only mild or in apparent disease [1][2][3][4]33,42,43]. This variation in disease severity among susceptible hosts may be related to the extent to which Y. pestis overcomes the initial host macrophage mediated stress. The host macrophages are the primary cells used by Y. pestis to evade host innate immune mechanisms and to multiply and disseminate systemically during the early stage of infection [1,13,[15][16][17][18][19][20][21]27]. The goal of our experimental study was to better understand Y. pestis responses to infection of macrophages from mice and dogs, as hosts experiencing severe or mild disease, respectively. We hypothesized that Y. pestis in macrophages of severely affected hosts likely overcome the intracellular antimicrobial stresses and multiplies intracellularly, whereas, in macrophages of hosts exhibiting only mild disease, Y. pestis fails to overcome the intracellular defense mechanisms and are eventually killed by these macrophages.
In our study, we found that a fraction of intracellular Y. pestis in host macrophages assumes filamentous bacterial morphology with multiple genome copies per bacterium likely induced as a result of macrophage associated stress. This filamentous morphologic change was frequently noted during the early stage of infection in both primary and tissue culture macrophages of both mice and dogs (Figs. 1, 2, 4 and 5, Table 1). Similar filamentous stress response have been observed for uropathogenic E. coli in mouse urinary bladder epithelial cells, Legionella spp. in Vero cells, Mycobacterium tuberculosis in human macrophage cell line THP-1, and Salmonella enterica serovar Typhimurium in mouse bone marrow and RAW264.7 macrophages [44][45][46][47][48][49]. The antibiotic gentamicin used in our infection protocol is also known to induce  filamentous changes in exposed bacteria [29]. However, in our study, when gentamicin was omitted from the infection protocol, filamentous Y. pestis were still observed supporting intracellular stress as the cause rather than gentamicin. Y. pestis filamentous morphologic response in macrophages may be an adaptive strategy to cope with intracellular antibacterial defense mechanisms in addition to stress relievers such as general stress regulation and inhibition of acidification of YCV [17,36,38]. For Salmonella typhimurium, filamentous morphologic changes in primary mouse bone marrow macrophages or in RAW264.7 cells are associated with exposure to NADPH oxidase, reactive oxygen species, nitric oxide or proteases associated with phagosome, and furthermore, this morphologic change is positively influenced by IFN-gamma priming of macrophages or presence of cationic antimicrobial peptides during the infection [47,48,50]. In support of filamentous morphologic change as a specific adaptive response induced by intracellular stress, the Y. pestis SOS DNA repair response to DNA damage occurring intracellularly is also known to cause the bacterium to undergo filamentous morphologic change by interfering with cell division process as observed in other bacteria [46,51,52]. Presumably, the filamentous morphologic change observed for intracellular Y. pestis is an adaptive strategy to aid survive in the hostile intracellular environment; this structural alteration may prevent the passing of damaged genomic copies to the daughter cells, by extending the time frame for DNA repair [53,54].
In addition to assumption of a filamentous bacterial morphology, Y. pestis also cause spacious vacuolar extension of YCV in mouse splenic and tissue culture macrophages (Figs. 1, 2, 4 and 5, Table 1). This vacuolar extension is consistent with previous report in J774A.1 macrophages infected with Y. pestis KIM6 in which YCVs were actively extended by the intracellular Y. pestis, and this extension process was dependent on presence of intact copies of PhoP-PhoQ transcriptional regulator [17]. Although the precise molecular mechanism governing this process is unknown, Y. pestis mediated extension of YCV likely benefits the bacterium by dilution of phagolysosomal content thereby reducing their antimicrobial activity and associated macrophage imposed stress.
Near coincidently with Y. pestis initiated spacious vacuolar extension of YCV in mouse macrophages, the filamentous Y. pestis reverted to coccobacillary bacterial morphology (Figs.1, 2, 4 and 5, Table 1). For mouse splenic macrophages, by 7.5 h p.i., the majority of intracellular Y. pestis converted back to coccobacilli, which were loosely confined within the spacious YCV. Reversion of Y. pestis from filamentous to coccobacillary morphology in mouse macrophages may reflect release of bacteria from intracellular stress. The molecular machinery which mediates the transition from filamentous to coccobacillary morphology is largely unknown; nevertheless, products of Y. pestis genes y2313, y2315 and y2316 may play a role in this process. These genes are expressed in mouse macrophage cell line J774A.1 at 4 h p.i, and mutational inactivation was associated with retention of filamentous intracellular Y. pestis for at least 24 h of infection [55].
In contrast to YCV changes in mouse macrophages, Y. pestis in dog macrophages did not induce vacuolar extension of YCVs, and Y. pestis retained its filamentous morphology throughout 27.5 h of infection in dog DH82 macrophages (Fig. 3, 4, and 5, Table 1). This suggests that Y. pestis in DH82 cells may not efficiently overcome intracellular stress, and consequently, the bacterium is susceptible to macrophage killing as evidenced by presence of disintegrating coccobacilli in dog YCV at 27.5 h p.i. (Fig. 5). In dog peripheral blood macrophages, Y. pestis did convert from filamentous forms to coccobacilli at 7.5 h p.i. but failed to modify YCV and remained in tight YCV, suggesting that intracellular Y. pestis are still controlled by macrophage defense mechanisms ( Figs. 1 and 2). These coccobacilli eventually died within YCVs due to macrophage killing. This failure of Y. pestis to overcome macrophage imposed stress during the initial intracellular parasitism phase of plague may result in less severe disease in dogs and coyotes [1,33,42,43].
Apart from the vacuolar extension of YCV, mouse and dog tissue culture macrophages markedly differed in the loss of macrophage viability from Y. pestis infection for the period of 27.5 h p.i. The infected RAW264.7 cells exhibited 45% cell lysis in contrast to approximately 10% for DH82 cells (Fig. 6). The high percentage of LDH leakage in RAW264.7 cells may be associated with spacious extension of YCV leading to lysis of infected macrophages. Furthermore, this finding in RAW264.7 cells agrees with the prevailing speculation that cell lysis is the primary mean of Y. pestis release from the infected macrophages [20]. In contrast, restriction of intracellular Y. pestis to the stress related filamentous morphology in tight, unmodified YCV for the entire infection may account for the failure of Y. pestis to induce lysis of DH82 cells. Similar observations have been made for filamentous Burkholderia pseudomallei in THP-1 cells [56].
In conclusion, our results demonstrate that Y. pestis in host macrophages assume filamentous morphology during initial stage of exposure to intracellular stress likely as a morphologic adaptation used by the bacterium to prolong its survivability under harsh conditions [54]. These filamentous Y. pestis in mouse macrophages eventually returned to coccobacillary morphology, which was associated with active vacuolar extension of YCV and subsequent macrophage cell lysis. In contrast, in dog macrophages, Y. pestis was either restricted to the stress-induced filamentous morphology for an extended period of time and then were killed by the macrophages or reverted to coccobacillary morphology but were eventually killed by the macrophages likely due to failure of the bacterium to induce spacious extension of YCV. Therefore, greater severity of infection and fatal outcome of Y. pestis infections in rodents and other highly susceptible hosts may be the consequence of Y. pestis overcoming macrophage associated stress during the initial phase of infection, whereas less severe disease in some canid rodent predators may be associated with macrophage restriction of the Y. pestis stress responsiveness during the intracellular parasitism phase of Y. pestis infection.

Materials and Methods
Bacterial strains and culture conditions Y. pestis strain KIM62053.1+ hms + psn + psa 2 (Dpsa2053.1) ybt + lcr 2 derived from KIM62053.1+ (KIM6+) was kindly provided by Dr. Robert Perry, University of Kentucky and used in all experiments except where noted [57]. For fluorescent imaging experiments, KIM62053.1+ was transformed with a modified green fluorescent protein expression plasmid (pGFPuv, Clontech, Mountain View, CA) by electroporation as described elsewhere [58,59]. Y. enterocolotica strain OADDL 91 was obtained from the Oklahoma State University College of Veterinary Medicine infectious diseases laboratory collection. For both Yersinia species, isolated colonies on Brain Heart Infusion (BHI) (Difco, Becton Dickinson Company, Franklin Lakes, NJ) agar plates grown for 36 h at 26uC were inoculated into BHI broth (Difco) and cultured overnight at 26uC with 160 rpm shaking.

Isolation of mouse splenic macrophages
Splenic macrophages were used instead of blood macrophages due to the difficulty of collecting sufficiently large volumes of blood from mice to isolate the required numbers of macrophages for experimentation. Splenic macrophages were isolated from 7 to 11 week old, female C57BL/6J mice (The Jackson Laboratory, Bar Harbor, Maine). Briefly, spleens from mice euthanized with 70% CO 2 under an Oklahoma State University Institutional Animal Care and Use Committee approved protocol were collected under sterile conditions in 10-15 mL of Dulbecco's Modified Eagle's Medium (DMEM) (Invitrogen, Grand Island, NY) supplemented with 20% fetal bovine serum (FBS) (Hyclone laboratories, Logan, UT) and 50 mg/mL of gentamicin sulfate (Sigma-Aldrich, St Louis, MO), and therein the same media, spleens were thoroughly minced using a sterile scalpel blade. The resulting cell suspension was passed successively through sterile nylon mesh of 160, 75, 15 mm sizes to yield tissue debris-free homogenous cell suspensions, and thereafter, these cell suspensions were adjusted to 2610 6 viable cells/mL in the same media. Finally, 6 mL aliquots of the cell suspensions were cultured overnight in 25 cm 2 , poly-D-lysine coated tissue culture flasks (Becton Dickinson Labware, Franklin Lakes, NJ) at 37uC with 5% CO 2 , followed by removal of nonadherent cells by washing with sterile phosphate buffered saline (PBS). The attached cells were further cultured for 3 days in fresh DMEM, FBS, and gentamicin media as above. Subsequently, the cell layers were changed into RPMI-1640 media (Sigma-Aldrich) supplemented with 20% FBS, 2 g/L sodium bicarbonate and no antibiotic for 2 days prior to use.

Isolation of dog peripheral blood derived macrophages
Blood was collected from healthy adult dogs via venipuncture under an Oklahoma State University Institutional Animal Care and Use Committee approved protocol and anticoagulated with 10% sodium citrate (Sigma-Aldrich). Anticoagulant peripheral blood was diluted 2-fold with PBS containing 50 mg/mL gentamicin, and the diluted blood was gently overlaid on sterile HistopaqueH-1083 (Sigma-Aldrich) at the ratio of 2:1 and centrifuged at 8006 g for 15 min at room temperature. From the multilayer separation, the buffycoat was collected by aspiration and washed thrice with sterile PBS containing 50 mg/mL gentamicin, and cells collected by centrifugation at 2506 g for 10 min at room temperature. The resulting cell pellets were resuspended in DMEM with 20% FBS and 50 mg/mL gentamicin at concentration of 2610 6 viable cells/mL, and then 6 mL aliquots were cultured overnight at 37uC with 5% CO 2 in 25 cm 2 poly-D-lysine coated flask, followed by removal of non-adherent cells by washing with sterile PBS. The cell layers were further cultured for 3 days in fresh DMEM, FBS, and gentamicin media as above. Subsequently, gentamicin was removed and the cells were kept in RPMI-1640 media with 20% FBS with no antibiotic for another 2 days prior to use.

Tissue culture cells and growth conditions
Mouse macrophage cell line RAW264.7 was provided by Dr. Guolong Zhang, Department of Animal Science, Oklahoma State University (ATCC, Manassas, VA; [60]), and dog macrophage cell line DH82 [61] was provided by Dr. Susan E. Little, Department of Veterinary Pathobiology, Oklahoma State University. Both cell lines were cultured at 37uC with 5% CO 2 tension in RPMI-1640 media with 10% FBS.

Infection of primary macrophage cells
For infection of primary macrophage cells, Y. pestis strain KIM6-2053.1+ was grown in BHI broth at 26uC as described above, and the inocula quantified by OD 600 nm as compared with a standard curve for CFU/OD 600 nm . At the time of infection, aliquots of 6 mL of inocula containing 4.0-5.4610 6 Y. pestis in RPMI-1640 with 10% FBS media were added to each 25 cm 2 flask containing 0.8-1.1610 6 viable macrophages as trypan blue exclusion assay carried out on the corresponding replica samples, yielding a multiplicity of infection of 5:1. To initiate the infection at zero time, the flasks were centrifuged at 8006 g for 3 min to enhance macrophage-bacteria contact and then incubated at 37uC under 5% CO 2 concentration for 30 min. Subsequently, adherent macrophages were washed gently thrice with 6 to 7 mL of sterile PBS and treated with 6 mL of RPMI-1640 with 10% FBS media containing 50 mg/mL gentamicin for 2 h under the same incubation conditions to kill extracellular bacteria. At the end of gentamicin exposure, macrophages were once again washed gently thrice with 6 to 7 mL of PBS and then 6 mL of antibiotic-free RPMI-1640 with 10% FBS media were added. From the resulting infection, samples were collected for light and transmission electron microscopic studies at various post-infection intervals. At 2.5 h p.i., samples were collected following removal of the gentamicin media and PBS washes by treating with 0.05% trypsin-EDTA (Mediatech Inc., Manassas, VA) at 37uC for 1 to 3 min; however, at 7.5 and 27.5 h p.i., 2 h prior to sampling, the infected adherent macrophages were treated with RPMI-1640 with 10% FBS media and 50 mg/mL of gentamicin for 2 h to kill any extracellular Y. pestis.

Infection of tissue culture cells
To prepare tissue culture cells for infection, RAW264.7 and DH82 cell monolayers grown in 75 cm 2 tissue culture flasks were released into the fresh RPMI-1640 with 10% FBS media using cell scraper (BD Falcon, Biosciences Discovery Labware, Bedford, MA) and number of cells per unit volume were counted using hemocytometer. Prior to use in assays, tissue culture cell viability was assessed by trypan blue exclusion assay with .95% viability required for subsequent use. Infection of RAW264.7 and DH82 macrophage cell lines with Y. pestis KIM6+ or Y. enterocolitica were carried out in 96-well flat-bottom plates (FALCON flat bottom polystyrene plate, Becton Dickinson Company). Prior to infection, 100 mL of cell suspensions containing 1610 6 RAW264.7 viable cells/mL of RPMI-1640 with 10% FBS were sub-cultured in wells A1 to A3 of the top row of 96-well flat-bottom tissue culture plates and in the same way DH82 cells were sub-cultured in wells A7 to A9. These plates were incubated at 37uC with 5% CO 2 concentration for 12 to 16 h. For infection, 100 mL of RPMI-1640 with 10% FBS media containing 5610 6 Y. pestis KIM6+ or Y. enterocolitica cultured overnight at 26uC was added to each well of A1-3 and A7-9 of all plates yielding a MOI of 5:1. The resulting infection was carried out at 37uC with 5% CO 2 for 30 min and then treated with 50 mg/mL gentamicin for Y. pestis or 100 mg/mL gentamicin for Y. enterocolitica for 2 h as above mentioned for primary macrophage infections. At 0 h p.i., replica of inocula and tissue culture cells with which the infection had started were taken as such for analysis. For 0.5 and 2.5 h p.i., samples were obtained immediately before adding gentamicin and antibiotic-free media, respectively. At 1.25 and 2.0 h p.i., samples were obtained after removal of the gentamicin media and washing thrice with PBS. Prior to 5.0, 7.5, 12.5, 18.5 and 27.5 h p.i. sampling, the tissue layers were treated for a second time with RPMI-1640 with 10% FBS media contained 50 mg/mL of gentamicin for Y. pestis or 100 mg/mL gentamicin for Y. enterocolitica for 2 h as mentioned above.

Infection of tissue culture cells in antibiotic-free media
To determine whether Y. pestis intracellular filamentous morphologic change occurred in the absence of gentamicin, RAW264.7 and DH82 cells were infected for 30 min as described above, but instead of treating with gentamicin to kill extracellular bacteria, the adherent macrophages were washed three times with 100 mL of PBS to remove extracellular bacteria and 100 mL of HBSS containing 10% FBS was added to wells. Infected macrophages were then incubated for 2 or 4.5 h and the morphology of intracellular Y. pestis observed by light microscopy.

Determination of colony forming units
CFUs were determined using a novel high-throughput assay developed by Nizet and colleagues and modified for use with Y. pestis in which infected tissue culture cells in 96-well plates are lyzed, serial dilution, and microcolonies of released intracellular bacteria are grown in soft agar in the 96-well plates and counted to determine the CFUs [59,62]. At the time of sampling, media from wells was aspirated, adherent tissue culture cells washed gently thrice with 100 mL of PBS, and the final wash transferred laterally to wells in the A row of the plate. Following washing and transfer of the final wash, 100 mL of 0.1% Triton X-100 in sterile PBS was added to wells A1-3 and A7-9, and the plates were incubated for 10 min at 37uC with 160 rpm on an oscillating shaker to allow lysis and release of intracellular Y. pestis into the media. Eighty mL of PBS were added to wells B1 to H12, and the cell lysates and wash solutions in row A were serially diluted 5-fold in PBS by transferring 20 mL from each row to the next from row A to row H. Subsequently, 80 mL of bacterial suspension left in each well of 96-well plate was overlaid gently with 120 mL of 0.83% Bacto-agar (Becton Dickinson, Franklin Lakes, NJ) maintained at 45 to 48uC prior to dispensing. Finally, the plates were incubated at 26uC for 16 to 24 h, and the resulting growth of micro-colonies counted using an inverted light microscope. Before adopting this microplate based colony counting, accuracy of the method was compared with conventional CFU determinations using a 10 cm agar plate method, which is described in detail by Wendte et. al., 2011 [59].

Determination of genomic equivalents
To determine the total number of Y. pestis comprised of culturable, viable but non-culturable, and dead bacteria in samples at various post-infection time points, GEs were determined using PCR. A 90 bp fragment from the single copy Y. pestis specific 'fur' gene was amplified using forward primer 59-TCT GGA AGT GTT GCA AAA TCC TG-39 and reverse primer, 59-AAG CCA ATC TCT TCA CCA ATA TCG-39. PCR reactions were carried out with 300 nM of each primer in 20 mL volume using the Fast SYBR green master mix (Applied Biosystems, Carlsbad, CA) according to the manufacture's instruction (initial enzyme activation at 95uC for 20 sec, followed by 40 cycles of denaturation at 95uC for 3 sec and annealing/extension at 60uC for 30 sec). Standard curves of threshold cycle (ct) values versus bacterial number were prepared by spiking known CFUs of Y. pestis strain KIM6+ into samples containing 1610 6 RAW264.7 or DH82 cells/mL. The assay was initiated immediately after adding the bacteria into the cell suspension by first heat bursting at 95uC for 10 min, and after centrifugation at 1,0006g for 1 min, 2 mL of supernatant was used to determine the ct-value using a 7500 Fast Real-Time PCR system (Applied Biosystems, Carlsbad, CA). For samples containing infected macrophages, the assay was initiated by heat burst, and the resulting samples were assayed as described above.

Determination of infected macrophage counts
To determine the number of macrophages at a particular infection interval, counts of macrophages were conducted. At each sampling period, infected RAW264.7 and DH82 cells in wells of replica plates were released by gentle pipetting and dilution in 0.4% trypan blue-PBS dye. Using a hemocytometer and microscope, the number of viable and non-viable macrophages was determined.

Determination of the percent infectivity
At each infection interval, samples of Y. pestis infected RAW264.7 and DH82 cells were cytospun (Statspin Cytofuge, Norwood, MA) onto glass microscope slides at low speed for 5 min and stained with Wright Giemsa stain. Using a light microscope at a 1,0006 magnification, approximately 400 macrophages per sample were examined to count the number of macrophages which had at least one detectable intracellular bacterium, and from these values, the percent infectivity was calculated.

Calculation of CFUs and GEs per macrophage
At each infection interval, number of CFUs and GEs per macrophage was calculated from the CFUs, GEs and macrophage number per mL and percent infectivity of the corresponding interval using the formulas: CFUs per macrophage~C FUs=mL Total macrophages=mL |

Fractional percent infectivity
GEs per macrophage~G Es=mL Total macrophages=mL |

Fractional percent infectivity
Here, total macrophages/mL represents the sum of trypan blue positive and negative cells per mL at each infection interval.

Macrophage cytotoxicity assay
Replica samples of RAW264.7 and DH82 cells infected with Y. pestis were used for the cytotoxicity assay. At each infection interval, the activity of the eukaryotic cytoplasmic enzyme lactate dehydrogenase (LDH) in the extracellular media was measured as an indicator of macrophage cell lysis. The LDH activity was measured using the CytoTox-ONE TM Homogenous Membrane Integrity Assay kit (Promega, Madison, WI) in a POLAR star OPTIMA (BMG Labtechnologies Inc., Cary, NC) at excitation and emission of 560 and 590 nm, respectively. From these values, percent cytotoxicity was calculated using the formula: % CytotoxicityL DH release from infected cells{release from uninfected cells LDH release from positive control{release from uninfected cells

|100
Positive control release is the maximum LDH release from uninfected macrophages induced by cell lysis with 0.1% Triton X-100.

Microscopic studies
Morphological features of intracellular Y. pestis and of infected macrophages were examined using microscopy.
Light microscopy. Y. pestis infected macrophage cell lines, RAW264.7 and DH82, and primary macrophage isolates from mouse spleen and dog peripheral blood were spun onto microscopic slides at slow speed for 5 min using a cytocentrifuge (Statspin Cytofuge), and the slides stained with Wright Giemsa stain. The stained slides were examined at 1,0006 magnification using light microscopy.
Transmission electron microscopy. Primary macrophage isolates from mouse spleen and dog peripheral blood and tissue culture cell lines RAW264.7 and DH82 were infected as described above and sampled at 2.5, 7.5 and 27.5 h p.i. During sampling, infected primary macrophage isolates were released into RPMI-1640 with 10% FBS media at the intended sampling intervals by using trypsin as explained elsewhere, then collected at 2506 g for 10 min, and finally fixed with 2.5% (vol/vol) glutaraldehyde in PBS for 1 h at room temperature. After fixing, the cells were changed into PBS as noted above and kept at 4uC until processing. In case of tissue culture macrophages, infected cells in 25 cm 2 flasks were fixed therein for 15 min on ice with 2.5% (vol/vol) glutaraldehyde in PBS. Subsequently, the fixed cells were transferred using a CytoOneH cell scraper (USA Scientific Inc., Ocala, FL) into the above noted glutaraldehyde solution to fix further for 45 min at room temperature. Thereafter, the fixed cells were collected by centrifuging at 2506 g for 10 min at room temperature, suspended in PBS and maintained at 4uC for further processing. For this processing, cells were postfixed with 1% osmium tetroxide (Polysciences Inc., Warrington, PA) in PBS for 1 h. Subsequently, cells washed thrice with PBS were en bloc stained with 1% aqueous uranyl acetate (Ted Pella Inc., Redding, CA) for 1 h, dehydrated in ascending grades of ethanol and then embedded in Eponate resin (Ted Pella Inc.). Ultra-thin sections of 90 nm were prepared using a Leica Ultracut UCT ultramicrotome (Leica Microsystems Inc., Bannock-burn, IL) and stained with uranyl acetate and lead citrate (Sigma-Aldrich). Finally these sections were examined in a JEOL 1,200 EX transmission electron microscopy (JEOL USA Inc., Peabody, MA).

Morphometric analysis of TEM images
Electron microscopic images were assessed using gimpH photoediting software version 2.6 (http://www.gimp.org/). Analysis was conducted on 9 or more randomly selected single cell images for each infected cell type and infection interval. Assessment of filamentous bacterial morphology was based on quantitative comparison of the 1 mm measure bar on images to determine the bacterial length. Bacteria longer than 5 mm, double the maximum length of normal Y. pestis (0.5-0.861-2.5 mm), were considered to be filamentous Y. pestis. Bacteria exhibiting normal morphology without intracellular vacuolation or irregular electron dense granulation were scored as intact Y. pestis. Intact and filamentous Y. pestis per infected RAW264.7 or DH82 cell were quantified based on these criteria. For the infected macrophages, the percentage of cells which had spacious extension of YCV was quantified.

Statistical analysis
To determine statistical significance for comparison of two means, Student's t-test was used. For comparisons of more than two means, one-way ANOVA and subsequent 'Tukey HSD' post hoc methods were employed. The resulting analysis was reported as significant difference at p-value 0.01 or 0.05 for a given comparison.