Unraveling neutrophil– Yersinia interactions during tissue infection

The human and animal pathogens Yersinia pestis, which causes bubonic and pneumonic plague, and Yersinia pseudotuberculosis and Yersinia enterocolitica, which cause gastroenteritis, share a type 3 secretion system which injects effector proteins, Yops, into host cells. This system is critical for virulence of all three pathogens in tissue infection. Neutrophils are rapidly recruited to infected sites and all three pathogens frequently interact with and inject Yops into these cells during tissue infection. Host receptors, serum factors, and bacterial adhesins appear to collaborate to promote neutrophil– Yersinia interactions in tissues. The ability of neutrophils to control infection is mixed depending on the stage of infection and points to the efficiency of Yops and other bacterial factors to mitigate bactericidal effects of neutrophils. Yersinia in close proximity to neutrophils has higher levels of expression from yop promoters, and neutrophils in close proximity to Yersinia express higher levels of pro-survival genes than migrating neutrophils. In infected tissues, YopM increases neutrophil survival and YopH targets a SKAP2/SLP-76 signal transduction pathway. Yet the full impact of these and other Yops and other Yersinia factors on neutrophils in infected tissues has yet to be understood.


Introduction
The study of the Gram-negative bacterial pathogens Yersinia pestis, Yersinia pseudotuberculosis, and Yersinia enterocolitica has been at the forefront of cellular and molecular pathogenesis for over four decades. Y. pestis, a recently emerged pathogen evolved from Y. pseudotuberculosis, is the causative agent of bubonic and pneumonic plagues and has produced several pandemics in the past 10,000 years 1 . These infections are associated with high mortality rates of 30% and over 90%, respectively, if not treated rapidly with antibiotics. Transmission of Y. pestis to cause bubonic plague occurs via flea bite into the intradermal skin layer, whereas transmission of pneumonic plague occurs via inhalation of Y. pestis into lungs. By contrast to the highly lethal infections caused by Y. pestis in humans, Y. pseudotuberculosis and Y. enterocolitica generally cause self-limiting gastroenteritis and mesenteric lymph adenitis in most otherwisehealthy humans, rarely spreading to cause systemic disease or fatal infections. Infections normally occur through ingestion of contaminated foods or liquids. All three Yersinia species infect a variety of mammals, including rodents and ungulates, and the enteric pathogens can be found in birds. Thus, the Yersiniae are "generalists", adept at surviving in many different hosts, and so have evolved virulence factors and pathogenic strategies that counteract immune systems of a variety of animals.
Over the past 35 years, the study of the Yersinia virulence factors-including bacterial adhesins, the type 3 secretion system (T3SS), effector proteins (Yops), Pla protease in Y. pestis, and iron acquisition systems-has revealed critical features of host-pathogen interactions (reviewed in 2-5). Notable recent advances include uncovering aspects of innate immunity that are triggered or suppressed (or both) by the T3SS and Yops in macrophages 6-17 and reconstructing the recent evolutionary progression from Y. pseudotuberculosis to Y. pestis 1,18-20 . Another critical feature of Yersinia-host cell interaction garnering attention is its interactions with neutrophils during various types of tissue infection. Neutrophils are critical cells of the innate immunity system and both sense pathogens resulting in release of signaling molecules, such as cytokines and alarmins, and kill invading microbes through a variety of mechanisms 21 . These killing mechanisms include phagocytosis, generation of reactive oxygen species, degranulation, and formation of neutrophil extracellular traps 21-23 . Effector Yops hamper a number of these processes in isolated neutrophils 2,3,24-30 , observations which are further driving current interest in how neutrophils interact with Yersinia in the context of infected tissues and other cell types. This mini-review highlights recent studies involving Yersinia-neutrophil interactions in murine tissues.

Yersinia spps target Yop injection to neutrophils during infection of tissues
Yersinia spps use the highly conserved T3SS to inject six or seven effector Yop proteins in host cells to cause disease in mammals 2,5,31 . In tissue infections using a β-lactamase reporter system, studies with Y. pestis, Y. enterocolitica, and Y. pseudotuberculosis have demonstrated that neutrophils are a major-and, typically, the primary-cell target for Yop injection 32-38 . That neutrophils are a significant target holds true regardless of whether the route of infection is oral, intravenous, or intranasal and the tissues examined are Peyer's patches, mesenteric lymph nodes, spleens, or lungs 32-37 .
There are, however, several exceptions, most notably early after tissue infection 32,39,40 . For example, 6 hours after intranasal infection with the virulent Y. pestis CO92 strain, alveolar macrophages are the primary injected cell type and comprise over 50% of injected cells whereas neutrophils comprise about 15% of the injected population 36 . This balance shifts by 12 hours when neutrophils start invading tissues at higher numbers and become over 70% of the injected population 36 . Similarly, 1 day after intravenous infection with Y. enterocolitica, macrophages are more highly targeted in the spleen but by day 2 neutrophils become equally targeted 39 . After oral infection with Y. pseudotuberculosis, levels of injected macrophages and neutrophils in the mesenteric lymph nodes are comparable 5 days after oral gavage 32 . Finally, in splenic infections, while neutrophils are enriched significantly for injection, frequently a comparable or even higher absolute number of B cells are injected with Yops 32,33,35,39 . The high injection levels of B cells may occur because they comprise the majority of total cells in spleensover 60% compared with the much lower numbers of neutrophils and macrophages-so Yersinia may stochastically encounter them more than any other cell type 32,33,35 . For insights into the Yop interactions with B and T cells, readers are referred to the following 38,41-44 .

What factors are important for Yersinia targeting Yops into neutrophils? Proximity
There are both physiological and molecular explanations for why neutrophils are a major target of Yersinia Yop injection during infection. One significant reason is location of the cells to the bacteria since tight binding of Yersinia to cells is required for Yop injection 25,45,46 , and neutrophils migrate to inoculation sites within hours or days after infection and are typically the closest cell type associated with the bacteria. After infection with the enteric Yersinia pathogens, neutrophils migrate to and eventually encase the bacterial colonies in the Peyer's patches or spleens between 24 and 72 hours after oral or intravenous infection [47][48][49] . In these tissues, pyogranulomas form containing tightly packed Y. pseudotuberculosis immediately surrounded by neutrophils with macrophages forming an outer ring of cells 47 . It is noteworthy that, in some cases, neutrophils associate more rapidly with yop mutants than with wild-type Yersinia. For instance, within a day after oral inoculation with wild-type Y. pseudotuberculosis, Y. enterocolitica, or yop mutants, more neutrophils are found in association with the yop mutants than the wild-type Yersinia, indicating that early interactions of wild-type Yersinia with resident tissue cells delay chemotaxis of neutrophils to the bacteria 48,49 .
After intradermal inoculation, by either a flea bite or needle inoculation, neutrophils are detected within 50 minutes to 7 hours of inoculation 50-52 . Likewise, in lung infections with Y. pestis or Y. pseudotuberculosis, neutrophils migrate to tissue sites within 12 to 48 hours, depending on the strain 36,37,53,54 .
Yet several lines of evidence suggest that proximity is not the only factor critical for the preponderance of neutrophils found injected with Yops. In one study, in which neutrophils and inflammatory monocytes were depleted from tissues, fewer overall cells are targeted for injection rather than different cell populations 32 . This suggests that the nature of bacterial growth within tissues or interactions with the remaining innate immune cells in tissues (or both) play a role in the number of injected cells during tissue infection.

Receptors, adhesins, and serum factors
Ex vivo studies show that the selectivity for injection into neutrophils is recapitulated in single-cell splenic and lung homogenates infected with Yersinia when bacteria are limiting, indicating that there are specific receptor-ligand interactions that are favored between the Yersinia and neutrophils 32,34,35,37,39,55 . Blockage of complement receptor 3 (CR3) on neutrophils from ex vivo splenic homogenates significantly reduced injection into neutrophils by Y. pestis in splenocytes, demonstrating that Y. pestis uses this receptor 55 and suggesting that the CR3 receptor, which is enriched in neutrophils, plays a role in promoting Y. pestis-neutrophil interactions over other cell types in these lysates. This has yet to be evaluated in a mouse model of infection with Y. pestis. But findings with Y. pseudotuberculosis also show a role for complement or serum factors in directing injection of Yops into neutrophils in isolated mouse splenocytes and in mouse infections in spleens after intravenous but not in intranasal infections 34,37 . Elegant in vivo studies with β1-depleted mice demonstrate that Y. enterocolitica uses β1 integrins to inject Yops into many different cell types in infected spleens, although this receptor usage was not specific to neutrophils 39 .
The enteric Yersinia bacterial adhesins YadA, Invasin, or Ail (or a combination of these) are important for injection in mouse tissues 34,39 . Under conditions where adhesin-mutants and the wild-type strain were recovered at comparable numbers, a ΔailΔinvΔyadA triple mutant in Y. pseudotuberculosis and a yadA mutant in Y. enterocolitica injected fewer cells after intravenous infection than the isogenic wild-type strains 34,39 . In the case of Y. enterocolitica, very few neutrophils are detected in tissues correlating with very few injected cells and this is similar to findings with Y. pseudotuberculosis 32,39 . After infection with the ΔailΔinvΔyadA triple mutant in Y. pseudotuberculosis, the spectrum of injected cells was not changed 34 . However, treatment with cobra venom factor both restores virulence of the triple mutant and causes a significant shift in spectrum of cells targeted by the triple-mutant but not the wild-type strain 34 . Specifically, in mice treated with cobra venom factor (which is a complement-activating protein that depletes complement regulatory proteins and ultimately complement), the triple mutant injects more B cells and fewer neutrophils 34 . This result shows that a combination of serum factors and bacteria adhesins influences cells targeted for injection in Y. pseudotuberculosis. However, it remains to be determined whether the nature of the pyogranuloma formed under these conditions is the same as or different from that formed by the wild-type strain 47 and whether this explains the altered spectrum of injected cells. Nonetheless, Ail and YadA have long been recognized in in vitro studies to interact with serum factors and to bind cells and promote injection 45,56-62 ; these in vivo studies demonstrate their importance for injection of Yops in tissue infection.

Do neutrophils matter in tissue infection?
Given that Yersinia expends energy injecting Yops into neutrophils and appears to have co-opted CR3 or serum factors (or both) in mouse tissues to enhance injection into neutrophils, the question arises "Are neutrophils important to contain Yersinia infection?" If neutrophils are important, one would predict increased colony-forming units (CFUs), increased disease symptoms, and decreased time to morbidity in their absence. At first glance, the results of the classic approach of depleting neutrophils and measuring infection outcomes are mixed.
In intradermal models of infected mice depleted of neutrophils with 1A8, an antibody recognizing Ly6G found on mature neutrophils 63 , the CFUs of Y. pestis in the skin increase significantly in the absence of neutrophils, yet the total number of bacteria replicating in the draining lymph node does not change 64 . Likewise, CFUs of Y. pestis in lymph tissues remain constant after depletion with RB6-8C5 52 , an antibody that depletes Gr-1-expressing cells, including mature and immature neutrophils and subsets of inflammatory monocytes, dendritic cells, and T cells 65,66 . These results support the ideas that neutrophils are important for curbing bacterial growth in the skin, but not the lymph nodes, and that neutrophils are not essential for dissemination from skin to lymph tissues. Some debate exists about whether Y. pestis disseminates from skin to lymph nodes by hitchhiking in neutrophils, macrophages, or dendritic cells or a combination of these (reviewed in 67). But these most recent studies support the idea that Y. pestis can disseminate to lymph nodes independently of neutrophils.
Depletion of neutrophils with 1A8 in lung infection with Y. pestis, as with lymph node infection, does not result in changes in bacteria counts 24 or 48 hours after infection 36 . Surprisingly, symptoms of disease progression and time to death are reduced in the absence of neutrophils, although the difference in time to death is not statistically significant 36 . Consistent with these findings, the histopathology of mice treated with 1A8 showed less damage to lungs with intact alveoli structure whereas untreated mice had necrotizing pneumonia 36 . However, artificially increasing the numbers of neutrophils in lungs prior to infection significantly attenuates Y. pestis infection 53 . Combined, these results show that, early after infection, high levels of neutrophils stop infection, but once Y. pestis reaches a certain stage-either in number or in modulating early host responses or both-the bactericidal activities of neutrophils are effectively nullified and, in fact, their continued migration into lungs wrought the damage observed during infection.
The impact of neutrophils on restraining infection by the enteric Yersinia pathogens is equally mixed. In an oral infection model of Y. pseudotuberculosis, depletion of neutrophils with 1A8 or RB6-8C5 resulted in significantly more growth at day 1 post-infection with wild-type and yopE, yopH, and yopK mutant strains, and disease symptoms were worse on subsequent days 48,68 . Likewise, increasing the numbers of neutrophils in Peyer's patches or increasing their activation in spleens by depletion of dendritic cells results in increased clearance of wild-type Y. enterocolitica 49,69 . Increasing neutrophils in tissues also suppresses Y. pseudotuberculosis yop mutant colonization but not wild-type colonization in oral infection 70,71 . Although these results point to some differences between the enteric Yersiniae interactions with neutrophils, they indicate that neutrophils can control early seeding or dissemination events in infection (or both). However, neutrophil depletion does not always increase Y. pseudotuberculosis growth in tissues. Three days after oral inoculation, fewer wild-type Y. pseudotuberculosis were detected overall in neutrophil-depleted mice (by luminescence) compared with non-depleted tissues despite worsening disease symptoms 48,68 . After intravenous infection with Y. pseudotuberculosis, the number of bacteria recovered 3 days from mock-depleted, 1A8-treated or RB6-8C5-treated mice was comparable despite the observation that 1A8-or RB6-8C5treated mice appeared more ill and reached morbidity faster 26 . Thus, although the growth of Y. pseudotuberculosis is not always increased in the tissues examined in the absence of neutrophils, the overall health of the mice typically worsens.
Overall, these results are consistent with the idea that Yersinia handles intimate interactions with neutrophils effectively once infection in a tissue is established, but the bacteria are more susceptible to neutrophils early in infection, such as soon after inoculation or when disseminating to new tissues. Supporting the idea that Yersinia spps are well designed to withstand neutrophil onslaught in tissues is the observation that a number of attenuated Yersinia mutants grow significantly better in neutrophil-depleted mice, indicating that the function of these proteins is to inactivate neutrophils or withstand the bactericidal activities of neutrophils. Importantly, adhesin mutants yadA and ail and several yop mutants such as yopH, yopE, and yopK mutants 26,37,48,72 all colonize significantly better in some tissues in neutrophil-depleted mice than in wild-type mice. (Not every mutant is restored for growth in the absence of neutrophils 73,74 ; for example, some are restored in the absence of both neutrophils and inflammatory monocytes 73,75 and some have been tested only in the absence of both 75 .)

What are the consequences to neutrophils after Yop injection in mouse infections?
Several elegant studies have examined the transcriptome of cells surrounding Yersinia microcolonies by using RNA sequencing (RNA-seq) 76,77 . When dual-tissue RNA-seq was used to evaluate the host cell and bacterial responses to infection of Y. pseudotuberculosis in the Peyer's patches, a number of host transcripts associated with infection were strongly induced; this is indicative of the pronounced neutrophil infiltrate that occurs after infection 43,48,77 . These included metal ion sequestration, inflammatory responses, acute-phase responses, and coagulative activities 77 . These findings shed light into the overall host response to infection in tissues which are composed predominately of neutrophils, but the findings do not distinguish the cells specifically in contact with bacterial microcolonies. The β-lactamase reporter system 32,33,35 can also be used to distinguish and isolate injected from non-injected neutrophils in infected tissues. Via such an approach, YopH, a tyrosine phosphatase, was found to target the Slp-76/SKAP-2/PRAM pathway in neutrophils during tissue infection 26 . This pathway is critical for reactive oxygen production of neutrophils after integrin stimulation 78 , providing a possible role for YopH. This approach can be further exploited to determine direct from indirect consequences of Yersinia-neutrophil interactions in tissues.
Via laser capture microdissection, the inflammatory lesions in the lungs induced by Y. pestis were parsed on the basis of proximal (and presumably containing many cells injected with Yops) and distal areas to Y. pestis microcolonies 76 . These transcriptomes were compared with each other and with the transcriptome of bone marrow neutrophils from uninfected (representing not activated) and infected mice. Remarkably, the transcriptomes of cells proximal to the bacteria are most similar to bone marrow neutrophils from uninfected mice; that is, both resemble non-activated cells with higher expression of pro-survival signals and lower expression of chemotaxis/migration genes than the more distally located cells 76 . Strikingly, YopM expression changes the physiology of neutrophils in tissue infection but not the bacterial burden. In histological sections of mice infected with a yopM mutant, cells appear anucleated and express more apoptotic markers 76 , indicating that YopM contributes to the pro-survival state of the neutrophils yet this is not sufficient to impact bacterial survival.

What are the consequences to Yersinia after contact with neutrophils in mouse tissues?
Changes to Y. pseudotuberculosis and Y. pestis gene expression in lymph tissues that contain high numbers of neutrophils have been analyzed by microarray analysis and RNA-seq, respectively 77,79 . Notably, genes required for metal ion acquisition, nitric oxide (NO) stress responsiveness, and (in Y. pseudotuberculosis) carbohydrate use were mostly highly upregulated in tissues compared with 37°C broth-grown cultures 77,79 . Many of these pathways are also critical for survival of Y. pestis in a rat buboe model 80 . Collectively, these results point toward a local lymph environment where the bacteria experience high NO stress and restrictive ion and metabolic conditions. Y. pestis and Y. pseudotuberculosis respond to this restrictive metabolic environment in different ways; Y. pseudotuberculosis induces carbohydrate use genes and the upper part of glycolysis, whereas Y. pestis uses anaerobic respiration 77,79,80 .
Direct observation of Y. pseudotuberculosis expressing fluorescent reporter constructs that are responsive to different environmental cues has permitted further dissection of bacterial responses in tissues 47 . Specifically, at the periphery of microcolonies, Y. pseudotuberculosis expresses higher levels from the hmp promoter, an NO responsive gene, and yopE, a T3SS gene. Higher NO expression is more uniformly observed in the outer ring of the microcolony, yet inducible nitric oxide synthase (iNOS) was not detected in the immediately adjacent cells, indicating that NO diffuses from a distance 47 . By contrast, high expression from the yopE promoter was sporadically observed in individual cells on the periphery, suggesting that these cells are in direct contact with neutrophils and therefore upregulating yopE transcription 47 . These findings are consistent with increased copy number of the plasmid containing the T3SS upon contact with host cells 81 .

Emerging models: Yersinia-neutrophil interactions in murine tissues
Many facets of Yersinia-neutrophil interactions have yet to be unraveled, but a working model of Yersinia-neutrophil interactions in infected tissues is beginning to emerge. Neutrophils are recruited rapidly to infected tissues, albeit sometimes after a delay relative to recruitment by a yop mutant. This delay suggests that very early Yop injection into resident tissue cells may modulate chemokine and cytokine release, delaying neutrophil recruitment. However, neutrophils rapidly become the most proximal cell type to Yersinia, surround them, and in turn are efficiently injected with Yops by Yersinia. Higher expression of stress response genes and T3SS promoters is observed in line with increases in copy number of the pYV plasmid and increases in ion sequestration genes in host cells. Injection disarms neutrophils without triggering significant cell death. Rather, the immediately adjacent cells adopt a pro-survival and low migration state that fails to reduce bacterial growth.
Understanding how different Yops collaborate to modulate neutrophil activities in tissue infection is ongoing. Because it is easier to obtain human primary neutrophils in large quantities relative to mice, most studies investigating Yersinia-neutrophil interactions have used isolated human neutrophils 24,29,[82][83][84][85][86] . It is important to be aware that murine and human neutrophils have notable differences and thus findings in one system cannot be inferred to occur in another 87,88 . Furthermore, human neutrophils are typically harvested from peripheral blood whereas mouse neutrophils are collected from either the bone marrow or the peritoneal cavity after being elicited by an irritant, such as casein or thioglycolate. Therefore, these cells are in different stages of development and have the potential to respond to bacteria differently. Nonetheless, studies in either system are important to understand how Yersinia, through manipulation of neutrophils, thwarts the orchestrated mammalian host cell response to infection at an organismal, tissue, cellular, and molecular level.

Author contributions
The author reviewed the literature and drafted and edited the review.

Grant information
The author was supported by National Institutes of Health (NIH) grants NIAID R01 AI113166, AI107055, R21 AI128093, and U19 AI131126.
The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.