Application of genome-wide metabolomic models to reveal the secret of longevity in elevated fatty acid oxidation

Whole-genome metabolomics model reveals secret of longevity: elevated fatty acid oxidation
Article title: System-level metabolic modeling facilitates unveiling metabolic signature in exceptional longevity

Journal: Aging Cell

Impact factor: 9.
304

Cooperative client: Kunming Institute of Zoology, Chinese Academy of Sciences,

provided by Baiqu Bio Service: Off-target Whole Metabolism

Research Background
It is well known that metabolic control plays a critical role in regulating healthspan and longevity in various organisms, but is critical to the systemic metabolic profile of centenarians, a paradigm for healthy aging and longevity in humans Little is known
.
At the same time, how to well describe the metabolic state at the biological system level of interest is still a major challenge in systems metabolism research
.


To address this challenge and better understand the metabolic mechanisms of healthy aging, we developed a genome-wide precision metabolic modeling (GPMM) approach that quantitatively integrates transcriptomic, proteomic, and kinematic data Network Predictive Modeling
.
Benchmark analysis showed that GPMM successfully characterized the characteristic metabolic reprogramming of the NCI-60 cell line; the R2 between prediction and experimental measurement was 0.
86 compared to existing methods, greatly improving the performance of the modeling
.

Using this method, we examined the metabolic network of the Chinese centenarians and determined that elevated fatty acid oxidation (FAO) was the most prominent metabolic feature of these long-lived individuals
.


Research resultIn this study, we developed a genome-wide precise metabolomic modeling (GPMM) approach to address the challenge of 'low precision'
.
The method quantitatively integrates enzyme kinetic information from knowledge bases and enzyme levels from transcriptomic and proteomic data into metabolic models
.
Specifically, we first curated general human metabolomic models and used information from the literature to set upper limits on the uptake rates of major nutrients in the blood
.
A simplified Recon 3D model was subsequently constructed to maximize the number of reactions with Kcats and minimize the number of reactions without Kcats
.
Next, we quantitatively integrated enzyme kinetic parameters and gene expression levels to constrain the upper and lower bounds for each reaction
.
Finally, through flux variation analysis (FVA), reactions at zero flux were removed to obtain a single model, followed by Markov Chain Monte Carlo (MCMC) sampling to identify metabolic changes and key regulators
.


Notably, GPMM enables some in silico analyses not included in the state-of-the-art COBRA method toolbox (COBRA toolbox v3.
0), and thus can be broadly applied to metabolomics engineering and therapy
.
Furthermore, GPMM can be used for personalized metabolomic modeling of precision medicine
.

Since the input transcriptome may be noisy during experiments or mapping, we analyzed the robustness of GPMM to demonstrate whether GPMM has the ability to tolerate gene expression noise
.
We first construct a noise-induced transcriptome by adding random values ​​(i.
e.
artificial noise) to the raw expression data (i.
e.
true expression values) for each gene
.
Specifically, we introduced 1%, 5%, and 10% noise into the gene expression data of the NCI-60 cell line to construct noise-induced transcriptomes
.
Then, metabolomic modeling was performed on the real and noise-induced expression datasets and flux results were compared between the two datasets
.
The results showed that the R-squared was greater than 0.
98 for each cell line at 1%, 5%, or 10% noise (Fig.
1b)
.
We performed 100 inductions of 5% gene expression noise in the NCI-60 cell line H460 and performed metabolomic modeling using GPMM
.
We obtained an average R2 of 0.
984 between real samples and noise-induced samples (Fig.
1c)
.
Some important fluxes in cancer cells, such as ATP production and lactate secretion, were also consistent across these 100 simulations (Fig.
1d)
.
These results show that the GPMM algorithm is a robust method
.


The Warburg effect is one of the most important cancer hallmarks in NCI-60 cells
.
Therefore, we compared the predicted lactate secretion fluxes with experimental values ​​and found that GPMM predicted lactate secretion in NCI-60 cells well (R2=0.
86, p=2.
2e-24) (Fig.
1e)
.
To elucidate the metabolomic profile of centenarians to better understand why these individuals are able to delay or avoid many severe age-related disease afflictions, we aimed to study the metabolism of centenarians with longevity in a sample from Hainan Province
.
This cohort included 76 centenarians, 54 children of centenarians, and 41 spouses of children of centenarians
.


Next, we applied GPMM to investigate the metabolomic signature of longevity in this cohort
.
In total, we established 171 individual metabolic models of GPMM based on leukocyte transcriptome information
.
Each model contains 3977 reactions, which can be divided into 4 functional components: nutrient uptake (22 reactions), metabolic transport (2478 reactions), enzymatic catalysis reactions (1103 reactions), and secretion and demand reactions (374 reactions) (Fig.
2a)
.


By comparing differences in metabolic profiles between centenarians and young controls (i.
e.
, F1SPs), and adjusting for the effects of age and gender, we obtained 343 up- and 90 down-regulated fluxes
.
The most striking features of all four metabolic processes consistently showed that long-chain fatty acid beta-oxidation (FAO) was elevated in centenarians
.
For example, in the nutrient absorption component, long-chain fatty acids (i.
e.
, octadecenoic acid and octadecenoic acid) and oxygen uptake were significantly elevated in CENs (Fig.
2b); in the metabolite transport component, long-chain fatty acid oxidation (i.
e.
, Peroxisome and mitochondrial transport in subcellular organelles was also elevated in these long-lived individuals (Fig.
2d)
.
Likewise, the enzymatic catalytic components showed that cellular fatty acid storage (i.
e.
, triacylglycerol synthesis), FAO, pyruvate metabolism, branched amino acid metabolism, tricarboxylic acid (TCA) cycle, and oxidative phosphorylation were all elevated in centenarians (Fig.
2c)
.
Consistently, in the secretion and demand components, we found that centenarians released more carbon dioxide and less TCA intermediate metabolites (Fig.
2e)
.


Given the critical role of long-chain fatty acid oxidation in carbon catabolomics, we explored upstream and downstream reactions to this process to determine whether the observed elevation was limited to long-chain fatty acid oxidation or was present in other carbon catabolic pathways in
.
Surprisingly, we found that upstream reactions of long-chain fatty acid oxidation, including fatty acid uptake, activation, and transport, were all elevated in centenarians (Fig.
2f)
.
Downstream reactions of long-chain fatty acid oxidation, half of the tricarboxylic acid cycle (TCA cycle) flux (4 out of 8 reactions) and more than half of the oxidative phosphorylation complexes (3 out of 5 reactions) in centenarians Significantly increased in old age (Fig.
2f)
.

Since higher systemic levels of long-chain fatty acid oxidation lead to higher tissue uptake and consumption of fatty acids, we hypothesized that serum long-chain fatty acid concentrations should be lower in centenarians
.
By generating and analyzing serum metabolomic data from the same long-lived cohort, we obtained 505 metabolites
.
After removing metabolites associated with aging or gender effects, we found 83 down-regulated and 53 up-regulated metabolites in centenarians (Fig.
3a)
.
Among the down-regulated metabolites, 80.
7% (67/83) were fatty acid-like metabolites (FAL) (Fig.
3b)
.


Interestingly, differential abundance score (DA) analysis revealed that 71% (5/7) of fatty acid-like metabolite (FAL) families were significantly down-regulated in centenarians, including phosphatidic acids (PAs), phosphatidylethanolamines (PEs) , phosphatidylcholines (PCs), and long-chain fatty acid sphingomyelin (SM) (Fig.
3c)
.
Specifically, among centenarians, 100% (3/3) of PAs, 100% (15/15) of PEs, 81.
2% (26/32) of PCs, and 100% (1/1) of SMs were significantly decreased (Fig.
3d)
.
In addition to FAL metabolites, free long-chain fatty acids (eg, trans-vaccinic acid and palmitic acid) were also significantly reduced in centenarians (p = 0.
002 and 3.
9e-4) (Fig.
3e,f)
.
Compared with F1SP, F1s had significantly lower trans-vaccine levels (Fig.
3e, p = 0.
004) and palmitic acid levels (Fig.
3f, p = 0.
05)
.
These results suggest that some fatty acid signatures of centenarians, such as reductions in free fatty acids, may be inherited
.
Interestingly, among the upregulated metabolites, the most prominent were bile acids, a group of fatty acid uptake metabolites (Fig.
3c)
.
These results suggest that decreased serum fatty acid concentrations were the most striking feature in our centenarian metabolomic data
.
This observation also explains our previous epidemiological investigation, which found that centenarians had lower total cholesterol compared with F1SPs
.


Study Conclusions
Evidence from serum metabolomics supports this observation
.
Given that elevated fatty acid oxidation declines with normal aging and is impaired in many age-related diseases, studies suggest that elevated levels of fatty acid oxidation have the potential to be a novel feature of healthy aging in humans
.

Article/Aqu Metabolomics