You are on page 1of 43

New insights into aging and obesity-associated aging characteristics of female

subcutaneous adipose tissue revealed by integrative analysis of multi-omics data

Zichao Lia,1, Shun Wangb,1, Shaojie Liuc,1, Ziwen Xud, Xiaowei Yie, Yunchuan Wang a, Jiaqi Liu a, Liang

Luo a, Xiaozhi Bai a, Xujie Wang a, Peng Ji a, Hongtao Wang a,*, Dahai Hua,*

a
Department of Burns and Cutaneous Surgery, Xijing Hospital, Fourth Military Medical University,

710032 Xi’an, Shaanxi, China

b
Department of Computer Science, Xiamen University, 316005 Xiamen, Fujian, China

c
Department of Urology, Xijing Hospital, Fourth Military Medical University, 710032 Xi’an, Shaanxi,

China

d
Fourth Military Medical University, 710032 Xi’an, Shaanxi, China

e
Department of Oto-Rhino-Laryngology, West China Hospital, West China Medical School, Sichuan

University, 610041Chengdu, Sichuan, China.

*Corresponding authors: wanght@fmmu.edu.cn (Hongtao Wang); hudhai@fmmu.edu.cn (Dahai.Hu)

1
These authors contributed equally to this work.

Abstract

Background: The subcutaneous adipose tissue (SAT) plays a role in the redistribution of adipose tissue

and systemic functions during aging. However, the changes in SAT during aging are not well

understood.

Methods: We integrated the multi-omics expression profiling data of 37 female individuals (age 23-72)

retrieved from the GEO database to construct the aging profiles of SAT. Weighted gene correlation

network analysis (WGCNA) was employed to determine the co-expression patterns of differential

genes, aging-specific biological processes, and regulatory networks at different aging stages. The

differential co-expression (DiffCoEx) analysis was further performed to explore the effects of obesity

on the SAT during aging.

Findings: A landscape of the aging profiles for differential genes, miRNAs, and gene methylation data

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
was first provided. Moreover, we provided an improved understanding of the age-dependent variations

in the biological states of the SAT. The integrative analysis indicated that only the regulatory

relationship of differentially expressed genes and miRNAs (DEGs-DEMs)was significant. While there

was no evidence confirming that the differential expression of senescence-related genes could be

regulated by methylation in SAT aging. Our research mainly elucidated that in normal aging, SAT

primarily experiences the immunosenescence, alterations of metabolism, and defective angiogenesis in

the different aging stages. Moreover, a higher risk of metabolic disorders, developing chronic

nonhealing wounds, and accelerated senescence in aging of obese females was revealed by the

DiffCoEx analysis of SAT.

Interpretation: Aging and obesity mutually influence the pathological states and decreased clinical

applications of the SAT. We additionally show that SAT is a potential novel model for studying aging

and aging-related metabolic disorders in the future.

Funding: This work was supported by the National Natural Science Foundation of China (No.

81971835; No. 81772071; No. 81772072).

Keywords: female subcutaneous adipose tissue, aging, obesity, multi-omics profiling, Weighted gene

correlation network analysis (WGCNA), differential co-expression (DiffCoEx) analysis.

Research in context

Evidence in this study

In the aging process, the systemic adipose tissue experienced the redistribution and aging related

function alterations. Previous studies on the aging related changes in the biological status of the

adipose tissue mainly focused on the visceral and ectopic fat. However, there are no systematic studies

on the changes of the biological states of SAT in aging. The development of obesity is partly attributed

to the excessive accumulation of SAT and leads to many disorders. Since the excessive deposited SAT

constitutes a dysfunctional state for the obese population, whether and how obesity affects the aging

process of SAT is important and an interesting direction to study.

Added value of this study

SAT was first utilized as the biological model to explore the aging process and the interaction between

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
aging and obesity. Our research originally integrated multi-omics microarray data to provide the

landscape of the aging spectrum. In the normal aging process, SAT experienced the specific aging

related biological processes, which led to many pathological states and decreased clinical application of

SAT. Further analysis identified the characteristic the biological processes of SAT aging for the obese

population compared to the non-obese population.

Implications of all available evidence

SAT is an ideal model to study human aging, obesity, and the interaction relationships between them.

SAT aging and obesity are two mutually reinforcing process and both of them could cause certain

metabolic disorders and the limitations of clinical application. Our results provide a potential novel

approach for the frontier of targeted therapy of aging and obesity.

1. Introduction

Aging is characterized by progressive decline of systemic physiological functions and represents a

major risk factor for chronic illnesses, including cardiovascular disease, metabolic disorders, and

cancers1. As age advances, systemic adipose tissues also undergo aging-related changes, especially fat

redistribution, which results in diminished subcutaneous adipose tissue (SAT) in the lower-body and

increased visceral and ectopic fat2. Adipose tissue dysfunction during aging is associated with

adipocyte senescence, which leads to disrupted cellular structures, as well as impaired insulin response

and secretory profiles3-4. Previous studies have reported age dependent functional variations in visceral

adipose tissue (VAT) and ectopic fat5. Chronic low-grade inflammation, metabolic disorders like insulin

resistance, and interference with homeostasis are aging-related features associated with VAT in human

and animals5-6. Despite storing about two-thirds of systemic fat and having multiple physiological

functions, few studies have investigated the widely distributed SAT during aging.

Obesity is often indicated by a body mass index (BMI) ≥30 and is associated with excessive

accumulation of white adipose tissue viscerally and subcutaneously. Due to its increasing incidence,

obesity is a major public health concern. In the US it is estimated that by 2030, one in two adults will

have obesity and one in four will have severe obesity7. Multiple studies have reported that during

obesity, dysfunctional excessive preadipocytes, VAT adipogenesis, and ectopic fat cause metabolic and

inflammation disorders8. Obesity is also associated with increased risk of type 2 diabetes,

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
cardiovascular disease, and cancer3, 8-9. Here, we hypothesize that obesity influences all aspects of fat in

physiology, including aging. However, there is little evidence on the effects of obesity on the aging

process of adipose tissue.

Here, we evaluated differential gene expression, miRNA and methylation data to profile female SAT

aging in Youth to Middle age. Analysis of aging-related genes revealed that female SAT aging is

characterized by immunosenescence, metabolic changes and defective angiogenesis. Our analysis also

identified the core molecular regulatory networks associated with female SAT aging and the

characteristics of SAT aging under the influence of obesity. Our data provide a better understanding of

the biological changes in SAT aging in non-obese and obese females. SAT is widely used in plastic and

reconstructive surgery 10-11


and there is evidence that many systemic metabolic diseases are associated

with SAT disorders 12-13. Therefore, our findings may guide targeted treatment of SAT aging associated

metabolic disorders, and improve outcomes for fat grafting in elderly and obese patients. Due to its

accessibility, SAT may be an ideal model for human aging studies.

2. Materials and methods

2.1. Microarray data

mRNA (GSE25401), miRNA (GSE25470) and methylation (GSE24884) datasets were downloaded

from GEO (Sheet S1 and Table S1). These datasets contained 56 female obesity samples14-15, of which

37 were selected and split into 6 groups according to age (youth: 20-30 years old, middle age: 40-50

years old, Elder: ≥60 years old) and obesity status (non-obese and obese).

2.2. Aging spectrums construction based on DEGs, DEMs and DMGs

Aging spectrums were constructed on the basis of differential gene expression profiles, miRNAs and

methylation, and visualized by heatmaps. Differentially expressed genes (DEGs) and differentially

expressed miRNAs (DEMs) between groups were identified using RankProd package16 in R. P ≤0·05

was set as cutoff threshold for DEGs and DEMs. Differential gene methylation (DMGs) was identified

using limma R package17 based on p ≤0·05 and differential mean methylation levels of ≥5% between

groups.

2.3. Integrative analysis of multi-omics interaction patterns

Molecular regulatory networks were constructed to uncover underlying intermolecular interactions

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
between the identified core aging related biological processes of SAT. Protein-protein interaction (PPI)

analysis for DEGs was done using STRING18 to construct differential co-expression network and

regulatory networks. Network node colors correspond to co-expression modules. Target interaction

relationships between DEMs and 3’UTRs of DEGs were downloaded from mirWalk219. Spearman

correlation coefficient between the expression patterns of DEMs and DEGs <-0·5 indicated existence of

regulatory relationships in networks. The same threshold of Spearman correlation coefficient was set to

identify DEGs-DMGs regulatory relationships. Interaction and regulatory networks between

DEGs-DEGs and DEGs-DEMs were visualized using CytoScape20.

2.4. Weighted gene co-expression network analysis (WGCNA)

We obtained the DEGs in different age groups of obese and non-obese females, and required their

union set which can be considered as aging-related DEGs (4560 DEGs) for further constructing the

scale-free gene co-expression network using “WGCNA” package 21


(Figure S3A). In this scale-free

network, only a few nodes were highly connected to others. Here, we used the signed correlation

method to transform the correlation coefficient of co-expression between nodes as follows: adjacency

(between nodes) = (one+correlation)/two, which weakened negative correlations and enhanced positive

ones. Next, the soft thresholding power, β, was set to 14 based on the ‘pick soft threshold’ function of

WGCNA. This soft thresholding power enabled the generation of an appropriate scale-free

co-expression network. The dynamic tree cut method was used to hierarchically cluster genes using the

dissimilarity matrix (1-TOM)22. The minimum cluster size required 50 genes, and the clustered

modules with high similarity were merged with a height cut-off of 0·4.

2.5. Differential gene co-expression analysis using DiffCoEx

To quantify differences in overall aging-related co-expression patterns in SAT between non-obese and

obese females, we performed differential co-expression (DiffCoEx) analysis based on WGCNA23.

Adjacency matrices for non-obese and obese cases microarray datasets were calculated separately and

adjacency values between nodes defined as follows: if correlations coefficient was >0, adjacent values

equaled the correlation coefficient; if correlations coefficient was ≤0, adjacent values were set as 0.

Topological overlap matrix was derived from the matrix of differences in powered adjacencies between

non-obese and obese cases and used as a distance metric. Genes were clustered by average hierarchical

clustering, so gene co-expression modules were partitioned with unique and similar variation patterns

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
between the 2 populations. DiffCoEx modules were further identified using the hybrid method of

dynamic tree cutting. Hierarchical clustering tree was cut at a height of 0·94. The minimum cluster size

of each module was 20 and merged cut height ≤0.2.

2.6. GSEA and GOBP enrichment analysis

The genes of co-expression modules from WGCNA and DiffCoEX analyses were intersected with

DEGs in different age groups for further enrichment analyses (Figure S4A-B. Next, Gene Set

Enrichment Analysis (GSEA)24 and Gene Ontology Biological Process (GOBP) enrichment analyses of

the selected DEGs in different groups were performed using Cluster Profiler R package25. Gene

expression profiling matrix and DEGs sets were divided according to age groups and obesity status.

Aging related biological-processes pathways with p ≤0 · 05 were identified as biological processes

involved in SAT aging.

3. Results

3.1. Multi-omics aging profiles of subcutaneous adipose tissue

The multi-omics expression profiling data obtained from 37 females were grouped as shown in Table1.

Our analyses offered an improved understanding of age-dependent variations in gene, miRNA and gene

methylation expression profiles during SAT aging. The SAT aging profiles for non-obese cases

uncovered 1860 DEGs (1148 (Youth vs Middle Age), 833 (Middle Age vs Elder)), 56 differentially

expressed miRNA (DEMs) (34 (Youth vs Middle Age), 25 (Middle Age vs Elder)) and 332 differential

gene methylation (DMGs) (204 (Youth vs Middle Age), 164(Middle Age vs Elder)) between the age

groups (p ≤0 · 05). Additionally, SAT aging-related expression profiles were evaluated in obese cases.

During aging, 1623 DEGs (711 (Youth vs Middle Age), 1019 (Middle Age vs Elder)), 95 DEMs (40

(Youth vs Middle Age), 60 (Middle Age vs Elder)) and 348 DMGSs (249 (Youth vs Middle Age), 205

(Middle Age vs Elder)) in all age groups (p ≤0 · 05; Table 2, Figure 1). A panorama of SAT aging

miRNAs and gene methylation spectrums for non-obese cases is shown in Figure S1A and Figure S2A

based on DEMs and DMGs. Because the DEGs aging spectrum is too broad, representative DEGs that

may be involved in the aging process are shown (Figure 3B)

3.2. Integrative analysis of multi-omics aging profiles

To elucidate the regulatory relationships between DMGs-DEGs-DEMs, we performed an integrative

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
analysis of multi-omics expression profiles for each case based on aging profiles. Remarkably, rich

regulatory relationships emerged between DEGs and DEMs. 500 regulatory relationships with

moderate correlation intensity ((r<-0·5, p≤0·05) between DEGs and DEMs in non-obese cases were

subjected to further study (Figure S1B). Based on differential gene methylation profiles and gene

expression profiles, none of the DEGs were considered to be regulated by methylation during SAT

aging. (Figure S2B). Although 111 DMGs were identified between different age groups, there was no

powerful evidence confirming the existence of regulatory patterns between aging associated DEGs and

DMGs in non-obese cases, so as to obese female (r<-0·5, p ≤0·05; Figure S2B, Table S2, S3).

Moreover, 119 and 308 methylated genes were identified in non-obese and obese cases (Sheet S2,S3).

In non-obese cases, these aging-irrelevant methylated genes were mainly involved in molecular

functions, cell signals and communication, cellular response, and metabolism (Figure S2C-D)

3.3. Identification of SAT aging related DEGs expression patterns

To further investigate the expression pattern of aging-related genes, we used WGCNA to identify

significant aging-related gene modules among 4560 genes (Figure S3A). Similar expression patterns

were grouped into modules via hierarchical average linkage clustering (Figure S3B). To ensure

relatively balanced scale independence and mean connectivity, a soft-thresholding power of 14 was set

for network construction (Figure S3C-D). From the gene expression profiles of non-obese cases, this

analysis identified 11 gene modules involving 4138 genes (Table S4, Figure 2A) and 10 modules,

closely associated with the aging process but with different patterns in SAT selected (Figure 2B). Thus,

together with the DEGs uncovered in the above analysis, 1754 DEGs remained for analysis at different

aging stages.

3.4. Aging-related specific biological processes of SAT

Next, we analyzed the selected enrichment DEGs using gene ontology biological processes (GOBP) to

identify aging-related pathways. Interestingly, comparison of the aging process between Youth-Middle

age and Middle Age-Elder, revealed shared specific BP terms. Significantly enriched GOBP pathways

in Youth or Middle age groups showed that advancing age was associated with an upregulation of

pathways related to inflammatory states (GO 0006954, GO 0001816, GO 0032943, GO 0042116),

tumor necrosis factor (GO 0032640, GO 0071706), lipid deposition and metabolism (GO 0008610, GO

0006629, GO 0044255, GO 0006633, GO 0032365). The Middle age group was also associated with

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
reduced immune response (GO 0070227, GO 0071887, GO 0002700, GO 0002274), energy

metabolism (GO 0046034, GO 0140053, GO 0006091) and angiogenesis (GO 1905288, GO 1905459)

(Figure 2C). In the aging process between Middle age and Elder, GOBP pathways associated with

carbohydrate metabolism (GO 0044262, GO 0006006, GO 0010675, GO 0006109, GO 0016052)

emerged as the main processes in the Elder group. Additionally, lower enrichment of functional

pathways implied a weakening of immune responses (GO 0045321, GO 0046649, GO 0002520, GO

0006955) and energy metabolism (GO 0046034) during aging (Figure 2C) Gene set enrichment

analysis (GSEA) was further used to characterize aging-related biological processes. During the SAT

aging process, GOBP enrichment pathways were associated with inflammatory states, immune

response, lipid metabolism, carbohydrate and energy metabolism. Additionally, the aging process was

accompanied by specific fat cell differentiation (GO 0045444) from Youth to Middle age, and B cell

mediated immunity (GO 0019724) from Middle age to Elder (Figure 2D)

3.5. SAT aging related interaction network and core nodes

Based on the aging related molecules and identified GOBP pathways (Sheet S4), we constructed an

interaction network to show the regulatory relationship between selected DEGs and DEMs. This

analysis highlighted 7 subnetworks that were then used to explore the regulatory patterns of our

focused aging related biological processes (immune response, inflammatory states, chronic

inflammatory cells, TNF, energy and carbohydrate metabolism and lipid metabolism) more explicitly

(Figure 3A). This process identified 70 significant core genes. A heatmap was then drawn from the

interaction network and associated gene modules based on WGCNA. Similarly, 28 miRNAs presented

in heatmap were considered core nodes, as they could modulate expression of core genes, altering the

aging related biological processes in the interaction network (Figure 3B-C)

3.6. SAT aging related module differences in obese and non-obese females

To explore SAT aging pattern differences in obese and non-obese females, differential co-expression

(DiffCoEx) analysis was done for aging-related genes. Of the 4560 genes analyzed, 539 were assigned

arbitrarily to one of 7 modules with various colors denoting various correlations representing different

expression patterns in obese and non-obese females (Figure 4A). The green, black, blue, brown, and

yellow modules contained 240 genes that were significantly co-expressed during the SAT aging process

in non-obese females. The red and turquoise modules, containing 299 genes, were notably co-expressed

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
in obese females during aging (Table S5). Bubble plot analysis showed that different aging stages

between obese and non-obese females were accompanied by different co-expression module patterns

(Figure 4B). The differential co-expression network was visualized using Cytoscape (Figure 4C).

3.8. Aging related functional enrichments and regulatory patterns in obese and non-obese females

Next, GOBP enrichment analysis for specific co-expression modules in each group revealed that SATs

of obese ad non-obese females experienced different biological processes during aging (Sheet S5).

During aging, non-obese female SAT experienced improved production of monocyte chemotactic

protein-1 (MCP-1, GO 0010243) while age progression from Youth to Middle Age encountered

increasing chronic inflammatory responses. Meanwhile, up-regulated lipid metabolism related

pathways (GO 1900076, GO 0031670) and down-regulated DNA duplication (GO 0044271) were also

significant. Additionally, highly enriched glycolytic fermentation (GO 0048523) was upregulated

during progression from Middle Age to Elder group (Figure 5A).

Our analysis found that during age progression from Middle Age to Elder groups, obese females SATs

experienced increased inflammatory mediators (GO 0009888, GO 0048869), immune response

suppression (GO 0099111, GO 0032984), wound healing (GO 0002697), mitochondrial gene

expression (GO 0030705), angiogenesis (GO 0035509, GO 0031622) and superoxide dismutase

activity (GO 0062012) (Figure 5A) Interestingly, based on GOBP analysis of pathways involved in

specific co-expression modules during aging, most aging related alterations of GOBP pathways in

non-obese females were identified in progression from Youth to Middle Age. For obese females, they

present in progression from Middle Age to Elder.

Based on DiffCoEx analysis, we constructed the molecular regulatory networks on our focused

aging-related biological processes under the influence of obesity. Each part in the networks revealed

the interaction patterns of genes and miRNAs involved in aging-related core biological alterations

(immune response, inflammatory state, mitochondria-related proteins, oxidative stress and lipid

metabolism) (Figure 5B). Finally, 8 genes and 2 miRNAs involved in various aging related biological

functions or at the center of the networks, were chosen for display (Figure 5C-D). The selected

molecules were regarded as key nodes with significantly different co-expression patterns and might be

the decisive factors in respective SAT aging processes in obese and non-obese females.

4. Discussion

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
SAT, along with visceral and yellow fat, constitute the systemic adipose tissue. Previous studies have

shown that during visceral fat aging, dysfunctional adipogenesis alters normal physiology, promoting

various pathological conditions, including metabolic disorders . Dynamic redistribution between


26-27

SAT and VAT during aging implies that as with VAT, SAT-associated physiological changes also

occur28. However, few studies have investigated SAT aging profiles. Here, we first profiled the SAT

aging landscape using multi-omics analysis, and identified specific aging-related biological processes,

interaction networks and associated core molecules that may be considered factors in SAT aging.

Aging is associated with immunosenescence, which weakens immunity and enhances susceptibility to

infections and cancer29. Immunosenescence in SAT is linked to altered immune and inflammatory

processes in aging. GOBP analysis of immune related pathways and regulatory patterns of highlighted

core molecules (Figure 2C, 3A) found that increased apoptosis and decreased lymphocytes and

leukocytes activation contribute suppressed immune response during aging. Additionally, development

of chronic low-grade SAT inflammatory state also characterized progression from Youth to Middle Age,

accompanied by elevated cytokine production as well as mononuclear and macrophage infiltration

(Figure 2C). We find that highly expressed proinflammatory factors, including IL6 and cytokines, are

closely linked to lipotoxicity and metabolic disorders30-31. This inflammatory SAT state may cause

regional chronic pain and dysfunction of parenchymal cells adjacent to inflammatory adipose tissue32-33.

Immunosenescence was associated with immune suppression, particularly humoral immunity, during

progression of Middle Age to Elder (Figure 2C-D).

Insulin resistance, an important feature of type 2 diabetes34-35, is associated with SAT. During

progression from Youth to Middle Age, SAT TNF production and mitochondrial gene expression may

promote insulin resistance (Figure 2C)36-37. We find that increased TNF superfamily cytokines

expression correlates with aging and may promote insulin resistance by interfering with response to

insulin. Besides, redundant TNFs and cytokines may accelerate fat catabolic processes and suppress

preadipocytes differentiation38, causing release of more fatty acids, lipid metabolites, and

proinflammatory factors into surrounding tissues and blood, promoting insulin resistance39-40. Previous

studies have shown that insulin resistance and diabetes are closely associated with impaired

mitochondrial function in humans and rodents41-42. Remarkably, we observed significantly reduced

mitochondrial gene expression in SAT during aging (Figure 2C), which may impair mitochondrial fat

oxidation and cooperate with TNF to promote insulin resistance.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
In this study, we characterized altered metabolic patterns and decreased ATP production in SAT during

aging and observed that metabolism transitioned from enhanced lipid metabolism to enhanced

carbohydrate metabolism (Figure 2C, 3A). Various studies have shown that body fat redistribution

during aging increases trunk fat (represented by abdominal fat), and reduces peripheral SAT43-44.

During progression from Youth to Middle Age, increased lipid and fatty acid biosynthesis, and reduced

lipid transport may synergistically contribute to excessive deposition of abdominal SAT (Figure 2C),

which may affect energy metabolism and lead to abdominal obesity associated pathologies in Middle

Age. In later aging processes, significantly upregulated carbohydrate metabolism may be a

compensatory mechanism for ATP supply changes following decreased fat mass in the Elder group

(Figure 2C).

Previous studies show that defective angiogenesis is common in the Elder group45. Consistently, SAT

angiogenesis is suppressed during aging, enhancing vascular smooth muscle cell apoptosis in

progression from Youth to Middle Age (Figure 2C). Excessive deposition of abdominal SAT and

reduced angiogenesis in Middle Age contribute to regional hypoxia and chronic inflammation9. Thus,

aging related defective angiogenesis is unfavorable for wound healing and fat graft survival after

autologous fat transplantation46-47.

We find that excessive VAT and SAT accumulation contributes to obesity, which characterizes SAT

aging. Our data implies that SAT between obese and non-obese females exhibits different aging

features at specific stages. DiffCoEx analysis revealed that changes in aging-related biological

processes in non-obese females mainly ocurred during progression from Youth to Middle Age. In obese

females the changes were more significant during pregression from Middle Age to Elder (Figure

5A-B).

DiffCoEx analysis showed that SAT immunosenescence and inflammatory state presented different

aging-associated variation patterns in obese and non-obese females. In non-obese females, the

characteristically SAT upregulated monocyte chemotactic protein-1 (MCP-1) may mediate chronic

inflammation during progression from Youth to Middle Age (Figure 5A)48. In obese females, during

progression from Middle age to Elder, suppression of natural killer cell differentiation and the

biological activity of immunoglobulin contribute to nonspecific immunity immunosenescence. Our

data showed that obesity aggravates SAT aging-related inflammation, mainly by enhancing infiltration

by inflammatory cells and cytokines (Figure 5A). Furthermore, reduction in HDAC4, VAV1 and TNF

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
expression in obese people suppressed development and activation of T and B cells, and promoted SAT

immunosenescence and inflammation49-50 (Figure 5C). Besides, inflammation, along with impaired

wound healing-related inflammatory responses and suppression of SAT angiogenesis, may cooperate to

promote chronic wounds and lower fat graft survival in obese females51-52 (Figure 5A)

DiffCoEx analysis shows that dysfunctional SAT metabolism and accelerated aging are remarkable

aging characteristics in obese females. Since compensatory lipid catabolism and metabolism, and

glycolytic fermentation were impaired in obese females, regional females SAT may have experiencd

excessive lipid deposition and insufficient energy supply. Additionally, increased lipid production and

deposition was associated with enhanced compensatory LPL expression in Middle Age for non-obese

female but not in obese female. Decreased FGF19 expression in obese people suppressed lipid

metabolism53 (Figure 5C). The 2 factors contribute excessive regional lipid accumulation and lipid

metabolism dysfunction in obese females54. Moreover, elderly obese females are more likely to develop

type II diabetes. Our data suggest that decreased SAT mitochondrial gene expression during aging may

reduce fatty acid beta-oxidation levels and elevate the risk of metabolic disorders for obese females55.

Additionally, during progression from Middle Age to Elder, obese female SAT was in an accelerated

aging status attributable to increased oxidative stress due to reduced superoxide dismutase activity56

(Figure 5A).

SAT is widely distributed in the human body and exhibits aging-related features that are desirable in

aging models57-58. Systemic aging is accompanied by immunosenescence, inflammation and impaired

angiogenesis as demonstrated in numerous studies5, 59, including ours. SAT metabolic change is unique,

and may improve our understanding of metabolic variation patterns associated with aging and

aging-related metabolic disorders. Integrative analysis of SAT multi-omics aging profiles implied that

there was little regulatory relationship between aging associated DEGs and DMGs. Therefore, we

believe that methylation has little influence on the expression of aging related genes during SAT aging

in obese and non-obese females. Relative to non-obese females, methylation is more likely to modulate

gene expression in obese individuals (Sheet S3). These findings suggest that future studies on aging

related regulatory relationships should focus on DEGs-DEMs interaction networks and the differential

gene methylation characteristics during SAT aging in obese and non-obese females.

Our study has the following limitations. First, the multi-omics expression profiling data used were all

obtained from GEO. Since no project fully matched our research direction, only relevant sections of the

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
dataset were selected for further exploration. Second, all data analyzed in this study were collected

from females and therefore may not reflect SAT in males. Third, sampling was limited to abdominal

SAT and therefore is not representative of peripheral SAT. Fourth, we did not uncover strong evidence

that genes in our focused biological processes were regulated by methylation, and only the interaction

network between genes and miRNAs was demonstrated. Fifth, we didn’t perform experimental

validation of key molecules and pathways. Despite these limitations, the robustness of the selected

datasets and appropriate statistical models, insured reliable conclusions. We will next perform single

cell sequencing using a similar experimental design and experimental verification to validate our

observations.

In conclusion, herein, for the first time, we systematically describe the entire landscape of aging-related

molecules, biological processes, and core regulatory networks for the SAT (Fig.6). In the aging process,

gene expression and methylation profiles exhibited variation patterns, while there was no distinct

regulatory relationship between them. SAT showed specific aging-related characteristics in the different

aging stages. We established that immunosenescence, chronic inflammatory state, impaired

mitochondria, alterations of the metabolic patterns, and defective angiogenesis are typical physiological

characteristics of the female SAT associated with aging. These aging-related physiological

characteristics of the SAT were shown to contribute to numerous pathological states (such as insulin

resistance, regional excessive lipid deposition, and chronic pain) and decrease the fat graft survival,

further affecting the clinical application of SAT. Moreover, comparing the aging patterns of SAT

between obese and non-obese females, we found that the characteristics of SAT aging in obese females

were more remarkable from the middle age to the elderly age. The aging process of SAT for obese

females was more likely characterized by nonspecific immune impairment, reduced metabolic

compensatory, exacerbated impaired mitochondria and angiogenesis, and increased oxidative stress.

Therefore, the SAT of obese females showed an accelerated aging process with higher risks of type 2

diabetes and developing chronic nonhealing wounds compared with non-obese females during aging.

Since the aging SAT has biological processes similar to the aging systemic tissue and has its specific

metabolic characteristics, we speculate that SAT is a potential novel biological model to explore aging

and aging-related disorders. This is the first systematic study exploring the aging process of the human

SAT and the relationship between the aging process and obesity. Our findings provide novel insights

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
into the physiological changes of SAT during aging, which will guide future anti-aging and obesity

research.

Acknowledgement

We thank the following institution for giving us powerful support for raw data collection and

multi-omics data analysis: Department of Burns and Cutaneous Surgery, Xijing Hospital, Fourth

Military Medical University. The authors thank Freescience Academic Services

(freescience@zju.edu.cn) for English language editing and review services.

Funding source

This study was funded by the National Natural Science Foundation of China (No. 81772071; No.

81971835; No. 81772072).

Declaration of Competing Interests

All authors declare no competing interests.

Author contributions

DHH, HTW and ZCL contributed to the study and experimental design. ZCL, SW and SJL contributed

to raw data collection. ZCL, SW, and XWY contributed to the multi-omics data analysis. ZCL,YCW,

JQL, LL and XJB contributed to the quality control and verification of data. SJL, XJW and PJ

contributed to the additional data analysis. ZWX and SW contributed to the quality control of the figure

and table. ZCL, SJL, and SW wrote the manuscript. All authors read and approved the final manuscript.

References

[1]. Khan, S.S., B.D. Singer and D.E. Vaughan, Molecular and physiological manifestations and

measurement of aging in humans. Aging Cell, 2017. 16(4): p. 624-633.

[2]. Kuk, J.L., et al., Age-related changes in total and regional fat distribution. Ageing Research

Reviews, 2009. 8(4): p. 339-348.

[3]. Lumeng, C.N. and A.R. Saltiel, Inflammatory links between obesity and metabolic disease. J Clin

Invest, 2011. 121(6): p. 2111-7.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
[4]. Stout, M.B., et al., Physiological Aging: Links Among Adipose Tissue Dysfunction, Diabetes, and

Frailty. Physiology (Bethesda), 2017. 32(1): p. 9-19.

[5]. Pararasa, C., C.J. Bailey and H.R. Griffiths, Ageing, adipose tissue, fatty acids and inflammation.

Biogerontology, 2015. 16(2): p. 235-48.

[6]. Taketani, H., et al., Aging-associated impairment in metabolic compensation by subcutaneous

adipose tissue promotes diet-induced fatty liver disease in mice. Diabetes Metab Syndr Obes, 2019. 12:

p. 1473-1492.

[7]. Ward, Z.J., et al., Projected U.S. State-Level Prevalence of Adult Obesity and Severe Obesity. N

Engl J Med, 2019. 381(25): p. 2440-2450.

[8]. Perez, L.M., et al., 'Adipaging': ageing and obesity share biological hallmarks related to a

dysfunctional adipose tissue. J Physiol, 2016. 594(12): p. 3187-207.

[9]. Mancuso, P. and B. Bouchard, The Impact of Aging on Adipose Function and Adipokine

Synthesis. Front Endocrinol (Lausanne), 2019. 10: p. 137.

[10]. Purcell, C. and S.M. Taylor, Idiopathic facial lipoatrophy in a healthy middle-aged woman: a case

report. Journal of Otolaryngology - Head & Neck Surgery, 2019. 48(1): p. 1-4.

[11]. Cansancao, A.L., et al., Brazilian butt lift performed by board certified Brazilian plastic surgeons:

Reports of an expert opinion survey. Plast Reconstr Surg, 2019.

[12]. Alba, D.L., et al., Subcutaneous Fat Fibrosis Links Obesity to Insulin Resistance in Chinese

Americans. The Journal of Clinical Endocrinology & Metabolism, 2018. 103(9): p. 3194-3204.

[13]. Bódis, K. and M. Roden, Energy metabolism of white adipose tissue and insulin resistance in

humans. European Journal of Clinical Investigation, 2018. 48(11): p. e13017-n/a.

[14]. Son, K.H., et al., Age-related accumulation of advanced glycation end-products-albumin,

S100beta, and the expressions of advanced glycation end product receptor differ in visceral and

subcutaneous fat. Biochem Biophys Res Commun, 2016. 477(2): p. 271-6.

[15]. Hirose, H., et al., Effects of Aging on Visceral and Subcutaneous Fat Areas and on Homeostasis

Model Assessment of Insulin Resistance and Insulin Secretion Capacity in a Comprehensive Health

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
Checkup. J Atheroscler Thromb, 2016. 23(2): p. 207-15.

[16]. Kuk, J.L., et al., Waist circumference and abdominal adipose tissue distribution: influence of age

and sex. Am J Clin Nutr, 2005. 81(6): p. 1330-4.

[17]. Keenan, C.R. and R.S. Allan, Epigenomic drivers of immune dysfunction in aging. Aging Cell,

2019. 18(1): p. e12878.

[18]. Krautbauer, S., et al., Free fatty acids and IL-6 induce adipocyte galectin-3 which is increased in

white and brown adipose tissues of obese mice. Cytokine, 2014. 69(2): p. 263-71.

[19]. Isakson, P., et al., Impaired preadipocyte differentiation in human abdominal obesity: role of Wnt,

tumor necrosis factor-alpha, and inflammation. Diabetes, 2009. 58(7): p. 1550-7.

[20]. Hysing, E.B., et al., Detection of systemic inflammation in severely impaired chronic pain

patients and effects of a multimodal pain rehabilitation program. Scand J Pain, 2019. 19(2): p. 235-244.

[21]. Hefetz-Sela, S. and P.E. Scherer, Adipocytes: impact on tumor growth and potential sites for

therapeutic intervention. Pharmacol Ther, 2013. 138(2): p. 197-210.

[22]. de Mutsert, R., et al., Associations of Abdominal Subcutaneous and Visceral Fat with Insulin

Resistance and Secretion Differ Between Men and Women: The Netherlands Epidemiology of Obesity

Study. Metab Syndr Relat Disord, 2018. 16(1): p. 54-63.

[23]. Gotoh, H., et al., Contribution of subcutaneous fat accumulation to insulin resistance and

atherosclerosis in haemodialysis patients. Nephrol Dial Transplant, 2009. 24(11): p. 3474-80.

[24]. Sleigh, A., et al., Mitochondrial dysfunction in patients with primary congenital insulin resistance.

J Clin Invest, 2011. 121(6): p. 2457-61.

[25]. Acosta, J.R., et al., Increased fat cell size: a major phenotype of subcutaneous white adipose tissue

in non-obese individuals with type 2 diabetes. Diabetologia, 2016. 59(3): p. 560-70.

[26]. Zoller, V., et al., TRAIL (TNF-related apoptosis-inducing ligand) inhibits human adipocyte

differentiation via caspase-mediated downregulation of adipogenic transcription factors. Cell Death Dis,

2016. 7(10): p. e2412.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
[27]. Miller, M.R., et al., Levels of free fatty acids (FFA) are associated with insulin resistance but do

not explain the relationship between adiposity and insulin resistance in Hispanic Americans: the IRAS

Family Study. J Clin Endocrinol Metab, 2012. 97(9): p. 3285-91.

[28]. Gao, D., et al., Interleukin-1beta mediates macrophage-induced impairment of insulin signaling in

human primary adipocytes. Am J Physiol Endocrinol Metab, 2014. 307(3): p. E289-304.

[29]. Wang, C.H. and Y.H. Wei, Role of mitochondrial dysfunction and dysregulation of Ca(2+)

homeostasis in the pathophysiology of insulin resistance and type 2 diabetes. J Biomed Sci, 2017.

24(1): p. 70.

[30]. Gonzalez-Franquesa, A. and M.E. Patti, Insulin Resistance and Mitochondrial Dysfunction. Adv

Exp Med Biol, 2017. 982: p. 465-520.

[31]. Sepe, A., et al., Aging and regional differences in fat cell progenitors - a mini-review. Gerontology,

2011. 57(1): p. 66-75.

[32]. Zamboni, M., et al., Effects of age on body fat distribution and cardiovascular risk factors in

women. Am J Clin Nutr, 1997. 66(1): p. 111-5.

[33]. Costantino, S., et al., Epigenetics and cardiovascular regenerative medicine in the elderly. Int J

Cardiol, 2018. 250: p. 207-214.

[34]. Upputuri, P.K., et al., Recent developments in vascular imaging techniques in tissue engineering

and regenerative medicine. Biomed Res Int, 2015. 2015: p. 783983.

[35]. Eto, H., et al., The fate of adipocytes after nonvascularized fat grafting: evidence of early death

and replacement of adipocytes. Plast Reconstr Surg, 2012. 129(5): p. 1081-92.

[36]. Kanneganti, T.D. and V.D. Dixit, Immunological complications of obesity. Nat Immunol, 2012.

13(8): p. 707-12.

[37]. Paulo, E., et al., Adipocyte HDAC4 activation leads to beige adipocyte expansion and reduced

adiposity. J Endocrinol, 2018. 239(2): p. 153-165.

[38]. Hirasaka, K., et al., Deficiency of Cbl-b gene enhances infiltration and activation of macrophages

in adipose tissue and causes peripheral insulin resistance in mice. Diabetes, 2007. 56(10): p. 2511-22.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
[39]. Zhao, R., et al., Inflammation in Chronic Wounds. Int J Mol Sci, 2016. 17(12).

[40]. Rodrigues, M., et al., Wound Healing: A Cellular Perspective. Physiol Rev, 2019. 99(1): p.

665-706.

[41]. Lan, T., et al., FGF19, FGF21, and an FGFR1/beta-Klotho-Activating Antibody Act on the

Nervous System to Regulate Body Weight and Glycemia. Cell Metab, 2017. 26(5): p. 709-718.e3.

[42]. Castellano-Castillo, D., et al., Adipose Tissue LPL Methylation is Associated with Triglyceride

Concentrations in the Metabolic Syndrome. Clin Chem, 2018. 64(1): p. 210-218.

[43]. Heinonen, S., et al., Impaired Mitochondrial Biogenesis in Adipose Tissue in Acquired Obesity.

Diabetes, 2015. 64(9): p. 3135-45.

[44]. Zhang, Y., et al., Obesity-induced oxidative stress, accelerated functional decline with age and

increased mortality in mice. Arch Biochem Biophys, 2015. 576: p. 39-48.

[45]. Hackl, M., et al., miR-17, miR-19b, miR-20a, and miR-106a are down-regulated in human aging.

Aging Cell, 2010. 9(2): p. 291-296.

[46]. Ghatreh-Samani, M., et al., Oxidative stress and age-related changes in T cells: is thalassemia a

model of accelerated immune system aging? Central European Journal of Immunology, 2016. 1: p.

116-124.

[47]. Xu, L., et al., Ablation of PPARgamma in subcutaneous fat exacerbates age-associated obesity

and metabolic decline. Aging Cell, 2018. 17(2).

[48]. Coue, M., et al., Natriuretic peptides promote glucose uptake in a cGMP-dependent manner in

human adipocytes. Sci Rep, 2018. 8(1): p. 1097.

[49]. Jensen, C.H., et al., The imprinted gene Delta like non-canonical notch ligand 1 (Dlk1) associates

with obesity and triggers insulin resistance through inhibition of skeletal muscle glucose uptake.

EBioMedicine, 2019. 46: p. 368-380.

[50]. Del, C.F., et al., RankProd 2.0: a refactored bioconductor package for detecting differentially

expressed features in molecular profiling datasets. Bioinformatics, 2017. 33(17): p. 2774-2775.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
[51]. Ritchie, M.E., et al., limma powers differential expression analyses for RNA-sequencing and

microarray studies. Nucleic Acids Res, 2015. 43(7): p. e47.

[52]. Szklarczyk, D., et al., STRING v11: protein-protein association networks with increased coverage,

supporting functional discovery in genome-wide experimental datasets. Nucleic Acids Res, 2019.

47(D1): p. D607-D613.

[53]. Dweep, H. and N. Gretz, miRWalk2.0: a comprehensive atlas of microRNA-target interactions.

Nat Methods, 2015. 12(8): p. 697.

[54]. Shannon, P., et al., Cytoscape: a software environment for integrated models of biomolecular

interaction networks. Genome Res, 2003. 13(11): p. 2498-504.

[55]. Zhang, B. and S. Horvath, A general framework for weighted gene co-expression network

analysis. Stat Appl Genet Mol Biol, 2005. 4: p. Article17.

[56]. Langfelder, P., B. Zhang and S. Horvath, Defining clusters from a hierarchical cluster tree: the

Dynamic Tree Cut package for R. Bioinformatics, 2008. 24(5): p. 719-20.

[57]. Tesson, B.M., R. Breitling and R.C. Jansen, DiffCoEx: a simple and sensitive method to find

differentially coexpressed gene modules. BMC Bioinformatics, 2010. 11: p. 497.

[58]. Subramanian, A., et al., Gene set enrichment analysis: a knowledge-based approach for

interpreting genome-wide expression profiles. Proc Natl Acad Sci U S A, 2005. 102(43): p. 15545-50.

[59]. Yu, G., et al., clusterProfiler: an R package for comparing biological themes among gene clusters.

OMICS, 2012. 16(5): p. 284-7.

Figure legends

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
Fig.1 Identification of differentially expressed mRNAs, miRNAs and methylation in aging processes
for both obese and non-obese female. Volcano Plots showed the up-regulated (log2(fold change)>0) and
down-regulated (log2(fold change)<0) differential molecules respectively in each group. Venn diagram
represented unique differential molecules in each aging stage and the shared differential molecules that
are up-regulated or down-regulated at the same time in two aging stages, i.e. YG(Youth Group) vs
MG(Middle Age Group) and MG(Middle Age Group) vs EG(Elder Group) .

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
Fig.2 The co-expression patterns of genes and visualization of core GOBP pathways of SAT in aging
process. a. Visualization of gene cluster tree. Based on the adjacency-based dissimilarity of the
hierarchical clustering genes, the tree diagram corresponding to color was established by the dynamic
tree cutting method recognition module. 11 modules were identified with specific co-expression
patterns and notated with different colors. b. 10 aging-related gene co-expression modules with
different enrichment status of SAT in each group. Enrichment Score indicted the ratio of DEGs in a
specific module to all genes involved in the WGCNA network. Fisher P Value also presented the
enrichment level of DEGs in each gene module with a significant level< 0.05. c. Functional enrichment
analysis for selected DEGs. 30 GOBP pathways for up- or down-regulated DEGs were used to
demonstrated with bubble plot. Gene counts(n) showed the overlap between DEGs and the significant
genes from a pathway. P value represented the enrichment status of the selected DEGs, with a
significant level< 0.05. d. Gene set enrichment analysis of genes based on age groups. GOBP items,
normalized enrichment score (NES) and P value were also presented in chart.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
Fig.3 The regulatory networks and heatmaps of core molecules. a. The regulatory networks of
DEGs-DEmiRNAs involved in the core biological processes in SAT aging. The color of each genes in
the network corresponds to its positioned co-expression modules based on WGCNA. b. Heat map of
DEGs involved in core biological processes in different age groups. DEGs were distributed in different
color groups, in consistence with their co-expression modules. c. Heat map of DEmiRNAs involved in
core biological processes in different age groups. DEmiRNAs were assigned in different color groups
based on the positioned co-expression modules of their regulated DEGs.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
Fig.4 DiffCoEx analysis based on aging related genes between obese and non-obese female. a. The
comparative correlation heatmap showed the differentially co-expressed modules between obese and
non-obese female based on aging related genes. The upper diagonal of the main matrix shows a
correlation between pairs of genes among the non-obese female (dark color corresponds to higher
correlations; light, lower correlations). The lower diagonal of the heatmap shows a correlation between
the same gene pairs among the obese female. Color bars at the edge of the square showed modules in
which genes co-expressed differently between obese and non-obese female (second column). b. 7
aging-related differential co-expression modules with different enrichment status of SAT in each group.
Enrichment Score indicted the ratio of DEGs in a specific module to all genes involved in the
DiffCoEx network. The significant level of Fisher P Value is < 0.05. c. Differential co-expression
network of SAT between obese and non-obese female. Nodes in different colors represented genes in
different modules. Red lines showed stronger correlation in obese female than in non-obese female;
blue lines show stronger correlation for non-obese female. The thickness of the connections between
two nodes represented the correlation strength.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
Fig.5 Specific aging related alteration of biological processes and core genes in obese or non-obese
female based on DiffCoEx analysis. a. On basis of DiffCoEx analysis, specific aging related GOBP
enrichment pathways for obese or non-obese female were demonstrated with bubble plots. Gene
counts(n) showed the overlap between DEGs and the significant genes from a pathway. P value
represented the enrichment status of the selected DEGs, with a significant level< 0.05. b. The
regulatory networks of DEGs-DEmiRNAs involved in the core biological processes based on DiffCoEx
analysis of aging process between obese and non-obese female. The color of each gene in the network
corresponds to its positioned modules based on DiffCoEx analysis. “*” represented the core molecules
potentially performing the important functions in SAT aging. c. Box plots of expression of selected
core genes of SAT for both obese and non-obese female in different age groups. d. Box plots of
expression of selected core miRNAs of SAT for both obese and non-obese female in different age
groups.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
Fig.6 The redistribution and alterations of biological processes of SAT, and high-risk diseases related to
adipose tissue status. First, we showed the redistribution of SAT in aging. More SAT was accumulated
in Middle Age among non-obese female. Hypertrophy and a disorganized array of adipocytes were
exhibited among obese female. Second, we showed the changes in biological status of SAT in different
age groups and listed the high-risk diseases, which was increased caused by the variations of biological
status of SAT in aging process. Third, we showed the specific aging-related variation patterns of SAT
affected by obesity and more significantly increased risks of presented diseases were tagged for obese
female.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
DEmRNA DEmiRNA DMR

Obese Nonobese Obese Nonobese Obese Nonobese

Youth 3 6 3 6 3 6

Middle Age 11 9 11 9 11 9

Elder 3 5 3 5 3 5

Table1 The expression profiling data were obtained from 37 female, which were divided into groups
according to ages and obese status.

Non-obese population Obese population

YG vs.MG MG vs. EG YG vs.MG MG vs. EG

Up-regulated 557 436 388 451


DEmRNAs
Down-regulated 591 397 323 568

Up-regulated 12 14 6 35
DEmiRNAs
Down-regulated 22 11 34 25

Up-regulated 117 100 150 135


DMRs
Down-regulated 87 64 99 70

Table 2 Description of DEGs, DEMs and DMGs for SAT in different age groups among non-obese and
obese female.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
Supplementary tables, figures and legends

Annotated
Data set Platform Probe
molecules

GPL6244
mRNA Profiling GSE25401 32331 16180
(Affymetrix Human Gene 1.0 ST)

GPL8786
miRNA Profiling GSE25470 (Affymetrix Multispecies 847 805
miRNA-1 Array)

GPL8490
Methylation Profiling GSE24884 (Illumina HumanMethylation27 27578 12701
BeadChip)

Table S1 The data sets platforms probes and annotated molecules of the multi-omics chip data
including gene profiling, miRNA profiling and gene methylation profiling.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
Spearman Corr.(r) P value

AKAP6 -0.308270677 0.185681755

TOR2A -0.248120301 0.290194328

AMELX -0.239097744 0.308552435

MIA2 -0.060150376 0.801607069

DHX33 0.012030075 0.961971391

ZNF512 0.012431351 0.962231851

ZNF408 0.081203008 0.73347559

MYEOV 0.159458454 0.501889551

ASB9 0.181954887 0.440913257

BRSK1 0.184962406 0.433273062

AADACL2 0.210526316 0.371305188

CD1C 0.231578947 0.324387987

RNF14 0.326315789 0.160224304

ERRFI1 0.333834586 0.150382557

PRPH2 0.357894737 0.121794087

BTBD2 0.393984962 0.086660675

Table S2. Spearman correlation coefficient (r) and P value of the correlation between the expression
of 16 selected DEGs and methylation level among non-obese female. “r < −0.5, P ≤ 0.05” denoted
statistical significance.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
Spearman Corr.(r) P value

IL17A -0.279411765 0.276432988

TRIM55 -0.237745098 0.356747173

MUC1 -0.205103015 0.429704159

IL23R -0.031862745 0.905925224

NFKBIB -0.019607843 0.94347712

ALG13 0.056372549 0.831480414

COMT 0.115196078 0.659518278

NR5A2 0.139705882 0.592010128

ADD3 0.151960784 0.559403278

SMCO4 0.178921569 0.490666292

ABCC9 0.18872549 0.466762825

COL3A1 0.208333333 0.420812044

TLR3 0.367647059 0.14710262

ITLN2 0.424019608 0.091305667

UNC5D 0.705882353 0.002128774

Table S3. Spearman correlation coefficient (r) and P value of the correlation between the expression
of 15 selected DEGs and methylation level among obese female. “r < −0.5, P ≤ 0.05” denoted statistical
significance.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
WGCNA Module turquoise blue brown yellow green red

Gene Numbers 526 509 505 411 407 405

WGCNA Module black pink magenta purple greenyellow out of modules

Gene Numbers 387 294 257 245 192 422

Table S4 The numbers of co-expression DEGs in each gene module based on WGCNA for non-obese
female.

WGCNA Module black blue brown green

Gene Numbers 38 80 43 39

WGCNA Module yellow red turquoise out of modules

Gene Numbers 40 38 261 4021

Table S5 The numbers of differential co-expression genes in each gene module based on WGCNA
between obese and non-obese female.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
Fig. S1 Aging spectrum of miRNA and the DEMs-DEGs interaction relationships. a. Heat map of the
expression of 56 DEMs in different age groups among non-obese female. b. Flowchart for exploring
the DEMs-DEGs interaction relationships among non-obese female.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
Fig. S2 Aging spectrum of gene methylation, flowchart and functional enrichments. a. Heat map of the
expression of 111 DMGs in different age groups among non-obese female. b. Flowchart for exploring
the regulatory relationships between aging related gene expression and methylation of SAT. “r < -0.5,P
≤0.05” in Spearman correlation analysis denoted statistical significance. c and d. Top 20 GOBP
pathways of the aging-irrelevant methylated genes among non-obese(c) and obese female (d),
respectively.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
Fig. S3 Analysis of the network topology for adjacency matrix weighting parameters (power). a. 4560
genes united by the 10 parts were considered as the potential aging-related DEGs for WGCNA.
b.Hierarchical average linkage clustering for the above screened DEGs. Branches of the dendrogram
represent DEGs with similar expression patterns. c and d. The x-axis represents soft threshold (power),
and the y-axis represents the scale free fitting index and connectivity for each power. The
soft-thresholding power for network construction was set at 14.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
Fig.S4 Flowchart for target co-expression DEGs and DiffCoEx DEGs. a. Flowchart for the target
co-expression DEGs in non-obese female. The Up- and Down-regulated co-expression modules of
WGCNA were separately intersected with the corresponding co-expression DEGs in different age
groups for further enrichment analysis. b. Flowchart for the target DiffCoEx DEGs between non-obese
and obese female. The DiffCoEx modules of WGCNA were intersected with the corresponding DEGs
profiling in different groups divided by age and obese status.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
Sheet S1 Detailed information of the 37 samples and their sample ID (GSM) corresponding to
multi-omics data.

Sheet S2 Spearman correlation coefficient (r) and P value of the correlation between gene methylation
profiling and gene expression profiling among non-obese female.

Sheet S3 Spearman correlation coefficient (r) and P value of the correlation between gene methylation
profiling and gene expression profiling among obese female.

Sheet S4 The description of the aging-related biological processes among non-obese female based on
GOBP enrichment analysis.

Sheet S5 The description of specific aging-related biological processes based on DiffCoEx analysis
between non-obese and obese female.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
er has not been peer reviewed. Electronic copy available at: https://ss
er has not been peer reviewed. Electronic copy available at: https://ss
er has not been peer reviewed. Electronic copy available at: https://ss
er has not been peer reviewed. Electronic copy available at: https://ss
er has not been peer reviewed. Electronic copy available at: https://ss
er has not been peer reviewed. Electronic copy available at: https://ss
  DEmRNA DEmiRNA DMR
Obese Nonobese Obese Nonobese Obese Nonobese
Youth 3 6 3 6 3 6
Middle Age 11 9 11 9 11 9
Elder 3 5 3 5 3 5
Table1 The expression profiling data were obtained from 37 female, which were divided into groups
according to ages and obese status.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869
    Non-obese population Obese population

    YG vs.MG MG vs. EG YG vs.MG MG vs. EG

Up-regulated 557 436 388 451


DEmRNAs
Down-regulated 591 397 323 568

Up-regulated 12 14 6 35
DEmiRNAs
Down-regulated 22 11 34 25

Up-regulated 117 100 150 135


DMRs
Down-regulated 87 64 99 70

Table 2 Description of DEGs, DEMs and DMGs for SAT in different age groups among non-obese
and obese female.

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=3666869

You might also like