ABSTRACT
The mechanisms by which interferon gamma (IFN-γ) controls the replication of cytosolic pathogens independent of responses, such as the generation of reactive oxygen species/reactive nitrogen species (ROS/RNS), have not been fully elucidated. In the current study, we developed a model using Francisella tularensis, the causative agent of tularemia, in which pathways triggered by IFN-γ commonly associated with bacterial control were not required. Using this model, we demonstrated that IFN-γ-mediated production of itaconate and its ability to impair host mitochondrial function, independent of activity on the pathogen, were central for the restriction of bacterial replication in vitro and in vivo. We then demonstrate that IFN-γ-driven itaconate production was dispensable, as directly targeting complex II using cell membrane-permeable metabolites also controlled infection. Together, these findings show that while reprogramming of mitochondrial metabolism is a key factor in IFN-γ control of intracellular bacteria, the development of antimicrobial strategies based on targeting host mitochondrial metabolism independent of this cytokine may be an effective therapeutic approach.
INTRODUCTION
Interferon gamma (IFN-γ) is a key cytokine in host defense against bacterial pathogens. IFN-γ signaling activates numerous antimicrobial programs, including the production of reactive oxygen species (ROS) and reactive nitrogen species (RNS), generation of antimicrobial peptides and cytokines, increase in phagolysosomal acidification, and reprograming of host metabolic pathways to support inflammatory cytokine production (1–3). Full activation of IFN-γ-associated antimicrobial programs often requires a secondary immune stimulus, such as Toll-like receptor (TLR) and/or tumor necrosis factor receptor (TNFR) activation (2). Evasion or subversion of IFN-γ-induced antimicrobial responses is a known characteristic of successful or more virulent intracellular bacterial pathogens. For example, evasion of TLR activation through surface antigen modification, upregulation of ROS/RNS scavengers, and establishment of a tolerant intracellular niche by escaping or manipulating the phagolysosome are some but not all of the documented mechanisms by which successful pathogens evade the host response (2, 4). Consequently, several prominent mechanisms induced by IFN-γ to control bacterial replication are not always sufficient at limiting infection, as has been reported with multiple pathogens (5–10). Therefore, determining alternative mechanisms by which the host constrains pathogen replication through IFN-γ-mediated signaling will assist in the development of novel antimicrobial and vaccine strategies.
Francisella tularensis is a Gram-negative bacterium and the causative agent of tularemia. As a primary virulence strategy, F. tularensis suppresses inflammatory cascades early during infection, allowing for rapid unchecked replication within the cytosolic compartment of host macrophages (11, 12). Murine models have clearly shown an important role for IFN-γ in the control of both attenuated and virulent forms of F. tularensis, though attenuated strains are significantly more sensitive to this mechanism of host defense (6, 13, 14). However, there are several examples in which more common pathways triggered by IFN-γ reported to control intracellular bacterial replication are not required for the restriction of F. tularensis infection (6, 15). Therefore, comparative studies between attenuated and virulent strains of F. tularensis infection may serve as a useful tool to identify new mechanisms by which IFN-γ controls intracellular bacterial replication.
In addition to more commonly reported mechanisms of control of intracellular pathogens, such as ROS/RNS production, recent reports have suggested that exposure of cells to IFN-γ influences host metabolic pathways to promote macrophage activation and inflammation (3, 16). Therefore, we hypothesized that IFN-γ reprogramming of host metabolism toward a hostile metabolic environment would be the primary mechanism for the control of bacterial infection. We provide evidence demonstrating that IFN-γ-stimulated reprogramming of mitochondrial metabolism and production of a specific metabolite contributed to the control of bacterial replication in vitro and in vivo. Furthermore, we demonstrate that selective targeting of mitochondrial metabolism with specific inhibitors in the absence of IFN-γ is an effective strategy to control virulent F. tularensis infection.
RESULTS
IFN-γ controls attenuated but not virulent F. tularensis infection in BMDM.Although variation in their sensitivity to IFN-γ-mediated killing has been reported for different strains of F. tularensis, a direct side-by-side comparison has not been shown. Moreover, the majority of studies focus on cells treated with IFN-γ prior to infection, which does not mimic the in vivo environment in which a fraction of cells would first be infected before the production of IFN-γ would be triggered. In our passive infection model, <10% of our cells are infected at the outset before IFN-γ is introduced (data not shown), allowing us to determine the effects of the cytokine on both infected and uninfected cells as the infection progresses. Initially, we established if attenuated live vaccine strain (LVS) and virulent SchuS4 had differential susceptibilities to IFN-γ-mediated killing following infection of bone marrow-derived macrophages (BMDM) in vitro. BMDM were infected with SchuS4 or LVS and were treated with recombinant IFN-γ 3 h postinfection, and intracellular bacterial loads were determined over 48 h. IFN-γ treatment did not control SchuS4 infection within the first 24 h (Fig. 1A). IFN-γ triggered modest control of SchuS4 replication by 48 h after infection but did not limit any cytotoxicity associated with SchuS4 infection at this time point (Fig. 1B). Conversely, IFN-γ readily controlled LVS replication within the first 24 h (Fig. 1C) and eliminated the cytotoxicity associated with infection at 48 h (Fig. 1D). IFN-γ control of LVS infection was dose dependent (see Fig. S1A in the supplemental material). These data establish the ability of virulent, but not attenuated, F. tularensis to resist IFN-γ-mediated antimicrobial responses early during infection in BMDM.
IFN-γ controls attenuated but did not cure virulent F. tularensis infection in BMDM. (A) CFU recovered from SchuS4-infected (MOI = 10) BMDM treated or not with IFN-γ (100 U/ml) at 3, 24, and 48 h. (B) Extracellular LDH activity in cell supernatants collected from SchuS4-infected (MOI = 10) or mock-infected BMDM treated or not with IFN-γ (100 U/ml) at 48 h. (C) CFU recovered from LVS-infected (MOI = 10) BMDM treated or not with IFN-γ (100 U/ml) at 3, 24, and 48 h. (D) Extracellular LDH activity in cell supernatants collected from LVS-infected (MOI = 10) or mock-infected BMDM treated or not with IFN-γ (100 U/ml) at 48 h. Data are presented as the mean ± standard error of the mean (SEM) (n = 3). ***, P < 0.001 indicates significance between untreated and IFN-γ-treated groups using a Student t test (A and C) or a 2-way ANOVA for multiple comparisons (B and D).
Common IFN-γ-mediated host responses are dispensable for the control of LVS infection.It is not clear outside in vitro infection of peritoneal macrophages what host factors are required to mediate IFN-γ-directed control of LVS replication (14, 15). Increased production of cytokines has been reported in LVS relative to SchuS4 (17), which may provide secondary signals that participate in driving antimicrobial mechanisms. More specifically, activation of MyD88 through various TLRs or cytokine signaling receptors in conjunction with IFN-γ signaling has been implicated in the control of bacterial infections through the production of ROS/RNS (18, 19). Therefore, we first assessed the contribution of MyD88 to the control of LVS infection using MyD88−/− mice. The presence of MyD88 did not significantly contribute to IFN-γ control of LVS infection (Fig. 2A). Another possibility for IFN-γ-mediated control of LVS was that LVS triggered tumor necrosis factor alpha (TNF-α) independently of MyD88-mediated signaling. TNF-α is well known to synergize with IFN-γ to drive ROS/RNS production (18, 20). While we were unable to detect TNF-α in cell supernatants from LVS-infected BMDM (data not shown), we could not rule out the possibility that TNF-α present at subdetectable levels was synergizing with IFN-γ to control LVS infection. Therefore, we also assessed the contribution of TNF-α to the control of LVS infection following IFN-γ treatment in TNF-α−/− BMDM. IFN-γ significantly controlled LVS infection in TNF-α−/− BMDM similarly to wild-type (WT) controls (Fig. 2B). In light of these findings, we hypothesized that it was possible that IFN-γ was able to effectively evoke the production of RNS or ROS independently of secondary signals. Thus, we next determined the specific contributions of RNS and ROS in the control of LVS infection using BMDM from mice deficient in glycoprotein 91 (gp91) and/or nitric oxide synthase 2 (NOS2) (gp91−/−, NOS2−/−, or gp91/NOS2−/−). Once again, IFN-γ significantly controlled LVS infection similarly to the WT in all of these strains (Fig. 2C to E). Similar results were observed when the lowest concentration of IFN-γ that controls infection was used (0.1 U/ml) (data not shown), indicating that these findings are not contingent upon dose. Thus, the generation of ROS/RNS via gp91 and NOS2 was not the primary mechanism by which IFN-γ controls LVS-infected BMDM.
Common mechanisms of bacterial control induced by IFN-γ are dispensable for control of LVS infection. (A to F) CFU recovered from LVS-infected (MOI = 10) BMDM from WT or MyD88−/− (A), TNF-α−/− (B), gp91phox−/− (C), NOS2−/− (D), gp91phox−/−/NOS2−/− (E), or IFNAR−/− (F) mice treated or not with IFN-γ (100 U/ml) at 3, 24, and 48 h. Data are presented as the mean ± SEM (n = 3). ***, P < 0.001 indicates significance between untreated and IFN-γ-treated groups. No statistical significance was observed between strains with IFN-γ treatment using 2-way ANOVA for multiple comparisons.
As an alternative to ROS/RNS production, IFN-γ may support type I IFN signaling or depletion of tryptophan as alternative mechanisms to control intracellular pathogens (18, 21). However, neither the absence of type I IFN signaling nor supplementation with tryptophan in WT BMDM impacted the ability of IFN-γ to control LVS replication (Fig. 2F and S1B). Together, these data support the notion that control of LVS infection following recombinant IFN-γ treatment must be occurring independent of many of the more commonly known IFN-γ mechanisms associated with the restriction of replication of intracellular pathogens.
IFN-γ reprograms host central metabolism with LVS but not SchuS4 infection.Given that we were unable to attribute IFN-γ-mediated control of LVS to many of the previously described mechanisms, we sought out novel possibilities to explain this phenomenon. An alternative mechanism by which IFN-γ may control infection is through reprogramming of the host metabolism to promote a hostile environment for cytosolic pathogens. We have previously established that SchuS4 increases mitochondrial function in BMDM early during infection as a central virulence mechanism, whereas LVS failed to trigger similar mitochondrial reprogramming (22). Considering the connection between bacterial virulence and mitochondrial function, we hypothesized that varied susceptibility to IFN-γ between LVS and SchuS4 may be related to differential impact on host cell metabolic processes. Therefore, we first determined if IFN-γ affected host cell metabolism and, if so, whether these changes were further modulated during infection with LVS or SchuS4. IFN-γ treatment increased basal respiration rates in BMDM; however, maximal respiration rates were decreased in a dose- and time-dependent manner concomitant with a shift toward glycolysis (Fig. S2A to F). Next, we measured changes in mitochondrial function in BMDM infected with SchuS4 or LVS with or without IFN-γ treatment. At the multiplicity of infection (MOI) used here, we have previously reported that Francisella spp. do not independently contribute to the oxygen consumption rates within 7 h after infection; therefore, this time point was selected to evaluate extracellular flux (22). Consistent with our earlier report, infection of BMDM with SchuS4 increased mitochondrial respiration (Fig. 3A), characterized by a significant increase in basal respiration and proton leak (Fig. 3C and D), maintenance of maximal respiration rates (Fig. 3E), and increased nonmitochondrial respiration (Fig. 3F). The addition of IFN-γ to SchuS4-infected cells did not significantly change any parameters of mitochondrial function relative to infected cells only (Fig. 3A and C to F). Conversely, LVS infection caused a statistically significant decrease in mitochondrial maximal respiration rates compared to uninfected and SchuS4-infected cells (Fig. 3B and E). Impairment of mitochondrial function within LVS-infected cells was further exacerbated following the addition of IFN-γ. Specifically, we observed significantly reduced basal respiration in LVS- and IFN-γ-treated cells compared to LVS-infected and uninfected controls, as well as SchuS4-infected cells treated with IFN-γ (Fig. 3C). Additionally, we observed significantly reduced maximal respiration rates compared to SchuS4-infected cells treated with IFN-γ (Fig. 3E). Together, these results indicate that LVS infection alone is driving a metabolic reprogramming mechanism that synergizes with IFN-γ to impair host mitochondrial respiration and that this is distinct from changes observed among SchuS4-infected cells. Furthermore, these data suggest that LVS is unable to effectively manipulate host mitochondrial metabolism in the presence of IFN-γ compared to SchuS4.
IFN-γ reprograms host central metabolism with LVS but not SchuS4 infection. (A) Mito stress test of SchuS4-infected BMDM treated or not with IFN-γ (100 U/ml) at 7 h. (B) Mito stress test of LVS-infected BMDM treated or not with IFN-γ (100 U/ml) at 7 h. Oxygen consumption rate (OCR) traces shown in panels A and B were from the same study but subdivided into SchuS4 or LVS infection for clarity. The OCR traces were further dissected into individual features of mitochondrial function, including basal respiration (C), proton leak (D), maximal respiration (E), and nonmitochondrial respiration (F). (G) Glycolysis, corresponding to the extracellular acidification rate (ECAR) of SchuS4- and LVS-infected BMDM treated or not with IFN-γ (100 U/ml) at 7 h. The OCR traces are representative of 3 independent experiments. The results for individual features including basal respiration, proton leak, maximal respiration, nonmitochondrial respiration, and glycolysis are shown as the mean ± SEM (n = 3 to 6). *, P < 0.05; **, P < 0.01; ***, P < 0.001 indicate significance using a 2-way ANOVA for multiple comparisons. Oligo, oligomycin; Rot/Ant, rotenone/antimycin.
The ATP demands of the cell following a loss of mitochondrial respiration can be met by increasing glycolysis. Increased glycolysis is also required for optimal host cell activation of inflammatory responses (3, 23). Therefore, we hypothesized that a loss of mitochondrial respiration following LVS infection would result in a shift to glycolysis. We also speculated that infection with SchuS4 would not trigger an increase in glycolysis and that cells infected with this bacterium would be refractory to IFN-γ-driven glycolysis. Consistent with this hypothesis, LVS infection significantly increased glycolysis compared to that in the untreated controls (Fig. 3G). IFN-γ treatment did not further enhance acidification rates in LVS-infected cells. Conversely, SchuS4 infection did not increase glycolysis over that in mock-infected groups (Fig. 3G). Furthermore, SchuS4- but not LVS-infected cells showed a statistically significant impairment in the upregulation of glycolysis triggered by IFN-γ (Fig. 3G). These data clearly support our central hypothesis that attenuated LVS differentially modulates host cell metabolism compared to SchuS4. Moreover, LVS-triggered reprogramming of host cell metabolism in the presence of IFN-γ is consistent with the evolution of an inflammatory response. Finally, SchuS4, but not LVS, infection impaired IFN-γ-associated reprogramming of mitochondrial function and glycolysis.
LVS infection decreases ETC complex II respiration.Central to mitochondrial respiration is the activity of four distinct electron transport chain (ETC) complexes. Electrons can be channeled through the electron transport chain via two distinct routes involving initiation at either complex I or complex II. Impairment of either one of these routes or a combination of the two might be responsible for the decreased respiration seen in LVS infection or after IFN-γ treatment (24, 25). To assess the relative contributions of these pathways, we measured mitochondrial ETC complex I and II activity by permeabilizing the plasma membrane and assessing respiratory flux in the presence of selective complex I or II substrates (22). Treatment with IFN-γ and infection with SchuS4 in the absence of IFN-γ resulted in increased ATP-driven or state 3 respiration rates through complex I relative to the mock controls (Fig. 4A). Conversely, LVS infection did not increase complex I state 3 respiration rates. Treatment of SchuS4-infected cells with IFN-γ reversed the elevated state 3 respiration rates back to levels observed in uninfected controls but had an insignificant impact on state 3 respiration rates in LVS-infected cells (Fig. 4A). Together, these data support minimal inhibitory effects of IFN-γ on complex I activity during infection with F. tularensis. No significant effects of IFN-γ treatment or SchuS4 infection (in the absence of IFN-γ) on complex II activity were observed. However, LVS infection significantly reduced complex II activity relative to uninfected and SchuS4-infected cells (Fig. 4B). A general and significant reduction in complex II respiration rates was observed in both SchuS4- and LVS-infected cells treated with IFN-γ; however, respiration rates were significantly lower in LVS-infected cells treated with IFN-γ than in SchuS4-infected cells treated with IFN-γ (Fig. 4B). In addition to a loss of complex II activity, we observed an overall increase of the NADH/NAD+ ratio in LVS-infected cells (Fig. S3), consistent with significant reprogramming of the tricarboxylic acid (TCA) cycle and upregulation of glycolysis. Altogether, these data indicate that decreased complex II activity was the primary component that drove reduced respiration following LVS infection and IFN-γ treatment.
LVS infection decreases ETC complex II respiration. (A) State 3 respiration of complex I in BMDM infected with SchuS4 or LVS treated or not with IFN-γ (100 U/ml) at 7 h. (B) State 3 respiration of complex II in BMDM infected with SchuS4 or LVS treated or not with IFN-γ (100 U/ml) at 7 h. Data are representative of 3 independent experiments. Error bars represent the SEM for 6 to 8 technical replicates. *, P < 0.05; **, P < 0.01; ***, P < 0.001 indicate significance using a 2-way ANOVA for multiple comparisons.
LVS infection increases itaconate levels in BMDM, which contributes to the control of infection.Once we established that perturbation of mitochondrial function among LVS-infected cells was associated with an impairment of complex II activity, we next determined the mechanism by which reduced mitochondrial respiration occurs. Generation of RNS and, to a lesser extent, ROS, generated via NOS2 and NADPH oxidase 2 (NOX2), respectively, may drive mitochondrial dysfunction by binding directly to iron-sulfur clusters or sulfhydryl groups of mitochondrial components (26–28). However, NOS2 and NOX2 did not significantly contribute to IFN-γ-triggered control of LVS (Fig. 2C to E). Furthermore, we observed reduced maximal respiration rates in gp91−/−, NOS2−/−, and gp91/NOS2−/− BMDM following IFN-γ treatment, suggesting that the influence of IFN-γ on complex II function was independent of these components (Fig. S4A to C). Therefore, there are potentially alternative mechanisms by which IFN-γ may reprogram mitochondria to contribute to the control of LVS replication. It has been reported that IFN-γ promotes increased cis-aconitate decarboxylase (ACOD1, also known as interferon response gene-1) activity (29). Enzymatic activity of ACOD1 results in the production of itaconate from the TCA cycle intermediate cis-aconitate, which in turn may bind the ETC complex II, thereby inhibiting its activity (30). Therefore, we first investigated the possibility that LVS and/or IFN-γ treatment may promote the production of itaconate in BMDM. Infection with LVS significantly increased ACOD1 transcription above mock levels, and transcription was synergistically enhanced with IFN-γ treatment (Fig. 5A). Consistent with upregulation of ACOD1 transcription, both LVS and IFN-γ treatment significantly increased intracellular itaconate levels above those of mock-infected/treated cells, and LVS infection with IFN-γ treatment acted synergistically to increase itaconate in BMDM (Fig. 5B). Conversely, SchuS4 did not significantly increase itaconate levels above those of mock-infected cells. While SchuS4 did not impair IFN-γ-mediated induction of itaconate, we did not observe synergistic production of this metabolite as observed in LVS-infected and IFN-γ-treated cultures (Fig. 5B).
LVS infection increases itaconate levels in BMDM, which contributes to the control of infection. (A) ACOD1 mRNA levels as indicated by the fold change in BMDM whole-cell lysates 24 h following infection with LVS (MOI = 50) with or without IFN-γ treatment (100 U/ml). (B) Itaconate levels in BMDM whole-cell lysates 24 h following infection with LVS or SchuS4 (MOI = 50) with or without IFN-γ treatment (100 U/ml). (C) State 3 respiration rates for complexes I and II in permeabilized BMDM treated with increasing doses of itaconate. (D) Percent change in maximal respiration rates relative to uninfected controls in BMDM from WT or ACOD1−/− mice 7 h following infection with LVS (MOI = 50) with or without IFN-γ treatment (100 U/ml). (E) LVS grown in MMH broth over 24 h with exposure to increasing doses of itaconate. The starting inoculation was 105 CFU. (F and G) CFU recovered from LVS-infected (MOI = 10) BMDM from WT or ACOD1−/− mice treated with or without IFN-γ at 0.1 U/ml (F) or 100 U/ml (G) at 3, 24, and 48 h. (A, B, and D to G) Data are shown as the mean ± SEM (n = 3), which is a representative experiment of 3 separate repeats with 6 to 8 technical replicates each. (A to D) *, P < 0.05; **, P < 0.01; ***, P < 0.001 indicate significance using a 1- or 2-way ANOVA for multiple comparisons. (F and G) *, P < 0.05; **, P < 0.01 indicate significance between WT and ACOD1−/− groups treated with IFN-γ using a 2-way ANOVA for multiple comparisons.
Next, we confirmed whether itaconate could be responsible for impaired complex II activity in BMDM. There was a dose-dependent decrease in complex II, but not complex I, state 3 respiration rates following the treatment of permeabilized BMDM with itaconate (Fig. 5C). Next, we performed extracellular flux analysis on LVS-infected C57BL/6NJ or ACOD1−/− BMDM, which lack the ability to produce itaconate, to determine if itaconate promoted mitochondrial dysfunction during infection. LVS infection alone and LVS infection with IFN-γ treatment caused a significant reduction in maximal respiration rates in WT but not ACOD1−/− BMDM (Fig. 5D). Together, these data support the notion that itaconate inhibition of mitochondrial complex II contributes to decreased mitochondrial respiration following LVS infection in vitro.
Itaconate has been reported to inhibit isocitrate lyase in bacterial metabolism as a direct antimicrobial mechanism against several pathogens (31–33). However, F. tularensis does not contain the gene for isocitrate lyase, making direct itaconate activity against F. tularensis unlikely (34). Consistent with this notion, itaconate added to LVS cultured in modified Mueller-Hinton (MMH) broth during early log phase did not inhibit replication (Fig. 5E). Given that itaconate did not have direct killing activity on LVS, we hypothesized that the control of LVS replication by itaconate and, by extension, IFN-γ was a consequence of inhibited complex II activity to promote an antimicrobial environment. Therefore, we next determined if the production of itaconate within host macrophages contributed to IFN-γ-mediated control of LVS replication. In the presence of 0.1 U/ml IFN-γ, bacterial loads were greater in ACOD1−/− BMDM than in the WT over 48 h (Fig. 5F). Similar results were observed within the first 24 h following infection when cells were treated with 100 U/ml IFN-γ; however, control of infection was lost by 48 h (Fig. 5G).
To determine whether itaconate production through IFN-γ was an antimicrobial mechanism that could be applicable to other intracellular pathogens, we also infected WT and ACOD1−/− mice with Listeria monocytogenes. We observed a modest but significant contribution of ACOD1 to IFN-γ control of L. monocytogenes (Fig. S5). These data support a more general role for IFN-γ-driven itaconate production in the control of intracellular pathogens in vitro.
Itaconate is critical for the control of LVS in vivo.The data presented above suggest a partial role for IFN-γ-induced itaconate in the control of LVS replication in vitro. However, the in vivo compartment is a more dynamic system, and our in vitro findings may be accentuated following in vivo infection. Therefore, we next assessed the contribution of itaconate to the control of LVS following intranasal infection with LVS in WT and ACOD1−/− mice. We first compared the ability of mice to survive either a sublethal (102 CFU/mouse) or lethal (104 CFU/mouse) dose of LVS. All ACOD1−/− and WT mice survived infection with 102 CFU (Fig. 6A). However, all ACOD1−/− mice infected with 104 CFU succumbed to infection, whereas approximately 10% of WT mice survived (Fig. 6A). Furthermore, ACOD1−/− mice had a statistically significant decreased mean time to death relative to the WT (−1.949 ± 0.3551 days; *, P < 0.0001). We suspected that ACOD1−/− mice succumbed earlier to infection due to their inability to control bacterial replication. ACOD1−/− mice died within 4 to 5 days following infection with 104 CFU, which would not allow for a comprehensive examination of the control (or lack thereof) of LVS replication and dissemination over time. Therefore, we infected mice with a sublethal dose of LVS (102 CFU) and assessed bacterial burden within the lungs, spleen, and liver over the course of 3 weeks. LVS bacterial loads were significantly greater in ACOD1−/− mice in the lungs at day 14 and in the spleen and liver at days 7 and 14 than in the WT controls (Fig. 6B to D). By day 21 after infection, all mice had cleared bacteria from the liver and spleen. In contrast, a residual number of bacteria were still present in the ACOD1−/− lungs at this time point (Fig. 6B). Increased bacterial burdens among ACOD1−/− mice on days 7 and 14 after infection were associated with greater production of inflammatory cytokines/chemokines (Fig. 6E to M and S6). Interleukin 4 (IL-4) and IL-10 were not present at detectable levels, consistent with an unconstrained inflammatory response in the ACOD1−/− mice (data not shown).
Itaconate is critical for the control of LVS in vivo. (A) Survival of WT and ACOD1−/− mice following infection with 102 or 104 CFU delivered via intranasal instillation. (B to D) Recovered CFU from the lungs (B), spleens (C), or livers (D) of WT and ACOD1−/− mice over time following infection with 102 CFU delivered via intranasal instillation. (E to M) Cytokine or chemokine IL-12p40 (E to G), MCP-1 (H to J), or IL-1α (K to M) concentration in whole lung, spleen, or liver lysates shown in panels B to D. Data are shown as the mean ± SEM (n = 3 to 4 mice/experiment) pooled from two separate experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001 indicate significance using a Student t test.
In addition to bacterial load and inflammatory cytokine/chemokine production, we assessed pathological changes associated with LVS infection in ACOD1−/− mice in the lungs, liver, and spleen compared to the WT over time. Neither WT nor ACOD1−/− mice demonstrated pulmonary, hepatic, or splenic changes in the absence of infection. Minimal differences in disease pathology were observed between WT and ACOD1−/− mice in the liver over the course of the infection or between the lung and spleen at days 4 and 7 postinfection (Fig. S7A to C). In contrast, at day 14 postinfection, ACOD1−/− mice began to display marked to severe pulmonary lesions and extensive edema, while WT mice lesions ranged from minimal to marked (Fig. S7B). The spleens of ACOD1−/− mice also showed distinct differences in histopathology at day 14 postinfection as demarcated by reticuloendothelial hyperplasia, as well as significant extramedullary hematopoiesis compared to the essentially normal presentation of splenic architecture observed in WT animals at this time point (Fig. S7C). By 21 days after infection, there were no significant differences in the lungs, liver, or spleen of WT and ACOD1−/− mice, consistent with control of bacterial infection (Fig. S7A to C). In summary, following sublethal infection, ACOD1−/− mice exhibit greater transient inflammation and pathology associated with increased bacterial burden, confirming a role for itaconate in disease pathogenesis in vivo with LVS infection.
Complex II impairment controls SchuS4 infection.We have established that one mechanism by which IFN-γ contributes to the control of LVS replication is via the production of itaconate, which negatively modulates complex II within the mitochondria. Therefore, we hypothesized that IFN-γ would not be required for the control of infection if complex II was directly targeted using cell-permeable itaconate. Moreover, this would suggest that other cell-permeable metabolites that inhibit complex II, such as dimethyl malonate (DMM), would be effective at controlling F. tularensis. To test this hypothesis, we added 4-octyl-itaconate (4OI) or DMM and assessed their ability to control intracellular replication of SchuS4 in primary mouse and human macrophages. The doses of 4OI and DMM used in these studies were selected based on minimal toxicity observed in macrophages after incubation for 48 h, as measured by the lactate dehydrogenase (LDH) assay (Fig. S8). We also initially confirmed that these agents did not directly act on bacterial replication when added to SchuS4 cultured in broth. No significant differences were evident between 4OI-, DMM-, and vehicle-treated SchuS4 broth cultures after 24 h when drugs were added during early log phase (Fig. S9A and B). Next, we assessed if the addition of 4OI or DMM impacted SchuS4 replication in mouse and human macrophages. 4OI treatment of SchuS4-infected mouse BMDM resulted in statistically significant control of bacterial replication within the first 24 h of infection in a dose-dependent manner (Fig. 7A). However, DMM controlled intracellular bacterial replication to a greater extent than did 4OI in mouse BMDM, which was maintained up to 48 h (Fig. 7B). In human macrophages, we observed the opposite pattern, in which 4OI controlled infection to a greater extent than did DMM (Fig. 7C and D), and the doses of DMM required to achieve control were at least 10-fold higher than those required for control in mouse macrophages.
Addition of cell-permeable itaconate to host cells controls SchuS4 infection. (A and B) CFU recovered over 48 h from SchuS4-infected (MOI = 10) BMDM that were treated with increasing doses of 4OI (A) or DMM (B). (C and D) CFU recovered over 48 h from SchuS4-infected (MOI = 10) primary human macrophages that were treated with increasing doses of 4OI (C) or DMM (D). (E to G) CFU recovered from the lung, spleen, and liver of vehicle- or DMM-treated mice (250 μg/kg) at days 2 and 4 postinfection. Data are shown as the mean ± SEM (n = 3) for all in vitro studies. For in vivo studies, data are shown as the mean ± SEM (n = 4 to 5 mice/experiment) pooled from two separate experiments. *, P < 0.05; **, P < 0.01; ***, P < 0.001 indicate significance of treated groups relative to vehicle (A to C) using a one-way ANOVA for multiple comparisons. *, P < 0.05 indicates significance relative to vehicle control (E to G) using a Student t test.
It has been reported that 4OI treatment does not lead to an increase in intracellular itaconate in mouse macrophages unless a secondary stimulus, such as lipopolysaccharide (LPS), is present (35). Therefore, we speculated that the differential control of SchuS4 between mouse and human macrophages in response to 4OI treatment was due to species variation in the upregulation of intracellular itaconate. Treatment with 4OI induced a significant increase in intracellular itaconate in both mouse and human macrophages (Fig. S9C and D). However, human macrophages displayed a faster and greater increase in intracellular itaconate following 4OI treatment, consistent with superior control of SchuS4 in these cells. Furthermore, 4OI treatment reduced mitochondrial maximal respiration rates in both treated human and mouse macrophages, consistent with impairment of complex II activity due to increased intracellular itaconate (Fig. S9E and F). Collectively, these data demonstrate that the induction of metabolites via IFN-γ is not required for control of intracellular replication, and direct supplementation of metabolites that drive inhibition of mitochondrial complex II can restrict intracellular replication of bacteria in vitro.
We next assessed if complex II inhibitors could control infection in vivo. For these studies, we treated mice with DMM, rather than 4OI, based on its superiority in controlling infection in mouse macrophages in vitro (Fig. 7B). Daily treatment with DMM resulted in significantly reduced bacterial burdens in liver and spleen at day 2 postinfection, suggesting a delay in dissemination (Fig. 7E to G). However, by day 4, we observed no significant differences in bacterial burden in these tissues, indicating that the therapeutic benefits of DMM may be restricted to early pathogenesis mechanisms or may have been overcome upon systemic dissemination and establishment of a fulminant infection. Additionally, we observed an unexpected outcome of adipose tissue toxicity that was restricted to the epididymal fat pads of DMM-treated mice (data not shown), potentially contributing to the control, or lack thereof, of bacterial burdens. Nevertheless, these studies support the central notion that treatment with complex II inhibitors can control bacterial infection in vivo and support further development of therapeutic strategies using these inhibitors.
DISCUSSION
IFN-γ activates numerous pathways that can facilitate the detection and control of intracellular microbial pathogens (18). However, well-described antimicrobial pathways elicited by IFN-γ are not always sufficient at controlling highly virulent pathogens. In support of this notion, we and others confirm that ROS/RNS production via NOX2 and inducible nitric oxide synthase (iNOS), respectively, are dispensable or offer only a minor contribution for early control of F. tularensis and may only partially contribute to IFN-γ control of a variety of other intracellular microbes (5, 15, 36–38). Furthermore, iron restriction, tryptophan depletion, and type I IFN responses, which have been attributed to participate in IFN-γ control of multiple pathogens, are not significant contributors to the control of F. tularensis infection (6, 15). Many IFN-γ-mediated responses target pathogens within the host phagosomal compartment; however, IFN-γ is reported to not limit phagosomal escape of F. tularensis (6). These features make F. tularensis a model bacterial pathogen for elucidating novel mechanisms induced by IFN-γ that restrict bacterial replication within the cytosolic space. In the current study, we demonstrate that metabolic reprogramming of host macrophages by IFN-γ is a mechanism to restrict the cytosolic replication of F. tularensis. Furthermore, we made the surprising finding that upon identification of appropriate modulation of mitochondrial function, IFN-γ was dispensable for the control of intracellular bacterial replication.
IFN-γ signaling has been known for some time to facilitate a reduction in mitochondrial respiration during infection, primarily through NOS2 and generated RNS, which can bind to iron sulfur clusters within the ETC (16, 26, 27). However, RNS generation is typically licensed through activation of MyD88 signaling (18). Our data support a MyD88- and associated ROS/RNS-independent mechanism for IFN-γ to reprogram mitochondrial metabolism. In agreement with these findings, recent reports indicate that IFN-γ on its own can facilitate metabolic reprogramming that is not dependent upon RNS and is linked to the upregulation of glycolysis and stabilization of hypoxia-inducible factor 1 alpha (HIF-1α) (3, 23). We also observed upregulation of glycolysis independent of RNS production (data not shown). However, we extended these findings by demonstrating an additional role for IFN-γ in driving itaconate production during infection as a mechanism to inhibit mitochondrial respiration via complex II. Therefore, it is feasible that IFN-γ signaling modulation of core metabolic processes during infection promotes both inflammation and the creation of a metabolic state adverse to cytosolic pathogens.
Impaired ETC complex II activity has been implicated in the repurposing of the mitochondria for biosynthetic processes and as a signaling platform to regulate inflammation (39). Our results strongly indicate that IFN-γ and LVS infection drive itaconate production and associated complex II inhibition, which facilitated a metabolic and inflammatory state hostile to F. tularensis. Conversely, SchuS4 limits the production of itaconate through a process that is not fully understood but is likely multifactorial. Differential modulation of mitochondrial metabolism by SchuS4 capsule and lipids compared to LVS is a potential mechanism, as well as variation in the ability of lipids between strains to suppress IFN-γ-associated transcriptional regulators of ACOD1 (22, 40–42). Regardless, once itaconate is produced, it does not control bacteria by exhibiting direct activity on the bacterial glyoxylate pathway, as has been reported with other pathogens (31–33). These conclusions are supported by the absence of isocitrate lyase in the genome of F. tularensis and the inability of itaconate to control bacterial replication in broth (34). A similar response was reported in which itaconate contributed to the control of Mycobacterium tuberculosis infection in mutants engineered to lack the isocitrate lyase gene (43). The specific mechanism by which reduced complex II activity results in the control of intracellular bacterial replication has not been determined. There are several possibilities by which the modulation of complex II could impair bacterial replication, including mitochondrial ROS and stabilization of HIF-1α to drive proinflammatory cytokine production (3, 23, 44, 45). In the current study, our observation of decreased complex II activity, but minimal alteration in complex I activity, during LVS infection, as well as an increased NADH/NAD+ ratio, supports a potential role for mitochondrial ROS that can accompany TCA cycle reprogramming (46). Alternatively, inhibition of mitochondrial respiration may allow for buildup of other TCA cycle metabolites with immunomodulatory activity and/or direct antimicrobial activity, such as citrate or succinate (47). Furthermore, itaconate-associated inhibition of complex II may deprive F. tularensis of critical substrates required for growth, such as glycerol, the utilization of which by the bacterium has been linked to fatty acid oxidation (48).
In addition to our observation that IFN-γ-driven itaconate production contributes to the control of bacterial replication in vitro, to our knowledge, the studies presented herein are the first to demonstrate a role for itaconate in the pathogenesis of F. tularensis in vivo. Loss of itaconate was associated with greater bacterial burden, inflammation, and tissue damage in infected tissues than in the WT controls. Of particular interest was the extensive acute necrotizing pneumonia and edema observed in the lung compartment at day 14 postinfection in ACOD1−/− mice compared to WT mice. This suggests that there is a contribution of itaconate not only to the control of bacterial replication but in the regulation of inflammation-dependent tissue damage. The lack of immunosuppressive cytokines IL-4 and IL-10 in infected tissues further underscores the unconstrained nature of the inflammatory response in ACOD1−/− mice, likely leading to greater tissue damage. Itaconate has been proposed to play important roles in both upregulation and resolution of inflammation following LPS exposure (35). We observed significantly higher concentrations of a wide variety of proinflammatory cytokines in ACOD1−/− mouse lungs at day 14 postinfection with LVS, suggesting dysregulation of the inflammatory response. Similar results were observed in an M. tuberculosis mouse model of infection, including higher bacterial loads and cytokine production and increased pathology (43). While these similarities were evident, the distinguishing feature of our model is that it provides clear evidence that itaconate manipulation of host cell metabolism, and not that of the bacterium, provides a similar in vivo outcome.
Our in vivo findings further underscore the power of using metabolites to target host mitochondria as a viable strategy to combat infection, including those mediated by microbes in which the metabolites have no known bacterial targets. Specifically, we observed that complex II was a central target for therapeutic intervention with cell-permeable metabolites. Importantly, IFN-γ was not required, and mechanistically general modulators of complex II were sufficient, for the control of SchuS4. For example, both 4OI and DMM contributed to restriction of SchuS4 replication although with various degrees of effectiveness in mouse and human macrophages. Differences in responses to 4OI or DMM between species may be indicative of variations in hydrolysis rates of these compounds by esterases to their active metabolites and/or variation in complex II regulatory networks. Furthermore, in addition to direct inhibition of complex II, DMM can influence malonate-related fatty acid (FA) synthesis/oxidation (49), which may be more integral to mouse macrophage function. Further studies characterizing mouse and human primary macrophage regulatory elements of complex II and dependence on specific metabolic networks will be informative in selecting treatments that appropriately modulate mitochondrial function to control infection.
In summary, the findings from this study define a novel mechanism by which IFN-γ contributes to the control of intracellular pathogens through impacting mitochondrial function. This may be broadly applicable to multiple infections mediated by different pathogens. For example, there is evidence that IFN-γ control of infection mediated by Legionella, Brucella, Chlamydia, Mycobacterium, and Rickettsia spp. may occur independent of ROS/RNS or other prominent mechanisms (8, 33, 36–38, 50), and itaconate has also been implicated in the control of some of these organisms (33, 43). However, the activity of this metabolite on host cell function was not specifically identified. It is tempting to speculate that itaconate and potentially other metabolites induced by IFN-γ may drive mitochondrial reprogramming resulting in the control of these highly virulent pathogens. Identification of these metabolites and their ability to modulate mitochondrial function open a door for the development of novel antimicrobial therapies in which IFN-γ may be dispensable. This concept of direct targeting of complex II using cell-permeable metabolites in the absence of IFN-γ may have important implications for combating infection in individuals with defects in their ability to produce or respond to IFN-γ.
MATERIALS AND METHODS
Mice.C57BL/6J, C57BL/6NJ, ACOD1−/−, TNF-α−/−, gp91phox−/−, and NOS2−/− mice were purchased from Jackson Laboratories (Bar Harbor, ME). gp91/NOS2−/−, MyD88−/−, and IFNAR−/− mice were obtained from colonies maintained in-house at Rocky Mountain Laboratories. All animals were used between 6 and 8 weeks of age and were housed in an animal biosafety level 2 (BSL-2) facility at Rocky Mountain Laboratories. Food and water were provided ad libitum, and all protocols involving animals were done in accordance with the Animal Care and Use Committee guidelines. All guidelines were approved by the Animal Care and Use Committee at Rocky Mountain Laboratories.
Bacteria.Francisella tularensis subsp. tularensis (SchuS4) was provided by Jeannine Peterson (Centers for Disease Control and Prevention, Fort Collins, CO). Francisella tularensis subsp. holarctica LVS was provided by Karen Elkins (Food and Drug Administration, Silver Spring, MD). Listeria monocytogenes was provided by Megan Smithey (University of Arizona, Tucson, AZ). All bacteria were prepared for experiments from frozen stocks, as previously described (51). All experiments involving select agents were performed under approved biosafety protocols under BSL-3 containment at the Rocky Mountain Laboratories.
Bone marrow-derived macrophage generation.BMDM were generated as previously described (17). Briefly, mouse femurs were flushed and cells resuspended in complete Dulbecco’s modified Eagle medium (cDMEM) supplemented with glutamine, HEPES, nonessential amino acids (Thermo Fisher Scientific, Carlsbad, CA), and 10% fetal bovine serum (Atlas Biologicals, Fort Collins, CO). Cultures were supplemented with 20 ng/ml recombinant murine macrophage colony-stimulating factor (rM-CSF) (Peprotech, Rocky Hill, NJ) in a T-75 cm2 flask at 37°C with 5% CO2. Forty-eight hours later, nonadherent cells were collected, pelleted, and resuspended in fresh cDMEM plus rM-CSF. Forty-eight hours later, the culture medium was exchanged with fresh cDMEM plus rM-CSF. Twenty-four hours later, cells were collected by scraping and were pelleted and plated into 6-well plates (8 × 105 cells/well in 2 ml cDMEM plus rM-CSF) for analysis of metabolites, 48-well tissue culture plates (1 × 105 cells/well in 0.5 ml cDMEM plus rM-CSF) for infections, or 96-well Seahorse bioanalyzer plates (8 × 104/well in 80 μl in cDMEM plus rM-CSF) (Agilent Technologies, Santa Clara, CA) for analysis of mitochondrial bioenergetics. All cells were used 24 h after plating.
Generation of monocyte-derived human macrophages.Human macrophages were prepared from apheresis-prepared monocytes, as described previously (52). Human monocytes, enriched by apheresis, were obtained from peripheral blood samples provided by the Department for Transfusion Medicine and the NIH Clinical Center (Bethesda, MD) under a protocol approved by the NIH Clinical Center institutional review board. Signed informed consent was obtained from each donor, acknowledging that the donation would be used for research purposes by intramural investigators at the NIH. Following culturing with granulocyte M-CSF (GM-CSF; Peprotech), cells were removed by 0.25% trypsin (Thermo Fisher) and seeded in 48-well plates at 6 × 104 cells/well for infection assays, 6-well plates at 4.5 × 105 cells/well for determination of metabolites, or a seahorse assay plate at 2 × 104 cells/well for assessment of mitochondrial bioenergetics. All cells were used within 24 h of plating.
In vitro infection.Macrophages were infected with SchuS4 or LVS, as previously described (17). Briefly, medium was removed from the BMDM and reserved. F. tularensis was added at a multiplicity of infection (MOI) of 10 in 200 μl cDMEM (or RPMI for human cells) per well for 48-well plates or 500 μl cDMEM/well for 6-well plates. L. monocytogenes was added at an MOI of 0.001 in 200 μl cDMEM/well for 48-well plates. Mock-infected cells were incubated with cDMEM. Cells were incubated for 90 min at 37°C with 5% CO2, bacteria were removed, and cDMEM was supplemented with 50 μg/ml gentamicin (Thermo Fisher) added to each well at 0.5 ml/well for 48-well plates or 2 ml/well for 6-well plates. Cells were incubated for 45 min at 37°C with 5% CO2, and then the medium was removed, wells were washed three times with phosphate-buffered saline (PBS), and reserved cDMEM was added back to the wells. Cells were lysed via removal of medium and the addition of 200 μl/well sterile water for enumeration of intracellular bacterial loads at the time points indicated in the figures. Lysates were serially diluted in PBS, plated on MMH agar, and incubated at 37°C with 5% CO2 for 24 to 48 h for the enumeration of colonies. Where indicated in each figure, BMDM were treated immediately following infection with IFN-γ (Peprotech). Also, where indicated, dimethyl-malonate (Millipore-Sigma) or 4-octyl-itaconate (Cayman Chemical) was added immediately after infection and again at 24 h following infection. Cell death was measured at 48 h following infection using the commercially available CytoTox 96 nonradioactive cytotoxicity assay (Promega), according to the manufacturer’s instructions.
Extracellular flux assay by whole-cell analysis.BMDM were infected with SchuS4 or LVS at an MOI of 50 for 7 h for analysis of extracellular flux. Recombinant IFN-γ (100 U/ml; Peprotech) was added immediately following the addition of the bacteria to the wells. For assessment of 4OI or DMM inhibition of mitochondrial function, mouse and human macrophages were treated for 2 h prior to the assay. One hour prior to analysis, medium was removed, leaving 40 μl of residual volume at which the cell cultures were prepared to undergo either the Mito stress test or the glycolysis stress test. For the Mito stress test, BMDM were washed 2× with 200 μl of Seahorse Mito stress assay medium (minimal DMEM with 25 mM glucose, 2 mM sodium pyruvate, and 2 mM l-glutamine; Agilent Technologies). Next, 140 μl of Mito stress assay medium was added to each well to make the final well volume 180 μl. Cells were then incubated for 1 h at 37°C in a non-CO2 incubator. Mitochondrial function was then assessed on a Seahorse XFe96 bioanalyzer (Agilent Technologies) in real time following injection of oligomycin (1.5 μM; Millipore-Sigma), carbonyl cyanide-p-trifluoromethoxyphenylhydrazone (FCCP; 2.0 μM; Cayman Chemical), antimycin (0.5 μM; Millipore-Sigma), and rotenone (0.5 μM; Millipore-Sigma), following the standard manufacturer’s instructions. Specific mitochondrial parameters were calculated as previously described (22). Briefly, basal respiration is the basal oxygen consumption rate (OCR) minus the nonmitochondrial OCR, proton leak is the OCR following injection of oligomycin minus nonmitochondrial OCR, maximal respiration is OCR following FCCP injection minus nonmitochondrial OCR, and nonmitochondrial OCR is OCR following injection of rotenone/antimycin. For the glycolysis stress test, BMDM were washed 2× with 200 μl of Seahorse glycolysis stress assay medium (minimal DMEM with 1 mM l-glutamine; Thermo Fisher Scientific). Next, 140 μl of glycolysis stress assay medium was added to each well to make the final well volume 180 μl. Cells were then incubated for 1 h at 37°C in a non-CO2 incubator. Glycolysis was then assessed on the Seahorse XFe96 bioanalyzer in real time following injection of glucose (10 mM), oligomycin (1.5 μM), and 2-deoxy-glucose (2-DG; 50 mM; Millipore-Sigma) following the standard manufacturer’s instructions. Glycolysis was calculated as the extracellular acidification rate (ECAR) following the injection of glucose. Normalization of seahorse data was performed by calculating the percent change of treated/infected groups relative to the maximal respiration or basal glycolytic rate of untreated cells, respectively.
Extracellular flux assay in permeabilized cells.For direct measurement of mitochondrial function, BMDM were infected with SchuS4 or LVS at an MOI of 50 and treated with IFN-γ as described above, but prior to analysis on the Seahorse XFe96 bioanalyzer, cells were permeabilized as previously described (22). Briefly, immediately prior to the assay, cDMEM was removed except for 40 μl of residual volume, and cells were washed once with 300 μl of 1× MAS buffer (220 mM mannitol, 70 mM sucrose, 10 mM KH2PO4, 5 mM MgCl2, 2 mM HEPES, 1 mM EGTA [pH 7.2] plus 0.2% [wt/vol] fatty acid-free bovine serum albumin [BSA]). MAS buffer (1×) containing 1 nM plasma membrane permeabilizer (PMP) reagent (Agilent Technologies) and 4 mM ADP (Millipore-Sigma), and complex I- or complex II-specific substrates were added to the washed wells to achieve a final volume of 180 μl. For complex I substrate utilization, 1× MAS buffer containing glutamate/malate (10 mM/10 mM) was used. For complex II substrate utilization, 1× MAS buffer containing succinate/rotenone (10 mM/2 μM) was used. Plates were immediately placed on the Seahorse XFe96 bioanalyzer, and mitochondrial function assessed using mix/wait/measure times of 0.5 min/0.5 min/2 min. No non-CO2 incubation period or instrument equilibration steps were done prior to the start of the assay to minimize the time the permeabilized cells were in nonionic medium. Three basal (state 3) measurements of OCR were obtained, and then oligomycin was injected (1.5 μM) to induce state 4o. Three consecutive measurements of state 4o OCR were obtained.
Liquid chromatography-mass spectrometry analysis of metabolites.Mouse macrophages were infected with LVS or SchuS4 and treated with IFN-γ (100 U/ml), as described above. Twenty-four hours later, medium was removed, and each well was washed with 4 ml of 0.9% NaCl. Wells were then flooded with 0.4 ml of ice-cold methanol for 1 min. Subsequently, 0.4 ml of ice-cold water was added to each well, and the cells were scraped. The cell debris suspension from each well was collected, and 0.4 ml of ice-cold chloroform was added. Samples were shaken vigorously at 4°C for 20 min and centrifuged at 14,000 × g for 20 min. The upper aqueous layer was collected and concentrated approximately 10× via SpeedVac evaporation. Samples were analyzed using an Acclaim organic acid column (5 μm, 4 mm by 150 mm) and an Agilent 1200 series liquid chromatograph (LC) monitored by a diode array detector at 210 nm. Samples were loaded on the column with 98% 50 mM phosphate (pH 2.68)–2% acetonitrile and eluted with an 8-min gradient to 40% acetonitrile. The itaconate peak was identified by comparison to a standard and by mass spectrometry using an Agilent 6100 single quadrupole instrument. Peak area and chromatograms were normalized to the sum of all peak integrals within each sample. For analysis of itaconate following 4OI treatment, both human and mouse macrophages were treated with increasing doses of 4OI, and at 0, 2, and 5 h posttreatment, medium was removed, cells were washed with 0.9% NaCl, and isolation of metabolites was performed as described above.
In vivo infections with LVS.C57BL/6NJ (WT) or ACOD1−/− mice were intranasally infected with either a sublethal (102 CFU/25 μl saline) or lethal (104 CFU/25 μl saline) dose of LVS. Actual inoculum concentrations were confirmed by plating a portion of the inoculum on MMH agar plates. Following infection, mice were monitored regularly and euthanized when they met endpoint criteria, according to guidelines approved by the Animal Care and Use Committee at Rocky Mountain Laboratories (RML). Additional mice that received the sublethal vaccinating dose were euthanized at 4, 7, 10, 14, and 21 days postinfection and lungs, liver, and spleen collected for both pathology evaluation, determination of bacterial burden, and inflammatory cytokines. Briefly, organs were collected aseptically and placed in ice-cold tissue lysis buffer (150 mM Tris-HCl, 5 mM EDTA, 10 mM Trizma base) supplemented with a 1:100 dilution of phosphatase inhibitor cocktail I, phosphatase inhibitor cocktail II, and protease inhibitor cocktail III (all from AG Scientific, San Diego, CA). Organs were immediately homogenized by grinding tissues through a sterile S/S type 304 no. 60 wire mesh screen (Billeville Wire Cloth Co., Cedar Grove, NJ) using a 5-ml syringe plunger. A portion of the resulting homogenate was immediately serially diluted in PBS and plated on MMH agar for enumeration of bacterial loads. The remaining homogenates were centrifuged at 12,000 rpm for 20 min. Supernatants were then collected for analysis of TNF-α, IL-10, IL-4, monocyte chemoattractant protein 1 (MCP-1), RANTES, and macrophage inflammatory protein 1 alpha (MIP-1α) by a Cytometric bead array using an LSRII multiparameter flow cytometer and FCAP Array Software (all from BD Biosciences), according to the manufacturer’s instructions. IL-12p40 (BD Biosciences) or IL-1β and IL-1α (R&D Systems) were detected using a commercially available enzyme-linked immunosorbent assay (ELISA), according to the manufacturer’s instructions.
In vivo infections with SchuS4.C57BL/6J mice were intranasally infected with SchuS4 (∼25 CFU/25 μl saline). Actual inoculum concentrations were confirmed by plating a portion of the inoculum on MMH agar plates. Eight hours following infection, mice were treated with DMM (250 μg/kg of body weight or 400 μg/kg) or vehicle (PBS) in 200 μl via intraperitoneal (i.p.) injections. Mice were then treated i.p. every 24 h following the initial treatment with DMM or PBS and harvested at day 2 and day 4 postinfection for analysis of CFU in various tissues, as described above. Following infection, mice were monitored regularly and euthanized when they met endpoint criteria according to guidelines approved by the Animal Care and Use Committee at RML.
Quantitative PCR.Total RNA was isolated from LVS- or mock-infected BMDM in 24-well tissue culture plates 24 h postinfection using the quantitative PCR (qPCR) RNeasy kit (Qiagen), following the manufacturer’s instructions. Quantitative PCR was performed using TaqMan gene expression assays (Life Technologies) and primers for ACOD1 (Irg-1, Mm01224532_ml) and TATA box binding protein (Tbp, Mm00446973_ml) on a QuantStudio 6 Flex real-time PCR system (Thermo Fisher). Immune-responsive gene 1 (IRG-1) signal was normalized to TATA box binding protein (TBP) and the fold change calculated relative to mock-infected groups using the ΔΔCT method.
Histopathological examination of tissues.Tissues were fixed in 10% neutral buffered formalin with two changes of formalin for a minimum of 7 days. Tissues were placed in cassettes and processed with a Sakura Tissue Tek VIP-5 processor, on a 12-h automated schedule, using a graded series of ethanol, xylene, and Paraplast Xtra. Embedded tissues were sectioned at 5 μm and dried overnight at 42°C prior to staining. Tissue sections were stained with hematoxylin and eosin (H&E) and examined on an Olympus BX51 light microscope equipped with an Olympus DP722 camera and associated cellSens Dimension 1.4.1 software. Tissues were assessed by a board-certified pathologist. Lesions were scored from 0 to 5 based on the criteria shown in Table S1 in the supplemental material.
NADH/NAD+ assay.BMDM were plated at 8 × 104 cells/well in 80 μl cDMEM in a 96-well Seahorse assay culture plate. BMDM were infected with LVS (MOI = 50) and treated or not with IFN-γ (100 U/ml), and after 7 h, relative NADH and NAD+ levels were measured using the NAD/NADH-Glo assay (Promega), according to the manufacturer’s instructions. Briefly, cells were washed twice with 200 μl of PBS and then lysed in 0.2 N NaOH containing 1% dodecyltrimethylammonium bromide (DTAB). For the assessment of NAD+ levels, 25 μl of 0.4 N HCl was added to 50 μl of cell lysate, which was then heated for 15 min at 60°C. Samples were neutralized by addition of 25 μl of Trizma base prior to the addition of the luminescent detection reagent. For the assessment of NADH levels, no acid was added to the cell lysates prior to addition of the luminescent detection reagent. Samples were incubated at room temperature for 30 min and then assessed using a luminometer.
Statistical analysis.Statistical analysis involved comparisons of means using one- or two-way analysis of variance (ANOVA), followed by Tukey’s or Dunnett’s multicomparison test to compensate for increased type I error. Unpaired Student's t test was utilized for simple comparisons between two groups. An unpaired, nonparametric Mann-Whitney test was used for comparisons of histopathological scores. Statistical power was >0.8 to determine sample size, and statistical significance was defined as a probability of type I error occurring at <0.5% (P < 0.05). The minimum number of experimental replicates was 3 for in vitro studies. For in vivo studies, experiments were repeated twice with 3 to 5 mice per group for each experimental replicate, and data were pooled (6 to 10 mice total) for statistical analysis. Statistical analysis was performed using Prism 8.0.
ACKNOWLEDGMENTS
We thank Karin Peterson and Heinz Feldman for providing two strains of gene knockout mice used in these studies. We also thank Lydia Roberts for helpful discussion in the development and execution of this work.
This work was supported by the Intramural Research Program of the National Institutes of Health, National Institute of Allergy and Infectious Diseases.
Conceptualization, F.J., R.B., and C.M.B.; Methodology, F.J., R.B., B.S., D.S., and C.M.B.; Investigation, F.J., R.B., B.S., T.W., D.S., and C.M.B.; Writing – Original Draft, F.J. and C.M.B.; Writing – Reviewing & Editing, F.J. and C.M.B.
We declare no conflicts of interest.
FOOTNOTES
- Received 17 September 2019.
- Returned for modification 23 October 2019.
- Accepted 12 November 2019.
- Accepted manuscript posted online 18 November 2019.
Supplemental material is available online only.
- This is a work of the U.S. Government and is not subject to copyright protection in the United States. Foreign copyrights may apply.
