Longevity-Associated Transcription Factor ATF7 Promotes Healthspan by Suppressing Cellular Senescence and Systematic Inflammation

Aging is characterized by persistent low-grade systematic inflammation, which is largely responsible for the occurrence of various age-associated diseases. We and others have previously reported that long-lived people (such as centenarians) can delay the onset of or even escape certain major age-related diseases. Here, by screening blood transcriptome and inflammatory profiles, we found that long-lived individuals had a relatively lower inflammation level (IL6, TNFα), accompanied by up-regulation of activating transcription factor 7 (ATF7). Interestingly, ATF7 expression was gradually reduced during cellular senescence. Loss of ATF7 induced cellular senescence, while overexpression delayed senescence progress and senescence-associated secretory phenotype (SASP) secretion. We showed that the anti-senescence effects of ATF7 were achieved by inhibiting nuclear factor kappa B (NF-κB) signaling and increasing histone H3K9 dimethylation (H3K9me2). In Caenorhabditis elegans, ATF7 overexpression significantly suppressed aging biomarkers and extended lifespan. Our findings suggest that ATF7 is a longevity-promoting factor that lowers cellular senescence and inflammation in long-lived individuals.


INTRODUCTION
Aging is a major risk factor for several diseases that limit health span, including neurodegenerative diseases, metabolic syndromes, cancer, and cardiovascular disease [1]. Persistent low-level chronic inflammation associated with aging is largely responsible for these diseases [2,3].
Inhibiting cellular senescence and/or age-related inflammation may delay aging and slow the development of age-associated diseases. For example, selective elimination of SnCs (senolytic), or suppression of SASP secretion (senomorphic) via transgenic or pharmaceutical strategies can slow aging and age-related diseases, thereby extending healthy lifespan [6].
Although aging is an irreversible process with progressive loss of physiological integrity, impaired function, and increased vulnerability to death, some longlived species display a delayed aging phenotype, including long-lived humans [7]. As a paradigm of successful human aging, long-lived individuals (LLIs) can reach an extremely advanced age (>90 years) without developing serious age-related diseases, thus living a healthier and longer lifespan [8]. We previously reported that LLIs exhibit enhanced autophagy-lysosomal activity [9], lower ribosomal biogenesis [10], and superior metabolic profiles [11,12], which may contribute to their healthy aging. Levels of inflammatory signaling increase during aging, which is linked to many age-related diseases [2]. However, centenarian cells show an antiinflammatory phenotype [13]. Recently, Sayed et al. constructed an inflammatory clock of aging and found that centenarians had a lower inflammatory clock index associated with longevity [5]. In the current study, we found that LLIs displayed lower levels of proinflammatory factors such as IL6 and TNFα. However, how long-lived centenarians gain these survival and healthy aging advantages remains unclear. Based on unbiased transcriptomic and interaction analyses, we identified activating transcription factor 7 (ATF7) as a potential contributor to the survival advantage of LLIs. ATF7 can bind to the cyclic AMP response element (CRE), which belongs to the ATF/CREB superfamily [14][15][16]. ATF7 activity is growth-and stress-initiated, and its transcriptional function strictly requires phosphorylation by mitogen-activated protein kinases (MAPKs) [17]. In the absence of stress, ATF7 represses the transcription of its target genes, including those encoding factors involved in innate immunity, by recruiting H3K9 di-and trimethyltransferases G9a, EDET/SET-DB1, and Suv39h1 [18,19]. In the presence of stress, ATF7 is phosphorylated by p38MAPK, leading to the release of ATF7 and methyltransferases from target genes, thereby activating transcription [19]. During aging, the relatively preserved innate immune response overwhelms the more variable adaptive immune response, leading to higher production of proinflammatory mediators [20]. These circulating proinflammatory factors can, in turn, keep the immune system in a state of chronic low-level activation [3].
In this study, we found that relatively low levels of inflammation were accompanied by an up-regulation of ATF7 in LLIs. Intriguingly, ATF7 was not upregulated during the process of cellular aging, but rather was decreased. We demonstrated that ATF7 can suppress cellular senescence and systematic inflammation by modulating nuclear factor kappa B (NF-κB) signaling and dimethylation levels of histone H3K9 (H3K9me2). We also demonstrated that ATF7 extended the lifespan of Caenorhabditis elegans (C. elegans). These findings suggest that ATF7 may be an important contributor to longevity by suppressing cellular aging and systematic inflammation.

Study subjects, RNA sequencing (RNA-seq), network construction, and centrality analysis
As described in our prior study [10], we sampled 135 females from longevous families in Hainan Province, southern China, and constructed 76 and 59 sequencing libraries using poly(A) capture and ribosomal RNA depletion methods, respectively. The RNA-seq reads were filtered to obtain clean reads [21], aligned to the human (hg19) and C. elegans (WBcel235) reference genomes [22], and quantified to determine gene abundance [23]. Differentially expressed gene (DEG) analysis was performed using the 'limma' package [24], with the effects of library type and cell composition considered (Gene expression ~ 0 + Group + Lymph + Neu + Mid). The 'removeBatchEffect' function in 'limma' was used to remove the effect of library type on longevous family transcriptome data. The DESeq2 function in the R platform was used to detect DEGs between different experimental groups. A parameter false discovery rate (FDR) cutoff of less than 0.05 was considered statistically significant. Pearson correlation coefficient (r) was used to determine correlations between differently expressed transcription factors (TFs) and immune-related genes, with a Benjamini-Hochberg adjusted P-value cutoff of less than 0.05. The immune-related gene list was downloaded from the ImmPort database (www.immport. org/home) and TFs were obtained from the footprintDB and TRRUST databases (http://floresta.eead.csic.es/ footprintdb/index.php; www.grnpedia.org/trrust/). Coexpression network plots were constructed using Cytoscape (v3.9.0) [25] and centrality analysis was conducted using the cytoHubba program in Cytoscape. Functional enrichment analysis was performed using "Metascape" [26] and "DAVID" [27,28]. Raw RNA-seq data of replicative senescent human embryonic lung fibroblasts (WI-38), human neonatal foreskin fibroblasts (HFF and BJ), and oncogene-induced senescent human embryonic lung fibroblasts (IMR-90) were collected from the ''GEO Repository'' (www.ncbi.nlm.nih.gov/geo/) (accession numbers GSE63577, GSE64553, GSE61130).

Senescence-associated beta-galactosidase (SA-β-gal) staining
SA-β-gal staining was performed using a Senescence Cells Histochemical Staining Kit (CS0030-1kt, Sigma-Aldrich, USA) according to the manufacturer's instructions. Briefly, cells were washed with phosphatebuffered saline (PBS) and fixed at room temperature for 7 min, then incubated in staining working solution overnight at 37 °C. Cells were visualized using a Nikon eclipse Ti inverted microscope and a minimum of 200 cells were counted.

Quantitative real-time polymerase chain reaction (qRT-PCR)
Total RNA was extracted using TRIzol reagent (15596018, Invitrogen, USA) and reverse transcribed into cDNA using a RevertAid First Strand cDNA Synthesis Kit (K1622, Thermo Fisher Scientific, USA) according to the manufacturer's protocols. qRT-PCR was performed using 2 × Tsingke® Master qPCR Mix (TSE201, Tsingke, China), with actin as the endogenous normalization control. Gene expression levels were calculated using the 2 −ΔΔCt method [31]. All primers used are presented in Supplementary Table 1.

Western blotting
Cells were lysed using RIPA lysis buffer (P0013B, Beyotime, China) supplemented with 1% phenylmethylsulfonyl fluoride (ST506, Beyotime, China) on ice. Protein extract (30 μg) was separated by sodium dodecyl-sulfate polyacrylamide gel electrophoresis (SDS-PAGE) and subsequently transferred to polyvinylidene fluoride (PVDF) membranes (162-0177, Bio-Rad, USA). The membranes were incubated with appropriate primary antibodies at 4 ℃ overnight, then with horseradish peroxidase (HRP)-labeled goat anti-rabbit or anti-mouse immunoglobulin G (IgG) (H + L) secondary antibodies at room temperature for 1 h. Protein bands were visualized using an ECL Detection Kit (UEL-S6009L, US Everbright, China). Quantification of blots was performed using ImageJ software. Detailed information on primary antibodies used in western blotting is provided in Supplementary Table 2.

EdU incorporation assay
Cell proliferation was analyzed using a Cell-Light EdU Apollo567 In Vitro Kit (C10310-1, Ribobio, China). Briefly, cells (8 × 10 3 ) were seeded into each well of a 96well plate and incubated for 24 h with 5 μM EdU at 37 °C. The cells were then added with 1 × reaction buffer for 30 min. Images were captured using a Cytation 5 Cell Imaging Multi-Mode Reader (BioTek).

Cellular reactive oxygen species (ROS) detection
Cellular ROS detection was performed using CellROX® Oxidative Stress Reagent (C10422, Thermo Fisher Scientific, USA) according to the manufacturer's instructions. Briefly, cells (8 × 10 3 ) were seeded into each well of a 96-well plate, followed by the addition of CellROX® Reagent at a final concentration of 10 μM and incubation for 30 min at 37 °C. Images were captured using a Cytation 5 Cell Imaging Multi-Mode Reader (BioTek). Quantification was performed using ImageJ software.

Chromatin immunoprecipitation (ChIP) analysis
ChIP analysis was performed using a SimpleChIP® Enzymatic Chromatin IP Kit (9003, Cell Signaling Technology, USA) according to the manufacturer's instructions. Chromatin was crosslinked with 1% formaldehyde in cell medium at room temperature for 10 min. Glycine was then added to the medium to terminate the reaction. After washing in ice-cold PBS, cells were lysed and sonicated into 150-900-bp fragments. Immunoprecipitation was performed overnight at 4 °C with anti-histone H3K9me2 and anti-histone H3. Normal rabbit IgG was used as a negative control. The immunocomplexes were washed using ChIP-Grade Protein G Magnetic Beads (9006, Cell Signaling Technology, USA) and incubated at 65 ℃ overnight following the addition of 5 M NaCl and Proteinase K to reverse the crosslinks. Free precipitated DNA was further purified. The DNA samples were used for qRT-PCR. All primers are presented in Supplementary Table 3.

Co-immunoprecipitation
The HDF cells were lysed in IP-lysis buffer (50 mM Tris-HCl, 150 nM NaCl, 1 mM ethylenediaminetetraacetic acid, 0.2% NP40, and 10% glycerol) for 30 min on ice. After centrifugation at 12 500 rpm and 4 °C for 15 min, the supernatants were incubated in anti-ATF7 or control IgG. The immunocomplexes were then collected using Protein-A-agarose beads (11134515001, Roche, Switzerland) and subjected to western blotting with anti-G9a antibodies.

Cleavage Under Targets and Tagmentation (CUT&Tag) assay
CUT&Tag assay was performed using a Hyperactive ® Universal CUT&Tag Assay Kit for Illumina (TD903, Vazyme, China) according to the manufacturer's instructions. Briefly, 1 × 10 5 cells were collected, washed in wash buffer, and added to 10 μL of activated ConA Beads. Bead-bound cells were incubated overnight at 4 °C with anti-histone H3K9me2. Normal rabbit IgG was used as a negative control. The cells were washed and incubated with rabbit secondary antibodies at room temperature for 1 h. After washing, the cells were incubated with pA/G-Tnp at room temperature for 1 h, then with TruePrep Tagment Buffer L (TTBL) and Dig-300 buffer for fragmentation at 37 °C for 1 h. The released chromatin fragments were extracted, and libraries were generated using TruePrep® Index Kit V2 for Illumina (TD202, Vazyme, China). Reads were aligned to the human reference genome hg19 using Bowtie2 (v.2.4.1) [32]. The 'picard.jar' program in GATK (v4.1.8.0) was used to remove duplicated reads [33]. For each replicate, peaks were first called using MACS2 (v.2.1.2) [34] with the parameters: -q 0.05 -B. Differential binding analysis of peaks was performed using the Bioconductor package 'DiffBind'. The 'annotatedPeak' function in the R package 'CHIPseeker' [35] was used to annotate peaks. Peaks were visualized using Integrative Genomics Viewer (IGV) [36].

Lifespan assay
All C. elegans lifespan experiments were conducted at 20 ℃ in accordance with standard protocols using HT115/RNA interference (RNAi) (with addition of Amp). Briefly, worms were branched to obtain synchronized eggs, which were cultured to hatch and used for lifespan assays. At L4 molting, animals were transferred to plates containing 20 M 5-fluoro-2'deoxyuridine (FUDR), which kills their progeny at the embryo stage but does not significantly affect lifespan. The plates were refreshed daily. Approximately 50 worms were used for each treatment. The first day of adulthood was recorded as day 1 and dead worms were counted from day 5. Worms that did not respond to nose touch were picked out of the bacterial lawn, and worms that did not move were scored as dead.

Body bending assay
Worms were pretreated as in the lifespan assays. On the day of examination, 100 μL of M9 buffer was pipetted onto the surface of an empty 35-mm NGM plate. A single worm was added to the buffer and allowed 30 s to recover from the transfer. The number of thrashing movements over a 30-s period was counted (four times per worm, 15-20 worms per treatment). One worm from one treatment was counted, then one worm from another treatment was counted, and so on, with the cycle repeated to avoid environmental or operational interference.

Lipofuscin accumulation
Autofluorescence of intestinal lipofuscin was measured as an index of senescence in day 10 adults. Worms were randomly selected from each bacterial lawn and washed three times with M9 buffer. The worms were then placed on 5% agar pads coated with 10 mM sodium azide in M9 buffer to initiate paralysis. Lipofuscin autofluorescence images were taken using blue excitation light (405−488 nm) with 4',6-diamidino-2-phenylindole (DAPI) and laser confocal scanning microscopy. Fluorescence was quantified using ImageJ software to determine lipofuscin levels. Three independent experiments were performed with more than 30 worms for each bacterial species each day.

Pharyngeal pumping
Wild-type and ATF7-overexpressing worms were synchronized by timed oviposition onto an empty vector (control) and ATF7 double-stranded RNA (dsRNA)expressing HT115. Day 1 adult worms were transferred to control and RNAi plates with FUDR. Pumping rates were determined on day 10 of adulthood. Terminal pharyngeal bulb pumps were counted at 30-s intervals for each worm on the bacterial lawn. The pumping rate per minute was averaged for 10 worms per treatment. Pumping rates were assessed in three independent replicate experiments.

RNAi in C. elegans
RNAi ATF7 was generated using primers GAATTCC TGCAGCCCCGCGTCCAATTGATGGGC and ACGC GTGGATCCCCCCACAGATAACAATTTGGAAATG TCC and linked to L4440 by infusion (SmaI). The E. coli strain HT115 (DE3) expressing ATF7 dsRNA was grown overnight at 37 °C in Luria-Bertani (LB) broth containing 100 μg/mL ampicillin, then spread on nematode growth medium (NGM) plates containing 100 μg/mL ampicillin and 1 mM isopropyl 1-thio-β-D-galactopyranoside (IPTG). The RNAi-expressing bacteria were then grown at 25 °C overnight. Synchronized L4 larvae were placed on the plates at 20 °C. RNAi was induced from L4 unless otherwise noted. L4 adult animals were used for further experiments.

Statistical analysis
Statistical analyses were conducted using GraphPad Prism v9 (GraphPad Software). Data are presented as means ± standard error of the mean (SEM). Normal distribution of data was determined using the Shapiro-Wilk test. Comparisons were performed using two-tailed Student's t-test or Mann-Whitney U test. Survival curves were compared, and P-values were calculated using logrank (Mantel-Cox) analysis. P < 0.05 was defined as statistically significant.

Centenarians display lower inflammation with possible ATF7 involvement
Peripheral blood mononuclear cells (PBMCs) from centenarians show reduced IL6 mRNA levels [13]. To determine overall gene expression changes in proinflammatory factors in LLIs, we analyzed RNA-seq data in a cohort of longevous families [10]. In total, 2 870 significant DEGs were identified between LLIs and younger controls (YCs) (P < 0.05) (Fig. 1A). Among the DEGs, a portion of major proinflammatory factors, including IL6, CXCL8 (IL8), TNF, IGFBP4, and IGFBP6 [37][38][39], were down-regulated in LLIs compared to YCs (Fig. 1B). Consistent with the RNA-seq results, cytokine array showed that LLIs had significantly lower levels of MMP9 and CCL4 (MIP1b) and lower levels of proinflammatory factors, including IL6, IL1A, and TNFα, compared to the sex-matched YCs (Fig. 1C). As proinflammatory factors are primarily regulated at the transcriptional level, we focused on differentially expressed TF genes that may have both anti-inflammatory and potentially anti-aging functions. We performed network centrality analysis to prioritize candidate TF genes [40,41], and identified five TF genes related to inflammatory regulation (Fig. 1D). Among them, ELF1, IRF3, ZNF410, and ARNT may promote the development of inflammation or senescence [42][43][44][45]. ATF7 not only plays a repressive role in innate immunity [18], but also responds to intrinsic and extrinsic stresses [17], and thus was chosen as the candidate gene (Fig. 1B, D). We confirmed the expression level of ATF7 in LLIs and YCs using qRT-PCR (Supplementary Fig. 1A) and explored its potential roles and mechanisms in the following experiments.

Manipulation of ATF7 alters cellular senescence progression
As ATF7 was upregulated in LLIs compared to YCs, we investigated whether it affects cellular senescence, one of the major hallmarks of aging. Unlike the case in LLIs, the protein level of ATF7 was progressively down-regulated during serial passage in HDF cells, accompanied by the up-regulation of p16 (CDKN2A), a marker of cellular senescence ( Fig. 2A). Similarly, ATF7 was down-regulated in senescent WI-38 fibroblasts, HFF and BJ fibroblasts, and IMR-90 fibroblasts compared to nonsenescent cells (Supplementary Fig. 1B) (online published data) [46][47][48]. The downregulation of ATF7 was partly due to increased phosphorylation by p38MAPK ( Supplementary Fig. 2A), which can cause instability and degradation [49,50]. Notably, we found that application of SB203580, an inhibitor of p38MAPK, decreased the degradation of the ATF7 protein ( Supplementary Fig. 2B).  Next, we examined the effect of ATF7 on HDF senescence. Overexpression of ATF7 significantly decreased senescence markers, including p16 (CDKN2A) protein levels and SA-β-gal-staining positive cells, and increased H3K9me2 levels compared to control HDFs ( Fig. 2B-C). In contrast, knockdown of ATF7 with shRNA reversed the above senescence markers (Fig. 2D-F) and inhibited DNA synthesis indicated by EdU staining (Fig.  2G). These results were validated in the IMR-90 fibroblasts ( Supplementary Fig. 3A-B), suggesting a suppressive role of ATF7 in cellular senescence. As DNA damage and ROS are primarily responsible for cellular senescence, we tested ROS levels in ATF7-overexpressed or down-regulated cells. As shown in Fig. 2H, compared to untreated control HDFs, overexpression and downregulation of ATF7 significantly decreased and increased ROS levels in HDFs, respectively.

ATF7 represses secretion of SASP
SnCs affect surrounding cells/tissues by secreting SASP, which consists of inflammatory factors, extracellular modifiers, and growth factors [51]. To test whether ATF7 regulates SASP and to identify the pathways involved in senescence following ATF7 manipulation, we performed transcriptome analysis of ATF7-overexpressing and control HDFs. In total, 827 DEGs were identified, including 534 down-regulated and 293 up-regulated DEGs in the ATF7-overexpressing HDFs relative to the control cells. KEGG pathway enrichment analysis indicated that the down-regulated DEGs were enriched in inflammatory pathways, including cytokine-cytokine receptor interaction, PI3K-Akt signaling pathway, MAPK signaling pathway, JAK-STAT signaling pathway, and NF-κB signaling pathway (Fig. 3A). Up-regulated DEGs were enriched in the TGF-β signaling pathway (Fig. 3B), among which ID1, ID2, ID3, ID4, SMAD6, SMAD7, and SMAD9 are negative regulators of inflammation [52][53][54][55]. These findings suggest that ATF7 may play an inhibitory role in modulating HDF inflammation. SnC-secreted SASP is responsible for multiple age-related pathologies [56]. Based on transcriptome data, 13 down-regulated SASP factors, together with upstream inflammationregulatory genes (e.g., STAT3, NFKB2, and RELB), were confirmed using qRT-PCR (Fig. 3C, D). In addition, ATF7 overexpression significantly reduced the mRNA expression levels of IL8 and GM-CSF (Fig. 3D). In contrast, compared to the control cells, knockdown of ATF7 up-regulated several SASP factors and downregulated genes in the TGF-β signaling pathway, i.e., ID1, ID2, SMAD6, and SMAD7 ( Supplementary Fig. S4A-C). To verify the inhibitory role of ATF7 on SASP, we analyzed SASP release in culture medium using antibody arrays. As shown in Fig. 3E, a panel of inflammatory factors, including IL6, IL8, TNFα, GM-CSF, IL2, IL5, and IFNγ, were significantly decreased in ATF7overexpressing HDFs compared to the controls. (B) Qualification results of (A). (C) Co-immunoprecipitation of ATF7 and G9a. Cell lysates of HDFs were immunoprecipitated with anti-ATF7 antibody or control IgG, with immunocomplexes then subjected to western blotting using G9a (n = 2 biological replicates). (D) ChIP assay with qRT-PCR detection of H3K9me2 abundance in SASP genes at ATF7-binding sites of promoter regions in control and ATF7-overexpressing HDFs (n = 4 biological replicates for IL6 and p16, n = 5 biological replicates for other genes). OE, overexpression. Statistical analyses were performed using two-sided Student's t-test.
Notably, the release of TNFα, GM-CSF, and IL2 was markedly inhibited (>50% reduction) and IL5 was completely inhibited in the HDFs with ATF7 overexpression relative to the control cells (Fig. 3E). The effect of ATF7 on suppressing SASP in SnCs mimicked that in LLIs, in which ATF7 up-regulation was accompanied by lower levels of SASP factors compared to YCs (Fig. 1B-C). These findings suggest that ATF7 can repress cellular senescence and SASP secretion in SnCs and may contribute, at least partially, to lower inflammation in LLIs. analysis of N2 and ATF7-overexpressing or RNAi worms exposed to P. aeruginosa PA14. Worms were fed with PA14 for 2 days (g) or 10 days (h) after reaching adulthood (n = 3 biological replicates/group). OE, overexpression. Statistical analyses were performed using two-sided Student's t-test.

ATF7 restrains SASP by inhibiting NF-κB signaling and modulating H3K9me2
To determine how ATF7 inhibits SASP, we focused on the inflammatory pathways enriched in KEGG analysis.
Based on immunoblotting, we found that the phosphorylation level of P65, a canonical indicator of NF-κB signaling activity, was significantly decreased in ATF7-overexpressing HDFs compared to control HDFs ( Supplementary Fig. S5A). NF-κB is a recognized regulator of SASP production [6,57]. Therefore, we investigated upstream signals in this pathway and found that the phosphorylation levels of P65 and IKKα/β were diminished in ATF7-overexpressed cells compared to the control cells, whereas the total protein levels of P65 and IKKα/β remained unchanged (Fig. 4A, B).
In addition to its role in regulating NF-κB signaling, ATF7 is also reported to regulate inflammatory gene expression epigenetically [18,50]. As shown in Fig. 2B, ATF7 overexpression up-regulated H3K9me2, which negatively regulates SASP factors IL6 and IL8 [58]. To investigate whether ATF7 regulates SASP through H3K9me2, we performed chromatin immunoprecipitation plus qRT-PCR analysis using H3K9me2 antibodies. As shown in Fig. 4C and Supplementary Fig. 5B, coimmunoprecipitation revealed that ATF7 could directly bind G9a, which regulates H3K9 dimethylation. These results suggest that ATF7 may recruit G9a to regulate H3K9me2 and silence its target genes, as reported previously [18,50]. The ChIP-qRT-PCR analysis also showed that ATF7 overexpression up-regulated H3K9me2 levels around the promotor regions of several SASP factors, including IL8, IL6, CXCL1, CXCL2, CXCL3, CCL8, and TNF, as well as the upstream regulator STAT1, but not p16 (Fig. 4D). These data were validated using the CUT&Tag assay, which showed a high signal-to-noise ratio ( Supplementary Fig. 6) [59]. Thus, these results suggest that ATF7 suppresses SASP via both NF-κB and H3K9me2.

Manipulation of ATF7 affects lifespan and survival in C. elegans
As overexpression of ATF7 delayed cellular senescence, suppressed SASP production in vitro, and was associated with longevity in LLIs, we wondered whether it may function to promote lifespan. We tested this using C. elegans, a model species widely used to determine molecular mechanisms of lifespan [60]. As shown in Fig.  5, transgenic ATF7 overexpression extended the C. elegans lifespan by 10.3% compared to the control, whereas RNAi knockdown of ATF7 decreased lifespan by 17.2% (Fig. 5B). Consistently, compared to the control, overexpression of ATF7 reduced intestinal lipofuscin and promoted body locomotion and pharyngeal pumping (Fig.  5C, D), applied as aging markers in C. elegans [61]. In contrast, knockdown of ATF7 accelerated aging in C. elegans compared to the control group (Fig. 5C, D).
To explore how ATF7 regulates lifespan and aging in C. elegans, we performed RNA-seq analysis of ATF7overexpressed or silenced C. elegans and control C. elegans. Gene Ontology (GO) enrichment analysis showed that the up-regulated genes in ATF7overexpressed C. elegans relative to control C. elegans were mainly enriched in innate immune response, whereas the down-regulated genes in ATF7-silenced C. elegans were mainly enriched in innate immune response and defense responses against gram-positive bacteria (Fig. 5E, F). The downregulation and up-regulation of innate immune-related genes in ATF7-silenced and overexpressing worms support the regulatory role of ATF7 in innate immunity [18,62,63]. Moreover, most upregulated DEG orthologs in humans play roles in modulating immune function, depending on different physiological/pathophysiological contexts (Supplementary Table 4). Importantly, as ATF7 is closely associated with innate immunity, we suspect it may influence the survival of pathogen-challenged C. elegans. Indeed, in the presence of Pseudomonas aeruginosa (PA14), transgenic ATF7 overexpression enhanced C. elegans survival, whereas RNAi knockdown of ATF7 reduced survival (Fig. 5G, H). Overall, these results suggest that ATF7 can extend the lifespan and promote healthy aging in C. elegans, in part by regulating innate immunity genes.

DISCUSSION
Persistent systemic chronic inflammation can lead to a series of age-related diseases [2,3]. Inflammaging is a well-recognized source of age-related inflammation [3]. As an important contributor to inflammaging in vitro and in vivo, SnC-secreted SASP can cause several age-related diseases by maintaining a proinflammatory internal microenvironment [64]. Accumulating evidence suggests that attenuating SASP production and age-related inflammation can alleviate age-related diseases and extend healthspan [2,[65][66][67]. Our previous study and others have shown that LLIs can delay (or avoid) the onset of major age-related diseases [68][69][70][71]. We therefore wondered whether this survival advantage is achieved by maintaining better levels of inflammation. Here, by screening a panel of inflammatory factors, we found relatively low levels of inflammation in our longevity cohort. We also identified the ATF7 transcriptional factor as a key upstream regulator in maintaining lower inflammation in LLIs. Importantly, manipulation of ATF7 modulated cellular senescence and C. elegans lifespan, suggesting it may be a novel and conserved contributor for longevity across various species, including humans.
In the current study, ATF7 was highly expressed in LLIs, but its expression was gradually lost during cellular senescence, suggesting it may be a longevity-associated factor rather than an aging-related factor. As a transcription repressor, ATF7 is involved in the regulation of cell cycle progression [72,73], heterochromatin formation [16,18,74], telomere shortening [49,75], and innate immunity [18,62,63,76]. Our results showed that overexpression of ATF7 suppressed cellular senescence and various inflammatory factors, indicating a possible role in delaying aging in individuals. Indeed, ATF7 has been reported to affect aging in mice, with its depletion found to promote senescence of embryonic fibroblasts and shorten lifespan [77]. Here, we provide evidence that ATF7 acts as a longevity-promoting factor in humans by inhibiting cellular senescence and SASP secretion. ATF7 influences senescence via three pathways. First, ATF7 represses activation of NF-κB signaling. Second, ATF7 silences target genes, including SASP-related genes, by recruiting histone H3K9 dimethyltransferase G9a. These two pathways synergistically decrease SASP factor production. Third, ATF7 directly inhibits ROS production, thus repressing senescence. These pathways may synergistically lead to lifespan extension in C. elegans and humans.
Cellular senescence can be induced by various stimuli, including DNA damage, oncogene activation, oxidative stress, chemotherapy, mitochondrial dysfunction, and epigenetic changes [78]. Our data showed that ATF7-dependent senescence may be mediated through its regulation of oxidative stress and epigenetic changes (Fig. 6). SASP factors are primarily regulated by transcription [79]. Upstream regulators of SASP, such as p38MAPK, JAK2/3, TGF-β, and PI3K-Akt, converge into two transcriptional mechanisms, i.e., the NF-κB pathway and C/EBP family [6]. While our results showed that ATF7 significantly represses NF-κB activity, whether C/EBP is involved remains to be investigated.
ATF7 is also reported to suppress a subset of genes associated with innate immunity by recruiting the H3K9 dimethyltransferase G9a [18]. For example, high H3K9me2 levels around the promoters of SASP factors IL6 and IL8 have been shown to decrease their induction [58]. Our data indicated that overexpression of ATF7 markedly increased H3K9me2 levels, and ATF7 directly bound to and recruited G9a to dimethylate H3K9, further suppressing genes encoding SASP factors (Fig. 6). These findings suggest that ATF7 plays a role in delaying cellular senescence and inhibiting SASP production, which may function to regulate individual aging. Furthermore, our results showed that overexpression and depletion of ATF7 delayed and accelerated aging, and extended and shortened C. elegans lifespan, respectively. However, the exact mechanism by which ATF7 protects C. elegans from aging has not yet been determined as the immune system of C. elegans differs from that of humans, although preliminary RNA-seq data suggest that innate immune functions may be involved. Taken together, our results suggest that ATF7 plays an important role in promoting healthy aging and longevity by regulating cellular senescence and inflammatory genes (Fig. 6).
In conclusion, we identified an up-regulation of transcription factor ATF7 in LLIs and demonstrated its role in promoting longevity in a model organism. ATF7 inhibited senescence and SASP factors by suppressing NF-κB pathway activity and increasing H3K9 dimethylation, suggesting that it may inhibit persistent systemic chronic inflammation and promote lifespan in LLIs. However, further research is necessary to determine whether ATF7 may be a target for reducing age-related pathology associated with chronic inflammation.