Research ArticleImmunology

Mitochondrial reactive oxygen species enable proinflammatory signaling through disulfide linkage of NEMO

See allHide authors and affiliations

Science Signaling  12 Feb 2019:
Vol. 12, Issue 568, eaar5926
DOI: 10.1126/scisignal.aar5926

Mitochondrial ROS in proinflammatory signaling

In addition to releasing proinflammatory cytokines that orchestrate immune responses to pathogens, macrophages also generate reactive oxygen species (ROS) that directly target the pathogen and specific host signaling pathways. Herb et al. found that infection with Listeria monocytogenes caused mouse macrophages to generate mitochondrial ROS (mtROS) that entered the cytosol and induced the formation of intermolecular disulfide bonds in nuclear factor κB (NF-κB) essential modulator (NEMO), a component of the inhibitor of κB kinase (IKK) complex. This covalent linkage of NEMO was required for activation of the ERK1/2 and NF-κB pathways and for the subsequent secretion of proinflammatory cytokines. Thus, disulfide linkage of NEMO is critical for activation of the IKK complex and can be triggered by the infection-induced production of mtROS.

Abstract

A major function of macrophages during infection is initiation of the proinflammatory response, leading to the secretion of cytokines that help to orchestrate the immune response. Here, we identify reactive oxygen species (ROS) as crucial mediators of proinflammatory signaling leading to cytokine secretion in Listeria monocytogenes–infected macrophages. ROS produced by NADPH oxidases (Noxes), such as Nox2, are key components of the macrophage response to invading pathogens; however, our data show that the ROS that mediated proinflammatory signaling were produced by mitochondria (mtROS). We identified the inhibitor of κB (IκB) kinase (IKK) complex regulatory subunit NEMO [nuclear factor κB (NF-κB) essential modulator] as a target for mtROS. Specifically, mtROS induced intermolecular covalent linkage of NEMO through disulfide bonds formed by Cys54 and Cys347, which was essential for activation of the IKK complex and subsequent signaling through the extracellular signal–regulated protein kinases 1 and 2 (ERK1/2) and NF-κB pathways that eventually led to the secretion of proinflammatory cytokines. We thus identify mtROS-dependent disulfide linkage of NEMO as an essential regulatory step of the proinflammatory response of macrophages to bacterial infection.

INTRODUCTION

Reactive oxygen species (ROS) have long been considered as toxic molecules that on the one hand are important for antimicrobial immunity but on the other hand cause collateral damage to organelles, cells, and tissues. There is now a growing appreciation that ROS can also influence specific cellular signaling pathways (1, 2). Superoxide (O2), the common precursor of all ROS, and hydrogen peroxide (H2O2) reversibly oxidize specific thiol groups in cysteine or methionine residues and thereby alter protein function (1, 2). The main sources of ROS production are members of the dihydronicotinamide adenine dinucleotide phosphate (NADPH) oxidase (Nox) family and mitochondria, where escape of electrons from the mitochondrial electron transport chain (ETC) results in the formation of O2. Both ROS produced by Nox family members, such as the phagocyte NADPH oxidase Nox2, and ROS produced by mitochondria contribute to antimicrobial immunity and to specific cellular signaling pathways, suggesting overlapping or redundant functions. The source of ROS and the signaling pathways that are regulated by them largely depend on the specific cell type and stimulus (1, 2).

One of the major functions of macrophages during infection is the secretion of proinflammatory cytokines (3) that helps to activate and orchestrate the immune response to the invading pathogen. The proinflammatory response leading to cytokine secretion is activated after recognition of the invading pathogen through pathogen-associated molecular patterns (PAMPs). For this purpose, macrophages use a large array of pattern recognition receptors (PRRs) (4), of which the Toll-like receptor (TLR) family is the best studied. Stimulation of TLRs activates a complex signaling cascade that culminates in the activation of the nuclear factor κB (NF-κB) and mitogen-activated protein kinase (MAPK) pathways that trigger proinflammatory cytokine secretion (5). This signaling cascade is initiated through the formation of a complex consisting of myeloid differentiation primary response 88 (MyD88); interleukin-1 (IL-1) receptor–associated kinase 4 (IRAK4), IRAK1, or IRAK2 (IRAK1/2); and tumor necrosis factor (TNF) receptor–associated factor 6 (TRAF6). TRAF6 then catalyzes the formation of polyubiquitin chains on both TRAF6 itself and on IRAK1/2. Recruitment of transforming growth factor–β (TGF-β)–activated kinase 1 (TAK1) to polyubiquitinated TRAF6 triggers TAK1-mediated activation of the p38 and c-Jun N-terminal kinases 1 and 2 (JNK1/2) MAPK pathways (6). By contrast, the extracellular signal–regulated protein kinases 1 and 2 (ERK1/2) MAPK pathway and the NF-κB pathways are activated by the inhibitor of κB (IκB) kinase (IKK) complex (6).

The IKK complex consists of the dimeric kinase subunits IKKα and IKKβ (IKKα/β) and the regulatory subunit NF-κB essential modulator (NEMO), also known as IKKγ (7, 8). NEMO is of particular importance for IKK complex activation because it both positively and negatively regulates both IKKα/β kinase activation and substrate interactions. The primary functional unit of the IKK complex is a constitutive, noncovalently linked dimer of NEMO associated with a homo- or heterodimer of IKKα/β. In resting cells, these minimal functional units are organized into higher-order lattice-like structures through direct interactions and the binding of NEMO to ubiquitin chains (9). In response to proinflammatory stimuli, these lattice-like structures are reorganized into condensed supramolecular complexes that promote activation of IKKα/β by phosphorylation (9, 10). Phosphorylated IKKβ then induces activation of the ERK1/2 and NF-κB pathways (7, 8). It induces NF-κB signaling by phosphorylating IκB proteins and the p65 subunit of NF-κB, also known as RelA (7, 8). IκB phosphorylation leads to its proteasomal degradation, thus releasing NF-κB to translocate into the nucleus, bind to DNA, and induce transcription of NF-κB target genes. NF-κB p65 phosphorylation enhances transactivation (11). Phosphorylated IKKβ induces ERK1/2 signaling by phosphorylating p105 and tumor progression locus 2 (TPL-2) (6). Phosphorylation of p105 results in proteolysis, leading to maturation into p50 and release of TPL-2. Upon release from the inactivating complex with p105 and further activation by IKK2-mediated phosphorylation, TPL-2 phosphorylates MAPK kinases 1 and 2 (MEK1/2), which in turn phosphorylate ERK1/2.

Regulation of the proinflammatory response to individual PAMPs that activate individual TLR signaling pathways, such as lipopolysaccharide (LPS), has been extensively studied (5). However, much less is known about the response to complex stimuli such as pathogenic bacteria that simultaneously activate multiple signaling pathways. Listeria monocytogenes is a Gram-positive bacterial pathogen that has been used to unravel various immunological processes (12, 13). Because L. monocytogenes specializes in escaping phagolysosomal degradation in macrophages by lysing the phagosomal membrane with its pore-forming toxin, listeriolysin O (LLO), and two phospholipase C proteins, recognition of L. monocytogenes infection not only involves surface PRR such as TLRs, scavenger receptors, and C-type lectin receptors but also involves cytosolic PRRs such as nucleotide-binding oligomerization domain–containing–like receptors and retinoic acid–inducible gene I–like receptors. Moreover, L. monocytogenes can also be recognized by some phagocytic receptors, such as opsonic receptors and integrins. One of the consequences of the simultaneous activation of these receptors is a robust induction of proinflammatory cytokine secretion, making L. monocytogenes infection an ideal model to study the proinflammatory response of macrophages to infection with a bacterial pathogen.

Depending on the context, ROS can either inhibit or stimulate the proinflammatory response of macrophages (14). Deficiency for the phagocyte NADPH oxidase Nox2 does not only lead to severe immunodeficiency but also to a hyperinflammatory phenotype with increased secretion of several proinflammatory cytokines (15), indicating that Nox2-derived ROS can influence proinflammatory signaling. Whether ROS produced by mitochondria contribute to the proinflammatory response of macrophages remains an open question because several studies have come to different conclusions (1619). In particular, the underlying molecular mechanisms by which mitochondrial ROS (mtROS) may regulate proinflammatory signaling and cytokine secretion have remained elusive.

Here, we used L. monocytogenes as a model pathogen to investigate whether and how Nox- or mitochondria-derived ROS regulate the proinflammatory response of macrophages to bacterial infection. Using genetic models, as well as chemical antioxidants and specific inhibitors, we identified mtROS as crucial stimulators of proinflammatory signaling leading to cytokine secretion by L. monocytogenes–infected macrophages. We delineated the pathway leading to mtROS production and identified the redox-dependent intermolecular covalent linkage of NEMO through disulfide bonds formed by Cys54 and Cys347 as a mechanism by which mtROS license activation of the IKK complex and thereby activation of the ERK1/2 and NF-κB pathways that eventually lead to the secretion of proinflammatory cytokines.

RESULTS

Cytosolic ROS are required for proinflammatory cytokine secretion by infected macrophages

Peritoneal macrophages from mice responded to infection with L. monocytogenes in vitro by secreting proinflammatory cytokines such as IL-1β, IL-6, and TNF (Fig. 1A). The ROS scavenger N-acetyl cysteine (NAC) reduced secretion of these cytokines by about 80% (Fig. 1A and fig. S1A), indicating that ROS play an important role in the proinflammatory response of L. monocytogenes–infected macrophages. Scavenging of ROS with NAC also reduced cytokine secretion by macrophages coincubated with nonpathogenic bacteria such as Gram-positive Bacillus subtilis or Gram-negative Escherichia coli (fig. S1B), indicating that ROS are generally required for the secretion of proinflammatory cytokines upon challenge of macrophages with bacteria. Furthermore, cytokine secretion after activation of individual TLRs with agonists specific for the heterodimeric TLRs, such as TLR2/1 (Pam3CSK4) or TLR2/6 (FSL-1), or the homodimeric TLRs, such as TLR3 (polyinosinic-polycytidylic acid [poly(I:C)]), TLR4 (LPS), or TLR9 (CpG-rich DNA), also was sensitive to scavenging of ROS with NAC (fig. S1C). Together, these data indicate that ROS played an important role in sescretion of proinflammatory cytokines by macrophages challenged with bacteria.

Fig. 1 Cytosolic ROS are required for proinflammatory cytokine secretion by infected macrophages.

(A) Peritoneal macrophages were infected with L. monocytogenes (L.m.) at a multiplicity of infection (MOI) of 1 in the presence or absence of the ROS scavenger NAC. Secretion of the proinflammatory cytokines IL-1β, IL-6, and TNF was quantified by enzyme-linked immunosorbent assay (ELISA) (n = 12 to 20 independent experiments). (B to E) Macrophages were infected at a MOI of 1 or 10. Both the kinetics of ROS production (line graphs) and the area under the curve (AUC, bar graphs), as a measure for the total amount of ROS produced, are shown. Extracellular ROS production was quantified by measuring isoluminol chemiluminescence (n = 7 independent experiments) (B). Cytosolic ROS production was quantified by measuring DCF fluorescence (n = 9 independent experiments) (C). Extracellular ROS production by peritoneal macrophages from wild-type (WT) and Nox2−/− mice was quantified by measuring isoluminol chemiluminescence (n = 5 independent experiments) (D). Cytosolic ROS production by macrophages from WT and Nox2−/− mice was quantified by measuring DCF fluorescence (n = 5 independent experiments) (E). RLU, relative light units. (F) Macrophages from WT and Nox2−/− mice were infected at a MOI of 1 in the presence or absence of NAC. Secretion of IL-1β, IL-6, and TNF was quantified by ELISA (n = 9 independent experiments). Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 by Student’s t test. n.i., not infected; n.s., not significant; n.d., not detectable.

ROS can be produced and act in different subcellular locations such as the phagosome lumen and the extracellular milieu (extracellular ROS) or the cytoplasm (intracellular ROS), and both extracellular and intracellular ROS can contribute to proinflammatory signaling in macrophages (14). Because NAC is cell permeable and therefore scavenges both extracellular and intracellular ROS, we next investigated the subcellular source of the ROS required for the secretion of proinflammatory cytokines by infected macrophages. Extracellular ROS can be measured with cell-impermeable dyes such as isoluminol (20), whereas intracellular ROS can be assessed by cell-permeable dyes such as the 2,7-dichlorofluorescein derivative 6-carboxy-2,7-dichlorodihydrofluorescein diacetate, di(acetoxymethyl ester) (referred to as DCF throughout this manuscript), which exclusively measures ROS in the cytosol (21). Infection of macrophages with L. monocytogenes resulted in a substantial increase in both extracellular (Fig. 1B) and cytosolic ROS production (Fig. 1C). Similarly, coincubation of macrophages with B. subtilis or E. coli induced an increase in extracellular and cytosolic ROS production (fig. S1, D and E). L. monocytogenes–infected macrophages from Nox2 knockout (Nox2−/−) mice did not generate any extracellular ROS at all (Fig. 1D), indicating Nox2 as the exclusive source of extracellular ROS. By contrast, the increase in cytosolic ROS production induced by L. monocytogenes infection was not compromised in Nox2−/− macrophages (Fig. 1E), showing that Nox2-derived ROS did not contribute to the increase in cytosolic ROS in L. monocytogenes–infected macrophages.

Cytokine secretion by L. monocytogenes–infected Nox2−/− macrophages was normal for IL-6 and TNF and even increased for IL-1β (Fig. 1F), indicating that extracellular Nox2-derived ROS are dispensable for the secretion of proinflammatory cytokines by infected macrophages. ROS scavenging with NAC reduced the secretion of cytokines by L. monocytogenes–infected Nox2−/− macrophages to a similar degree as in infected wild-type macrophages (Fig. 1F), further supporting that cytosolic ROS, rather than Nox2-derived extracellular ROS, were required for the secretion of proinflammatory cytokines by infected macrophages.

Cytosolic ROS are produced by complex III of the mitochondrial ETC

In addition to Nox2, the Nox family includes six other enzymes that could be responsible for the production of the cytosolic ROS that are required for proinflammatory cytokine secretion: Nox1, Nox3, Nox4, Nox5, Duox1, and Duox2. Because Nox3 is only found in the inner ear and Nox5 is absent in mice (22), neither was addressed in this study. L. monocytogenes–infected macrophages from Nox1−/−, Nox4−/−, Duox1−/−, and Duox2−/− mice produced normal amounts of cytosolic ROS (fig. S2A) and secreted normal amounts of proinflammatory cytokines (fig. S2B) compared to macrophages from wild-type mice. Moreover, simultaneous deficiency for Nox1 to 4 in macrophages deficient for p22phox (p22phox−/−), the common catalytic subunit of Nox1 to 4, compromised neither the production of cytosolic ROS (fig. S2A) nor cytokine secretion (fig. S2B). Furthermore, neither xanthine oxidase inhibition with allopurinol nor inducible nitric oxide synthase (iNOS) inhibition with N-nitroarginine methyl ester (L-NAME) altered cytosolic ROS (fig. S2C) or cytokine secretion (fig. S2D) by L. monocytogenes–infected macrophages. These results indicate that none of the Nox family members, xanthine oxidase, or iNOS were involved in cytosolic ROS production and proinflammatory cytokine secretion by L. monocytogenes–infected macrophages.

Another major source of cytosolic ROS is the mitochondrial ETC (23). Rotenone-mediated blockade of electron transfer from complex I to complex III of the ETC nearly completely abrogated the infection-induced production of cytosolic ROS and the secretion of all ILs, chemokines, and growth factors tested (Fig. 2, A to C, and fig. S3A), suggesting that secretion of proinflammatory cytokines by infected macrophages required mtROS production.

Fig. 2 Cytosolic ROS are produced by complex III of the mitochondrial ETC.

(A) Cytosolic ROS production of peritoneal macrophages in the presence or absence of the mitochondrial ETC complex I inhibitor rotenone was quantified by measuring DCF fluorescence (n = 7 independent experiments). (B) Secretion of IL-1β, IL-6, and TNF in the presence or absence of rotenone was quantified by ELISA (n = 9 to 17 independent experiments). (C) Secretion of the indicated cytokines in the presence or absence of rotenone was quantified by ProcartaPlex multiplex cytokine assay (n = 2 independent experiments). IFN-γ, interferon-γ; MIP-1α, macrophage inflammatory protein-1α; MCP-1, monocyte chemoattractant protein 1; GRO-α, growth-regulated oncogene α; G-CSF, granulocyte colony-stimulating factor; GM-CSF, granulocyte-macrophage colony-stimulating factor. (D) ROS production and release into the mitochondrial matrix were quantified by measuring MitoSOX Red fluorescence (n = 12 independent experiments). ROS production into the mitochondrial matrix was induced with rotenone as a positive control. (E) Cytosolic ROS production in the presence or absence of the superoxide scavenger 4-hydroxy-2,2,6,6-tetramethylpiperidinyloxyl (TEMPOL) was quantified by measuring DCF fluorescence (n = 2 independent experiments). (F) Secretion of IL-1β, IL-6, and TNF in the presence or absence of TEMPOL was quantified by ELISA (n = 4 independent experiments). (G) Macrophages were preincubated for 1 hour in the presence or absence of the hydrogen peroxide scavenger ebselen before infection. Cytosolic ROS production was quantified by measuring DCF fluorescence (n = 3 independent experiments). (H) Macrophages were preincubated for 1 hour in the presence or absence of ebselen before infection. Secretion of IL-1β, IL-6, and TNF was quantified by ELISA (n = 6 independent experiments). (I) Secretion of IL-1β, IL-6, and TNF in the presence or absence of the superoxide dismutase 1 (SOD1) inhibitor lung cancer screen 1 (LCS-1) was quantified by ELISA (n = 10 independent experiments). Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 by Student’s t test.

The mitochondrial ETC produces ROS that are released either into the mitochondrial matrix through complexes I, II, and III or into the mitochondrial intermembrane space through complex III (23). From there, the mtROS can reach the cytosol by diffusion (23). ROS production and release into the mitochondrial matrix can be measured with MitoSOX Red, a superoxide indicator that is specifically targeted to the mitochondrial matrix. Although L. monocytogenes infection substantially increased cytosolic ROS production, it did not increase ROS production into the mitochondrial matrix (Fig. 2D). Furthermore, scavenging of ROS specifically in the mitochondrial matrix with the mitochondrial matrix–targeted version of the O2 scavenger 2,2,6,6-tetramethylpiperidinyloxyl (MitoTEMPO) (fig. S3B) or by expression of the H2O2-decomposing enzyme catalase in the mitochondrial matrix [macrophages from mtCAT transgenic (mtCATtg) mice] (24) reduced neither cytosolic ROS (fig. S3, C and D) nor cytokine secretion (fig. S3, E and F). Similarly, inhibition of complex II of the ETC with malonate neither reduced cytosolic ROS production (fig. S3G) nor cytokine secretion (fig. S3H). By contrast, blocking electron transfer from complex III to complex IV with antimycin A strongly increased the production of cytosolic ROS (fig. S3I). Together, these data indicate that infection-induced mtROS were not released into the mitochondrial matrix but directly into the mitochondrial intermembrane space through complex III of the ETC.

mtROS released into the mitochondrial intermembrane space can enter the cytosol in the form of either O2 or H2O2 (23). In contrast to O2, which requires voltage-dependent anion channels to cross the mitochondrial outer membrane, H2O2 can leave the mitochondrial intermembrane space by crossing the mitochondrial outer membrane by diffusion (23). Specific scavenging of cytosolic O2 in L. monocytogenes–infected macrophages with TEMPOL neither altered cytosolic ROS (Fig. 2E) nor markedly reduced cytokine secretion (Fig. 2F). By contrast, specific scavenging of cytosolic H2O2 with ebselen strongly reduced both cytosolic ROS (Fig. 2G) and cytokine secretion (Fig. 2H), indicating that L. monocytogenes infection–induced mtROS reach the cytosol as H2O2. In support of this notion, inhibition of SOD1 with the SOD1-specific inhibitor LCS-1 nearly completely impaired cytokine secretion by L. monocytogenes–infected macrophages (Fig. 2I). Because SOD1 converts O2 into H2O2 both in the cytosol and in the mitochondrial intermembrane space, these data further support H2O2 as the cytosolic mtROS required for proinflammatory cytokine secretion. Collectively, these data indicate that infection of macrophages triggered O2 production into the mitochondrial inner membrane space by complex III of the ETC. There, O2 was converted by SOD1 into H2O2 that diffused into the cytosol, where it enabled the secretion of proinflammatory cytokines.

L. monocytogenes infection induces mtROS production through TLR2

We next addressed the source of the signals that induce mtROS production after infection. L. monocytogenes uses the pore-forming toxin LLO to induce fragmentation of the mitochondrial network, thereby impairing mitochondrial function, in HeLa cells (25), and perturbation of mitochondrial function can result in increased mtROS production. L. monocytogenes deficient for LLO (Δhly) or for prfA, the transcriptional master regulator of hly, and other virulence factors (ΔprfA) induced mtROS production to a similar extent as did wild-type L. monocytogenes (fig. S4A), thus excluding a role for L. monocytogenes virulence factors in the induction of mtROS. Furthermore, the mitochondrial network was not fragmented in L. monocytogenes–infected macrophages (fig. S4B). The mitochondrial membrane potential, an indicator of mitochondrial health, was not reduced after L. monocytogenes infection (fig. S4C). Instead, mitochondria were slightly hyperpolarized, indicating increased activity of the ETC. Thus, mtROS production induced by L. monocytogenes infection of macrophages was most likely not a result of perturbed mitochondrial function.

Excessive influx of Ca2+ into mitochondria can also cause increased mtROS production (26). Increasing cytosolic Ca2+ with the Ca2+ ionophore ionomycin was sufficient to induce mtROS production in uninfected macrophages (fig. S4D). Furthermore, macrophages infected with L. monocytogenes in the absence of extracellular Ca2+ nearly completely failed to produce mtROS (fig. S4E), indicating that influx of extracellular Ca2+ was essential for the induction of mtROS production in response to infection. However, inhibition of Ca2+ entry into mitochondria by inhibiting the mitochondrial calcium uniporter with ruthenium red or its derivative Ru360 did not reduce mtROS production induced by L. monocytogenes infection (fig. S4F), suggesting that Ca2+ induced mtROS production by acting as a second messenger rather than by overloading mitochondria.

Because TNF stimulation can increase mtROS production (27), we investigated whether TNF was involved in mtROS production by infected macrophages. Neither prolonged exposure to TNF before infection nor stimulation with TNF at the time of infection increased mtROS production (fig. S4G). Moreover, macrophages from TNF−/− mice produced normal amounts of mtROS (fig. S4H), IL-1β, and IL-6 (fig. S4I). These data indicate that TNF was not involved in mtROS production by L. monocytogenes–infected macrophages.

Salmonella typhimurium triggers antibacterial mtROS production through the TLR4-MyD88-TRAF6 signaling pathway (28). Therefore, we investigated whether L. monocytogenes infection induced mtROS production through TLR signaling. The production of mtROS after L. monocytogenes infection was strongly, although not completely, reduced in macrophages from TLR2−/− mice (Fig. 3A). However, macrophages from MyD88−/− mice completely failed to induce mtROS production in response to L. monocytogenes infection (Fig. 3B), indicating that, in addition to TLR2, other MyD88-dependent TLRs also contributed to the induction of mtROS production after L. monocytogenes infection. Likewise, in macrophages from mice deficient for TRAF6 (TRAF6−/−), a signal transducer downstream of many PRR receptors (29), mtROS production in response to L. monocytogenes infection was completely abrogated (Fig. 3C). Cytokine secretion after L. monocytogenes infection was markedly reduced in TLR2−/− macrophages and almost completely abolished in MyD88−/− and TRAF6−/− macrophages (Fig. 3, D to F). These data indicate that mtROS production and cytokine secretion by L. monocytogenes–infected macrophages were predominantly triggered by the TLR2-MyD88-TRAF6 signaling pathway.

Fig. 3 L. monocytogenes infection induces mtROS production through TLR2.

(A to C) Cytosolic ROS production by peritoneal macrophages from WT or TLR2−/− (A), MyD88−/− (B), or TRAF6MYEL-KO (C) mice was quantified by measuring DCF fluorescence [n = 7 independent experiments (A), n = 7 independent experiments (B), and n = 6 independent experiments (C)]. (D to F) Secretion of IL-1β, IL-6, and TNF in WT or TLR2−/− (D), MyD88−/− (E), or TRAF6MYEL-KO (F) macrophages was quantified by ELISA [n = 5 independent experiments (D), n = 3 independent experiments (E), and n = 4 independent experiments (F)]. Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 by Student’s t-test.

mtROS promote proinflammatory signaling through the ERK1/2 and NF-κB pathways

Because signaling through TLR2, MyD88, and TRAF6 in L. monocytogenes–infected macrophages was required for both mtROS production and cytokine secretion, we hypothesized that mtROS act on the proinflammatory signaling pathways leading to cytokine production. The main pathways leading to the production of cytokines are the ERK1/2, JNK1/2, and p38 MAPK pathways and the pathways leading to activation of NF-κB. Preventing mtROS generation either by scavenging of ROS with NAC or by inhibiting mtROS production with rotenone reduced the phosphorylation of ERK1/2 (Fig. 4, A and B) and the upstream regulators MEK1/2 (fig. S5A) in L. monocytogenes–infected macrophages by more than 50%, indicating that mtROS were required for signaling through the ERK1/2 pathway in this context. By contrast, the phosphorylation of JNK1/2 was only slightly decreased, and the phosphorylation of p38 remained unchanged (Fig. 4, A and B). Therefore, we hypothesized that the ERK1/2 pathway was the major pathway leading to the secretion of cytokines in L. monocytogenes–infected macrophages. Inhibition of the ERK1/2 pathway with the MEK1/2 inhibitor PD98059 was sufficient to completely block the secretion of IL-1β and IL-6 and to reduce TNF secretion (Fig. 4C). These data indicate that mtROS were required for proinflammatory signaling leading to cytokine secretion by L. monocytogenes–infected macrophages mainly because they promote signaling through the ERK1/2 pathway.

Fig. 4 mtROS promote proinflammatory signaling through the ERK1/2 and NF-κB pathways.

(A and B) At the indicated time points after infection of peritoneal macrophages in the presence or absence of NAC or rotenone, phosphorylation of ERK1/2 (n = 3 independent experiments), JNK1/2 (n = 5 to 7 independent experiments), p38 (n = 6 to 8 independent experiments), and NF-κB p65 (n = 4 to 6 independent experiments) and degradation of IκBα (n = 4 to 7 independent experiments) were analyzed by Western blot of whole-cell extracts (A) and quantified by densitometry (B). p.i., post infection. (C) Secretion of IL-1β, IL-6, and TNF in the presence or absence of the MEK1/2 inhibitor PD98059 was quantified by ELISA (n = 3 independent experiments). (D) At 1 hour after infection in the presence or absence of NAC or rotenone, in vitro binding of NF-κB in nuclear extracts to a DNA probe was analyzed by electrophoretic mobility shift assay (EMSA) and quantified by densitometry (n = 3 independent experiments). Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 by Student’s t test.

MAPKs induce cytokine secretion in conjunction with the NF-κB pathway (30). Degradation of IκBα, a prerequisite for the release of the p65 subunit of NF-κB, was not reduced in infected macrophages in the absence of mtROS generation (Fig. 4, A and B). However, phosphorylation of NF-κB p65 (Fig. 4, A and B), which is necessary for optimal induction of NF-κB target genes (11), translocation of NF-κB p65 from the cytosol into the nucleus (fig. S5B), and DNA binding by NF-κB (Fig. 4D), were substantially reduced in infected macrophages in the absence of mtROS generation, indicating that mtROS also promote signaling leading to activation of the NF-κB pathway.

mtROS enable IKK complex activation by inducing intermolecular covalent linkage of NEMO through disulfide bonds formed by Cys54 and Cys347

Signaling through both the ERK1/2 and the NF-κB pathways is initiated by the IKK complex, whereas the JNK1/2 and p38 pathways are activated independently of the IKK complex (6). For IKK complex activation, phosphorylation of IKKα and, particularly, of IKKβ is essential (6). Phosphorylation of IKKβ in L. monocytogenes–infected macrophages was reduced by more than 50% in the absence of mtROS generation (Fig. 5A), indicating that activation of the IKK complex required mtROS-dependent signals.

Fig. 5 mtROS induce intermolecular covalent linkage of NEMO through disulfide bonds formed by Cys54 and Cys347.

(A) At the indicated time points after infection of peritoneal macrophages in the presence or absence of NAC or rotenone, the phosphorylation status of IKKβ (n = 6 to 8 independent experiments) was analyzed by Western blot of whole-cell extracts and quantified by densitometry. (B) At the indicated time points after infection in the presence or absence of rotenone, covalent linkage of NEMO through disulfide bonds was analyzed by nonreducing SDS–polyacrylamide gel electrophoresis (PAGE) and Western blot, and quantified by densitometry (n = 3 independent experiments). MW, molecular weight. (C) Covalent linkage of NEMO through disulfide bonds was stabilized by amine-to-amine cross-linking with BS3, analyzed by standard reducing SDS-PAGE and Western Blot, and quantified by densitometry (n = 5 independent experiments). (D and E) Macrophages were transfected with mRNA encoding WT NEMO (NEMOWT) or a NEMO mutant in which redox-sensitive Cys54 and Cys347 are mutated to alanine (NEMOC54/347A) before infection. Covalent linkage of NEMO through disulfide bonds was analyzed by nonreducing SDS-PAGE and Western blot and quantified by densitometry [n = 2 independent experiments (D) and n = 5 independent experiments (E)]. (F) The presence of IKKα and IKKβ in the covalently linked NEMO-containing complex was investigated by nonreducing SDS-PAGE of cell extracts and Western blotting for IKKα and IKKβ (n = 3 independent experiments). Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, and ***P < 0.001 by Student’s t test.

The phosphorylation status of IKKβ, as well as that of several other kinases involved in proinflammatory signaling, is determined by the relative activities of upstream kinases and specific phosphatases like protein phosphatase 2A (PP2A) and PP2Cβ (8). Like many other phosphatases, PP2A (31) and PP2Cs (32) are sensitive to oxidative inactivation, which favors the phosphorylation of their substrates. Therefore, we investigated whether oxidative inactivation of phosphatases by mtROS was required for proinflammatory signaling leading to cytokine secretion by L. monocytogenes–infected macrophages. We tested whether phosphatase inhibition could rescue impaired cytokine secretion by mtROS-deficient macrophages. Neither inhibition of protein tyrosine phosphatases (PTPs) with orthovanadate or PTP inhibitor I nor inhibition of serine-threonine phosphatases with ocadaic acid, calyculin A, tautomycin, or sanguinarine rescued infection-induced cytokine secretion by mtROS-deficient macrophages (fig. S6). Thus, mtROS did not promote proinflammatory signaling by inactivating redox-sensitive phosphatases.

An important regulator of IKK complex activation is its subunit NEMO (7). The primary functional unit of NEMO is a constitutive, noncovalently linked dimer. Covalent links between NEMO monomers formed by disulfide bonds between the redox-sensitive cysteines Cys54 and Cys347 have been shown to stabilize the NEMO dimer and increase its affinity to IKKβ in vitro (3335). Therefore, we investigated whether mtROS produced by L. monocytogenes–infected macrophages enabled IKK complex activation by inducing disulfide bond–mediated covalent linkage of NEMO dimers.

Nonreducing SDS-PAGE under conditions that disrupt noncovalent interactions but preserve covalent links such as disulfide bonds showed that naïve macrophages did not contain disulfide-linked NEMO molecules (Fig. 5B). Upon infection of macrophages with L. monocytogenes, however, NEMO was recruited into a disulfide-linked complex of about 200 kDa (Fig. 5, B and C). mtROS produced in response to L. monocytogenes infection were crucial for the formation of this complex (Fig. 5B). Expression of a NEMO mutant in which the redox-sensitive cysteines Cys54 and Cys347 are mutated to alanine (NEMOC54/347A) and that therefore cannot form disulfide-linked dimers (33) impaired NEMO complex formation in response to infection (Fig. 5, D and E, and fig. S7, A and B). The other components of the IKK complex, IKKα and IKKβ, were not included in the complex (Fig. 5F). Together, these data indicate that mtROS produced in response to L. monocytogenes infection induced disulfide linkage of NEMO through Cys54 and Cys347.

Similar to deficiency for mtROS production, NEMOC54/347A expression specifically impaired infection-induced IKKβ and ERK1/2 phosphorylation (Fig. 6A) and DNA binding of NF-κB (Fig. 6B), as well as secretion of proinflammatory cytokines (Fig. 6C). Although these macrophages still expressed endogenous wild-type NEMO (NEMOWT), the effect of NEMOC54/347A expression was distinct, indicating a strong dominant-negative effect on NEMO complex formation. Notwithstanding, we decided to corroborate these results by reexpressing NEMOWT or NEMOC54/347A in NEMO-deficient cells. Like peritoneal macrophages, bone marrow–derived macrophages (BMDMs) (fig. S8, A to C) and mouse embryonic fibroblasts (MEFs) (fig. S9, A to C) strictly required mtROS for cytokine secretion and NEMO complex formation upon L. monocytogenes infection, and both NEMO−/− BMDMs (fig. S8, D and E) and NEMO−/− MEFs (fig. S9, D and E) were completely defective for infection-induced IKK complex activation and subsequent proinflammatory signaling leading to cytokine secretion. Reexpression of NEMOWT, but not that of NEMOC54/347A, completely rescued infection-induced IKK complex activation and subsequent proinflammatory signaling leading to cytokine secretion in NEMO−/− BMDMs (fig. S8, D and E) and in NEMO−/− MEFs (fig. S9, D and E). These data show that mtROS-mediated disulfide linkage of NEMO through Cys54 and Cys347 was absolutely crucial for IKK complex activation and subsequent proinflammatory signaling leading to cytokine secretion upon infection.

Fig. 6 mtROS-mediated disulfide linkage of NEMO is crucial for activation of proinflammatory signaling leading to cytokine secretion.

(A to C) Peritoneal macrophages were transfected with mRNA encoding NEMOWT or NEMOC54/347A before infection. At the indicated time points after infection, the phosphorylation of IKKβ, ERK1/2, JNK1/2, p38, and NF-κB p65 and the degradation of IκBα were analyzed by Western blot and quantified by densitometry (n = 3 independent experiments) (A). In vitro binding to a DNA probe of NF-κB in nuclear extracts of macrophages infected for 1 hour was analyzed by EMSA and quantified by densitometry (n = 3 independent experiments). The dotted line indicates where intervening bands have been excised (B). Secretion of IL-1β, IL-6, and TNF was quantified by ELISA (n = 4 independent experiments) (C). n.t., not transfected. (D) Macrophages were transfected with mRNA encoding WT IKKβ (wtIKKβ) or constitutively active IKKβ (caIKKβ) and then infected with L. monocytogenes or stimulated with LPS. Secretion of IL-1β, IL-6, and TNF in the presence or absence of NAC or rotenone was quantified by ELISA (n = 3 independent experiments). Data are shown as mean ± SEM. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001 by Student’s t test.

If mtROS permit proinflammatory signaling leading to cytokine secretion exclusively by enabling activation of the IKK complex through disulfide linkage of NEMO, expression of a caIKKβ mutant (36) should allow for cytokine secretion even in the absence of mtROS production. Expression of caIKKβ resulted in infection-induced proinflammatory cytokine secretion even by mtROS-deficient macrophages (Fig. 6D and fig. S10, A to C), indicating that mtROS were not required for processes downstream of IKK complex activation. Moreover, mtROS deficiency also did not impair processes directly upstream of IKK complex activation such as the degradation of IRAK1 (fig. S11A) or the posttranslational modification of NEMO observed upon infection (fig. S11B).

Collectively, these data indicate that mtROS produced in response to L. monocytogenes infection enabled activation of the IKK complex and subsequent signaling through the ERK1/2 and NF-κB pathways leading to cytokine secretion by inducing intermolecular disulfide linkage of NEMO through the redox-sensitive cysteines Cys54 and Cys347.

On the basis of our data, we propose the following model for licensing of the proinflammatory response of L. monocytogenes–infected macrophages by mtROS (Fig. 7): Recognition of L. monocytogenes by TLR2 in conjunction with other MyD88-dependent TLRs induces activation of IRAK1/2 and TRAF6. TRAF6 then induces mtROS production. mtROS are produced as O2 and released into the mitochondrial intermembrane space by complex III of the ETC and converted into H2O2 by SOD1, thus allowing them to enter the cytosol by diffusion. In the cytosol, mtROS mediate intermolecular covalent linkage of NEMO through disulfide bonds formed by the redox-sensitive cysteines Cys54 and Cys347, which allows for activation of the IKK complex. Last, the fully active IKK complex activates the ERK1/2 and NF-κB pathways, resulting in proinflammatory cytokine secretion.

Fig. 7 Model of how mtROS enable proinflammatory signaling in infected macrophages through disulfide linkage of NEMO.

On the basis of our data, we propose the following model for the licensing of proinflammatory cytokine secretion by mtROS in L. monocytogenes–infected macrophages: (1) Recognition of L. monocytogenes by TLR2, in conjunction with other MyD88-dependent TLRs, induces activation of IRAK1/2 and TRAF6. (2) TRAF6 induces mtROS production. mtROS are produced as O2 by complex III of the ETC and released into the intermembrane space, where they are converted into H2O2 by SOD1, which allows them to enter the cytosol by diffusion. (3) In the cytosol, mtROS mediate intermolecular linkage of NEMO through disulfide bonds between the redox-sensitive cysteines Cys54 and Cys347. (4) Covalent linkage of NEMO molecules allows activation of the IKK complex. (5) The fully active IKK complex activates the ERK1/2 and NF-κB pathways, resulting in (6) proinflammatory cytokine production and secretion. IMS, intermembrane space.

DISCUSSION

Using L. monocytogenes, a pathogen that has been used previously to unravel various immunological processes (12, 13), we identify ROS produced by mitochondria as crucial regulators of the proinflammatory response to bacterial infection. Mechanistically, we found that mtROS-mediated disulfide linkage of NEMO was an essential regulatory step in proinflammatory signaling leading to cytokine secretion. Specifically, mtROS mediated covalent linkage of NEMO through disulfide bonds formed by Cys54 and Cys347, which is essential for activation of the IKK complex, and subsequent signaling through the ERK1/2 and NF-κB pathways that eventually leads to proinflammatory cytokine secretion.

ROS have been reported to act in various signaling pathways (1, 2). The subcellular source of the ROS and the signaling pathways that are regulated by them largely depend on the specific cell type and stimulus (1, 2). Here, we used L. monocytogenes infection to model the proinflammatory response of macrophages to a complex stimulus, a pathogenic bacterium, which simultaneously activates multiple signaling pathways. Because the phagocyte NADPH oxidase Nox2 is the predominant source of ROS produced by macrophages, Nox2-derived ROS often take center stage in the antimicrobial response of macrophages. However, our data clearly show that the proinflammatory response of L. monocytogenes–infected macrophages depended on ROS produced by mitochondria. Scavenging ROS with NAC or selective blockade of electron transfer through the mitochondrial ETC with rotenone nearly completely abrogated the production of cytosolic ROS and proinflammatory signaling leading to cytokine secretion. By contrast, Nox2, other Nox or Duox enzymes, xanthine oxidase, or iNOS were not required for cytosolic ROS production and cytokine secretion. Thus, our data show that mtROS are of critical importance for the proinflammatory response leading to secretion of cytokines by L. monocytogenes–infected macrophages.

Previously, several studies had used LPS as a specific PAMP to address the question of whether and how mtROS contribute to proinflammatory cytokine secretion, yet came to different conclusions (1618). In these studies, mtROS were found to be required for TNF secretion upon hypoxia but not upon LPS stimulation (16); required for increased LPS-induced secretion of IL-1β but not TNF, and actually decreased secretion of IL-10 (18); or partly reduced LPS-induced secretion of TNF, IL-6, IL-8, and IL-10 (17). Our data indicate that cytokine secretion in response to the TLR4 agonist LPS, to the TLR3 agonist poly(I:C), and to the nonpathogenic E. coli was partially sensitive to mtROS scavenging. However, cytokine secretion in response to the TLR2/1 agonist Pam3CSK4, the TlR2/6 agonist FSL-1, the TLR9 agonist CpG-rich DNA, or Gram-positive bacteria such as L. monocytogenes or B. subtilis was much more sensitive to mtROS scavenging. Thus, our data suggest that mtROS are of particular importance for the proinflammatory response to stimuli associated with Gram-positive bacteria.

Stimuli associated with Gram-negative pathogens induce ROS production into the mitochondrial matrix (17, 28), which in turn is required for proinflammatory cytokine secretion (17, 18). Our data show that L. monocytogenes infection did not induce the release of ROS into the mitochondrial matrix and that cytokine secretion by L. monocytogenes–infected macrophages did not require ROS release into the mitochondrial matrix. By contrast, L. monocytogenes–induced mtROS were released directly into the mitochondrial intermembrane space by complex III of the ETC. Thus, our data reveal a distinct response to Gram-negative pathogens also in the location of mtROS production, indicating different mitochondrial targets for the signals induced by Gram-positive and Gram-negative pathogens, respectively.

The signaling functions of ROS are exerted mostly by O2 and H2O2, with H2O2 being by far the more stable signaling intermediate (1). Our data indicate that it is the production of cytosolic H2O2 that is required for proinflammatory signaling leading to cytokine secretion. Scavenging of O2 in neither the cytosol nor the mitochondrial matrix had any effect on cytokine secretion by L. monocytogenes–infected macrophages. By contrast, scavenging of H2O2 in the cytosol, but not in the mitochondrial matrix, and inhibition of SOD1 strongly impaired proinflammatory cytokine secretion. Thus, the proinflammatory response of macrophages to L. monocytogenes infection depends on cytosolic H2O2 originating from mitochondria-produced O2.

The molecular mechanisms by which mtROS mediate the proinflammatory response of macrophages so far had remained uncharacterized. Inactivation of redox-sensitive phosphatases by mtROS has been suggested as a potential mechanism (17). In L. monocytogenes–infected macrophages, inhibition of PTPs or serine-threonine phosphatases did not restore cytokine secretion by mtROS-deficient macrophages. Instead, our data indicate that the critical mtROS-dependent step in proinflammatory signaling leading to cytokine secretion is the covalent linkage of NEMO through disulfide bonds formed by the redox-sensitive cysteines Cys54 and Cys347, which allows activation of the IKK complex. This conclusion is supported by our findings that: (i) L. monocytogenes infection resulted in linkage of NEMO through disulfide bonds; (ii) deficiency for mtROS production impaired this disulfide linkage of NEMO and IKK complex activation, proinflammatory signaling through ERK1/2 and NF-κB, and cytokine secretion; (iii) expression of the NEMOC54/347A mutant (33) that cannot undergo disulfide linkage at Cys54 and Cys347 phenocopied mtROS deficiency; and (iv) expression of a constitutively active mutant of IKKβ (36) that bypasses NEMO regulation of IKK complex activation completely restored secretion of proinflammatory cytokines in mtROS-deficient macrophages. Thus, our data introduce mtROS-mediated disulfide linkage of NEMO as an essential regulatory step required for activation of the IKK complex and, thereby, for enabling ERK1/2 and NF-κB signaling leading to proinflammatory cytokine secretion.

The specific consequences of disulfide linkage for NEMO structure, function, and interactions with its numerous binding partners remain to be elucidated. In particular, the precise composition of the covalently linked 200-kDa complex that is formed upon infection through the formation of the disulfide bonds between NEMO molecules remains to be investigated. Cys54 and Cys347 have been shown to mediate disulfide linkage of NEMO monomers into a covalently linked dimer of about 100 kDa in H2O2-treated cells (3335). Our data now show that, in infected cells, Cys54 and Cys347 mediate additional covalent interactions. The other subunits of the IKK complex, IKKα/β, were not covalently linked to NEMO upon infection. Because dimerization of NEMO dimers to form a 200-kDa complex has been described (37), Cys54 and Cys347 may mediate the formation of a covalently linked dimerized dimer of NEMO. Alternatively, they may covalently link NEMO to proteins other than IKKα/β. Collectively, our data elucidate the molecular mechanism by which mtROS license proinflammatory cytokine secretion and identify disulfide linkage of NEMO as an essential regulatory step required for activation of the IKK complex and subsequent signaling through the ERK1/2 and NF-κB pathways.

MATERIALS AND METHODS

Mice

Nox2−/− (38) and Nox4−/− mice (39) were a gift of R.P. Brandes (Goethe University, Germany). Nox1−/− mice (40) were a gift of K.-H. Krause (University of Geneva, Switzerland). p22phox−/− mice (41) were a gift of J. Woo (Stanford University Medical Center, USA). Duox1−/− mice (42) were a gift of M. Geiszt (Semmelweis University, Hungary). TLR2−/− (43), mtCATtg (24), and Duox2−/− (44) mice were purchased from the Jackson laboratory. MyD88−/− (45), TRAF6fl/fl/LysMCre+/− (TRAF6MYEL-KO) (46), and NEMOfl/fl (47) mice were a gift of M. Pasparakis (University of Cologne, Germany). TNF−/− mice (48) were a gift of S. Nedoskasov (German Rheumatism Research Center, Germany). All mice were backcrossed at least 10 times to the C57BL/6 background. Mice were kept under specific pathogen–free conditions at the animal facilities of the Medical Centre of the University of Cologne. Experiments were performed in accordance with the Animal Protection Law of Germany in compliance with the Ethics Committee at the University of Cologne with 8- to 20-week-old mice.

Bacteria

In vivo–passaged L. monocytogenes, strain EGD-e, serotype 1/2a, and the isogenic deletion mutants Δhly and ΔprfA (49) were cultured in brain heart infusion (BHI) medium until mid-log phase. Heat killing of L. monocytogenes by incubation in phosphate-buffered saline (PBS) for 30 min at 70°C was verified by plating on blood agar plates. B. subtilis subsp. spizizenii were cultured in BHI medium, and E. coli K12 DH5-α were cultured in lysogeny broth (LB) medium.

Culture and infection of peritoneal macrophages, BMDMs, and MEFs

Mouse peritoneal macrophages were enriched from peritoneal cells by magnetic cell sorting using CD11b MicroBeads (Miltenyi Biotec) according to the instructions of the manufacturer as previously described (50). Macrophages were seeded into well plates and allowed to adhere overnight in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% fetal calf serum (FCS).

For generation of BMDMs from NEMOfl/fl mice, bone marrow cells were prepared from tibias and femurs of 7- to 12-week-old mice and cultured for 6 days in VLE-RPMI 1640 medium supplemented with 10% FCS, penicillin (100 U/ml), streptomycin (100 μg/ml), 2 mM Hepes, 200 nM sodium pyruvate, and recombinant macrophage colony-stimulating factor (20 ng/ml) (PeproTech). Then, BMDMs were seeded into well plates in VLE-RPMI 1640 medium supplemented with 10% FCS.

Mycoplasma-free MEFs from wild-type and NEMO−/− mice (47) were a gift of M. Pasparakis (University of Cologne, Cologne, Germany) and cultured in DMEM supplemented with 5% FCS, penicillin (100 U/ml), and streptomycin (100 μg/ml). For infection, MEFs were seeded into well plates and allowed to adhere overnight in DMEM supplemented with 5% FCS.

Peritoneal macrophages were infected at a MOI of 1 unless stated otherwise, BMDMs at a MOI of 1 or 5, and MEFs at a MOI of 5 or 25. Infection was synchronized by centrifugation at 850g for 5 min at 4°C. After three wash steps, peritoneal macrophages were incubated in Hanks’ balanced salt solution (HBSS) with Ca2+ and Mg2+ supplemented with 5% heat-inactivated normal mouse serum (Dunn Lab), BMDMs in VLE-RPMI 1640 medium supplemented with 10% FCS, and MEFs in DMEM supplemented with 5% FCS. At 1 hour after infection of MEFs, gentamycin (0.5 μg/ml) was added to prevent proliferation of extracellular L. monocytogenes. Where indicated, NAC (50 mM), rotenone (100 μM), 4-hydroxy-2,2,6,6-tetramethylpiperidinyloxyl (TEMPOL) (500 μM), ebselen (20 μM), allopurinol (100 μM), l-NAME (100 μM), MitoTEMPO (500 μM), malonate (100 μM), antimycin A (30 μM), carbonyl cyanide 3-chlorophenylhydrazone (CCCP) (50 μM), ionomycin (100 μM), RuRed (10 μM), Ru360 (5 μM), sanguinarine (1 nM) (all from Sigma-Aldrich), PD98059 (20 μM), calyculin A (100 μM), PTP inhibitor I (5 μM), orthovanadate (1 μM), okadaic acid (1.5 nM), tautomycin (10 nM) (all from Merck Millipore), LCS-1 (7.5 μM) (Calbiochem), or TNF (10 ng/ml) (R&D Systems) was added to the medium before or after infection as indicated. All substances were used at concentrations confirmed to be nontoxic for peritoneal macrophages by CyQUANT Direct Cell Proliferation Assay (Thermo Fisher Scientific).

Analysis of cell viability

CyQUANT Direct Cell Proliferation Assay was used to determine cell viability in accordance with the manufacturer’s instructions (Thermo Fisher Scientific). Macrophages lysed with 0.1% Triton X-100 were used as 100% dead control for each individual sample.

Quantification of cytokine secretion

Cytokines in supernatants of peritoneal macrophages infected for 5 hours, or BMDM or MEF infected for 24 hours were measured according to the instructions of the manufacturer by ELISA (R&D Systems) or by ProcartaPlex multiplex cytokine assay (Affymetrix). Where indicated, cytokine production was stimulated for 5 hours with Pam3CSK4 (10 μg/ml), FSL-1 (5 μg/ml), ultrapure LPS from E. coli O111:B4 (5 μg/ml), ultrapure flagellin from S. typhimurium (10 μg/ml), imiquimod (5 μg/ml), CpG-rich DNA oligonucleotides ODN 1668 (5 μg/ml) (all purchased from InvivoGen), or poly(I:C) (5 μg/ml) (Calbiochem) instead of infection. Absorbance was measured with a Tecan Infinite M 1000 (Tecan Group), and ProcartaPlex bead fluorescence was measured with a Luminex 100 system (Luminex Corporation).

Quantification of ROS production

Peritoneal macrophages were seeded in white or black 96-well plates for luminescence or fluorescence measurements, respectively. Macrophages were infected in ice-cold HBSS. To analyze extracellular ROS production, macrophages were incubated in 50 μM isoluminol (Sigma-Aldrich) and horseradish pereoxidase (3.2 U/ml) (Merck Millipore) in HBSS after infection. To analyze cytosolic ROS production, macrophages were incubated in 20 μM DCF (Thermo Fisher Scientific) in HBSS for 15 min at 37°C before infection. To analyze ROS production into the mitochondrial matrix, macrophages were incubated in 5 μM MitoSOX Red (Thermo Fisher Scientific) in HBSS for 15 min at 37°C before infection. Because Nox2-derived ROS oxidized MitoSOX Red already in the extracellular milieu, leading to false-positive MitoSOX Red fluorescence, Nox2−/− macrophages were used for these experiments. After infection, plates were transferred on ice to the respective plate reader preheated to 37°C. Isoluminol luminescence or DCF fluorescence was measured at 1-min intervals using a TriStar2 LB 942 Multimode Plate Reader (Berthold Technologies), and MitoSOX Red fluorescence was measured using a Tecan Infinite M 1000 microplate reader (Tecan Group).

Immunoblotting

Macrophages were lysed in radioimmunoprecipitation assay buffer [50 mM tris-HCl, 150 mM NaCl, 0.1% Nonidet P-40, 0.5% sodium desoxicholate, 1% SDS, 0.5% benzonase endonuclease (Merck Millipore), and protease and phosphatase inhibitor cocktails] and subjected to SDS-PAGE and Western blot as previously described (50).

Covalent linkage of NEMO through disulfide bonds was analyzed under nonreducing conditions as previously described (33). Briefly, macrophages were lysed in AT buffer lacking dithiothreitol (DTT) but containing 20 mM N-ethylmaleimide (NEM) (Sigma-Aldrich). Samples were heated in SDS sample buffer containing only 0.2% (w/v) final concetration (f.c.) SDS and no β-mercaptoethanol and then were subjected to standard SDS-PAGE and Western blot.

For cross-linking of cellular proteins, cell lysates in AT buffer lacking DTT but containing 20 mM NEM were incubated for 30 min at room temperature (RT) with 0.1 nM BS3 [bis(sulfosuccinimidyl)suberate] (Thermo Fisher Scientific). Then, 20 mM tris was added for 15 min at RT to quench cross-linking reactions. Samples were heated in SDS sample buffer containing 0.2% (w/v) f.c. SDS and 10% β-mercaptoethanol and subjected to standard SDS-PAGE and Western blot.

Monoclonal antibody against β-actin (1:10,000; catalog no. A2228) was from Sigma-Aldrich. Monoclonal antibodies recognizing phospho-IKKα/β (Ser176/180) (1:1000; catalog no. 2697), phospho-p38 MAPK (Thr180/182) (1:1000; catalog no. 4511), phospho-p65 (Ser536; 1:1000) (catalog no. 3033), IκBα (catalog no. 4812), IKKβ (1:1000; catalog no. 8943), IRAK1 (1:1000; catalog no. 4504), or FLAG (1:1000; catalog no. 8146) and polyclonal antibodies recognizing phospho-JNK (Thr183/185) (1:1000; catalog no. 9251), phospho-MEK1/2 (Ser217/221) (1:1000; catalog no. 9121), phospho-ERK1/2 (Thr202/204) (1:1000; catalog no. 9101), or IKKα (1:1000; catalog no. 2682) were from Cell Signaling Technology. The monoclonal antibody recognizing NEMO/IKKγ (1:1000; catalog no. MA1-41046) was from Invitrogen. Because NEMO band intensity strongly increased after infection when NEMO also becomes slightly larger in size (Fig. 5C and fig. S11B), we speculate that this antibody preferentially binds to posttranslationally modified NEMO. In addition, the apparent increase in total NEMO upon infection under complex-preserving conditions such as cross-linking (Fig. 5C) or nonreducing conditions (Fig. 5, B and E, fig. S8C, and fig. S9C) is probably due to the presence of multiple posttranslationally modified NEMO molecules in the complex, even further improving antibody binding due to increased avidity. Band intensities were quantified using ImageJ software (National Institutes of Health).

Immunofluorescence staining and microscopy

Peritoneal macrophages were allowed to adhere to sterile glass coverslips overnight. At the indicated time points after infection, cells were fixed in 3% paraformaldehyde in PBS for 20 min at RT. Staining of L. monocytogenes was performed as previously described (50); extracellular L. monocytogenes were excluded from the analysis. Nuclei were stained with 4′,6-diamidino-2-phenylindole and NF-κB p65 with a monoclonal antibody (catalog no. sc-8008, Santa Cruz Biotechnology). Samples were mounted on glass microscopic slides in ProLong Gold antifade reagent (Invitrogen) and analyzed using an IX81 fluorescence microscope (Olympus). In each experiment, at least 100 infected cells per time point and condition were analyzed.

For visualization of FLAG-tagged proteins, macrophages were transfected and fixed as described above. Macrophages were permeabilized with 0.25% Triton X-100. After blocking with 10% bovine serum albumin in PBS, FLAG-tagged proteins were stained using a monoclonal antibody recognizing FLAG (1:1000; catalog no. F1804, Sigma-Aldrich) at 37°C for 2 hours. Samples were mounted on glass microscopic slides and analyzed as described above.

Electrophoretic mobility shift assay

NF-κB activity was assessed as previously described (51). Briefly, nuclear extracts were prepared and normalized for protein content. The NF-κB–specific oligonucleotides containing two tandemly arranged NF-κB binding sites of the HIV-1 long terminal repeat enhancer (5′-ATCAGGGACTTTCCGCTGGGGACTTTCCG-3′) were end-labeled with [γ-32P]adenosine 5´-triphosphate using polynucleotide kinase (Roche). EMSA assays were performed by incubating 5 μg of nuclear extracts with 500 ng of poly(I:C) (Roche) in binding buffer [5 mM Hepes (pH 7.8), 5 mM MgCl, 50 mM KCl, 5 mM DTT, and 10% glycerol; 20 μl final volume] for 20 min at RT. Then, 2 × 104 cpm end-labeled double-stranded oligonucleotide probe was added, and the reaction mixture was incubated for 7 min. The samples were fractionated by electrophoresis through a 6% polyacrylamide gel in low ionic strength buffer (0.25× tris-borate–EDTA buffer).

Transfection

mRNAs for transfection were generated by in vitro transcription using a HiScribe T7 ARCA mRNA kit (with polyadenylate tailing) (New England Biolabs) and 1.25 mM 5mCTP and pseudo-uridine 5´-triphosphate (Jena Bioscience) according to the instructions of the manufacturer. 5′ triphosphates were removed by antarctic phosphatase (New England BioLabs) treatment. The MEGAclear Transcription Clean-Up Kit (Qiagen) was used for purification of mRNA.

FLAG-tagged NEMO constructs [Addgene catalog no. 27268 for NEMOC54/347A (33) and the corresponding wild-type control] already contained a T7 promotor sequence in correct orientation and thus were simply linearized with Xba I before being used as DNA template. The T7 promotor sequence was added to FLAG-tagged IKKβ constructs [Addgene catalog no. 11103 for wild-type IKKβ and catalog no. 11104 for caIKKβ (IKK-2S177/181E) (36)] by polymerase chain reaction (PCR) using 5′-GAAATTAATACGACTCACTATAGGGTTGA-TCTACCATGGACTACAAAGACG-3′ as the forward primer and 5′-GAGGAAGCGAGAGCT-CCATCTG-3′ as the reverse primer. The T7 promotor sequence was added to the Cre recombinase sequence from a pPGK-Cre plasmid (52) by PCR using 5′-GAAATTAATACGAC-TCACTATAGGGGCAGCCGCCACCATGTCCAATTTACTGACCGTAC-3′ as the forward primer and 5′-CTAATCGCCATCTTCCAGCAGGC-3′ as the reverse primer.

Peritoneal macrophages and MEFs were transfected with 200 ng of mRNA, and BMDMs were transfected with 100 ng of mRNA complexed to jetMESSENGER (Polyplus-transfection) in accordance with the manufacturer’s instructions for 6 hours (NEMO and IKKß mRNA) or 48 hours (Cre mRNA).

Mitochondrial staining

For fluorescence microscopic analysis of the mitochondrial network of peritoneal macrophages, mitochondria were stained with the mitochondrial membrane potential-sensitive dye MitoTracker Red CMXRos (Thermo Fisher Scientific) at 100 nM for 15 min at 37°C before infection. Where indicated, macrophages were treated with 50 μM CCCP (Sigma-Aldrich) as a positive control for mitochondrial network fragmentation.

Analysis of mitochondrial membrane potential

Peritoneal macrophages were incubated in 50 nM tetramethylrhodamine, ethyl ester (TMRE) (Invitrogen) in HBSS with Ca2+ and Mg2+ for 15 min at 37°C and then infected with L. monocytogenes or treated with 50 μM CCCP. TMRE fluorescence was measured at 60-min intervals using a Tecan Infinite M 1000 microplate reader (Tecan Group).

Statistical analysis

For statistical analysis, data were subjected to unpaired two-tailed Student’s t test. P values of less than 0.05 were considered significant.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/12/568/eaar5926/DC1

Fig. S1. ROS are required for proinflammatory cytokine secretion by macrophages.

Fig. S2. Nox family enzymes, xanthine oxidase, and iNOS do not play a role in infection-induced cytosolic ROS production or cytokine secretion.

Fig. S3. Cytosolic ROS are produced by complex III of the mitochondrial ETC.

Fig. S4. mtROS production after L. monocytogenes infection does not depend on mitochondrial damage, Ca2+ overload, or TNF signaling.

Fig. S5. mtROS promote proinflammatory signaling through the ERK1/2 and NF-κB pathways.

Fig. S6. mtROS do not promote proinflammatory cytokine secretion by oxidative inactivation of phosphatases.

Fig. S7. Expression and transfection rates of NEMOWT and NEMOC54/347A.

Fig. S8. mtROS-mediated disulfide linkage of NEMO is crucial for IKK complex activation in BMDMs.

Fig. S9. mtROS-mediated disulfide linkage of NEMO is crucial for IKK complex activation in MEFs.

Fig. S10. Expression and transfection rates of wild-type IKKß and caIKKβ.

Fig. S11. mtROS are not required for infection-induced IRAK1 degradation or posttranslational modification of NEMO.

REFERENCES AND NOTES

Acknowledgments: We thank T. Gilmore (Boston University, Boston, Massachusetts, USA) for providing wild-type and NEMOC54/347A constructs and M. Pasparakis (University of Cologne, Cologne, Germany) for providing NEMO−/− MEF. We also thank D. Grumme, U. Karow, A. Machova, and S. Schramm for technical assistance and D. Rockhoff, M. Ackermann, L. Fischer, A. Jansen, and P. Brünker for support in animal caretaking. Funding: This work was supported by the Deutsche Forschungsgemeinschaft (DFG) (SFB 670). Author contributions: M.H. conducted and analyzed most of the experiments. A.G. conducted and analyzed immunofluorescence experiments. K.W. conducted and analyzed EMSA experiments. A.F. and A.W. provided technical expertise. O.K., O.U., and M.K. provided expertise and feedback. O.U. and M.K. secured funding. M.S. conceived the study, designed the experiments together with M.H., and wrote the manuscript together with M.H. and M.K. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper or in the Supplementary Materials.
View Abstract

Stay Connected to Science Signaling

Navigate This Article