Elevated CO₂ suppresses specific Drosophila innate immune responses and resistance to bacterial infection

Abstract

Elevated CO₂ levels (hypercapnia) frequently occur in patients with obstructive pulmonary diseases and are associated with increased mortality. However, the effects of hypercapnia on non-neuronal tissues and the mechanisms that mediate these effects are largely unknown. Here, we develop Drosophila as a genetically tractable model for defining non-neuronal CO₂ responses and response pathways. We show that hypercapnia significantly impairs embryonic morphogenesis, egg laying, and egg hatching even in mutants lacking the Gr63a neuronal CO₂ sensor. Consistent with previous reports that hypercapnic acidosis can suppress mammalian NF-κB-regulated innate immune genes, we find that in adult flies and the phagocytic immune-responsive S2* cell line, hypercapnia suppresses induction of specific antimicrobial peptides that are regulated by Relish, a conserved Rel/NF-κB family member. Correspondingly, modest hypercapnia (7–13%) increases mortality of flies inoculated with E. faecalis, A. tumefaciens, or S. aureus. During E. faecalis and A. tumefaciens infection, increased bacterial loads were observed, indicating that hypercapnia can decrease host resistance. Hypercapnic immune suppression is not mediated by acidosis, the olfactory CO₂ receptor Gr63a, or by nitric oxide signaling. Further, hypercapnia does not induce responses characteristic of hypoxia, oxidative stress, or heat shock. Finally, proteolysis of the Relish IκB-like domain is unaffected by hypercapnia, indicating that immunosuppression acts downstream of, or in parallel to, Relish proteolytic activation. Our results suggest that hypercapnic immune suppression is mediated by a conserved response pathway, and illustrate a mechanism by which hypercapnia could contribute to worse outcomes of patients with advanced lung disease, who frequently suffer from both hypercapnia and respiratory infections.

An average human produces 450 L of CO₂ per day (1) and elevated levels of CO₂ (hypercapnia) in the pulmonary and/or circulatory system are associated with worse outcomes in patients with cystic fibrosis (CF) (2) and chronic obstructive pulmonary disease (COPD) (3, 4), currently the fourth leading cause of death in the US (5). In vitro and animal studies have shown that hypercapnia can suppress mammalian inflammatory responses (6–8), including NF-κB-regulated cytokine production (9–11), which could contribute to the poor outcomes of patients with COPD and CF who frequently suffer from both hypercapnia and bacterial lung infections (12, 13). Hypercapnic immune suppression may be evolutionarily conserved because hypercapnia, in combination with hypoxia, also suppresses innate immune responses in shrimp, oysters, and crabs (14–16). However, the cellular pathways that respond to CO₂ and the physiological effects of hypercapnia are poorly understood (reviewed in refs. 17–19).

CO₂-sensing pathways in animals have been best defined in the nervous system (17, 20, 21). In Drosophila, the olfactory receptors Gr63a and Gr21a are both required for avoidance behavior to CO₂ (22, 23). In C. elegans, CO₂ avoidance is mediated by specific neurons (24, 25). The best characterized non-neuronal sensor of CO₂ to date is soluble adenylyl cyclase (26, 27, reviewed in ref. 28). However, soluble adenylyl cyclases have been lost in many evolutionary lineages, including plants, yeast, Drosophila, and C. elegans (29). Therefore, responses to CO₂ in these species must use other mechanisms. We sought to develop a genetically and molecularly tractable system for defining non-neuronal CO₂ response pathways. We chose Drosophila because of extensive conservation of the hypoxia (30), nitric oxide (NO) (31), and nearly all other major signal transduction pathways. Further, Drosophila possesses a well-characterized, multicomponent innate immune system controlled by conserved signaling pathways that include NF-κB-family transcription factors (32, 33). The organism's powerful in vivo and in vitro genetics combined with the ability to rapidly assay immune and physiological responses has resulted in manifold contributions to our understanding of mammalian innate immunity and immunity in general. These include identification of the Toll family of receptors (34) and RNAi as an antiviral mechanism (35, 36), functioning of NF-κB and Immune Deficiency (IMD)/TNF pathways (32, 33, 37), defining roles of NO, Wnt, and insulin signaling in immunity (31, 38, 39), increasing understanding of host tolerance versus resistance in pathogenesis (40), and describing immunological roles of autophagy (41). Drosophila has been especially valuable for investigating interactions between the immune system and environmental factors, including endosymbionts, mating, circadian rhythm, feeding, and native microbiota (e.g., 42, 43).

In this report we provide evidence that Drosophila melanogaster has specific physiological responses to hypercapnia. Notably, hypercapnia suppresses expression of a subset of antimicrobial peptides regulated by highly conserved NF-κB pathways, and there is a concomitant decrease in resistance of adult flies to specific bacterial pathogens. These results establish Drosophila as a general model for defining non-neuronal CO₂ signaling pathways and as a specific model for investigating suppression of innate immune responses by hypercapnia.

Results

Hypercapnia Causes Specific Effects on Drosophila Physiology Independent of Known Neuronal CO₂-Sensing Pathways.

To determine whether hypercapnia affects Drosophila physiology, we exposed flies to 13% CO₂ (P_CO2 94 mm Hg) and 19.5% CO₂ (P_CO2 140 mm Hg), while maintaining O₂ at 21% (see SI Text). These CO₂ levels are well below the >35% concentration at which CO₂ becomes anesthetic (44). Hypercapnia causes concentration-dependent defects in embryonic development. Thirteen percent CO₂ results in 20% of embryos having moderate defects such as malformations of the airway system (compare Fig. 1 A and B) and significantly slows hatching of eggs laid by normocapnic mothers (Fig. 1 D and E). At 19.5% CO₂, there is severe disruption of embryonic development with approximately 30% of exposed embryos showing large-scale patterning and morphogenesis defects (Fig. 1C), and >70% of eggs failing to hatch (Fig. 1E). The life span of adult flies is not affected by 13% CO₂ (Fig. 3A), but hypercapnia does cause a concentration-dependent reduction in the number of eggs that females lay (Fig. 1F). These effects of CO₂ on development and fertility are consistent with those we recently reported in C. elegans (45), with the exception that 19% CO₂ increases worm life span. Importantly, flies homozygous for a null mutation in the neuronal CO₂ receptor Gr63a (22, 23) are as sensitive to hypercapnia as wild-type flies in the egg hatching and laying assays (Fig. 1 D–F). Thus, many physiological effects of hypercapnia are mediated by an as-yet uncharacterized CO₂ response pathway(s).

Fig. 1.

Hypercapnia affects Drosophila development and physiology independently of neuronal CO₂ sensing. (A–C) Drosophila embryonic development is disrupted by hypercapnia as revealed by luminal staining of the embryonic tracheal (airway) system whose morphogenesis requires interaction with many distinct tissues. Culturing wild-type embryos in 13% CO₂ causes moderate defects in 20% of embryos (B, n = 111), while 19.5% CO₂ causes severe defects in 28% of embryos (C, n = 71) and increases the prevalence of moderate defects to 41%. Moderate defects (B) include breaks in the main airways (arrowheads) and missing or ectopic interconnections of dorsal branches (arrows). Severely abnormal tracheal development reveals gross embryonic patterning and/or morphogenesis defects (C). Representative images for each phenotype are shown. (D and E) Hypercapnia (24 h, D) decreases hatching of eggs laid in normocapnia by wild-type (WT) or mutant flies homozygous for a null allele in the neuronal CO₂ receptor, Gr63a. In 13% CO₂, over 90% of WT embryos hatch after 48 h (E), but in 19.5% CO₂ only approximately 30% hatch. (F) Hypercapnia reduces the number of eggs laid in 48 h by WT and Gr63a-null females mated in normocapnia. (G) As in mammalian cells (69), hypercapnia (1 h) causes endocytosis of the Na,K-ATPase in S2 cells. (H) Suppression of the antimicrobial peptide (AMP) Diptericin by hypercapnia in S2* cells is concentration-dependent. (Protein levels of AMPs were not assessed because antibodies against Drosophila AMPs are not available.) *, P < 0.05; **, P < 0.005.

Fig. 3.

Hypercapnia decreases resistance of Drosophila to specific bacterial infections. (A) Hypercapnia does not affect Drosophila life span. (B–H) Hypercapnia slightly increases death of flies inoculated with sterile PBS (B) or with E. coli (E), but significantly increases mortality at CO₂ levels as low as 7% after inoculation with A. tumefaciens (C), the human pathogen S. aureus (D), and the Drosophila natural pathogen E. faecalis (F). Immune suppression does not require the neuronal CO₂ receptor Gr63a (G). (H) Pretreatment of flies with 9% CO₂ before S. aureus infection in air is sufficient to increase mortality, even when flies are cultured in air after inoculation. For A–H, unless otherwise noted, flies were exposed to indicated CO₂ level for 24 h before inoculation and returned to hypercapnia until end of assay. We show representative results for the lowest CO₂ levels at which significant effects on mortality were consistently observed. All experiments were done in triplicate and the trial with the middle P value is shown. (I–L) Hypercapnia increases the bacterial load for strains causing increased mortality during hypercapnia. Horizontal lines show medians. Calculation of P values described in SI Text. CFU, colony forming units. (M–P) Effects of hypercapnia on bacterial growth. Note that S. aureus growth is dramatically reduced in 7% CO₂ even though hypercapnia increases mortality of flies infected with S. aureus. Error bars smaller than the data-point symbols are not shown.

We next asked whether Drosophila shares with mammals any specific response to hypercapnia. We previously reported that in human alveolar epithelial cells, hypercapnia causes endocytosis of the Na,K-ATPase (46). Analysis of surface abundance of Na,K-ATPase in Drosophila S2 cells reveals that hypercapnia also causes concentration-dependent endocytosis of the sodium pump in S2 cells (Fig. 1G). This result supports the existence of cell-autonomous CO₂ responses that do not depend on the neuronal CO₂-sensing pathways. That both Drosophila and human cells endocytose their Na,K-ATPase in response to hypercapnia suggests that some CO₂ responses are conserved between mammals and flies.

Hypercapnia Causes Specific Effects on Gene Expression.

To investigate the molecular basis of the physiological effects of hypercapnia and identify CO₂-responsive promoters to use as markers for dissecting CO₂-signaling pathways, we performed microarray analysis on adult flies. Similar to the limited changes in gene expression seen in neonatal mice raised in CO₂ (47), exposure of adult flies to 13% CO₂ for 24 h causes fewer than 500 genes to be up-regulated or down-regulated >1.5-fold, and fewer than 10 by >10-fold. Importantly, the regulated genes define discrete physiological functions. All of the up-regulated gene ontology (GO) families have GO annotations relating to metabolic functions (Fig. S1A), and 67% of the down-regulated gene ontology families have either immune- or fertility-related GO annotations (Fig. S1B). The strong down-regulation of the chorion and vitelline membrane genes required for egg production (Fig. S2A) is consistent with the dramatic decrease in fecundity of flies in hypercapnia. Critically, CO₂ does not induce genes characteristic of responses to hypoxia, heat shock, or oxidative stress (Fig. S2 B–D), indicating that the responses to elevated CO₂ are mediated by distinct pathways. Together, these results show that in Drosophila, hypercapnia does not cause global changes in gene expression, but instead has a very specific transcriptional signature.

Hypercapnia Suppresses Select Antimicrobial Peptide Genes.

The microarray data show that the down-regulated genes with immune-related GO annotations that are enriched in elevated CO₂ conditions are antimicrobial peptides (AMPs), which are important effectors of the Drosophila innate immune system (Fig. 2A). This observation parallels previous reports indicating that hypercapnic acidosis can suppress innate immune responses in mammals (6–11, 48, 49). Given the prevalence of pulmonary infections in CF and COPD patients (12, 13) and that hypercapnia is a risk factor for mortality in both of these debilitating diseases (2–4), we focused on defining the effects of elevated CO₂ on the fly innate immune system. We confirmed and extended the microarray results using quantitative real-time reverse transcriptase PCR (qPCR) and found that hypercapnia can indeed suppress expression of AMPs in adult flies (Fig. 2B).

Fig. 2.

Hypercapnia down-regulates specific antimicrobial peptides. (A and B) Hypercapnia (24 h) suppresses specific AMPs in vivo in unchallenged adults as shown by microarray analysis (A) and confirmed by quantitative PCR (B). (C and D) Hypercapnia suppresses specific AMPs in vitro (10 h CO₂ for C and D; PGN challenge at 5 h in D). *, P < 0.05; **, P < 0.005. PGN = E. coli peptidoglycan; Drs = Drosocin; Att = Attacin; Dpt = Diptericin; Cec = Cecropin; Def = Defensin; Drm = Drosomycin; Mtk = Metchnikowin.

To identify a simplified in vitro system for investigating hypercapnic immune suppression, we tested the effects of hypercapnia on the expression and induction of AMPs in Drosophila S2* cells, a phagocytic immune-responsive cell line whose viability we determined to be unaffected by 13% CO₂ (Fig. S3A). In flies, AMPs are regulated by one or both of the well characterized and conserved Toll and TNF-like IMD pathways (32, 50). Since S2* cells in culture lack the extracellular components required to activate the Toll receptor, E. coli peptidoglycan (PGN) induces the IMD pathway. However, there is likely to be some cross-talk to intracellular parts of the Toll pathway as well (51, 52). PGN challenge of S2* cells induces expression of all AMPs in air and CO₂ (Fig. 2 and Fig. S4D), but in hypercapnia the induced levels of some AMPs are suppressed (Fig. 2 C and D and Fig. S4D). Diptericin (Dpt), Attacin, and Drosomycin are consistently suppressed 3- to 5-fold, whereas Metchnikowin is not (Fig. 2 C and D). Importantly, hypercapnia also differentially suppresses AMP expression in whole flies (Fig. 2 A and B), and the responses in S2* cells reasonably approximate those observed in vivo. Further, as treating S2* cells with even a modest 5% CO₂ causes a 50% suppression of Dpt induction (Fig. 1H), S2* cells provide a powerful in vitro tool for elucidating the molecular details of CO₂-response pathways.

Hypercapnia Increases Mortality to Specific Bacterial Infections.

Antimicrobial peptides are a crucial element in the fly's defense against pathogens (53). We therefore asked whether hypercapnia reduces the ability of flies to survive infection with bacteria, including the natural Drosophila pathogen E. faecalis (54), the human pathogen S. aureus, which is commonly isolated from COPD and CF patients (13, 55), and A. tumefaciens, which has previously been reported to increase mortality in flies lacking a conserved component of NF-κB immune pathways (56). Elevated CO₂ slightly decreases survival of flies inoculated with sterile PBS (Fig. 3B) or with E. coli (Fig. 3E), which is snot considered a Drosophila pathogen (53). Wounding modestly reduces average life span (compare Fig. 3 A and B air conditions), but hypercapnia alone does not (Fig. 3A). Therefore, the marginally decreased survival in elevated CO₂ after non-pathogenic inoculations suggests that hypercapnia could interfere with wound-healing, which is closely linked to AMP expression (57). Survival of flies infected with E. faecalis (Fig. 3F) was significantly decreased by moderately elevated CO₂ (13% CO₂). Strikingly, CO₂ levels as low as 7% increase mortality to A. tumefaciens and S. aureus (Fig. 3 C and D). Furthermore, elevated CO₂ increases mortality to A. tumefaciens in flies lacking Gr63a (Fig. 3G), indicating that hypercapnic immune suppression is mediated by mechanisms independent of known neuronal CO₂-sensing pathways.

A critical issue is whether decreased survival of infected flies results from effects of hypercapnia on the host or the pathogen. To test for host effects, we exposed flies to 9% CO₂ for 24 h and returned them to normocapnia after inoculation with S. aureus. Hypercapnic pretreatment of the host significantly increases mortality after infection, although the pathogen was never exposed to elevated CO₂ (Fig. 3H). Thus, hypercapnia compromises the ability of flies to combat bacterial infections independently of effects of hypercapnia on the pathogen.

We also investigated whether hypercapnia affects pathogens by measuring their growth rates in LB media at elevated CO₂ levels. Growth of E. coli and E. faecalis in media is not significantly affected by elevated CO₂ (Fig. 3 M and N). However, despite the increased mortality of flies infected with A. tumefaciens and S. aureus, growth of these bacteria is actually markedly reduced by hypercapnia (Fig. 3 O and P). Because decreased bacterial growth would be expected to improve fly survival, the observed increase in mortality underscores the deleterious effects of hypercapnia on Drosophila host defenses.

Hypercapnia Decreases the Resistance to Bacterial Infections.

The survival of a host during infection is determined by its ability to resist the growth of pathogens and its ability to tolerate the presence of an infection. To distinguish whether resistance or tolerance is affected by hypercapnia, we determined bacterial loads in flies infected with E. coli, E. faecalis, and A. tumefaciens (see SI Text). Hypercapnia increases bacterial loads in WT and Gr63a mutant flies infected with strains that increase mortality (Fig. 3 J–L). Thus, hypercapnia reduces resistance to some bacterial infections. These results are consistent with hypercapnic acidosis increasing lung bacterial load in a rat pneumonia model (8), and with hypercapnic hypoxia increasing bacterial load in infected marine invertebrates (15).

Hypercapnia Rapidly Suppresses Innate Immune Responses in Vitro.

To further investigate the nature of hypercapnic immune suppression, we determined how hypercapnia affects the kinetics of innate immune responses in S2* cells. Hypercapnia suppresses rather than delays innate immune responses of PGN-challenged S2* cells, as the magnitude of the responses at all times is reduced (Fig. S4A). This suppression exhibits rapid onset (Fig. S4B) and recovery (Fig. S4C).

Hypercapnic Immunosuppression Is Not Mediated by NO or Acidosis.

As discussed above, the effects of hypercapnia on Drosophila physiology and the immune responses of S2* cells are not mediated via neuronal CO₂ sensing, and hypercapnia acts via pathways distinct from hypoxia, heat shock, and oxidative stress. Because non-neuronal CO₂-sensing pathways have not been defined, we tested possible roles of candidate signaling pathways in hypercapnic immune suppression. Nitric oxide (NO) has important roles in innate immune function in Drosophila and mammals (31, 58), and hypercapnia has been proposed to modulate NO-dependent pathways and inflammatory oxidants (59). However, hypercapnic suppression of Dpt in S2* cells is unaffected by either a NO synthase inhibitor or a NO donor (Fig. 4A), indicating that hypercapnic immune suppression is not mediated via signaling by NO.

Fig. 4.

Hypercapnia suppresses immune responses independent of pH, NO, and proteolytic activation of Rel. (A) Suppression of Diptericin (Dpt) in PGN-challenged S2* cells is not attenuated or enhanced by the NO synthase inhibitor L-NAME (1 mM) or the NO donor SNAP (1 mM). (B) Hypercapnia suppresses Dpt induction 5-fold when the culture medium is adjusted to pH7.0 with NaOH, but Dpt induction is only suppressed approximately 2-fold by Mops at pH6.5 and by NaCl and Hepes at pH7.0. (C) Hypercapnia does not alter Relish (Rel) cleavage in response to PGN challenge. GFP immunoblot of whole-cell lysates of S2 cells expressing GFP-Rel. (D and E) Models for hypercapnic suppression of innate immune effectors (see text for discussion). *, P < 0.05; **, P < 0.005 between CO₂ condition and equivalently treated air condition (A) or untreated air condition (B).

Another candidate mediator of hypercapnic immune suppression is acidosis, which is known to regulate immune responses [reviewed in (60, 61)]. CO₂ reacts with water to form carbonic acid (H₂CO₃) and previous investigations of hypercapnic innate immune suppression in mammalian immune cells have implicated—but not proven—acidosis as the critical mediator (9, 11, 62). Here, we show that elevated CO₂ levels can suppress immune responses in S2* cells independent of extracellular pH effects. Acidifying S2* cell culture media with 19 mM Mops in normocapnia causes only a 2-fold suppression of Dpt expression, less than half that caused by 13% CO₂, which is associated with an equivalent decline in pH (Fig. 4B). Notably, 25 mM NaCl or Hepes at neutral pH causes suppression equal to that of acidosis, making it difficult to discern acidotic from ionic or non-specific effects. We definitively demonstrate effects of CO₂ independent of extracellular acidosis by maintaining the S2* cell media at neutral pH during exposure to 13% CO₂ and finding that hypercapnia suppresses Dpt induction to the same extent seen in hypercapnic acidosis (Fig. 4B, black bars). Thus, in Drosophila cells, hypercapnia causes a consistent pattern of immune suppression distinct from the effects of extracellular acidosis. Importantly, this result parallels our results with mammalian alveolar macrophages that demonstrated that CO₂ suppresses expression of particular NF-κB regulated cytokines independent of acidosis (48).

Hypercapnia Acts Downstream of, or in Parallel to, Relish Activation.

How does hypercapnia modulate innate immune responses? We used cultured S2 cells to investigate how hypercapnia suppresses AMPs by examining PGN-induced activation of Relish (Rel), a homolog of the mammalian Rel/NF-κB transcription factors. Rel acts in the IMD pathway and contains an N-terminal Rel homology domain (RHD) that is endoproteolytically cleaved from a C-terminal inhibitory IκB-like region in response to PGN (63). The RHD then binds κB-sites on AMP promoters and induces their expression (32). In S2 cells exposed to 13% hypercapnia, there is no reduction in the PGN-induced proteolytic cleavage of Rel (Fig. 4C). Therefore, hypercapnia inhibits expression of Rel targets downstream of, or in parallel to, proteolytic activation of Rel. Together with work in human and mouse macrophages showing that hypercapnia suppresses NF-κB targets independent of acidosis and without blocking degradation of IκBα (48), our results support the idea of a conserved mechanism of hypercapnic immune suppression.

Discussion

We have systematically investigated mechanisms that could mediate responses to elevated CO₂. The effects of hypercapnia are present in flies lacking the Gr63a gustatory CO₂ receptor and are therefore mediated by mechanisms that are distinct from known neuronal CO₂-sensing pathways, and that can act cell autonomously. These effects are not simply a consequence of cellular damage because hypercapnia does not induce changes in gene expression characteristic of other environmental insults such as hypoxia, oxidative stress, or heat shock, nor does hypercapnia decrease fly life span or cell viability.

In Drosophila, the majority of gene expression changes in hypercapnia are accounted for by increased expression of metabolic genes and by down-regulation of reproductive and immune genes. This specific profile of changes is distinguishable from those of other immune-related insults such as Mycobacterium infection, which down-regulates metabolic genes and up-regulates the innate immune response (Fig. S5), and from Pseudomonas infection, which appears to suppress AMPs differently than CO₂ (Fig. S6). Interestingly, despite some distinctions, many metabolic and other genes regulated by starvation are similarly regulated by hypercapnia (Fig. S7). The effects of CO₂ on Drosophila innate immunity are most likely not due to decreased feeding, since we find a similar pattern of AMP down-regulation in S2* cells as in adult flies. We conclude that the gene expression changes we observe are specific and unique to hypercapnia.

To begin dissecting the mechanisms of immune gene regulation by hypercapnia, we have investigated its effects on the TNF-like IMD/Rel pathway. Our results show that hypercapnia inhibits expression of Rel targets downstream of, or in parallel to, proteolytic cleavage of the IκBα-like domain of Rel. This result parallels the findings of Wang et al. (48) who show in mammalian macrophages that hypercapnia suppress NF-κB-regulated genes without affecting proteolysis of IκBα. We propose two models for how a CO₂ response pathway might suppress transcription of Drosophila Rel targets such as Dpt. In the first model, a CO₂ response pathway may negatively regulate one of the components of the Rel transcription complex or block Rel nuclear import (Fig. 4D). Alternatively, and analogous to the regulation of hypoxia-responsive genes by the HIF transcription factor (reviewed in ref. 64), a CO₂-response pathway would activate a hypothetical CO₂-responsive factor (CO₂RF) that binds specific CO₂-response elements (CO₂RE) and then inhibit the Rel complex or directly suppress Dpt expression (Fig. 4E). The conceptual difference between the two models is that, in the first, genetic removal of the component of the Rel complex targeted by the CO₂-response pathway would significantly alter Dpt transcription even in the absence of CO₂. In contrast, removal of the CO₂-responsive factor in the second model would render the promoter unresponsive to CO₂ regulation, without affecting basal or induced transcription of Dpt in normocapnia.

Why should elevated CO₂ levels cause the physiological responses we observe? One possibility is that CO₂ serves as a read-out of metabolic activity, with elevated CO₂ indicating excessive metabolic load. We have shown that hypercapnia suppresses select physiological functions that are known to be metabolically demanding, including immune responses (65), egg-laying, and Na,K-ATPase activity, which consumes up to 40% of the energy supply in some cell types (66). This may allow for the reallocation of energy to more immediate needs such as powering the flight muscles. Consistent with this hypothesis, our microarray data reveal that all of the Drosophila gene ontology families up-regulated by hypercapnia are involved in metabolism (Fig. S1A) and there is commonality between the genes regulated during hypercapnia and starvation (Fig. S7). Furthermore, our previous work has shown that in mammalian cells, hypercapnia activates AMP-activated protein kinase (AMPK), a central regulator of metabolism (46). AMPK is activated by conditions that deplete ATP reserves, such as hypoxia or rapid muscle contraction. We suggest that in some tissues, CO₂ could act as a diffusible signal that regulates metabolism. A metabolically active tissue would produce CO₂ that could decrease activity of a responding tissue to preserve energy resources for the source tissue.

Hypercapnic suppression of innate immune and inflammatory responses has important implications for human health. Previous reports have suggested that hypercapnia can have beneficial effects when “permissive hypercapnia” regimens are used during mechanical ventilation (reviewed in ref. 67). Reduced inflammation is also observed when CO₂ is used as an insufflatant during laparoscopic surgery (68). In these situations, which do not involve defenses against pathogens, hypercapnic suppression of inflammatory responses may reduce tissue damage and improve outcomes. However, when pathogens are present, as in patients with COPD or cystic fibrosis, immune suppression by hypercapnia could increase susceptibility to, or exacerbate consequences of, bacterial infections. Experimental support of this possibility is provided by O'Croinin et al. (8) who recently showed that in a bacterial pneumonia model, rats exposed to 5% CO₂ had higher lung bacterial counts and more structural damage than normocapnic controls. Together, the Drosophila and rat results suggest that hypercapnia is not simply a marker of disease status, but rather that it can actively contribute to the progression of infection. The current lack of understanding of the molecular pathways that mediate CO₂ responses severely limits our ability to assess and intervene in pathological situations involving hypercapnia. Our findings establish Drosophila as a genetically and molecularly tractable system in which to define non-neuronal CO₂ response pathways that modulate host defenses and other physiologically important processes.

Materials and Methods

See SI Text for additional detail.

CO₂ exposures performed in BioSpherix C-Chambers (BioSpherix Ltd) fitted with a ProCO₂ regulator supplied with 20% CO₂/21% O₂/59% N₂. Unless otherwise noted, 5 h before PGN challenge, cells were resuspended in pre-equilibrated media.

Cell Culture.

Cells were primed with 1 μM ecdysone 15 h before CO₂ treatment and challenged with 100 ng/mL PGN for 5 h (Fig. 2D) or 2.5 h (Figs. 1H and 4 A and B, and (Fig. S4 B and C). Surface biotinylation was performed as previously described (69). L-NAME (1 mM) or 1 mM SNAP was added 10 h or 3 h, respectively, before CO₂ treatment as described in ref. 70. Media pH was adjusted by adding reagents to the indicated final concentrations: 19 mM Mops, 25 mM NaCl, 12.5 mM Hepes, and 25 mM NaOH (pre-equilibrated in CO₂ overnight).

Bacterial infection and CFU counts were performed as described previously (71, 72). Kaplan–Meier survival plots and P values generated using GraphPad Prism software from combined triplicates for a total of approximately 75 flies per survival curve. Inocula CFUs shown in Fig. S8.

Developmental and Fecundity Assays.

Developmental defects were assessed in embryos exposed to elevated CO₂ for 19 h. For the egg hatching assay, eggs laid by flies in air were collected for 1–2 h, counted, transferred to CO₂ conditions, and scored at 24 h or 48 h. Fecundity was determined as described in ref. 73.

Whole Fly Microarrays.

Triplicates of at least 30 5-day-old male and female flies were used as described in ref. 74. National Center for Biotechnology Information GEO accession number is GSE17444.

Statistical Analyses.

Two-tailed Student's t tests were used unless noted. Error bars indicate standard deviation. Air vs. CO₂ CFUs, t tests were performed on the natural logarithms of CFU counts as described in ref. 75.