Yersinia pseudotuberculosis YopH targets SKAP2-dependent and independent signaling pathways to block neutrophil antimicrobial mechanisms during infection.

Yersinia suppress neutrophil responses by using a type 3 secretion system (T3SS) to inject 6–7 Yersinia effector proteins (Yops) effectors into their cytoplasm. YopH is a tyrosine phosphatase that causes dephosphorylation of the adaptor protein SKAP2, among other targets in neutrophils. SKAP2 functions in reactive oxygen species (ROS) production, phagocytosis, and integrin-mediated migration by neutrophils. Here we identify essential neutrophil functions targeted by YopH, and investigate how the interaction between YopH and SKAP2 influence Yersinia pseudotuberculosis (Yptb) survival in tissues. The growth defect of a ΔyopH mutant was restored in mice defective in the NADPH oxidase complex, demonstrating that YopH is critical for protecting Yptb from ROS during infection. The growth of a ΔyopH mutant was partially restored in Skap2-deficient (Skap2KO) mice compared to wild-type (WT) mice, while induction of neutropenia further enhanced the growth of the ΔyopH mutant in both WT and Skap2KO mice. YopH inhibited both ROS production and degranulation triggered via integrin receptor, G-protein coupled receptor (GPCR), and Fcγ receptor (FcγR) stimulation. SKAP2 was required for integrin receptor and GPCR-mediated ROS production, but dispensable for degranulation under all conditions tested. YopH blocked SKAP2-independent FcγR-stimulated phosphorylation of the proximal signaling proteins Syk, SLP-76, and PLCγ2, and the more distal signaling protein ERK1/2, while only ERK1/2 phosphorylation was dependent on SKAP2 following integrin receptor activation. These findings reveal that YopH prevents activation of both SKAP2-dependent and -independent neutrophilic defenses, uncouple integrin- and GPCR-dependent ROS production from FcγR responses based on their SKAP2 dependency, and show that SKAP2 is not required for degranulation.


Introduction
The three pathogenic species of Yersinia that are associated with human infections are the causative agent of the pneumonic and bubonic plague, Yersinia pestis, and the gastrointestinal pathogens Y. pseudotuberculosis (Yptb) and Y. enterocolitica [1][2][3]. These species harbor a 70kb virulence plasmid encoding a Type 3 Secretion System (T3SS) that injects a set of 6-7 effector proteins, called Yops (Yersinia outer proteins), into mammalian cells [1,[4][5][6][7]. Yops are essential for virulence and interfere with many antimicrobial responses in immune cells, yet their functions in tissue infections are not completely understood [5,[8][9][10][11]. Infection with Yptb in tissues is characterized by formation of Yptb microcolonies that directly interface with neutrophils that have migrated to sites of infection [12][13][14]. Consistent with their physical proximity to Yptb microcolonies in tissues, neutrophils are typically the cells that are most frequently injected with Yops [15][16][17]. Yptb encodes two bacterial adhesins, Invasin and YadA, which bind either directly or indirectly to β1 integrins expressed on neutrophils [18,19]. This binding facilitates Yop-injection into neutrophils during tissue infection, whereas in the absence of Yops, the binding via these adhesins triggers phagocytosis of Yptb [15,20,21]. In addition to integrin receptor-mediated interactions, Yersinia activates other receptors on neutrophils during infection, including G-protein coupled receptors (GPCRs) [22][23][24][25]. Activation of integrin receptors and GPCRs in turn leads to a variety of neutrophil antimicrobial responses including reactive oxygen species (ROS) production and proteolytic enzyme release from granule stores, also known as degranulation.
SKAP2 is a cytosolic adaptor protein comprised of a N-terminal coiled-coil dimerization domain region, a proximal pleckstrin homology domain (PH), an interdomain containing two tyrosine-based signaling motifs and a C-terminal Src homology 3 (SH3) domain that mediates its binding to PRAM-1 and ADAP [45][46][47][48]. SKAP2 is a substrate for Scr kinases, and was recently found to be essential in regulating TNFα-, selectin-and CXCL1-mediated integrin activation [49]. In addition, SKAP2 regulates neutrophil functions including migration to sites of sterile inflammation, neutrophil extracellular trap (NET) formation, and integrin-mediated ROS production [49].
Here we investigate whether YopH is required to prevent ROS production during Yptb infection of mouse tissues and whether SKAP2 must be inactivated for a ΔyopH mutant to survive and thus is an important YopH target. We find that while inactivation of SKAP2-regulated pathways is a key function of YopH, YopH has additional functions independent of SKAP2 that enable further inhibition of both neutrophil ROS production and degranulation. Our studies also shed light on SKAP2 function, demonstrating its essential role in neutrophil ROS activation after integrin or GPCR-stimulation, but not in degranulation or migration of neutrophils to infected tissues sites.

Growth of a ΔyopH mutant is restored in mice lacking a functional NADPH oxidase complex and YopH is sufficient to block ROS production from murine neutrophils
Our previous work [36] demonstrated that several signaling proteins that are important for ROS production downstream of integrin signaling, including SKAP2, PRAM-1, SLP-76, Vav1, and PLCγ2 [49-54], were hypo-phosphorylated in YopH-injected neutrophils isolated from infected spleens. However, whether YopH plays a role in supporting Yptb growth during infection by suppressing ROS production is unknown. To test this, C57BL/6J and gp91 phox-/mice were infected intravenously (I.V.) or intratracheally (I.T.) with an equal mixture of IP2666 WT-Yptb and a ΔyopH mutant. Two days post-infection spleens and lungs were harvested and the competitive index (C.I.) was determined. By contrast with previous work on ΔyopE [67], the ΔyopH mutant competed significantly better in the spleens and lungs of gp91 phox-/mice than C57BL/6J mice (Fig 1A and 1B). The total CFU of Yptb recovered from gp91 phox-/mice were comparable to numbers recovered from C57BL/6J mice, but the numbers of the recovered ΔyopH mutant rose significantly in the gp91 phox-/mice (S1A Fig). These results support the idea YopH inactivates one or more pathways leading to the generation of ROS and thereby protects Yptb from superoxide-mediated killing during tissue infection.
To examine the ability of YopH to block ROS production in neutrophils, extracellular ROS production was measured after Yptb infection of neutrophils isolated from bone marrow (BM) of BALB/c mice (S1B Fig). BM neutrophils were infected with WT-Yptb, ΔyopH, Δ5 (a mutant lacking 5 effector Yops, H, E, O, M, J) or Δ5+pYopH, a strain expressing only YopH at native levels from an arabinose inducible plasmid (S1 Table). There were no differences in amount of YopH secretion between Δ5+pYopH and WT-Yptb (S2 Fig). YopH was necessary to block ROS production induced by WT-Yptb infection (Fig 1C and 1D) because infection with the ΔyopH strain generated high levels of ROS that were equivalent to infection with the Δ5 strain ( Fig 1D). In addition, YopH was the only Yop needed to block ROS released by BM neutrophils (Fig 1E and 1F) because infection with Δ5+pYopH blocked ROS production as much as WT-Yptb infection ( Fig 1F). Combined, these results indicate that YopH blocks production of extracellular ROS by neutrophils induced by Yptb infection.

Growth of a ΔyopH mutant in the spleen is partially restored in the absence of SKAP2 and is further enhanced in neutropenic mice
SKAP2 is essential for integrin-stimulated ROS production in neutrophils [49], is dephosphorylated in neutrophils injected with YopH during tissue infection [36], and associates with , (E-F) WT-Yptb +pBAD, Δ5+YopH (a strain expressing only YopH at native levels from an arabinose inducible plasmid),or Δ5+pBAD, and monitored for extracellular ROS production using isoluminol chemiluminescence assay. (D and F) Total chemiluminescence was determined by computing the area under the curve (AUC) using GraphPad Prism version 7. Mean ±SD of a representative experiment from 2-3 independent experiments done in 2-3 replicates is presented. Statistical significance was calculated using (A) the Mann-Whitney t-test, (B) one-way ANOVA followed by Tukey's Post-test after log 10 transformation of the data, or (D, F) one-way ANOVA followed by Tukey's Post-test for area under the curve (AUC).  [15,20,68,69] and that YopH blocks ROS production (Fig  1), led us to evaluate whether SKAP2 influences Yptb survival in the presence and absence of YopH. BALB/c (WT) and BALB/c-Skap2KO (Skap2KO) mice were co-infected I.V. with an equal mixture of WT-Yptb and a ΔyopH mutant. Three days post-infection spleens were collected, and the C.I. was determined. The ΔyopH mutant colonized Skap2KO mice almost 10-fold more efficiently compared to WT mice while overall bacterial burdens were similar (Fig 2A, S3A Fig), suggesting that SKAP2 may function in a YopH sensitive pathway(s) that serves to limit growth of a ΔyopH mutant in vivo. Although, Skap2KO mice have normal levels of circulating neutrophils [70], their extravasation into tissues occurs at significantly lower rates in a sterile model of inflammation [49]. Thus, it was possible that the improved survival of the ΔyopH mutant in the Skap2KO mice was due to reduced rates of neutrophil recruitment to infected sites. Analysis of the kinetics of neutrophil recruitment in the spleens of Yptb-

PLOS PATHOGENS
YopH inactivates SKAP2 and ROS to promote Yersinia growth in tissues infected WT and Skap2KO mice revealed no differences in the recruitment of neutrophils and bacterial burdens were comparable between the WT and Skap2KO mice at the tested time points (Fig 2B-2K, S3B Fig). Collectively, these results support the idea that the restoration of ΔyopH mutant growth in Skap2KO mice (Fig 2A) is due to a functional defect in Skap2-deficient neutrophils that increases the survival of a ΔyopH mutant. This suggests that SKAP2 mediates signaling from receptors stimulated by Yptb infection and participates in a subset of relevant antimicrobial neutrophil functions normally targeted by YopH.
The ΔyopH mutant was still defective for survival compared to WT-Yptb in Skap2KO mice suggesting that Skap2KO neutrophils may retain some killing activities against the ΔyopH mutant. To determine whether YopH targets pathways, which are independent of SKAP2, in neutrophils, additional cohorts of mice from each genotype were subjected to 1A8 antibodymediated depletion of neutrophils. Growth of the ΔyopH mutant was significantly higher in neutrophil-depleted Skap2KO mice compared to Skap2KO mice (Fig 2A), indicating that Skap2KO neutrophils retain some antimicrobial activities and that YopH targets additional SKAP2-independent pathways in neutrophils. Notably, ΔyopH growth was comparable in neutrophil-depleted WT and neutrophil-depleted Skap2KO mice, highlighting that the lack of SKAP2 activity in other innate immune cells does not enhance survival of the ΔyopH mutant in the absence of neutrophils. Thus, YopH targets both SKAP2-dependent and SKAP2-independent antimicrobial processes.

YopH blocks SKAP2-dependent and -independent ROS production by neutrophils
Since ΔyopH growth was restored in gp91 phox-/mice we evaluated the ability of YopH to block ROS production downstream of three major receptor classes (integrin receptors, G-protein coupled receptor (GPCR) and FcγR) known to activate the NADPH oxidase complex, two of which may be stimulated during Yptb infection [18,19,[22][23][24][25]. BM WT neutrophils were infected with WT-Yptb, ΔyopH, Δ5, or Δ5+pYopH and stimulated with poly-RGD (a polymer of arginine-glutamate-aspartate) to trigger integrin receptors, primed with E. coli lipopolysaccharide (LPS) and then stimulated with formyl-methionyl-leucyl-phenylalanine (fMLP) to trigger GPCR, or stimulated with IgG immune complex (IC) to trigger Fcγ receptor (FcγR) (Fig 3A-3C) [49, 55,[71][72][73][74][75][76]. In all cases, WT-Yptb and Δ5+pYopH inhibited ROS production similarly demonstrating that YopH was sufficient for ROS inhibition. In addition, high levels of ROS were detected in BM neutrophils infected with Δ5 or ΔyopH (Fig 3A-3C) demonstrating that YopH was also necessary for ROS inhibition when other Yops are present. Unlike stimulation with integrin or Fcγ receptors (Fig 3A and 3C), ROS production from ΔyopHinfected from BM neutrophils that were exposed to LPS and fMLP was less than Δ5-infected cells, indicating that another Yop acts to partially reduce ROS production in the absence of YopH downstream of GPCR stimulation ( Fig 3B). Further, ROS production after infection with Δ5 and exposure to LPS and fMLP was significantly higher than after exposure to LPS and fMLP alone ( Fig 3B). This suggests that infection with Δ5 augments ROS production through activation of other receptors, possibly due to bacteria-mediated crosslinking of and signaling by integrin receptor in the absence of Yops (Fig 1C-1F) [18,19].
Given our findings that Skap2KO mice retain some killing activities against the ΔyopH mutant (Fig 2A), we next tested the ability of Skap2KO neutrophils to produce ROS when stimulated through integrin receptor, GPCR and FcγR (Fig 3D-3F). Skap2KO neutrophils failed to generate high levels of ROS after poly-RGD and LPS + fMLP exposure (  To verify that defects in ROS production resulted from a receptor-mediated signaling defect and not an inability of Skap2KO neutrophils to activate the NADPH oxidase complex, WT and Skap2KO neutrophils were stimulated with phorbol myristate acetate (PMA), which bypasses receptor-mediated activation and directly activates protein kinase C (PKC), to trigger assembly of the NADPH oxidase complex [55,73,74]. Indeed, PMA-stimulated WT and Skap2KO neutrophils produce equivalent levels of ROS (Fig 3D-3F). Together, these results indicate that a functional NADPH oxidase complex can assemble in Skap2KO neutrophils, but that SKAP2 signaling is required for integrin receptor and GPCR-mediated YopH is necessary to block ROS from all three receptors while Skap2KO neutrophils fail to produce ROS following integrin receptor and GPCR but not FcγR stimulation. Respiratory burst of (A-C) WT BM neutrophils that were either uninfected or infected with IP2666 WT-Yptb, Yptb-Δ5+pBAD, Yptb-Δ5+pYopH or Yptb-ΔH at a MOI of 20:1, or (D-F) BM neutrophils from WT and Skap2KO mice was measured using an isoluminol chemiluminescence assay. BM neutrophils were added to (A, D) poly-RGD coated surface for 30 min, (B, E) primed with 10μg/ml LPS for 20 min followed by stimulation with 1μM fMLP for 10 min, or (C, F) added to IC coated surface for 30 min and monitored for extracellular ROS production. (D-F) PMA (1μM) was added to WT and Skap2KO neutrophils as a positive control. Total chemiluminescence was determined by computing the area under the curve for the duration of the experiment for each condition after subtracting the uninfected, unstimulated control plated on FBS. Statistical significance was calculated using (A-C) one-way ANOVA followed by Tukey's Post-test or (D-F) Student t-test. Only statistically significant comparisons are shown. Results presented are the mean ± SD of triplicate measurements. Data is a representative of at least three independent experiments done in triplicates. https://doi.org/10.1371/journal.ppat.1008576.g003

PLOS PATHOGENS
YopH inactivates SKAP2 and ROS to promote Yersinia growth in tissues ROS production in neutrophils. Since these two receptors may be stimulated during infection with Yptb, these results suggest that the inability of Skap2KO neutrophils to generate ROS contributes to the partial growth restoration of ΔyopH observed during in vivo infections (Fig 2A). While FcγR is unlikely to be activated by Yptb during infection of a naïve host, this result nonetheless raises the possibility that neutrophils have other receptors that may recognize Yptb, signal and function independently of SKAP2, leading to ROS generation in the absence of YopH. This superoxide production could potentially contribute to partial elimination of the ΔyopH mutant in Skap2KO mice.

YopH targets degranulation downstream of integrin receptor, GPCR and FcγR, and degranulation is SKAP2-independent
To investigate the ability of YopH to inhibit additional antimicrobial activities and evaluate whether these activities are controlled by SKAP2, we assessed YopH-dependent inhibition of degranulation in WT and Skap2KO neutrophils. In human and murine neutrophils, YopH partially contributes to the inhibition of neutrophil degranulation as mutants infected with either ΔyopH or ΔyopE are partially defective in degranulation [40,41]. The ability of YopH to prevent tertiary degranulation was tested by infecting WT BM neutrophils with WT-Yptb, ΔyopH, Δ5 or Δ5+pYopH and stimulating them with poly-RGD, fMLP, or IC (Fig 6A-6C). The amount of matrix metalloproteinase-9 (MMP-9), a component of tertiary granules, released in the cell-free supernatants was quantified by ELISA. Western blot analysis with anti-Rho-GDI antibody confirmed that the observed release of MMP-9 was not due to neutrophil lysis (S6A Fig). Degranulation was completely inhibited by WT-Yptb after stimulation with poly-RGD ( Fig 6A) compared to uninfected controls, and was partially blocked by WT-Yptb after exposure to LPS+fMLP or IC (Fig 6B and 6C), indicating the combined actions of the

PLOS PATHOGENS
YopH inactivates SKAP2 and ROS to promote Yersinia growth in tissues Yops are insufficient to halt degranulation under receptor stimulation (Fig 6B and 6C). Interestingly, degranulation was higher when cells were stimulated with either fMLP or IC and infected with Δ5 compared to stimulation of receptors in the absence of infection (S6B Fig), suggesting that simultaneous Δ5 binding via integrin receptor and stimulation of GPCR or FcγR enhances degranulation levels. YopH was sufficient to block tertiary granule release as Δ5+pYopH-infected neutrophils released MMP-9 to the same level as WT-Yptb downstream of integrin, GPCR, and FcγR (Fig 6A-6C). Consistent with previous results [40, 41], significantly more degranulation was observed after infection with ΔyopH compared to WT-Yptb, but significantly less compared to infection with Δ5 (Fig 6A-6C). This demonstrates that YopH plays a role in reducing degranulation in murine neutrophils but is not the sole factor involved. Of note, SKAP2 was dispensable for MMP-9 release with approximately 6 ng/ml, 20 ng/ml and 25 ng/ml MMP-9 released after integrin, GPCR and FcγR stimulation, respectively (Fig 6D-6F). Supernatants of PMA-stimulated neutrophils showed 50% degranulation compared to triton X-100 treatment (S6C Fig). Combined, these results indicate that SKAP2 is dispensable for tertiary degranulation and suggests that SKAP2KO neutrophils may control growth of the ΔyopH mutant through degranulation.

Skap2KO neutrophils are defective for phagocytosis of Yptb
Finally, given that YopH participates in the ability of Yersinia to resist phagocytosis in isolated human neutrophils [29, 37, 39, 81, 82], we evaluated the ability of Skap2KO neutrophils to phagocytose WT-Yptb, ΔyopH, or a strain defective for the type 3 secretion system, ΔyscF. Using a gentamicin protection assay, we found that all three strains were phagocytosed at a

PLOS PATHOGENS
YopH inactivates SKAP2 and ROS to promote Yersinia growth in tissues similar, low level by Skap2KO BM neutrophils and at significantly lower levels compared to that of WT BM neutrophils (Fig 7). This suggests that Skap2KO BM neutrophils are defective for invasin-β1-mediated phagocytosis regardless of the presence of YopH or the T3SS. As expected, a significantly higher percentage of ΔyscF was phagocytosed in WT BM neutrophils than the WT-Yptb and ΔyopH mutant (Fig 7). The ΔyopH mutant was consistently phagocytosed by WT BM neutrophils more frequently than WT-Yptb, but this difference was not statistically significant suggesting that YopH has a modest role in preventing phagocytosis in WT BM neutrophils.
In summary, YopH inhibits ROS production, reduces degranulation and has a modest effect on phagocytosis of WT-Yptb. Skap2KO neutrophils have defects in ROS production and phagocytosis, but retain the ability to migrate to sites of Yptb infection and degranulate. Collectively, these results establish a role for SKAP2-mediated neutrophil signaling and its antimicrobial function towards Yptb and reveal how YopH acts to counter it.

Discussion
By exploring the interplay between Yptb effector YopH and the neutrophil adaptor protein, SKAP2, during tissue infection, we demonstrate the role YopH plays in protecting Yptb from neutrophil-mediated ROS production as well as the role SKAP2 plays in mediating neutrophil antimicrobial functions. The finding that inactivation of SKAP2 partially restores the growth of a ΔyopH mutant indicates that one or more SKAP2-controlled pathways must be inactivated

PLOS PATHOGENS
YopH inactivates SKAP2 and ROS to promote Yersinia growth in tissues What is YopH doing during infection? Our results indicate that YopH plays at least two important roles in blocking antimicrobial functions of neutrophils: inhibiting ROS production and reducing degranulation (Fig 8). A ΔyopH mutant survived significantly better in gp91 phox-/mice than in WT mice, and YopH fully blocked ROS production downstream of integrin receptor, GPCR and FcγR stimulation in neutrophils. Furthermore, YopH interfered with the phosphorylation cascade triggered downstream of FcγR stimulation, including phosphorylation of Syk, SLP-76, PLCγ2 and ERK1/2. Syk was not identified as a YopH target in a previous study; however, in that case, neutrophils were activated with bacteria alone [36] raising the possibility that Syk may be inactivated by YopH in vivo if it is activated through concurrent stimulation of several receptors at infectious loci. During tissue infection, other SKAP2-independent pathways may be triggered to generate ROS, which YopH can block, thus a ΔyopH mutant is susceptible to the ROS produced. In addition, in some cases NETosis depends on ROS production [65]. So our findings that YopH is critical to block ROS production support the idea that YopH may also block NETosis observed after infection of human neutrophils with a TT3S-minus strain but not WT Yptb strain [88]. We also found that degranulation was not dependent on SKAP2 after integrin and GPCR activation, receptors triggered by Yersinia  [18, 19, 22-25]. Thus, degranulation could contribute to the partial elimination of the ΔyopH mutant observed in tissue infection of the Skap2KO mice (Fig 2A), because YopH was sufficient to reduce MMP-9 release from all three receptors tested (Fig 6A-6C). In addition, others have shown that YopH and YopE block degranulation-dependent release of lactoferrin, a  After integrin engagement, YopH inhibits signal propagation that is required for the activation of the NADPH oxidase complex and degranulation while SKAP2 is essential for ROS and phagocytosis and for maximal ERK1/2 phosphorylation, but not for degranulation or phosphorylation of Syk, SLP-76 and PLCγ2. (B) After GPCR stimulation of LPS-primed neutrophils with fMLP, YopH inhibits ROS and degranulation, where SKAP2 is required for ROS production, but not degranulation. (C) YopH inhibits ROS and degranulation after FcγR stimulation and reduces phosphorylation of Syk, SLP-76, PLCγ2 and ERK1/2, but SKAP2 is not required for ROS production, degranulation, nor phosphorylation of these signaling proteins. https://doi.org/10.1371/journal.ppat.1008576.g008

PLOS PATHOGENS
YopH inactivates SKAP2 and ROS to promote Yersinia growth in tissues are subjected to numerous innate signals, such as integrin stimulation from Yptb binding, and LPS and fMLP shed from Yptb. While it is tempting to speculate that YopH prevents phagocytosis in tissue infection as has been observed in isolated phagocytes [39], internalized ΔyopH was recovered in infected tissues at levels comparable to WT-Yptb [36], suggesting that YopH does not contribute significantly to anti-phagocytosis activities of Yptb in this model of Yptb tissue infection.
Precedence for more than one Yop blocking similar antimicrobial functions by targeting different proteins in receptor-mediated signaling cascades is abundant [5] and suggests that Yptb uses its arsenal of 6-7 Yops to block antimicrobial functions triggered by activation of different receptors and/or to alter antimicrobial responses of different cellular microdomains. Here, YopH was both necessary and sufficient to block ROS production downstream of two receptor classes; integrin and FcγR, however, its absence only partially crippled GPCR-mediated ROS production after infection with a ΔyopH mutant. These finding are consistent with the essential role of YopH in blocking integrin-mediated calcium flux, but not GPCR-mediated calcium release [38], and show that other Yops can compensate after activation of certain receptors. One Yop that could be acting with YopH to block GPCR-mediated ROS is YopO, as it has been reported to target Rho, Rac, and Gαq-dependent signaling pathways of GPCR [89][90][91][92][93][94][95]. In contrast to the near complete restoration of growth of a ΔyopH mutant in gp91 phox-/mice reported here, a ΔyopE mutant was not restored for growth although a chimeric YopE that retained its Rac1 GAP but was deficient in other GAP activities was significantly restored for growth in gp91 phox-/mice compared to C57BL/6J mice [67]. Thus, YopH appears to play a major role in protecting Yptb from ROS production during tissue infection, in part through its ability to block SKAP2-controlled pathways, while YopE may play a more prominent role at the phagocytic cup in inactivating Rac1 and RhoG rather than in inactivating Rac2 [67,[96][97][98][99][100].
Neutrophils kill bacteria using several different mechanisms including phagocytosis, ROS production, degranulation and NET production [101][102][103]. This raised the question of what anti-microbial functions are controlled by SKAP2, as Skap2KO mice partially restored ΔyopH growth. Together, our findings uncouple several key integrin-and GPCR-dependent antimicrobial responses from FcγR-dependent responses by virtue of their SKAP2-dependency for ROS production, and show that this adaptor is critical for some, but not all integrin-and GPCR-driven antimicrobial functions in neutrophils (Fig 8). SKAP2 may be part of a phospho-relay and/or could act a scaffolding function needed for full ERK1/2 phosphorylation [49]. ERK1/2 is required for phosphorylation of the p47 phox and p67 phox subunits of the NADPH oxidase complex in response to various stimuli in human neutrophils and therefore for ROS production [76][77][78][79]. The scaffolding function of SKAP2 may be essential to form stable signaling complexes downstream of integrin receptor and/or GPCR, but not FcγR. Hence, the defect of ROS in the integrin pathway may not be due to alteration of the phosphorylation state or activity of the proximal signaling proteins, but rather due to displacement of these proteins from the plasma membrane in the absence of SKAP2. In line with this hypothesis, the membrane localization of the SKAP2 binding partner, WASp, but not its phosphorylation, is abolished in Skap2KO neutrophils. This suggests that phosphorylation alone is not the only critical mediator of its activity, and possibly other SKAP2-associated-protein activities [49].
The fact that SKAP2 is critical for GPCR mediated ROS production was surprising, as the majority of the proteins known to interact with or predicted to be in the same signaling pathway as SKAP2 in neutrophils and other immune cells are dispensable downstream of GPCR; these include Syk, SLP-76, PRAM-1, and PLCγ2 [51, 53,54,104], suggesting that SKAP2 signals through a different signaling pathway downstream of GPCR (Fig 8B). In addition, SKAP2 is not required for calcium flux in neutrophils following chemokine stimulation [49].

PLOS PATHOGENS
YopH inactivates SKAP2 and ROS to promote Yersinia growth in tissues However, previous studies employed cytochalasin B to amplify the fMLP signal [51, 53,54,104], rather than LPS to prime neutrophils. Therefore, SKAP2 could be required for signaling pathways downstream of TLR4 receptor and/or downstream of GPCR in ROS production. Consistent with these findings, the proliferative response to LPS in Skap2KO B cells is strongly impaired, highlighting the critical role of SKAP2 in TLR4-mediated activation [70]. On the other hand, LPS-induced cytokine production was normal in Skap2KO dendritic cells, supporting the different functional requirement of SKAP2 in LPS activation in different immune cell types and functions [105]. Further, it is critical to evaluate these pathways as well as the role of SKAP2 in antimicrobial functions in human neutrophils to determine if these findings are applicable to humans as mouse and human neutrophils differ [55,75,103,106].
While SKAP2 is necessary for limiting colonization of ΔyopH mutant, it does not appear to limit neutrophil migration to spleens in our infection models. This contrasts with the defect observed in a sterile inflammation model of hepatic necrosis where Skap2KO neutrophils were defective for migration [49]. It is noteworthy that the time courses of the two studies were different. In the sterile inflammation model, neutrophil recruitment was first observed after 30 min of tissue injury and infiltration reached maximum by 4 hours, whereas in our infection model no migration was observed for the first 24 hours in the spleen. Consistent with our results, neutrophils lacking SLP-76, which functions downstream of integrin receptors, also showed normal migration in vivo in response to Staphylococcus aureus infection, yet were defective in a sterile inflammation model of acute kidney injury [51,107]. The difference in SKAP2-dependency in the context of microbial infection versus sterile inflammation may be due to multiple factors, including different chemokines released upon infection, a different adaptor protein compensating for the absence of SKAP2 in the infection model, or additional SKAP2-independent receptors activated during infection for tissue extravasation [51,107].
In conclusion, Yptb employs YopH to target both SKAP2-dependent and independent signal-transduction pathways to cripple antimicrobial functions of murine neutrophils. Skap2KO neutrophils migrate efficiently to infected tissues, but only partially control the growth of a ΔyopH mutant. SKAP2 is required for ROS production after integrin and GPCR stimulation but not for degranulation, indicating that different antimicrobial functions can be triggered by activation of the same receptor but are not all dependent on SKAP2. However, since all these activities are blocked by YopH, future work is focused on uncovering the additional signaltransduction pathways that are essential YopH targets during tissue infection.

Bacterial strains and growth conditions
Strains used in this work are listed in S1 Table. All mouse infections used Yptb IP2666 or derivatives while some neutrophil assays were performed with Yptb YPIIIpIB1 and derivatives as described in Figure Legends. The kanamycin resistant (Kan R ) IP2666ΔyopH mutant was constructed by introducing a yopH deletion into IP2666-NdeI-Kan R as previously described [30]. Yptb were grown overnight at 26˚C in 2xYT media with aeration. For animal infections, overnight cultures were diluted 1:40 in 2xYT media incubated at 26˚C with aeration for 8 hours, after which they were diluted 1:100 and incubated overnight with aeration and the OD 600 measured. The bacteria were diluted in PBS to achieve the required input dose for intravenous (I. V.) or intratracheal (I.T.) inoculation. For infection of neutrophils, Yptb were grown under Yop-inducing conditions as described [36]. Strains containing pBAD plasmids were grown in presence of 20 μg/ml chloramphenicol and 50 mM arabinose was added to the culture prior to the shift to 37˚C and maintained in medium with antibiotics and inducer throughout the assays. OD 600 of each culture was measured and strains were diluted in KRP buffer (Krebs-

PLOS PATHOGENS
YopH inactivates SKAP2 and ROS to promote Yersinia growth in tissues Ringer-Phosphate: PBS-Corning; Cat# 21040CV + 1 mM CaCl 2 + 1.5 mM MgCl 2 ) with 5 mM glucose or 5 mM arabinose depending on bacterial strains used in infections. BM neutrophils were infected at the indicated MOI.  week old BALB/c (Taconic labs), BALB/c-Skap2KO [48], C57BL/6J (Jackson labs), and C57BL/6J gp91 phox-/-(Jackson labs) mice were used for infections. Competition infections were performed using an equal mixture of unmarked and Kanamycin-resistant marked IP2666 strains. C57BL/6J and gp91 phox-/were I.V. or I.T. infected with 1x10 3 or 5x10 3 colony forming units (CFU), respectively, with a 1:1 mixture of WT-Yptb IP2666 and YptbΔyopH-Kan R . Two days post infection spleens and lungs were isolated, weighed, homogenized and plated on L agar containing 0.5μg/mL irgasan. BALB/c and Skap2KO mice were inoculated I.V. with 1x10 3 WT-Yptb IP2666 and Kan R marked ΔyopH and mice were sacrificed three days post-infection. The competitive index (C.I.) (C.I. = (CFU ΔyopH /CFU WT ) output /(CFU ΔyopH /CFU WT ) input ) was calculated as described [108] with the following modifications. In some cases, the CFU for each strain were determined by subtracting the number of yopH-Kan R colonies from the total number of colonies recovered from L-irgasan plates to increase the limit of detection. For neutrophil (Ly6G + ) cell depletions, mice were intraperitoneally injected with 100μl of 50μg/mL of 1A8 antibody (α-Ly6G + ) (Fisher; Cat # bdb551459) 16 hours prior to and 24 hours post-infection. Depletion was confirmed by staining spleen homogenates with α-GR1 + and α-CD11b + and analysis by FACS as previously described [21]. For determining the migration kinetics of neutrophils, 8 week-old female BALB/c and Skap2KO were either mock infected or IV infected with 1x10 3 WT-Yptb-GFP strain (chloramphenicol, Cm R ). Spleens were collected at 24, 33, 48 hours post-infection for CFUs and immunohistochemistry.

Immunofluorescence and image analysis
Spleen tissues collected at 24, 33 and 48 hours were fixed in 4% paraformaldehyde for 4 hours at room temperature. Tissues were frozen-embedded in Sub Xero freezing media (Mercedes Medical; Cat # MER 5000) and cut into 10 μm sections using a cryostat microtome. Sections were permeabilized using 100% methanol followed by blocking with 2% BSA in PBS for 1 hour at room temperature. Ly6G + cells were detected by staining sections with 1:100 rat antimouse monoclonal Ly6G (clone 1A8) antibody for 1 hour, washing 3 times with PBS, then incubating with 1:500 goat anti-rat Alexa Fluor 594 antibody (Invitrogen Cat# A-11007) and 1:1000 DAPI (Invitrogen) for 1 hour, and washing 3 times with PBS. Finally, samples were mounted on glass slides using ProLong Gold antifade (Invitrogen, Cat# P36930) and cured overnight in the dark at room temperature. Tissues were imaged using a 40X objective on a Nikon A1R confocal microscope (Nikon Instruments Inc.), with 0.5 μm z-steps using a filter set for DAPI (Ex/Em:350/470nm), GFP (Ex/Em:488/525nm) and mCherry (Ex/Em:561/ 595nm). Volocity software 6.3 was used for 3D reconstruction image quantification. For quantification of the microcolony size by volume, uninfected control images were used to set the threshold of the GFP intensity signal and objects smaller than 1μm 3 were excluded. Volume was determined by the summation of the individual signals in the green channel after applying the threshold limits. Quantification of neutrophils was determined by determining the sum of the red fluorescent signal intensity (mCherry channel).

Neutrophil isolation
BM neutrophils were isolated as previously described [36]. Briefly, BM was harvested from tibia and femur bones in HBSS without Ca 2+ or Mg 2+ (HBSS-/-). Red blood cells were lysed

PLOS PATHOGENS
YopH inactivates SKAP2 and ROS to promote Yersinia growth in tissues using sterile endotoxin free HyClone cold water (GE Heathcare; Cat# SH30529.FS) followed by addition of HBSS-/-to restore normal tonicity. Neutrophils were isolated using a three step, 55%, 65%, 75%, Percoll density gradient and centrifuged at 480 x g for 30 min without brakes. The cells at the 65-75% interface were collected, washed 2 times with HBSS-/-, and rested at room temperature for one hour. Purity of neutrophils ranged from 82-89% by FACS analysis with anti-CD11b, anti-Ly6C and anti-Ly6G antibodies (BioLegend). Neutrophils were then re-suspended in KRP neutrophil buffer to the desired concentration incubated for 20 min at room temperature and shifted to 37˚C in 5% CO 2 for 10-15 min prior to initiation of the experiment.

Neutrophil ROS assays
For stimulating integrin receptors, wells were coated with 15μg/mL poly-RGD (Sigma; Cat# F5022) for 2 hours at 37˚C, and then washed 2 times with PBS. To stimulate FcγR, wells were coated with 20μg/ml human serum albumin (Sigma; Cat # A9511) for 1 hour at 37˚C, washed 2X with PBS, incubated with 10% FBS for 30 min, washed 2 times with PBS, then incubated with a 1:400 dilution of rabbit anti-human serum albumin (Sigma; Cat # A0433) in 10% FBS for 1 hour at 37˚C, followed by washing 2 times with PBS. Control wells were coated with 10% FBS for 30 min, then washed 2 times with PBS. Assays for GPCR receptor were performed in the absence of Mg 2+ to decrease the possibility of integrin dependent responses and surfaces were coated with 10% FBS to minimize adhesion to mimic cells in suspension. For stimulating GPCR, BM neutrophils were suspended in KRP without Mg 2+ and primed with 10μg/mL LPS (Sigma; Cat# L4516) for 20 min then added to 10% FBS coated wells without centrifugation to maintain cells in suspension followed by stimulation using 1μM fMLP (Sigma; Cat # F3506). BM neutrophils suspended in KRP buffer at 37˚C were loaded with 50 μM isoluminol (Sigma; Cat# A8264) and 15 U/ml of HRP (Sigma; Cat # P6782) as previously described [109] and 2 x10 5 neutrophils/well were added to 96-well white high-bound plates (Thermo Fisher; Cat# 3922) coated with 10% FBS or different ligands and centrifuged at 190 x g for 1 min. For infections, bacteria resuspended in KRP buffer were added to wells at a MOI of 20:1 simultaneously with neutrophils when stimulating integrin and Fcγ receptors. The bacteria and neutrophils were spun down on the ligand coated plates for 1 min to uniformly induce receptor binding. For GPCR stimulation, bacteria were added to neutrophils with LPS priming and kept in suspension. PMA (1μM) (Sigma; Cat# P1585) was used as a positive control. Chemiluminescence was measured using a BioTek Synergy HT microplate reader. ROS detection using cytochrome c (Sigma; Cat # C-7752) was performed as previously described [109][110] with the following modifications. BM neutrophils (1X10 5 per 50 μl) were added to 50 μl 200μM cytochrome c in 96 well high-binding Immulon 4HBX plates (Fisher; Cat# 3855) coated with ligands or FBS. Absorbance at 550nm and 490nm were recorded and the difference used to calculate the nanomoles of superoxide produced using an extinction coefficient of 2.11x10 4 M -1 cm -1 for cytochrome c. The total amount of ROS for each condition was corrected by subtracting its superoxidase dismutase (SOD) control [109].
Neutrophil degranulation assay 2 x 10 5 BM neutrophils/well were plated on 96 well Immulon 4HBX plates coated with poly-RGD, IC or 10% FBS and incubated at 37˚C, 5% CO 2 . Infections were carried out at a MOI of 20:1 as described for ROS assays. After 3 hours, to determine the total amount of MMP-9, the control wells were treated with 0.01% triton X-100, without removing the supernatant, to lyse the cells and the supernatant collected. To determine the amount of released MMP-9, plates were placed on ice for 5 min to stop the reaction and then centrifuged for 10 min at 4˚C at 190

PLOS PATHOGENS
YopH inactivates SKAP2 and ROS to promote Yersinia growth in tissues x g. Supernatants were collected and stored at -80˚C until analysis by ELISA using DuoSet (R&D Systems; Cat # DY6718) for mouse MMP-9, or by western blot using an antibody against Rho-GDI (Cell Signaling; Cat# 2564S). The remaining cells in the wells were collected and lysed for western blot analysis to determine the total amount of Rho-GDI using an antibody against Rho-GDI. For comparisons between WT and Skap2KO neutrophils the absolute amount of MMP-9 detected in supernatants was reported for the stimulated receptor after subtracting the unstimulated (FBS-coated wells) control values. The amount of MMP-9 released after infection with Yptb strains was reported as fold-change compared to the control of unstimulated, uninfected wells.

Western blot analysis
Cells were incubated at 37˚C, 5% CO 2 for 5 min or 10 min for FcγR or integrin stimulation respectively. In some cases, BM neutrophils were infected with YPIII WT-Yptb or ΔyopH for 30 min and then spun down on the ligand coated surface. Cells were lysed in 1X Novex buffer and 2x10 6 cell equivalents were resolved on a 4-12% NuPAGE gel (Invitrogen; Cat# NP0335BOX) in MOPS buffer. Proteins were transferred to Immobilon-FL filters and subjected to Western blot analysis with the following primary antibodies at a dilution of 1:

Gentamicin protection assay
YPIII WT-Yptb, ΔyopH or ΔyscF strains were grown under Yop-inducing conditions and used to infect 1 x10 4 /well BM neutrophils at MOI of 10:1 in 96 well Immulon 4HBX plates coated with 10% FBS for 30 min at 37˚C, 5% CO 2 . Gentamicin was added to a final concentration of 100μg/ml for 60 min, and then cells were washed three times with PBS. BM neutrophils were centrifuged at 190 x g for 5 min between each wash. BM neutrophils were lysed with 100 μl 0.5% Triton-X-100 for 10 min followed by addition of 100μl 2xYT, and lysates were plated at different dilutions to determine the number of intracellular bacteria. Results are reported as the percent of the recovered CFU from gentamicin-treated cells divided by the non-gentamicin-treated control.

Ethics statement
This study was performed in accordance with the recommendations in the Guide for Care and Use of Laboratory Animals of the National Institutes of Health. The Tufts University Institutional Animal Care and Use Committee (IACUC) reviewed the study protocol under approval number B2015-35 and B2018-10.

PLOS PATHOGENS
YopH inactivates SKAP2 and ROS to promote Yersinia growth in tissues Supporting information S1 Fig. Yptb CFU in spleen of C57BL/6J or C57BL/6J gp91 phox-/mice and assessment of bone marrow neutrophil purity. (A) C57BL/6J or C57BL/6J gp91 phox-/mice were I.V. inoculated with a 1:1 mixture of IP2666 WT-Yptb and ΔyopH-Kan R . Spleen were collected 2 days post-infection and CFU for WT-Yptb and ΔyopH-Kan R was determined by plating on selective and non-selective plates. Statistical significance was calculated using one-way ANOVA with Sidak's multiple comparison test after log 10 transformation of data. (B) Bone marrow was harvested from the tibia and femur bones of mice and neutrophils were isolated using Percoll density gradient method. Cells at the 65-75% interface were collected and stained using with α-CD11b, α-Ly6G and α-Ly6C and analyzed by flow cytometry. Bone marrow isolated neutrophils were more than 80% pure.

S6 Fig. Release of neutrophil granules is not due to cell lysis and degranulation induced after infection and receptor stimulation is significantly higher than stimulation of single receptors alone. (A) Supernatants and cell lysates (2 x 10 5 cells) from infected and stimulated
samples were immunoblotted with anti-Rho-GDI as an indicator of cell lysis. A positive control for degranulation was treatment with PMA, and the negative control was cells added to FBS-coated wells. Rho-GDI was detected in the supernatants of cells treated with triton X-100. (B) BM neutrophils (2x10 5 ) were stimulated with poly-RGD, fMLP or IC to stimulate integrin, GPCR, or FcγR respectively and left uninfected or simultaneously infected with YptbΔ5 at a MOI of 20:1 for 3 hrs. Cell free supernatants from equivalent numbers were analyzed by ELISA for MMP-9 release. Data presented are expressed as fold increase relative to control uninfected and unstimulated samples. The data represent the means ± SEM from 3 independent experiments done in 2-3 replicates. (C) WT-BALB/c and Skap2KO BM neutrophils were stimulated for 3 hrs by 1μM PMA. Cell-free supernatants were analyzed for MMP-9 by ELISA. The data represent the means ± SEM from 3 independent experiments done in triplicate. Statistical significance was calculated using Student t-test. (TIFF) S1