IRG1 and Inducible Nitric Oxide Synthase Act Redundantly with Other Interferon-Gamma-Induced Factors To Restrict Intracellular Replication of Legionella pneumophila

Legionella pneumophila is one example among many species of pathogenic bacteria that replicate within mammalian macrophages during infection. The immune signaling factor interferon gamma (IFN-γ) blocks L. pneumophila replication in macrophages and is an essential component of the immune response to L. pneumophila and other intracellular pathogens. However, to date, no study has identified the exact molecular factors induced by IFN-γ that are required for its activity. We generated macrophages lacking different combinations of IFN-γ-induced genes in an attempt to find a genetic background in which there is a complete loss of IFN-γ-mediated restriction of L. pneumophila. We identified six genes that comprise the totality of the IFN-γ-dependent restriction of L. pneumophila replication in macrophages. Our results clarify the molecular basis underlying the potent effects of IFN-γ and highlight how redundancy downstream of IFN-γ is key to prevent exploitation of macrophages by pathogens.

IFN-␥ (28). NOX2 partners with phagosomal oxidase components to generate reactive oxygen species, which, like NO, can cause direct toxicity to phagocytized pathogens in neutrophils and macrophages (30,31). IRGM1 and IRGM3 are antimicrobial GTPases that participate in the disruption of membrane-bound, pathogen-containing compartments within phagocytes in the case of Toxoplasma (32,33) and Chlamydia (34). To date, however, a nonredundant role for IRGM1 and IRGM3 in disruption of the L. pneumophila-containing vacuole has not been established. Remarkably, Pilla et al. observed that macrophages derived from QKO mice retained restriction of L. pneumophila replication when stimulated with IFN-␥ (28). This study implicated the bacterial lipopolysaccharide (LPS) detector caspase 11 (CASP11), encoded by the gene Casp4, in some of the residual IFN-␥-dependent restriction of L. pneumophila replication in macrophages (28). Upon binding of bacterial lipopolysaccharide in the cytoplasm, CASP11 can trigger host macrophage pyroptosis, an inflammatory form of cell death (35,36).
Recently, Naujoks et al. implicated immune-responsive gene 1 (IRG1), encoded by the gene Acod1, in the IFN-␥-dependent immune response to L. pneumophila, demonstrating that driving Acod1 expression in macrophages was sufficient to suppress L. pneumophila replication (29). However, this study did not address whether macrophages deficient in IRG1 were impaired in the ability to restrict L. pneumophila when stimulated with IFN-␥. Like iNOS, IRG1 generates a potentially toxic metabolite (itaconate) and contributes to metabolic changes that occur in inflamed macrophages (37)(38)(39).
We recently described a mutant strain of L. pneumophila (ΔflaA ΔuhpC) that is able to replicate in macrophages treated with 2-deoxyglucose (2DG), an inhibitor of mammalian glycolysis (40). This strain allows us to probe the role that host cell metabolism plays in the immune response to L. pneumophila. In the present study, we use a combination of preexisting knockout mouse models, pharmacological treatment with 2DG and other drugs, CRISPR/Cas9 genetic manipulation of immortalized mouse macrophages, and primary BMMs from novel strains of CRISPR/Cas9-engineered mice to survey the factors required for IFN-␥-dependent restriction of L. pneumophila in macrophages. Ultimately, we demonstrate that iNOS and IRG1 are redundant in terms of IFN-␥-dependent restriction of L. pneumophila. Further, we identify six IFN-␥inducible factors, iNOS, IRG1, CASP11, NOX2, IRGM1, and IRGM3, which are responsible for the entirety of the IFN-␥-dependent restriction of L. pneumophila in macrophages.
2-Deoxyglucose partially reverses IFN-␥-dependent restriction of L. pneumophila in BMMs. We next investigated the possibility that IFN-␥ may act to restrict L. pneumophila not through induction of any single antimicrobial factor but by changing the metabolic landscape of the host macrophage to be unsuitable for the metabolic  needs of L. pneumophila. Specifically, we hypothesized that disruption of host macrophage glycolysis with 2-deoxyglucose (2DG) would interfere with IFN-␥-dependent restriction observed in BMMs. Macrophages infected with L. pneumophila and IFN-␥stimulated macrophages increase rates of aerobic glycolysis, which can be measured by increased consumption of glucose and increased secretion of lactate (1,40,46). First, we confirmed that 2DG blocked the increased glycolysis observed in L. pneumophilainfected BMMs stimulated with IFN-␥ ( Fig. 2A). We next tested the implication of glycolysis inhibition in IFN-␥-stimulated BMMs in terms of L. pneumophila replication. 2DG is metabolized to 2DG-phosphate (2DGP) in mammalian cells, which is directly antimicrobial (40). However, by taking advantage of a newly identified strain of L. pneumophila resistant to the direct antimicrobial effect of 2DG(P) in BMMs (ΔflaA ΔuhpC L. pneumophila) (40), we observed that addition of 2DG to BMMs rescued L. pneumophila replication in IFN-␥-treated macrophages by 100-to 200-fold (Fig. 2B). This result initially suggested that increased macrophage glycolysis plays a direct role in bacterial restriction. However, previous studies have demonstrated that while L. pneumophila has the capacity to metabolize glucose, it does not rely on glucose or glucose derivatives to fuel its replication in broth and is largely indifferent to perturbations in BMM glycolysis during infection (40,(47)(48)(49). To further probe the role that host macrophage glycolysis plays in terms of IFN-␥ restriction of L. pneumophila, we tested other conditions under which host macrophage glycolysis is impaired. BMMs lacking hypoxia-inducible factor 1␣ (HIF1␣) fail to upregulate glycolysis in response to inflammatory stimuli and have a defect in IFN-␥-mediated control of M. tuberculosis (22). We observed that HIF1␣-deficient BMMs resembled wild-type BMMs in terms of IFN-␥-dependent restriction and 2DG rescue of L. pneumophila replication (Fig. 2C). Replacement of glucose with galactose, which inhibits increased glycolysis in IFN-␥stimulated BMMs (22,50), also did not alter the ability of IFN-␥ to restrict or of 2DG to Interferon Gamma Bacterial Restriction ® rescue L. pneumophila replication (Fig. 2D). Further, IFN-␥ was able to mediate bacterial restriction, and 2DG was able to reverse this restriction, in BMMs cultured in glucosefree medium lacking any added sugar (Fig. 2D). Finally, we tested whether other inhibitors of glycolysis, 3-bromopyruvate (3BP) and sodium oxamate (NaO), recapitulated the effects of 2DG. Neither 3BP nor NaO reversed IFN-␥-dependent restriction of ΔflaA ΔuhpC L. pneumophila (the 2DG-resistant strain) or ΔflaA L. pneumophila (see Fig. S1 in the supplemental material). Together, these data indicate that glycolysis induction is not required for IFN-␥-mediated restriction of L. pneumophila replication in BMMs. This result suggests that effects of 2DG other than glycolysis inhibition are responsible for its interference with the cell-intrinsic IFN-␥-dependent immune response to L. pneumophila in BMMs. Some, but not all, unfolded protein response stimuli reverse IFN-␥-dependent inhibition of L. pneumophila. To determine potential "off-target" effects of 2DG that could be responsible for reversal of IFN-␥-mediated restriction of L. pneumophila, we performed a transcript profiling experiment. We measured global transcript abundance in BMMs exposed to TLR2 agonist Pam3CSK4 or infected with L. pneumophila and then stimulated with IFN-␥ Ϯ 2DG. Pathway analysis of transcripts upregulated under 2DG conditions indicated induction of endoplasmic reticulum stress, also known as the unfolded protein response (UPR) ( Fig. S2 and Table S1). 2DG is thought to trigger the UPR due to interference with protein glycosylation pathways in the endoplasmic reticulum (51). This led us to hypothesize that induction of the UPR perturbs IFN-␥dependent restriction of L. pneumophila replication in BMMs. In fact, we observed that other drugs that trigger UPR stress, including geldanamycin, brefeldin A, and dithiothreitol, also rescued L. pneumophila replication in IFN-␥-stimulated BMMs by ϳ10to 50-fold (Fig. 3A). However, not all drugs that trigger the UPR rescued L. pneumophila replication in IFN-␥-treated BMMs. For example, treatment of BMMs with the potent UPR inducers tunicamycin or thapsigargin did not reverse IFN-␥-mediated restriction of L. pneumophila replication (Fig. 3B).
One effect of UPR stress is arrest of protein translation via the PERK/EIF2␣ pathway, which can be reversed by the drug ISRIB (52). Importantly, ISRIB treatment did not interfere with 2DG-or geldanamycin-mediated rescue of L. pneumophila replication in IFN-␥-stimulated BMMs (Fig. 3C). This result indicates that reversal of IFN-␥-mediated restriction does not result from a global block in translation. We confirmed that UPR stimuli and ISRIB induced UPR-associated transcripts and inhibited ATF4-associated transcripts, respectively, via global transcript profiling (Fig. S3) (53). Taken together, these results suggest that while some UPR-triggering drugs can partially reverse IFN-␥-dependent restriction of L. pneumophila replication in BMMs, induction of the UPR is not sufficient to interfere with IFN-␥-mediated restriction. Additionally, the rescue of bacterial replication in IFN-␥-stimulated BMMs by UPR-triggering drugs does not act exclusively through general inhibition of protein translation.
IFN-␥ fully restricts L. pneumophila in BMMs lacking IRG1 but is only partially restrictive in BMMs lacking both IRG1 and iNOS. Our analysis above revealed that certain drugs that trigger the UPR rescue L. pneumophila replication in IFN-␥-stimulated BMMs, while others do not. We speculated that we could use these stimuli as a filter to look for transcripts associated with a restrictive versus permissive macrophage state. Using this logic to filter results from RNA sequencing (RNAseq) analysis of BMMs stimulated with Pam3CSK4 Ϯ IFN-␥ Ϯ UPR stimuli, we identified two genes, Nos2 (encoding iNOS) and Acod1 (encoding IRG1), whose transcript levels were elevated under restrictive conditions and lowered under permissive conditions (Fig. S4A). Since iNOS deficiency has no effect on IFN-␥-mediated control of L. pneumophila replication (e.g., Fig. 1B), we speculated that IRG1 may restrict L. pneumophila replication in IFN-␥-stimulated BMMs, as suggested (but not directly tested) previously (29). Using immortalized BMMs derived from C57BL/6 mice that inducibly express Cas9 (iCas9), we targeted Acod1 and Ifngr1 with guide RNAs to generate BMMs that lack expression of IRG1 and IFN-␥ receptor, respectively (Table 1 and Fig. S5). In comparison with Ifngr1-targeted BMMs, which failed to restrict L. pneumophila when stimulated with IFN-␥, we observed that Acod1-targeted immortalized BMMs retained the ability to restrict L. pneumophila replication upon stimulation with IFN-␥ (Fig. 4A). We next generated primary BMMs from Acod1 Ϫ/Ϫ mice derived on the C57BL/6NJ background (38). Similarly to immortalized BMMs, primary Acod1 Ϫ/Ϫ BMMs displayed intact IFN-␥dependent restriction of L. pneumophila (Fig. 4B). These results suggest that IRG1 activity alone is not required for restriction of L. pneumophila in IFN-␥-stimulated macrophages.
We next tested the hypothesis that iNOS and IRG1 activities are redundant in terms of restricting L. pneumophila in IFN-␥-stimulated BMMs. In line with this hypothesis, we observed a partial (ϳ10-fold) loss of restriction in Acod1 Ϫ/Ϫ BMMs treated with the iNOS inhibitor 1400W, indicating that in the absence of IRG1, iNOS function is required to mediate full restriction of L. pneumophila in IFN-␥-stimulated BMMs (Fig. 4C).
pneumophila replication (Fig. 4D to F). In an effort to pinpoint the additional genetic factors that mediate IFN-␥-dependent restriction of L. pneumophila in BMMs, we made use of existing QKO mice (28). We designed an experiment to test the hypothesis that the six factors implicated across our observations (iNOS and IRG1) and the studies by Pilla et al. (28) (QKO and CASP11) and Naujoks et al. (29) (IRG1) comprise the entirety of the IFN-␥-coordinated response to L. pneumophila in macrophages. We employed CRISPR/Cas9 to target Casp4 and Acod1 (Table 1) in QKO mouse embryos to generate three novel mouse strains: QKO mice that also lack functional CASP11 (QKO/C11), QKO mice that also lack functional IRG1 (QKO/IRG1), and QKO mice that additionally lack both CASP11 and IRG1 (6KO). As previously reported, and in line with our observations in Nos2 single-knockout BMMs, we observed that IFN-␥-dependent restriction of L. pneumophila in QKO BMMs was largely intact, indicating that no gene disrupted in these cells is absolutely required for restriction of L. pneumophila (Fig. 5A and B) (28). QKO/C11 BMMs did not lose IFN-␥-mediated restriction relative to QKO BMMs (Fig. 5A  Interferon Gamma Bacterial Restriction ® and B). In contrast, we observed a striking (ϳ100-fold) loss of restriction in QKO/IRG1 BMMs and a total loss of L. pneumophila restriction in 6KO BMMs stimulated with IFN-␥ ( Fig. 5A and B). We verified that QKO/IRG1 and 6KO BMMs do not display general defects in phagocytosis and endosome acidification (Fig. S6A). Additionally, we observed a significant reduction of nonpathogenic ΔdotA L. pneumophila (54, 55) CFU following phagocytosis by QKO/IRG1 and 6KO BMMs, indicating that early, IFN-␥independent antimicrobial responses are intact in these macrophages (Fig. S6B).
If the ability of 2DG to rescue L. pneumophila in IFN-␥-treated BMMs acts through inhibition of iNOS and IRG1, we would expect 2DG to have no effect in BMMs lacking expression of these factors. However, we observed that 2DG retained the ability to partially rescue L. pneumophila replication (by slightly less than 10-fold) in Nos2 Ϫ/Ϫ Acod1 Ϫ/Ϫ BMMs stimulated with IFN-␥ (Fig. 5C). Intriguingly, the rescue effect of 2DG was absent in QKO/IRG1 BMMs (Fig. 5C). Transcript profiling did not reveal an inhibitory effect of 2DG on expression of the other genes disrupted in the QKO/IRG1 background (Fig. S4B). This result indicates that 2DG may mediate some beneficial metabolic effect for L. pneumophila independently of regulating activity of iNOS and IRG1; however, these effects may require the other factors disrupted in the QKO background.
In sum, these results further underscore our previous observation that iNOS and IRG1 are redundant in terms of mediating a large proportion of the IFN-␥-dependent restriction of L. pneumophila in BMMs, as the addition of IRG1 deficiency to the QKO background profoundly disabled IFN-␥-mediated restriction. Further, our data reveal a partial role for CASP11 in control of L. pneumophila restriction in IFN-␥-stimulated BMMs, given the differences observed between QKO/IRG1 and 6KO BMMs. Finally, our results demonstrate that the six factors disrupted in 6KO BMMs, or a subset of those six that includes iNOS, IRG1, and CASP11, coordinate the entirety of the IFN-␥-dependent, cell-intrinsic control of L. pneumophila observed in BMMs.

DISCUSSION
Our results support a model in which IFN-␥ restricts L. pneumophila replication in mammalian macrophages through activation of multiple redundant factors, including iNOS and IRG1. To date, no study has identified any single IFN-␥-stimulated gene that fully accounts for the ability of macrophages to restrict L. pneumophila replication when stimulated with IFN-␥. Even QKO macrophages, which lack three other potentially antimicrobial factors in addition to iNOS, largely maintain the ability to restrict L. pneumophila, reinforcing the notion that redundant mechanisms contribute to IFN-␥mediated bacterial control in macrophages.
Our recent identification of a strain of L. pneumophila resistant to the direct antimicrobial effect of 2DG when growing in BMMs (40) allowed us to test the hypothesis that global disruption of macrophage metabolism interferes with the antimicrobial effects of IFN-␥. Indeed, 2DG partially reversed the restriction of L. pneumophila replication by IFN-␥ in BMMs. However, neither the glycolysis inhibition activity nor the UPR induction activity of 2DG, per se, appears to underlie the ability of this drug to subvert the antimicrobial effect of IFN-␥. Instead, 2DG appears to regulate the IFN-␥-dependent induction of iNOS and IRG1 via some as-yet-unidentified mechanism. In addition, there appears to be some effect of 2DG independent of iNOS and IRG1 regulation, suggesting that the drug interferes with the antimicrobial activities of IRGM1, IRGM3, and/or NOX2 in IFN-␥-stimulated macrophages. Ultimately, experimentation with 2DG and other stimuli that reversed IFN-␥-mediated restriction of L. pneumophila led us to the hypothesis that both iNOS and IRG1 are sufficient, and therefore redundant, in terms of mediating IFN-␥-coordinated immune response to L. pneumophila in macrophages.
A complex picture is emerging in terms of the role of IRG1 and the metabolite it produces, itaconate, during inflammation and infection. A direct antimicrobial role for itaconate via poisoning the bacterial glyoxylate pathway has been suggested for M. tuberculosis and L. pneumophila (29,37). IRG1 was shown to be an essential component of the immune response to M. tuberculosis, as Acod1 Ϫ/Ϫ mice succumbed more rapidly than wild-type mice to infection (56). However, IRG1 appeared to be required for regulation of non-cell-autonomous pathological inflammation, and there was no evidence for cell-intrinsic antimicrobial effects of itaconate (56). IRG1 has also been demonstrated to be protective in a model of Zika virus infection in neurons (57).
Interestingly, other studies have demonstrated anti-inflammatory effects of itaconate on myeloid cells, suggesting it may act as part of a negative feedback loop to control inflammation (39,58). Beyond production of itaconate, the disruption of oxidative metabolic pathways caused by IRG1 activity may promote antimicrobial metabolic shifts in macrophages. Ultimately, diverse cell-intrinsic and intercellular roles for IRG1 and itaconate likely contribute to the immune response to a broad array of pathogens. Our data demonstrate that IRG1 is not essential for the cell-intrinsic immune response to L. pneumophila in macrophages treated with IFN-␥. However, our data are consistent with the observation that IRG1 activity may be sufficient to restrict L. pneumophila replication, as previously reported (29). Both NO generated by iNOS and itaconate generated by IRG1 may be directly antimicrobial to L. pneumophila in macrophages stimulated with IFN-␥. Alternately or additionally, iNOS and IRG1 may act to restrict L. pneumophila replication via coordinating global changes in macrophage metabolism that restrict access to key bacterial metabolites or otherwise render the host macrophage inhospitable for bacterial growth.
Adding IRG1 and CASP11 deficiency to the QKO background revealed further layers of redundancy in the immune response to L. pneumophila coordinated by IFN-␥. While QKO/C11 macrophages did not differ significantly from QKO macrophages in terms of IFN-␥-mediated bacterial restriction, we observed a profound loss of restriction in QKO/IRG1 macrophages, beyond what we observed in primary Nos2 Ϫ/Ϫ Acod1 Ϫ/Ϫ macrophages. This result indicates that factors other than iNOS disrupted in the QKO background may play a role in limiting L. pneumophila in IFN-␥-stimulated macrophages. For example, IRGM1-deficient macrophages displayed a partial loss of IFN-␥dependent restriction of L. pneumophila (59). In agreement with the results of Pilla et al. (28), our data suggest that a role exists for CASP11 in the IFN-␥-mediated immune response to L. pneumophila, given the complete inability of IFN-␥ to restrict L. pneumophila replication in 6KO macrophages versus QKO/IRG1 macrophages (which retain CASP11). In combination with the data showing that Casp1/11 Ϫ/Ϫ and QKO/C11 BMMs retain IFN-␥-mediated bacterial restriction, this result demonstrates that the activity of CASP11 is also redundant, at least with the activities of iNOS and IRG1. A role for CASP11 may be less apparent in our experiments using immortalized macrophages, which may be impaired in cell death pathways in addition to other major physiological differences from primary cells.
In sum, our study reveals a more comprehensive picture of the factors that are required to coordinate the IFN-␥-dependent immune response to L. pneumophila. We have not determined whether all six of the genes disrupted in 6KO BMMs cells are required to fully exert IFN-␥-dependent cell-intrinsic restriction of L. pneumophila or a subset of the six that includes iNOS, IRG1, and CASP11. Nonetheless, we are encouraged that among the numerous genes transcribed in IFN-␥-stimulated macrophages, we have narrowed the field that mediate cell-intrinsic control of L. pneumophila to six candidates. While all of the gene products disrupted in the 6KO background could function directly as antimicrobial effectors, we also note the possibility that some or all may function as upstream regulators and thus affect L. pneumophila indirectly.
IFN-␥ is an essential component of the immune response to bacterial pathogens beyond L. pneumophila. Thus, the implications of this study extend beyond furthering our understanding of the immune response to L. pneumophila, an accidental pathogen of mammals that did not evolve to evade the human immune response. Our work reveals fundamental redundancy in the IFN-␥-dependent immune response to potentially pathogenic environmental microbes. Dissecting these overlapping innate immune strategies reveals the complexity and comprehensiveness of the innate immune barrier posed to novel environmental microorganisms by mammalian macrophages and IFN-␥. Further, a more detailed understanding of how IFN-␥ can mediate bacterial restriction in host cells may inform studies of how "professional" pathogens, such as M. tuberculosis, S. enterica, and L. monocytogenes, have evolved to avoid or subvert these effects of IFN-␥.

MATERIALS AND METHODS
Ethics statement. We conducted experiments in this study according to guidelines established by the Guide for the Care and Use of Laboratory Animals of the National Institutes of Health (60) under a protocol approved by the Animal Care and Use Committee at the University of California, Berkeley (AUP-2014-09-6665).
Mouse CRISPR. We generated QKO/C11 and QKO/IRG1 mice by pronuclear injection of Cas9 mRNA and guide RNAs into fertilized embryos of QKO mice as described previously (62). Founder male mice heterozygous for mutation in either Casp4 or Acod1 were backcrossed once onto the QKO background, and offspring were intercrossed to generate QKO/C11, QKO/IRG1, and 6KO mice. The Acod1 mutation was determined by amplifying a fragment of genomic DNA surrounding the cut site targeted in Acod1 exon 2 (forward primer AACTCTGGGAATGCCAGCTC and reverse primer GGAGCCACAACAGGGATCAA, yielding an ϳ440-bp PCR product) and Sanger sequencing, which revealed a three-nucleotide deletion (TTC) and a one-nucleotide insertion (A) at the cut site in mutant DNA, resulting in a frameshift mutation and premature stop codon. The Casp4 (encoding CASP11) mutation was determined by amplifying genomic DNA surrounding the cut sites indicated by both guide RNAs (forward primer GGGGCTCTGAAAAGGTGTGA and reverse primer TCTAGACACAAAGCCCATGT, revealing an ϳ520-bp band in wild-type DNA and an ϳ290-bp band in mutant DNA, indicating a missing ϳ230-bp fragment in mutant genomic DNA. iCas9 CRISPR. We cloned template DNA for the indicated guide RNAs into a pLX-sgRNA construct additionally containing blasticidin resistance (Addgene plasmid 50662). We transfected constructs into HEK293T cells along with lentivirus packaging vector pSPAX2 (Addgene plasmid 12260) and lentivirus envelope vector VSV-G (Addgene plasmid 8454). We used the resulting virus particles to transduce immortalized wild-type C57BL/6 cells that express doxycycline-inducible SpCas9 enzyme (generated using Addgene plasmid 50661). We cultured transduced cells in 3.0 g/ml blasticidin (InvivoGen) and 5.0 g/ml doxycycline (Sigma) for at least 2 weeks prior to use in experiments.
Bacterial strains, infection, and stimulation of BMMs. LP02 is a thymidine auxotroph derived from LP01, a clinical isolate of L. pneumophila (63). Generation of ΔflaA, ΔflaA ΔuhpC, ΔdotA, and luminescent strains of L. pneumophila has been described previously (40,41). We cultured all strains of L. pneumophila in AYE {ACES [N-(2-acetamido)-2-aminoethanesulfonic acid]-buffered yeast extract broth} or on ACESbuffered charcoal-yeast extract (BCYE) agar plates at 37°C. For measurement of intracellular L. pneumophila growth by luminescence or by CFU, we plated 1.0 ϫ 10 5 BMMs/well in opaque white TC-treated 96-well microtiter plates and infected them with L. pneumophila at a multiplicity of infection of 0.05. One hour postinfection by centrifugation at 287 ϫ g, we replaced the medium of infected BMMs with medium Ϯ stimulation at indicated concentrations. At the indicated times following infection, we measured bacterial growth by detection of luminescence at ϭ 470 using a SpectraMax L luminometer (Bio-Rad) or by dilution of infected cultures on BYCE agar plates for enumeration of CFU. Pam3CSK4 and E. coli-derived lipopolysaccharide (LPS) were purchased from InvivoGen. We used 2-deoxyglucose (Abcam), brefeldin A (BD), 1400W (Cayman Chemical), 3-bromopyruvate, sodium oxamate, galactose, geldanamycin, dithiothreitol, tunicamycin, thapsigargin, and ISRIB (all from Sigma) as indicated. Recombinant mouse IFN-␥ (ThermoFisher) was used at 6.0 ng/ml (60 U/ml) unless otherwise indicated. We performed lactate and glucose measurement with kits purchased from Sigma according to the manufacturer's instructions.
Phagocytosis and endosome acidification assay. We plated BMMs at a density of 1.0 ϫ 10 5 cells/well in opaque white TC-treated 96-well microtiter plates. Following pretreatment with 1.0 M cytochalasin D (Sigma) or dimethyl sulfoxide (DMSO), we exposed BMMs to pHrodo red zymosan bioparticles (ThermoFisher) according to the manufacturer's instructions. We measured fluorescence