Aging mechanisms—A perspective mostly from Drosophila

Abstract A mechanistic understanding of the natural aging process, which is distinct from aging‐related disease mechanisms, is essential for developing interventions to extend lifespan or healthspan. Here, we discuss current trends in aging research and address conceptual and experimental challenges in the field. We examine various molecular markers implicated in aging with an emphasis on the role of heterochromatin and epigenetic changes. Studies in model organisms have been advantageous in elucidating conserved genetic and epigenetic mechanisms and assessing interventions that affect aging. We highlight the use of Drosophila, which allows controlled studies for evaluating genetic and environmental contributors to aging conveniently. Finally, we propose the use of novel methodologies and future strategies using Drosophila in aging research.


| TOWARD RIGOROUS DEFINITIONS OF THE TERMS USED IN AGING RESEARCH
In the past century, global life expectancy has almost doubled, owing to improved control of infectious diseases and enhanced quality of life. 1 The United Nations predicts that by 2050, the global geriatric population (age of 80 years old and above) will triple that of 2015. 2 Given the expected rise in the proportion of the elderly to the young, studies to elucidate mechanisms that promote healthy aging will be of paramount impact. Indeed, there has been a remarkable upsurge in the number of the articles related to aging in the past decade ( Figure 1), reflecting intense interest in aging research. However, a plethora of challenges and questions remain to be addressed.

| Aging contribution to aging-related diseases
One of the most challenging questions in aging research is to distinguish the mechanisms and even phenotypes of aging from those of aging-related diseases. Aging is a risk factor for a multitude of chronic conditions and diseases, including cardiovascular diseases, osteoporosis, dementia, osteoarthritis, type 2 diabetes, cancer, and chronic obstructive pulmonary disease (COPD). 3 The diversity of these diseases suggests that the cellular and physiological changes that determine the onset of aging-related diseases are highly complex. While some suggest that aging is a distinct syndrome that should be considered separately from aging-related diseases, others argue that these two overlap in underlying mechanisms to varying degrees. 2,4 The lack of consensus may underscore the necessity to characterize the natural aging process. 5 These aging-related diseases can also manifest in those who are neither old in age nor physiologically aged. For example, cancers could also occur among young people. 6 Another example is Alzheimer's disease (AD), in which early-and late-onset are considered to have separate etiologies, 7 perhaps due to different genetic backgrounds. As such, it is beneficial to have models for evaluating the relative degree of contribution of aging to diseases.
Rothman's 1976 causal pie model 8 was initially employed to model causal inference in the field of epidemiology. We propose that this method can also be applied to describe the causal relationship between aging and aging-related diseases (Figure 2A), and this application may aid scientists in describing and quantifying the public health relevance to aging. We suggest considering aging or the cellular and physiological consequence of aging as a "component cause" (ie, a slice of a pie representing a factor that is not sufficient on its own to lead to the disease onset) of an aging-related disease, which explains an increased incidence among older individuals. However, since the disease can also occur in young people, in this model, aging on its own is not considered as a "necessary cause" (ie, a slice present in all the pies, representing a factor that must be present for the disease onset). The "sufficient causes" (ie, complete pies that represent combinations of all the factors that contribute to the disease onset) thus represent different pathways toward the disease onset. The impact of targeting aging on public health can be conceptualized by reducing the proportion of "sufficient causes." Methods have been developed to apply epidemiological data to causal pie models to quantify the burden of different classes of "sufficient causes" and the effect of interventions on a given "component cause," [9][10][11] which may also be used for the aging field.
F I G U R E 1 Overall trend in aging-related publications. The bar graph shows an exponential increase in the articles in the PubMed database with the keywords, "aging/ageing" in the title or abstract (from a search conducted in January 2020) F I G U R E 2 Possible models to evaluate the contribution of the molecular consequences of aging to the onset of aging-related diseases. A, Applying Rothman's 1976 causal pie model 8 to understanding the mechanisms of the onset of aging-related diseases and the public health implications of targeting aging. B, Possible strategy for isolating the effect of aging on the outcome of various aging-related diseases. It may also be important to elucidate possible multiplicative effects or interactions of aging with other variables, in order to address the degree of contribution of aging to aging-related diseases. C, The current vs trending approach to tackle aging-related diseases. A trending approach focuses on preventing aging-related diseases by targeting aging itself, rather than therapeutic interventions directed at one specific aging-related disease at a time Considering the diversity in aging-related diseases, finding shared molecular mechanisms may aid in isolating the effect of aging ( Figure 2B). Such approach is not trivial, as the effect of aging may be additive, or multiplicative, depending on other variables. Previous therapeutic strategies focus on one aging-related disease at a time and after its onset. In contrast, targeting aging can prevent or delay the onset of an array of aging-related diseases simultaneously, therefore becoming an important alternative ( Figure 2C). Studies using model organisms such as Drosophila can help provide mechanistic insights in the aging process and its interaction with diseases.

| Lifespan and healthspan as distinct aging phenotypes
While "lifespan" refers to the time from birth to death, "healthspan" refers to the length of healthy life. Current goals for aging research emphasize on elucidating the interventions that delay the onset of diseases to extend healthspan. 1,5 However, the extension in healthspan is only meaningful when being related to the lifespan.
Recently, sub-definitions of healthspan extension-"chronological" vs "proportional"-were coined. 12 "Chronological" healthspan extension describes a delayed onset of morbidity proportionally with a prolonged lifespan. Therefore, the total morbidity time is shortened. "Proportional" healthspan extension refers to the total morbidity time that remains the same despite a delay in the onset due to extended lifespan.
In human cohort studies, it is challenging to quantify the healthspan extension due to the difficulty in obtaining health records for all different competing morbidities, as well as a lack of a clear consensus of the conditions that constitute morbidity and warrant intervention. For example, geriatric patients often experience circadian rhythm shift, 13,14 however, this symptom is often unreported or not considered as a significant morbidity to be included in measuring healthspan.
Thus, a recent publication 5  Successful Aging Index, 16 and Healthy Aging Score. 17 It also summarizes the evidence found in model organisms such as mice and Caenorhabditis elegans. Below, we propose to extend this chart by including relevant Drosophila phenotypes and the assays measuring them (part 2b).

| Drosophila as a model organism for aging studies
The trend using model organisms in aging research have shifted from ascertaining only longevity (lifespan) to also include physiological assessments (healthspan). 18 Drosophila melanogaster is a genetically tractable model organism with a long history in aging research. The first documented aging experiment was conducted in 1916, which assessed the effect of temperature and food composition. 19 Using Drosophila, the pro-or anti-aging effects of environmental factors and pharmacological interventions can be easily assessed. Since a large number of flies with the same genotype can be investigated under controlled conditions, it allows scientists to conduct statistically powerful experiments at low cost.
The fly model is also amenable to largescale screens for new interventions and test the combinations of interventions quickly.
Human cohort studies of environmental factors are notoriously susceptible to confounding and reverse causation. The Drosophila model system can help validate the findings in human studies and establish causality. The short lifespan is another major advantage.
The typical lifespan of Drosophila in the optimistic temperature of 25 C is approximately 60 days on average. Increasing the temperature to 29 C leads to a reduced lifespan, while decreasing it to 18 C does the opposite. 20 Moreover, Drosophila is a convenient genetic system with numerous mutants and transgenic tools. For example, the wellestablished UAS/GAL4 system allows tissue-specific transgene ectopic expression 21 and temporal control can also be achieved using a temperature or hormone-inducible variant of the system. 22,23 RNAi knockdown studies can be conducted easily using a curated collection of RNAi lines that target most transcripts. 24 In addition, the CRISPR-Cas9 tools for gene editing in vivo have been developed in Drosophila. 25 Using these methods, scientists have effectively conducted the mechanistic studies, which are challenging in humans, to understand conserved genetic pathways implicated in human aging.
Such pathways include sirtuins and other chromatin regulators (Table 2), and those involved in regulating inflammation, 26 oxidative stress detoxification, 27 insulin signaling, 28 and mammalian Target of Rapamycin (mTOR) pathway. 29 2.2 | Applying fly-specific assays to the understanding of human healthspan index Various assays have been developed in Drosophila for observing aging phenotypes and aging-related disease models, which may relate to human healthspan ( 5 ; Table 1). For example, survival curve measurements are standard methods for assessing lifespan. The assessment of aging-related functional decline or functional senescence aids the quantification of the healthspan extension resulting from specific interventions. 30 Investigators may combine both measurements to evaluate how an intervention affects lifespan and healthspan.
Among commonly used assays for assessing healthspan are those that measure movement. These include the negative geotaxis assay for evaluating climbing ability, [31][32][33] and a variety of flight duration, 33,34 and exploratory activity tests. 33,35,36 The movement weakening during aging may be due to skeletal muscle deterioration or cognitive decline and can be further examined using histological

Cognitive function
Sensory perception GO:0007600-sensory perception Response to light, olfaction and taste 44 Variations of olfactory maze tests [278][279][280] Taste preference assay 281 Odor-stimulated movement real-time imaging 282 Short-term memory processing speed GO:0007614-short-term memory Response to light, olfaction, and taste 44 Variations of olfactory maze tests [277][278][279] Various olfactory and taste memory tests [45][46][47] Sleep defective GO:0030431-sleep Drosophila Activity Monitor (DAM) device and accompanying analysis software called ShinyR-DAM 283 Locomotor activity monitoring and phase-response curves 284 methods. For example, staining of actin filaments help visualize the loss of muscle fiber integrity, 32 and various neuronal markers allow the detection of the loss of specific neurons. [37][38][39][40] Other aging-related neurological phenotypes can be also quantified, including brain degeneration vacuoles, 31,41,42 retinal degeneration and loss of ommatidium photoreceptor integrity, 41,42 and neuromuscular junction abnormalities. 42,43 Aging-related cognitive decline can be assessed by evaluating the fly response to light, olfaction and taste, 44 and memory and learning. [45][46][47] Reproductive decline also occurs during aging and can easily be investigated by quantifying egg-laying or the number of progeny relative to copula copulation. 48,49 Additionally, germ line stem cells can be stained and quantified. 50,51 Cardiac function assays include monitoring heart rate 52 and cardiac tube wall movements. 53 Heat shock survival kinetics 41 or survival in hypoxic environment 54 can be used to assess the resistance to stress. Another commonly used assay is to evaluate oxidative stress resistance through the survival on paraquat. 55 Assays were also developed for assessing the impact of social interaction on lifespan. 56

| Dietary Restriction
It is well-established that DR enhances lifespan, and this appears to be conserved across species ranging from yeast to invertebrates to rodents. However, whether this conservation is simply due to reduced caloric intake remains elusive. Several Drosophila studies have shown that the effect was determined by the relative composition of yeast, sugar and essential amino acids (EAAs), rather than a mere reduction in calories. [57][58][59][60][61] One study found that supplementing with EAA negated the lifespan extending effect of DR, thus isolating the relevant dietary component, 60 and interestingly, this effect was accompanied by reduced fecundity. 60 Another study found that a short-term DR was also effective, suggesting that the intervention later in life in humans might be a valuable method. 62

| Sirtuins and Dietary Restriction
The link between aging and the sirtuin family of histone deacetylases These studies exemplify the strength of combining genetic tools with pharmacological treatments to dissect molecular mechanism of aging.

| Insulin/IGF and JNK/foxo signaling, metabolism, and dietary composition
It is well-established that reducing insulin/insulin-like growth factor (IGF) signaling extends lifespan across species, including humans and Drosophila, and that one major signaling pathway mediating the effect is the c-Jun N-terminal Kinase (JNK)/Forkhead box and sub-group O (foxo) pathway. [75][76][77] The JNK/foxo pathway is a conserved mechanism that confers protection against stress, as well as counters the activity of Insulin/IGF signaling. 28 Consistent with this notion, a study found that in the fly head, the number of genes regulated by the Foxo transcription factor significantly decreased with age, and many Foxotargeted genes were also altered in their expression profile (up-or downregulation) during aging. 78

| EPIGENETIC MECHANISMS IMPLICATED IN DROSOPHILA AND HUMAN AGING
The mechanisms involving epigenetic regulation (histone modifications, DNA methylation, noncoding RNAs, RNA modifications) are among the major models of aging, presented as the "hallmarks" of aging 83 and the "seven pillars" of aging by the NIH steering conference on aging. 84 The genetic heritability of longevity in humans is estimated to be only 15% to 40%, 85 suggesting the likely significant contribution of epigenetic and environmental factors. In this section, we present major studies describing different epigenetic mechanisms in aging, particularly those conserved between Drosophila and humans. These studies are summarized in Tables 2 to 4. 3.1 | Heterochromatin redistribution and epigenetic changes during aging The eukaryotic genome is highly organized with nucleosomes (DNA with histone proteins, H2A, H2B, H3, and H4) as a fundamental unit of chromatin. The histone proteins with various types of chemical modifications are highly conserved across species. 86 Chromatin consists of heterochromatin and euchromatin regions. [86][87][88] Heterochromatin are gene-poor, transcriptionally silent, and highly condensed regions, which are generally characterized by histone hypoacetylation and the enrichment of repressive histone marks, H3K9me2 and H3K9me3, and HP1α proteins. 86,88 In contrast, euchromatin regions T A B L E 2 Histone-related mechanisms conserved between Drosophila and human aging

Species Epigenomic modifications References
Histone methylation-transcriptional repression marks

Epigenomic modifications References
H3K27me3 Drosophila • H3K27me3 increased and broadened regions (AU: what regions? ED) with age, especially in the head • Reduction of H3K27me3 by the deletion of PRC2 genes correlates with up-regulation of glycolytic genes and increased lifespan • The gene-edited deletion of specific PRC2 components encoded by Dme-esc, Dme-E(z), Dme-Pcl, Dme-Su(z)12, and PCR1 component Dme-Su(z)2 increased lifespan [149] • Increased H3K27me3 with age in the head, which was alleviated in Dme-mir-34 mutants • miR-34 targets PRC2 components. [228] • Increased H3K27me3 with age.
• Heterozygous mutations in the core subunits of PRC2, Dme-E(z), and Dme-esc, increased longevity • Mutations in the PR-silencing antagonist, trithorax Dme-trx suppressed the H3K27me3 elevation/longevity effect of Dme-E(z) mutation • Dme-E(z) mutants showed stress resistance phenotypes and de-repression of well-characterized PC-target gene Dme-Abd-B, and metabolic stress resistance gene, Dme-Odc1 [150] • Dme-E(z) heterozygous mutants have sex-specific extension of lifespan and healthspan (resistance to hyperthermia, oxidative stress, and endoplasmic reticulum stress; enhanced fecundity) • Transcriptome profiling of Dme-E(z) heterozygous mutants found altered expression of 239 genes involved with metabolism, immune response, cell cycle, and ribosome biogenesis [151] Human • Increased bulk H3K27me3 in HSCs and progenitor cells with age [152] • Reduced H3K27me3 and PRC2 enrichment at the CDKN2A (INK4/ARF) locus results in transcriptional activation, followed by events leading to senescence and SAHF formation [159] Histone methylation-transcriptional activation marks

H3K4me3
Drosophila • Genome-wide overall increase in H3K4me3 peaks in aged flies [91] • H3K4me3 demethylase Dme-lid mutants had male-specific reduced-lifespan [168] Human • Genome-wide H3K4me3 assessment found that neurons isolated from prefrontal cortex had 600 loci with H3K4me3 peaks in samples from infants compared to 100 loci in old adults (>60 years) • The H3K4me3 peaks that were specific to infant samples consisted mainly of genes involved with neurogenesis, neuronal growth, and differentiation genes, that suggest cellular plasticity [292] H3K36me3 Drosophila • Genome-wide overall increase in H3K36me3 peaks in aged flies [91] Human • Blood DNA from individuals with Sotos syndrome (OMIM 117550), harboring Hsa-NSD1 H3K36 methyltransferase loss-of-function mutations, showed accelerated DNA cytosine methylation aging "clock" signature [169] Histone acetylation

Drosophila
• Midlife (premortality plateau) phenotypes-increased oxygen consumption, reduced histone deacetylase inhibitor sensitivity, increased ATP citrate lyase (encoded by Dme-ATPCL) activity, leading to elevated acetyl-CoA associated with increased histone acetylation, and transcriptional alterations • Decreasing Dme-ATPCL activity, or decreasing H4K12-specific acetyltransferase: Dme-chm, alleviated aging-associated changes and increased longevity [293] • Dietary restriction conditions known to extend lifespan via Dme-Sirt2 histone deacetylasemediated effects also delayed the age-associated increase in TE transcription in the head and fatbody • Dme-Sirt2 overexpression repressed TE elevation in the head and fatbody seen during aging [93] • Sirtuin-activating compounds, resveratrol, and fisetin increased lifespan and prolonged fecundity. • The longevity effect depended on functional Dme-Sirt2 [73] • Dme-Sirt2 overexpression increased lifespan • Dme-Sirt2 downregulation blocked the lifespan-extending effect of caloric restriction and Rpd3 mutation [71] are relatively gene-rich, transcriptionally active, less condensed, and enriched with histone hyperacetylation and active histone marks such as H3K4me2 and H3K4me3. 86,88 Heterochromatin redistribution in the aging process has been  Together, these studies strongly support that H3K9 methylation and • Strong dose-dependent effect of Dme-Sirt2 overexpression (when two to fivefold overexpressed) on increasing lifespan • Dme-Sirt2 overexpression increased the expression of the Puc protein phosphatase (encoded by Dme-puc) and the heatshock response gene, Dme-DnaJ-H • Puc protein phosphatase operates in a JNK signaling pathway [174] • Caloric restriction and Dme-HDAC1 (Rpd3) heterozygous loss-of-function mutation increased lifespan [176] • Ubiquitous Dme-Mi-2 RNAi knockdown increased lifespan Mi2 is a component of the NuRD chromatin remodeling complex that includes HDAC1 and HDAC2 and thus the observed effects may be related to histone acetylation [179] Human • Dme-SIRT3 intronic enhancer polymorphism was found to be associated with old age, in which the allele that results in no enhancer activity was absent in >90 year old males • A later meta-analysis found inconclusive results [294] • Dme-SIRT3 polymorphism was associated with elderly survivorship [295] Chromatin remodeling

NuRD Drosophila
• Ubiquitous Dme-Mi-2 RNAi knockdown increased lifespan [179] Human • Reduced expression of NuRD components in Hutchinson-Gilford progeria syndrome (HGPS, OMIM 176670), HGPS-derived cells and those from older donors • Reduced HDAC1 activity (encoded by Hsa-HDAC1) was also found (the NuRD chromatin remodeling complex that includes HDAC1 and HDAC2), suggesting that the mechanism could also be related to histone acetylation [296] TSURUMI AND LI mechanism was also demonstrated in mammals to suppress cancer development, 99,100 highlighting the advantage of using Drosophila for discovering conserved mechanisms.
These findings in Drosophila is consistent with an aging model in mammals, [101][102][103]

| Nuclear lamina and heterochromatin
Nuclear lamina (NL), a lining of the inner nuclear envelope, consists of filaments, lamins, and lamin-associated proteins, and is directly tethered with heterochromatin. 124 Heterochromatin domains of the genome are anchored to the NL, allowing them to reside in the nuclear periphery. It has been suggested that the spatial positioning of the heterochromatin is mechanistically important for their repressiveness of transcriptional activity. 125 The lamin-B receptor (LBR) forms a complex with HP1α, demonstrating a direct link. 126  In Drosophila, studies have also provided evidences for the relevance of the NL in natural aging. [132][133][134] A progressive reduction in LMNB expression in the fat body was observed with aging, and moreover, found to contribute to chronic inflammation and gut hyperplasia. 134 Intriguingly, knocking down LMNB in young adult or larval fat body resulted in reduced heterochromatin, and an increase in retrotransposon expression and DNA damage. 132,133 Collectively, these studies provide evidence of the importance of NL in controlling nuclear architecture, heterochromatin, TE activation, genome instability and chronic inflammation in aging. This again suggests the conservation between Drosophila and human aging.

| RNAi pathway related to heterochromatin and TE silencing
One of the major roles of the RNAi pathway is in regulating heterochromatin-induced silencing of repetitive DNA loci, such as GWAS association Hsa-SIRT1 rs7096385-T GWAS on atrial fibrillation [298] Type 2 diabetes is associated with increased expression of MIR34A in the heart Hsa-MIR34A Type 2 diabetes mellitus gene expression Cardiac muscle tissue [230] (Continues)

Species Molecular mechanism and aging phenotypes Species-Gene Concepts References
Overexpressed MIR34A in the lungs of patients with idiopathic pulmonary fibrosis, and functional studies found it promotes senescence and reduces cell proliferation TEs, pericentric satellite repeats, and telomeric repeats. 135

| PIWI/piRNA related to heterochromatin and TE silencing
It is well-established that the PIWI/piRNA pathway plays a crucial role in silencing TEs in the germline of Drosophila 144 by regulating heterochromatin formation. 145 More recently, a Drosophila study has found that Piwi is crucial for suppressing age-related TE expression in intestinal stem cells and the maintenance of epithelial homeostasis, thus demonstrating its new role in somatic stem cell maintenance. 146 The PIWI pathway has recently been appreciated as a key player in cancer as well. 147

| Silencing histone methylation mark, H3K27me3
It is well-established that H3K27me3 promotes gene silencing and is associated with polycomb repressive complex (PRC) 1 and 2. 148  Age-related hearing impairment [299] GWAS association to Alzheimer diagnosis accelerated after diagnosis of cognitive impairment Hsa-ADARB2 rs10903488-?
GWAS on age at menarche [302] of specific genomic areas with aging-associated augmentation of H3K27me3, [149][150][151] and this increase is particularly high in the head. 149 Loss-of-function mutations in the genes encoding specific PRC1 and PRC2 components suppressed the elevation of H3K27me3 during aging and improved lifespan and healthspan. Among PRC components,

Enhancer-of-zeste (E[z]) mutants have been particularly well-character-
ized. 150,151 Observed healthspan effects include improved stress resistance to hyperthermia, oxidative stress and endoplasmic reticulum stress, as well as enhanced fecundity. 150,151 The stress resistance was accompanied by the upregulation of many target genes including the stress resistance gene, Ornithine Decarboxylase 1 (Odc1). 150 Further genome-wide transcriptome profiling of these animals found altered expressions of 239 genes that were mainly involved in metabolism, immune response, cell cycle, and ribosome biogenesis. 150 Further characterization of the H3K27me3 regulation during aging may find potential targets for slowing down the aging in the epigenome.
In human studies, a bulk increase in H3K27me3 was also detected with aging, when assessing hematopoietic stem cells (HSCs) and progenitor cells. 152  There are other loci that can be affected by augmented levels of H3K27me3. 158 For example, a study in cell culture demonstrated that replicative senescence and SAFH formation resulted from a reduced level of H3K27me3 at the CDKN2A (INK4/ARF) locus. 159 Metformin, an anti-diabetic drug, was found to be a specific inhibitor of the H3K27me3 demethylase, Kdm6A, 158 and the cells treated with Metformin had a global increase in H3K27me3. Therefore, metformin may be applicable as an anti-aging intervention by inhibiting the T A B L E 4 Conserved mechanisms involving nuclear and chromosomal architecture implicated in aging in Drosophila, mouse, and human

Drosophila
• Increased oocyte meiotic segregation errors and nondisjunction with heterozygous mutation in cohesin subunit Dme-SMC1 with aging. [303] • Increased susceptibility to aging-related meiotic segregation errors and nondisjunction with heterozygous mutation in Dme-ord with reduction of centromere-proximal heterochromatin. [304]

Mouse
• Female mice deficient in the meiosis-specific cohesin SMC1β encoded by Mmu-Smc1b show nondisjunction and age-dependent loss of meiotic sister chromatid cohesion [305] Human • Downregulation of the meiosis-specific cohesin subunits encoded by Hsa-REC8 and Hsa-SMC1B, in women aged 40 and over compared with 20 years. [306] • Multiple lines of evidence suggest decreased meiotic sister chromatid cohesion with age and support that it plays key roles in the maternal age effect on meiotic segregation.
Reviewed in [307,308] Integrity of the nuclear lamina

Drosophila
• Aging-related progressive reduction in lamin-B (encoded by Dme-Lam) in the fat body was found to contribute to chronic inflammation and gut hyperplasia. • Depletion of lamin-B in the young/larval fat body via RNAi Dme-Lam results in reduced amount of heterochromatin, and increase in retrotransposon expression and DNA damage [132][133][134] Human • Hutchinson-Gilford progeria syndrome (HGPS, OMIM 176670), characterized by premature aging phenotype, is caused by various germline mutations in Hsa-LMNA encoding lamin A/C. [129,130] • In senescent dermal fibroblasts and keratinocytes, Hsa-LMNB1 encoded lamin-B1 protein expression is reduced. • Hsa-LMNB1 overexpression induces senescence. [131] • Lamina-associated domains (LADs) are genomic regions that form molecular contacts with nuclear lamins. LADS have many of the properties of heterochromatin at the periphery of the cell nucleus. [309] H3K27me3 demethylation at the locus such as CDKN2A. However, it is critically important to note that inactivating this locus may drive oncogenesis, therefore, a balanced activity or expression is crucial. 160,161 3.1.7 | Gene activation marks, H3K4me3, and H3K36me3 In the study that found an overall decrease in the repressive H3K9me3 mark in aging Drosophila, the active marks, H3K4me3 and H3K36me3 showed an overall increase. 91 H3K4me3 that located at the promoter regions are associated with transcriptionally active genes. 162 H3K36me3 is enriched in the gene bodies, regulating transcriptional fidelity and alternative splicing events. 163,164 Studies demonstrated that both activating marks can antagonize H3K27me3, [165][166][167] and the interplay between these histone marks might be important during aging. For instance, a mutation in the trithorax (trx) encoding a H3K4 methyltransferase was found to negate the lifespan increasing effect of E(z) mutation. 150 Elucidating the loci related to these marks during aging, including the identification of the regulatory enzymes and other factors can help find the interventions to suppress aging-related phenotypes. It is easy to conduct genetic studies in this area using the Drosophila model.
H3K4me3 and H3K36me3 appear to be also relevant to human aging brain. A genome-wide H3K4me3 profiling found that in neurons isolated from infant prefrontal cortex, H3K4me3 peaks in several hundred loci, compared to approximately a hundred in samples from old people (>60 years). 168 The H3K4me3 peak sites in infants consisted mainly of the genes involved in neurogenesis, neuronal growth, and differentiation. The link between H3K36 methylation and human aging was also observed in Sotos syndrome patients who had a mutation in a gene encoding H3K36 methyltransferase. 169 Future studies of H3K4me3 in other tissues and H3K36me3 in natural aging are expected to yield informative results. Concomitantly, assessing transcript levels and alternative splicing isoforms would be useful.

| Sirtuin as an anti-aging regulator
The Sirtuin family of NAD-dependent histone deacetylases (HDACs) has been in the limelight in aging studies because its stimulation or overexpression has been shown as a promising avenue for anti-aging. 170,171 Although it is well-established that Sirtuins mediate the longevity effect of caloric restriction (diet condition studies are described in detail in Section 2.3), its role played in Drosophila lifespan has been controversial. Whereas one study reported no effect, 172 some others observed lifespan extension, but only when the expression level of Sirtuin 2 (Sirt2) was two to fivefold upregulated. 71,173,174  Sirturins also regulate nonhistone targets. In both Drosophila and mammalian studies, p53 protein has been identified to be a major target of deacetylation by SIRT2. 178

| Nucleosome remodeling regulators in aging
The interaction between the NuRD-HDAC complex and PRC might be important in regulation of nucleosome remodeling during aging across species. Increased lifespan was found in Drosophila with Mi2 knockdown. 179 Mi2 protein is a component of the nucleosome remodeling and deacetylase (NuRD) complex that also includes HDACs and conserved across species. 180 This protein was initially found to participate in the PcG-mediated repression at Hox clusters. 181 The deacetylation of H3K27ac facilitates the methylation at the H3K27 residue, and this antagonistic mechanism was previously observed in both mammalian cells 182 and in Drosophila. 183 191 as well as applying the model longitudinally. 192,193 In Drosophila, C elegans, and yeast species, the presence of 5mC in the genome has been long debated. [194][195][196][197] In Drosophila, 5mC was found relatively more abundantly in early embryos, 195,198 but with a low level in adult tissues. 195,[197][198][199][200][201][202][203] In its genome, only one methyl- Dnmt1. Mt2/Dnmt2 was initially named as a DNMT, but later found to act predominantly as a tRNA m 5 C methyltransferase, 204-207 although in the context of tRNA-DNA hybrids, it can also act on DNA. 208 In Drosophila, the overexpression of Mt2/Dnmt2 increases lifespan, however the mechanism remains elusive. 141 The minimal presence of 5mC in adult flies suggest that it likely does not mimic methylation changes in 5mC in human aging, thus not a suitable model for the aging clock.

| 6-methyldeoxyadenosine
It was found recently that N 6 -6-methyldeoxyadenosine (6mA) is another DNA methylation mark that is abundantly present in both the human 209 and Drosophila genome. 210 In Drosophila, 6mA was demonstrated to be regulated by DNA 6mA demethylase (DMAD), and its presence was correlated with TE expression in the ovary. 210 A different study showed that 6mA transcriptionally regulated zelda via the recruitment of Jumu, a 6mA reader, to facilitate the maternal-tozygotic transition. 211 It was also demonstrated that 6mA might dynamically regulate the genes involved in neurodevelopment and neuronal functions and TE activity in the brain. 212 However, the role of 6mA in aging is still lacking in any organism. Given the relevance of TE in aging, 105 6mA and its regulators warrant further investigation for its role as a functional substitute for 5mC and a potential aging clock in Drosophila. It is also of great interest to delineate the similarities and differences between the mechanistic roles of 6mA and 5mC.

| Noncoding RNAs
The transcriptional dysregulation due to chromatin alterations has been shown to affect both protein-coding gene transcripts and noncoding RNAs. Studies of different noncoding RNAs involved in Drosophila aging, and their relevance to human aging are summarized in

| Long noncoding RNAs (lncRNAs)
LncRNAs can act as an enhancer to facilitate transcription, as well as a decoy to prevent or guide the recruitment of transcription factors and chromatin modifying factors, and the interactions between DNA methylation or histone modification mechanisms and lncRNAs are evident. 213 For instance, MALAT1 lncRNA, which was found to be upregulated in various cancer types in humans, 214 can regulate the recruitment of histone-lysine N-methyltransferase enzyme, EZH2 to promote H3K27 methylation at specific genomic loci. 215 LncRNAs can also act as a scaffold to promote the formation of protein complexes, which regulate various processes such as transcription, alternative splicing, translation, rRNA maturation, microRNAs binding to their target mRNAs, and signaling molecule phosphorylation. 213 Moreover, they can serve as a precursor to small RNAs. 213

| Small noncoding RNAs
In a Drosophila Hungtington's disease model that overexpressed the human pathogenic polyglutamine disease protein, a knockout in miR-34 resulted in reduced lifespan with accelerated brain aging, whereas its upregulation extended lifespan and alleviated neurodegeneration. 41 The Ecdysone-induced protein 74EF gene, Eip74EF was found to be a target of miR-34 and upregulated in the miR-34 deletion strain. 41 Another study confirmed the disruption of the Ecdysone signaling in miR-34 knockouts and reported a defect in innate immune response. 227 These observations suggest a new role of the Ecdysone pathway in aging. Another study showed that miR-34 targeted PRC2 components, Polycomblike (Pcl) and Su(z)12, and that in miR-34 deletion animals, the H3K27me3 accumulated in the brain and the gene expression profile was associated with advanced aging. 228 Also, in this study, miR-34 upregulation was shown to lead to the alleviation of neurodegeneration induced by pathogenic polyglutamine protein overexpression. 228 The miR-34 family is conserved across species including humans.
The human miR-34a (member of the miR-34 family) was found to be upregulated with aging in peripheral blood mononuclear cells (PBMCs) and target SIRT genes. 229 Given the importance of maintaining the level of Sirtuins (described in part 3b.iv. above), targeting miR-34a may be a feasible anti-aging intervention. Other studies have also shown the relevance of miR-34a to cancer and diabetes. [230][231][232] These observations underscore the importance of miR-34a in aging across species.
Other miRNAs relevant to aging include miR-125 and miR-9a. The loss of miR-125 in Drosophila was found to reduce lifespan, impact climbing activity, and increase neurodegeneration in the brain. 31 miR-9a was found to regulate the maintenance of male germline stem cell. 233 The human homologs of these miRNAs have been implicated in aging and cancer in humans. 229 250 Specifically, for AGO2 mRNA, both its abundance and the m 6 A enrichment were reduced. m 6 A was also found to regulate other biological processes relevant to aging in humans, including self-renewal in stem cells, 251 the circadian rhythm, 252 promoting cancer stem cells, 253 and enhancing tumorigenicity. 254 Other significant RNA modifications including 5-methylcytosine and pseudouridylation might be also involved in aging. [255][256][257] As mentioned above in Section 3.2, in Drosophila, the overexpression of Mt2/ Dnmt2, which acts mainly as a tRNA m 5 C methyltransferase, [204][205][206] increases lifespan. 141 Conversely, knocking down TRDMT1/Dnmt2 in human fibroblasts increased oxidative stress, DNA damage, miRNAs targeting transcripts related to proliferation, resulting in senescence. 258 In Drosophila, 2 0 Ome becomes enriched in specific miRNAs with age. 142 The study also reported that more miRNAs become associated with Ago2, but not with Ago1 during aging. 142 Indeed, Ago2 mutation led to decreased lifespan and neurodegeneration accompanied with increased brain vacuoles. 142 Thus, further studies of RNA modification mechanisms and their relation to the Ago2 loading in the context of aging may be important.
In addition to RNA base modifications, RNA editing has been recently described to play a role in aging. An interplay between RNA m 6 A modification and RNA adenosine-to-inosine (A-to-I) base editing was found; the presence of the two alterations appeared to be negatively correlated. 243 In Drosophila, a hypomorphism of the adenosine deaminase acting on RNA (ADAR) gene, which encodes an A-to-I editor, was found to cause the extended lifespan. This was accompanied in neurons with increased levels of histone modifications that are associated with heterochromatic silencing. 259 Human cohort studies found single nucleotide polymorphisms (SNPs) in the orthologues, ADARB1 and ADARB2 to be associated with longevity, 260  mechanisms and human aging-related heterogeneity with high resolution. 191 The trending approach, not just in the aging field, but widely, involves emerging technologies that allow molecular signature assessment of single cells, in addition to bulk tissues. The obvious limitation for the human studies is that not all tissues and cells can be harvested for investigation.
Moreover, an integrative -omics obtaining information from a wide variety of types of molecular signatures (genomics, epigenomics, transcriptomics, proteomics, metabolomics, metagenomics, and other -omics) is an emerging approach that provides comprehensive and unbiased information. 265

| Other uses of the Drosophila system in aging research
As indicated in Table 1, Drosophila, with the plethora of healthspan phenotypes that are relevant to human aging, has been shown to contribute significantly to the aging studies in vivo. It would also be advantageous for meticulous quantification of "proportional" vs "chronological" healthspan extension, which is not easy in human studies.
Drosophila have been used extensively as a model to elucidate fundamental mechanisms of all aspects of biology and there are still endless discoveries that could be made in this highly characterized Research Fellowship #84293. We would like to acknowledge Ankita Banerjee for her input and aid editing.

CONFLICT OF INTEREST
The authors declare that they have no conflict of interest.