Type 3 secretion system induced leukotriene B4 synthesis by leukocytes is actively inhibited by Yersinia pestis to evade early immune recognition

Subverting the host immune response to inhibit inflammation is a key virulence strategy of Yersinia pestis. The inflammatory cascade is tightly controlled via the sequential action of lipid and protein mediators of inflammation. Because delayed inflammation is essential for Y. pestis to cause lethal infection, defining the Y. pestis mechanisms to manipulate the inflammatory cascade is necessary to understand this pathogen’s virulence. While previous studies have established that Y. pestis actively inhibits the expression of host proteins that mediate inflammation, there is currently a gap in our understanding of the inflammatory lipid mediator response during plague. Here we used the murine model to define the kinetics of the synthesis of leukotriene B4 (LTB4), a pro-inflammatory lipid chemoattractant and immune cell activator, within the lungs during pneumonic plague. Furthermore, we demonstrated that exogenous administration of LTB4 prior to infection limited bacterial proliferation, suggesting that the absence of LTB4 synthesis during plague contributes to Y. pestis immune evasion. Using primary leukocytes from mice and humans further revealed that Y. pestis actively inhibits the synthesis of LTB4. Finally, using Y. pestis mutants in the Ysc type 3 secretion system (T3SS) and Yersinia outer protein (Yop) effectors, we demonstrate that leukocytes recognize the T3SS to initiate the rapid synthesis of LTB4. However, several Yop effectors secreted through the T3SS effectively inhibit this host response. Together, these data demonstrate that Y. pestis actively inhibits the synthesis of the inflammatory lipid LTB4 contributing to the delay in the inflammatory cascade required for rapid recruitment of leukocytes to sites of infection.


Introduction
Yersinia pestis causes the human disease known as the plague.Although typically characterized as a disease of our past, in the aftermath of the 3 rd plague pandemic, Y. pestis became endemic in rodent populations in several countries, increasing the potential for spillover into human populations through contact with infected animals and fleas [1][2][3].Human plague manifests in three forms: bubonic, septicemic, or pneumonic plague.Bubonic plague resulting from flea transmission arises when bacteria colonize and replicate within lymph nodes.Septicemic plague results when Y. pestis gains access to the bloodstream, either directly from a flea bite or via dissemination from an infected lymph node, and results in uncontrolled bacterial replication and sepsis.Finally, secondary pneumonic plague, wherein Y. pestis disseminates to the lungs via the blood, results in a pneumonia that can promote direct person-to-person transmission via aerosols.While treatable with antibiotics, if left untreated, all forms of plague are associated with high mortality rates, and the probability of successful treatment decreases the longer initiation of treatment is delayed post-exposure [3][4][5][6].Regardless of the route of infection, one of the key virulence determinants for Y. pestis to colonize the host is the Ysc type 3 secretion system (T3SS) encoded on the pCD1 plasmid [5,7].This secretion system allows direct translocation of bacterial effector proteins, called Yops, into host cells [5,8,9].The Yop effectors target specific host factors to disrupt normal host cell signaling pathways and functions [10][11][12][13][14][15].Because the T3SS and Yops are required for mammalian but not flea infection, the expression of the genes encoding these virulence factors are differentially expressed within these two hosts [5,8,16,17].The primary signal leading to T3SS and Yop expression is a shift in temperature from that of the flea vector (<28˚C) to that of the mammalian host (>30˚C).During mammalian infection, Y. pestis primarily targets neutrophils and macrophages for T3SSmediated injection of the Yop effectors [18][19][20].The outcomes of Yop injection into these cells include inhibition of phagocytosis, reactive oxygen species (ROS) synthesis, degranulation by neutrophils, and inflammatory cytokine and chemokine release required to recruit circulating

PLOS PATHOGENS
Inhibition of leukotriene B4 by Yersinia pestis neutrophils to infection sites [21][22][23][24][25][26].Importantly, previous work suggests that inhibition of neutrophil influx and establishing a non-inflammatory environment is crucial for Y. pestis virulence [27,28].Therefore, defining the molecular mechanisms used by Y. pestis to subvert the host immune response is fundamental to understanding the pathogenesis of this organism.Moreover, defining the host mechanisms targeted by Y. pestis to inhibit inflammation can also provide novel insights into how the host responds to bacterial pathogens to control infection.
A cascade of events tightly regulates inflammation to ensure rapid responses to control infection and effective immune resolution after clearance of pathogens to limit tissue damage [29,30].This inflammatory cascade is initiated by synthesizing potent lipid mediators and is sustained and amplified by the subsequent production of protein mediators [31,32].Polyunsaturated fatty acid (PUFA)-derived lipid mediators are potent modulators of the innate and adaptive immune responses [30,33].Of these, the eicosanoids, including the leukotrienes and the prostaglandins, are key regulators of the inflammatory cascade during infection [31,32].Leukotriene B4 (LTB 4 ) is rapidly synthesized from arachidonic acid upon activation of 5-lipoxygenase (5-LOX), cytosolic phospholipase A 2 (cPLA2), 5-LOX activating protein (FLAP), and LTA 4 hydrolase (Fig 1A) [34].Upon synthesis and release, LTB 4 is recognized by the high affinity BLT1 receptor on immune cells to promote chemotaxis and initiate the inflammatory cascade leading to production of pro-inflammatory cytokines and chemokines [31,32,[35][36][37][38][39].Together these inflammatory mediators promote the recruitment of circulating leukocytes to infected tissue [32].Importantly, because of its critical role in initiating the inflammatory cascade, disruption in the timely production of LTB 4 can slow the subsequent downstream release of cytokines and chemokines and the ability of the host to mount a rapid inflammatory response required to control infection.
Despite active proliferation of Y. pestis within the lungs in the mouse model, there appears to be an absence of pro-inflammatory cytokines, chemokines, and neutrophil influx for the first 36 hours of primary pneumonic plague [11][12][13][14][15].This phenotype dramatically differs from pulmonary infection with attenuated mutants of Y. pestis lacking the T3SS or Yop effectors or by other pulmonary pathogens, such as Klebsiella pneumoniae, which induce significant inflammation within 24 hours of bacterial exposure [11][12][13][14][15]. Surprisingly, despite the importance of lipid mediators in initiating the inflammatory cascade, the role of inflammatory lipids during plague has not been previously investigated.However, using human peripheral blood neutrophils, Pulsifer et al. previously demonstrated that Y. pestis can actively inhibit the synthesis of LTB 4 in vitro in a T3SS/Yop-dependent manner, suggesting that LTB 4 synthesis may be inhibited during plague [26].In this study, we expand on these observations by investigating LTB 4 synthesis by the mammalian host in response to Y. pestis.Using the murine model of plague, we demonstrate dysregulation in the production of LTB 4 by Y. pestis and provide the lipidomic profile of other host inflammatory lipids during the initial 48 h of pneumonic plague.We further show that exogenous treatment with LTB 4 inhibits bacterial proliferation in the murine model.Using Y. pestis mutants, we also discovered that leukocyte interactions with the Y. pestis T3SS triggers LTB 4 synthesis, but synthesis is inhibited by multiple Yop effectors secreted via the same T3SS.Together, these data suggest that modulation in the production of host inflammatory lipids is an additional virulence mechanism used by Y. pestis to inhibit the rapid recruitment of immune cells needed to control infection.

LTB 4 synthesis is delayed during pneumonic plague
Based on our previous observations that Y. pestis inhibits LTB 4 synthesis by human neutrophils [26], we sought to determine if LTB 4 was synthesized during infection using the murine

PLOS PATHOGENS
Inhibition of leukotriene B4 by Yersinia pestis

PLOS PATHOGENS
Inhibition of leukotriene B4 by Yersinia pestis model of pneumonic plague.C57BL/6J mice were intranasally infected with Y. pestis KIM5+ and LTB 4 was measured in the lungs during the non-inflammatory stage of disease (6, 12, and 24 h post-infection).We did not observe a statistically significant increase in LTB 4 synthesis at any time point, with only 3 of the 15 samples having elevated LTB 4 concentrations over the entire 24 h period (Fig 1B).Moreover, we did not observe a significant increase in 20-hydroxy LTB 4 (Fig 1C), which is the direct degradation product of LTB 4 (Fig 1A).To confirm these results, LTB 4 was measured in the lungs from a second independent group of C57BL/6J mice, this time expanding the analysis to include the pro-inflammatory stage of disease (36 and 48 h post-infection).Again, we did not observe statistically significant increases in LTB 4 or 20-hydroxy LTB 4 during the first 24 h of infection (Fig 1D and 1E).However, by 36 h post-infection, both lipids were statistically elevated compared to uninfected samples (p � 0.05).Moreover, we observed a significant increase in 5-HETE as early as 6 h post-infection (Fig 1F; p � 0.01), which can result if 5-LOX does not complete the synthesis of LTA 4 from arachidonic acid [40,41].While the synthesis of LTB 4 appears absent during the first 24 h of infection, the synthesis of other inflammatory lipids increased during the same period, including the prostaglandins, which are another group of eicosanoids whose synthesis is regulated via the cyclooxygenase pathway (Fig 1G -1I).Globally, we observed significant changes in the synthesis of 63 lipids during pneumonic plague, including lipids generally considered to be pro-inflammatory (18 lipids), anti-inflammatory (41 lipids), or pro-resolving (4 lipids) (S1 Table) [42,43].Together these data indicate LTB 4 synthesis is delayed during pneumonic plague.
BLT1 -/-mice are not more susceptible to pneumonic plague than C57BL/6J mice LTB 4 is recognized by the high-affinity G-protein coupled receptor BLT1, which is expressed primarily by innate and adaptive immune cells [38,44].LTB 4 -BLT1 engagement leads to host inflammatory immune responses such as chemotaxis, cytokine release, phagocytosis, and ROS production that contribute to the clearance of pathogens [35,45].Mice deficient in the expression of BLT1 cannot effectively respond to LTB 4 signaling and are generally more susceptible to infection [39,46,47].Because we did not observe LTB 4 synthesis during the early stages of pneumonic plague, we hypothesized that BLT1 -/-mice would not be more susceptible to Y. pestis infection.To test this hypothesis, we intranasally infected C57BL/6J and BLT1 -/-mice with Y. pestis KIM5+ and measured bacterial numbers in the lungs at 12 and 24 h post-infection.Bacterial numbers were not significantly higher in BLT1 -/-mice than C57BL/6J mice (Fig 2A).Furthermore, independent experiments with a Y. pestis strain with a luciferase bioreporter (Y.pestis CO92 Lux pcysZK ), which allows for monitoring bacterial proliferation via optical imaging and host survival in the same group [48], showed no significant differences between the two mouse lines in bacterial proliferation at later time points or in the mean-time to death (Fig 2B and 2C).These data indicate that the loss of LTB 4 -BLT1 signaling in BLT1 -/-mice does not impact the infectivity of Y. pestis, further supporting that LTB 4 synthesis and signaling is disrupted during pneumonic plague.

Exogenous LTB 4 treatment limits Y. pestis proliferation in vivo
Because LTB 4 synthesis and signaling appears to be disrupted during infection, we next asked if exogenous administration of LTB 4 could alter infection.To test this hypothesis, we used a previously described peritoneal model that allows for accurate administration of LTB 4 and easy recovery of both elicited leukocytes and bacteria via lavage [49].As described previously, intraperitoneal administration of LTB 4 resulted in an increase in the neutrophil population within the peritoneal cavity in C57BL/6J mice as early as 1 h post-administration (Figs 3A and S1) [49].When challenged intraperitoneally with Y. pestis KIM5+ after LTB 4 administration, we observed a significant decrease in the number of viable bacteria recovered from LTB 4treated animals, approaching the limit of detection, as compared to PBS-treated animals 3 h p � 0.05).Moreover, bacterial clearance was dependent on LTB 4 signaling, as LTB 4 treatment of BLT1 -/-mice did not alter bacterial or neutrophil numbers at 3 h post-infection compared to PBS-treated C57BL/6J mice (Fig 3B and 3C).Together, these data indicate that LTB 4 -mediated recruitment and activation of leukocytes can improve the host response to Y. pestis.

Neutrophils do not synthesize LTB 4 in response to Y. pestis
Because neutrophils are robust sources of LTB 4 [35], are the primary cells with which Y. pestis interacts during the first 24 h of pneumonic plague [19], and Y. pestis inhibits LTB 4 synthesis by human neutrophils [26], we next sought to determine if the LTB 4 response by neutrophils differed between Y. pestis and other bacteria.When bone marrow-derived neutrophils (BMNs) from C57BL/6J mice were stimulated with E. coli, S. enterica Typhimurium, or a K. pneumoniae manC mutant (unable to synthesize a capsule), LTB 4 synthesis was significantly induced within 1 h of infection (Fig 4A; p � 0.0001).However, infection with Y. pestis did not elicit LTB 4 synthesis, even when the MOI was increased to 100 bacteria per neutrophil (Fig 4B).Similar phenotypes were observed during infection of human peripheral blood neutrophils (hPMNs), recapitulating our previously published data for Y. pestis (Fig 4C and 4D) [26].Importantly, the absence of LTB 4 synthesis did not appear to be due to Y. pestis induced cell death, as no significant changes in cell permeability or cytotoxicity were observed during Y. pestis infections at an MOI of 20 when compared to uninfected neutrophils (S2A and S2B

Y. pestis actively inhibits LTB 4 synthesis
Seven Yop effectors are secreted via the T3SS [21][22][23][24], and Pulsifer et al. previously showed Yop effector-mediated inhibition of LTB 4 synthesis in human neutrophils by YpkA, YopE, YopJ, YopH, and YopT at an MOI of 100 [26].However, Yop inhibition of LTB 4 synthesis by murine neutrophils has not been previously investigated, nor whether the same Yop effectors are sufficient to inhibit LTB 4 synthesis at a lower MOI.Therefore, murine and human neutrophils were infected at an MOI of 20 with a Y. pestis mutant strain that expresses the T3SS but lacks all seven Yop effectors (Y.pestis T3E) [50].In contrast to Y. pestis infected cells, we observed a significant increase in LTB 4 synthesis in response to the Y. pestis T3E strain, indicating that the Yop effectors inhibit synthesis (Fig 5A and 5B; p � 0.0001).Moreover, when neutrophils were simultaneously infected with Y. pestis and the Y. pestis T3E mutant or Y. pestis and the K. pneumoniae manC mutant, LTB 4 levels were significantly lower than Y. pestis T3E or K. pneumoniae manC only infections (Fig 5C -5F; p � 0.0001).To determine if individual Yop effectors were sufficient to inhibit synthesis, murine neutrophils were infected with Y. pestis strains that expressed only one Yop effector [50].LTB 4 synthesis was significantly decreased if Y. pestis expressed YpkA, YopE, YopH, or YopJ, and an intermediate phenotype was observed during infection with a strain expressing YopT (Fig 5G).These phenotypes recapitulated those previously reported for human neutrophils [26].Together these data confirm that Y. pestis is not simply evading immune recognition but is actively inhibiting LTB 4 synthesis via the activity of multiple Yop effectors.

Neutrophils synthesize LTB 4 in response to the Y. pestis T3SS in the absence of the Yop effectors
Components of the T3SS are pathogen-associated molecular patterns (PAMPs) that are recognized by innate immune cells [7,51,52], suggesting that T3SS interactions with neutrophils may be responsible for the synthesis of LTB 4 during infections with the Y. pestis T3E strain.To test this hypothesis, we infected murine and human neutrophils with a Y. pestis strain lacking the pCD1 plasmid encoding the entire Ysc T3SS [Y.pestis T3 (-) ].Unlike infections with Y. pestis T3E, we did not observe an increase in LTB 4 synthesis by neutrophils during interactions with Y. pestis T3 (-) compared to uninfected or Y. pestis infected cells, even after 2 h of infection (Fig 5A and 5B).Importantly, Y. pestis T3 (-) infection did not appear to result in increased neutrophil cell permeability or cytotoxicity (S2 Fig) .To independently test that the T3SS is required to induce LTB 4 synthesis, Y. pestis T3E was cultured under conditions that alter the expression of the T3SS prior to infection of neutrophils [5,8,16].Measuring expression of the LcrV protein as a proxy for overall T3SS expression confirmed decreased T3SS expression in cultures grown at 26˚C compared to 37˚C (Figs 6A and S3A).As predicted by the Y. pestis T3 (-) data, LTB 4 synthesis was not observed from neutrophils infected with Y. pestis T3E strains grown at 26˚C, while synthesis was induced from bacteria cultured at 37˚C (Fig 6B).No difference in bacterial viability between any of the Y. pestis strains was observed during the time frame of the experiment, diminishing the possibility that differences in neutrophil killing of the bacteria was responsible for these phenotypes (S3B Fig) .Finally, neutrophils were infected with a Y. pestis T3E yopB mutant, which retains the other pCD1 encoded genes, but is defective in expression of the translocase that directly interacts with the host cell and is required for injection of the effector proteins [8,[53][54][55][56].Similar to the Y. pestis T3 (-) strain, the Y. pestis T3E yopB mutant did not induce LTB 4 synthesis, but synthesis was restored by yopB complementation (yopB::cyopB) (Fig 6C).Together, these data indicate that neutrophils recognize components of the Y. pestis T3SS as PAMPs, leading to the induction of LTB 4 synthesis, but only in the absence of the Yop effectors.

Y. pestis inhibition of LTB 4 synthesis is conserved during interactions with other leukocytes
In addition to neutrophils, two other lung resident leukocytes that can produce LTB 4 are mast cells and macrophages [33].To determine if Y. pestis inhibits LTB 4 synthesis by these two cell types, bone marrow-derived mast cells and macrophages were isolated from C57BL/6J mice and infected with Y. pestis, Y. pestis T3E, or Y. pestis T3 (-) .We observed no synthesis of LTB  similar to that of mast cells stimulated with crystalline silica, a potent inducer of LTB 4 synthesis [57,58].LTB 4 synthesis by mast cells was also dependent on the presence of the T3SS, as the Y. pestis T3 (-) strain did not induce LTB 4 synthesis (Fig 7A).For macrophages, previous reports indicate that polarization influences the ability to produce LTB 4 , with M1-polarized macrophages better able to synthesize LTB 4 in response to bacterial ligands than M2-polarized cells [59].Therefore, we measured LTB 4 synthesis of both M1-and M2-polarized macrophages (S4 Fig).Again, we observed no significant synthesis of LTB 4 by either macrophage population during interactions with Y. pestis, even after 4 h post-infection (Fig 7B and 7C).However, significant synthesis of LTB 4 was observed in M1-polarized macrophages in response to the Y. pestis T3E strain, which was dependent on the presence of the T3SS (Fig 7B; p � 0.0001).As suggested by previous reports [60], we did not observe significant changes in LTB 4 synthesis by M2-polarized macrophages during interactions with any of the Y. pestis strains tested (Fig 7C).Together, these data indicate that mast cells and M1-polarized macrophages can synthesize LTB 4 in response to the Y. pestis T3SS, but the activity of the Yop effectors inhibits this response.

Discussion
A hallmark manifestation of plague is the absence of inflammation during the early stages of infection, which is critical to Y. pestis virulence [10,17,28,52].While Y. pestis has been shown to actively dampen the host immune response, there is a gap in our understanding of the role of lipid mediators of inflammation during plague.This study sought to better define the host inflammatory lipid mediator response during pneumonic plague and expands our current understanding of how Y. pestis manipulates the immune system.During the earliest stages of infection, the host appears unable to initiate a timely LTB 4 response (Fig 1).Moreover, we demonstrated that exogenous treatment with LTB 4 can alter the host response to Y. pestis (Fig 3), suggesting that LTB 4 manipulation by Y. pestis contributes to disease outcome.Because LTB 4 is a potent chemoattractant crucial for rapid inflammation [31,32,61], a delay in LTB 4 synthesis during plague likely has a significant impact on the ability of the host to mount a robust inflammatory response needed to inhibit Y. pestis colonization.First, in the absence of LTB 4 , sentinel leukocytes will not undergo autocrine signaling via LTB 4 -BLT1.Because LTB 4 -BLT1 engagement activates antimicrobial programs in leukocytes [31,32,46,[62][63][64], the absence of autocrine signaling diminishes the ability of sentinel leukocytes directly interacting with Y. pestis to mount an effective antimicrobial response to kill the bacteria.LTB 4 synthesis is also regulated by BLT1 signaling, and autocrine signaling is required to amplify the production of LTB 4 needed to rapidly recruit additional tissue-resident immune cells to the site of infection [31,32,63,65,66].Therefore, the normal feed-forward amplification of LTB 4 synthesis, which is key for a rapid response to a bacterial infection, will also be inhibited by Y. pestis.Second, because LTB 4 is required for neutrophil swarming [65,67,68], Y. pestis will also inhibit this key inflammatory mechanism [69].Neutrophil swarming is required to contain bacteria at initial sites of infection [70,71].Thus, while individual neutrophils may migrate towards sites of Y. pestis infection, effective neutrophil swarming of large populations of neutrophils will be diminished.Finally, LTB 4 is a diffusible molecule that can induce the inflammatory cascade in bystander cells [32,72].Thus, while Y. pestis can inhibit cytokine and chemokine expression by cells with which it directly interacts [11,15], inhibition of LTB 4 synthesis likely also delays subsequent release of molecules by cells that do not directly interact with the bacteria.Together with the bacteria's other immune evasion mechanisms, inhibition of LTB 4 synthesis is likely another significant contributor to the generation of the non-inflammatory environment associated with the early stages of pneumonic plague [10,11,15].Incorporating these new LTB 4 data with published findings from other laboratories [10,11,15], we have updated our working model of Y. pestis inhibition of inflammation during pneumonic plague (Fig 8).
These studies also revealed that components of the T3SS trigger LTB 4 synthesis by leukocytes.Because our previous work with human samples indicated that neutrophils synthesize LTB 4 in response to Y. pestis in the absence of the T3SS [26], we were initially surprised that we did not observe LTB 4 synthesis by murine neutrophils to the Y. pestis T3 (-) strain.However, when we infected human neutrophils with lower MOIs, we observed that they also did not synthesize LTB 4 in the absence of the T3SS (Fig 5B).Under these infection conditions, neutrophils from both species only produced LTB 4 in response to Y. pestis expressing the T3SS but none of the Yop effectors.These data support that components of the T3SS are PAMPs produced by Y. pestis that are not only recognized by macrophages [73] but also by neutrophils.While previous studies have indicated that the Yop effectors are PAMPs in neutrophils [74,75], to our knowledge the data presented here represent the first example that non-effector components of the Y. pestis T3SS can also be recognized as a PAMP by neutrophils.In macrophages, in the absence of the Yop effectors, interactions with the T3SS, notably the translocon proteins YopB and YopD, induce NLRP3-dependent activation of the caspase 1 inflammasome, IL1-β secretion, and pyroptosis [54,76], suggesting that inflammasome activation may contribute to LTB 4 synthesis during interactions with Y. pestis T3E.However, whether inflammasome activation is required for the Y. pestis T3SS-mediated LTB 4 synthesis remains unclear, as LTB 4 synthesis in response to other stimuli is not dependent on inflammasome activation [57,77,78].Interestingly, infection of neutrophils with a strain of Y. pestis that only expresses YopK, which has been reported to inhibit NLRP3 inflammasome activation in macrophages [53,54], does not inhibit LTB 4 synthesis (Fig 4C) [26], supporting the possibility that LTB 4 synthesis may not be dependent on inflammasome activation in neutrophils.Future studies using neutrophils from mice defective in specific NLRs and caspases will allow us to definitively determine if inflammasome activation is required for LTB 4 synthesis in response to the Y. pestis T3SS.We have also confirmed that four Yop effectors, YpkA, YopE, YopJ, and YopH are sufficient to inhibit LTB 4 synthesis by both human and murine neutrophils.Synthesis of LTB 4 requires MAPKand Ca 2+ -dependent activation of cPLA2 and 5-LOX [34,79].Previous work, primarily in macrophages, has shown that both of these signaling pathways are efficiently inhibited by these four Yop effectors [8,20,[80][81][82][83][84], suggesting that subversion of MAPK and Ca 2+ signaling by Y. pestis is responsible for inhibition of LTB 4 synthesis.Supporting this hypothesis, Pulsifer et al. demonstrated that inhibition of ERK phosphorylation by YopJ is sufficient to inhibit LTB 4 synthesis by human neutrophils [26].Defining the specific molecular mechanisms employed by YpkA, YopE, and YopH to inhibit LTB 4 synthesis will be important in better understanding Y. pestis virulence and is a long-term goal.
One of the key antimicrobial mechanisms inhibited by the Yop effectors is phagocytosis [5,16,85,86], and Hedge et al. have previously shown that phagocytosis of crystalline silica is required for LTB 4 synthesis in that model of sterile inflammation [57].These data raise the possibility that inhibition of phagocytosis by Y. pestis may not only inhibit bacterial killing, but LTB 4 synthesis and rapid initiation of inflammatory programing in neutrophils.Studies to delineate the contribution of phagocytosis to LTB 4 synthesis are ongoing, but the differences in LTB 4 synthesis by cells infected with Y. pestis T3E and Y. pestis T3 (-) suggest that phagocytosis alone is not sufficient to trigger LTB 4 synthesis in the absence of proper PAMPs, in this case components of the T3SS.Moreover, the lack of LTB 4 synthesis in response to Y. pestis T3 (-) also differed from what we observed for other gram-negative bacteria without a T3SS (E. coli and K. pneumoniae), indicating that Y. pestis may also mask other potential gram-negative PAMPS that would typically be recognized by neutrophils.These data support that Y. pestis has evolved both active (via the Yop effectors) and passive mechanisms to evade immune recognition and induction of LTB 4 synthesis.It is worth noting that unlike human neutrophils, murine neutrophils did not appear to synthesize LTB 4 during infections with the T3 (-) strain at high MOIs (S5 Fig) .Differences in neutrophil responses between the two species have been well documented [87][88][89][90][91], but these observations merit further investigation into LTB 4 responses by human neutrophils using higher MOIs to determine if human neutrophils are able to recognize other PAMPs during Y. pestis infection.
Finally, while we focused primarily on LTB 4 in this study, we also observed changes in the synthesis of other lipids during plague that merit future considerations (S1 Table ).The rapid cyclooxygenase response raises questions about whether prostaglandins are protective or detrimental during pneumonic plague.Historically, prostaglandins were thought to promote inflammation, but these mediators appear more nuanced under closer scrutiny and can just as likely inhibit inflammation, as well as participate in normal development physiology, without eliciting inflammation [42,43,92].The prostaglandins we observed as being significantly elevated during the non-inflammatory stage of pneumonic plague-PGA 2 , PGD 2 , PGE 2 , and PGJ 2 -have been shown to inhibit inflammation in various models, especially as concentrations increase [42,43,[93][94][95].PGE 2 can inhibit NADPH oxidase activity during infection with K. pneumoniae, which suppressed bacterial killing [96], and directly counteracts the proinflammatory activities of LTB 4 [97,98].The phagocytic index of LTB 4 -stimulated rat alveolar macrophages (AMs) is reduced when co-stimulated with PGE 2 [98].Moreover, AMs treated with PGE 2 showed a 40% reduction in LTB 4 synthesis when stimulated with an ionophore known to induce a strong LTB 4 response [97].This inhibition of LTB 4 by PGE 2 is suspected to be via an increase in second messenger cAMP that activates protein kinase A (PKA), which has been shown to inhibit LTB 4 synthesis [97,99].Together, these data suggest that the elevated levels of prostaglandin synthesis observed during pneumonic plague may contribute to the blunted LTB 4 response by the host.
In conclusion, we have defined the kinetics of the key inflammatory lipid mediator LTB 4 during pneumonic plague, which revealed a blunted response during the early stages of infection.Furthermore, we have shown that Y. pestis actively manipulates LTB 4 synthesis by leukocytes via the activity of Yop effectors to generate a beneficial inflammatory outcome to the pathogen.These discoveries warrant further research into the role of lipids, and subsequent manipulation of their synthesis by Y. pestis, to fully understand the molecular mechanisms Y. pestis has evolved to manipulate the mammalian immune response.

Ethics statement
All animal work was approved by the University of Louisville Institutional Animal Care and Use Committee (IACUC Protocol #22157).Use of human neutrophils was approved by the University of Louisville Institutional Review Board guidelines (IRB #96.0191) and written consents for use were obtained.

Bacterial strains
Bacterial strains used in this study are listed in S2 Table .For mouse infections, Y. pestis was grown at 26˚C for 6-8 h, diluted to an optical density (OD) (600 nm) of 0.05 in Bacto brain heart infusion (BHI) broth (BD Biosciences Cat.No. 237500) with 2.5 mM CaCl 2 and then grown at 37˚C with aeration for 15-18 h [100].For cell culture infections, Y. pestis was cultured with BHI broth for 15-18 h at 26˚C in aeration.Cultures were then diluted 1:10 in fresh, warmed BHI broth containing 20 mM MgCl 2 and 20 mM Na-oxalate and cultured at 37˚C for 3 h with aeration to induce expression of the T3SS.Bacterial concentrations were determined using a spectrophotometer and diluted to desired concentrations in 1 × Dulbecco's phosphatebuffered saline (DPBS) for mouse infections or fresh medium for in vitro studies.Concentrations of bacterial inoculums for mouse studies were confirmed by serial dilution and enumeration on BHI agar plates.

Mouse infections
All animal work was performed at least twice to ensure reproducibility.6-8 week-old C57BL/ 6J or BLT1 -/- [49] male and female mice were infected with Y. pestis KIM5+ or Y. pestis CO92 LUX pcysZK .For lipid measurements, mice were anesthetized with ketamine/xylazine and administered 20 μL of Y. pestis KIM5+ suspended in 1× DPBS to the left nare as previously described [48,100].Mice were monitored for the development of moribund disease symptoms twice daily and humanely euthanized when they met previously approved end point criteria.At 6, 12, 24, 36, or 48 h, mice were humanely euthanized by CO 2 asphyxiation and lungs were harvested and lung masses recorded.Lungs were transferred to a 2 mL tube pre-filled with 2.8 mm ceramic beads (VWR, Cat.No. 10158-612), flash frozen on dry ice, and stored at -80˚C until preparation for lipid analysis.For CFU studies, mice were humanely euthanized by CO 2 asphyxiation at 12 or 24 h and lungs were harvested.Lungs were transferred to Whirl Pak's containing 1 mL of 1 x DPBS, and gently homogenized using a serological pipette.Homogenized tissues were serial diluted and plated onto BHI agar.After 2 days of incubation at 26˚C, bacteria were enumerated.For optical imaging and survival curves, mice were infected with Y. pestis CO92 LUX pcysZK and monitored for bacterial proliferation as a function of bioluminescence by optical imaging and for the development of moribund disease.At each time point, mice were anesthetized with isoflurane and imaged using the IVIS Spectrum imaging system (Caliper Life Sciences, Hopkinton, MA).Average radiance (photons/s/cm2) was calculated for the lungs as previously described [48].For the exogenous LTB 4 treatment, mice were intraperitoneally injected with 1 x DPBS or 10 nmol LTB 4 (Cayman Chemical Cat.No. 20110).At 1 h post-treatment, mice were administered 10 5 CFU of Y. pestis KIM5+ via intraperitoneal injection.At 3 h post infection, mice were humanely euthanized, and the peritoneal cavity was washed and collected using 2 lavages of 1 mL of 1 x DPBS.Lavages were used for CFU enumeration or neutrophil quantification by flow cytometry.

Lipid extraction and quantification by LC-MS
To quantify LTB 4 abundance from whole lungs, lungs were thawed with 1.8 mL of ice cold 75% methanol + 0.1% BHT for 3 minutes.Lungs were then homogenized with a Bead Ruptor 4 (OMNI) at speed 5 (5 m/s) for 4 cycles of 45 seconds with 1-minute pauses in which the lungs were placed on ice.Tissue debris was then centrifuged for 10 min at 1,500 x g at 4˚C.The supernatant (~1.5 mL) was then transferred to a fresh eppendorf tube, incubated at 4˚C for 24 h to inactivate Y. pestis and extract lipids.After successful inactivation, samples were removed from BSL3 containment and stored at -80˚C.Lipid extraction was then performed as previously described [101].For the expanded global lipid analysis, lungs were thawed with 1.5 mL of ice cold 1 x DPBS + HALT protease and phosphatase inhibitor cocktail for 3 minutes.Lungs were then homogenized with a Bead Ruptor 4. Tissue debris was then centrifuged for 10 min at 1,500 x g at 4˚C.The supernatant (~1.5 mL) was then transferred to a fresh eppendorf tube.From this, 250 μL of supernatant was combined with 750 μL of 100% methanol + 0.1% BHT (final concentration of 75%) and incubated at 4˚C for 24 h to inactivate Y. pestis and extract lipids.After confirmation of successful inactivation of Y. pestis, lipids were extracted and quantified by the Wayne State University Lipidomics Facility as previously described [102].The extracted samples were analyzed for the fatty acyl lipidome using standardized methods as described previously [103,104].

Flow cytometry
To quantify the neutrophil population from peritoneal lavages, cells were labeled with anti-Ly6G antibody (1:400; BD Pharminogen Cat.No. 551460) and anti-CD11b antibody (1:600; Biolegend Cat.No. 101212) for 1 h on ice, in the dark.Cells were pelleted and resuspended in 1% PFA.Single cell suspensions were generated by straining with 70 μM mesh prior to analysis on the flow cytometer.Neutrophils were identified as cells with high expression of Ly6G and CD11b and data is represented as the percent of the population that were classified as neutrophils.An example of the gating strategy is shown in S1 Fig.

Cell isolation and cultivation
Human neutrophils were isolated from the peripheral blood of healthy, medication-free donors, as described previously [105].Briefly, white blood cells were isolated from whole blood using a 6% dextran solution.Neutrophils were then separated from monocytes using a percoll gradient of 42% and 50.5%.RBCs were then lysed from the neutrophil containing layer using 0.2% NaCl for 30 seconds and followed by a quench with 5 mL 1.6% NaCl.Neutrophil isolations yielded � 95% purity and were used within 1 h of isolation.Murine neutrophils were isolated from bone marrow of 7-12-week-old mice using an Anti-Ly-6G Microbeads kit (Miltenyi Biotec Cat.No. 130-120-337) per the manufacturer's instructions.Neutrophil isolations yielded � 95% purity and were used within 1 h of isolation.Macrophages were differentiated from murine bone marrow in DMEM supplemented with 1 mM Na-pyruvate and 10% FBS for 6 days.Macrophages were either polarized with 10 ng/mL of GM-CSF (M1; Kingfisher Biotech Cat.No. RP0407M) or with 30% L929 conditioned media and 10 ng/mL of M-CSF (M2; Kingfisher Biotech Cat.No. RP0462M) throughout the differentiation.The medium was replaced on days 1 and 3 (adapted from [106]).Polarization was confirmed by qRT-PCR, as previously described [107], using markers for M1 and M2 phenotypes, TNF-α and Il-10, respectively (S4 Fig) .Murine mast cells were isolated and differentiated from bone marrow as previously described [108].Briefly, isolated bone marrow cells were resuspended in BMMC culture medium [DMEM containing 10% FCS, penicillin (100 units/mL), streptomycin (100 mg/mL), 2 mmol/L L-glutamine, and 50 mmol/L β-mercaptoethanol] supplemented with recombinant mouse stem cell factor (SCF) (12.5 ng/mL; R&D Systems Cat.No. 455-MC) and recombinant mouse IL-3 (10 ng/mL; R&D Systems Cat.No. 403-ML).Cells were plated at a density of 1 x 10 6 cells/mL in a T-75 cm 2 flask.Nonadherent cells were transferred after 48 hours into fresh flasks without disturbing the adherent (fibroblast) cells.Mast cells were visible after 4 weeks of culture and propagated further or plated for experiments in DMEM without antibiotics.

Leukocyte infections
Human neutrophils were resuspended in Kreb's buffer (w/ Ca 2+ and Mg 2+ ) then adhered to 24-well plates for 30 min that were coated with pooled human serum prior to infection (wells were washed twice with 1 x DPBS prior to plating the cells).Murine bone marrow neutrophils were resuspended in RPMI + 5% FBS then adhered to 24-well plates for 30 min that were coated with FBS prior to infection (wells were washed twice with 1 x DPBS prior to plating the cells).Neutrophils were infected at a multiplicity of infection (MOI) of 20, 50, or 100 and incubated for 1 h in a cell culture incubator at 37˚C with a constant rate of 5% CO 2 .Co-infections were performed at a final MOI of 20 (MOI of 10 for each strain). 1 h post-infection, supernatants were collected, centrifuged for 1 min at 6,000 x g, and supernatants devoid of cells were transferred to a fresh eppendorf tube.Macrophages were adhered to 24-well plates in DMEM + 10% FBS 1 day prior to infection.Macrophages were infected at an MOI of 20.At 4 h postinfection, supernatants were collected, centrifuged for 1 min at 6,000 x g, and supernatants devoid of cells were transferred to a fresh eppendorf tube.Mast cells were adhered to 24-well plates in DMEM only for 1 h prior to infection.Mast cells were infected at an MOI of 20 or treated with crystalline silica (100 mg/cm 2 ).At 2 h post-infection supernatants were collected, centrifuged for 1 min at 6,000 x g, and supernatants devoid of cells were transferred to a fresh eppendorf tube.All infections were synchronized by centrifugation (200 x g for 5 min).All samples were stored at -80˚C until ELISA.

Measurement of LTB 4 by enzyme-linked immunosorbent assay
Supernatants of neutrophils, macrophages, and mast cells were collected and measured for LTB 4 by ELISA per manufacturer's instructions (Cayman Chemicals Cat.No. 520111).

Cell viability assays
To determine leukocyte permeability, cells were incubated with trypan blue for 5 min and trypan blue exclusion was measured using SD100 counting chambers (VWR Cat.No. MSPP-CHT4SD100) and a cell counter (Nexcelom Cellometer Auto T4).To determine leukocyte cytotoxicity, lactate dehydrogenase (LDH) was measured from leukocyte supernatants using the CytoTox 96 Non-Radioactive Cytotoxicity kit (Promega Cat.No. g1780) per the manufacturer's instructions.

Bacterial viability assays
To measure bacterial viability during interactions with neutrophils, murine neutrophils were resuspended in RPMI + 5% FBS then adhered to 96-well white bottom plates for 30 min coated with FBS prior to infection (wells were washed twice with 1 x DPBS prior to plating the cells).Neutrophils were infected at an MOI of 20, centrifuged for 5 min at 200 x g, and bacterial

PLOS PATHOGENS
Inhibition of leukotriene B4 by Yersinia pestis viability was measured as a function of bioluminescence using a plate reader (BioTek Cytation 1 imaging reader).

Measurement of LcrV by western blot
Bacterial strains were cultured with BHI broth for 15-18 h at 26˚C in aeration.Cultures were then diluted 1:10 in fresh warmed BHI broth containing 20 mM MgCl 2 and 20 mM Na-oxalate and cultured at 37 or 26˚C for 3 h. 1 OD 600 of bacterial pellets were collected and resuspended in 1 x SDS-PAGE loading buffer, boiled for 10 min, and 0.1 OD 600 was separated on a 10% SDS-PAGE gel.As a positive control, 0.2 g of recombinant LcrV protein was used (BEI resources Cat.No. NR-32875).Samples were immunoblotted with polyclonal anti-LcrV antibody diluted to 1:4,000 (BEI Resources Cat.No. NR-31022).Anti-goat IgG HRP secondary antibody was diluted to 1:5,000 (Bio-Techne Cat.No. HAF017).Densitometry was performed using ImageJ software to compare LcrV bands between samples [109].

Statistics
For all studies, male and female mice or human donors were used and no sex biases were observed for any phenotype.All in vivo experiments were repeated at least twice and in vitro experiments at least 5 times.Where noted in the figure legends, figures may represent the combined data from multiple biologically independent experiments.For in vitro experiments, each data point represents data from biologically independent experiments performed on different days.Where appropriate and as indicated in the figure legends, statistical comparisons were performed with Prism (GraphPad) using one-way analysis of variance (ANOVA) with Dunnett's or Tukey's post hoc test, T-test with Mann-Whitney's post hoc test, or Log-Rank analysis.P values � 0.05 were considered statistically significant and reported.For LC-MS analysis of lipids, a LIMMA-Moderated T-test was performed using a modified version of a previously published protocol using R packages [110][111][112].Briefly, raw data were transformed by taking logarithmic base 2 followed by quantile normalization.Missing values were then ascribed using a singular value decomposition method.Lipids missing > 40% of the values were excluded from subsequent analysis.Finally, differentially abundant lipids (p � 0.05) were further filtered by fold-change (FC) criteria (1 < log 2 FC < 1) and multiple comparisons testing with a false discovery rate.

Fig 2 .Fig 3 .
Fig 2. BLT1 -/-mice are not more susceptible to pneumonic plague than C57BL/6J mice.(A) C57BL/6J (green circles) or BLT1 -/-(purple squares) mice were infected intranasally with 10 x the LD 50 of Y. pestis KIM5+ and lungs were harvested at 12 and 24 h post-infection.Bacterial proliferation within the lungs was determined by CFU enumeration.Each symbol represents an indivdual mouse and the box plot represents the median of the group ± the range.Combined data from two independent experiments.(B-C) C57BL/6J (green circles) or BLT1 -/-(purple squares) mice were infected intranasally with 10x the LD 50 of Y. pestis CO92 LUX pcysZK .(B) Bacterial proliferation in the lungs as a function of bioluminescence.Each symbol represents an indivdual mouse and the box plot represents the median of the group ± the range.Combined data from two independent experiments.(C) Survival curves of mice from B (n = 15).For A and B, T-test with Mann-Whitney's post hoc test indicated no statistically significant (ns) differences between C57BL/6J and BLT1 -/-groups.For C, Log-Rank anlysis revealed no statistically signficant (ns) differences in surival between the two groups.https://doi.org/10.1371/journal.ppat.1011280.g002 Fig).Even at an MOI of 100, while cell permeability appeared slightly elevated in Y. pestis-infected murine neutrophils compared to uninfected cells (S2C Fig; 9% vs. 28%), overall cytotoxicity was lower in Y. pestis-infected cells (S2D Fig; 12% vs. 4%).Similarly, Y. pestis did not induce elevated permeability or cytotoxicity in human neutrophils (S2E and S2F Fig).These data demonstrate that while neutrophils can rapidly synthesize LTB 4 in response to other bacterial pathogens, neither murine nor human neutrophils appear to synthesize LTB 4 in response to Y. pestis.

Fig 6 .Fig 7 .
Fig 6.Neutrophils synthesize LTB 4 in response to the Y. pestis T3SS in the absence of the Yop effectors.(A) Relative expression of LcrV based on western blots normalized to total protein loaded from bacteria cultured at 37 or 26˚C.(B) Murine neutrophils (BMNs) were infected (MOI of 20) with Y. pestis (Yp) or mutants that either lacked the Yop effectors (T3E) or the Yop effectors and the T3SS [T3(-)] cultured at 37 or 26˚C and LTB 4 was measured at 60 min postinfection.(C) Murine neutrophils (BMNs) were infected (MOI of 20) with Yp, T3E, T3(-), a yopB mutant in the T3E background (ΔB), or ΔB complemented with yopB (ΔB::B) cultured at 37˚C and LTB 4 was measured at 60 min post-infection.Each symbol represents an independent biological infection and the box plot represents the median of the group ± the range.One-way ANOVA with Tukey's post hoc test compared to each condition for A and B and Dunnett's post hoc test compared to uninfected for C. ns = not significant, *** = p � 0.001, # = p � 0.0001.https://doi.org/10.1371/journal.ppat.1011280.g006

Fig 8 .
Fig 8. Working model for inhibition of the inflammatory cascade during plague.(A) Normal response by sentinel leukocytes results in rapid production of LTB 4 that leads to autocrine signaling, neutrophil swarming, and induction of cytokine and chemokine release.(B) Y. pestis inhibits the production of LTB 4 via the action of the Yop effectors, which delays resident neutrophil recruitment and subsequent production of cytokines and chemokines needed for inflammation.https://doi.org/10.1371/journal.ppat.1011280.g008