Macrophages, dendritic cells and other innate immune cells are involved in inflammation and host defense against infection. Metabolic shifts in mitochondrial dynamics may be involved in Toll-like receptor agonist-mediated inflammatory responses and immune cell polarization. However, whether the mitochondrial morphology in myeloid immune cells affects anti-tumor immunity is unclear. Here we show that FAM73b, a mitochondrial outer membrane protein, has a pivotal function in Toll-like receptor-regulated mitochondrial morphology switching from fusion to fission. Switching to mitochondrial fission via ablation of Fam73b (also known as Miga2) promotes IL-12 production. In tumor-associated macrophages, this switch results in T-cell activation and enhances anti-tumor immunity. We also show that the mitochondrial morphology affects Parkin expression and its recruitment to mitochondria. Parkin controls the stability of the downstream CHIP–IRF1 axis through proteolysis. Our findings identify mechanisms associated with mitochondrial dynamics that control anti-tumor immune responses and that are potential targets for cancer immunotherapy.
Mitochondria generate cellular energy that can alter their morphology through biogenesis and the opposing processes of fission and fusion1, 2. Maintenance of dynamic mitochondrial networks has crucial effects on signaling transduction, ATP production, iron–sulfur cluster biogenesis, and calcium buffering3, 4. Mitochondrial morphology switching is also implicated in cell survival, apoptosis, and cellular metabolic homeostasis5, 6. Both the outer membrane (OM) and inner membrane (IM) are involved in mitochondrial fusion and fission. Mitofusin 1 (MFN1) and MFN2 are essential for OM fusion and maintenance of mitochondrial morphology7, 8. Optic atrophy 1 (OPA1) controls IM fusion and protects cells with mitochondrial dysfunction due to Mfn1 deficiencies9,10,11. Two OM proteins, named FAM73a (also known as mitoguardin 1) and FAM73b, are required for mitochondrial fusion, and they function by regulating phospholipid metabolism via mitochondrial phospholipase D (MitoPLD)12. Although the physiological functions of mitochondria are linked to their morphology13, mitochondrial dynamics in immune responses are not clear owing to the embryonic lethality of MFN1/2 double knockout (KO) or OPA1 mutant mice. However, FAM73a and FAM73b KO mice are viable and exhibit only moderately decreased body weight and body fat. Therefore, FAM73a and FAM73b KO mice are suitable models to evaluate the role of mitochondrial dynamics in immune homeostasis and host defense.
Mitochondria have essential functions in both innate and adaptive immunity. Mitochondria are catabolic organelles and are the major source of cellular ATP and ROS, which are important in innate immune responses to cellular damage, stress, and infection14,15,16. Mitochondria also host signaling modulators such as mitochondrial antiviral signaling protein (MAVS) and evolutionarily conserved signaling intermediate in Toll pathway, mitochondrial (ECSIT) to control pattern recognition receptor (PRR)-mediated type I interferon induction and inflammatory responses17,18,19,20,21,22. Additionally, mitochondria-mediated metabolic changes are associated with immune cell polarization, particularly lymphocyte homeostasis and memory T-cell generation23. T-cell differentiation to T helper type 1 (Th1), Th2, and Th17 subpopulations preferentially utilizes glycolysis rather than mitochondrial OXPHOS24, 25, and T regulatory (Treg) cells have distinct metabolic demands, which are dependent on both lipid metabolism and OXPHOS24, 25. Polarization of macrophages also involves different metabolic pathways, with aerobic glycolysis important for M1 macrophages and fatty acid oxidation (FAO)-driven mitochondrial oxidative phosphorylation important for differentiation of M2 macrophages26, 27.
IL-12 family cytokines are mainly produced by myeloid cells, and they control adaptive immune responses, especially T-cell differentiation28. IL-12 p35, IL-12 p40, and IL-23 p19 are proinflammatory cytokines produced by dendritic cells, macrophages and fibroblasts in response to microbial pathogens and tumors29, 30. IL-12 and IL-23 expression is associated with epigenetic modifications31 and various transcription factors, such as c-Rel, IRF5, and IRF130. Genetic evidence indicates that LPS-induced IL-12 p35 expression is reduced in Irf1 −/− macrophages. In vitro data show that monoubiquitinated C terminus of HSC70-interacting protein (CHIP; also known as STUB1) is required for proteolytic degradation of IRF132. Furthermore, CHIP is associated with Parkin, which mediates selective autophagy of depolarized mitochondria33, 34. Therefore, here, we investigate the effect of mitochondrial membrane remodeling on IL-12 production and the activity of the CHIP–IRF1 axis in anti-tumor immune responses.
Here we characterize a proteolysis-dependent mechanism of inflammatory cytokine production regulated by mitochondrial fission. Genetic ablation of the fusion mediator Fam73b in macrophages and dendritic cells promotes TLR-induced IL-12 expression and inhibits IL-10 and IL-23 expression. Macrophage-derived IL-12 promotes anti-tumor T-cell responses in vivo in mouse melanoma and MCA-induced fibrosarcoma models. Myeloid cell but not T cell conditional knockout mice have enhanced Th1 responses. Fam73b or Mfn1/Mfn2 depletion causes severe mitochondrial fragmentation and degrades monoubiquitinated CHIP. Furthermore, mitochondrial fission promotes accumulation and recruitment of Parkin, which directly induces monoubiquitinated CHIP degradation and stabilizes the crucial downstream transcription factor IRF1. Our data highlight an unappreciated role of mitochondrial morphology in macrophage polarization and identify an associated signal transduction network.
Mitochondrial dynamics involved in macrophage polarization
To evaluate whether mitochondrial dynamics are involved in macrophage polarization, we stimulated wild-type (WT) bone marrow-derived macrophages (BMDMs) with the TLR4 ligand lipopolysaccharide (LPS) and examined the mitochondrial morphology. Confocal microscopy revealed that LPS-treated macrophages rapidly and predominantly exhibited punctate mitochondria (Fig. 1a). The intensity of mitochondrial fragmentation was dependent on LPS concentration (Supplementary Fig. 1a). Additionally, the mitochondrial network maintained fission status until 12 h after stimulation (Supplementary Fig. 1b). Ultrastructural analysis using electron microscopy (EM) also indicated that LPS treatment led to small, diverse mitochondria dispersed throughout the cytoplasm (Fig. 1b). Morphometric analysis revealed significantly more mitochondria that occupied a comparable portion of the cellular area (Supplementary Fig. 1c, d). Time-lapse microscopy also showed that mitochondria quickly switched to the fission stage within 2 h (Fig. 1c). We further examined the expression levels of several critical regulators of mitochondrial dynamics12, 35. We found that the canonical fusion mediators Mfn1/Mfn2 and Fam73b were suppressed by LPS stimulation, with reduced phosphorylation of the fission factor Drp1 (Fig. 1d, e). As reported, mitochondrial fission reduces Ca2+ uptake and intramitochondrial Ca2+ diffusion36. Cytosolic Ca2+ rise activates the cytosolic phosphatase calcineurin that normally interacts with Drp1. Calcineurin-dependent dephosphorylation of Drp1 regulates its translocation to mitochondria37. Similar results were obtained when the TLR3 ligand poly(I:C) was used to stimulate cells (Supplementary Fig. 1b, e). M1 macrophage differentiation is enhanced by interferon-γ (IFN-γ) treatment38. In contrast to TLR agonists, IFN-γ alone did not induce extensive fragmentation of mitochondria in BMDMs (Fig. 1f).
IL-4 is known to promote M2 polarization, which is involved in tissue repair and resistance to parasitic helminthes38. Recent research has demonstrated that M2 macrophages enhance the mitochondrial oxygen consumption rate (OCR) and spare respiratory capacity (SRC)27, which suggests that IL-4 might promote mitochondrial fusion. Thus, we treated BMDMs with IL-4 for 12 h and assessed themitochondrial morphology. The data revealed that the mitochondria in M2 macrophages formed elongated tubules, which occupied more cytoplasmic area (Fig. 1g, h, Supplementary Fig. 1c, d). Furthermore, only Fam73b levels were increased compared to other genes, such as Mfn1 and Mfn2 (Fig. 1i, Supplementary Fig. 1f). These results suggest a potentially essential role of Fam73b in mediating mitochondrial morphology during macrophage polarization.
FAM73b negatively regulates anti-tumor innate immunity
To study the in vivo function of FAM73b in the immune system, we employed a gene targeting approach (KO first) to delete the Fam73b gene in mice (Supplementary Fig. 2a). Immunoblot (IB) analysis revealed loss of FAM73b expression in different types of immune cells in germline FAM73b KO mice (Supplementary Fig. 2b). FAM73b KO mice did not show significant abnormalities in thymocyte development, although they had a moderate increase in the frequency of peripheral CD8+ T cells and all memory T-cell populations (Supplementary Fig. 3a, b). The percentage of Treg cells was comparable both in the spleen and inguinal lymph node (iLN) (Supplementary Fig. 3c). Additionally, deletion of Fam73b had no significant effect on the frequency of macrophages, neutrophils or dendritic cells in the bone marrow (BM) and spleen (Spl) (Supplementary Fig. 3d).
Although macrophage development was normal, FAM73b deficiency also triggered a high degree of mitochondrial fragmentation as other cell types12. This fission stage could not be rescued by IL-4 treatment as the WT control (Fig. 2a, b). FAM73b KO BMDMs also contained more mitochondria with a comparable occupied area (Supplementary Fig. 1c, d). Additionally, FAM73b deficiency did not affect mitochondrial biogenesis, and cells had a similar mtDNA level (Supplementary Fig. 1e). Interestingly, IL-4 still promoted mitochondrial fusion in Mfn1/2 −/− mouse embryonic fibroblasts (MEFs) (Fig. 2c). This result suggests that Mfns are dispensable for the mitochondrial morphology switch under polarization stress. Collectively, these data indicate that Fam73b functions as a crucial regulator of mitochondrial dynamics during macrophage polarization.
To further investigate the function of FAM73b in regulating the in vivo immune response, we employed a murine melanoma model in which B16 melanoma cells were inoculated into mice. Fam73b deletion profoundly suppressed tumor growth and increased the survival rate of the tumor-bearing mice (Fig. 2d, e). Consistently, FAM73b KO mice had enhanced IL-12 and reduced IL-10 serum levels (Fig. 2f). Moreover, the T-cell-derived cytokine IFN-γ was also upregulated (Fig. 2g). FACS analysis revealed an increased frequency of IFN-γ-positive CD4+ and CD8+ effector T cells both in the draining lymph node (dLN) and the spleen of tumor-bearing FAM73b KO mice (Fig. 2h).
Because the development of primary tumors involves long term interactions between the immune system and tumors, it still remains unknown how the excessive and enhanced production of IL-12 induced by FAM73b deficiency impacts the development of primary tumors. Therefore, we employed a well-characterized methylcholantrene (MCA)-induced fibrosarcoma model. We treated the WT and FAM73b KO mice with 800 μg of MCA and monitored tumor formation for up to 150 days. FAM73b KO mice clearly developed fibrosarcoma at a significantly lower rate of incidence (Fig. 2i). FAM73b KO mouse serum also displayed a profound increase in IL-12 and IFN-γ serum levels (Fig. 2j).
FAM73b functions in myeloid cells to regulate tumor growth
To delineate the roles of FAM73b in various immune cells, we bred Fam73b-flox mice with mice expressing Cre recombinase driven by the Cd4 promoter or lyz2 promoter to generate T-cell- or myeloid cell-specific KO mice. Conditional KO mice exhibited a significant FAM73b deficiency in T cells (Supplementary Fig. 1g) or macrophages (Supplementary Fig. 1h). In the B16 melanoma model, T-cell-specific KO mice had similar tumor growth and death ratio to those of WT mice (Fig. 3a, b). Consistently, we found that T-cell-specific Fam73b ablation had no effect on T-cell subtype differentiation in vitro (Fig. 3c). RNAseq analysis also demonstrated that FAM73b-deficient naive CD4+ T cells exhibited gene profiles similar to those of WT cells (Fig. 3d, Supplementary Data 1).
Macrophages are the dominant leukocyte population in the inflammatory microenvironment of tumors39. IL-4 from the tumor microenvironment induces cathepsin protease activity in tumor-associated macrophages (TAMs)40. To clarify the role of FAM73b in TAMs, we first evaluated the effect of the tumor microenvironment on the mitochondrial morphology. As shown in Supplementary Fig. 4a, b, FAM73b promoted mitochondria switching to the fusion stage when co-cultured with B16 melanoma cells. This fusion was restored by IL-4 neutralizing antibody. In contrast to T-cell-specific KO mice, myeloid cell-KO mice suppressed tumor growth and maintained the survival rate (Fig. 3e, f). Compared to WT hosts, deficiency of FAM73b in myeloid cells led to an increased frequency of infiltrating CD8+ T cells in the tumors, together with a reduction in TAMs and myeloid derived suppressor cells (MDSCs) (Fig. 3g). However, enhanced CD8+ T-cell recruitment was not due to a reduction of Treg cells (Fig. 3h). The tumors and draining lymph node (dLN) of myeloid cell-KO mice contained an increased frequency of IFN-γ-positive CD4+ and CD8+ effector T cells (Fig. 3i). qRT-PCR assays revealed an increase in Il12a levels specifically in tumor-infiltrating macrophages from myeloid cell-KO mice (Fig. 3j). Similar to macrophages, FAM73b deficiency in dendritic cells also led to tumor growth resistance, coupled with an increased survival rate (Supplementary Fig. 3c, d). Collectively, these results support an unexpected role for mitochondrial fission in innate cell-mediated anti-tumor immunity.
Mitochondrial fission promotes TLR-mediated IL-12 induction
To elucidate the molecular mechanism by which Fam73b regulates innate immunity, we analyzed the majority of the transcriptome changes in mRNA abundance. As shown in Fig. 4a and Supplementary Data 2, we identified 786 differentially expressed genes between WT and FAM73b KO macrophages with LPS stimulation (>2 fold). Without stimulation, different genes were mainly enriched in regulating cell survival and development (Fig. 4b and Supplementary Fig. 5a). Furthermore, FAM73b deficiency significantly enhanced Il12a induction and severely suppressed Il10 and Arg1 when responding to TLR stimulation (Fig. 4b and Supplementary Fig. 5b). Cytokine induction was also confirmed by qRT-PCR analysis (Fig. 4c). Additionally, FAM73b appeared to be a common regulator of Il12a, because FAM73b also suppressed Il12a induction by the polyI:C (TLR3 ligand), R848 (TLR7 ligand), and CpG (TLR9 ligand) (Fig. 4d and Supplementary Fig. 6a, b). Similar results were obtained with ELISAs that detected the secreted cytokines (Fig. 4e). In contrast, FAM73b is required for production of specific markers in M2 polarization (Supplementary Fig. 6c). As seen in macrophages, FAM73b deficiency in bone marrow-derived dendritic cells (BMDCs) also promoted Il12a induction (Fig. 4f). A previous report claimed that FAM73a was also required for mitochondrial fusion12. Just as with FAM73b, FAM73a KO macrophages produced more IL-12 than macrophages from their WT littermates (Supplementary Fig. 6d, e), suggesting that FAM73a and FAM73b heterotypic complexes promote mitochondrial fusion and negatively regulate Il12a induction.
To further test whether mitochondrial fission is involved in Il12a enhancement, WT, and FAM73b KO macrophages were cultured with the fission inhibitor Mdivi-1 before LPS treatment. This drug significantly promoted mitochondrial fusion both in WT and Fam73b macrophages (Supplementary Fig. 6f). Mdivi-1 also erased differences in Il12a and Il10 production (Fig. 4g). We further evaluated the function of other fusion-related molecules, such as Mfn1/Mfn2 and Opa1. Due to the embryonic death of these global KO mice, we assessed the cytokine induction in Mfn1/Mfn2- or Opa1-deleted MEFs. The results indicated that loss of Mfn1/Mfn2 or Opa1 also accelerated Il12a transcription when MEFs were transfected with Lipofectamine-packaged poly(I:C) (Fig. 4h, i). These results demonstrate an unexpected and pivotal role of mitochondrial dynamics in regulating IL-12 production.
Fam73b deficiency promotes IRF1 accumulation
Previous studies revealed that BMDCs from Stat1 −/− and Ifnar −/− mice exhibit strongly reduced IL-12 p70 but not IL-12 p40 secretion41. Thus, Fam73b ablation may improve type I IFN induction to promote Il12a expression. In contrast, Fam73b deficiency caused a significant reduction in inducible Ifnb expression by various stimulators (Supplementary Fig. 7a), without affecting TBK1 and IRF3 phosphorylation (Supplementary Fig. 7b). The results indicated that Ifnb may not be the reason for the enhanced Il12a levels. Mitochondrial fission-related ROS production generally acts as a signal to trigger pro-inflammatory cytokines42. To further assess the role of ROS, we first measured ROS production using FACS. The results revealed that Fam73b ablation impaired ROS induction (Supplementary Fig. 7c). Furthermore, the increase in Il12a was not restored when cells were incubated with hydrogen peroxide (H2O2) (Supplementary Fig. 7d). Recently, several groups found that mitochondria-mediated metabolic change was involved in establishing immune cell phenotype26. Therefore, we compared oxygen consumption by WT and FAM73b KO BMDMs using extracellular flux analysis. FAM73b KO cells exhibited severely decreased OXPHOS activity at the basal level, but the difference was impaired when cells were treated with LPS (Supplementary Fig. 7f), while the relative extracellular acidification rate (ECAR) values were substantially similar between WT and FAM73b KO cells (Supplementary Fig. 7g). WT and FAM73b KO macrophages did not respond to either the FAO inhibitor etomoxir (ETO) or the protonophore uncoupler FCCP (Supplementary Fig. 7h). Our data support the conclusion that the phenotypic differences between WT and FAM73b KO macrophages were correlated with metabolic reprogramming mediated by mitochondria.
Recent evidence has revealed that the balance between the utilization of glycolysis and mitochondrial respiration is controlled by NF-κB, which is also crucial for Il12a induction43. To elucidate the role of NF-κB in FAM73b KO-mediated Il12a augmentation, we examined MAP kinase (MAPK) and canonical NF-κB activation. Immunoblot detection indicated that the increase in Il12a is not due to hyperactivation of IKK, based on the level of phosphorylated IKKs, or its substrates IκBα and p105 (Fig. 5a). Fam73b deficiency also did not appreciably affect activation of the three major MAPK families (Fig. 5b). Furthermore, loss of FAM73b was dispensable for nuclear translocation of NF-κB members (c-Rel, p50, and p65), as revealed by the similar levels of these proteins in the nucleus (Fig. 5c). Our results suggested that Il12a augmentation was not due to hyperactivation of these conventional signaling pathways.
Previous studies revealed that the transcription factor IRF1 is required for IL-12 p35 induction and for suppressing IL-10 and IL-23 production, which is consistent with the phenotypes caused by FAM73b deficiency44, 45. Interestingly, immunoblot analyses revealed that both FAM73b (Fig. 5d) and FAM73a KO (Supplementary Fig. 8a) macrophages displayed a higher IRF1 protein level. Comparable levels of Irf1 mRNA indicated that the IRF1 increase was associated with posttranslational modification (Fig. 5e). A proteasome inhibitor, MG132, largely rescued the IRF1 level in WT macrophages, suggesting that mitochondrial fission stabilized IRF1 by suppressing its proteolysis process (Fig. 5f). Consistently, loss of IL-12 mRNA expression was observed in activated macrophages from IRF-1 KO mice (Fig. 5g). Further, to test the role of IRF1 aggregation, we used two different shRNAs to knockdown the Irf1 gene in both WT and FAM73b KO BMDMs (Supplementary Fig. 8b). IRF1 silencing specifically impaired Il12a induction and erased the difference between WT and KO cells (Fig. 5h). Our results suggest that IRF1 plays an essential role in mitochondrial dynamics-mediated inflammatory responses. FAM73b KO BMDMs also had profoundly less abundant IRF1 K48-linked ubiquitination (Fig. 5i), suggesting a requirement of mitochondrial fission in suppressing IRF1 ubiquitination.
Mitochondrial fission impairs CHIP monoubiquitination
An E3 STUB1, known as CHIP, has been reported to mediate IRF1 protein degradation via ubiquitination32. A recent study established that a fraction of the CHIP population is monoubiquitinated in vivo, which has functional consequences for CHIP activity46. Therefore, we compared the CHIP monoubiquitination levels between WT and FAM73b KO BMDMs. Surprisingly, monoubiquitinated CHIP was severely abolished in the cytoplasm of FAM73b KO BMDMs (Fig. 6a). CHIP knockdown elevated IRF1 accumulation and Il12a production in WT cells, but little or no effect was observed without Fam73b (Fig. 6b and Supplementary Fig. 8c). A monoubiquitination site, K2, of CHIP that resides in an N-terminal extension is conserved in mammals (Supplementary Fig. 8d). To investigate the function of this site, we knocked out endogenous CHIP in the murine macrophage cell line RAW264.7 using a CRISPR/CAS9 system. CHIP KO RAW264.7 cells were reconstituted with WT and mutant CHIP K2A. Mutant CHIP lacked monoubiquitination, which impaired its capacity to promote IRF1 K48-linked polyubiquitination (Fig. 6c, d). Additionally, mutant CHIP did not cripple Il12a induction as was seen in the WT control (Fig. 6e). MG132 treatment indicated that degradation of monoubiquitinated CHIP mediated by mitochondrial fission mainly occurred through the proteolysis pathway (Fig. 6f). Furthermore, FAM73b KO BMDMs had profoundly more abundant CHIP K48-linked polyubiquitination when proteasome activity was blocked (Fig. 6g).
To examine the interaction between mitochondria and CHIP, we performed immunofluorescence image analysis using confocal microscopy. The results showed that CHIP did not locate at mitochondria in KO cells. This dissociation was rescued by MG132 treatment (Fig. 6h). Interestingly, subcellular fractionation studies also revealed that only monoubiquitinated CHIP was associated with mitochondria (Fig. 6i). Furthermore, Mdivi-1 treatment restored the monoubiquitinated CHIP level (Fig. 6j). CHIP recruitment to the mitochondria was also rescued by Mdivi-1 (Fig. 6k) in FAM73b KO BMDMs. Together, these results establish that fusion mitochondria-associated monoubiquitinated CHIP is indispensable for IRF1 K48-linked polyubiquitination and degradation via proteolysis.
Fission mitochondria degrade monoubiquitinated CHIP
CHIP is known to be associated with Parkin, which is required for clearing damaged mitochondria to maintain cellular homeostasis47. Surprisingly, Fam73b- and Mfn1/2-deleted cells had high levels of Parkin (Fig. 7a and Supplementary Fig. 8e), suggesting that mitochondrial fission triggered increased Parkin expression. However, comparable p62 and light chain 3 (LC3α/β) levels indicated that CHIP degradation was not due to mitophagy activation (Fig. 7a). The increase in Park2 mRNA abundance further indicated that mitochondrial fission enhanced the Parkin transcriptional level (Fig. 7b). Upregulated Parkin is selectively recruited to fission mitochondria in FAM73b-deficient cells (Fig. 7c). Interestingly, Mdivi-1 not only restored CHIP monoubiquitination but also suppressed Parkin expression (Fig. 7d). To evaluate the role of Parkin in IL-12 induction, we collected Parkin KO BMDMs and measured various cytokine levels. Parkin deficiency suppressed IL-12 expression but promoted IL-10 and IL-23 induction (Fig. 7e). Loss of Parkin also upregulated CHIP monoubiquitination and inhibited IRF1 expression (Fig. 7f). Moreover, Parkin silencing significantly rescued the stability of monoubiquitinated CHIP and erased the phenotype driven by Fam73b depletion (Fig. 7g, h). Together, all of these data suggest that Parkin accumulation contributes to FAM73b KO-mediated phenotypes.
To clarify the relationship between Parkin and CHIP, we first evaluated their interaction. The results indicated that overexpressed Parkin mainly associated with monoubiquitinated CHIP but not regular CHIP in HEK293T cells (Fig. 7i). Furthermore, coexpression of CHIP with Parkin strongly induced CHIP polyubiquitination in HEK 293T cells (Fig. 7j). Taken together, these data suggest that mitochondrial fission upregulated the expression and recruitment of the crucial E3 ubiquitin ligase Parkin, which colocalizated with CHIP and induced monoubiquitinated CHIP degradation.
Mitochondria are dynamic organelles that frequently divide and fuse2, 48. Previous evidence has revealed that mitochondria not only sustain immune cell homeostasis but are also necessary for launching immune responses. However, how mitochondrial dynamics determine the subtypes of immune responses remains poorly investigated. In this study, we clarified an essential role of the mitochondrial morphology in promoting IL-12 induction in innate immunity. The outer mitochondrial membrane protein FAM73b promotes mitochondrial fusion and suppresses TLR-stimulated IL-12 expression. Elevation of the FAM73b level is significantly associated with various types of stimulation. However, the classical fusion regulators Mfn1, Mfn2, and Opa1 also induce phenotypes similar to Fam73b in macrophages. These results indicate that multiple mitochondrial dynamics-related proteins may be involved in innate immunity regulation. Interestingly, our data showed that Fam73b deficiency functioned as a negative regulator of TLR-induced type I IFNs, which was different from MFN1/2 KO cells49. Therefore, type I IFN regulation by Mfn1/Mfn2 depends on their individual functions but not the morphology switch directly.
Current research suggests that mitochondrial dynamics control the balance between metabolic pathway engagement and T-cell fate23. However, as shown in Fig. 3a–d, Fam73b depletion did not have an intrinsic role in T-cell activation. In contrast to the Opa1 −/− T cells, CD4+ or CD8+ T cells from T-cell-specific KO mice did not exhibit any increases in memory T cells or differences in CD4+ T-cell differentiation. These data indicate that the potential functions of mitochondrial dynamics in T cells remain incompletely investigated. These differences indicated that the phenotypes observed in Opa1 −/− mice might be due to its individual functions but not the mitochondrial morphology.
Mitochondrial dynamics influence cellular function through multiple mechanisms, such as cell survival and metabolic output. Mitochondrial fusion is known to be important for efficient FAO via lipids, but fission promotes aerobic glycolysis50, 51. Furthermore, according to a previous finding, M2 macrophages exhibit enhanced FAO, while M1 macrophages rely instead on aerobic glycolysis27. However, few studies have elucidated the relationship between mitochondrial dynamics, metabolic changes and macrophage polarization. Our study found that Fam73b deficiency promoted M1-like phenotypes without an increase in ECAR. This result suggests a novel mechanism of mitochondrial morphology regulating macrophage polarization. Additionally, molecular evidence revealed that mitochondrial fission facilitated IL-12 induction through regulation of the CHIP–IRF1 axis stability via Parkin. The Pink1–Parkin pathway has been demonstrated to promote mitochondrial fission by downregulating MFN1 and OPA152. However, MFN1/2 or FAM73b deficiency also enhanced Parkin levels. These results suggest that mitochondrial fission is part of a feedback loop for promoting the Parkin level and maintaining mitochondria quality. ER stress has been shown to be important for activation of the unfolded protein response (UPR) and thereby for upregulation of Parkin expression53. FAM73b deficiency may enhance Parkin transcription by inducing ER stress. Based on our findings, we propose a model explaining how mitochondrial fission regulates TLR-stimulated Il12a expression (Supplementary Fig. 9).
In conclusion, our study presents a new concept of cancer immunotherapy through modulation of the mitochondrial morphology. Physically, TLRs and the suppressive cytokine IL-4 promotes mitochondrial fission and fusion, respectively, by regulating the Fam73b level. Mitochondrial fission is mediated by various molecules and specifically enhances Il12a induction and inhibits suppressive regulators such as Il10 and Arg1. Both germline and myeloid cell-specific Fam73b ablation profoundly enhances T-cell responses to tumor growth. Our data further suggest a central function of mitochondrial fission in regulating IRF1 stability and promoting monoubiquitinated CHIP degradation. Moreover, upregulated Parkin recruits to fission mitochondria and inhibits CHIP–IRF1 axis signal transduction, which is in turn important for enhancing IFN-γ induction for Th1 polarization in vivo. Our findings demonstrate a novel signaling network that controls the innate immunity response and has profound therapeutic implications for cancer treatment.
Mice were maintained under specific-pathogen-free conditions in a controlled environment of 20 °C–22 °C, with a 12/12 h light/dark cycle and 50–70% humidity; food and water were provided ad libitum. Fam73a-targeted mice were produced by transcription activator-like effector nucleases (TALEN) from the FVB/N strain. Parkin KO mice (Strain ID: 006582, C57BL/6 background) were provided by Dr. Shuo Yang from the Jackson Laboratory. IRF1 KO mice (Strain ID: 002762, B6.129S2 background) were purchased from the Jackson Laboratory. The heterozygous FAM73b-knockout first mice (Strain ID: Fam73btm1a(KOMP)Wtsi) were originally obtained from the UC DAVIS Knockout Mouse Project (KOMP) Repository. Male and female heterozygous FAM73b-knockout first mice (Fam73b +/−) were mated to each other to produce control WT and homozygous KO mice (Fam73b −/−). In FAM73b KO first mice, exon 3 of the Fam73b gene was targeted using a FRT-LoxP vector (Supplementary Fig. 2a). Fam73b-floxed mice were generated by crossing the FAM73b KO first mice with FLP deleter mice (Rosa26-FLPe; Jackson Laboratory). The Fam73b-floxed mice were further crossed with Cd4-Cre and Lyz2-Cre mice (all from Jackson Laboratory, C57BL/6 background) to generate T cell conditional FAM73b KO (Fam73b f/f Cd4-Cre), myeloid cell conditional FAM73b KO (Fam73b f/f Lyz2-Cre), and dendritic cell conditional FAM73b KO (Fam73b f/f Itgax-Cre) mice. Heterozygous mice were bred to generate littermate controls and KO (or conditional KO) mice for experiments. In the animal studies, the WT and multiple KO mice are randomly grouped. Outcomes of animal experiments were collected blindly and recorded based on ear-tag numbers of 6- to 8-week-old experimental mice. Genotyping was performed as indicated in Supplementary Fig. 2b, f, g. All animal experiments were conducted in accordance with protocols approved by the Institutional Animal Care and Use Committee of Zhejiang University.
Antibodies and reagents
Antibodies targeting IKBα (C-21, 1:1000), Fam73b (S-12, 1:1000), p65 (C-20, 1:1000), Lamin B (C-20, 1:1000), ERK (K-23, 1:2000), phospho-ERK (E-4, 1:1000), JNK2 (C-17, 1:1000), p38 (H-147, 1:1000), IKKα (H-744, 1:1000), ubiquitin (P4D1, 1:1000), p105/p50 (C-19, 1:1000), IRF1 (M-20, 1:1000), CHIP (H-231, 1:1000), HSP60 (H-1, 1:3000), TBK1 (108A429, 1:1000), IRF3 (SC-9082, 1:1000), LC3α/β (SC-292354, 1:1000), and c-Rel (sc-71, 1:1000), as well as a control rabbit IgG (sc-2027), were from Santa Cruz Biotechnology. Antibodies targeting phospho-DRP1 (Ser637, #3455, 1:1000), DRP1 (D6C7, 1:1000), Mitofusin-2 (D1E9, 1:1000), phospho-IκBα (Ser32, 14D4, 1:1000), phospho-JNK (Thr180/Tyr185, #9251, 1:1000), phospho-p38 (Thr180/Tyr182, 3D7, 1:1000), phospho-p105 (Ser933, 18E6, 1:1000), phospho-IKKα/β (Ser176/180, 16A6, 1:1000), phospho-TBK1 (Ser172, D52C2, 1:1000), phospho-IRF3 (Ser396, 4D4G, 1:1000), and K48-linkage-specific polyubiquitin (D9D5, 1:8000) were purchased from Cell Signaling Technology Inc. Anti-actin (C-4, 1:10,000) was from Sigma. HRP-conjugated anti-HA (HA-7) and anti-FLAG (M2) were from Sigma-Aldrich. Antibodies targeting p62 (ab56416, 1:1000) and Parkin (ab15954, 1:1000) were from Abcam. Fluorescence-labeled antibodies are listed in the section describing the flow cytometry and cell sorting procedures.
Mouse CHIP and Parkin cDNA were amplified from splenic mouse mRNA using PCR and inserted into the pCLXSN(GFP) retroviral vector. HA-Ub, Myc-Ub, and subtypes of Ub expression plasmids were provided by Dr. Shao-cong Sun. pGIPZ lentiviral vectors encoding a non-silencing shRNA control and two different Irf1, Stub1, and Park2 shRNAs were designed and produced by Invitrogen.
LPS (derived from Escherichia coli strain 0127:B8) and CpG (2216) were from Sigma-Aldrich. R848 and Poly I:C was from Amersham, and recombinant murine M-CSF was from Peprotech.
All the uncropped scans of the western blots are shown in Supplementary Fig. 10.
Flow cytometry and ICS
Single-cell suspensions from B16 melanoma, spleens or draining lymph nodes were subjected to flow cytometry using CytoFlex (Beckman Coulter) and the following fluorescence-labeled antibodies from eBioscience: PB-conjugated anti-CD4 and anti-CD11c; PE-conjugated anti-B220 and anti-F4/80; PerCP5.5-conjugated anti-Gr-1 (Ly6G); APC-conjugated anti-CD62L; APC-CY7-conjugated anti-CD11b and anti-CD8; and FITC-conjugated anti-IFNγ, anti-CD44 and anti-Foxp3. DAPI was from Life Technologies, and MitoSpyTM Orange CMTMRos was from BioLegend.
For intracellular cytokine staining (ICS), the tumor-infiltrating T cells were stimulated for 4 h with PMA plus ionomycin in the presence of monensin and then subjected to intracellular IFN-γ and subsequent flow cytometry analysis. All FACS data were analyzed by FlowJo 7.6.1. All FACS gating/sorting strategies are shown in Supplementary Fig. 11.
BMDMs (5 × 105) were collected and seeded on 12-well plates containing 70% alcohol-pretreated slides for starvation overnight. Cells treated with or without stimulation as indicated were stained with 250 nM MitoSpyTM Orange CMTMRos for 20 min and fixed with 4% paraformaldehyde (PFA) for 20 min. Then, the cells were washed with PBS three times and stained with 10 µg/ml DAPI. All the samples were imaged on a LSM710 (Carl Zeiss) confocal microscope outfitted with a Plan-Apochromat ×63 oil immersion objective lens (Carl Zeiss). The data were collected using Carl Zeiss software ZEN 2010. To quantify the mitochondrial morphology of MitoSpyTM Orange CMTMRos-stained macrophages, scoring was blindly analyzed with Image-Pro software. Short was defined as cells with a majority of mitochondria less than 7 μm and long as cells in which the majority of mitochondria were more than 7 μm.
Transmission electron microscopy
WT and Fam73b KO macrophages were washed in PBS and fixed in 2.5% GA on ice for 15 min. Then, the cells were scraped and put into a 1.5 ml EP tube. The GA solution was refreshed, and the cells were suspended and incubated at 4 °C overnight. Then, the cells were embedded into agarose gel. The gel was cut into small pieces, washed in PBS and post-fixed in 1% osmic acid for 1–2 h. Then, the samples were washed in PBS and dehydrated in a gradient ethanol series (50%, 75%, 85%, 95% and 100% ethanol), each for 15 min. Then, the samples were embedded in Epon resin. Embedded samples were cut into 60-nm ultrathin sections, and sections were counterstained with uranyl acetate and lead citrate. All the samples were observed using a Hitachi HT7700 electron microscope.
BMDMs (2 × 105) were plated in glass chamber slides (LabTek) and starved overnight. Cells were stained for 20 min with 250 nM MitoSpyTM Orange CMTMRos (BioLegend). Imaging was started immediately after addition of 500 ng/ml LPS on a DeltaVision RT system with SoftWorx software (Applied Precision) at 37 °C and 10% CO2. Images were acquired every 10 min for 2 h. The data were exported as uncompressed AVI files and processed with Premiere Pro.
B16 model of melanoma
WT, FAM73b KO, Fam73b-lyz2 cre/+, Fam73b-Itgax cre/+and Fam73b-cd4 cre/+ mice administered a s.c. injection of 2 × 105 B16 melanoma cells. Tumor size is presented as a square caliper measurement calculated based on two perpendicular diameters (mm2). The injected mice were monitored for tumor growth every other day, and based on protocols approved by the IACUC of Zhejiang University. The maximum sizes of tumors are limited up to 225 mm2 by ethical permission. Thus, the mice were sacrificed and defined as lethality when the tumor size reached to 225 mm2. For analysis of tumor growth rate and tumor-infiltrating immune cells, a lower number (2 × 105) of B16 cells were injected to prevent lethality during the course of the experiment. To minimize individual variations, littermate WT and different Fam73b conditional KO mice were used. Mice were randomly selected for tumor injection, and analysis of tumor size was performed in a blinded fashion. To analyze tumor-infiltrating immune cells, tumor tissues were treated with 0.25 mg/ml collagenase A (Sigma-Aldrich) and 25 U/ml DNase (Roche Diagnostics, Indianapolis, IN) for 20 min at 37 °C, and the cells were passed through a plastic mesh. The resulting dissociated cells were collected by centrifugation, resuspended in Red Blood Cell Lysis Buffer for 3 min, and washed twice in PBS. The cells were subjected to flow cytometry analysis, and TAMs and MDSCs were gated based on CD11b+F4/80+ or CD11b+Gr-1high surface markers, respectively.
WT and Fam73b KO mice were given a s.c. injection of a high dose of MCA (800 μg), as indicated in the figure legends, in the right flank. The mice were monitored for fibrosarcoma development weekly over the course of 150 days. Tumor size is presented as a square caliper measurement calculated based on two perpendicular diameters (mm2). The mice were monitored for tumor growth every week. Mice with tumors larger than 225 mm2 were sacrificed and recorded as having lethal tumors based on protocols approved by the Institutional Animal Care and Use Committee of Zhejiang University.
Generation of BMDCs and BMDMs
Bone marrow cells isolated from the WT or dendritic cell-specific KO mice were cultured in RPMI 1640 medium containing 10% FBS supplemented with GM-CSF (10 ng/ml) for 7 days. The differentiated DCs were stained with Pacific blue-conjugated anti-CD11c and isolated with a FACS sorter. In some experiments, BMDMs were generated using MCSF-supplemented medium.
WT and Fam73b KO macrophages were washed with PBS twice and stained with DCFH-DA (10 μM; Beyotime) in DMEM without FBS for 20 min at 37 °C. After being washed with PBS twice, the cells were harvested and analyzed by flow cytometry (Cytoplex, Beckman Coulter).
The OCR and ECAR were measured in XFp extracellular flux analyzers (EFAs) (Seahorse Bioscience) using a XFp Cell Mito Stress Test Kit and a XFp Glycolysis Stress Test kit, respectively. The following parameters were used in the assays: seed cells 8 × 104 per well, Oligomycin 1.0 μM, FCCP 1.0 μM, Rotenone/antimycin A 0.5 μM, glucose 10 mM, and 2-DG 50 mM as indicated.
Lentiviral particles were produced by transfecting HEK293T cells (using the calcium-phosphate method) with a pGIPZ lentiviral vector encoding either a non-silencing shRNA or Irf1-, Stub1-, or Park2-specific shRNA, along with the packaging vectors psPAX2 and pMD2. BMDMs, differentiated using M-CSF-supplemented medium for 5 days, were infected with the lentiviruses for 8 h. After 72 h, the infected cells were enriched via flow cytometric cell sorting (based on GFP expression) and subsequently used for experiments.
ELISA and qRT-PCR
Supernatants of in vitro cell cultures were analyzed via ELISA using a commercial assay system (eBioScience). For qRT-PCR, total RNA was isolated using TRIzol reagent (Molecular Research Center, Inc.) and subjected to cDNA synthesis using RNase H-reverse transcriptase (Invitrogen) and oligo (dT) primers. qRT-PCR was performed in triplicate using an iCycler Sequence Detection System (Bio-Rad) and iQTM SYBR Green Supermix (Bio-Rad). The expression of individual genes was calculated with a standard curve and normalized to the expression of Actb. The gene-specific PCR primers (all for mouse genes) are shown in Supplementary Table 1.
Cells were pretreated with MG132 for 2 h and then lysed with a Nonidet P-40 lysis buffer (50 mM Tris-HCl, pH 7.5, 120 mM NaCl, 1% Nonidet P-40, 1 mM EDTA, and 1 mM DTT) containing 6 M urea and protease inhibitors. IRF1 or CHIP was isolated by IP with antibodies targeting IRF1 (M-20) and CHIP (H-231). The ubiquitinated proteins were detected by IB using an anti-ubiquitin (Santa, P4D1) or anti-K48-linked ubiquitin (Millipore, 05–1307) antibody.
Statistical analysis was performed using Prism software. No data are excluded from the analyses. Two-tailed unpaired t-tests were performed. P values <0.05 were considered significant, and the level of significance is indicated as *P < 0.05 and **P < 0.01. In the animal studies, a minimum of four mice were required for each group based on the calculated number necessary to achieve a 2.3-fold change (effect size) in a two-tailed t-test with 90% power and a significance level of 5%. All statistical tests are justified as appropriate, and the data meet the assumptions of the tests. The variance is similar between the statistically compared groups.
Sequence data that support the findings of this study are available from the authors and have been deposited in the National Center for Biotechnology Information (NCBI) BioProject with the primary accession code PRJNA359723. All other data are available from the authors upon reasonable requests.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
We thank Dr. Shao-cong Sun, Dr. Zongping Xia, Dr. Bin Zhao, and Dr. Hong-bing Shu for the expression vectors. This study was supported by the National Natural Science Foundation of China (81572651/81771675), Fundamental Research Funds for the Central Universities (2016QN81013), the National Natural Science Foundation of China (31471315/ 31671457), the Thousand Young Talents Plan of China, and the Zhejiang University Special Fund for Fundamental Research.