Evasion of IFN-γ Signaling by Francisella novicida Is Dependent upon Francisella Outer Membrane Protein C

Background Francisella tularensis is a Gram-negative facultative intracellular bacterium and the causative agent of the lethal disease tularemia. An outer membrane protein (FTT0918) of F. tularensis subsp. tularensis has been identified as a virulence factor. We generated a F. novicida (F. tularensis subsp. novicida) FTN_0444 (homolog of FTT0918) fopC mutant to study the virulence-associated mechanism(s) of FTT0918. Methods and Findings The ΔfopC strain phenotype was characterized using immunological and biochemical assays. Attenuated virulence via the pulmonary route in wildtype C57BL/6 and BALB/c mice, as well as in knockout (KO) mice, including MHC I, MHC II, and µmT (B cell deficient), but not in IFN-γ or IFN-γR KO mice was observed. Primary bone marrow derived macrophages (BMDM) prepared from C57BL/6 mice treated with rIFN-γ exhibited greater inhibition of intracellular ΔfopC than wildtype U112 strain replication; whereas, IFN-γR KO macrophages showed no IFN-γ-dependent inhibition of ΔfopC replication. Moreover, phosphorylation of STAT1 was downregulated by the wildtype strain, but not the fopC mutant, in rIFN-γ treated macrophages. Addition of NG-monomethyl-L-arginine, an NOS inhibitor, led to an increase of ΔfopC replication to that seen in the BMDM unstimulated with rIFN-γ. Enzymatic screening of ΔfopC revealed aberrant acid phosphatase activity and localization. Furthermore, a greater abundance of different proteins in the culture supernatants of ΔfopC than that in the wildtype U112 strain was observed. Conclusions F. novicida FopC protein facilitates evasion of IFN-γ-mediated immune defense(s) by down-regulation of STAT1 phosphorylation and nitric oxide production, thereby promoting virulence. Additionally, the FopC protein also may play a role in maintaining outer membrane stability (integrity) facilitating the activity and localization of acid phosphatases and other F. novicida cell components.


Introduction
Francisella tularensis, a Gram-negative facultative intracellular bacterium, causes the disease tularemia [1]. Of the F. tularensis subspecies (subsp) described to date, those responsible for human disease include the highly virulent subsp. tularensis (type A) and the moderately virulent subsp. holarctica (type B) [2]. Although Francisella novicida (also referred to F. tularensis subsp. novicida [3]) is avirulent in humans, it behaves similarly to F. tularensis in human and mouse macrophages [4], shares biochemical features with F. tularensis [3,5] and exhibits an LD 50 in mice similar to F. tularensis via the pulmonary route [2,6]. There also are notable differences between F. novicida and F. tularensis, such as the LPS of F. novicida being more stimulatory [7,8,9]. However, F. novicida has been an attractive model for the study of the virulence mechanisms of F. tularensis, given the genetic similarity (98.1% homology between sequences common to U112 and SCHU S4) to F. tularensis [10], and similar lifestyle within host cells.
Francisella can infect a variety of cell types including pulmonary epithelial cells [11,12], dendritic cells [13], and hepatocytes [14], but exhibits a strong tropism for replication within macrophages [13]. The bacterial components responsible for the marked virulence of Francisella are still largely unknown. Among the most studied virulence factors are the Francisella pathogencity island (FPI) genes, which encode a secretion system required for phagosome escape and intramacrophage growth [15]. Additional virulence factors have been identified, in particular a 58 kDa outer membrane lipoprotein (FTT0918). This protein is truncated (expressed as a fusion protein) in the live vaccine strain (LVS) derived from F. tularensis subsp. holarctica [16], and reintroduction of FTT0918 into LVS restored virulence to a level indistinguishable from a wildtype subsp. holarctica strain [17]. Moreover, deletion of FTT0918 in F. tularensis subsp. tularensis also attenuates virulence [16]. Given the importance of this outer membrane lipoprotein (designated FopC) in F. tularensis virulence and pathogenicity, we generated a defined F. novicida FTN_0444 (homolog to FTT0918, 99% identity in aa sequence) mutant, designated KKF332, and characterized the contribution of this protein to the virulence of this microorganism. Our results demonstrate that FopC plays a role in inhibition of IFN-cmediated host immune defense(s), thereby promoting virulence.

Results
Construction of the F. novicida fopC mutant, KKF332 In order to evaluate the role of FopC protein (FTN_0444) as a virulence factor, we generated a fopC mutant of F. novicida using a previously described Targetron insertion method [18]. Sitespecific insertion of a 0.9 kb-Ll.LtrB intron between nt. 444025 and 444026 of the F. novicida U112 genome resulted in a fopC mutant, designated here as KKF332. Schematic representation of the mutated fopC locus in KKF332 is shown in Fig. 1A. Disruption and inactivation of the fopC gene in KKF332 was confirmed by Southern and Western blot analyses, respectively. Southern blot analysis (Fig. 1B) using an Ll.LtrB intron-specific probe [18] detected a 5.2 Kb band which corresponded to the NdeI restriction digestion product of the KKF332 fopC gene locus, but not to the parental U112 strain. Western blot analysis (Fig. 1C) using murine anti-FopC polyclonal antibodies demonstrated the presence of a 58 kDa band in wildtype U112 cells, but not in fopC mutant cells further confirming the lack of FopC protein expression in KKF332.

Virulence of the fopC mutant is attenuated in a murine pulmonary tularemia model
The fopC mutant was assessed for virulence in a murine pneumonic tularemia model. To determine the LD 50 of the fopC mutant using this model, BALB/c mice (6/group) were inoculated intranasally with an increasing dosage of strain KKF332. As shown in Fig. 2A, the LD 50 of the fopC mutant was between 10 4 and 10 5 CFU, which was approximately 1000 fold greater than that of the wildtype U112 strain (,10 CFU or less; [19]). As expected, all mice inoculated intranasally with 10 3 CFU of U112 (,100 LD 50 ) succumbed to infection by day 5.
To further examine replication and dissemination of the fopC mutant in comparison to the wildtype F. novicida strain, BALB/c mice were challenged i.n. with 10 3 CFU of strains KKF332 or U112, and bacterial burdens in the lungs, liver and spleen determined. Both fopC mutant and wildtype strain recovery were comparable in the lungs of infected mice at 1-2 days after inoculation and both strains also could be recovered at similar levels in the liver and spleen on the second day after challenge (Fig. 2B). However, on day 3, bacterial burdens were significantly (p,0.05) lower (1-2 logs) in all organs (lungs, liver and spleen) of DfopC infected mice in contrast to mice infected with the wildtype strain, which all succumbed to infection by day 5. In contrast, all fopC mutant infected mice survived and the organism was recovered at in increasing number in the lungs until day 7, with the number of recovered bacteria decreasing at day 14. Additionally, decreasing fopC mutant burdens were observed in the liver and spleen between day 3 and day 14. Collectively, reduced mortality and decreased bacterial burdens in the lungs, liver, and spleen demonstrate that fopC mutant virulence is attenuated, and that host immune responses could largely control mutant strain replication.

IFN-c mediates fopC mutant attenuation in mice
Since IFN-c has been shown to be a predominant component of the anti-Francisella protective immune response [20,21], we compared survival of BALB/c IFN-c +/+ and IFN-c 2/2 mice following i.n. challenge with 10 3 CFU wildtype U112 or fopC mutant. As shown in Fig. 2C, both IFN-c +/+ and IFN-c 2/2 mice challenged with U112 succumbed to infection within 6 days following challenge. However, only IFN-c 2/2 , not IFN-c +/+ , mice succumbed within 7 days following DfopC challenge. BALB/c IFNc +/+ mice challenged with DfopC survived for at least 30 days following challenge. These results suggest that FopC may be required to mediate resistance of F. novicida to elimination by endogenous IFN-c. We also analyzed whether mice deficient in endogenous IFN-c signaling (IFN-cR 2/2 mice) were susceptible to KKF32 challenge similar to the IFN-c 2/2 mice. Moreover, we used mice deficient in b2-microglobulin (MHC I 2/2 mice), MHC II (MHC II 2/2 mice), or B cells (mmT mice), all lacking components that have been shown to contribute to inducible IFN-c production. We challenged C57BL/6 and different knockout (KO) mouse strains (MHC-I 2/2 , MHC-II 2/2 , mmT and IFN-cR 2/2 ) intranasally with 10 3 CFU 112 (Fig. 3A) or fopC mutant (Fig. 3B). All strains of mice succumbed to wildtype U112 infection by day 6. In contrast, the C57BL/6 mice and three of four KO mouse strains (MHC-I 2/2 , MHC-II 2/2 , and mmT) infected with the fopC mutant all survived for the entire period of the experiment (30 days). The notable exception was IFN-cR 2/2 mice, which all succumbed to infection with the fopC mutant by 8 days.
Examination of replication and dissemination profiles indicated wildtype U112 strain replication and kinetics of dissemination to the liver and spleen were similar in C57BL/6 and IFN-cR 2/2 mice; although, higher (1-1.5 log 10 ) numbers of bacteria were recovered from the spleen and liver of IFN-cR 2/2 mice at days 2 and 3 (Fig. 4A). In comparison, the fopC mutant could be detected in the liver of IFN-cR 2/2 mice one day earlier than in the liver of C57BL/6 mice, and bacterial loads were significantly higher (1-2.5 log 10 ) in all tissues of IFN-cR 2/2 mice at day 3 post challenge. Collectively, increased susceptibility of IFN-cR 2/2 mice to the fopC mutant suggests that IFN-c was one of the host components responsible for the attenuation of KKF332. These results demonstrate that IFN-c signaling may play an important role in controlling the clearance of the fopC mutant following bacterial infection. Moreover, these findings are also in agreement with our previous observations [22] showing that CD4 + T cells are not required in the effector phase of antibody-mediated IFN-cdependent anti-Francisella immunity. Thus, the primary source of IFN-c to confer protective immunity against the fopC mutant may come from innate immune cells such as macrophages and natural killer cells.

IFN-c mediates inhibition of replication of the fopC mutant within primary macrophages
Mononuclear phagocytes are an important component of innate defenses against F. tularensis [1] and a primary site for intracellular bacterial replication [23]. Thus, we compared replication of wildtype U112 and fopC KKF332 strains in C57BL/6 and IFN-cR 2/2 macrophages, previously stimulated with rIFN-c in order to better understand how IFN-c signaling may control bacterial replication. Bone marrow derived primary macrophages (BMDM, 2610 5 /well) prepared from C57BL/6 and IFN-cR 2/2 mice were infected (MOI 10 or 100) with either U112 or KKF332 in the presence or absence of 1 ng/ml rIFN-c and viable bacteria within the macrophages recovered at 3 and 24 h after inoculation. Since initial (3 h) bacterial uptake/replication was comparable under all conditions tested (data not shown), our studies were based on comparisons carried out at 24 hours. Both wildtype and fopC mutant bacteria replicated comparably in C57BL/6 and IFN-cR 2/2 BMDM in the absence of rIFN-c (Fig. 5A, 10 MOI). However, addition of rIFN-c to C57BL/6 BMDM resulted in greater inhibition of intracellular fopC mutant replication (2.5-3.0 log 10 reduction) than intracellular U112 replication (1.0-1.5 log 10 reduction). Importantly, rIFN-c mediated inhibition in BMDM was not observed in IFN-cR 2/2 macrophages, demonstrating that  were challenged i.n. with 10 3 CFU U112 or increasing inocula (10 3 -10 6 CFU) of fopC mutant. All animals were monitored daily for survival for 30 days. (B) BALB/C mice were challenged i.n. with 10 3 CFU wildtype U112 or fopC mutant. Organs (lungs, liver and spleen) were collected on days 1, 2, and 3 from mice (n = 3 mice at each time point) infected with U112 and on days 1, 2, 3, 7 and 14 postchallenge from mice infected with the mutant, homogenized, and dilution plated for bacterial enumeration. Results are presented as log 10 CFU per organ (lower detection limit was approximately 1000 CFU/organ). Differences between replication of U112 and fopC strain on day 3 are significant with P values ,0.05 represented by * in the figure. Death by day 7 of all mice resulting from U112 challenge is represented by §. (C) BALB/c IFNc +/+ and IFN-c 2/2 mice (8/group) were challenged i.n. with 10 3 CFU wildtype U112 or fopC mutant. All animals were monitored daily for survival over 30 days. Difference in survival between fopC mutant challenged BALB/c and IFN-c 2/2 mice was significant at a p,0.001. activation of the IFN-c signaling pathway is responsible for the observed inhibition of bacterial replication. Moreover, these results also indicate that the fopC mutant is more susceptible to IFN-c-mediated killing within BMDM, when compared to the U112 parental strain. Addition of rIFN-c to C57BL/6 BMDM also resulted in increased inhibition of fopC mutant replication compared to the wildtype U112 at 100 MOI (data not shown).
Recently, it has been reported that F. novicida downregulates the IFN-c signaling pathway to evade host immunity [24]. Because we observed greater inhibition of replication of the fopC mutant than of the wildtype strain within rIFN-c treated macrophages, we sought to determine the effect of the fopC mutant on the IFN-c signaling pathway by examining phosphorylation of STAT1 (signal transducer and activator of transcription-1), a critical molecule in the IFN-c pathway that binds and activates transcription of responding genes (such as iNOS), as well as at the levels of nitric oxide (NO) produced in BMDM infected with either the wildtype U112 or the fopC mutant. Minimally phosphorylated STAT1 (p-STAT1) was observed in mock infected BMDM or those infected with either wildtype or fopC mutant strain in the absence of rIFN-c treatment (Fig. 5B); whereas, addition of rIFN-c induced p-STAT1 in mockinfected BMDM. However, despite addition of rIFN-c to BMDM infected with the wildtype U112 strain, phosphorylation of STAT1 was greatly suppressed (Fig. 5B) in agreement with previous studies from Roth and Parsa [24,25]. In contrast, rIFN-c treatment of BMDM infected with the fopC mutant resulted in p-STAT1 levels similar to that seen in rIFN-c treated mock infected cells.
Given that STAT1 activation can result in induction of NO, we measured NO levels in culture supernatants of infected cells as previously described. We observed significantly increased NO production in rIFN-c-treated BMDM infected with the fopC mutant compared to rIFN-c-treated cells infected with wildtype U112 (Fig. 5C). Increased NO levels following rIFN-c treatment in fopC mutant infected BMDM is all the more significant given that the level of NO induced in the absence of IFN-c by fopC was significantly lower than in BMDM infected with U112 (Fig. 5C). These results suggest that the increased levels of NO production in rIFN-c treated BMDM may contribute to the greater inhibition of  replication observed for the fopC mutant as compared to the wildtype strain. Collectively, these results suggest a role for FopC in evading host IFN-c-mediated responses.

Characterization of the fopC mutant
Based upon predicted Signal Peptidase II cleavage site data, FopC is considered to be a lipoprotein [26]. Membrane lipoproteins can serve as structural proteins, enzymes, transporters, or immune evasion molecules [27]. The fopC mutant, KKF332, replicated similarly to U112 in both supplemented TSB and Chamberlain's medium within a 36 h growth period, and exhibited similar sensitivity to several classes of antibiotics (Table  S1). However, biochemical characterization of KKF332 using substrate embedded API-ZYM strips revealed aberrant localization of acid phosphatase activity in the fopC mutant (Table S2). Acid phosphatase activity was observed associated with the whole cell fraction in the wildtype U112 strain; whereas, minimal acid phosphatase activity was found in the culture supernatant. In striking contrast, acid phosphatase activity of the fopC mutant was observed predominantly in the culture supernatant, and less associated with the whole cell fraction. This suggests that FopC contributes to the normal localization of acid phosphatases.
Currently, five acid phosphatase genes have been identified within the U112 genome and among these, AcpA is the major acid phosphatase associated with the outer membrane [28]. In order to determine if FopC is involved in proper localization of AcpA, Western immunoblot analyses utilizing AcpA antisera were carried out on U112 and KKF332 cell pellets and filtered culture supernatants. These analyses revealed that AcpA was more abundant in U112 pellets than in fopC mutant (Fig. 6A, WB). In contrast, AcpA was only found in the supernatant of the fopC mutant strain, but not in the supernatant of the wildtype U112 strain (Fig. 6B, WB). The presence of AcpA in the fopC mutant culture filtrate was confirmed further by proteomic analysis of the filtrate using Matrix Assisted Laser Desorption/Ionization-Time of Flight mass spectrometry (Table S3). Moreover, there was a greater abundance of different proteins in the supernatant of the fopC mutant strain than in the supernatant of the wildtype U112 strain (Fig. 6B, Gel); although growth conditions and CFUs used to generate the respective supernatants were similar. Some of the proteins (Table S3) in the fopC culture supernatant were identified by proteomic analysis with molecular weights ranging from 9 kDa to 100 kDa. These proteins consist of putative cytoslic proteins (e.g. DNA-binding protein HU-beta, isocitrate dehydrogenase), periplasmic proteins (e.g. beta-lactamase class A), membrane proteins (OmpA family protein, AcpA), and previously identified secreted proteins (ChiA, ChiB, CpbA, PepO) [29] suggesting proteins were released into culture through the ''leaky'' membrane of the fopC mutant. Western blot analysis of another identified protein, IglC, also revealed similar protein localization pattern to AcpA (Fig. 6A and 6B). These results suggest that FopC protein may play a role in maintaining outer membrane stability (integrity) since in the absence of FopC, acid phosphatases (especially AcpA) and a variety of other proteins normally cell associated could now be found in the supernatant. To further confirm that the membrane of the fopC mutant is less stable than the parental U112 strain, we compared the susceptibility of both bacterial strains to polymyxin B (PMB) treatment. Polymyxin B is a cationic cyclic peptide with a fatty acid tail, which interacts with Gramnegative bacterial membranes and affects membrane stability [30,31,32]. As shown in Fig. 7C, the fopC mutant replicated in a similar manner to U112 in the absence of PMB, but demonstrated significantly (p,0.01) slower growth in the presence of 400 mg/ml PMB at 24 and 48 h after bacterial inoculation. Moreover, the sensitivity of F. novicida to PMB was dose-dependent, and the fopC mutant exhibited greater sensitivity to PMB than U112 even at a low concentration (25 mg/ml [p,0.01], Fig. 7D) suggesting decreased F. novicida membrane stability in the absence of the FopC protein.

Complementation of KKF332 with the fopC gene restores virulence
In order to confirm that phenotypes associated with the KKF332 strain are due only to the loss of FopC, we transformed this strain with a plasmid (pfopC) that expresses fopC from its native promoter. Expression of FopC in the complemented strain (fopC/ pfopC) was confirmed by Western blot analysis using anti-FopC antibody (Fig. 7A). Given the marked difference in acid phosphatase localization between the wildtype U112 and fopC mutant strains, the cell pellet and culture supernatant material of pfopC complemented strain was analyzed for localization of acid phosphatase. As previously shown, wildtype U112 acid phosphatase activity is predominantly cell associated with little activity in the culture supernatant; whereas, acid phosphatase activity of the fopC mutant strain is observed to be predominantly in the supernatant, with little cell associated activity (Fig. 7B). When fopC KKF332 was complemented with pfopC, localization of acid phosphatase activity resembled that of the wildtype strain, i.e., high cell-associated levels of activity, in contrast to low activity levels observed in the culture supernatant (Fig. 7B). Furthermore, resistance to PMB damage was restored in the fopC complemented (fopC/pfopC) strain similar to the level of wildtype U112 (Fig. 7C and D). Complementation of fopC mutant with pfopC also restored inhibition of p-STAT1 to a level similar to wildtype U112 in IFNc stimulated BMDM (Fig. 8A), and restored intramacrophage replication in IFN-c treated BMDM to a level similar to that observed for the wildtype strain (Fig. 8B). The inhibition of fopC mutant replication in IFN-c treated BMDM might largely be due to production of nitric oxide, an effector of IFN-c signaling since addition of 0.25 mM N G -monomethyl-L-arginine (L-NMMA), an NOS inhibitor, led to an increase of fopC replication to a level similar to that in the BMDM unstimulated with rIFN-c (Fig. 8B). Finally, pfopC complementation restored virulence. C57BL/6 and IFN-cR 2/2 mice (6/group) were challenged i.n. with 10 3 CFU U112, the fopC mutant (KKF332) strain, or complemented KKF332 carrying pfopC (fopC/pfopC); and monitored for 30 days. As shown in Fig. 8C, all C57BL/6 mice succumbed to infection with wildtype U112 within five days; whereas, all mice survived infection with the fopC strain in agreement with the results presented above. Importantly, complementation of the fopC mutant with pfopC fully restored virulence (p,0.001 compared to fopC mutant challenged mice), with all mice succumbing to infection by day 6. Additionally, IFN-cR 2/2 mice infected with U112, the fopC mutant (KKF332) strain, or KKF332 complemented with pfopC succumbed to infection within eight days of infection as expected and consistent with earlier results. These results demonstrate that the attenuated virulence phenotype of KKF332 is specifically dependent upon loss of FopC. Thus, complementation of the KKF332 mutant with pfopC restored acid phosphatase activity localization, inhibition of IFN-c-mediated responses, as well as levels of virulence observed for the wildtype organism in the murine pulmonary tularemia model.

Discussion
Francisella, as with other obligate intracellular and facultative intracellular bacteria, has evolved various strategies to evade early host immune defenses including inhibition of phagolysosomal fusion [33,34] and early acidification [4,35,36]. Once the organism escapes the phagosome, it replicates in the host cytosol and induces apoptosis and lysis of infected cells. Additionally, the infected host cells and extracellular bacteria lead to dissemination of Francisella from the lung during pulmonary infection. In this report, we describe the role of a F. novicida outer membrane protein, FopC, in evasion of IFN-c signaling and virulence of this organism in an experimental mouse model.
Evasion of IFN-c-mediated host immune defenses has been characterized for a number of viruses [37,38] as well as for several intracellular bacteria [39,40,41]. For example, Salmonella typhimurium evades killing in activated macrophages by inhibition of IFN-cinduced nitric oxide production. This evasion is dependent on the membrane associated protein NirC, while Chlamydia trachomatis employs a different mechanism to evade IFN-c-mediated host defense to achieve persistent infection [42]. IFN-c-mediated indoleamine 2,3-dioxygenase upregulation and tryptophan starvation affect the developmental cycle of Chlamydia trachomatis [42]. In the presence of exogenous indole, expression of functional chlamydial tryptophan synthase has been shown to effectively overcome IFN-c-mediated killing [43]. In contrast, F. tularensis infections are more acute and rapidly fatal. Tridandapani and colleagues [24] have shown that F. novicida infection of murine macrophages results in suppression of IFN-c-induced STAT1 expression and phosphorylation. Our results now demonstrate a plausible mechanism by which bacterial components are involved in the F. novicida-mediated inhibition of IFN-c-mediated signaling. IFN-c-treated primary macrophages exposed to the fopC mutant exhibited greater levels of p-STAT1 as early as 3 h compared to the wildtype strain. Moreover, fopC mutant-infected macrophages also produced greater NO in the supernatant, and exhibited greater inhibition of intracellular bacterial replication. This inhibition of fopC replication was abrogated by the presence of a NOS inhibitor suggesting that the IFN-c signaling effector, nitric oxide, was important for control of fopC mutant infection.
During phenotypic characterization of the DfopC with API-ZYM, we noticed that acid phosphatase activity, which is normally cell-associated in the wildtype strain, was largely found in culture supernatants of the fopC mutant. This raises the question whether the aberrant acid phosphatase phenotype of the fopC mutant plays a role in modulation of IFN-c signaling. It is quite plausible based on (1) precedential evidence supporting the involvement of phosphatase(s) in inhibition of IFN-c signaling; for example, Vaccina virus inhibits IFN-c signaling via virus-associated phosphatase activity (VH1) [37]; (2) a recent publication showing F. novicida AcpA was capable of dephsophorylating other host defense proteins [44]; and (3) our observation of macrophages infected with a strain lacking four acid phosphatases (including AcpA) showing no impairment of p-STAT1 formation, similar to the fopC mutant (data not shown).
One interesting phenotype of the fopC mutant strain is that many proteins are present in the culture supernatant of the DfopC strain that are either absent or present at much lower levels in supernatants from the wildtype strain. This suggests that FopC functions to stabilize the outer membrane and that its absence leads to outer membrane destabilization and release of multiple membrane-associated proteins such as AcpA. Additional evidence in the published literature appears to support membrane-stabilization by FopC. Previous studies of Lindgren et al. [45] demonstrated that a FTT0918 (fopC) mutant of F. tularensis subsp. tularensis exhibits an inability to acquire iron bound to siderophores. We suggest that destabilization of the outer membrane in the absence of FTT0918 (FopC) might lead to loss of the siderophore receptor, thus resulting in the observed iron deficient phenotype. Experiments are currently being carried out to test this hypothesis.
In summary, findings presented here indicate that the FopC protein of F. novicida is a virulence determinant and functions to maintain membrane integrity and the proper localization of factors such as AcpA which may be one of several F. novicida bacterial components involved in the dephosphorylation of STAT1 to inhibit IFN-c-mediated signaling. Together, our findings support the importance of FopC in facilitating avoidance of immunosurveillance and promoting intracellular bacterial growth.

Intranasal challenge and bacterial dissemination
Mice were first anesthetized with 3% isoflurane, using a rodent anesthesia system (Harvard Apparatus, Holliston, MA), and then inoculated intranasally (i.n.) with respective doses of F. novicida strains in 25 ml of phosphate-buffered saline (PBS). The actual CFU administered in each experiment was determined by dilution plating of inocula on supplemented TSA. For survival experiments, animals were monitored daily for morbidity and mortality. For bacterial dissemination assays, mice were challenged i.n. with 10 3 CFU of U112 (lethal dose) or KKF332 (sublethal dose). At 1, 2, and 3 days postchallenge with U112 infected mice and at 1, 2, 3, 7, and 14 days with KKF332 infected mice (3 mice at each time point), lungs, livers, and spleens were collected into 3 ml of Dulbecco's modified Eagle's medium containing 4.5 g/L glucose, L-glutamine, sodium pyruvate (DMEM; Mediatech, Fairfax, VA) and supplemented with 10% (v/v) fetal bovine serum (D10; HyClone, Logan, UT), homogenized, adjusted to a final volume of 5 ml and dilution plated on supplemented TSA for bacterial enumeration. Survival data indicate no significant differences between the two strains (BALB/c and C57BL/6) of mice used in this study with regard to U112 or fopC mutant susceptibility. None of the U112-infected mice survived past day 7.

Antibodies
For generation of polyclonal antisera against FopC, 10 mg recombinant FopC protein was emulsified in an equal volume of complete Freund's adjuvant and injected intraperitoneally into 4 to 6-week-old, female BALB/c mice. After 2 weeks, mice were boosted with the same amount of protein with incomplete Freund's adjuvant. Anti-FopC sera were prepared 2 weeks post boost. Antibodies specific for phosphotyrosine-STAT1 were purchased from Cell Signaling Technology (Beverly, MA). b-actin antibody was purchased from Sigma-Aldrich (St. Louis, MO). AcpA antiserum was a gift from Dr. Thomas Reilly (University of Missouri).

Infection of macrophages
Primary macrophages were prepared from murine bone marrow cells as previously described [51]. These bone marrow derived macrophages (BMDM) were washed with DMEM and resuspended to a density of 10 6 cells/ml in D10. For intramacrophage growth assays, the suspension was aliquoted in 200 ml volumes per well in 96-well tissue culture plates and incubated for 2 h with U112 or KKF332 bacteria at a multiplicity of infection (MOI) of 10 or 100 bacteria/macrophage followed by 1 h gentamicin (GIBCO BRL, Grand Islands, N.Y.) treatment (20 mg/ml) to kill extracellular organisms. Wells were replenished with D10 and cells were grown at 37uC in a 5% CO 2 incubator for an additional 21 h. Measurements of intramacrophage bacteria at 24 h post inoculation were performed as previously described [52]. In IFN-c signaling assay experiments, BMDM were seeded into 6well plates at a concentration of 10 6 cells/well in 1 ml D10 overnight, and treated with 1 ng/ml rIFN-c (eBioscience) for 2 h followed by inoculation with 10 MOI wildtype or mutant bacteria. A dose titration experiment was initially carried out to determine the optimal IFN-c concentration for intramacrophage replication and 1 ng/ml concentration was used in subsequent IFN-c signaling studies because of the maximal (but not saturated) effect on control of DfopC replication. BMDM were lysed after 3 h incubation by adding 100 ml SDS sample buffer (62.5 mM Tris-HCl, pH 6.8, 2% w/v SDS, 10% glycerol, 50 mM DTT, 0.01% w/v bromophenol blue) per well, sonicated for 15-20 sec, and centrifuged (12,0006g, 10 min). Culture filtrates (30 ml) of these cells were analyzed for p-STAT1 by Western blotting using monoclonal antibody per manufacturer's recommendations (Cell Signaling Technology).

API-ZYM and acid phosphatase assays
U112 or mutant strains were grown with an inoculum of 100 ml (OD 600 = 0.5) in 30 ml Chamberlain's medium [53] for 36 hrs. Bacteria were collected by centrifugation and adjusted to an OD 600 of 0.3. The culture supernatants (following centrifugation) were filtered through 0.22 mm filters, and concentrated to 0.5 ml using a 10,000 MW cut-off Centripreps (Millipore, Billerica, MA). Bacteria (65 ml) or the concentrated culture filtrates were added to API-ZYM strips (bioMeriux, Inc, Durham, NC), incubated for 4 h at 37uC and analyzed for enzymatic profiles according to manufacturer's instructions. Quantitative acid phosphatase activity was determined by release of p-nitrophenol (pNP) from pnitrophenylphosphate substrate (pNPP; Sigma Chemical, St. Louis, Mo.) [54]. Briefly, 20 ml concentrated culture filtrate or OD 600 = 1 of bacteria were added to 100 ml 50 mM citric acid buffer (pH 5) containing 7.6 mM pNPP in a 96-well plate, incubated at 37uC for 5 h, and the production of pNP measured at 410 nm using a microplate reader (mQuant; Biotek Instruments, Winooski, VT).

Determination of nitrite concentration
The concentration of nitrite in cell supernatants was estimated using the Griess reagent [55]. Briefly, 100 ml supernatant was mixed with 50 ml Griess reagent and incubated at room temperature in the dark for 30 min. Absorbance was measured at 520 nm and nitrite concentrations determined by extrapolation using a NaNO 2 standard curve (1-100 mM).

Inhibition of bacterial growth by polymyxin B
Log phase F. novicida bacterial strains were used to inoculate 200 ml Chamberlain's medium (to an OD 600 = 0.05) per well containing various concentration of polymyxin B sulfate salt (0-1.6 mg/ml, Sigma) in 96 well polystyrene microplates. Bacteria were allowed to growth for up to 48 h in a 37uC incubator and the turbidity of bacterial cultures was measured at 600 nm at 4, 8, 24 and 48 h after inoculation using a microplate reader.

Statistical analysis
SigmaStat (Systat Software Inc., San Jose, CA) was used to perform all tests of significance. Statistical analyses for survival experiments were performed using the Kaplan-Meier test. The Student's t test was used to determine differences in in vitro and in vivo bacterial replication. Data are all presented as mean values 6 the respective standard deviation.

Supporting Information
Table S1 Inhibition of F. novicida growth by various classes of antibiotics. F. novicida wildtype U112 and fopC mutant strain were grown to 0.6 OD 600 , diluted to a Mcfarland 1 standard and 100 ml of each culture was confluently spread onto Mueller-Hinton agar plates, and the antibiotic discs placed on the plates. The plates were incubated at 37uC overnight and the zones of inhibition in mm measured around each disc. Data presented above is the average of readings from three discs of each antimicrobial. (DOCX)

Table S2
Enzymatic characterization of wildtype U112 and fopC mutant using the API-ZYM kit. * Intensity of the color reactions was graded from 0 to 5 according to an API-ZYM color reaction chart. (DOCX)