Enhanced insulin‐regulated phagocytic activities support extreme health span and longevity in multiple populations

Abstract The immune system plays a central role in many processes of age‐related disorders and it remains unclear if the innate immune system may play roles in shaping extreme longevity. By an integrated analysis with multiple bulk and single cell transcriptomic, so as DNA methylomic datasets of white blood cells, a previously unappreciated yet commonly activated status of the innate monocyte phagocytic activities is identified. Detailed analyses revealed that the life cycle of these monocytes is enhanced and primed to a M2‐like macrophage phenotype. Functional characterization unexpectedly revealed an insulin‐driven immunometabolic network which supports multiple aspects of phagocytosis. Such reprogramming is associated to a skewed trend of DNA demethylation at the promoter regions of multiple phagocytic genes, so as a direct transcriptional effect induced by nuclear‐localized insulin receptor. Together, these highlighted that preservation of insulin sensitivity is a key to healthy lifespan and extended longevity, via boosting the function of innate immune system in advanced ages.


| INTRODUC TI ON
The rapidly aging population around the world, particularly in developed countries, has urged policy makers to switch the focus of healthcare from delivering adequate treatments to small populations of patients with acute episodes of single diseases to providing sustainable measures for maintaining good health and avoid the development of multiple chronic age-related conditions in the general population (Beard et al., 2016). The immune system plays a central role in many processes of age-related non-communicable diseases such as cardiovascular diseases, type 2 diabetes, and dementia (Rea et al., 2018). Activated immune functions, which frequently describe as inflammation, has been recognized as part of their pathophysiologies (Rea et al., 2018). However, accumulating evidence challenges this assumption and suggests that the immune system may instead get mounting adaptive responses to chronic stressors, prolonging the chances of survival of an organism (Rubinow & Rubinow, 2017;Soo et al., 2022). To address this argument, one possible way is to investigate the immune signatures in long-lived individuals (LLIs; mean age >95 year old) and centenarians, the "aging champions" who achieved successful human aging and exhibited medical histories with remarkably low incidences of common age-related disorders (Borras et al., 2016). Since inflammaging and immunosenescence is a common feature of chronological aging in ordinary people contributing to enhances risks of mortality at advanced age (Borras et al., 2016); this proposes that a better functioning immune system, with stronger pro-survival and stress handling abilities, are likely at play in shaping extreme longevity.
The immune system can be schematically seen as two divisions.
The ancestral/innate arm is mainly represented by monocytes, natural killer (NK) and dendritic cells (DC); whereas the adaptive arm is represented by the B and T lymphocytes. As if a functioning immune system requires a homeostatic balance between the two, gene expression profiles of circulating immune cells would likely reveal important clues that are crucial for achieving healthy aging. A recent single-cell transcriptomic study reported that expansion of cytotoxic CD4 T cells is a unique immune signature among supercentenarians (Hashimoto et al., 2019); whereas previous bulk transcriptome studies proposed that shift in lymphocyte to myeloid cell ratio (Karagiannis et al., 2022), enhanced autophagy-lysosomal function (Xiao et al., 2018); reduction in ribosomal biosynthesis (Xiao et al., 2022), or upregulated apoptotic Bcl-xL (Borras et al., 2016) is however crucial to successful aging. It remains unclear if any common immune features unique to extreme longevity exist among LLIs regardless to their origins; and whether the associated molecular signatures can provide insights for practical translations.
By harnessing the wealth of single-cell and bulk transcriptome datasets available in the public repositories (Table 1); we uncovered that significant induction of innate immune monocytes with enhanced lysosomal and phagocytic activity is a previously unrecognized, common, and unique immune signature among LLIs from various geographical origins and ethnicities. The life cycle of these monocytes in LLIs is enhanced and primed to a M2-like macrophage phenotype. Monocytes are the major immune cells that express insulin receptor (INSR). Functional characterization revealed an insulin-signaling centric immunometabolism network which supports multiple aspects of phagocytosis. Such reprogramming is associated to a skewed trend of DNA demethylation, particularly at the promoter regions of multiple phagocytic genes, so as a direct transcriptional effect induced by the nuclear INSR. Together, these findings highlighted that preservation of insulin sensitivity hence an active innate monocyte-driven phagocytic activity is a defense mechanism in safeguarding healthy lifespan and extended longevity.

| A consensually activated innate immune signature in healthy LLIs
It has been reported that aspects of the aging process could be inferred from the gene expression profiles of peripheral blood cells.
To investigate if common health-related signature exists among LLIs or centenarians of different geographical origins; initial analyses of three independent sets of white blood cells transcriptomic data, obtained from a total of 261 disease-free, healthy LLIs and 127 young controls (i.e., immediate descendant's spouses, F1SP) originated from Cheng Mai (ChM) (Xiao et al., 2018), Lingshui (LS) (Xiao et al., 2022), Lingao (LG) (Xiao et al., 2022) Counties of the Hainan province were performed (Table S1). With the geographical-specific sets of differentially expressed genes (DEGs) (Figure 1a-c); commonly dysregulated DEGs (i.e., cDEGs), including 317 up-and 449 downregulated genes, were identified (Table S2, nominal p < 0.05). Indeed all these commonly deregulated genes were also identified by an alternative way of analysis if samples from all these geographical sites were first integrated together via batch effect correction (Table S2, adjusted p < 0.01). This therefore validated the robustness of these commonly deregulated genes. The non-randomness of the cDEGs was confirmed by a premutation test via a random selection and overlapping estimation of the same number of genes as found in each set of DEGs. Such calculation revealed that a significantly lower number of genes (i.e., only 27 up-and 22 down-regulated) were expected to be common ( Figure S1A). Further comparison was made with the list of DEGs (i.e., 709 up-and 1511 down-regulated) obtained from the nonagenarians in the Netherlands' Leiden (NL) Longevity Study (Passtoors et al., 2012). Around 19.87% of up-and a higher 43.65% of down-regulated genes from the Hainan cDEGs were indeed common to those from the Netherland's dataset (Figure 1d), suggesting a certain degree of similarities in blood profiles exist among these individuals of different ethnicity. Among the commonly upregulated were those that promote M2 polarization [e.g., Macrophage scavenger receptor-1 (MSR1) , ATP-binding cassette subfamily G member-1 (ABCG1) (Sag et al., 2015), sphingomyelin synthase-2 (SGMS2) (Deng et al., 2021) (Rotthier et al., 2010)]; whereas those commonly downregulated were those that promote apoptosis [e.g., caspase-6 (CASP6) (Cowling & Downward, 2002), TNF receptor superfamily member 10a (TNFRSF10A; aka death receptor 4, DR4) (Micheau, 2018)] (Table S3). Next, with a focus on our Hainan discovery cohorts, subsequent gene functional analysis using the Metascape tool revealed that the over-represented biological pathways identified from each independent Hainan DEGs set were also very similar to one another ( Figure S1B, Table S4). Common upregulated pathways concurrently supported the innate immune responses, such as monocyte differentiation, activation of lysosomes, metal ion transport, and cytoskeleton remodeling for phagocytosis. In the contrary, activities related to the adaptive immune response mediated by the T-and B-cells became less robust ( Figure 1e, Figure S1C,D).
As these non-diseased, healthy LLIs or nonagenarians have lived their lives to the extreme of human longevity, it is expected that their immune system are also subjected to changes related to the normal aging process, in addition to the changes underlying extreme longevity. To differentiate between the two, a larger scale analysis was performed with whole blood transcriptome profiles of 755 healthy subjects aged between age 20 and 79 years obtained from the genotype-tissue expression (GTEx) database ( Figure 1f,g, Figure S1E). With a total of 4148 DEGs categorized into 36 groups based on their expression patterns along the age scale ( Figure S1E), 14 groups were selected due to their consensual up-or down-ward trends observed at age 50 years and beyond, plus one group (Group 16) that revealed strong and robust gene upregulation at age 70-79 ( Figure 1f,g). These genes, together (Up: 1166; Down: 1175), were then defined as the "GTEx-common age-associated genes" (Table S5).
For any genes occurred on this list, along with the 1497 common age-associated genes identified from a meta-analysis of six European ancestry (EA) studies (38.2-78.2 year old) (Peters et al., 2015), were filtered out from cDEGs list curated from the Hainan cohort ( Figure 1h), leaving a net of 159 up-and 211 down-regulated genes that truly reflected changes associated with extreme longevity ( Figure 1i, Table S6). Subsequent gene functional characterization by the Metascape tool of this extreme longevity and healthy aging gene list (LHA) identified an insulin signaling-centric immunometabolism network ( Figure S2A). At the metabolic side, insulin signaling facilitates the biosynthesis of sphingolipid (Bryan et al., 2015), amino and nucleotide sugars (Tafesse et al., 2015) that known to facilitate the recognition of foreign particles by macrophage and Data type: Bulk RNA-sequencing, HepG2 Reference citation: PMID 30955890 TA B L E 1 (Continued) neutrophils. At the immune side, the signaling also activates a set of downstream pathways that are known to facilitate phagocytosis, including plasma membrane organization, vesicle-mediated plasma membrane transport, Wnt signaling and tyrosine kinase signaling network ( Figure S2B,C, Table S7).
Mounting evidence had indicated that the immune system adapts and reprograms upon chronic stress via initiating an allostasis response (Rubinow & Rubinow, 2017) that allows physiological recalibration and creates conceptual spaces for adaptive responses that may result in new biological set points in order to support survival (Rubinow & Rubinow, 2017). Such remodeling is frequently found in non-communicable, age-related chronic conditions, including the early mild cognitive decline (MCI) and advanced Alzheimer's disease (AD), type 2 diabetes (T2DM), and coronary artery disease (CAD).
Surprisingly, as compared to their respective age-matched controls,  Figure S3C,D). These patterns suggested that despite being non-diseased and healthy, the peripheral blood profiles of LLIs were pre-set to an activated status resembling to those who had been adapted to tolerate chronic stresses.

| Enhanced lysosomal-phagocytic function in professional phagocytes is associated with healthy lifespans and extreme longevity
The blood contains many subtypes of white blood cells and they could be differentially regulated in the context of aging. To delineate the specific cell-type contribution in the common immune signatures identified among LLIs and Netherland nonagenarians, blood cell type-specific molecular characters were profiled with reference to single-cell transcriptomes (Wilk et al., 2021). Among a total of 61,880 while blood cells explored (Figure 2a), 22 major cell types were identified based on their established markers ( Figure S4) (Wilk et al., 2021). From there, pre-defined gene lists, such as the LHA genes, DEGs list curated from the European ancestry (EA) studies (EA-DEG), and genes clustered in selected groups (Figure 1f,g) were mapped to different cells types (Table S8). As in details, list of downregulated LHA genes and those broadly defining the common age effects ( Table S8).
To further characterize the differentially expressed gene signature patterns in these cells, each gene was assigned with a score based on its relative expression in every single cell, followed by unsupervised clustering of these genes into different gene clusters ( Figure 2d, Table S9). Four major clusters were substantially F I G U R E 1 Conserved longevity and healthy aging (LHA) genes identification across multiple long-lived populations. (a-c) Volcano plots of the differentially expressed genes (DEGs) between long-lived individuals (LLIs) and their children's spouse (F1SP) from Chengmai (a), Lingao (b), and Lingshui (c). Genes with p-values smaller than 0.05 and fold changes >1.2 were regarded as significantly changed (Red: upregulated; Blue: downregulated). (d) Upset plot of the shared genes across the entire Hainan cohort. The commonly up-and down-regulated genes were marked as ChM-LS-LG-cUp and ChM-LS-LG-cDn, respectively. These were compared to those identified from the Netherlands' Leiden (NL) Longevity Study (NLUp: upregulated; NLDn: downregulated). (e) Over-representation analysis (ORA) of common LHA genes identified from (d) using the Metascape platform. Every single node represented an enriched term and two nodes were linked if their Kappa similarities were higher than 0.3. Similar functional terms were clustered together and were displayed using the same color. Node size was proportional to the number of enriched genes. (f-g) Expression pattern plots of the common aging-associated genes in the human whole blood transcriptomes obtained from GTEx database. The upregulated and downregulated common aging-associated genes were manually extracted, if the trend of changes persist uniformly >50 years old (y.o.). (h) Volcano plot of the shared genes across the Hainan cohort. Commonly found age-associated genes identified from the GTEx were colored by red and blue corresponded to the up-and down-regulation during normal aging. Yellow dots points represented the common aging associated-genes identified from the 14,983 individuals of European ancestry (EA) studies. (i) Volcano plot of the LHA genes after eliminating the common aging-associated genes showed in (H).
identified. Majority of the upregulated LHA genes enriched in classic professional phagocytes, including the monocytes and DCs (Cluster 1), so as the lymphocytic NK cells and CD8-positiive cytotoxic T cells (CD8 T) (Cluster 2) whose major protective mechanism is also related to phagocytosis. For the downregulated LHA genes, these were over-represented in both T (Cluster 3) and B cells (Cluster 4) ( Figure 2d). Gene functional characterization of upregulated LHA genes in Cluster 1 using the Metascape tool revealed their predominant roles in mediating phagocytosis at both the cell membrane (i.e., plasma membrane organization, vesicle-mediated transport, sphingolipid metabolism, integrin-mediated signaling) and intracellular levels (i.e., N-glycan biosynthesis, lysosome, RHO GTPase cycle) Table S10).
In contrast to Cluster 1, Cluster 2 is a small cluster, the related genes were more associated with RHO GTPase signaling and microtubule-based process for phagocytic engulfment enriched in NK cells and cytotoxic CD8+ T cells ( Figure 2e, Table S10).

| Activated phagocytic monocytes prime a M2-like alternatively activated macrophage signature in LLIs
Apart from the above qualitative changes, the relative quantities of these immune cell populations were estimated by the CIBERSORTx deconvolution calculation using their respective cell type-specific markers ( Figure S7). Among all the detectable cell types, monocytes ratio was significantly increased among all three groups of Hainan LLIs as compared to their F1SP. In contrary, the reverse trend was found in the Naïve CD4 T cells ( Figure S7A). Indeed, similar downward trend of the latter was observed as well during the normal aging process ( Figure S7B). This inferred that the increase in monocyte cellular ratio, but not the changes in naïve CD4 T cells, was the only quantitative change closely associated to extreme longevity. To further investigate the details of such changes, markers related to events occur during the monocyte life cycle were analyzed. Referencing previous studies which reported that colony-stimulating factor 1 receptor (CSF1R) that promotes proliferation of monocytes, macrophages, and dendritic cells was significantly downregulated during the course of ordinary aging (Duong et al., 2022;Hearps et al., 2012), here this gene was however found to be upregulated among LLIs as compared to their suggested that increased phagocytosis capacity-an M2-like signature-is generally observed in brain infiltrated peripheral macrophages in advanced disease; and that resembles the prediction from the circulating monocytes of the LLIs ( Figure S3). Together, these alternatively confirmed that the innate immune system of the diseasefree individuals who lived to extreme longevity is pre-activated and reprogrammed to withstand chronic stresses.

| Enhanced insulin sensitivity and oxidative phosphorylation is associated with the functional reprogramming of the monocytes
The study of energy expenditure has deep roots in understanding aging and lifespan in all species. Reduced energy production from metabolic active tissues and organ is one of the prominent drivers of aging. Therefore, energy conservation via streamlining the limited resources to sustain essential functions had been developed, for instance a decline in cellular ribosomal biogenesis during the post-developmental phase is a mechanism for promoting healthy aging and longevity (Xiao et al., 2022). Indeed, such energy-saving changes were also observed during the course of aging (Figure 2).
Immune defense is energy demanding, competitions for limited  (Table S11). With an emphasis on the energy-generating OXPHOS pathway, these genes were indeed also highly expressed in monocytes of supercentenarians from Japan (P CD14 = 2.07 e-5, NES CD14 = 1.83, P CD16 = 1.13 e-7, NES CD16 = 1.80) (Figure 4b,c, Table S11), suggesting that is a cellular-specific feature in promoting extreme longevity. To further validate so, transcriptome profiles of monocytes harvested from 1183 individuals of common lifespans ranging from 44 to 83 years old were analyzed (Reynolds et al., 2014).
By the weighted gene co-expression network analysis (WGCNA), a total of 36 modules characterized based various gene expression pattern identified (Figure 4d). Within which 16 modules were significantly correlated to chronological aging (Figure 4e). Genes involved in OXPHOS were indeed clustered in modules whose expression trends were negatively correlated to age in individuals of ordinary life spans (Figure 4f). This validated that the enhanced OXPHOS is a unique feature in monocytes related to extreme longevity.
As hinted from our early functional analysis, the LHA gene set characterizing the signatures both the overall peripheral blood profiles of LLIs indeed highlighted an insulin signaling immunometabolism network ( Figure S2). Insulin signaling has multifaceted stimulatory effects on OXPHOS (Nisr & Affourtit, 2014). This is also realized in insulin resistance-a condition where insulin signaling is compromised-can lead to the opposite effects. Indeed, INSR was Puchalowicz & Rac, 2020). Together, these data suggested that the enhanced phagocytic signatures of the innate immune monocytes and the descendent M2-like macrophages was likely supported by a robust insulin signaling and OXPHOS network, for fulfilling the elevated energy demand from active phagocytosis.

| LLI phagocytic genes are targets of insulin receptor and insulin-regulated transcription factors
Recent study revealed that activated INSR can directly associ-  (Mathys et al., 2019) dataset. Immune cell cluster was further isolated and re-clustered using a higher resolution to dissociate the T cells, peripheral macrophage (P.macrophage), and brain resident-microglia with their respective markers. Ex: excitatory neurons; In: inhibitory neurons; Opc: Oligodendrocyte progenitor cells. (i) Evaluation of the relative expression levels of the LHA genes across the distinct cell types in the human brain. The average gene expression of the LHA genes between cell types was compared and were further clustered to identify the peripheral macrophage-specific expressed LHA genes, as highlighted by the dash box. (j) Tukey boxplots for the peripheral macrophage associated-LHA genes found being (j) up-and (k) down-regulated during the course of cognitive impairment. Values represent median z-score of gene expression. Statistical significance was calculated by one-way ANOVA. Ross & Cantrell, 2018) and DAP12 cytokine signaling network (Tomasello & Vivier, 2005) needed to facilitate the maturation of lymphocyte-lineage of the adaptive immune system was likely deactivated, hinted by the corresponding gene hypermethylation status ( Figure 5e, Table S13). These methylome-based changes therefore alternatively supported findings from the transcriptomic analyses performed in other independent datasets.
With the same approach, a more refined analysis was performed on the genomic regions located at proximity to the curated list of LHA genes, which revealed a total of 119 differentially methylated regions (DMR) (Figure 5f).  ; the Kruppel-like factor-4 (KLF4) that governs monocyte differentiation (Feinberg et al., 2007); and the activating transcription factor-3 (ATF3) that promotes macrophage migration and M2 polarization via suppressing the M1-like fate (Sha et al., 2017) were only ones significantly induced (Figure 5j, Figure S12). Notably, these transcription factors were also direct targets of nuclear INSR, where robust binding sites were identified at their core promoter regions ( Figure 5k). Together, both changes observed in the global DNA methylome, and transcription factor abundance support potential influence of insulin, explaining how it may regulate the innate immune system of LLIs.

| DISCUSS ION
Our study constitutes a holistic analysis, illustrating the changes in immune functional network and molecular signatures associated with extreme health span and longevity common to multiple groups of "aging champions" across continents. By performing an integrated analysis using multiple blood transcriptome and DNA methylome data from subjects of different geographical and ethnicity backgrounds; this has led to an unexpected finding on the insulinsensitive, activated innate immune monocytes, with enhanced lysosomal and phagocytic activities. The findings on the activation of the innate immune monocytes have not been reported before, and such changes are found to be supported by an insulin signalingdriven immunometabolic network that remained sensitive and active among these subjects. The robust insulin action is tailored to the INSR-presenting monocytes, supporting their intrinsic metabolic demands crucial for further differentiation and polarization as M2like macrophages with robust lysosomal and phagocytic activities.
The relationship between immunity and longevity, particularly extreme longevity, has been established (Sansoni et al., 2008).
Here we revealed that the peripheral blood immune signatures of disease-free healthy individuals who lived to extreme longevity were indeed resembling to those who were at much younger age but suffering from various chronic conditions. This observation was in fact counter-intuitive to the main stream assumption that an altered immune signature is part of the pathoetiologies. Indeed, accumulating evidence emerged recently challenges this perspective, rather it is proposed that the immune signatures seen could be a reflection of a continual adaptive responses and physiologic recalibration to chronic stressors, leading to new biological set points that favors organismal survival under stressful conditions (Rubinow & Rubinow, 2017). In lower organisms like Caenorhabditis elegans, a recent study indicated that enhanced stress resistance among long-lived mutants is a key in determining longevity (Soo et al., 2022). Here, our findings also re-emphasize the fact that inflammation as part of the normal immune defenses has been selected through the evolution, its critical role in supporting survival of an organism shall not be overlooked.
From this perspective, here we speculate that the common changes observed in individuals of extreme longevity are indeed pre-set to withstand chronic stress, even before the emergence of chronic diseases. The new set points are associated with a better preservation F I G U R E 4 Enhanced insulin sensitivity and oxidative phosphorylation is associated with the functional reprogramming of the monocytes. (a) Dot plot revealing the relative expression level of genes involved in KEGG metabolic pathways in various immune cells (as identified in Figure 2a). Significance was evaluated using GSEA. Red and blue dots respectively represented a high or low normalized enrichment score (NES). Point size corresponded to the p-value. Significance was considered when p-value was smaller than 0.05, and would be indicated as a dot, or else as a plus symbol. (b-c) Genes involved in oxidative phosphorylation were enriched in both CD14 and CD16 monocytes harvested from Japanese supercentenarians as compared to the corresponding controls. this perspective, the preference in boosting the innate immune function over that of the adaptive system proposes an underlying metabolic advantage during the extreme aging condition.
Differential energy requirements between the two systems have been proposed (McDade et al., 2016). As compared to the opposite trends in the acquired immune system, the innate immune system imposes a lower upfront developmental cost, but higher operating costs when being activated. In disease-free long-living individuals whose body nutritional functions have also simultaneously suffered from an age-related decline, this inevitable condition shall favor the energy investment toward the development of innate immune defenses as a barricade toward unknown encounters of pathogens and infections (McDade et al., 2016). Along the same vein, pathway analysis indeed also suggested a less active ribosomal biosynthesis network among the adaptive immune lineages. This could indeed be metabolically beneficial by sparing the limited fuel metabolites for the rest of the body immune function, as indeed even at resting conditions, nearly half of ribosomal RNA synthesized in the lymphocytes are degraded without even participating in any defense-related protein synthesizing purposes (Cooper, 1970). Moreover, reduced protein synthesis/turnover in ribosomes is a conserved mechanism to alleviate cellular senescence stress in multiple cell types (Xiao et al., 2022). Functionally, monocytes, DCs and macrophages also play important upstream roles in regulating T cell fate as well, via presenting various antigens, secretion of cytokines, and other co-stimulatory signals (Chu et al., 2020).
Phagocytosis is an integral part of the innate immune system.
Immunologically-silent phagocytosis of foreign pathogens and autologous apoptotic cells is crucial to maintaining tissue homeostasis, wound healing, and innate immune balance. While phagocyte dysfunction has been associated with tissue aging and potentially related to age-related diseases; its direct linkage to lifespan and longevity remains elusive, except in one study reported in laboratory mice (Guayerbas & De La Fuente, 2003). Here we provide a direct evidence in human illustrating that enhanced monocyte phagocytic activities is the key immune signature that distinguishes diseasefree aging champions from those who lived to only normal lifespans.
Moreover, our network-based analysis revealed that an activated insulin signaling-centric immunometabolic network is a potential driver underlying the enhanced lysosomal-phagocytic activities in the insulin-sensitive monocytes. Recently, it has been proposed that the innate immune cells are also capable in building immune memory characteristics. This "trained immunity" describes a persistent hyperresponsive phenotype of innate immune cells after brief stimulation (Tercan et al., 2021). In this essence, insulin signals could be a potential underrecognized stimulus, introducing memories in monocytes and macrophages. Such effect is likely maintained by distinct epigenetic and metabolic mechanism that persists for at least months, if not for years due to the associated cellular reprogramming effect.
The effect of insulin on epigenetics has been illustrated reversely by the case of insulin resistance, which is associated with global DNA hypermethylation (Zhao et al., 2012). illustrating the expression trends of insulin-responsive LHA genes after insulin stimulus. Bar plot on the right representing the ratio of insulin-sensitive LHA genes with INSR binding sites to the total number of genes in each clusters (Defined in Figure 2e). (c) Comparison of global methylation profiles from nonagenarians and middle-aged individuals (Heyn et al., 2012). (d) Volcano plot of the global methylation state between nonagerians and middle-aged controls. Entries with absolute delta beta value greater than 0.1 and adjusted p-value smaller than 0.01 were considered as significant. (e) Functional enrichment of the genes adjacent to hypo-and hyper-methylated regions. Genes in proximal to methylated regions were extracted from the HumanMethylation450 annotation file, enrichR web server was used to analyze the hypo and hyper-methylated associated genes. (f) Volcano plot of the differentially methylated regions across the DNA sequence of LHA genes. Differentially methylated regions (DMR) with absolute beta value variation larger than 0.1 and the adjusted p-value smaller than 0.01 were considered as significant. in the Genome Sequence Archive    (Passtoors et al., 2012).
Datasets for defining the GTEx-common age-associated genes-RNA-seq raw counts matrix of the whole blood cells from the healthy person as well as their phenotypes were downloaded from the genotype-tissue expression (GTEx) project (v8). About 1497 chronological age-associated genes identified by individuals of European ancestry were obtained from the corresponding report (Peters et al., 2015).  (>50-year-old) were manually selected as the common aging gene.

| Identification of differentially expressed genes (DEGs) in LLIs
As a result, 15 groups were selected.
To validate the robustness of the manual selection approach, we have also custom-devised a statistical method/metric to identify the age-associated pattern based on the mean value of the scale scores of each age group generated by degPatttern (Table S14). By calculating the slope value using the following formula: For any slope value of each individual group that is greater than "0" means a consistent variation beyond 50 years old. Using this approach, 16 patterns were suggested to be selected (Table S14, orange highlights). This "calculated" group list was indeed in an almost perfect match (14/16 matched) with our "manually" selected list (i.e., Group name highlighted in red in Table S14 for easy comparison), except only three clusters, i.e., Groups 40, 32, and 16.
For Group 40, it was selected from the calculation, yet it was not included in our analysis as it contains only two genes, so that was omitted in subsequent analyses. For Group 32, since the mean Z-score of gene abundance fell within similar range at 60-69 (−0.16494) and 70-79 (−0.20438), implying no obvious changes between the two age groups so that was omitted from the subsequent analysis as well. Our decision on selecting Group 16 was based on the fact that genes in it were highly and consistently induced in the age 70-79 period, as reflected by the mean Z-score of gene abundance difference, which shall play significant roles in advanced ordinary aging processes (Figure 1f). Consensus gene modules in CD14 monocytes across 1202 samples with age range from 44 to 83 years old were defined using the weighted gene co-expression network analysis (WGCNA) (v1.71).
The normalized probe expression matrix was downloaded from GEO using GEOquery (2.58.0) and was adjusted by the concentration value for each cell type using the remove batch effect function in limma (v3.46.0). Unsupervised clustering between samples identified 19 outliers and were subsequently removed. The blockwiseModules function was used to identify the co-expressed modules with the following parameters: unsigned topological overlap matrix, minimum size of 30 probe for each module, dissimilarity threshold with 0.1, weighting parameter β was set as eight based on pickSoftThreshold function. The first principal component of each module was used to evaluate their relationship to age. TCA cycle and OXPHOS pathways associated genes were extracted from MSigDB (v7.4), and their enrichment significance were evaluated using hypergeometric test.

| Deconvolution of immune cell types between LLIs and controls
To gain insight into the cellular composition variation in LLIs, CIBERSORTx web server (Newman et al., 2019)

ACK N OWLED G M ENTS
The authors would like to gratefully acknowledge the following parties. The results published here are in whole or in part based on data obtained from (1) the processed count matrix provided by Fu-Hui Xiao and Qing-Peng Kong (HRA000656 and CRA000515); (2) the processed count matrix obtained from AD Knowledge

CO N FLI C T O F I NTE R E S T S TATE M E NT
The authors have no conflict of interest to declare.

DATA AVA I L A B I L I T Y S TAT E M E N T
All analysis have been carried using freely available software packages. Custom code used to analyse the RNA-seq data and datasets generated and/or processed in the current study is available on https://github.com/KimCh ow-Lab/LLIIm mune/.