Cellular stress signaling activates type-I IFN response through FOXO3-regulated lamin posttranslational modification

Neural stem/progenitor cells (NSPCs) persist over the lifespan while encountering constant challenges from age or injury related brain environmental changes like elevated oxidative stress. But how oxidative stress regulates NSPC and its neurogenic differentiation is less clear. Here we report that acutely elevated cellular oxidative stress in NSPCs modulates neurogenic differentiation through induction of Forkhead box protein O3 (FOXO3)-mediated cGAS/STING and type I interferon (IFN-I) responses. We show that oxidative stress activates FOXO3 and its transcriptional target glycine-N-methyltransferase (GNMT) whose upregulation triggers depletion of s-adenosylmethionine (SAM), a key co-substrate involved in methyl group transfer reactions. Mechanistically, we demonstrate that reduced intracellular SAM availability disrupts carboxymethylation and maturation of nuclear lamin, which induce cytosolic release of chromatin fragments and subsequent activation of the cGAS/STING-IFN-I cascade to suppress neurogenic differentiation. Together, our findings suggest the FOXO3-GNMT/SAM-lamin-cGAS/STING-IFN-I signaling cascade as a critical stress response program that regulates long-term regenerative potential.


Main
Stem cells persist over the total mammalian lifespan to maintain tissue homeostasis by replacing damaged or lost cells and the deterioration of their differentiation capacity is one of the key components of organismal aging 1 . Elevation of stress response pathways is a hallmark of aging tissues which also promotes the depletion of adult stem cells by inducing senescence or cell death [2][3][4] . Intrinsic and extrinsic cell-stressors such as DNA damage, mitochondrial dysfunction, loss of proteostasis, and the inflammatory tissue milieu contribute to an increased stress response. In particular, oxidative stress contributes to the functional decline of stem cells by inflicting damage to cellular macromolecules ultimately leading to cytostasis or cytotoxicity 1 . Model organism for precocious stem-cell depletion or dysfunction emphasize the role of key molecules involved in oxidative stress response (i.e. Atm 5 , Foxo3 6-10 , Prdm16 11 ) in maintaining stem-cell reserves. Nevertheless, the direct cellular consequence of stress response that translates as molecular aging of stem cells remains broad and non-specific.
Among the many organs, the brain is particularly vulnerable due to its high oxygen consumption, unusual enrichment of polyunsaturated fatty acids as well as the presence of excitotoxic amino acids 12 . As a result, neural stem/progenitor-cells (NSPCs) constantly face stressful challenges and decline in their neurogenic potential in adult brains 13,14 . However, the underlying mechanism by which elevated stress response regulates NSPC fate remains poorly understood.
FOXO transcription factors play evolutionarily conserved roles in a wide range of biological processes from aging to metabolism, not only by sensing stress but also through promoting stress resistance 15 . For example, previous studies indicated that FOXO is required for long-term regenerative potential of the hematopoietic stem cell (HSC) by regulating the response to physiologic oxidative stress and quiescence 7 . In the central nervous system, FOXO expression not only serves a key role in preserving neural stem cell pools 9, 10 but also protects neurons against age-related axonal degeneration across species [16][17][18] . But despite these advances, there still lacks a mechanistic understanding of how aging-related oxidative stress affects FOXO activation systematically and whether and how that contributes to the neuroprotective responses.
The type-I interferon (IFN-I) response is an innate immune response that can be induced by a number of pattern recognition receptors 19 . Among them, cytosolic DNA fragments are recognized by cyclic GMP-AMP synthetase (cGAS) which initiates reaction of GTP and ATP to form cyclic GMP-AMP (cGAMP) cGAMP, a ligand of the signaling adaptor stimulator of interferon genes (STING, TMEM173). The binding of cGMP to STING activates TANK Binding Kinase 1 (TBK1) kinases-mediated phosphorylation of transcription factor IRF3 that triggers IFNα/β production and subsequent IFN response [20][21][22][23] . Increased IFN-I response has been shown to promote NSPC quiescence and suppress neurogenic differentiation 13,24 . Interestingly, a recent study revealed that IFN-I signaling is elevated in the brain of aged humans and animals and correlates with increased oxidative stress 13 . But the connection between oxidative stress and IFN-I response is unclear.
Here, we report that oxidative stress-induced FOXO3 activation promotes transcriptional upregulation of N-methyltransferase GNMT to deplete S-adenosyl-L-methionine (SAM). Using the NSPC system, we further uncovered that reduction of intracellular SAM availability disrupts nuclear lamin maturation that eventually leads to cytosolic DNA leakage, cGAS/STING activation and IFN-I response. As a result, NSPC enters quiescence at the expense of neurogenic differentiation. These findings established FOXO3-GNMT/SAM-lamin-cGAS/STING-IFN-I signaling cascade as a critical stress response program that protects NSPC from detrimental environmental insults.

High redox potential-mediated cellular stress activates IFN-I pathway
The neurogenic differentiation potential of NSPCs declines under iatrogenic insults, traumatic injuries, or inflammatory stress conditions 13,25 . Among these, to determine how oxidative stress signal impacts NSPC differentiation, we subjected the cultured murine NSPCs to either pro-oxidant agent paraquat (PQ) or anti-oxidant Nacetylcysteine (NAC). Measurement by a ratiometric Grx-roGFP2 sensor confirmed that PQ treatment induced a marked elevation of intracellular redox potential relative to the mock-treated NSPCs, whereas NAC treatment led to a significant reduction (Fig. 1a, b).
Importantly, compared to the mock-treated control NSPCs, we found that NSPCs under PQ but not NAC treatment, exhibited marked reduction in production of TUBB3 or doublecortin (DCX)-positive newly born neurons when induced to differentiate (Fig. 1c), suggesting a regulatory role of oxidative stress response on neurogenic differentiation.
To determine the signaling pathway that underlies the oxidative stress-induced neurogenic decline, we next performed gene expression profiling against mock-treated control and NSPCs following 48 h redox preconditioning. Gene set enrichment analysis (GSEA) of differentially regulated genes revealed type-I Interferon (IFN-I) signaling as one of the most enriched signature pathways (Fig. 1d, e). Quantitative real-time PCR (qRT-PCR) further confirmed transcriptional upregulation of major IFN-I signaling downstream surrogates, including Ifnb1, Isg15, Socs1, Usp18, and Nos2 (Fig. 1f).
Moreover, ELISA analysis further revealed that secretion of the key IFN-I response effector IFNβ was also strongly elevated in PQ-treated NSPCs (Fig. 1g). To test whether activation of IFN-I signaling accounts for redox stress-induced neurogenic decline, we treated the NSPCs with IFNβ. Indeed, addition of IFNβ alone was sufficient to suppress neurogenic differentiation of NSPCs (Fig. 1h). These data collectively suggest that oxidative stress signaling regulates neurogenic differentiation through IFN-I pathway.

FOXO3 is required for ROS-induced IFN-I response
FOXO proteins are major regulators of physiological oxidative stress response partly, because they modulate the transcriptional expression of ROS-scavenging enzymes 26,27 .
To determine the role of FOXO3 in ROS-induced IFN-I response, we next analyzed how FOXO3 depletion impacts oxidative stress-induced IFN-I signaling activation.
In comparison, the PQ treatment-induced IFN-I response and its upstream and downstream signaling activation were evidently attenuated in PQ treated NSPCs depleted FOXO3 (sg-Foxo3) (Fig. 2b-d), suggesting that FOXO3 plays a crucial role in regulation of ROS-induced IFN-I response.
FOXO3 integrates a variety of cellular signals that modulate its transcriptional activity 28 .
To examine whether activation of FOXO3 by itself was sufficient to trigger IFN-I response independently of oxidative stress, we transduced NSPCs with an adenoviralencoded activated mutant form of FOXO3 (FOXO3 TA , triple alanine form 29 ) that was exclusively located in the nucleus (Fig. 2e, f). qRT-PCR analysis of the FOXO3 TA transduced NSPCs revealed a markedly increased expression of ISGs as compared to the control adenovirus-infected NSPCs (Fig. 2g), indicating that FOXO3 is directly responsible for oxidation stress-induced IFN-I activation.

Oxidation of FOXO3 activates IFN-I response
Previous reports suggest that ROS signaling activates FOXO by inducing its nuclear translocation 30,31 . Indeed, we found that ROS treatment of NSPCs not only stimulated FOXO3 nuclear retention and activation, but also led to an elevated FOXO3 protein expression ( Fig. 3a-c). By contrast, treatment with the anti-oxidant NAC promoted FOXO3 cytoplasmic shuttling and protein degradation and reduced FOXO3 transcriptional activity. The nucleo-cytoplasmic shuttling of FOXO reportedly is controlled through a combination of post-translational modifications, particularly AKTmediated phosphorylation that promotes its cytoplasmic sequestering 28 . Consistently, ROS treatment caused a significant reduction of FOXO3 phosphorylation at threonine 32/serine 256, raising a possibility that oxidative stress may induce FOXO3 nuclear translocation by impeding its phosphorylation 32,33 .
Reversible cysteine thiol oxidation is a well-known mechanism that regulates signaling cascades and protein activities under redox stress conditions 34 . Immunoprecipitation followed by blotting analysis against cysteine sulfenic acid (Cys-SOH) confirmed a strong elevation of FOXO3 sulfenylation in ROS treated NSPCs compared to the controls (Fig. 3e). Since mammalian FOXO3 contains a highly conserved Cys residue (Cys31) adjacent to threonine (Thr) that is subjected to AKT phosphorylation (Fig. 3d), we next asked whether the Cys31 oxidation affected AKT-dependent Thr32 phosphorylation. To this end, we reconstituted the FOXO3-null NSPCs with a lentiviral construct encoding green fluorescent protein (GFP)-tagged either wild-type (WT) or Cys31 to alanine FOXO3 mutant (C31A). Immunoblot analysis showed that ROS treatment induced a strong reduction of Thr32 phosphorylation in FOXO3 WT but not the C31A FOXO3 mutant compared to the mock-treated control cells (Fig. 3f), suggesting that oxidation at Cys31 may impedes Thr32 phosphorylation. In line with this, fluorescence live-imaging of the FOXO3 localization indicated that the ROS-induced nuclear translocation of GFP-tagged FOXO3 C31A mutant was significantly compromised as compared to that of the wild-type FOXO3 (Fig. 3g). Concurrently, qRT-PCR analysis of ROS-treated NSPCs revealed upregulation of oxidative stress-induced FOXO3 downstream anti-oxidant genes (i.e. Sod2, Sesn3) as well as markedly subdued ISGs in C31A mutant transduced NSPCs as compared to wild-type FOXO3-transduced control cells (Fig. 3h, i). Notably, although C31A mutation compromised ROS-induced FOXO3 nuclear shuttling and activity, it did not affect FOXO3 nuclear translocation upon treatment with either PI3K inhibitor (GDC0941) or AKT inhibitor (MK2206) (Extended Data Fig. 1a, b). These findings together indicate that Cys thiol oxidation and its associated inhibitory function on FOXO3 phosphorylation is a key mechanism that underlies ROS-induced FOXO3 and its downstream signaling activation (Fig. 3j).

Compromised lamin processing upon oxidative stress invokes IFN-I response
The IFN-I signaling is a cellular innate immune response and is often triggered by the cytosolic DNA-sensing cGAS/STING pathway 35 . To examine whether ROSinduced IFN-I activation is mediated by aberrant cytosolic DNA appearance, we treated the cGAS-GFP-expressing NSPCs with pro-oxidant PQ. Fluorescence microscopic analysis of the PQ treated cells revealed an increased nuclear leakage as represented by appearance of lobulated nuclei and inappropriate formation of cytoplasmic cGAS-GFP-containing DNA foci (Fig. 4a), suggesting a compromised nuclear envelope integrity.
The nuclear lamina is essential for the maintenance of nuclear shape and mechanics, and its dysregulation causes nuclear envelopathies and accumulation of cytosolic chromatin fragments 36 . As an essential component of nuclear lamina, the maturation of functional lamin A/B from newly synthesized prelamin A/B follows a multistep process of posttranslational modification that involves farnesylation and methylation of its C-terminal cysteine before proteolytic cleavage of its C-terminal 15 amino acids (Fig. 4b). To test whether oxidative stress impacts lamin distribution, we stably transduced NSPCs with N-terminal GFP-tagged prelamin A or mCherry-tagged prelamin B1. Strikingly, compared to the control cells in which the tagged lamin proteins dispersed evenly along the nuclear envelopes, we found that a large portion of PQtreated NSPCs exhibited an irregular lamin distribution, reminiscent of protein aggregation (Fig. 4c). To test whether disrupted lamin processing activated IFN-I response upon oxidative stress, we stably expressed the cysteine 585 to serine prelamin B1 mutant (LMNB1 CS ) in NSPCs, that is defective of prelamin maturationessential farnesylation and methylation 37 . Immunofluorescence analysis of the cells indicated that LMNB1 CS mutant protein, which was negative to mature lamin B1-specific 8D1 monoclonal antibody 38 , formed the aggregate-like nucleoplasmic foci similar to the ones observed in wild-type lamin B1-transduced NSPCs under ROS treatment (Fig. 4d).
qRT-PCR analysis further revealed that expression of LMNB1 CS mutant alone activated IFN-I response and downstream gene expression and that ROS treatment could further enhance its effect on ISGs expression (Fig. 4e). Conversely, expression of a C-terminus deletion form of mature lamin A mutant (LMNA m ) strongly attenuated the ROS-induced IFN-I signaling and ISGs expression (Fig. 4f, g). These findings together strongly suggest defective lamin processing as an underlying cause of IFN-I response under oxidative stress.

ROS-induced intracellular SAM depletion disrupts lamin maturation
Lamin maturation requires isoprenylation and methylation on the c-terminal cysteine residues 39 . To determine how ROS regulates lamin posttranslational modification, we performed targeted quantitative polar metabolomics profiling by liquid chromatographytandem mass spectrometry (LC-MS) on samples derived from control, pro-or antioxidant treated NSPCs. Among the 258 metabolites analyzed, we found that the turnover of SAM exhibited an inverse correlation with redox potential. Compared to the mock-treated control cells, treatment with the pro-oxidant PQ gave rise to a 3.3-fold reduction of cellular SAM and a 1.7-fold reduction of SAM to SAH ratio (

ROS regulates intracellular SAM through GNMT
SAM is a universal co-substrate involved in methyl group transfers 40 . Intracellular SAM levels are balanced by MAT2A-catalyzed synthesis and its consumption through multiple catabolic processes (Fig. 6a). Since our metabolite profiling revealed little change of intracellular methionine -the precursor for SAM (Fig. 6a), we next turned to the expression of the major enzymes involved in SAM metabolism. qRT-PCR analysis of control and ROS-treated NSPCs indicated that cellular expression of MAT2A, the enzyme that catalyzes the synthesis of SAM from methionine, remained relatively stable (Extended Data Fig. 3a). By contrast, among the key catabolic enzymes that catalyze the SAM to SAH conversion, we found that expression of glycine N-methyl transferase (GNMT) was markedly induced by PQ treatment but suppressed by anti-oxidant NAC To determine whether GNMT-regulated SAM depletion could also instigate nuclear leakage accompanied by cGAS/STING signaling activation, we transduced the DOXinducible GNMT expressing NSPCs with the cytosolic DNA fragment-sensing cGAS-GFP construct. Compared to the mock-treated control cells, DOX-treated NSPCs exhibited a significant elevation of cGAS-GFP-containing foci formation (24.4% ± 1.275% vs 0.7% ± 0.45%) (Fig. 6k). Immunoblot analysis of control and DOX-treated NSPCs further revealed a strong reduction of 8D1-positive mature lamin B1 protein level following GNMT induction (Fig. 6l), suggesting compromised lamin maturation.
Consistently, transduction of a mature lamin mutant (GFP-LMNA m ) in the DOX-GNMT NSPCs restored the nuclear envelop integrity and suppressed the GNMT inductionevoked IFNβ secretion as well as ISGs expression (Fig. 6m, n). Concordantly, enforced expression of a STING HAQ mutant 44 could also partially offset the effect of GNMT induction by attenuating the IFN-I response and downstream gene expression (Extended Data Fig. 3b, c). Collectively, these data indicate that GNMT is a key regulator of IFN-I response under ROS treatment.

Redox stress impacts NSPC neurogenic potential through FOXO3-regulated GNMT expression
We observed that the frequency of PQ-induced cGAS-GFP foci was suppressed by immunoprecipitation (ChIP) coupled with qRT-PCR confirmed that FOXO3 was 9.7 ± 2.7-fold enriched at GNMT promoter relative to the background gene desert (Fig. 7a).
Moreover, qRT-PCR analysis of NSPCs indicated that relative to control virus infected NSPCs, enforced expression of an active FOXO3 TA mutant was able to significantly enhance GNMT transcription (Fig. 7b), whereas CRISPR/Cas9-mediated depletion of endogenous FOXO3 suppressed GNMT mRNA expression (Fig. 7c). These data suggest that FOXO3 transcriptionally controls GNMT expression.
We next examined whether FOXO3 regulated ROS-initiated IFN-I response through GNMT. Treatment of NSPCs with pro-oxidant agent PQ promoted a marked increase of GNMT mRNA and protein expression relative to the mock-treated control cells (Fig. 7c,   d). This ROS-induced GNMT upregulation was significantly compromised in the NSPCs depleted of FOXO3. Notably, the FOXO3-depleted NSPCs exhibited a steady increase of cellular SAM levels compared to their respective controls before or after ROS treatment (Fig. 7e). Concordantly, knockdown of GNMT abolished FOXO3 TA expression-induced IFN-I response and downstream gene expression (Fig. 7f).
Finally, we went on to determine how FOXO3-GNMT/SAM-IFN-I signaling pathway regulates neurogenesis under oxidative stress condition. As expected, immunoblot and immunofluorescence analysis revealed that treatment of NSPCs with pro-oxidant PQ attenuated their neuronal differentiation capacity, as evidenced by the reduction of expression of neuronal marker TUBB3 and percentage of TUBB3-positive cell population relative to the mock-treated control cells (Fig. 7g-j). But further depletion of either FOXO3 or GNMT in the PQ-treated NSPCs reversed the ROS effect and was sufficient to restore their neurogenic potential (Fig. 7g-j). Notably, the FOXO3 depletionpromoted neuronal differentiation could be further blocked by DOX-induced exogenous GNMT expression, consistent with our finding that GNMT is a downstream effector of FOXO3 signaling (Fig. 7k). Considering the elevation of ROS in aging brain, we further examined type-I IFN stimulated gene expression in young and old (<60 year-old) and aged (>60 year-old) patient brain samples. Consistently, we observed a clear increase of ISGs along with GNMT mRNA expression in aged brains (Extended Fig. 6).
Altogether, our results suggest the FOXO3-GNMT/SAM-lamin-cGAS/STING-IFN-I signaling cascade as an important physiological stress response program that may protect the nervous system against acute oxidative insults (Fig. 7l).

Discussion
Alterations of the redox state, as in many brain pathologies, regulate the fate of NSPCs 45 . Our study revealed that cellular stresses including a higher redox potential are translated into IFN-I response via FOXO3-GNMT/SAM-lamin changes (Fig. 7l). In particular, we showed that redox potential controls NSPC function by altering IFN-I response through metabolic regulation of intracellular SAM availability. Mechanistically, our study uncovered a previously unidentified FOXO3 signaling cascade that functionally connects oxidative stress response with NSPC differentiation through SAMdepletion-induced IFN-I activation. Our findings of redox-dependent neurogenic regulation warrant future studies on the therapeutic rejuvenation of stress-impacted adult NSPCs.
FOXO transcription factors play a central role in a wide range of biological processes, including stress sensing and regulation of stress response 15 . Genetic studies from many organisms have repeatedly demonstrated the conserved insulin/IGF-PI3K-AKT-FOXO cascade as a major regulatory signaling pathway of aging and lifespan. In the central nervous system, expression of FOXO plays not only a key role in preserving neural stem cell pools 9, 10 , but also protects neurons against age-related axonal degeneration across species [16][17][18] . Despite these advances, there still lacks a mechanistic understanding of how oxidative stress affects FOXO activation systematically and whether and how that contributes to the neuroprotective responses. In the current study, we identified FOXO3 oxidation at the evolutionarily conserved Cys31 residue as a new regulatory mechanism that modulate redox-dependent FOXO3 nucleo-cytoplasmic shuttling and downstream signaling. Notably, a previous study reported that ROSinduced FOXO4 oxidation at Cys239 promotes its nuclear import by forming a disulfidedependent protein complex with transportin-1 31 . These findings suggest that redoxregulated nuclear shuttling is a conserved mechanism underlying FOXO-mediated oxidative stress response.
Our data indicate that FOXO3 mediates redox response through regulation of GNMT and downstream SAM catabolism. Enhanced SAM catabolism by GNMT extends the lifespan in Drosophila 46 . In the nervous system, GNMT-mediated SAM metabolism is required for the proliferative signaling of NSPC and hippocampal neurogenesis 47 . But the underlying mechanism is unclear. Here we found that treatment of NSPCs with prooxidants led to upregulated GNMT expression and reduction of intracellular SAM availability. SAM is a metabolite generated via the one-carbon metabolism and is the main methyl donor in cellular methylation reactions 40 . SAM depletion through dietary methionine restriction has been shown to modulate histone methylation and induce stem cell quiescence 42,43,48 . In our study, we found that not only GNMT-induced SAM depletion in NSPCs confers a global reduction of H3K4 methylation, but is also sufficient to trigger cGAS/STING signaling and IFN-I response through regulation of nuclear lamin maturation. These findings support FOXO3-GNMT/SAM axis as a stress responsive program that protect tissue homeostasis by orchestrating anti-oxidative function, metabolic rewiring, and gene expression.
Defective lamin processing is known to cause various human pathologies, particularly those related to aging. A truncated lamin A causes a premature aging syndrome of Generation of viral particles. Each sgRNA was cloned into lentiCrisprV2 vector following Zhang lab instructions 59 . In brief, annealed sgRNA oligos with T4 PNK enzyme was cloned into lentiCrisprV2 vector digested by BsmBI. pDONR221-GNMT was cloned into pInducer gateway destination vector by using LR clonase II enzyme mix. To Buffers consisted of 100% acetonitrile for mobile A, and 0.1% NH4OH/20 mM CH3COONH4 in water for mobile B. Gradient ran from 85% to 30% A in 20 min followed by a wash with 30% A and re-equilibration at 85% A. Metabolites were identified on the basis of exact mass within 5 ppm and standard retention times. Relative metabolite quantitation was performed based on peak area for each metabolite. All data analysis was done using in-house written scripts.

Statistical analysis.
We determined experimental sample sizes on the basis of preliminary data. All results are expressed as mean ± s.e.m. GraphPad Prism software (version 7, San Diego, CA) was used for all statistical analysis. Normal distribution of the sample sets was determined before applying unpaired Student's two-tailed t-test for two group comparisons. One-way ANOVA was used to assess the differences between multiple groups. The mean values of each group were compared by the Bonferroni's post-hoc procedure. Differences were considered significant when P<0.05.