Antioxidant systems, such as the thioredoxin-1 (Trx1) pathway, ensure cellular redox homeostasis. However, how such systems regulate development and function of myeloid cells is barely understood. Here we show that in contrast to its critical role in T cells, the murine Trx1 system is dispensable for steady-state myeloid-cell hematopoiesis due to their capacity to tap the glutathione/glutaredoxin pathway for DNA biosynthesis. However, the Trx1 pathway instrumentally enables nuclear NF-κB DNA-binding and thereby pro-inflammatory responses in monocytes and dendritic cells. Moreover, independent of this activity, Trx1 is critical for NLRP3 inflammasome activation and IL-1β production in macrophages by detoxifying excessive ROS levels. Notably, we exclude the involvement of the Trx1 inhibitor Txnip as a redox-sensitive ligand of NLRP3 as previously proposed. Together, this study suggests that targeting Trx1 may be exploited to treat inflammatory diseases.
Aerobic organisms must continuously prevent the oxidation of their cellular macromolecules. To achieve this, they have evolved a sophisticated system composed of antioxidant proteins that ensure redox homeostasis in the cell. The two main pathways that make up this cellular redox system are the thioredoxin (Trx) system and the glutathione (GSH) system (Arnér and Holmgren, 2000; Couto et al., 2016). GSH can donate electrons to both glutaredoxins (Grx) and glutathione peroxidases (Gpx), a member of which, namely Gpx4, has been recently shown to prevent ferroptotic cell death in T and B1 cells (Matsushita et al., 2015; Muri et al., 2019a). Both the Trx and GSH systems ultimately rely on the reducing power from NADPH, which in mammals is generated by the alternative oxidation of glucose in the pentose phosphate pathway (PPP) (Stanton, 2012). In turn, the Trx and the GSH/Grx systems provide reducing equivalents that sustain a number of cellular processes crucial for cell function, cell survival, cell proliferation, and redox-regulated signaling pathways. For instance, they fuel the reducing power for DNA synthesis by ribonucleotide reductase (RNR) during proliferation, for scavenging reactive oxygen species (ROS), and for the reduction of protein disulfides back to their reduced state (Camier et al., 2007; Holmgren, 1985; Holmgren, 2000; Muri et al., 2018; Potamitou et al., 2002; Prinz et al., 1997). In the cytosolic Trx1 system, Trx reductase 1 (TrxR1, encoded by the Txnrd1 gene) has the unique capacity to transfer electrons from NADPH to oxidized Trx1 (encoded by the Txn1 gene), thereby keeping Trx1 in its reduced state. Thioredoxin-interacting protein (Txnip) is an additional member of the Trx1 system, which negatively regulates Trx function (Arnér, 2009; Mustacich and Powis, 2000). In the GSH/Grx system, by contrast, glutathione reductase (Gsr) maintains the pool of cellular GSH in its reduced state, which in turn further reduces oxidized Grx (Lu, 2013). To which extent the Trx and the GSH/glutaredoxin systems compensate for each others functions in vivo remains unknown.
Macrophages and dendritic cells (DCs) secrete several inflammatory cytokines to orchestrate immune responses. Upon sensing microbial components via Toll-like receptors (TLR), they utilize the MyD88 adaptor to activate nuclear factor-κB (NF-κB)-dependent transcription of pro-inflammatory cytokines including IL-6 (encoded by the Il6 gene), IL-12p40 (encoded by the Il12b gene), TNF-α (encoded by the Tnfa gene) and IL-1β (encoded by the Il1b gene) (Akira and Takeda, 2004). Secretion of IL-1β, however, needs a second signal required for inflammasome assembly, caspase-1 or −11 activation, processing of the immature IL-1β precursor (pro-IL-1β), and subsequent release of the active and mature form of IL-1β (Martinon et al., 2002). A variety of different stimuli that activate inflammasome have been described in the field, especially for the canonical NLRP3 inflammasome (Broz and Dixit, 2016). Interestingly, cellular redox regulation and ROS production have been described to regulate both NF-κB activity (Morgan and Liu, 2011) and NLRP3 inflammasome function (Tschopp and Schroder, 2010). However, the molecular mechanisms of this redox regulation remain to be defined. In particular, the Trx-inhibitor Txnip has been proposed to activate the NLRP3 inflammasome in response to ROS (Zhou et al., 2010), although these results remain controversial (Masters et al., 2010). Therefore, the mechanism by which redox regulation is linked to NF-κB and inflammasome regulation is not fully resolved yet.
We have previously characterized the roles of the Trx1 and GSH/Grx1 systems in T- and B-cell immunity. Notably, we demonstrated that the Trx1 system is critically required to fuel reducing power for the sustainment of DNA biosynthesis during metabolic reprogramming in T but not in follicular B cells (Muri et al., 2018; Muri et al., 2019b). In the present study, we found that the Trx1 system is dispensable for the steady-state hematopoiesis of myeloid cells (i.e. neutrophils, monocytes, macrophages and DC subsets), which efficiently rearrange their redox system toward the GSH/Grx pathway to fuel proliferation when the Trx1 system is absent. Furthermore, we demonstrated how the Trx1 and Grx systems differentially regulate the inflammatory responses of bone marrow-derived DCs (BMDCs) and macrophages (BMDMs). Specifically, while the first utilize the reducing power of the Trx1 system to allow efficient NF-κB p65 transcription factor binding to its DNA response element, the latter need Trx1-dependent antioxidant functions to enable NLRP3 inflammasome formation and IL-1β release. Importantly, our data exclude a role of Txnip in NLRP3 inflammasome regulation as previously proposed (Zhou et al., 2010). In conclusion, these results suggest that therapeutic intervention aimed at blocking the Trx1 system may be beneficial to treat inflammatory diseases.
To investigate the requirement of the Trx1 system in myeloid cells during development and homeostatic maintenance, we crossed mice carrying tamoxifen (TAM)-inducible Rosa26-CreERT2 with mice carrying loxP-flanked Txnrd1 alleles to generate progeny (Txnrd1fl/fl;Rosa26-CreERT2) in which Txnrd1 is globally deleted upon TAM administration. Cre-mediated deletion in total bone marrow cells and in CD11b+ splenocytes of Txnrd1fl/fl;Rosa26-CreERT2 mice was complete at the genomic DNA and mRNA levels (Figure 1—figure supplement 1). TAM injection into Txnrd1fl/fl;Rosa26-CreERT2 mice completely abolished the development of T cells in the thymus, leading to a massive reduction of CD4+CD8+ double positive, CD4+ and CD8+ single positive thymocytes (Figure 1A), as shown previously (Muri et al., 2018). In striking contrast, numbers of eosinophils, neutrophils, monocytes and DCs were comparable in the bone marrow of TAM-treated Txnrd1fl/fl;Rosa26-CreERT2 and control (Txnrd1fl/fl) mice (Figure 1B and Figure 1—figure supplement 2A), indicating that the Trx1 system is dispensable for the process of hematopoiesis in these cells. Despite their short lifespan, monocytes, neutrophils and eosinophils were also present in the blood with a similar frequency in the absence and presence of Txnrd1 (Figure 1C and Figure 1—figure supplement 2B). Moreover, Txnrd1 deficiency also did not affect total numbers of alveolar macrophages, eosinophils, neutrophils, monocytes and conventional type 1 and 2 DCs (cDC1 and cDC2) in the lungs (Figure 1D and Figure 1—figure supplement 2C). Similarly, these populations were also unchanged in the spleen apart from a reduction in total numbers of cDC2 (Figure 1E and Figure 1—figure supplement 2D). Taken together, these results demonstrate that, in contrast to its critical role in T cells, the Trx1 system is dispensable for the development and the homeostatic maintenance of various types of myeloid-cell populations.
Since distinct myeloid populations were not reduced in the absence of the Trx1 system at the steady-state, we next investigated whether bone marrow lacking Txnrd1 could efficiently reconstitute the hematopoietic compartment of irradiated wild type (WT) hosts in a competitive situation with WT cells. To test this, we generated mixed-bone marrow chimeras by reconstituting the hematopoietic compartment of irradiated C57BL/6 mice (CD45.1+CD45.2+) with an equal ratio of congenically marked donor bone marrow cells from WT (CD45.1+) and TAM-treated Txnrd1fl/fl;Rosa26-CreERT2 (CD45.2+) or Txnrd1fl/fl (CD45.2+) mice as control (Figure 1F). Expectedly, Txnrd1-deficient bone marrow completely failed to refill the T-cell compartment of irradiated recipients (Figure 1G), consistent with our previous findings reporting a crucial role of Trx1 for T-cell proliferation (Muri et al., 2018). Notably, eosinophils, neutrophils, monocytes, DCs and macrophages lacking Txnrd1 were reduced by 30–50% in the bone marrow, blood, spleen and lungs, as compared to WT counterparts, thus indicating a small contribution of Txnrd1 for the expansion of myeloid-cell precursors that could be revealed in a competitive situation with WT cells (Figure 1H,I and Figure 1—figure supplement 3A,B), similar to B-cell development (Figure 1—figure supplement 3C; Muri et al., 2019b). Together, these results confirm that, in contrast to its crucial role during thymic T-cell development, the Trx1 pathway is largely dispensable for the development and the maintenance of myeloid cells.
To investigate whether the GSH/Grx system may compensate for the absence of the Trx1 system, thereby allowing normal development and homeostatic maintenance of myeloid cells, we took advantage of L-buthionine-sulfoximine (BSO), which is known to deplete GSH levels upon oral administration in vivo (Watanabe et al., 2003). Therefore, we first generated mixed-bone marrow chimeras as described above. After reconstitution, mice were injected with TAM to delete the Txnrd1 gene and subsequently treated with BSO over a period of 22 days to deplete GSH (Figure 2A). Expectedly, BSO-treated mice showed a significant reduction in total GSH levels compared to controls in the bone marrow and in the spleen (Figure 2B,C). In keeping with the data presented above, TAM administration led to a partial defect displayed by Txnrd1-deficient blood neutrophils, eosinophils and monocytes in a competitive setting with WT cells (Figure 2D,E and Figure 2—figure supplement 1). Interestingly, however, additional depletion of GSH strikingly reduced Txnrd1-deficient granulocytes within 5 days after oral BSO administration (Figure 2D,E and Figure 2—figure supplement 1). Neutrophils, eosinophils and monocytes lacking the Trx1 system from the bone marrow, lungs and spleen also displayed a similar sensitivity to GSH depletion as observed in the blood (Figure 2F–H). Although DCs were similarly affected, Txnrd1-deficient tissue resident macrophages in the lungs and in the spleen only showed a minor reduction upon BSO administration, consistent with their fetal origin and local persistence by slow turnover during adult life (Figure 2G,H). In keeping with the data presented above, Txnrd1-deficient CD4+CD8+ double positive, CD4+ and CD8+ single positive thymocytes were completely outcompeted by WT cells irrespective of BSO treatment (Figure 2I), consistent with the critical role of the Trx1 system and dispensability of the GSH/Grx system for T-cell proliferation (Muri et al., 2018). We further observed that oral administration of BSO to WT mice did not affect numbers of blood neutrophils and monocytes over time (Figure 2J). Moreover, the numbers of various myeloid populations in the bone marrow, spleen and lungs were not affected by GSH depletion in WT mice (Figure 2K and Figure 2—figure supplement 2), therefore suggesting that the GSH/Grx system is only utilized as a backup system to compensate for the absence of the Trx1 pathway. Overall, these data demonstrate that in striking contrast to T cells, myeloid cells can tap both the Trx and GSH/Grx pathways to sustain thiol-based redox reactions during development and homeostatic maintenance.
Since the Trx1 system was dispensable for steady-state hematopoiesis of myeloid cells, we next investigated whether Txnrd1-deficient cells could also efficiently expand during emergency myelopoiesis, where proliferative stress in the bone marrow and blood neutrophilia are induced by a sublethal dose of LPS (Boettcher et al., 2012). To test this, we intraperitoneally administered LPS twice in a 48 hr interval to TAM-treated Txnrd1fl/fl and Txnrd1fl/fl;Rosa26-CreERT2 mice before sacrificing them for analysis 24 hr later (Figure 3A). While both groups of mice increased the numbers of CD11b+Ly-6G-Ly-6Cintbone marrow precursors upon LPS administration, Txnrd1-deficient mice displayed a limited capacity (Figure 3B,C). Moreover, accumulation of neutrophils in peripheral blood was also impaired in the absence of the Trx1 system (Figure 3D,E). Overall, these results demonstrate that the GSH/Grx pathway does not efficiently compensate for the absence of the Trx1 system during emergency hematopoiesis when massive proliferation occurs due to LPS stress. The observed defect in expansion was therefore comparable to the one shown in competitive chimera settings.
So far, we have shown that the function of the Trx1 system in various myeloid cells unlike in T cells is largely dispensable for maintenance and homeostasis, and that it can be efficiently substituted by the GSH/Grx pathway. Next, we aimed at addressing whether the Trx1 system is important for the immune function of myeloid cells, such as the secretion of pro-inflammatory cytokines. Bone marrow harvested from TAM-treated Txnrd1fl/fl;Rosa26-CreERT2 and Txnrd1fl/fl littermate control mice was differentiated in the presence of granulocyte-macrophage colony-stimulating factor (GM-CSF) to obtain BMDCs in culture. We first confirmed that Txnrd1 deletion was complete at the mRNA level in knockout BMDCs (Figure 4A) and that cell survival was not compromised (Figure 4B). To study inflammatory responses, we stimulated Txnrd1-deficient and -sufficient BMDCs with various TLR ligands, namely CpG (TLR9 agonist), LPS (TLR4 agonist), LTA (TLR2 agonist), R837 (TLR7 agonist) and zymosan (TLR2 and dectin-1 agonist), and subsequently assessed cytokine secretion via ELISA. Interestingly, expression of Txn1 was upregulated upon TLR-driven stimulation, potentially indicating an involvement of the Trx1 pathway in inflammatory responses (Figure 4C). Indeed, while WT BMDCs readily produced pro-inflammatory cytokines including IL-6, IL-12p40 and TNF-α in responses to each of the stimuli, this was strikingly impaired in Txnrd1-deficient BMDCs (Figure 4D,E and Figure 4—figure supplement 1), which was also evident at the transcriptional level (Figure 4F,G). Similarly, the secretion of IL-1β by TLR-primed BMDCs upon ATP, alum (Al(OH)3) or nigericin stimulation was also defective in the absence of Txnrd1 due to impaired transcription of the Il1b gene (Figure 4H,I and Figure 4—figure supplement 2). Together, these data suggest that the Trx1 system positively regulates the transcriptional program triggered by TLR ligands.
Considering the important role of IL-12 in the polarization of T helper type 1 (Th1) cells (Langenkamp et al., 2000), we next wondered whether reduced IL-12 production by Txnrd1-deficient BMDCs would affect their ability to drive Th1 polarization of CD4+ T cells. Indeed, we measured a lower frequency of IFN-γ-producing CD4+ T cells when naïve CD4+ T cells were co-cultured with Txnrd1-deficient BMDCs (Figure 4J) compared to when they were co-cultured with WT controls. Therefore, the absence of the Trx1 system in BMDCs inhibits their capacity to induce pro-inflammatory, IFN-γ-producing Th1 cells.
We next aimed to elucidate the mechanism of impaired pro-inflammatory responses in Txnrd1-deficient BMDCs. Upon binding of microbial ligands, TLR associate with adaptor molecules such as MyD88 to activate the IKK complex and mitogen-activated protein kinases, such as extracellular receptor kinase (Erk), p38 and c-Jun N-terminal kinase (JNK), which in turn influence the activity of the transcription factors NF-κB, AP1 and CREB, thereby ultimately regulating expression of pro-inflammatory cytokines (Akira and Takeda, 2004). Interestingly, we observed intact TLR signaling in Txnrd1-deficient BMDCs upon stimulation with LPS or R837, as assessed by the phosphorylation and consequent degradation of IκB-α and phosphorylation of Erk1/2 (Figure 5A and Figure 5—figure supplement 1), thus supporting the idea that early downstream events upon TLR stimulation are not affected by the absence of the Trx1 system.
Proteolysis of IκB-α results in the release and translocation of NF-κB p65 and p50 to the nucleus, where it induces the transcription of genes encoding pro-inflammatory cytokines (Akira and Takeda, 2004). Since no degradation defects of IκB-α were observed in the absence of Txnrd1, we next analyzed the translocation of NF-κB into the nucleus by microscopy. However, we observed no significant defects in the nuclear translocation of NF-κB p65 between Txnrd1-deficient and sufficient BMDCs (Figure 5B,C). Apart from this, we noticed the presence of cytosolic p65 punctate in the absence of the Trx1 system (Figure 5—figure supplement 2), but the relevance and the potential cellular consequences of this pattern are unknown at present and need further investigations. Overall, the microscopy analysis demonstrates that Txnrd1 deficiency does not affect the nuclear translocation of NF-κB p65, therefore suggesting a downstream defect.
We next hypothesized that Txnrd1 deficiency might affect the binding of NF-κB p65 to its DNA response element. Therefore, we stimulated Txnrd1-deficient and sufficient BMDCs with LPS and performed NF-κB p65 chromatin immunoprecipitation (ChIP) followed by RT-PCR to investigate p65 DNA binding to the promoters of inflammatory genes. In keeping with our hypothesis, we observed reduced NF-κB p65 DNA binding to the promoters of the pro-inflammatory genes Il12b, Il1b and Il6, and of the notorious NF-κB target gene Nfkbia (encoding IκB-α) (Figure 5D). Furthermore, we validated our ChIP results using an ELISA-based method for detecting specific p65 transcription factor DNA binding in nuclear extracts (Figure 5E). Thus, these data indicate that the Trx1 system positively regulates transcription of pro-inflammatory cytokines by promoting the binding of NF-κB to its DNA response element in BMDCs.
As the Trx1 system is an important player in maintaining cellular redox homeostasis (Mustacich and Powis, 2000), we next investigated whether deletion of Txnrd1 would impact ROS levels and consequently cytokine secretion in BMDCs. Despite a significant increase in total cellular ROS levels (Figure 5—figure supplement 3A), the reduction in IL-12p40 secretion by Txnrd1-deficient BMDCs was not restored by the supplementation of a panel of antioxidants (Figure 5—figure supplement 3B,C). Importantly, we verified as a control that the antioxidant catalase scavenged excessive ROS in the absence of the Txnrd1 (Figure 5—figure supplement 3D). Together, these results suggest that the defect in pro-inflammatory cytokine secretion is not a consequence of a general increase of ROS but is due to the compromised Trx1-mediated regulation of the redox status of NF-κB, which interferes with NF-κB p65 binding to the DNA.
The BMDC culture is known to comprise both conventional DCs and monocyte-derived macrophages (Helft et al., 2015) with the latter only possessing the capacity of inflammasome activation and IL-1β release (Erlich et al., 2019). To address the role of the Trx1 system in pro-inflammatory responses of macrophages, we harvested bone marrow from TAM-treated Txnrd1fl/fl;Rosa26-CreERT2 and Txnrd1fl/fl littermate controls and generated BMDMs in culture with macrophage colony-stimulating factor (M-CSF). We first verified how the deletion of Txnrd1 would affect classical (M1) and alternative (M2) macrophage activation (Figure 6A), which are known to mediate inflammatory responses and tissue repair, respectively. Interestingly, we observed increased expression of Txnrd1 and Txn1 in M1 compared to M2 macrophages and in BMDMs stimulated with a variety of TLR ligands (Figure 6B,C and Figure 6—figure supplement 1A), consistent with a possible role in maintaining redox homeostasis in ROS-generating M1 macrophages; however, no difference in expression of the Trx1 system inhibitor Txnip was detected (Figure 6—figure supplement 1B). In keeping with the induction of the Trx1 system in M1 macrophages, we observed lower expression of the M1 markers Nos2, Cd38, Hif1a and Gpr18 in M1-polarized Txnrd1-deficient cells compared to the WT controls (Figure 6D and Figure 6—figure supplement 1C–E). A similar impairment in Nos2 expression was additionally confirmed at the protein level by flow cytometry (Figure 6E). However, Txnrd1 deletion did not affect M2 polarization as measured by the RNA expression of the known M2 markers Fizz1, Arg1 and Ym1 (Figure 6F and Figure 6—figure supplement 1F,G), and protein levels of Relmα, CD301b, PDL2 and CD206 (Figure 6G and Figure 6—figure supplement 1H–J). Importantly, we excluded that defective M1 polarization was a consequence of cell death, since Txnrd1-deficient and sufficient BMDMs underwent cell death to a similar extent (Figure 6H and Figure 6—figure supplement 1K). Taken together, these results demonstrate that the Trx1 system is induced by LPS/IFN-γ stimulation and promotes a M1-macrophage phenotype.
Since M1 macrophages secrete pro-inflammatory cytokines during immune responses, we next stimulated Txnrd1-deficient and sufficient BMDMs with a variety of TLR ligands. To our surprise and in striking contrast to the results obtained with BMDCs, IL-12p40 levels remained unaffected (Figure 6I), while production of IL-6 (Figure 6J) and TNF-α (Figure 6—figure supplement 2) was even increased in Txnrd1-deficient BMDMs compared to controls. Importantly, Txnrd1 gene deletion in knockout cells was confirmed at the mRNA level (Figure 6K). Along the same lines, Txnrd1fl/fl;Rosa26-CreERT2 BMDMs displayed significantly higher Il12b and Il6 expression at the transcriptional level compared to Txnrd1fl/fl control BMDMs (Figure 6L,M) upon stimulation, which may suggest the presence of a compensatory mechanism.
Overlapping activities of the Trx and GSH/Grx systems have been suggested in different biological responses, but their relative contribution in a particular response in distinct cell types is barely understood (Lillig and Holmgren, 2007). Interestingly, we observed that all Grx isoforms were higher expressed in BMDM compared to BMDC cultures (Figure 6N). Considering the fact that Txnrd1 is critical for the transcription of pro-inflammatory cytokines in BMDCs but not in BMDMs as shown above, we next wondered whether the alternative GSH/Grx system compensates for the absence of the Trx1 system in BMDMs. To address this, we treated Txnrd1fl/fl;Rosa26-CreERT2 and Txnrd1fl/fl BMDMs with BSO prior to stimulation with TLR ligands to deplete GSH and thereby interfere with Grx functions. Treatment of Txnrd1fl/fl;Rosa26-CreERT2 BMDMs with BSO drastically reduced IL-12p40 secretion to minimal levels (Figure 6O and Figure 6—figure supplement 3A), without affecting viability (Figure 6—figure supplement 3B). In addition, BSO‐treatment did not affect cytokine secretion by WT BMDMs (Figure 6O and Figure 6—figure supplement 3A), suggesting that the Trx1 system alone is sufficient for sustaining cytokine production.
We next investigated how Txnrd1 deficiency influences DNA binding activity of NF-κB to its response element in BMDMs. In keeping with no decrease in IL-6, IL-12p40 and TNF-α secretion in the absence of the Trx1 system, we observed that NF-κB p65 DNA binding activity was not affected by Txnrd1 deletion (Figure 6P). Interestingly, however, blocking the GSH/Grx pathway by BSO treatment in Txnrd1-deficient BMDMs turned out to completely abrogate the DNA-binding capacity of NF-κB p65, thereby confirming the compensatory activity of the GSH/Grx system in BMDMs (Figure 6P). Overall, here we show that while BMDCs are critically dependent on the Trx1 system for NF-κB-mediated pro-inflammatory cytokine secretion, BMDMs can utilize both the Trx1 and GSH/Grx systems to fuel redox reactions aimed at promoting NF-κB-binding activity and consequent transcription of pro-inflammatory target genes.
An intact signal one in Txnrd1-deficient BMDMs, in contrast to Txnrd1-deficient BMDCs, allowed us to study the quality of signal two and inflammasome activation in the absence of Txnrd1. Interestingly, release of mature IL-1β induced by the NLRP3 activators ATP and alum following priming with various TLR ligands was abrogated in Txnrd1-deficient BMDMs (Figure 7A and Figure 7—figure supplement 1A) as shown using an ELISA assay known to display a much higher sensitivity for mature IL-1β than for its pro-form (Dick et al., 2016). This was additionally confirmed by western blotting showing strikingly reduced release of processed IL-1β upon NLRP3 inflammasome activation, while levels of pro-IL-1β in cell lysates (Figure 7B and Figure 7—figure supplement 1B) and transcription of Il1b mRNA were unaffected in Txnrd1-deficient BMDMs upon TLR priming (Figure 7C). Consistent with a defective NLRP3 inflammasome, the processing of caspase-1 to its active form was also dramatically impaired (Figure 7B and Figure 7—figure supplement 1B). Overall, these data demonstrate that the Trx1 system promotes NLRP3 inflammasome activation and consequently secretion of the mature IL-1β.
Intraperitoneal administration of monosodium urate (MSU) is well known to elicit NLRP3-dependent production of mature IL-1β in mouse peritoneum leading to a massive infiltration of neutrophils (Martinon et al., 2006). In line with the in vitro defects, we found that Txnrd1-deficient mice display an impaired neutrophil influx (Figure 7D) and reduced IL-1β production (Figure 7E) upon MSU injection compared to WT mice. Therefore, these results further confirm that the Trx1 system plays a critical role in NLRP3 inflammasome activation and IL-1β secretion in vivo.
The Trx1 system is a major player in the control of cellular antioxidant responses (Mustacich and Powis, 2000). Therefore, we next hypothesized that Txnrd1 deficiency may influence ROS homeostasis and consequently negatively affect the normal processing of IL-1β by the NLRP3 inflammasome. To test this possibility, we first quantified cellular ROS in the absence of the Trx1 system by CM-H2DCFDA staining and observed that Txnrd1-deficient BMDMs accumulated ROS to a higher extent compared to control cells after treatment with LPS or R837 (Figure 7F,G). Notably, we additionally found that supplementation with a cell-permeable version of catalase, one of the main cellular antioxidants involved in the neutralization of hydrogen peroxide, can restore ATP- and alum-induced IL-1β production in Txnrd1-deficient BMDMs up to WT levels (Figure 7H and Figure 7—figure supplement 2). As a control, we verified that addition of catalase does not affect the expression of the Il1b gene (Figure 7—figure supplement 3). By contrast, other classical antioxidants, such as N-acetyl-L-cysteine (NAC), ascorbic acid (AA), dithiothreitol (DTT), and diphenyleneiodonium (DPI), did not rescue the impaired IL-1β production in the absence of the Trx1 system (Figure 7—figure supplement 4). As ROS are commonly described as inducers of the NLRP3 inflammasome (Tschopp and Schroder, 2010), we next aimed at better understanding why the antioxidant catalase restores IL-1β production in the absence of the Trx1 system. Therefore, we incubated Txnrd1-deficient and -sufficient BMDMs with increasing concentrations of catalase and measured IL-1β secretion and ROS levels upon LPS and ATP stimulation. We observed that only low concentrations of catalase restored defective IL-1β production in Txnrd1-deficient cells (Figure 7I) and concomitantly reduced ROS to levels observed in WT cells (Figure 7J). However, higher concentrations of the antioxidant catalase (described as ‘medium’ and ‘high’ in the graphs) more potently inhibited cellular ROS and consequently IL-1β secretion upon NLRP3 inflammasome activation in both WT and Txnrd1-deficient BMDMs (Figure 7I,J), in line with the general view of ROS as triggers of the NLRP3 inflammasome (Tschopp and Schroder, 2010). Altogether, these results demonstrate that the Trx1 pathway is crucial for IL-1β production in macrophages by prevention of an excessive accumulation of hydrogen peroxide.
Our data suggest that excessive levels of ROS impair inflammasome activation. Contrary to these observations, the Trx1 inhibitor Txnip has been described as a redox-sensitive ligand of the NLRP3 inflammasome that promotes mature IL-1β production in response to ROS (Zhou et al., 2010). Therefore, to reconcile this and comprehensively understand how the Trx1-Txnip system regulates IL-1β release, we generated Txnipfl/fl;Vav1-Cre mice by crossing mice with loxP-flanked Txnip alleles to mice expressing Cre recombinase from the Vav1 promoter, and differentiated in vitro BMDMs as described above. Although Txnip deletion was complete at the mRNA level (Figure 7—figure supplement 5A), Txnip-deficient BMDMs did not show any defects in IL-12p40 (Figure 7—figure supplement 5B) and IL-1β secretion (Figure 7K,L and Figure 7—figure supplement 5C) compared to WT controls. Moreover, administration of MSU into Txnip-deficient and sufficient mice elicited a comparable influx of neutrophils in the peritoneum (Figure 7M), thereby further supporting the dispensability of Txnip in NLRP3 inflammasome activation. Together, these data demonstrate that Trx1 but not Txnip is critical for NLRP3-dependent processing of IL-1β.
In this study, we assessed the role of the Trx1 system in the development, homeostatic maintenance, and inflammatory responses of myeloid cells. We found that the Trx1 pathway is not required for development and maintenance of tissue and blood monocytes, DCs, neutrophils, eosinophils and macrophages due to their flexibility in rearranging their redox system toward the GSH/Grx pathway to fuel cellular thiol-based reactions when the Trx1 system is compromised, in contrast to its critical role for the development of T cells (Muri et al., 2018). However, in a situation of emergency myelopoiesis driven by endotoxin, when proliferation of myeloid precursors bursts (Boettcher et al., 2012), the GSH/Grx pathway was insufficient to fully compensate for the absence of the Trx1 system and fewer neutrophils were found in blood of TAM-treated Txnrd1fl/fl;Rosa26-CreERT2 mice. This may be either due to less efficient donation of electrons to RNR by the GSH/Grx system at the last step of the PPP or due to a delay in mobilization of this backup system. Indeed, we previously reported a delay in engagement of the Grx pathway in germinal center responses of follicular B cells lacking Txnrd1 (Muri et al., 2019b).
Importantly, we have established a critical role of the Trx1 system for pro-inflammatory responses in both DCs and macrophages. Using BMDCs, we showed that Trx1 provides reducing equivalents for the redox-dependent binding of the NF-κB p65 transcription factor to target DNA in DCs. The transcription factor NF-κB serves as a crucial mediator of inflammatory responses by inducing the expression of numerous pro-inflammatory cytokine and chemokine genes among others (Liu et al., 2017). NF-κB regulation is tightly controlled both in the cytoplasm and in the nucleus, and many factors and conditions are known to control NF-κB induction including ROS (Gloire and Piette, 2009; Kabe et al., 2005; Schreck et al., 1991). Interestingly, while redox regulation of NF-κB activity by the Trx1 antioxidant pathway has been proposed, its contribution in inflammation remains controversial because both activating and inhibitory effects have been described (Isakov et al., 2014; Kabe et al., 2005; Lillig and Holmgren, 2007; Morgan and Liu, 2011). Our results demonstrate that the Trx1 system promotes the DNA-binding capacity of NF-κB in the nucleus rather than its activation in the cytoplasm or nuclear translocation. Trx1 promotes NF-κB DNA binding probably by structural modification of a component(s) (e.g. via reduction of disulfide bonds) of the NF-κB binding complex rather than through indirect effects of ROS, since supplementation with different antioxidants did not restore NF-κB activity in Txnrd1-deficient BMDCs.
GM-CSF-driven BMDCs are known to comprise a mixture of conventional DCs, monocyte-derived DCs and macrophages (Helft et al., 2015), with the latter population only capable of inflammasome-dependent IL-1β production (Erlich et al., 2019). Therefore, we aimed to verify our results in BMDMs generated with M-CSF, which are thought to consist of a pure population of macrophages. However, in contrast to BMDCs, we surprisingly found that BMDMs displayed normal NF-κB DNA binding activity and transcription of pro-inflammatory genes in the absence of the Trx1 system and showed that this is due to higher expression of the cellular Grx proteins in BMDMs compared to BMDCs. Indeed, we showed that BMDMs but not BMDCs have the capability to engage the Grx pathway when the Trx1 system is genetically deleted. Thus, Trx1 activates NF-κB transcription, and depending on the expression levels of Grx proteins in different myeloid-cell types, Txnrd1-deficiency may have more or less pronounced outcomes. Myeloid cells are known to generate high amounts of ROS by NADPH oxidases to fight intracellular microbes, which may explain the requirement of robust ROS scavenging systems (Lambeth, 2004).
Inflammasomes are cytoplasmic multiprotein complexes that trigger IL-1β processing and release (Broz and Dixit, 2016). Although Txnip has been suggested as a redox-sensitive ligand of NLRP3 that links ROS generation to NLRP3 inflammasome activation (Zhou et al., 2010), we here showed that NLRP3-dependent IL-1β production in macrophages is independent of Txnip both in vitro and in vivo. In contrast, our study identifies the Txnip-interacting partner Trx1 as a critical regulator of inflammasome-dependent IL-1β release in macrophages. In the absence of the Trx1 system, we observed drastically impaired IL-1β production in response to classical NLRP3 stimuli in vitro and during MSU-induced peritonitis in vivo, while pro-Il1b transcription was unaltered. Interestingly, Txnrd1-deficient BMDMs displayed excessive intracellular ROS levels upon TLR triggering, and decomposition of hydrogen peroxide to water and oxygen with a cell-permeable form of the enzyme catalase restored IL-1β production. Notably, other anti-oxidants lacked this capacity, indicating that local peroxide detoxification by Trx1 is critical for NLRP3 inflammasome activation. Thus, our findings suggest that moderate ROS levels act as triggers for the NLRP3 inflammasome in line with the known role of mitochondrial ROS released from stressed mitochondria (Latz et al., 2013; Zhou et al., 2011), while excessive ROS levels display an opposite effect and inhibit NLRP3 activation. Excessive ROS concentrations may lead to the oxidation of particular components of the inflammasome machinery (e.g. ASC, NLRP3, or caspase-1), with detrimental consequences for their ability to assemble and/or cleave IL-1β. Likely candidates might be oxidations on cysteine (Cys-disulfide; Cys-sulfenic acid; Cys-sulfinic acid; Cys-sulfonic acid) and on methionine (Met-sulfoxide), which are generally caused by oxidizing agents such as ROS. In keeping with this hypothesis, nitrogen oxide (NO) has been reported to induce NLRP3 and caspase-1 nitrosylation and thereby inhibit NLRP3 inflammasome activity (Kim et al., 1998; Mishra et al., 2013). Taken together, ROS must be kept within a restricted window of concentrations to allow inflammasome activation, where both an excess or an undersupply prevent it. A non-linear dose response to a stressor (e.g. ROS) with a certain effect at low doses and the reverse at high doses has been referred to as hormesis (Ristow, 2014; Ristow and Schmeisser, 2011). Our results are in line with an hormetic regulation of NLRP3 inflammasome activity by ROS, which promote its assembly at moderate concentrations, while they inhibit it at low and high concentrations.
While classically activated M1 macrophages are associated with inflammatory responses, alternatively activated M2 macrophages are described as anti-inflammatory, thereby promoting tissue remodeling and resolution of inflammation (Chinetti-Gbaguidi et al., 2015). Interestingly, extensive research has demonstrated that the distinct metabolic phenotypes of these two types of macrophages dictate their function. For instance, M1 macrophages undergo increased rates of glycolysis compared to M2 macrophages, allowing rapid ATP production and sustainment of the PPP in order to generate NADPH for ROS production and de novo nucleotide biosynthesis (Kelly and O'Neill, 2015; O'Neill, 2015; Russell et al., 2019). We found that LPS/IFN-γ but not IL-4 stimulation leads to an induction of the Trx1 system, which is in line with the observed impairment in M1 polarization in the absence of the Trx1 pathway. As discussed above, Trx1 donates electrons to RNR, which catalyzes the rate-limiting step of the PPP, leading to nucleotide biosynthesis (Holmgren and Sengupta, 2010). Therefore, the importance of the Trx1 system in M1 polarization is consistent with the high rates of the PPP in these cells. Why M1 macrophages require such enormous amounts of DNA building blocks despite their low proliferative capacity is unknown, although it has been suggested that they need them to generate different RNA populations, such as microRNAs and long non-coding RNAs (O'Neill et al., 2016). In Txnrd1-deficient M1 macrophages, reduced levels of these regulatory RNAs may impact proper M1 polarization. M2 macrophages, on the other hand, seem to suppress the PPP (Haschemi et al., 2012). Thus, this observation is fully in line with the dispensability of the Trx1 system in fueling the PPP during M2 polarization.
Our results show that the cellular redox systems of BMDCs and BMDMs differ considerably, with each providing valuable information to the mechanism. The defect of NF-κB mediated transcription (also referred to as signal 1 in the inflammasome field) in Txnrd1-deficient BMDCs did not allow to study inflammasome activity due to impaired transcription of pro-IL1β mRNA, which was intact in BMDMs and therefore allowing to unravel the function of Trx1 in NLRP3 activation. Importantly, we confirmed impaired NLRP3 inflammasome activity in vivo by treatment of Txnrd1-deficient mice with MSU. The differential redox regulation in BMDCs and BMDMs may reflect distinct requirements of the Trx1 and GSH/Grx pathways for the inflammatory response of DCs, tissue macrophages and monocyte-derived macrophages in vivo.
In conclusion, our study reveals a key role of the Trx1 system for NF-κB-mediated inflammatory responses as well as NLRP3 inflammasome-mediated IL-1β release. Consequently, these results unveil therapeutic opportunities for treatment of inflammatory diseases with Trx1 inhibitors.
Txnrd1fl/fl mice (Jakupoglu et al., 2005) were provided by Marcus Conrad (Helmholtz Zentrum, Munich, Germany) and were backcrossed for more than eight generations to C57BL/6. To generate Txnrd1fl/fl;Rosa26-CreERT2 mice, Txnrd1fl/fl mice were crossed with Rosa26-CreERT2 (Hameyer et al., 2007) mice. Txnipfl/fl mice (Yoshioka et al., 2007) were purchased from The Jackson Laboratory (Bar Harbor, Maine, USA) and crossed with B6.Cg-Tg(Vav1-Cre)A2Kio/J (Vav1-Cre) mice (de Boer et al., 2003). B6 Ptprca (CD45.1) animals were also purchased from The Jackson Laboratory (Bar Harbor, Maine, USA). About 6–12 week-old age- and sex-matched mice (either female or male) were used for the experiments. Mice were kept in individually ventilated cages under specific pathogen free conditions at the ETH Phenomics Center (EPIC; Zurich, Switzerland). All animal experiments were approved by the local animal ethics committee (Kantonales Veterinärsamt Zürich) and were performed according to local guidelines (TschV, Zurich) and the Swiss animal protection law (TschG). For deletion of the Txnrd1 gene in Txnrd1fl/fl;Rosa26-CreERT2 mice, animals were intraperitoneally injected with 2 mg TAM (Sigma-Aldrich) on two consecutive days and used for experiments at least 10 days later. For in vivo GSH depletion, L-buthionine-(S,R)-sulfoximine (BSO; Sigma-Aldrich) was supplemented in the drinking water at a concentration of 20 mM for the indicated time.
For bone marrow chimeras, C57BL/6 (CD45.1+CD45.2+) recipients were irradiated twice with 4.75 Gy with a 4 hr break in a RS 2000 (Rad Source Technologies Inc, Alpharetta, USA). The following day, mice were reconstituted by intravenous injection of 1−3 × 106 bone marrow cells of the donor mice (1 to 1 mixture of bone marrow from CD45.1+ and CD45.2+ donors). Animals were analyzed 8–10 weeks after reconstitution. Either Txnrd1fl/fl;Rosa26-CreERT2 (and Txnrd1fl/fl as control) donor mice or CD45.1+CD45.2+ recipient mice after reconstitution were intraperitoneally treated twice with 2 mg TAM to delete Txnrd1 gene.
Mice were intraperitoneally injected twice with 35 μg LPS (Ultrapure E. coli 0111:B4, InvivoGen) in a 48 hr interval and were sacrificed for analysis 24 hr later. Bone marrow and blood were then taken and processed for analysis.
Mice were intraperitoneally injected with 1.8 mg MSU crystals (InvivoGen) dissolved in 0.2 ml sterile PBS. After 8 hr, mice were euthanized by CO2 exposure, and peritoneal cavities were washed with 10 ml PBS + 2 mM EDTA (Sigma-Aldrich). Neutrophil infiltration in peritoneal lavage fluids was subsequently assessed by CD11b and Ly-6G flow cytometry staining, whereas concentration of IL-1β was determined by ELISA.
Mice were sacrificed by an intraperitoneal overdose of sodium pentobarbital. Organs were removed and then processed according to the following procedure. Lungs and spleens were minced and then digested for 45 min at 37°C in IMDM medium (Life Technologies) containing 2 mg/ml of type IV collagenase (Worthington) and 0.02 mg/ml DNaseI (Sigma). All other organs were directly disrupted and passed through a 70 μm cell strainer (Corning). Bone marrow cells were flushed from femurs and tibia, and then directly passed through the 70 μm cell strainer. Leukocytes from the liver were isolated by using Percoll gradient centrifugation (GE Healthcare). ACK buffer was used for erythrocyte lysis for all organs.
Bone-marrow cells from femur and tibia of sex-matched 6–12 week-old mice were differentiated into BMDCs in RPMI-1640 medium (Gibco) supplemented with GM-CSF (supernatant from X63-GMCSF cell line), 2 mM L-glutamine (GE Healthcare), 10 mM HEPES (Lonza), 100 U/mL penicillin, 100 μg/mL streptomycin (Gibco), 10% FCS (Gibco). Fresh medium was added on day 3 and day 6 of culture, and non-adherent cells were harvested and used in experiments on day 7 of culture. BMDMs were cultured analogously, but with medium supplemented with 20 ng/ml recombinant M-CSF (PeproTech). Adherent BMDMs detached by washing plates with cold PBS + 2 mM EDTA (Sigma-Aldrich) were used in experiments. For differentiation of BMDMs toward M1 and M2, cells were polarized in the presence of ultra-pure LPS from Escherichia coli O111:B4 (20 ng/ml; InvivoGen) + IFN-γ (50 ng/ml; PeproTech) and IL-4 (20 ng/ml; PeproTech), respectively. For analysis of M1/M2 markers at the RNA and protein levels, BMDMs were polarized for 24 and 48 hr, respectively. For BMDC/BMDM stimulation, cells were generally primed with R837 (Tocris Bioscience), CpG (InvivoGen), Ultra-pure LPS from Escherichia coli O111:B4 (InvivoGen), lipoteichoic acid from Staphylococcus aureus (InvivoGen), and zymosan A (Sigma-Aldrich). Cells were generally primed with the indicated TLR ligands for 3 hr before stimulation for 4 hr with 200 μg/ml alum (Al(OH)3; SERVA Electrophoresis), or for 6 hr before stimulation for 1 hr with 2 mM adenosine 5’-triphosphate disodium salt (ATP; Sigma-Aldrich) or 5 μM nigericin sodium salt (Sigma-Aldrich) or as indicated in each figure legend. Other compounds used were applied overnight and consisted of the following: L-ascorbic acid (Sigma-Aldrich), L-buthionine-sulfoximine (BSO; Sigma-Aldrich), catalase-polyethylene glycol (catalase-PEG; Sigma-Aldrich), diphenyleneiodonium chloride (DPI; Sigma-Aldrich), DL-dithiothreitol (DTT; Sigma-Aldrich), and N-acetyl-L-Cysteine (NAC; Sigma-Aldrich). The concentrations at which all the compounds were used are indicated in each figure legend.
5 × 103 Txnrd1-deficient and sufficient BMDCs were co-cultured with GP61-80-specific, MACS-enriched naïve CD4+ T cells (25 × 103/well) in the presence of the indicated doses of GP61-80 peptide (GLNGPDIYKGVYQFKSVEFD, I-Ab-restricted). Co-cultures were performed in IMDM + GlutaMAX, 10% FCS, 100 U/ml penicillin, 100 μg/ml streptomycin, 50 μM β-mercaptoethanol (all Gibco). After 4 days, CD4+ T cells were analyzed for the intracellular expression of IFN-γ upon restimulation with phorbol 12-myristate 13-acetate (PMA; 10−7 mol/L; Sigma) + ionomycin (1 μg/ml; Sigma) in the presence of monensin (2 μg/ml; Sigma) for 4 hr at 37°C.
For dead cell exclusion, cells were stained with either eFluor 780 (eBioscience) or with the Zombie Aqua or Red Fixable Viability kits (Biolegend). Prior to surface staining with antibodies, Fc gamma receptors were blocked by incubating cells with anti-CD16/CD32 antibody (2.4G2, homemade). Antibodies for extracellular stains were then incubated with cells for 15 min in FACS buffer (PBS + 2% FCS). When intracellular staining was required (for IFN-γ, Relmα, CD206 and Nos2 stainings), cells were additionally fixed with 4% formalin, permeabilized with PBS supplemented with 2% FCS and 0.5% saponin (permeabilization buffer), and intracellularly stained for 30 min in permeabilization buffer. The anti-Relmα antibody was detected with an anti-rabbit secondary antibody labeled with the FITC fluorophore. To analyze cell death, cells were stained with Annexin-V-APC (BD Bioscience) and eFluor 780 (eBioscience) in Annexin-V binding buffer. Total cellular ROS were quantified by CM-H2DCFDA staining (Life Technologies) and measuring fluorescein signal after 30 min recovery in supplemented medium. Cells were acquired on LSRFortessa (BD Bioscience), or sorted on FACSAria III (BD Bioscience). Data were analyzed in FlowJo software (Tree Star). All antibodies and staining reagents with their respective dilutions used in this study can be found in the Key resources table.
To measure the NF-κB p65 transcription factor DNA binding activity, nuclear extracts from LPS-primed BMDCs or BMDMs were first isolated using the Nuclear Extraction Kit (Abcam), and the NF-κB binding activity was subsequently measured using the NF-κB p65 Transcription Factor Assay Kit (Abcam). Both kits were used following the manufacturer’s instructions. Values obtained from the Transcription Factor Assay Kit were normalized for the total protein concentrations in the nuclear extracts, which were determined using the Pierce BCA Protein Assay Kit (Thermo Scientific).
CD11b+ and CD4+ enrichments were achieved by positive selection using a MACS system with microbeads conjugated to monoclonal anti-mouse CD11b and CD4 antibodies (MACS, Miltenyi Biotec), respectively, following the manufacturer’s instructions.
To measure total GSH levels in cell lysates, the Glutathione Assay Kit (Sigma-Aldrich) was used following the manufacturer’s instructions.
Cytokines in supernatants were quantified by sandwich ELISA using the following pairs of capture and detection antibodies: B122 and 13-7112-85 (eBioscience) for IL-1β, C15.6 and C17.8 (eBioscience) for IL-12p40, MP5-20F3 and MP5-32C11 (Biolegend) for IL-6, and G281-2626 (BD Pharmingen) and biotinylated anti-TNF-α polyclonal antibody (13-7341-85; eBioscience) for TNF-α. After the coating and detection steps, alkaline phosphatase (AP)-conjugated streptavidin (SouthernBiotech) was provided, and the alkaline phosphatase p-nitrophenyl phosphate (pNPP; Sigma-Aldrich) substrate was subsequently added to each well. The plates were finally read at 405 nm.
Total RNA was extracted using TRIzol (Life Technologies), followed by reverse transcription using GoScript Reverse Transcriptase (Promega) according to the manufacturer’s instructions. Real-time quantitative PCR (RT-PCR) was performed using Brilliant SYBR Green (Stratagene) on an i-Cycler (Bio-Rad Laboratories) according to manufacturer’s protocol. Expression was normalized to the housekeeping gene Tbp for mRNA expression, or to genomic Txnrd1 for addressing DNA recombination efficiency in cells lacking the Txnrd1 gene. The sequences of all used primers are listed in Supplementary file 1.
5 × 106 BMDCs per genotype were fixed with 1% formaldehyde (10 min at 37°C), and cross-linking was stopped by adding glycine to a final concentration of 0.125 M. Cells were washed using ice-cold PBS, resuspended in swelling buffer (25 mM HEPES [pH 8], 1.5 mM MgCl2, 10 mM KCl, 0.5% NP-40) with protease inhibitors for 15 min, spun down and finally resuspended in 400 μl lysis buffer (PBS [1x], 1% SDS, 1% NP-40, 0.5% sodium deoxycholate) with protease inhibitors for 10 min. Sonication was subsequently performed using the Bioruptor (Diagenode; 30 cycles of 30 s on, 30 s off) at 4°C to shear chromatin into ca. 200–300 base-pair DNA fragments. Sonicated samples were diluted with 1.6 ml of lysis buffer without SDS (PBS [1x], 1% NP-40, 0.5% sodium deoxycholate). 40 μg sonicated chromatin in 1 ml volume was precleared by adding 50 μl of dynabeads (Thermo Scientific) and incubation on a wheel for 2 hr at 4°C. At this stage, 10% of chromatin was taken as input DNA. Supernatant was then collected, 3 μg of p65 antibody (Biolegend) was added, and samples were incubated overnight at 4°C on a wheel. Chromatin-antibody complexes were then retrieved by adding 50 μl of dynabeads (Thermo Scientific) and incubation for 4 hr at 4°C on a wheel. Beads were washed once with ice-cold low-salt wash buffer (16.7 mM Tris-HCl [pH 8], 0.167 M NaCl, 0.1% SDS, 1% Triton-X), once with ice-cold high-salt wash buffer (16.7 mM Tris-HCl [pH 8], 0.5 M NaCl, 0.1% SDS, 1% Triton-X), twice using LiCl wash buffer (0.25 M LiCl, 10 mM Tris-HCl [pH 8], 1 mM EDTA [pH 8], 0.5% sodium deoxycholate, 0.5% NP40), and once with TE buffer (10 mM Tris-HCl [pH 8], 5 mM EDTA [pH 8]). DNA was then eluted in elution buffer (1% SDS, 100 mM NaHCO3) by gentle shaking for 30 min at 37°C, and cross-links were reversed by overnight incubation at 65°C. After RNase A and proteinase K treatment, DNA was purified via phenol/chloroform extraction and ethanol precipitation.
Cells were lysed on ice with RIPA buffer (20 mM Tris-HCl, pH 7.5, 150 mM NaCl, 5 mM EDTA, 1 mM Na3VO4, 1% Triton X-100, supplemented with protease inhibitor [Sigma-Aldrich] and phosphatase inhibitor [Sigma-Aldrich]). Cell debris was then removed by spinning for 10 minutes at 4°C. Protein concentrations were determined using the Pierce BCA Protein Assay Kit (Thermo Scientific). Whole cell extracts (30 μg of proteins) were fractionated by SDS-PAGE and transferred to a polyvinylidene difluoride (PVDF) membrane using a transfer apparatus according to manufacturer’s instructions (Bio-Rad). Membranes were blocked with 4% nonfat milk in TBST (50 mM Tris, pH 8.0, 150 mM NaCl, 0.1% Tween20) for 45 minutes, washed and incubated with primary antibodies (1:1000 in TBST with 4% BSA) at 4°C for 12 hours. After washing, membranes were incubated with a 1:2000 dilution (in TBST with 4% nonfat milk) of horseradish peroxidase-conjugated anti-rabbit or anti-mouse antibodies for 1 hour. Blots were washed with TBST three times and developed with the ECL system (Thermo Scientific) according to manufacturer’s instructions. All antibodies for immunoblotting used in this study can be found in the Key resources table.
BMDCs were stimulated with LPS (400 ng/ml) on glass coverslips for the indicated times. After stimulation, cells were fixed in 2% paraformaldehyde and subsequently permeabilized in 0.25% Triton X-100 in PBS. For staining, cells were incubated with anti-NF-κB p65 antibody (Santa Cruz) in 0.1% Triton X-100 containing 2% goat serum in PBS. Secondary antibodies were anti-rabbit IgG, ads-FITC (Southern Biotech). During incubation with the secondary antibody, cells were additionally stained for DNA with DAPI (Sigma-Aldrich) and for actin with phalloidin-647 (Thermo Scientific). The coverslips were mounted on glass slides with Vectashield mounting medium (AdipoGen) for immunofluorescence analysis. Microscopy was performed using an oil immersion objective (UPlanSApo, Olympus; Magnification 100×, NA = 1.4) on a DeltaVision microscope system coupled to a sCMOS camera (pco.edge 5.5, PCO). For quantification of NF-κB nuclear translocation, DAPI and actin stainings were used to create masks for nuclei and whole cell. NF-κB signal intensity within these masks were separately quantified.
Data were analyzed using either a Student’s t test (two-tailed, unpaired), or one-way ANOVA followed by either Tukey’s or Dunnett’s corrections, or a two-way ANOVA adjusted by Bonferroni's multiple comparison test. The data are represented as mean + standard deviation. The method of statistical evaluation and the significance levels are also described in each figure legend.
Focus on mammalian thioredoxin reductases--important selenoproteins with versatile functionsBiochimica Et Biophysica Acta (BBA) - General Subjects 1790:495–526.https://doi.org/10.1016/j.bbagen.2009.01.014
Physiological functions of thioredoxin and thioredoxin reductaseEuropean Journal of Biochemistry 267:6102–6109.https://doi.org/10.1046/j.1432-1327.2000.01701.x
Cutting edge: lps-induced emergency myelopoiesis depends on TLR4-expressing nonhematopoietic cellsThe Journal of Immunology 188:5824–5828.https://doi.org/10.4049/jimmunol.1103253
Inflammasomes: mechanism of assembly, regulation and signallingNature Reviews Immunology 16:407–420.https://doi.org/10.1038/nri.2016.58
Visualization of ribonucleotide reductase catalytic oxidation establishes thioredoxins as its major reductants in yeastFree Radical Biology and Medicine 42:1008–1016.https://doi.org/10.1016/j.freeradbiomed.2006.12.027
The role of glutathione reductase and related enzymes on cellular redox homoeostasis networkFree Radical Biology and Medicine 95:27–42.https://doi.org/10.1016/j.freeradbiomed.2016.02.028
Transgenic mice with hematopoietic and lymphoid specific expression of creEuropean Journal of Immunology 33:314–325.https://doi.org/10.1002/immu.200310005
Redox regulation of nuclear post-translational modifications during NF-kappaB activationAntioxidants & Redox Signaling 11:2209–2222.https://doi.org/10.1089/ars.2009.2463
ThioredoxinAnnual Review of Biochemistry 54:237–271.https://doi.org/10.1146/annurev.bi.54.070185.001321
Antioxidant function of thioredoxin and glutaredoxin systemsAntioxidants & Redox Signaling 2:811–820.https://doi.org/10.1089/ars.2000.2.4-811
The use of thiols by ribonucleotide reductaseFree Radical Biology and Medicine 49:1617–1628.https://doi.org/10.1016/j.freeradbiomed.2010.09.005
Suppression of the pro-inflammatory NLRP3/interleukin-1β pathway in macrophages by the thioredoxin reductase inhibitor auranofinBiochimica Et Biophysica Acta (BBA) - General Subjects 1840:3153–3161.https://doi.org/10.1016/j.bbagen.2014.07.012
Cytoplasmic thioredoxin reductase is essential for embryogenesis but dispensable for cardiac developmentMolecular and Cellular Biology 25:1980–1988.https://doi.org/10.1128/MCB.25.5.1980-1988.2005
Redox regulation of NF-kappaB activation: distinct redox regulation between the cytoplasm and the nucleusAntioxidants & Redox Signaling 7:395–403.https://doi.org/10.1089/ars.2005.7.395
Thioredoxin and related molecules--from biology to health and diseaseAntioxidants & Redox Signaling 9:25–47.https://doi.org/10.1089/ars.2007.9.25
T cell lipid peroxidation induces ferroptosis and prevents immunity to infectionThe Journal of Experimental Medicine 212:555–568.https://doi.org/10.1084/jem.20140857
Crosstalk of reactive oxygen species and NF-κB signalingCell Research 21:103–115.https://doi.org/10.1038/cr.2010.178
The thioredoxin-1 and glutathione/glutaredoxin-1 systems redundantly fuel murine B-cell development and responsesEuropean Journal of Immunology 49:709–723.https://doi.org/10.1002/eji.201848044
Protein levels of Escherichia coli thioredoxins and glutaredoxins and their relation to null mutants, growth phase, and functionJournal of Biological Chemistry 277:18561–18567.https://doi.org/10.1074/jbc.M201225200
The role of the thioredoxin and glutaredoxin pathways in reducing protein disulfide bonds in the Escherichia coli cytoplasmJournal of Biological Chemistry 272:15661–15667.https://doi.org/10.1074/jbc.272.25.15661
Extending life span by increasing oxidative stressFree Radical Biology and Medicine 51:327–336.https://doi.org/10.1016/j.freeradbiomed.2011.05.010
Immunometabolism at the interface between macrophages and pathogensNature Reviews Immunology 19:291–304.https://doi.org/10.1038/s41577-019-0124-9
NLRP3 inflammasome activation: the convergence of multiple signalling pathways on ROS production?Nature Reviews Immunology 10:210–215.https://doi.org/10.1038/nri2725
A novel model of continuous depletion of glutathione in mice treated with L-buthionine (S,R)-sulfoximineThe Journal of Toxicological Sciences 28:455–469.https://doi.org/10.2131/jts.28.455
Thioredoxin-interacting protein links oxidative stress to inflammasome activationNature Immunology 11:136–140.https://doi.org/10.1038/ni.1831
Tiffany HorngReviewing Editor; ShanghaiTech University, China
Satyajit RathSenior Editor; Indian Institute of Science Education and Research (IISER), India
In the interests of transparency, eLife publishes the most substantive revision requests and the accompanying author responses.
Antioxidant systems maintain cellular redox homeostasis, and this study showed that TrxR1, a key enzyme in the Thioredoxin system, influences emergency myelopoiesis, inflammatory gene induction in dendritic cells, and NLRP3 inflammasome activation. The findings indicated that TrxR1 activity modulates an optimal level of cellular ROS that is needed for normal myeloid cell functionality.
Decision letter after peer review:
Thank you for submitting your article "Thioredoxin-1 differentially promotes NF-κB target DNA binding and NLRP3 inflammasome activation independently of Txnip" for consideration by eLife. Your article has been reviewed by three peer reviewers, one of whom is a member of our Board of Reviewing Editors, and the evaluation has been overseen by Satyajit Rath as the Senior Editor. The reviewers have opted to remain anonymous.
The reviewers have discussed the reviews with one another and the Reviewing Editor has drafted this decision to help you prepare a revised submission.
Antioxidant systems maintain cellular redox homeostasis, yet how this affects myeloid cell biology is not well understood. Here Kopek et al. examine the effects of genetic deletion of Txnrd1, a key enzyme in the Thioredoxin system, on myeloid cell development/homeostasis, polarization, and effector functions.
The authors found that Txnrd1 deletion had little effect on steady state myeloid cell numbers, but led to significant reduction in a competitive setting with WT counterparts and during emergency myelopoiesis.
Upon exposure to Toll-like receptor agonists, Txnrd1-deficient bone marrow derived dendritic cells (BMDCs) displayed reduced induction of Il12p40, Il1b, and Il6, which was attributed to reduced p65 DNA binding. However, Txnrd1-deficient bone marrow derived macrophages (BMDMs) compensated by upregulating enzymes of the GSH/Grx system leading to no reduction of inflammatory gene induction.
Finally, the authors examined NLRP3 inflammasome activation in BMDMs. The authors found that Txnrd1 deficiency led to decreased NLRP3 inflammasome activation, which they attributed to increased ROS production, since treatment with cell permeable catalase rescued the defect.
1) The authors should obtain more support for their conclusion that excessive levels of ROS impair NLRP3 inflammasome activation given that many previous studies have reached opposite conclusions.
• The authors are advised to more definitely examine NLRP3 inflammasome activation by interrogating IL-1β and caspase-1 processing by western blot.
• The authors need to offer some explanation, ideally supported by experimental evidence, for the discrepant role of ROS in inflammasome activation in their data versus the literature. For example, is it possible to gradually titrate the amount of catalase added to either WT or Txnrd1 deficient BMDMs and to see dose dependent effects on DCFDA staining concomitant with different effects on inflammasome activation? In Figure 7H-K, can the authors confirm that addition of catalase does not affect Il1b gene induction? At the very least, the authors should propose experiments that could be done in future studies to confirm the differing effects of excessive versus moderate ROS levels on inflammasome activation.
2) Clarification is needed on a few points in the second part of the story addressing the role of the Trx system in NF-κB activation.
• The authors should confirm the defect in p65 DNA binding using p65 ChIP at the promoters of the inflammatory genes.
• What is the proposed mechanism for why Txnrd1 deficiency is more specific for NF-κB regulation than general oxidation? A control that demonstrates the general efficacy of the reducing agents seems warranted.
• It is not clear why nuclear NF-κB is unchanged (Figure 5B) but DNA binding is affected (Figure 5E) in the KO.
3) In the first part of the story, the authors conclude that reduced myeloid cell numbers in the absence of the antioxidant system(s) is due to defects in proliferation, but data to support this is lacking. The authors are advised to describe the data as defects in "maintenance" and "homeostasis".https://doi.org/10.7554/eLife.53627.sa1
Essential revisions:1) The authors should obtain more support for their conclusion that excessive levels of ROS impair NLRP3 inflammasome activation given that many previous studies have reached opposite conclusions.• The authors are advised to more definitely examine NLRP3 inflammasome activation by interrogating IL-1β and caspase-1 processing by western blot.
We thank the reviewers for this important comment. As suggested, we have now performed western blotting for IL-1β and caspase-1 in both the cell lysate and SN upon stimulation with LPS/R837 and ATP/Alum to interrogate NLRP3 inflammasome activation. Consistent with the impaired secretion of IL-1β in the absence of the Trx1 system measured by ELISA, we found reduced levels of both processed IL-1β and caspase-1 by western blot (see new Figure 7C and Figure 7—figure supplement 1B). Thus, these data are in line with a defect in the activation of the NLRP3 inflammasome when the Txnrd1 gene is deleted.
• The authors need to offer some explanation, ideally supported by experimental evidence, for the discrepant role of ROS in inflammasome activation in their data versus the literature. For example, is it possible to gradually titrate the amount of catalase added to either WT or Txnrd1 deficient BMDMs and to see dose dependent effects on DCFDA staining concomitant with different effects on inflammasome activation?
We agree with the reviewers that this is a critical point. First of all, it is important to mention that, as already described in the Discussion section, nitrogen oxide (a key nitrogen reactive species [RNS]) can nitrosylate NLRP3 and caspase-1, thus leading to NLRP3 inflammasome inhibition (see for example PMID: 9780184 and PMID: 23160153). One could imagine a similar mechanism with ROS and in particular with H2O2, where excessive level of this particular ROS may result in the oxidation/peroxidation of inflammasome components (such as ASC, NLRP3, and caspase-1 among others) or other inflammasome regulators, thus compromising their proper functionality. This would be consistent with the defect that we have observed in IL-1β and caspase-1 processing. We believe that our data/conclusions are not opposite to other studies in the field as we report that only excessive levels of H2O2 due to Txnrd1 deletion impair inflammasome function. To further strengthen this model, we followed the suggestions of the reviewers and titrated the amount of catalase to see dose dependent effects on DCFDA staining and IL-1β secretion (see new Figures 7I and 7J). Interestingly, we now show that only low levels of catalase could rescue IL-1β production in the absence on Txnrd1 by bringing ROS to the levels observed in WT cells. By contrast, increased catalase concentrations, which led to a more drastic inhibition of ROS, turned out to display the opposite effects and thus inhibit inflammasome formation in both WT and KO BMDMs. This observation is in line with the many reports in the literature that describe ROS as inducers of NLRP3 inflammasome. In conclusion, our data confirms that ROS are involved in NLRP3 inflammasome activation; however, they also highlight that an excess of ROS (for example due to the genetic ablation of the key antioxidant pathway Trx1) impairs NLRP3 activation. As mentioned, the underlying molecular mechanism is unknown and currently under investigation, but it could involve a general oxidation of inflammasome components which compromises their proper functionality.
In Figure 7H-K, can the authors confirm that addition of catalase does not affect Il1b gene induction?
We are thankful for the reviewers’ thoughtful suggestion. First of all, it is important to mention here that the levels of IL-1β mRNA in KO BMDMs, as we have already shown in our initial submission (see new Figure 7B), are higher compared to WT cells, despite a large impairment in IL-1β secretion (probably due to compensation mechanisms). Importantly, however, by performing the experiment suggested by the reviewers, we have observed that catalase treatment does not affect the transcriptional levels of IL-1β (after either LPS or CpG priming) in both WT and KO cells (see new Figure 7—figure supplement 3). Thus, these data suggest that catalase treatment promotes inflammasome activation rather than affecting priming in KO BMDMs.
At the very least, the authors should propose experiments that could be done in future studies to confirm the differing effects of excessive versus moderate ROS levels on inflammasome activation.
To date, various cellular and molecular events have been proposed as trigger for the NLRP3 inflammasome, such as potassium efflux, lysosomal destabilization and ROS among others. Although ROS are one of the first NLRP3 activators that have been identified, the exact mechanism by which the NLRP3 inflammasome senses ROS remains still unknown and is beyond the aim of this study. A clear example for the huge controversy of ROS in the inflammasome field is given by the role of the key antioxidant pathway Nrf2. In particular, Nrf2 has been associated with both a negative (e.g. PMID 27211851; PMID 29590092) and a positive (e.g. PMID: 24798340; PMID: 21484785) regulation of IL-1β responses. Thus, we strongly believe that understanding how ROS affect NLRP3 inflammasome activation and especially how various concentrations of ROS have different effects might be not so simple. As we have already mentioned before, we believe and our data suggest that moderate/medium ROS levels trigger or participate in the activation of the NLRP3 inflammasome, while excessive/damaging ROS concentrations are inhibitory because they may oxidize and compromise the functionality of factors that are needed for inflammasome formation. To test this in future studies, we are currently planning to determine protein oxidations caused by Txnrd1 deficiency using unbiased mass spectrometry. Likely candidates might be oxidations on cysteine (Cys-disulfide; Cys-sulfenic acid; Cys-sulfinic acid; Cys-sulfonic acid) and on methionine (Met-sulfoxide), which are generally caused by oxidizing agents such as ROS (this has now been added to the Discussion section). One should for example analyze the most significant differences in oxidation patterns between WT and KO cells, focusing on known inflammasome components and other regulators. Using this strategy, we actually aim to understand whether genetic ablation of a key antioxidant pathway leading to high and detrimental ROS levels shifts the redox balance and leads to oxidation of proteins involved in NLRP3 inflammasome function.
2) Clarification is needed on a few points in the second part of the story addressing the role of the Trx system in NF-κB activation.• The authors should confirm the defect in p65 DNA binding using p65 ChIP at the promoters of the inflammatory genes.
We agree with the reviewers that validating the results obtained with the ELISA-based method will strengthen the conclusion. Indeed, as suggested, we have now performed p65 ChIP-qPCR at the promoters of the inflammatory genes (the details about the utilized procedure can be found in the new method section). We tested four different genes (IL-12p40, IL-6, IL-1β and Nfkbia), which are all well-known NF-κB target genes, and we performed qPCR utilizing two different primer sets for each gene. The first primer pair (named as “promoter”) allowed us to amplify a fragment close to the NF-κB binding sites at the promoter region, whereas the second primer pair (named as “Ctr primers”) was used as a control to amplify a region several kilobases away from the NF-κB binding sites. As depicted in the new Figure 5D, we were able to confirm that Txnrd1-deficient BMDCs display a reduced NF-κB binding to target DNA.
• What is the proposed mechanism for why Txnrd1 deficiency is more specific for NF-κB regulation than general oxidation? A control that demonstrates the general efficacy of the reducing agents seems warranted.
We thank the reviewers for mentioning this important point. Unfortunately, it seems like we were not clear enough in our initial description of the data, which must have resulted in a misunderstanding of the reviewer(s). The reason why we studied the NLRP3 inflammasome phenotype exclusively in BMDMs is that they do not display any defects in IL-1β transcription. In contrast, BMDCs have a drastic impairment in TLR priming, thus rendering the investigation of NLRP3-inflammasome function more complicated. This said, our data do not indicate that “Txnrd1 deficiency is more specific for NF-κB regulation than general oxidation” in BMDCs; indeed, in BMDCs, we observed both a defect in NF-κB binding to target DNA and a general increase of ROS, which is likely leading to a defective NLRP3 activation as we have shown for BMDMs. Originally, we omitted these data from our manuscript to reduce complexity and facilitate readability and understanding of the story. Unfortunately, it seems like that we reached the opposite and created confusion of the reviewer(s). Therefore, we have now included these data in the new version of the manuscript (see new Figure 5—figure supplement 3). Here, we show that BMDCs also display higher ROS levels in the absence of Txnrd1, and that catalase-treatment, although it inhibits ROS formation, cannot restore cytokine production. Therefore, it looks like both BMDCs and BMDMs require Txnrd1 to inhibit ROS and allow inflammasome formation; by contrast, the NF-κB phenotype is exclusively observed in BMDCs, as BMDMs express high levels of glutaredoxins which can compensate for the absence of the Trx1 system. It is additionally important to mention that the binding of NF-κB to the target DNA is not ROS dependent (as catalase, which blocks ROS production, does not restore cytokine production) but specifically requires the Trx1 protein (or Grx as a backup), for example by providing reducing power for the regulation of NF-κB redox state (e.g. disulfide bond formation). This is actually reminiscent of what we have previously observed in T cells, where the enzymatic activity of Trx1, but not its general antioxidant function, was needed to deliver electrons to a specific protein, namely ribonucleotide reductase for the synthesis of dNTPs.
• It is not clear why nuclear NF-κB is unchanged (Figure 5B) but DNA binding is affected (Figure 5E) in the KO.
We thank the reviewers for raising this point and we are sorry for the lack of clarity. NF-κB induced inflammatory responses can be separated into at least three steps including (i) cytoplasmic NF-κB activation, (ii) nuclear translocation, and (iii) DNA binding followed by transcription of target genes, which are controlled by different biochemical mechanisms. We found reduced production of pro-inflammatory cytokines known to be controlled NF-κB in Txnrd1-deficient BMDCs. We showed that cytoplasmic NF-κB activation and nuclear translocation were unaffected indicating that DNA binding, which is differentially regulated, is impaired in the absence of Txnrd1. Indeed, in the first submitted version of the manuscript, we showed that DNA binding is impaired in Txnrd1-deficient BMDCs using a binding assay. In the revised version, as requested by the reviewers, we supported this finding by performing p65-CHIP analysis of NF-κB target genes.
What it actually remains to be answered in future studies would be why and how the p65 DNA binding is affected. In this regard, we believe that Trx1 controls and regulates the redox status of p65 (and possibly of other factors that participate in DNA binding). For instance, Trx1 may provide electrons to reduce disulfide bonds of p65, and this may promote the DNA binding activity of the transcription factor. By using mass spectrometry, one could better analyze and understand how exactly the Trx1 system regulates the redox state of p65.
3) In the first part of the story, the authors conclude that reduced myeloid cell numbers in the absence of the antioxidant system(s) is due to defects in proliferation, but data to support this is lacking. The authors are advised to describe the data as defects in "maintenance" and "homeostasis".
We thank the reviewers for pointing out this aspect that was not fully supported by the data in our manuscript. We have previously described the defect as a “proliferation defect” since this is what we have originally reported in the T cells (where the Trx1 system donates electrons for the synthesis of DNA building blocks during rapid proliferation; PMID: 29749372). Although we believe that a similar mechanism very likely also takes place in myeloid cells, we agree with the reviewers that the data in the current manuscript do not unambiguously prove it. Thus, we have now followed the reviewers’ suggestion and described the data in this manuscript as defects in “maintenance” and “homeostasis”.https://doi.org/10.7554/eLife.53627.sa2
- Manfred Kopf
- Manfred Kopf
The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.
We are grateful for research grants from ETH Zurich (ETH-23-16-2) and SNF (310030B_182829). We thank Marcus Conrad for providing Txnrd1fl/fl mice. We thank Peter Nielsen for designing primers and Madlen Müller for help during the ChIP experiment. We further thank members of the ETH Flow Cytometry Core Facility for cell sorting. We acknowledge the use of the Immgen database as an informative tool for our study.
Animal experimentation: All animal experiments were approved by the local animal ethics committee (Kantonales Veterinärsamt Zürich, licenses 25/2014, ZH054/18, ZH054/19 and ZH134/18), and performed according to local guidelines (TschV, Zurich) and the Swiss animal protection law (TschG).
- Satyajit Rath, Indian Institute of Science Education and Research (IISER), India
- Tiffany Horng, ShanghaiTech University, China
© 2020, Muri et al.
This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.