Cellular Reactive Oxygen Species Inhibit MPYS Induction of IFNβ

Many inflammatory diseases, as well as infections, are accompanied by elevation in cellular levels of Reactive Oxygen Species (ROS). Here we report that MPYS, a.k.a. STING, which was recently shown to mediate activation of IFNβ expression during infection, is a ROS sensor. ROS induce intermolecular disulfide bonds formation in MPYS homodimer and inhibit MPYS IFNβ stimulatory activity. Cys-64, -148, -292, -309 and the potential C88xxC91 redox motif in MPYS are indispensable for IFNβ stimulation and IRF3 activation. Thus, our results identify a novel mechanism for ROS regulation of IFNβ stimulation.


Introduction
ROS, including the superoxide anion, hydrogen peroxide and hydroxyl radicals, are generated from the incomplete reduction of oxygen [1]. Most intracellular ROS (90%) are generated in mitochondria due to electron leakage along the respiratory chain [2]. ROS are also generated in the endoplasmic reticulum (ER) during the unfolded protein response (UPR) [3]. Bacterial infections induce transient production of large amounts of ROS, a process called the oxidative burst, on the membrane of endosomes within phagocytes such as neutrophils and macrophages [4].
ROS are a double-edged sword. Potent ROS mediated oxidative stress causes irreversible cell damage and eventually cell death [5]. More modest elevation of ROS activates ''redoxin signaling'' which uses cysteine residues as redox sensors to mediate inflammatory responses [1,5]. Cys can be reversibly oxidized to sulfenic acids, S-glutathionylated or S-nitrosylated cysteines, or disulfide bonds [1]. S-glutathionylations of IRF3 and Cys-179 of the IKK-b subunit inhibit their activation [6,7]. Oxidation of catalytic Cys in caspases and protein tyrosine phosphatases (PTP) inhibits their activation [8,9]. Thus, ROS-mediated post-translational modifications on Cys regulate the biological activities of many proteins. High cellular ROS levels have been linked to ageing, human cancers, inflammatory, lung and cardiovascular diseases [5,10]. Antioxidants show protective effects for certain cancers and cardiovascular diseases [11].
MPYS is a four-transmembrane protein which was originally identified as a growth inhibitor that mediates anti-MHC II mAb induction of cell death in B lymphoma cells [12]. Later, it was found MPYS is a potent IFNb stimulator that is essential for innate immune responses to RNA and DNA viruses [13]. MPYS contains a potential redox-active Cys 88 -L-G-Cys 91 (C 88 xxC 91 ) motif at the N-terminus of the second TM ( Figure S3). The CxxC motifs are often found in the oxidoreductases and are essential for their catalysis of redox reactions [14]. Here we report that MPYS is a ROS sensor. Sustained cellular ROS cause MPYS oxidation and loss of its ability to activate IFNb expression.

ROS inhibit MPYS induction of IFNb expression
MPYS was initially located in mitochondria [12]. Mitochondria is the major source for intracellular ROS production due to electron leakage along the respiratory chain [2]. We asked if the function of MPYS can be regulated by ROS. Rotenone is a specific inhibitor of mitochondrial electron transport chain complex I [15]. 293T cells treated with rotenone exhibited increased cellular ROS level ( Figure 1a). MPYS is an IFNb stimulator. Overexpressing MPYS in 293T cells activates IFNb promoter [13]. However, we found that pre-treatment of 293T cells with rotenone completely abolishes MPYS induction of IFNb ( Figure 1b).
To test the hypothesis that ROS interfere with MPYS function, we expressed MPYS in two different lines: 293GT and 293MT. The 293GT line has constitutively higher endogenous ROS than the 293MT line ( Figure 1c). We found that in 293GT cells, MPYS lost its ability to activate IFNb (Figure 1d, 1e).
We next determined whether ROS compromise MPYS ability to transduce signals in response to a physiologic stimulus. MPYS is required for Listeria monocytogenes induced IFNb response ( Figure  S1c) [13]. We found that pre-treatment of rotenone significantly inhibited Listeria monocytogenes induction of IFNb production in RAW264.7 macrophage cells ( Figure 1f). As a control, Rotenone pre-treatment did not inhibit Poly (I:C) induced IFNb response ( Figure S1d). We conclude that sustained cellular ROS inhibit MPYS induction of IFNb.

Endogenous MPYS occurs predominantly as a nondisulfide-bonded homodimer
We showed previously that MPYS is predominantly dimeric in mouse B cells [12]. Using the reducible chemical crosslinker Dithiobis [succinimidyl propionate] (DSP), we found on nonreducing SDS-PAGE, DSP treated MPYS ran as a single band of ,80 kDa, the predicted size of a MPYS homodimer (Figure 2d, lane 1) while non-DSP treated mock (or DTT treated) MPYS ran at the monomer size (,40 kDa) (Figure 2d, lane 3, 2). We further found that Flag-tagged MPYS associates with differently tagged, i.e. HA-MPYS, when the two are co-expressed ( Figure S2c). Thus, the ,80 kDa band seen upon non-reducing SDS-PAGE gel analysis of DSP treated splenocytes is a MPYS homodimer.
Since in non-DSP treated cells this homodimer is sensitive to SDS (Figure 2d, lane 3), its ''constitutive'' dimerization is not mediated by intermolecular disulfide bonds. We conclude that MPYS exists in unperturbed cells primarily as non-disulfide bonded homodimer (Figure 3c, upper panel).

MPYS becomes a disulfide-linked homodimer under conditions of oxidative stress
We wanted to know whether the non-disulfide-bonded SDS sensitive MPYS dimer becomes intermolecular disulfide bonded upon oxidation.

MPYS cysteines are critical for IFNb stimulation
MPYS is an IFNb stimulator [13]. Here, we showed that ROS induces intermolecular disulfide bonds formation in MPYS homodimer and inhibits its ability to activate IFNb. We thus hypothesized that cysteines in MPYS are important for MPYS function. We mutate the critical cysteines involved in oxidized MPYS homodimer formation and examine their abilities to stimulate IFNb promoter. In addition, we measured phosphorylation of IRF3 which lies downstream from MPYS in the IFNb stimulation pathway [16].
Mutation of these Cys to Ser revealed that the Cys64 is absolutely required for IFNb reporter and IRF3 activation, while the Cys148 has moderate effect (Figure 4a, 4b). Mutation of the C 88 xxC 91 to S 88 xxS 91 or all four Cys (S4) eliminated MPYS ability to induce IRF3 phosphorylation and IFNb production (Figure 4c, 4d).

Discussion
In this report, we determined that MPYS, a potent stimulator IFNb production, is a ROS sensor. We further found that the IFNb stimulation by MPYS can be inhibited by high levels of ROS, and ROS stimulate MPYS oxidation as manifest by formation of intermolecular disulfide bonds within MPYS homodimers. We conclude that MPYS senses the Redox state of the cell and under conditions of high oxidative stress is ''turned off'', presumably by oxidation.
ROS play an important role in regulation of innate immune responses and the strength of the ROS signals determines the outcome of such responses [1]. For example, low levels of H 2 O 2 activate the p53 antioxidant response, whereas high levels trigger p53-dependent apoptosis [17]. The effect of ROS on MPYS function likely also depends on the levels of ROS. Rotenone treatment, which produces sustained cellular ROS, may cause irreversible modification on critical cysteines in MPYS thus inhibits the MPYS-mediated IFNb response. Interestingly, NLRX1, a negative regulator of the anti-viral response [18], triggers strong ROS production (comparable to the level triggered by TNFa) [19]. Both NLRX1 and MPYS can be found in mitochondria [12,19]. It is tempting to suggest that NLRX1 negatively regulates the IFNb response by producing strong ROS, inhibiting MPYS activity.
MPYS contains 6 critical cysteines important for IFNb stimulation. Among them, Cys-64, Cys-148 and C 88 xxC 91 are targets of cellular ROS. The other two Cys are adjacent to Arg (Cys 292 -Arg 293 and Cys 309 -Arg 310 ). The positive charge of Arg293 or Arg310 may lower the pKa of Cys292 or Cys309 so that they may also be targets of ROS [20]. The question is, then, why these cysteines are important for IFNb stimulation. Previous studies propose that MPYS/STING functions as an adaptor protein that recruits ser/thr kinase TBK1 [13,21]. We find that TBK1 associates with both oxidized and reduced MPYS ( Figure S4a). More importantly, both the C148S and C 88 xxC 91 mutant have normal TBK1 association ( Figure S4b). Considering the fact that the CxxC motifs are often found in the oxidoreductases and are essential for their catalysis of redox reactions [14], we hypothesize that MPYS may also have the oxidoreductase activity and TBK1 may be its substrate.
Previous over-expression studies have placed MPYS in the ER and mitochondrial outer membrane [12,13,16]. The mitochondrial outer membrane is physically and physiologically connected to the ER [22]. This physical link may facilitate Ca 2+ [23] and, we suggest, ROS communication between ER and mitochondria. We found that the ER stress inducer, Brefeldin A, also generated oxidized MPYS ( Figure S2a). Thus, MPYS may also act in ER stress sensing.
The mitochondrial intermembrane space (IMS) is connected to the cytosol by porins in the outer membrane of mitochondria which allow the diffusion of small ions like glutathione [24]. Thus the environment of IMS is less oxidizing than that of the ER lumen. Recently, a group of interacting mitochondrial proteins, including MPYS, VISA, NLRX1 and most recently Mitofusin 2, have been identified as key components of the innate intracellular viral sensing pathway [25] [18,26]. In light of our current discovery that the MPYS is a Redox sensor, we suggest that the antiviral response also utilizes mitochondrial ROS as second messenger.
In conclusion, we have shown that the IFNb stimulator, MPYS is a ROS sensor and its signaling function is regulated by ROS. Future studies need to be done to determine if MPYS is indeed an oxidoreductase and the in vivo biological significance of that activity.

Reporter Gene Assay
Cells were seeded in 24-well dishes (,1610 5 /ml) and transfected the following day using Effectene transfection reagent kit (Qiagen, 301427). Reporter assays were performed as previous described [30]. All experiments were repeated at least three times and results are representative. Error bars represent Standard deviation.

Measurement of intracellular ROS
Cells were washed and suspended in PBS, and then culture with H 2 DCFDA (5 mM) (Invitrogen, D399) for 20 min at 37uC. They were then washed in PBS and the fluorescence was measured by flow cytometry.
Listeria monocytogenes infection RAW264.7 cells expressing the IFNb-luc construct were infected with Listeria monocytogenes strain 10403S at multiplicity of five bacteria/cell. Cells were washed at 1 h after infection. Live bacteria were killed by addition of gentamicin. Whole cell lysates were prepared in RIPA buffer after infection. Luciferase activity was read with BD Monolight kit and Synergy reader. Figure S1 MPYS knockdown in RAW264.7 and K46 B cells. a, b. WCL from K46 (a) or RAW264.7 (b) cells expressing either the luciferase control or MPYS knock-down constructs [12] were fractionated on a non-reducing SDS-PAGE, and blotted with anti-MPYS Ab. N.S.: non specific staining. c. RAW264.7-IFNb-Luc cells expressing MPYS-knockdown (MPYS-KD) or control knock-down (Control) were infected with Listeria monocytogenes as in Materials and Methods. Luciferase activity was measured. P value was calculated by student T-test (one-tailed). d. RAW264.7-IFNb-Luc cells were first treated with rotenone for 16hrs. Poly (I:C) (2.5mg/ ml for 16hrs) was added into culture. Luciferase activity was measured as before. (TIF) Figure S2 Endogenous MPYS becomes a disulfidelinked homodimer under oxidative stress. a. RAW264.7 cells were treated with Brefeldin A (0.5mg/ml) for 20hr in culture. Cells were lysed in the RIPA buffer, fractionated using nonreducing SD-PAGE, and probed with anti-MPYS Ab. Oxidized MPYS is indicated. b. RAW264.7 cells were serum starved overnight. Cells were then harvested and lysed in RIPA buffer. The oxidized MPYS was detected as (a). c. 293MT cells were cotransfected with indicated plasmids. After 24 hrs, the cells were lysed in CHAPS buffer. Flag IP was performed. The blot was probed with anti-MPYS or anti-HA Ab. (TIF) Figure S3 Alignment of MPYS from multiple species. The human mpys cDNA described and used in this report was derived from a fetal liver library and found to differ by a single amino acid from that previously reported by [16] and [13] (indicated by arrow). In this sequence a G to A SNP (rs1131769) altered amino acid 232 from His (H) to Arg (R). The human population frequency data in the dbSNP database indicates that ,80% of human are homozygous and ,20% are heterozygous for this R232 allele. No European and only ,2% of Japanese or Sub-Saharan African are homozygous for the previous reported H232 allele. Thus the R232 allele we studied here is the most relevant to the human population and referred as wild-type MPYS in this study. Cysteine residues important for the IFNb stimulation were boxed. (TIF) Figure S4 The C148S and CxxC mutants have normal TBK1 association. a. Flag-TBK1 and MPYS were cotransfected into the 293MT cells. Flag proteins were precipitated. The immunoprecipitates were fractionated using non-reducing SDS-PAGE gels and blotted as indicated (left panel). WCL before or after the Flag IP were fractionated on a non reducing gel (right panel). Blots were probed with anti-MPYS Abs. b. Various HA tagged MPYS constructs were co-transfected with Flag-TBK1 into the 293MT cells. MPYS was immunoprecipitated by anti-HA MAb (16B12). Immunoprecipitates were fractionated and blots were probed with indicated Abs. (TIF)