Interferons Direct an Effective Innate Response to Legionella pneumophila Infection*

Legionella pneumophila remains an important opportunistic pathogen of human macrophages. Its more limited ability to replicate in murine macrophages has been attributed to redundant innate sensor systems that detect and effectively respond to this infection. The current studies evaluate the role of one of these innate response systems, the type I interferon (IFN-I) autocrine loop. The ability of L. pneumophila to induce IFN-I expression was found to be dependent on IRF-3, but not NF-κB. Secreted IFN-Is then in turn suppress the intracellular replication of L. pneumophila. Surprisingly, this suppression is mediated by a pathway that is independent of Stat1, Stat2, Stat3, but correlates with the polarization of macrophages toward the M1 or classically activated phenotype.

Both type I (e.g. IFN-␣/␤) and type II (IFN-␥) IFNs play an important role in the innate response to intracellular microbes (18). Although IFN-␥'s ability to classically activate macrophages toward an M1 phenotype has been intimately associated with antibacterial activity, more recent studies have also ascribed antibacterial activity to IFN-I (19 -22).
Characterization of IFN's antiviral activity led to the identification of the IFN-␣ receptor (IFNAR), the IFN-␥ receptor (IFNGR), and the JAK-STAT signaling paradigm, where STATs (Signal Transducers and Activators of Transcription) are transcription factors and JAKs (Janus Kinases) the receptor-associated kinases that activate them (23). Specifically, IFN-␥ directs the formation of a single transcription factor, the Stat1 homodimer, whereas IFN-I directs the activation of both Stat1 homodimers and ISGF-3 (IFN-stimulated gene factor 3; Stat1 ϩ Stat2 ϩ IRF9). Recent studies have underscored an important role for IRF-3 and IRF-7 in promoting activation of an IFN-I autocrine loop. This entails a sequential expression of IFN-␤ and IFN-␣, and is dependent on IFNAR. This autocrine loop plays an important role in innate immunity (23)(24)(25).
Both IFN-␥ and IFN-Is suppress L. pneumophila growth in murine macrophages (13,26,27). Our studies highlight a role for the IFN-I autocrine loop in the innate response to this bacterium. The ability of L. pneumophila to induce IFN-I expression was found to depend on IRF-3, yet be independent of the flagellin-Naip5 axis, as well as p50/cRel. Additionally, in contrast to the critical role Stat1 plays in the antibacterial response of IFN-␥, the ability of IFN-␣ to suppress L. pneumophila growth was found to be independent of both Stat1 and Stat2. Finally, the Stat1-independent protection afforded by IFN-␣ correlated with the induction of classically activated M1 macrophage markers.
Growth Curves-Ex vivo bacterial growth was evaluated by infecting (MOI ϭ 0.25) day 6 bone marrow-derived murine macrophages (BMMs) with post-exponential phase L. pneumophila in 24-well plates (2.5 ϫ 10 5 BMMs per well; (41)). Specifically, BMMs were infected with bacteria resuspended in RPMI. The zero hour, collected after 2 h of absorption, as well as all other time points were collected after washing infected BMMs with sterile water (10 s) and then lysing them in 1 ml of sterile water (5 min). Recovered bacteria were serially diluted in water and plated on CYE agar. Three days later, colony counts were enumerated. All infections were carried out in triplicate, and verified in at least three independent studies.
RT-PCR-Total RNA was prepared from day 6 BMMs before or after L. pneumophila infection (MOI ϭ 10) through lysis in Trizol (Invitrogen, Carlsbad, CA). 3-5 g of total RNA was reverse-transcribed (M-MLV; Invitrogen), as previously reported (32,37). For semi-quantitative analysis, cDNA was amplified for 28 cycles, fractionated on a 0.8% agarose gel and stained with ethidium bromide. For quantitative (Q) analysis, cDNA was amplified in an ABI Prism 7500 with SYBR green master mix (Applied Biosystems, Foster City, CA) and specific primers (see supplemental Table  S1). Gene expression was normalized to a ␤-actin control. For each primer set, Ct values and standard curves were generated by plotting log DNA concentration versus Ct values from 1:5 serial dilutions with SDS1.9.1 software (Applied Biosystems).

IFN-␣ and IFN-␥ Protect Macrophages from Infection with
L. pneumophila-To more carefully explore the ability of IFNs to suppress L. pneumophila replication in mice, BMMs were prepared from three distinct murine backgrounds, A/J, 129 and C57Bl/ 6J, as well as IFNGR and IFNAR knock-out mice (28). As anticipated, L. pneumophila replicated more effectively in A/J than C57Bl/6J BMMs. 129 macrophages exhibited an intermediate phenotype (see Fig. 1, A and B and Refs. 13,26,27). Consistent with previous reports, IFN-␥ pretreatment rendered all three genotypes, but not IFNGR(Ϫ/Ϫ) BMMs, resistant to infection (Fig. 1, A and D and Refs. 26,42,43). IFN-␣ also significantly suppressed L. pneumophila growth in all three genotypes, but not in IFNAR(Ϫ/Ϫ) macrophages (Fig. 1, B and D). Although IFN-␣ tended to be a bit less potent, (Fig. 1, A versus B, see also Figs. 4 and 7), it effectively suppressed the growth of both wild type and flagellin null (FlaϪ, panel E) L. pneumophila, in all genotypes (panels B and D; data not shown). This provided evidence that the response to IFN-Is is independent of Naip5 and the inflammasome (13). Intriguingly, L. pneumophila replicated significantly more efficiently in IFNAR(Ϫ/Ϫ) than the parental 129 macrophages (Fig. 1C), suggesting that L. pneumophila may stimulate IFN-I autocrine activity (see below and Refs. 13,27). The lack of an analogous response in IFNGR(Ϫ/Ϫ) BMMs, likely reflects the inability of macrophages to secrete autocrine IFN-␥ (data not shown).
L. pneumophila Infection Stimulates the Secretion of IFN-I-To determine whether L. pneumophila stimulated autocrine IFN-I production, IFN-␤ expression was evaluated by a sensitive Q-PCR assay (22,44). Both WT (denoted Lp) and FlaϪ L. pneumophila stimulated a robust 50 -100-fold induction of IFN-␤ in C57Bl/6J macrophages ( Fig. 2A). This response was maximal at 6 h after infection. However, dotA mutants, defective in type IV secretion and intracellular replication, did not stimulate IFN-␤ expression ( Fig. 2A). Similar results were obtained in A/J macrophages, indicating that the flagellin-Naip5 axis does not contribute to this response ( Fig. 2B and Ref. 13). Control studies revealed that WT (Lp) and FlaϪ L. pneumophila stimulated strong TNF induction at both 3 and 6 h, whereas the response to dotA was considerably more transient (Fig. 2C). Moreover, this was associated with considerably lower quantities of secreted TNF (supplemental Fig. S1A and Ref. 13,45)). Likewise, only WT L. pneumophila stimulated robust IL-1␤ secretion (supplemental Fig. S1B). These results underscore the need for sustained intracellular growth to stimulate a strong inflammatory response ( Fig. 2 and supplemental  Fig. S1). Moreover, differences in the pattern and kinetics of cytokine expression suggested that IFN-␤ and TNF expression are directed by distinct L. pneumophila sensor systems, which are also independent of the Naip5-Ipaf axis.
L. pneumophila IFN-␤ Expression Is IRF-3-dependent-Two transcription factors, NF-B and IRF-3, play an acute role in stimulating the expression of IFN-␤ (35, 46 -49). Electrophoretic mobility shift assays (EMSAs) revealed that p50 and cRel are the major NF-B transcription factors activated by LPS  and L. pneumophila in murine macrophages (supplemental Fig. S2, A-C). Moreover, dotA-dependent NF-B activation was considerably more transient than with either WT or FlaϪ L. pneumophila (supplemental Fig. S2C). To explore the potential role NF-B subunits may play in IFN-I expression, p50 and cRel double knock-out BMMs were exploited (35). Upon infection with L. pneumophila, p50/cRel double knock-out macrophages exhibited a pattern of IFN-␤ expression that was analogous to control C57Bl/6J cells (Fig. 3A). Curiously, the p50/ cRel double knock-out macrophages displayed increased IFN-␤ induction when they were infected with dotA or stimulated with LPS, underscoring a potentially complex counter regulatory role recently highlighted for NF-B (13,45,50,51). Moreover, L. pneumophila replicated normally in p50/cRel double knock-out BMMs (supplemental Fig. S2D). These observations indicate that NF-B does not signifi-cantlycontributetoL. pneumophiladependent IFN-I response.
IFN-␣ Stimulates a Unique Innate Immune Response-Next, we turned our attention to understanding how IFN-Is suppress L. pneumophila growth. Pretreatment with either 100 or 1000 units/ml of IFN-␣ rendered A/J BMMs resist-  OCTOBER 30, 2009 • VOLUME 284 • NUMBER 44 ant to L. pneumophila growth, in a dose-dependent manner (Fig. 4A). Similar results were obtained in 129 or C57Bl/6J macrophages (not shown). Moreover, IFN-␣ treatment impeded L. pneumophila replication, even when applied 24-h after infection, suggesting that intracellular bacteria represent the critical target of IFN-I activity (Fig. 4B). Consistent with this, IFN-␣ pretreatment did not affect the uptake of GFP-labeled L. pneumophila (not shown). Likewise, when GFP-L. pneumophila were exploited in a FACS-based assay, their replication was sensitive to IFN treatment (Fig. 4C). Consistent with growth curve studies (see Fig. 1), a modest level of GFP-L. pneumophila replication was observed in IFN-␣-treated BMMs, whereas IFN-␥ completely suppressed GFP-L. pneumophila growth at 24 and 48 h (Fig. 4C).
IFNs Activate M1 Macrophages-The antibacterial response to IFN-␥ correlates directly with the development of M1 or "classically activated" macrophages ((52); see Fig. 6), whereas polarization toward alternatively activated M2 macrophages had no effect on L. pneumophila growth (data not shown). As previously reported, this response was associated with an upregulation of MHC-II and CD86 surface expression, as well as induction of iNOS (53,55). Likewise, the ability of IFN-␥ to direct MHC-II, CD86, and iNOS expression was fully dependent on Stat1 (see Fig. 6, A and B). In contrast, the equally effective ability of IFN-␣ to induce these M1 markers was fully independent of Stat1 (Fig. 6B). Moreover, IFN-␣ directed M1 polarization was also independent of Stat2 (not shown). The close correlation between IFN-I stimulated, STAT-independent suppression of L. pneumophila growth and M1 polarization, suggests that the classical activation of macrophages is important for this innate response. These observations also highlight an unexpected functional dichotomy in the ability of type I and II IFNs to polarize macrophages toward the M1 phenotype and suppress L. pneumophila growth.
The relatively rapid, IFN-␣-stimulated and Stat1-independent induction of iNOS, a characteristic M1 marker, provided an opportunity to explore the potential role noncanonical signaling pathways may play in IFN-I response. Based on previous evidence of a role for p38 and PI-3 kinases in STAT-independent IFN-I response (56,57), the ability of IFN-␣ to induce iNOS expression was evaluated in the presence of two specific inhibitors, SB203580 and LY94002, respectively ( Fig. 6C and Ref. 56). Neither inhibitor blocked IFN-␣ stimulated expression of iNOS, raising the possibility that these two kinases also do not contribute to the IFN-I-dependent M1 polarization associated with L. pneumophila suppression. Consistent with this possibility, biochemical studies suggest that IFN-I does not effectively activate these kinases in primary murine macrophages (supplemental Fig. S4).
Nitric Oxide Accounts for Phenotypic Differences between Type I and II IFNs-NO production has not been ascribed a significant role in the innate response to L. pneumophila infection (26,58). To determine whether NO might contribute to the noncanonical IFN-I response associated with L. pneumophila suppression, additional studies were carried out. As previously reported, L. pneumophila failed to effectively stimulate NO production (26,58). However, pretreatment of cells with IFN-␥ and to a lesser extent with IFN-␣ led to a robust induction of NO (Fig. 7A). Moreover, this was blocked through the addition of either the iNOS-specific inhibitor, L-NIL, or a more general NOS inhibitor, L-NMMA (Fig. 7A). These observations raised the possibility that the more robust NO response stimulated by IFN-␥ might account for the modestly enhanced ability of IFN-␥ (versus IFN-␣) to suppress L. pneumophila growth (e.g. Figs. 1, 4C, and 7B). Consistent with this, the level of intracellular L. pneumophila replication was modestly increased when NO production was blocked in IFN-pretreated macrophages (Fig. 7B). Moreover, the change in L. pneumophila growth was considerably more striking in IFN-␥-pretreated macrophages, suggesting that NO production is likely to account for the more effective ability of IFN-␥ (versus IFN-␣) to suppress L. pneumophila replication. These studies also suggest that M1-associated responses that are independent of iNOS play a far more important role in the ability of IFNs to impede L. pneumophila growth.

DISCUSSION
Pattern recognition sensors systems evolved to detect the presence of pathogen associated molecular patterns (PAMPs) and initiate effective innate responses (24,59,60). Exploiting the murine system, investigators have identified two cytosolic sensors, Naip5/Birc1e and Ipaf, which specifically initiate an innate response to L. pneumophila flagellin (2,(7)(8)(9)11). This entails activation of the inflammasome, leading to caspase1 dependent secretion of IL-1␤ and IL-18, as well as the induction of pyroptosis (7,11,12). Curiously however, IL-1␤ and IL-18 do not appear to directly regulate L. pneumophila growth in macrophages (3,13). Moreover, the Naip5-Ipaf sensor system does not seem to participate in the secretion of other cytokines that have been associated with L. pneumophila infection (13,27).
Evidence that L. pneumophila replicated more robustly in IFNAR1(Ϫ/Ϫ) BMMs raised the possibility that the IFN-I autocrine loop may contribute to the innate response in WT macrophages, as recently suggested (13,27). Consistent with this, both WT and FlaϪ L. pneumophila potently stimulated IFN-␤ expression, analogous to what has been reported for L. monocytogenes, F. tularensis, as well as other microbial pathogens (34,44,61,62). Moreover, this response is largely independent of TLRs (13,14), IPS-1/MAVs (63), the Naip5-Ipaf-flagellin axis ( Fig. 2 and Ref. 13), IFNAR1, and Stat1 (supplemental Fig. S3A). Mechanistic studies determined that L. pneumophila stimulated IFN-␤ expression was dependent on IRF-3, but not p50/c-Rel (Fig. 3). Consistent with this, IFN-␤ expression also appeared to be dependent on TBK-1, the kinase that activates IRF-3 (see supplemental Fig. S3B and Ref. 44). Previous RNAi-based studies in human epithelial cells, which do not support robust L. pneumophila growth, have also implicated IRF-3, as well as IPS-1 in L. pneumophila-stimulated IFN-I expression (64). In summary, the innate response to L. pneumophila infection entails the activation of multiple pattern recognition systems that direct the secretion of several inflammatory cytokines, including IL-1␤, IL-18, TNF, and IFN-␤ (7,11,13). Although it seems likely that these pathways synergize to mount an effective innate response to L. pneumophila, IFN-I remains the single most effective cytokine in suppressing L. pneumophila growth (12,13).