Autolysosomal degradation of cytosolic chromatin fragments antagonizes oxidative stress–induced senescence

Oxidative stress-induced DNA damage, the senescence-associated secretory phenotype (SASP), and impaired autophagy all are general features of senescent cells. However, the cross-talk among these events and processes is not fully understood. Here, using NIH3T3 cells exposed to hydrogen peroxide stress, we show that stress-induced DNA damage provokes the SASP largely via cytosolic chromatin fragment (CCF) formation, which activates a cascade comprising cGMP-AMP synthase (cGAS), stimulator of interferon genes protein (STING), NF-κB, and SASP, and that autolysosomal function inhibits this cascade. We found that CCFs accumulate in senescent cells with activated cGAS-STING-NF-κB signaling, promoting SASP and cellular senescence. We also present evidence that the persistent accumulation of CCFs in prematurely senescent cells is partially associated with a defect in DNA-degrading activity in autolysosomes and reduced abundance of activated DNase 2α. Intriguingly, we found that metformin- or rapamycin-induced activation of autophagy significantly lessened the size and levels of CCFs and repressed the activation of the cGAS-STING-NF-κB-SASP cascade and cellular senescence. These effects of autophagy activators indicated that autolysosomal function contributes to CCF clearance and SASP suppression, further supported by the fact that the lysosome inhibitor bafilomycin A1 blocked the role of autophagy-mediated CCF clearance and senescence repression.

Oxidative stress-induced DNA damage, the senescence-associated secretory phenotype (SASP), and impaired autophagy all are general features of senescent cells. However, the cross-talk among these events and processes is not fully understood. Here, using NIH3T3 cells exposed to hydrogen peroxide stress, we show that stress-induced DNA damage provokes the SASP largely via cytosolic chromatin fragment (CCF) formation, which activates a cascade comprising cGMP-AMP synthase (cGAS), stimulator of interferon genes protein (STING), NF-B, and SASP, and that autolysosomal function inhibits this cascade. We found that CCFs accumulate in senescent cells with activated cGAS-STING-NF-B signaling, promoting SASP and cellular senescence. We also present evidence that the persistent accumulation of CCFs in prematurely senescent cells is partially associated with a defect in DNA-degrading activity in autolysosomes and reduced abundance of activated DNase 2␣. Intriguingly, we found that metformin-or rapamycin-induced activation of autophagy significantly lessened the size and levels of CCFs and repressed the activation of the cGAS-STING-NF-B-SASP cascade and cellular senescence. These effects of autophagy activators indicated that autolysosomal function contributes to CCF clearance and SASP suppression, further supported by the fact that the lysosome inhibitor bafilomycin A1 blocked the role of autophagy-mediated CCF clearance and senescence repression.
Hydrogen peroxide (H 2 O 2 ) or ionizing radiation can induce cellular senescence (1,2). This type of senescence is commonly referred to as oxidative stress-induced premature senescence (SIPS). 3 The hallmarks of SIPS include persistent DNA damage, autolysosome dysfunction, senescence-associated secretory phenotype (SASP), and endoplasmic reticulum stress (3,4). In addition, SIPS is associated with the loss of the nuclear lamina protein lamin B1 and the compromise of the integrity of the nuclear envelope (3,5,6). Concomitantly, nuclear membrane blebs that contain damaged chromatin fragments appear in senescent cells, which eventually partition into the cytoplasm and become cytosolic chromatin fragments (CCFs) (7,8). These CCFs contain damaged genomic DNA, ␥H2AX, and heterochromatin markers H3K9me3 and H3K27me3 but lack certain euchromatin markers, such as H3K9ac (7).
Cytoplasmic DNA can be recognized by the cytosolic DNA sensor cGMP-AMP synthase (cGAS). Following activation, cGAS synthesizes the second messenger molecule 2Ј-3ЈcGMP-AMP (cGAMP), which engages the stimulator of interferon genes (STING) (9 -12). Activation of cGAS-STING leads to two downstream pathways: type I interferon expression through interferon regulatory factor (IRF3) and pro-inflammatory responses through NF-B (NF-B) (13,14). The secretion of pro-inflammatory cytokines and other factors is a key feature of senescence, referred to as SASP (15,16). The mechanisms that activate SASP include a series of poorly understood events that are connected to DNA damage response and NF-B activation (17)(18)(19). Accumulating evidence indicates that SASP factors exert a non-cell-autonomous function and can alter the behavior of neighboring cells or play cell-autonomous roles and reinforce the senescence state (15,20).
Lysosomes play an important role in the degradation of CCFs (21)(22)(23). DNase 2 (also called DNase 2␣) is one of the lysosomal nucleases that have been well characterized (24). DNase 2␣-de-ficient mice spontaneously produce high levels of type I interferons and SASP factors, including tumor necrosis factor (Tnf-␣), interleukin 6 (Il6), and Il-1␤ (25)(26)(27)(28). Others and our previous results have suggested that the lysosome content is up-regulated in senescent cells, whereas their degradation activity is down-regulated (29,30). However, it remains unknown whether the impaired autolysosomal function contributes to oxidative stress-mediated CCF accumulation and subsequent SASP and cellular senescence.
In the present study, we show that autolysosomal dysfunction in DNA degeneration contributes to CCF accumulation in oxidative SIPS cells, where CCFs provoke SASP production through the aberrant activation of the cGAS-STING pathway. Conversely, restoration of DNA degeneration activity of autolysosome by activating autophagy can effectively accelerate the clearance of CCFs, thereby down-regulating cGAS-STING pathwaymediated SASP and preventing cellular senescence.

SASP is a feature and promoter of oxidative SIPS
The generation of SASP in senescent cells was first verified in an oxidative SIPS model we established previously (31) (Fig. S1, A-C). Gene expression profiling with RNA-Seq revealed that the expression of a lot of genes related to DNA replication and cell cycle was decreased in H 2 O 2 -treated NIH3T3 cells (Fig.  S1D). On the contrary, the expression of SASP genes was upregulated remarkably in those senescent cells, mostly representing the top up-regulated group based on gene ontology (GO) and gene set enrichment analysis (GSEA) (Fig. S2, A-C). The up-regulation of representative SASP genes was also confirmed by a qRT-PCR assay (Fig. S2D).
Then the paracrine role of SASP factors in senescence development was assessed. Proliferative NIH3T3 or MRC-5 cells were treated with conditioned medium (CM) from senescent NIH3T3 cells (CM-S) or control proliferative NIH3T3 cells (CM-C) to investigate the influence of SASP on proliferative cells (Fig. S3A). Compared with incubation with CM-C, the incubation with CM-S caused a 7-fold increase in senescenceassociated ␤-galactosidase (SA-␤-Gal) activity in NIH3T3 and 4-fold in MRC-5 cells, respectively (Fig. S3B). A similar conclusion was obtained based on a co-cultured experiment using Transwells (Fig. S3C). Co-culture with H 2 O 2 -treated NIH3T3 cells increased the percentage of SA-␤-Gal-positive cells in either NIH3T3 or MRC-5 cells (Fig. S3D). These results clearly demonstrate the important role of SASP and SASP-mediated paracrine function in senescence development.

DNA damage, CCF accumulation, and cGAS-STING pathway activation occur in senescent cells
It is interesting to explore the relationship between oxidative damage, CCFs, and SASP. We first examined the status of damaged nuclear DNA and proteins. Both 8-OHdG and ␥H2AX signals (two DNA damage markers) were temporally strengthened during senescence development (Fig. 1, A and B). Other signs of cellular oxidative damage were also observed, including disordered lamin B1-positive structure (Fig. 1C), increased protein levels of ␥H2AX and 53BP1, and decreased protein level of lamin B1 (Fig. 1E). Damaged chromatin was released from the nucleus to cytoplasm, where those DNA-containing blebs were called CCFs (DAPI ϩ ␥H2AX ϩ ). The release of nuclear chromatin to the cytoplasm was observed in senescent cells, and the percentage of CCF-positive cells was increased with senescence processing (Fig. 1D). In addition, the level of cytosolic dsDNA was also increased in senescent cells (Fig. 1F). These data suggest that oxidative stress induces DNA damage and promotes the release of nuclear chromatin and formation of CCFs.
Recent studies have proposed that CCFs could activate cGAS. Therefore, we addressed the relationship between oxidative DNA damage-prompted CCF formation and the cGAS-STING-SASP pathway in senescent cells. We observed that the cellular level of cGAMP, the product of cGAS enzyme, was increased in senescent cells (Fig. 1G). In addition, the dimeric STING protein was increased in senescent cells, together with the elevated phosphorylation of NF-B and IRF3 (Fig. 1H). Finally, we confirmed the increase in the mRNA of interferon ␤ (Ifn␤) and C-X-C chemokine ligand 10 (Cxcl10), two genes known widely as the downstream targets of the cGAS-STING pathway, in senescent cells (Fig. 1I). Together, these results suggest that oxidative DNA damage, CCF accumulation, and activation of the cGAS-STING signaling are collectively provoked in senescent cells, and these events were accompanied by SASP expression.

cGAS-STING pathway activation mediates SASP production and senescence process in an NF-Bdependent manner
Next, we sought to determine the causal relationship between CCF-evoked cGAS-STING activation and senescence progress. In NIH3T3 cells, we observed that cGAS knockdown repressed reduced H 2 O 2 -induced expression of SASP genes and senescence-associated genes, including p16 and p21 (Fig. 2, A-C). This effect of cGAS knockdown was coupled with the reduced activation of NF-B and the repressed senescence ( Fig.  2, C and D). However, cGAS knockdown had no effect on the CCF content after H 2 O 2 treatment (Fig. S4). Given that the transcription factor NF-B is a key regulator of SASP and also one of the major downstream targets of cGAS-STING signaling (13,18), we assessed the significance of NF-B signaling in cGAS-STING-involved cellular senescence. Quite clearly, NF-B inhibitor QNZ (EVP4593) treatment suppressed H 2 O 2induced expression of SASP factors and cellular senescence (Fig. 2, E-G). QNZ treatment also eased the exogenous cGAMP-induced activation of NF-B (p65 phosphorylation) and up-regulation of SASP factors (Fig. 2, H and I). It must be noted that QNZ did not repress the cGAMP-induced up-regulation of IFN␤, an IRF3 target gene, although IRF3 is known as another downstream target of the cGAS-STING pathway in different situations (13). Collectively, these results suggest that NF-B is critically involved in cGAS-STING function during the H 2 O 2 -induced SASP program and senescence.

CCF degradation capacity of autolysosomes is decreased in SIPS
Then we were interested in examining the mechanism underlying CCF accumulation in senescent cells. As the autophagic flux was impaired in senescent and genotoxic CCF degradation contributes to senescence suppression cells (31,32) in our previous reports, we asked whether a lysosomal defect in DNA degradation contributes to CCF accumulation in senescent cells.
We first examined the colocalization status of CCFs (DAPI ϩ ␥H2AX ϩ ) with autophagosomal and autolysosomal structures (LC3 ϩ , SQSTM1/p62 ϩ , or Lamp1 ϩ ). Notably, this kind of colocalization was observed, mostly shaped as blebs budding off of the nucleus in senescent cells (Fig. 3, A-C). To evaluate the DNA degradation capacity of autolysosome in senescent cells, we examined the expression of DNase 2␣, the unique DNA endonuclease inside of lysosomes. We found that the abundance of the activated form of DNase 2␣ was dramat-ically decreased after H 2 O 2 treatment, without an mRNA level change (Fig. 3, D-F). Notably, we found that the level of activated DNase 2␣ protein in isolated lysosomes from H 2 O 2treated cells significantly decreased compared with control cells (Fig. 3, G and H). These results suggest that the impairment of DNA degradation capacity of autolysosome occurs with senescence.
To further confirm this hypothesis, we inspected the DNA degradation activity of lysosomes by performing an in vitro assay processed by incubating isolated lysosomes with purified plasmid DNA (Fig. 3G). As shown, more intact plasmid DNA remained after the reaction when the DNA was incubated with

CCF degradation contributes to senescence suppression
lysosomes isolated from H 2 O 2 -induced senescent cells, compared to that incubated with lysosomes from proliferative cells. Correspondingly, the relative ratio of DNA degradation was less in that group (Fig. 3I). These data further indicated the decreased DNA degradation capacity in lysosomes of senescent cells. Taken together, these results suggest the autolysosomal localization of CCFs and the weakened DNA degradation capacity in senescent cells.

Restoring autolysosomal function decreases CCF accumulation and cGAS-STING signaling in senescent cells
Based on the clarification that impaired autolysosomal degeneration of DNA contributes to CCF accumulation during senescence, we then verified the effect of autolysosomal activa-tion on CCF elimination. Metformin and rapamycin are two typical activators of autophagy (31). Either of them significantly reduced the protein level of LC3 and SQSTM1/p62, which reflects the restoration of autophagy flux and the enhanced degradation capacity of autolysosomes (Fig. 4A). In parallel, these autophagy activators repressed H 2 O 2 -induced expression of ␥H2AX (Fig. 4A) and diminished the percentage of CCFpositive cells in senescent cells (Fig. 4, B and C). Correspondingly, metformin and rapamycin repressed the activation of cGAS-STING-NF-B signaling and inhibited the expression of SASP factors in senescent cells (Fig. 4, D and E). These autophagy activators also dramatically lessened the percentage of SA-␤-Gal-positive cells following H 2 O 2 treatment (Fig. 4F). In addition, incubation of proliferative cells with culture me-   dium from autophagic activated-senescent cells (metformin or rapamycin-treated) led to a decrease in SA-␤-Gal activity (Fig.  4G). A similar result was obtained based on the co-cultured experiment using the Transwell system (Fig. 4H). Taken together, these results suggest that autophagy restoration can effectively reverse oxidative stress-induced CCF accumulation and cGAS-STING signaling activation, parallel with its effect of preventing oxidative stress-induced senescence.

Autolysosome contributes to autophagy-mediated CCF degradation and subsequent repression of cGAS-STING-SASP cascade
To clearly disclose the executive platform where CCFs are degraded, the function of lysosomes was further investigated. The approach featured treatment of cells with bafilomycin A1, an inhibitor of lysosome acidification (33). Although rare in H 2 O 2 -treated cells (Fig. 5A, row 1), the co-localization of CCFs (DAPI ϩ ␥H2AX ϩ ) and autolysosomal (SQSTM1/p62ϩ) or lysosome structures (Lamp1 ϩ ) became observable in cells treated with metformin or rapamycin ( Fig. 5A and Fig. S5 (rows 2 and  4)). Intriguingly, the autophagic activator-induced colocalization of CCFs and lysosomes was blocked by bafilomycin A1 (Fig. 5A and Fig. S5 (row 2 versus row 3 and row 4 versus row 5)). Quantitatively, metformin and rapamycin reduced the percentage of CCF-positive cells, and this effect was neutralized by bafilomycin A1 (Fig. 5B). What is more, our data also revealed that the autophagy-mediated decrease in cytoplasmic dsDNA content was blunted by bafilomycin A1 (Fig. 5C). These pieces of data suggest that lysosome functions as the crucial matrix and effector for CCF degradation. Furthermore, we analyzed

CCF degradation contributes to senescence suppression
whether bafilomycin A1 could block autophagy activatorelevated DNA degradation in senescent cells. The results showed that the DNA degradation capacity of lysosome was enhanced by metformin and rapamycin, but this enhancement was suppressed by bafilomycin A1 (Fig. 5D). Consistently, the metformin-and rapamycin-elevated abundance of the activated form of DNase 2␣ was markedly blunted by bafilomycin A1 (Fig. 5D). These results reveal the essential contribution of lysosomes to the degradation of CCFs.
Going forward from above, we next analyzed the effects of lysosome function on cGAS-STING signaling and SASP. Notably, bafilomycin A1 significantly blunted autophagic activatorsuppressed phosphorylation of NF-B p65 subunit and the mRNA level of SASP factors (Fig. 5, E and F). Taken together, our results show that autolysosomal activation markedly promotes CCF degradation and lysosomal inhibition impede CCF degradation, demonstrating the contribution of lysosomes to CCFs clearance and the regulation of cGAS-STING-SASP axis and stress-induced premature senescence.

Discussion
Our findings demonstrated that oxidative stress causes CCF accumulation in premature senescent cells, which is associated with the defective DNA degeneration activity of autolysosome. This study also verified that CCFs activate cGAS-STING-NF-B signaling, SASP, and oxidative SIPS development. Of importance, our study revealed the effectiveness of autophagic flux restoration in the degradation of CCFs and thereby the down-regulation of the cGAS-STING-NF-B-SASP-SIPS cascade (Fig. 6). These findings not only provide new evidence for the pro-senescent role of oxidative stress-provoked CCF-cGAS-STING signaling and the molecular basis of SASP regulation, but also establish the possibility that autolysosomal function is an intrinsic defense force of cells for CCF clearance and therefore a negative regulator of SASP production. These findings may encourage further study directed toward exploring new ways against senescence and aging.
SASP is a common hallmark of different types of senescence; however, the signaling governing its production remains unknown. The most important progress in this area reported in recent years might be the findings that SASP production is a downstream event of cGAS-STING pathway, whereas the latter can be activated by DNA damage-induced CCFs (7,13,14). Such evidence is also harvested via this study, which includes the remarkable expression of SASP component genes, the accumulated CCFs in cytoplasm, and the activated cGAS-STING pathway. It is worthy of mention that our data were obtained by using conditioned medium and Transwell-based co-cultivation. This line of evidence particularly supports the idea that, in addition to being the product of senescent cells and working to aggravate the pro-senescent effect of other stimuli, SASP acts actually as an independent promoter of senescence. In other words, SASP can initiate senescence by itself. With regard to the mechanism of SASP on senescence, a well-accepted theory is that SASP can change the situation of generating cells own and the microenvironment surrounding the generating cells through auto-and paracrine effects, thus accelerating the development of senescence in a population (19,20).
The innate DNA sensor cGAS and its downstream adaptor STING have recently been proven to be crucial regulators of the SASP and senescence (7,13). Previous work demonstrated that the down-regulation of lamin B1 and the subsequent partial loss of nuclear integrity with the occurrence of chromatin breakage represents a characteristic of senescent cells, which allows together the damaged chromatin fragments to be released into the cytoplasm, resulting in CCF accumulation, which then activates cGAS and SASP production (7,8). Our data are consistent, as DNA damage and cytoplasmic DNA foci appeared in senescent cells and were associated with cGAS-STING path- CCF degradation contributes to senescence suppression way activation. Moreover, knockdown of cGAS displayed a compromised SASP factor expression and SA-␤-Gal activity. Interestingly, we found that the emergence of the SASP was at a relatively late stage (at day 2 after H 2 O 2 treatment) of senescence development, whereas the occurrence of DNA damage and CCF accumulation was earlier (at day 1 after H 2 O 2 treatment). This observation strengthened the implication of CCFs in cGAS activation, SASP production, and senescence processing.
To seek the innate antagonistic way in which cells prevent CCF accumulation, we paid attention on the autophagic lysosomal system, because it is the only cellular organelle that acts on macromolecules and large cellular component degeneration in the cytoplasm and contains both protease and nuclease (33). Most importantly, it has been confirmed that the function of this system is impaired in senescent cells, and the restoration of the function of the system is a promising strategy for senescence prevention (30). This opinion is encouraged by the work of Ivanov et al. (22), which demonstrates that cytoplasmic DNA accumulated during senescence can be cleared by the lysosome, likely through an autophagy pathway. Moreover, Takahashi et al. (34) reported that the depletion of cytoplasmic DNases, DNase 2␣ and Trex1, can elevate cGAS-STING activity and SASP expression in pre-senescent cells. Our data are in agreement with the others, as the abundance of activated DNase 2␣ protein declined in senescent cells, with the lowered dsDNA degradation capacity of lysosomes. More intriguingly, we found that autolysosomal activators, metformin and rapamycin, can effectively improve the dsDNA degradation capacity of lysosomes, leading to reduced CCF accumulation in the cytoplasm. Given that pro-DNase 2␣ is processed and activated by proteolytic enzyme cathepsins in lysosome and this enzyme needs an acidic condition to hydrolyze dsDNA (24,25,35), it is reasonable to deduce that the activity of DNase 2␣ largely depends on the function of autolysosomes, probably with regard to the activity of cathepsins, the status of the pH value, and so on. For this reason, we believe more attention should be paid to the DNA degradation capacity of the autolysosomal system, to further explore the regulatory mechanism upon DNase 2␣ activity. In addition, we deem the role of autophagy activation in senescence prevention to be relevant not only to its activity in protein degradation but also to its activity in dsDNA degradation. Given the viewpoint of CCFs and SASP in cells, autolysosmal function is an available mechanism for alleviating CCF accumulation-provoked SASP production and senescence development.
In conclusion, our findings demonstrate the important contributions of both CCF formation and autolysosomal defect to oxidative stress-induced senescence and also the connection between CCF accumulation and GAS-STING-SASP cascade activation, which is largely a response to senescence development. Our finding further reveals that the improvement of autolysosomal function is a promising strategy against CCF accumulation and therefore an inhibitory way to prevent SASP and cellular senescence. These findings expand our understanding of the relationship of different cellular processes that affect SASP production via mutual interaction, positively or negatively.

NIH3T3 cells (murine fibroblast cell lines) and MRC-5 cells (human fibroblast cell lines) were purchased from Shanghai
Institutes for Biological Sciences of the Chinese Academy of Sciences (Shanghai, China) and cultured in complete Dulbecco's modified Eagle's medium (DMEM) supplemented with 10% (v/v) FBS in a humidified atmosphere with 5% CO 2 at 37°C.
For senescence induction, a modified H 2 O 2 treatment protocol was established as reported previously (30,31). In brief, NIH3T3 cells seeded in 100-mm dishes with 5 ϫ 10 5 cells/dish density 24 h were trypsinized and suspended in PBS at 1 ϫ 10 6 cells/ml density and exposed to 400 M H 2 O 2 in an Eppendorf tube at 37°C for 45 min. During H 2 O 2 treatment, the tube was turned upside down gently every 5-10 min. H 2 O 2 treatment was terminated by discarding H 2 O 2 -added PBS after a 5-min centrifugation at 800 rpm and a washing process. Then the cells were cultured with complete medium for various durations. During the adhesion cultivation, the cells were subjected to different treatments that will be described in the individual figure legends.

SA-␤-Gal staining
Intracellular SA-␤-Gal activity was assayed using an SA-␤-Gal staining kit (catalog no. C0602, Beyotime, Beijing, China) according to the standard protocol, and senescent cells were identified as bluish green-stained cells under a phase-contrast microscope. SA-␤-Gal-positive cells as a percentage of total cells was determined by counting 1,000 cells in seven random fields for each group. The results were expressed as the mean of triplicates Ϯ S.D.

Generation of conditioned medium
The indicated cells (2 ϫ 10 6 ) were seeded in a 10-cm dish and incubated for 3 days in DMEM with 10% (v/v) FBS. At the end of CCF degradation contributes to senescence suppression treatments, cells were washed with PBS, and cultured with fresh medium. After 48 h of incubation, these culture medium, refer to as conditioned medium (CM) was collected and centrifugation at 1,500 rpm for 5 min and filtered through a 0.2-m pore filter. The medium volume/cell number proportion was 1 ml/1 ϫ 10 5 NIH3T3 cells.

Cell treatment with CMs
NIH3T3 or MRC-5 cells were seeded at a density of 2.0 ϫ 10 3 cells/cm 2 , and once grown to ϳ30% confluence, cells were incubated for 7-10 days with CM from control proliferative NIH3T3 cells, reported as CM-C, or with CM from H 2 O 2treated and autophagy-activated H 2 O 2 -treated NIH3T3 cells, reported as CM-S and CM-S-Met or CM-S-Rapa, with medium changes every 2 days (DMEM/CM ϭ 1:3). Then cells were analyzed by SA-␤-Gal staining.

Co-culture experiment
For transmembrane co-culture, proliferative cells and H 2 O 2treated cells were cultured together in a transmembrane system (Transwell, Corning, Lowell, MA) with 0.4-m pores. In shared medium co-cultures, H 2 O 2 -treated or proliferative cells were cultured in the lower well, and proliferative cells were cultured in the upper well of the Transwell plate. Proliferative cells were seeded at the same density for shared medium and contact co-culture groups for each run of the experiment. The conditions were designed to identify whether secreted factors from H 2 O 2 -treated cells lead to proliferative cells to senescence in the conditioned media. After 7-10 days of culture, cells were analyzed by SA-␤-Gal staining. Every experiment was repeated at least three times.

RNA-Seq
RNA was isolated from the control (CTL), H 2 O 2 -treated NIH3T3 cells using TRIzol (Takara, Shiga, Japan). Sequencing libraries were generated using the NEBNext Ultra TM RNA Library Prep Kit for Illumina (catalog no. E7530L, New England Biolabs) following the manufacturer's recommendations, and index codes were added to attribute sequences to each sample. The clustering of the index-coded samples was performed on a cBot cluster generation system using HiSeq PE Cluster Kit v4-cBot-HS (Illumina) according to the manufacturer's instructions. After cluster generation, the libraries were sequenced on an Illumina Hiseq 4000 platform, and 150-bp paired-end reads were generated. Sequencing data were processed using HTSseq online software from the Bioinformatics and Biostatistics Core Facility (EPFL). Heat maps were produced from normalized expression data using Cluster 3.0 for computation and Java Treeview for visualization.

GO and GSEA analyses
Gene transcripts expressed in NIH3T3 cells were analyzed with PANTHER (http://www.pantherdb.org). 4 Using PANTHER, protein classification was performed according to the ontological term of molecular function. For PANTHER analysis, we used statistics overrepresentation (i.e. the default setting) to com-pare classifications of multiple clusters of lists with a reference list to statistically identify the over-or underrepresentation of PANTHER ontologies. GSEA was applied to run on a list of genes ranked using the metric of "Diff of Classes." The genes contained in the SASP gene set were collected from previously published articles (16,36). The KEGG gene sets were obtained from the Molecular Signatures Database version 6.1 (http:// software.broadinstitute.org/gsea/msigdb) (37). 4

cGAMP quantitation by LC-MS/MS
cGAMP extraction was performed as reported (7), with slight modifications. NIH3T3 cells were cultured in 10-cm dishes (4 ϫ 10 6 cells), and the culture media were removed and replaced with 2 ml of 80:20 methanol/water. The dishes were incubated at Ϫ80°C overnight to promote protein precipitation, scraped, and transferred to 2-ml centrifuge tubes. Samples were subjected to three vortex, freeze/thaw cycles in liquid nitrogen, sonicated in an ice water bath at full power for 3 min, and clarified by centrifugation at 20,000 ϫ g for 20 min at 4°C, and the supernatant was concentrated by a bench top evaporator and subjected to LC-MS/MS. The pellets from the centrifugation step were dissolved in 1% (w/v) SDS followed by sonication, and the concentration of proteins was measured by a Pierce BCA protein assay kit, allowing normalization to total proteins. Dried supernatants were resuspended in 100 l of 0.1% (v/v) formic acid in water and subjected to solid-phase extraction. Liquid chromatography separation was performed using an AB QTPAP 5500 (Sciex) and ACQUITY UPLC BEH C18ϩ column (Waters, 1.7 m). Samples were maintained at 4°C, and the injection volume was 5 l. The aqueous mobile phase (A) was 0.1% formic acid in water, and the organic mobile phase (B) was acetonitrile. Analytes were eluted using the following gradient: 60% (v/v) A and 40% (v/v) B maintained over 1 min at 500 l/min. Ionization source parameters were optimized using a 2Ј-3Ј cGAMP standard (catalog no. tlrl-nacga23-02, InvivoGen, France) and set to a positive mode. Source parameters were as follows: spray voltage, 5500 V; vaporizer temperature, 500°C; sheath gas pressure, 60 psi; auxiliary gas pressure, 60 p.s.i. Individual reactions monitored and collision energies (CEs) were as follows: cGAMP m/z 675.1 3 m/z 524.0 (CE ϭ 30 V), m/z 312.0 (CE ϭ 30 V). Data analysis was performed using Xcalibur software (AB Sciex) and Prism (GraphPad).

Cytosolic dsDNA detection
For quantification of dsDNA, nuclear, cytosolic, and mitochondrial fractions were prepared using a mitochondria isolation kit (catalog no. 89874, Thermo Scientific, San Jose, CA) from 1 ϫ 10 7 cells. DNA from cytosol was purified by a DNA minikit (catalog no. DP316, TIANGEN, Beijing, China) and quantified by a QuantiFluor dsDNA kit (catalog no. E2670, Pro-mega, Madison, IL) using a Fluromax-4 spectrofluorometer (HORIBA Scientific). Every experiment was repeated at least three times.

In vitro assay for DNA degradation capacity of lysosomes
This assay basically followed a previous study (32). Briefly, lysosomes were isolated from ϳ1.0 ϫ 10 7 NIH3T3 cells using a lysosome isolation kit (catalog no. CMS10118.1, GENMED SCI-ENTIFICS, Inc., Wilmington, DE). For monitoring DNA degradation, isolated lysosomes (30 g of protein) were incubated with 0.5 g of purified plasmid DNA (pcDNA3.1) at 37°C for 5 min in 30 l of 0.3 M sucrose containing 10 mM MOPS buffer (pH 8.0) with an energy-regenerating system (10 mM ATP, 10 mM MgCl 2 , 2 mM phosphocreatine, and 50 g/ml creatine phosphokinase). After the incubation, the levels of DNA associated with the precipitated lysosomes (including DNA inside the lysosomes) were analyzed. DNA in agarose gels was detected using ExRed (catalog no. 7AA07D, ZOMANBIO, Wuhan, China).

siRNA transfection
Cells were transfected with siRNAs using JetPRIME transfection reagent according to the manufacturer's instructions. siRNA oligonucleotides were purchased from Sangon Biotech, and the sense strand sequences targeting the following mouse genes were as follows: m-cGAS-1, GGAUUGAGCUACAAGA-AUATT; m-cGAS-2, GCUGUAACACUUCUUAUCATT.

Immunofluorescence and quantification
Immunofluorescence was performed as described previously (31). In brief, cells were fixed in 4% (v/v) paraformaldehyde in PBS for 20 min at room temperature. Cells were washed twice with PBS and permeabilized with 0.5% (v/v) Triton X-100 in PBS for 15 min. After washing two times with PBS, cells were blocked in 5% BSA in PBS for 1 h at room temperature. Cells were then incubated with primary antibodies in 5% (w/v) BSA in PBS supplemented with 0.1% (v/v) Tween 20 (PBST) overnight at 4°C. The next day, the cells were washed four times with PBST, each for 10 min, followed by incubation with secondary antibody, in 5% (w/v) BSA/PBST for 1 h at room temperature. The cells were then washed four times in PBST, incubated with 1 g/ml DAPI in PBS for 10 min, and washed twice with PBS. The slides were imaged with a Nikon AIRMPϩ fluorescent confocal microscope. Cells with more than five visible spots at the expected location were considered positive. Over 200 cells from four randomly selected fields were analyzed. The following antibodies were used for immunofluorescence: 8-OHdG (1:300), ␥H2AX (Ser-139) (1:500), lamin B1 (1:200), LC3 (1:500), SQSTM1/p62 (1:500), and Lamp1 (1:500). Secondary antibodies, anti-mouse Alexa Fluor 594 (catalog no. A11005) and anti-rabbit Alexa Fluor 488 (catalog no. A11008) (Life Technologies, Inc., Carlsbad, CA), were used at 1:1,000 dilution. Every experiment was repeated at least three times.

Immunoblotting
Western blotting was performed as described previously (30). Whole-cell lysates were extracted in radioimmune precipitation assay lysis buffer (50 mM Tris-base, 1.0 mM EDTA, 150 mM NaCl, 0.1% (w/v) SDS, 1% (v/v) Triton X-100, 1% (w/v) sodium deoxycholate, 1 mM phenylmethylsulfonyl fluoride) supplemented with phosphatase inhibitor mixture (Thermo Scientific). The lysates were rotated at 4°C for 30 min, and supernatants were loaded. The protein concentration was determined by the BCA method (catalog no. CW20011S, CWBIO, Beijing, China). For immunoblotting of STING dimer, reducing reagent was not added to the lysates before loading. The lysates containing 30 g of proteins were loaded on the SDS-polyacrylamide gel and separated by electrophoresis, followed by blotting on a polyvinylidene difluoride membrane (catalog no. IPVH00010, Millipore, Germany). The target proteins were probed by corresponding primary antibodies with optimized conditions and then incubated with the secondary antibody. Immunological signals were surveyed via an electrochemical luminescence method, using the Immobilon Western Chemiluminescence horseradish peroxidase substrate kit (catalog no. 2023, Bio-Rad) and Fusion Solo Imaging System. The band intensities were quantified by FUSION-CAPT analysis software. Every experiment was repeated at least three times, and representative data are shown.

qRT-PCR
Total RNA was isolated from cultured cells using TRIzol (Takara, Shiga, Japan), and 1 g of total RNA was used for reverse transcription by cDNA Synthesis Super Mix (catalog no. B24403, Biotool, Shanghai, China), and then qRT-PCR was performed using SYBR Green qPCR Master Mix (catalog no. B21203, Biotool). PCRs were performed in triplicate, and the relative amount of cDNA was calculated by the comparative CT method using the 18S rRNA sequences as the control. The primer sequences used for qRT-PCR are shown in Table S1. Experiments were repeated three times.

Statistical analysis
Data are expressed as means Ϯ S.D. from at least three biological replicates. Statistical analysis was performed using Prism (GraphPad Software Inc.). The difference between control and treated samples was examined using Student's t test. The difference between multiple groups was examined by oneway analysis of variance with Bonferroni post hoc test.