IL-1α Signaling Initiates the Inflammatory Response to Virulent Legionella pneumophila In Vivo

Legionella pneumophila is a Gram-negative intracellular bacterial pathogen that is the causative agent of a severe pneumonia called Legionnaires’ disease. After inhalation of aerosolized bacteria, L. pneumophila can infect and replicate within lung alveolar macrophages. Intracellular replication of L. pneumophila in macrophages in vitro, and virulence of L. pneumophila in animal models, requires a type IV secretion system (T4SS) called the Dot/Icm system, which secretes bacterial effector proteins into the host cytosol. These effectors, >270 of which have been identified (reviewed in Ref. 1), are believed to be critical for establishment of the Legionella-containing vacuole, the specialized membrane-bound intracellular compartment in which L. pneumophila replicates. In addition to its essential role in facilitating intracellular bacterial replication, the L. pneumophila T4SS is also associated with the induction of several potent innate immune responses (reviewed in Ref. 2).

Legionnaires’ disease is characterized by robust infiltration of neutrophils and other immune cells into the lungs (3–5). Mice depleted of neutrophils exhibit an increased burden of L. pneumophila in the lungs (6–9). Furthermore, in vivo blockade of the CXCR2 chemokine receptor reduces the number of neutrophils recruited to the lungs of L. pneumophila–infected mice and increases the lethality of L. pneumophila infection (8). Despite the clear protective role of neutrophils in L. pneumophila infections, it is also believed that excessive neutrophil influx may be responsible for much of the pathology associated with Legionnaires’ disease (3, 5). Thus, infected hosts require mechanisms to carefully regulate the influx of neutrophils into tissues such that sufficient neutrophils are recruited to mediate pathogen clearance without causing excessive immune pathology. Despite the central role of neutrophils in Legionnaires’ disease, the mechanisms controlling neutrophil recruitment to the lung in response to L. pneumophila remain poorly understood.

Previous work has established that neutrophil recruitment to the lung in response to L. pneumophila requires bacterial expression of the Dot/Icm T4SS (6). In addition, the type I IL-1R (IL-1R) and its downstream signaling adaptor protein, MyD88, are also required (6, 10–13). TLRs, which also use the MyD88 signaling adaptor, appear to only have a modest role in neutrophil recruitment to the lung (10–12, 14), suggesting that IL-1R signaling is the main pathway leading to neutrophil recruitment to the lung in vivo. Two related cytokines, IL-1α and IL-1β, can both signal through the IL-1R. A previous study suggested that IL-1β is critical for neutrophil recruitment in response to L. pneumophila (6). It was proposed that infected macrophages generate IL-1β that signals through the IL-1R expressed by airway epithelial cells (AECs). IL-1R signaling in AECs amplifies the initial IL-1β signal by triggering the production of chemokines, such as CXCL1 and CXCL2, which stimulate the rapid and robust recruitment of neutrophils to the lung (6). However, no study has specifically addressed a possible role for IL-1α in mediating IL-1R–dependent neutrophil recruitment in vivo, and consequently, the relative role of IL-1α and IL-1β in response to L. pneumophila remains uncertain.

Both IL-1α and IL-1β lack classical signal peptides to target the proteins to the conventional secretory pathway, and the mechanism of their release from cells remains poorly understood. Production of IL-1β appears to require two steps. First, activation of the NF-κB transcription factor results in transcription of Il1b mRNA, which is then translated into pro–IL-1β protein. Release of mature IL-1β has then been shown, in most instances, to require the caspase-1 protease, which cleaves and activates IL-1β into its biologically active form (15, 16). Caspase-1 is itself activated within multiprotein complexes called “inflammasomes” (17, 18). L. pneumophila has been shown to stimulate IL-1β release primarily via the NAIP5/NLRC4 inflammasome that senses bacterial flagellin that is translocated into the host cell cytosol via the Dot/Icm T4SS (19–23).

In contrast to IL-1β, IL-1α does not require proteolytic processing by caspase-1 to be biologically active (24, 25). Nevertheless, in certain instances, inflammasome activation can promote the extracellular release of IL-1α, perhaps as a result of inflammasome-induced cell death (24, 26, 27). However, it is still unclear whether the inflammasome is required for IL-1α production in response to bacterial infections in vivo. Similar to Il1b, the Il1a gene can be transcriptionally induced by infection, but IL-1α may also be constitutively expressed in certain cell types (24, 28). Virulent (T4SS⁺) L. pneumophila has been shown to induce IL-1α production by macrophages in vitro as well as in lung infections in vivo (29). In contrast, ΔdotA L. pneumophila mutants, which lack an active T4SS, do not induce IL-1α in vitro or in vivo (29). Nevertheless, the precise mechanism of IL-1α production in response to L. pneumophila remains unclear. Previous studies have suggested that T4SS-dependent activation of p38 and JNK MAPKs are required to induce Il1a transcription (29, 30). Activation of MAPKs by L. pneumophila appears to be partially due to a T4SS-dependent inhibition of host protein synthesis (30). Five L. pneumophila T4SS–translocated effectors have been identified that inhibit host protein synthesis (31), and Myd88/Nod1/Nod2^−/− macrophages infected with a strain lacking these five effectors (∆5) exhibit diminished MAPK activation and reduced Il1a mRNA levels as compared with wild-type (WT)–infected macrophages (30). However, infection of WT macrophages with the ∆5 mutant still induces normal MAPK activation (30), implying that MyD88/Nod signaling can also contribute. The mechanism by which protein synthesis inhibition results in MAPK activation remains unknown, and moreover, it is not clear whether macrophages infected with the ∆5 L. pneumophila strain exhibit a defect in release of IL-1α protein.

In this study, we show that in response to infection with virulent L. pneumophila in vivo, IL-1α produced by hematopoietic cells is the dominant cytokine leading to neutrophil recruitment to the lung at early time points (0–12 h) postinfection. We find that IL-1α and IL-1β act redundantly at later time points because neither Il1a^−/− nor Il1b^−/− mice have defects in neutrophil recruitment or bacterial clearance in the lung 24 h postinfection. Interestingly, IL-1α is produced normally in mice lacking both caspase-1 and caspase-11, strongly implying that inflammasomes are not required for IL-1α production. Mice deficient in both Casp1/11 and Il1a phenocopied Il1r1^−/− mice, confirming that inflammasomes can compensate for a lack of IL-1α at late time points. Interestingly, we did not detect a defect in IL-1α production in macrophages infected with the L. pneumophila mutant lacking five bacterial effectors that block host translation (Δ5). Although the Δ5 mutant had no defect in IL-1α production, we find that translation inhibition in concert with TLR activation is sufficient to induce IL-1α in vitro and in vivo. Taken together with previous studies (29–31), these results suggest that an uncharacterized pathway is responsible for IL-1α production in response to L. pneumophila infection in vivo. Our results point to a critical role for IL-1α in initiating IL-1R–dependent neutrophil recruitment and inflammatory responses in vivo that is complementary to the established inflammasome/IL-1β signaling axis.

This study was carried out in strict accordance with the recommendations in the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health. The protocol was approved by the Animal Care and Use Committee at the University of California, Berkeley.

Except for bone marrow chimeras (see below), all mice were age matched at 6–8 wk old. Il1r1^−/− and C57BL/6 (B6) mice were purchased from The Jackson Laboratory. Casp1/11^−/− mice (32) were a gift from A. van der Velden and M. Starnbach (Harvard Medical School, Boston, MA). Il1a^−/− and Il1b^−/− mice have been described previously (33). Il1a/Casp1/11^−/− triple-knockout (TKO) mice were generated from crosses at University of California, Berkeley. B6.SJL-Ptprc^a/BoyAiTac (CD45.1) mice were purchased from Taconic. For bone marrow chimeras, 5- to 6-wk-old mice were irradiated twice with 600 rad 4 h apart and reconstituted with 1 × 10⁷ donor cells by injection into the tail vein. Chimeric mice were bled 11 wk after irradiation and reconstitution was assessed by flow cytometry of hematopoietic cells for expression of CD45.1 and CD45.2 using anti–CD45.1-FITC (eBioscience) and anti–CD45.2-PE (eBioscience) Abs. Twelve weeks after irradiation, chimeric mice were infected with L. pneumophila. All mice were specific pathogen free, maintained under a 12-h light/dark cycle (7 am to 7 pm), and given a standard chow diet (Harlan-irradiated laboratory animal diet) ad libitum.

Age-matched mice were anesthetized with ketamine and infected intranasally with 2 × 10⁶ LP01 or LP01ΔdotA in 20 μl PBS. In some experiments, mice were treated intranasally with exotoxin A (ExoA), Pam₃CSK₄, or both in 20 μl PBS, as described previously (31). Bronchoalveolar lavage (BAL) was performed by introducing 800 μl PBS into the trachea with a catheter (18 g, 1.3648 mm; BD Angiocath). Cells in the BAL fluid were pelleted, and cell-free BAL fluid was analyzed by ELISA. Total host cells in the lavage fluid were counted by staining cells with Guava Viacount (Millipore) and running samples on the Guava easyCyte Plus flow cytometer running CytoSoft5.3 software (Millipore). Lavage samples were stained with anti–Gr-1-PeCy7 and anti–Ly-6G-PE (eBioscience) and analyzed on a Beckman Coulter FC-500 analyzer. Absolute numbers of Ly-6G⁺Gr1⁺ cells were calculated by taking the percent double-positive cells determined by flow cytometry and multiplying by the total number of viable cells counted by the Guava easyCyte Plus flow cytometer. Bacterial burden in lungs was enumerated by hypotonic lysis of host cells in the lavage, followed by spiral plating onto buffered charcoal yeast extract plates with the Autoplate 5000 spiral plating system (Spiral Biotech). CFU/ml in BAL fluid was determined by a QCount Colony Counter (version 3.0; Advanced Instruments). BAL fluid mass was recorded prior to processing, and this mass was used to estimate the volume of recovered BAL fluid. Total CFU was then calculated by multiplying CFU/ml by the estimated volume of BAL fluid. When noted, mouse body temperature and weight were monitored postinfection with LP01. Mouse body temperature was measured by rectal probe and microtherma thermometer (Braintree Scientific). The probe was lubricated with a water-based lubricant (Astroglide) before use. Temperature and weight were measured at the same time daily.

For in vitro experiments, all L. pneumophila strains were derived from LP02, a streptomycin-resistant thymidine auxotroph derived from L. pneumophila LP01. The ΔdotA, ΔflaA, and Δ5ΔflaA strains were generated on the LP02 background and have been described previously (21, 30, 31). Mutants lacking one or more effectors were generated from LP02 by sequential in-frame deletion using the suicide plasmid pSR47S as described previously (34). Sequences of primers used for constructing deletion plasmids are listed in Supplemental Table IA. Unless otherwise noted, all strains used for in vitro infections were deficient for bacterial flagellin (ΔflaA), and thus, non–motile. L. pneumophila from the ΔflaA background were used in vitro to avoid activation of the NAIP5/NLRC4 inflammasome (19–23). For in vivo experiments, we used L. pneumophila WT strain LP01, a non–motile streptomycin–resistant strain derived from the original Philadelphia outbreak (35). The ΔdotA LP01 strain has been described previously (36).

Bone marrow–derived macrophages (BMDMs) were plated in 24-well plates at a density of 5 × 10⁵ cells/well and infected at a multiplicity of infection (MOI) of 1–3 (as indicated) by centrifugation for 10 min at 400 × g. In some experiments, macrophages were treated with ExoA (List Biological Labs), a synthetic bacterial lipopeptide (Pam₃CSK₄) (InvivoGen), or both. After 1 h of infection, medium was changed. All in vitro L. pneumophila infections were in the absence of thymidine to curtail bacterial replication.

At the indicated time posttreatment, supernatants or BAL fluid were collected, cleared by centrifugation, and analyzed by ELISA using paired IL-1α Abs (BD Biosciences and eBioscience) or paired IL-1β Abs (eBioscience and BD Biosciences). Recombinant IL-1α (eBioscience) or IL-1β (eBioscience) was used as a standard for each respective ELISA. Cytotoxicity was measured by evaluation of lactate dehydrogenase released from cells (37). Specific lysis was calculated as a percentage of lactate dehydrogenase released by detergent-lysed macrophages.

Macrophages were derived from the bone marrow of C57BL/6J mice (The Jackson Laboratory). Macrophages were derived by 8 d of culture in RPMI 1640 medium supplemented with 10% serum, 100 μM streptomycin, 100 U/ml penicillin, 2 mM l-glutamine and 10% supernatant from 3T3–M-CSF cells, with feeding on day 5. HEK293T cells were grown in complete medium (DMEM, 10% FBS, 100 U/ml penicillin, 100 μg/ml streptomycin, and 2 mM l-glutamine).

The library of 259 confirmed or putative secreted effector proteins has been described previously (38). Using the Gateway cloning system (Invitrogen), the library was cloned into a Gateway compatible murine stem cell virus 2.2 retroviral expression construct. We modified the murine stem cell virus 2.2 expression construct with an in-frame 6× Myc tag upstream of the cloned effectors to accommodate for non-AUG start codon usage in prokaryotes, and we removed the downstream internal ribosome entry site–GFP. HEK293T cells were plated at 2.5 × 10⁴ cells/well in 96-well tissue culture plates. Twenty-four hours after plating, cells were cotransfected using Lipofectamine 2000 (Invitrogen), following the manufacturer’s instructions, with a single library clone and the TK-Renilla luciferase reporter construct. Twenty-four hours after transfection, cells were lysed in passive lysis buffer (Promega) for 5 min at 25°C. Cell lysates were incubated with the Renilla luciferase substrate coelenterazine (Biotium), and luminescence was measured on a SpectraMax L microplate reader (Molecular Devices). The relative block in translation was measured by comparing Renilla luminescence in cells transfected with a control bacterial protein that does not block translation.

Macrophage RNA was isolated using an RNeasy kit (Qiagen), according to the manufacturer’s protocol. RNA samples were treated with RQ1 DNase (Promega) prior to reverse transcription with Superscript III (Invitrogen). cDNA reactions were primed with poly(dT). Quantitative PCR was performed as described previously (39) using a Step One Plus RT-PCR system (Applied Biosystems) with Platinum Taq DNA polymerase (Invitrogen) and EvaGreen (Biotium). Transcript levels were normalized to those of Rps17. The following primer sequences were used: for Il1a, 5′-ATGACCTGCAACAGGAAGTAAAA-3′ (forward) and 5′-TGTGATGAGTTTTGGTGTTTCTG-3′ (reverse); and for Rps17, 5′-CGCCATTATCCCCAGCAAG-3′ (forward) and 5′-TGTCGGGATCCACCT CAATG-3′ (reverse).

BMDMs (5 × 10⁵) were seeded in 24-well plates and infected with bacterial strains at an MOI of 3. Twenty-five minutes prior to labeling, macrophages were treated with 25 μg/ml chloramphenicol to inhibit bacterial translation. At 6 and 24 h postinfection, medium was removed and incubated with 25 μCi/ml [³⁵S]methionine (Perkin Elmer) in RPMI 1640 medium without methionine supplemented with 10% serum, 2 mM l-glutamine, 25 μg/ml chloramphenicol, and 10% supernatant from 3T3–M-CSF cells. Cells were labeled for 1 h, washed three times with cold PBS, and then lysed with radioimmunoprecipitation assay (RIPA) buffer supplemented with 2 mM Na₃VO₄, 1 mM PMSF, 25 mM NaF, and 1× Roche protease inhibitor mixture (no EDTA) (pH 7.2) for 10 min at 4°C. Total protein levels were measured by bicinchoninic acid assay, and equal amounts of protein were mixed with SDS sample buffer (40% glycerol, 8% SDS, 2% 2-ME, 40 mM EDTA, 0.05% bromophenol blue, and 250 mM Tris-HCl [pH 6.8]), boiled for 5 min, and then separated by SDS-PAGE. The gels were stained with Coomassie blue to show equal protein loading, dried, and exposed to a phosphor screen and visualized using a Typhoon Trio imager (GE Healthcare).

A previous report identified the IL-1R as a major signaling pathway that controls the recruitment of neutrophils to the lung in response to L. pneumophila (6). This paper proposed that IL-1β is the major ligand signaling through the IL-1R in L. pneumophila infections but did not specifically address a possible role of IL-1α and also did not examine the consequences of IL-1R deficiency on host health. Before addressing the relative importance of IL-1α and IL-1β, we first set out to confirm the previously proposed role of IL-1R signaling in L. pneumophila infection. We infected Il1r1–deficient (Il1r1^−/−) mice with WT L. pneumophila and examined the mice at 12, 24, and 48 h postinfection. Total numbers of Ly6G⁺Gr1⁺ cells (in this paper referred to as neutrophils) in BAL fluid were determined by flow cytometry, and bacterial burden was measured by plating for CFUs. Consistent with previous reports, Il1r1^−/− mice recruited reduced numbers of neutrophils to the lungs in response to L. pneumophila, with ∼10-, 5-, and 4-fold fewer neutrophils in Il1r1^−/− mice than WT mice at 12, 24, and 48 h postinfection, respectively (Fig. 1A–C). Interestingly, although there are significant defects in the number of neutrophils recruited to the lungs of Il1r1^−/− mice, the total number of cells in the BAL fluid of these mice does not significantly differ from WT mice (Fig. 1A–C). The similarity in overall numbers of cells in the BAL appears to be because postinfection with L. pneumophila, Il1r1^−/− mice harbor greater numbers of alveolar macrophages and CD45-negative/low cells that compensate for the decrease in neutrophils (Supplemental Fig. 1). One possible explanation for this is that, in WT mice, damaged or dead alveolar macrophages and CD45-negative/low cells are normally phagocytosed and thereby eliminated by neutrophils. Thus, with decreased neutrophils in Il1r1^−/− mice, alveolar macrophages and CD45-negative/low cells accumulate (Supplemental Fig. 1). In addition to decreased neutrophils, Il1r1^−/− mice harbor ∼5- and 17-fold higher CFU in BAL fluid over B6 controls at 24 and 48 h postinfection, respectively (Fig. 1B, 1C). Il1r1^−/− mice also have a slight increase in bacterial burden measured in BAL fluid at 12 h postinfection, but this difference is not dramatic, presumably because L. pneumophila does not have enough time to appreciably replicate or be cleared by the host at this time point (Fig. 1A).

Although our data confirm that IL-1R signaling is critical for neutrophil recruitment and elimination of bacteria from the lung, neutrophils are also believed to be key mediators of the immune pathology of Legionnaires’ disease. Therefore, we were interested to determine whether the decreased neutrophil response in Il1r1^−/− mice resulted in overall increased or decreased host health. To assay host health, we followed body temperature and weight loss in WT and Il1r1^−/− mice (Fig. 1D). Although both WT and Il1r1^−/− mice eventually recover from the infection, Il1r1^−/− mice show more severe weight loss and temperature decreases than WT mice after being infected with L. pneumophila over a 2-wk study (note that, in contrast to humans, mice typically exhibit a hypothermic response, rather than a fever, as a result of infection (40)). We have shown that Il1r1^−/− mice exhibit increased bacterial burden in the lung at 12, 24, and 48 h postinfection with L. pneumophila (Fig. 1). We therefore suggest that increased bacterial burden in Il1r1^−/− mice over the first week of infection is likely the cause of the decreased overall health of these animals in response to L. pneumophila infection. However, after ∼1 wk of infection, compensatory innate and/or adaptive immune responses likely control the infection. Overall, our results suggest that during the course of experimental L. pneumophila infection, the beneficial function of early neutrophil influx in bacterial clearance outweighs the potentially negative effects of neutrophil-mediated immune pathology. These data also establish an important role for IL-1R signaling in host health in addition to the previously established role for IL-1R signaling in neutrophil recruitment and bacterial clearance.

IL-1α and IL-1β are the only known agonists of the IL-1R. We therefore tested whether there was a correlation between IL-1α or IL-1β production and the recruitment of neutrophils to the lungs of infected mice. B6 mice were infected with WT L. pneumophila or an avirulent mutant strain of L. pneumophila that lacks a functional T4SS (ΔdotA). BAL fluid was harvested at 3, 6, 9, and 12 h postinfection and assessed for the presence of neutrophils, IL-1α, and IL-1β. The earliest in vivo Dot/Icm-dependent response was the production of IL-1α, which was first detectable at 3 h postinfection (Fig. 2A). By contrast, the earliest significant production of IL-1β (above that induced by ΔdotA) was not until 6 h postinfection (Fig. 2B), the same time that the Dot/Icm-dependent influx of neutrophils can first be detected (Fig. 2C). It is interesting to note that there seems to be an increase in the number of neutrophils found in the BAL fluid postinfection with the ΔdotA L. pneumophila strain at 3 h, although this difference is not statistically significant (Fig. 2C). Consistent with previous results, the ΔdotA L. pneumophila strain did not appreciably induce IL-1α production in the lung (Fig. 2A, 2B). Thus, although there may be a low level of Dot/Icm-independent neutrophil recruitment to the lung, this recruitment appears to be IL-1α independent and likely plays a minimal role in protecting the host from infection. Taken together, these data show that IL-1α production is largely Dot/Icm dependent and occurs prior to the recruitment of neutrophils to the lung. Our findings suggest a role for IL-1α in the early IL-1R–dependent and Dot/Icm-dependent recruitment of neutrophils to the lungs of L. pneumophila–infected mice.

We next tested whether the loss of IL-1α or IL-1β would have an effect on neutrophil recruitment. We infected WT (B6), Il1a^−/−, Il1b^−/−, Casp1/11^−/−, and Il1r1^−/− mice with WT L. pneumophila and measured neutrophil recruitment and bacterial burden at 12 h postinfection. As expected, Il1r1^−/− mice showed a strong defect in recruitment of neutrophils to the lung, whereas both the Il1b^−/− and Casp1/11^−/− mice showed no defect in neutrophil recruitment to the lung at 12 h postinfection, as compared with B6 mice (Fig. 3A). However, there was ∼17-fold decrease in the number of neutrophils recruited to the lung of Il1a^−/− mice as compared with B6 mice (Fig. 3A). Importantly, Il1a^−/− mice have no significant difference in the production of IL-1β in the BAL fluid of infected mice at 12 h postinfection (Supplemental Fig. 2A, 2B). These data suggest that IL-1α may be more important than IL-1β for the recruitment of neutrophils to the lung at 12 h postinfection. Interestingly, the defect in neutrophil recruitment in Il1a^−/− mice was not as pronounced as the defect seen in Il1r1^−/− mice. This suggests that, although IL-1β is not itself essential for neutrophil recruitment, it can partially compensate for the loss of IL-1α (addressed further below). The bacterial burden in the infected mice was very similar among all of the genotypes, likely because at 12 h postinfection, L. pneumophila has not had enough time to appreciably grow or be cleared by the host immune response (Fig. 3B). As an important control, measurement of IL-1α protein levels in the BAL fluid of infected mice demonstrated that only Il1a^−/− mice had defects in production of IL-1α in response to L. pneumophila infection (Fig. 3C). The amount of IL-1α detected in the BAL fluid of L. pneumophila–infected Il1r1^−/− mice is slightly higher than WT mice, likely due to an increase in bacterial burden caused by reduced neutrophil recruitment to the lungs of these mice (Fig. 3C). The increase in bacterial burden in Il1r1^−/− mice likely leads to more infected macrophages and thus an increase in the production of IL-1α. In addition, the loss of the IL-1R may result in less internalization of the IL-1α protein, resulting in higher extracellular accumulation. We also note that IL-1α is produced even in Casp1/11^−/− mice, indicating that, in response to L. pneumophila, IL-1α production in vivo can be independent of both caspase-1 and caspase-11 inflammasomes (Fig. 3C).

We hypothesized that the intermediate phenotype seen in the Il1a^−/− mice was due to low levels of inflammasome-dependent IL-1β production that are still capable of signaling through the IL-1R. We were unable to generate Il1a/b^−/− double-knockout mice because these genes are located directly next to each other on the chromosome. Thus, to test whether there is redundancy between IL-1α and IL-1β, we generated Il1a/Casp1/11^−/− TKO mice. These mice are predicted to be deficient in production of IL-1α and IL-1β, because production of biologically active IL-1β is generally believed to require caspase-1. We should note that the TKO mice are not only defective in IL-1β cytokine production, but they are also unable to undergo pyroptosis, a caspase-1/11–dependent form of lytic cell death, which has previously been shown to evict bacteria from their intracellular niche and render them susceptible to phagocytosis and killing by neutrophils (41, 42). The loss of pyroptosis could lead to an increased bacterial burden; however, at 12 h postinfection, we see very little differences in bacterial burden in the BAL fluid, and thus, we argue that the major defect in the TKO mice at 12 h postinfection is the loss of IL-1β processing and release (Fig. 3D, 3E). Consistent with a defect in IL-1α and IL-1β production, we find that in response to L. pneumophila infection, TKO mice produce almost no detectable IL-1α and very low levels of IL-1β in BAL fluid at 12 h postinfection (Supplemental Fig. 2C, 2D). Interestingly, TKO mice exhibited a large defect in neutrophil recruitment to the lung; in fact, these mice were as defective in neutrophil recruitment as Il1r1^−/− mice (Fig. 3D). These data suggest that IL-1α is the major cytokine required to signal through the IL-1R and recruit neutrophils to the lung at 12 h postinfection, although caspase-1/11–dependent signaling through the IL-1R (presumably mediated by IL-1β) can partially compensate for the loss of IL-1α. Furthermore, caspase-1 is usually considered to be essential for IL-1β processing (15, 16), although some previous reports have suggested that IL-1β can be generated in the absence of Casp1/11 (43, 44). Even though Il1a/Casp1/11^−/− TKO mice produced very low levels of IL-1β, TKO mice were as defective in neutrophil recruitment as Il1r1^−/− mice, implying that, at least in response to L. pneumophila, production of biologically active IL-1β requires caspase-1/11.

Il1r1^−/− mice exhibit reduced neutrophil recruitment that is sustained until at least 48 h postinfection (Fig. 1A). We therefore tested whether the loss of IL-1α would lead to a defect in neutrophil recruitment and an increase in bacterial burden at late time points. We infected WT, Il1a^+/−, and Il1a^−/− littermates with WT L. pneumophila and compared these mice to Casp1/11^−/− mice (Fig. 3F–H). At 48 h postinfection, Il1a^−/− mice had no defect in neutrophil recruitment to the lung and only a modest defect in control of bacterial burden (Fig. 3F, 3G). In addition, we saw no defect in neutrophil recruitment by Casp1/11^−/− mice at 48 h postinfection, suggesting that both IL-1α and IL-1β are capable of signaling through the IL-1R and can compensate for the loss of each other by 48 h postinfection. In fact, Casp1/11^−/− mice actually appeared to exhibit increased recruitment of neutrophils to the lung (Fig. 3F). However, despite the increased neutrophil recruitment, Casp1/11^−/− mice also exhibited increased bacterial burdens in the lung at 48 h postinfection (Fig. 3G). As mentioned previously, this counterintuitive result is likely explained by the loss of caspase-1/11–dependent pyroptosis, which has previously been shown to evict bacteria from their intracellular niche and render them susceptible to phagocytosis and killing by neutrophils (41, 42). Importantly, we find that Casp1/11^−/− mice produce IL-1α in response to L. pneumophila infection and actually induce significantly more IL-1α than WT mice; this increase is likely due to the loss of pyroptosis and subsequent increased bacterial burden in these mice (Fig. 3H).

IL-1α is inducible in hematopoietic cells but is also reported to be constitutively expressed by certain nonhematopoietic cells (24, 28). We therefore wished to determine whether the rapid production of IL-1α and the ensuing neutrophil influx required IL-1α production by hematopoietic or nonhematopoietic cells. We generated bone marrow chimeras in which WT B6.SJL (CD45.1⁺) mice were reconstituted with bone marrow from Il1a^−/− (CD45.2⁺) mice and vice versa. To confirm that our chimeras had been reconstituted to a high level, blood samples were collected and stained with Abs for CD45.1 and CD45.2 that marked WT- and Il1a^−/−-derived hematopoietic cells, respectively (Supplemental Fig. 3). Chimeric mice were infected with L. pneumophila, and BAL fluid was collected 12 h postinfection. Mice reconstituted with B6 hematopoietic cell populations produced IL-1α in response to L. pneumophila infection, whereas mice reconstituted with Il1a^−/− bone marrow failed to produce IL-1α (Fig. 4A). Importantly, the production of IL-1α correlated with the recruitment of neutrophils to the lung (Fig. 4B). Consistent with our previous findings (Fig. 3), we see little difference in the total CFU found in the BAL fluid of these mice at 12 h postinfection, although there was a slight increase in bacterial burden in mice that received Il1a^−/− bone marrow (Fig. 4C). These chimera experiments demonstrate that hematopoietic cells in the lung, presumably macrophages that have been infected with L. pneumophila, are responsible for the early production of IL-1α and subsequent recruitment of neutrophils to the site of infection.

Given the major role IL-1α plays in neutrophil recruitment, we next wanted to explore the molecular mechanism of IL-1α production by macrophages. We and others (31, 34, 45, 46) previously showed that L. pneumophila encodes five Dot/Icm-secreted effectors that inhibit host protein synthesis. A strain lacking these five effectors (∆5) was defective in the induction of a subset of inflammatory cytokines, including IL-23 and GM-CSF (31). Moreover, ∆5 was also defective in the transcriptional induction of the Il1a gene when the TLR and NOD-like receptor innate immune sensing pathways were severely hindered (infections of Myd88/Nod1/Nod2^−/− BMDMs) (30). The overall model emerging from our previous studies was that protein synthesis inhibition by virulent L. pneumophila produces a host cell stress response that leads to the production of inflammatory cytokines. Therefore, we asked whether the ∆5 L. pneumophila strain could still induce IL-1α protein release by WT BMDMs. We infected macrophages with the ∆5 L. pneumophila strain on the ∆flaA background (∆5∆flaA) and measured the production of IL-1α from these cells. We used L. pneumophila on the ΔflaA background to avoid the confounding effects of NAIP5/NLRC4 inflammasome activation by flagellin. Interestingly, we found that the ∆5∆flaA strain still induced production of significant amounts of IL-1α protein (Fig. 5A). This result is consistent with previous in vivo observations that showed that neutrophil recruitment is normal in response to the ∆5 mutant (30). We considered two possible explanations for the ability of the ∆5 mutant to induce IL-1α: 1) protein synthesis inhibition is not required for IL-1α production; or 2) residual protein synthesis inhibition by the ∆5 strain is sufficient to induce IL-1α. Consistent with the latter possibility and with our previous work (31), we found that the ∆5∆flaA strain still significantly inhibited host protein synthesis in BMDMs (as measured by incorporation of [³⁵S]methionine) as compared with infection with ΔdotAΔflaA, which does not block translation (Fig. 5B) (31). These results raised the possibility that L. pneumophila might encode additional effectors that inhibit host protein synthesis. To identify these effectors, we used a library of 259 known and putative secreted effectors (38) that we cloned into a mammalian expression vector. Each individual effector expression plasmid was cotransfected into 293T cells, along with a plasmid that constitutively expresses Renilla luciferase, and protein synthesis (as assessed by luminescence) was measured 24 h after transfection (Supplemental Fig. 3, Supplemental Table IB). As a positive control, this screen successfully identified the five previously described effectors that are known to block host translation (Lpg0437, Lpg1368, Lpg1488, Lpg2504, and Lpg2862) (Supplemental Table IB) (31). In addition, two other effectors that inhibit host protein synthesis were identified: Lpg0208, a serine/threonine kinase, and Lpg1489, a putative effector of unknown function (47). Lpg0208 and Lpg1489 were confirmed to inhibit protein synthesis in 293T cells, as measured by reduced [³⁵S]methionine incorporation upon overexpression of each effector (data not shown). However, deletion of these two additional effectors in the ∆5ΔflaA background to generate a strain we call ∆7ΔflaA did not significantly affect the ability of L. pneumophila to inhibit host protein synthesis in macrophages (Fig. 5B). The ∆7 strain also induced normal production of IL-1α in vitro (Fig. 5A). The residual ability of ∆7 L. pneumophila to inhibit host protein synthesis and/or induce IL-1α may therefore be due to additional effectors that were not present in our effector library. Alternatively, inhibition of host protein synthesis may result from the combined effects of multiple L. pneumophila effectors (which would not have been detected in our one-by-one effector screen) or the infection process itself.

The above results showed that induction of IL-1α by ∆7 L. pneumophila correlates with inhibition of host protein synthesis. We therefore wished to determine whether inhibition of host protein synthesis is sufficient to cause IL-1α release in vitro and in vivo. To recapitulate TLR signaling that occurs during L. pneumophila infection, BMDMs were treated with 10 ng/ml Pam₃CSK₄, a TLR2 ligand. This treatment induced transient intracellular IL-1α protein (Fig. 6A) but did not result in significant IL-1α release (Fig. 6B). BMDMs were therefore additionally treated with 50 ng/ml ExoA, a toxin made by Pseudomonas aeruginosa, which blocks translation by inhibiting the activity of elongation factor 2a (reviewed in Refs. 48 and 49). As with TLR stimulation, ExoA treatment alone was insufficient to induce IL-1α production. However, we found that treatment of BMDMs with both Pam₃CSK₄ and ExoA combined to induce release of IL-1α at 24 h postinfection (Fig. 6A, 6B). ExoA appeared to have two important effects that might explain its role in IL-1α release. First, in contrast to the transient induction of IL-1α induced by TLR signaling alone, additional treatment with ExoA caused the sustained presence of intracellular IL-1α protein at 24 h posttreatment, similar to what is seen in L. pneumophila infection (Fig. 6A). The sustained production of IL-1α protein was associated with a prolonged elevation of Il1a mRNA (Fig. 6C). Second, ExoA caused cell death by 24 h posttreatment (Fig. 6D), which may explain how intracellular accumulated IL-1α is released from these macrophages. ΔflaA and Δ5ΔflaA L. pneumophila–infected macrophages, which also experience a block in host protein synthesis, show sustained transcriptional induction, release IL-1α from the cell, and undergo cell death at 24 h postinfection (Fig. 6A–D). Importantly, the death of L. pneumophila–infected cells does not appear to depend on bacterial replication because cell death and IL-1α release still occurred when bacterial replication was curtailed by removal of thymidine from the medium. We speculate that inhibition of protein synthesis may be responsible for induction of host cell death. Protein synthesis inhibition and TLR stimulation also synergized to induce elevated IL-1α production in vivo (Fig. 6E). Taken together, these findings suggest that TLR activation in concert with translation inhibition can recapitulate IL-1α production and release in response to L. pneumophila infection and that this treatment is sufficient to induce release of IL-1α.

Legionnaires’ disease is an inflammatory pneumonia associated with a pronounced influx of neutrophils to the lung (3, 5). The recruitment of neutrophils to the lung is important for controlling bacterial burden; however, excessive neutrophil recruitment also can be detrimental to the host and may be responsible for immune pathologies associated with Legionnaires’ disease (3, 5). Thus, the host must tightly regulate the recruitment of neutrophils to the site of infection. In animal models of L. pneumophila infection, neutrophil recruitment has been shown to be important for protecting the host (6, 8, 9), yet the mechanism for this recruitment has remained unclear. A number of studies have demonstrated that MyD88 is an important host factor that protects mice from L. pneumophila infection (6, 10–13), and the IL-1R has been shown to be the critical receptor upstream of MyD88 that controls the recruitment of neutrophils to the lung in response to L. pneumophila infection (6). Indeed, it has been shown that IL-1R signaling is required in AECs to induce chemokines, such as CXCL1 and CXCL2, which then recruit neutrophils to the site of infection (6).

In our study, we identify the cytokine IL-1α as a critical initiator of IL-1R–dependent neutrophil recruitment to the lungs of L. pneumophila–infected mice. We find that IL-1α, but not IL-1β, precedes neutrophil recruitment to the lung, and we show that IL-1α is generated specifically by cells in the hematopoietic compartment (presumably infected macrophages). Given these data, we therefore propose a model by which IL-1α is produced by alveolar macrophages in response to virulent L. pneumophila and signals through the IL-1R on AECs, amplifying the original signal and generating chemokines, which recruit the initial wave of neutrophils to the lung. Importantly, at time points later than 12 h postinfection, IL-1α and IL-1β can both signal through the IL-1R and compensate for the loss of each other. Our data suggest that IL-1α is one of the earliest cytokines produced in response to L. pneumophila in vivo and thus initiates the recruitment of neutrophils and the inflammatory response to L. pneumophila in vivo.

Similar to L. pneumophila, Streptococcus pneumoniae leads to a severe pneumonia associated with massive influx of neutrophils. In mouse models of S. pneumoniae infection in the lung, Il1a/I1b^−/− double-knockout and Il1b^−/− mice are more susceptible to disease and have decreased clearance of bacteria from the lung (50). Moreover, Il1r1^−/− mice have increased bacterial burden in the lung and decreased neutrophil recruitment to the lung (51). Macrophage uptake of S. pneumoniae induces inflammasome activation and IL-1β release, which can signal to epithelial cells to recruit neutrophils by releasing the chemokine CXCL8 (51). Studies with S. pneumoniae suggest a model whereby activated macrophages secrete IL-1β, which signals through the IL-1R of AECs, thus leading to the production of chemokines, which recruit neutrophils to the site of infection (50, 51). This proposed mechanism is similar to the mechanism that we propose for L. pneumophila infections, except that it appears that IL-1α, rather than IL-1β, is the dominant cytokine early during L. pneumophila infections. These studies with S. pneumoniae suggest that amplification of early responses to infection by IL-1R signaling in AECs may be a conserved immune strategy important for recruiting neutrophils in response to bacterial infections. Importantly, the role for IL-1α in S. pneumoniae infections remains unclear.

In addition to S. pneumoniae, IL-1R signaling has been shown to be important for host protection from numerous pathogens, including Listeria monocytogenes (52–54), Mycobacterium tuberculosis (55–57), P. aeruginosa (58), Staphylococcus aureus (59), Klebsiella pneumoniae (60), and Candida albicans (61). In many of these infections, the mechanism by which IL-1R provides protection is not clear, and the relative roles of IL-1α and IL-1β have not been elucidated. One study that dissected the relative roles of IL-1α and IL-1β during M. tuberculosis infection found each cytokine played essential and nonredundant roles in vivo (62). This study, along with our results showing that IL-1α is of primary importance in early responses to L. pneumophila in vivo, suggest it will be worthwhile to examine more carefully the relative contributions of IL-1α and IL-1β in mediating IL-1R–dependent responses to other pathogens as well.

The molecular mechanism leading to IL-1α production has remained elusive (28). This is in stark contrast to IL-1β production, where intensive effort over the past decade has defined the mechanisms leading to IL-1β release downstream of inflammasome activation (reviewed in Ref. 18). Our data suggest that equal attention should be paid to the mechanisms of IL-1α production. Indeed, IL-1α has been shown to be induced in response to a number of bacterial pathogens including L. pneumophila (29, 30), L. monocytogenes (63), S. aureus (64), and M. tuberculosis (57, 62); however, the molecular mechanism of IL-1α production in response to these pathogens remains unsettled. Classic studies showed that IL-1α can be cleaved by the Calpain family of calcium-dependent proteases, but IL-1α does not appear to require processing to signal through the IL-1R (24, 25, 28). Some reports have suggested that IL-1α production in response to noninfectious stimuli such as toxins can involve activation of the caspase-1 or caspase-11 inflammasomes (24, 26, 27, 65). In addition, a previous report suggests that at 4 h postinfection Casp1/11^−/− mice have defects in IL-1α production in response to L. pneumophila infection in vivo (27). In contrast to these reports, our data show that Casp1/11^−/− mice have no defect in IL-1α production in response to L. pneumophila infection in vivo. This difference may be due to the different strains of L. pneumophila used in the two studies. Nevertheless, our results indicate that IL-1α and IL-1β can be produced via distinct but complementary pathways that provide alternative means to induce IL-1R signaling and immune defense in vivo. Given the critical importance of neutrophils in providing defense against numerous bacterial pathogens, it is perhaps to be expected that hosts would not rely on a single mechanism for activation of IL-1R signaling that could then be easily subverted or avoided.

Instead of a role for the inflammasome in IL-1α release, our data show that translation inhibition in concert with TLR stimulation is sufficient to induce IL-1α both in vitro and in vivo. Recent work from our laboratory and others have shown that, in mice and in C. elegans, translation inhibition can be sensed by the host and induce a number of immunological responses, including the production of proinflammatory cytokines (2, 30, 31, 66, 67). Previous research identified five L. pneumophila effectors that block host translation (31, 34, 45, 46). We previously found that this translation block induces a host stress response that can induce a subset of inflammatory cytokines, including IL-23 and GM-CSF (30, 31). Although the L. pneumophila ∆5 strain lacking the five effectors is partially defective in its ability to inhibit host protein synthesis (31) and is defective for IL-23 and GM-CSF induction, we confirmed in this study that cells infected with ∆5 L. pneumophila still experience a significant block in protein synthesis. Consistent with our finding that protein synthesis inhibition and TLR signaling are sufficient to induce IL-1α, we also find that ∆5-infected cells still produce IL-1α. In fact, even after identifying two novel bacterial effectors that block host translation and generating an L. pneumophila mutant (∆7) that lacks these effectors in addition to the original five effectors, we were still unable to abolish the Dot/Icm-dependent ability of L. pneumophila to inhibit protein synthesis and induce IL-1α. We propose several hypotheses to explain these results. First, there may be additional bacterial effectors in L. pneumophila that are not in our library of cloned effectors. Given that L. pneumophila is a generalist and has a multitude of natural hosts (68), it is possible that there is substantial additional redundancy encoded in the L. pneumophila genome. A second possibility is that there may not be a specific L. pneumophila effector that targets the host protein synthesis machinery; instead, the blockade of protein synthesis we observe may be the result of a host response to the infection process itself. Indeed, translation inhibition has long been recognized as a protective response during viral infections (69), and it is now evident that numerous bacterial infections can elicit host stress pathways that affect protein synthesis, for example via phosphorylation of eukaryotic initiation factor 2α (69). Last, it is possible that the ability of L. pneumophila to induce IL-1α is unrelated to protein synthesis inhibition. We tend not to favor this latter possibility because we found that inhibition of protein synthesis in conjunction with TLR signaling was sufficient to induce IL-1α, and moreover, it is clear that L. pneumophila infection results in both TLR signaling and inhibition of protein synthesis. Thus, although the mechanism of IL-1α production continues to elude the field, it seems likely that translation inhibition is at least one mechanism for IL-1α induction, even if other parallel mechanisms might also exist.

Taken together, our data show that IL-1α is a major cytokine responsible for the early recruitment of neutrophils to the lung in response to L. pneumophila infection from 0 to 12 h postinfection. We propose that a dominant role for IL-1α in protection against microbial infection may hold true for other pathogens, depending on the stage and mode of infection. Although much recent work has focused on the mechanisms of IL-1β production, our work suggests that IL-1α signaling can be as important, or indeed more important, than IL-1β signaling in vivo. Indeed, it is probably evolutionarily advantageous for hosts to encode multiple parallel pathways to induce IL-1R signaling, given the critical role that the IL-1R appears to play in orchestrating neutrophil recruitment and other immune responses in vivo.

We thank members of the Vance and Barton Laboratories for discussions and Norver J. Trinidad for technical assistance. We also thank April E. Price and Meghan A. Koch for experimental advice, discussions, and technical assistance. We thank Ralph Isberg for the L. pneumophila effector library.

The authors have no financial conflicts of interest.