(26d) Metabolic Network Analysis for Understanding the Biology of Aging

Metabolic network analysis for
understanding the biology of aging

Sudharshan Ravi*, Rudiyanto Gunawan*

*Institute of Chemical and
Bioengineering, ETH Zurich, Switzerland

*Swiss Institute of Bioinformatics,
Lausanne, Switzerland

Aging is a
complex and multifactorial process that causes progressive functional decline
and decreased ability of an organism to respond to stress. The complexity of
the ageing process has motivated applying a systems-oriented approach through
the creation and analysis of cellular networks. In this regard, recent human
omics profiling efforts such as Genotype-Tissue Expression (GTEx), functional
annotation of the mammalian genome (FANTOM), Human Protein Atlas and Human
Metabolic Atlas projects, have generated large-scale human datasets, which
would allow us to study human ageing process holistically. In this study, we
focused on the relationship between ageing and metabolism. While this
relationship has been well documented (Barzilai et al., 2012), most notably by
the longevity effects of caloric or dietary restriction, a genome-wide analysis
of human metabolism during aging has not yet been carried out. Here, we
leveraged on transcriptomics data from the GTEx project and the curated genome-scale
metabolic network model (Recon 2) (Thiele et al., 2013), to identify
age-related metabolic alterations in human. We were also able to identify aging
metabolic signatures in 11 human tissues. Our findings pointed to age-related
alterations in pathways pertaining to cellular energy generation,
branched-chain amino acid homeostasis, fatty acid metabolism and stress
response.

The GTEx database consists of RNA-sequencing measurements of 56202
transcripts collected from 714 subjects whose ages ranges from 20 to 79 years
old. Information on sample collection, quality control and normalization is
available elsewhere (Mele et al., 2015). We employed a linear mixed effects
model to determine the set of “aging genes”, i.e. genes whose expression were
altered with aging across different human tissues.

Y_ijk = b_i+ u_ij
T_j + x_ik (Sex)_k
+ z_i (Age)_k + g_k
+ e_ijk

where Y_ijk is the expression of
gene i in tissue j belonging to the subject k, u_ij T_j
and x_ik (Sex)_k
denote the fixed effects of tissue (T_j) and sex,
respectively, and g_k is the random
effect associated with each subject. To analyse tissue-specific gene expression
changes, the linear mixed effect model was applied on a subset of the
transcriptomic data belonging to that particular tissue, without the term u_ij T_j.
The number of differentially expressed genes with age (FDR<0.01), referred
to below as aging genes, for 11 major tissues are tabulated in Table 1.

Table 1. 
Number of differentially expressed genes

RECON 2 human genome-scale metabolic model consists of 2140 metabolic
genes that control 4821 metabolic reactions. The reactions are classified into
various subsystems based on their role and function. We mapped the aging genes
to the human genome-scale metabolic network by identifying the metabolic
reactions whose enzymes are among the aging genes (see Fig 1). Finally, we
performed Fisher exact tests to determine the over-representation of different
metabolic subsystems among the aging genes.

Fig. 1. Mapping the aging
genes to the genome scale metabolic model allows us to look at over-represented
subsystems by downregulated (highlighted in red) and upregulated (green) aging
genes.

Our analysis
pointed to significant age-related metabolic alterations of pathways pertaining
to cellular energy generation. In particular, tricarboxylic acid (TCA) cycle
and oxidative phosphorylation processes are overrepresented among the set of
downregulated aging genes (see Fig 2A). These pathways are also
over-represented among downregulated aging genes across multiple tissues,
despite the varying number of overlapping aging genes and the number of aging
genes among tissues. Metabolic pathways involved in the fatty acid oxidation
are also downregulated across all tissues. Indeed, impaired fatty acid
oxidation has previously been shown to be associated with mouse aging
(Houtkooper et al., 2011). Meanwhile, inositol phosphate metabolism pathway is
enriched among upregulated aging genes across different tissues (see Fig 2B).
This pathway regulates cell proliferation and is also involved in the energy
sensing pathway PI3K/Akt. Our analysis also showed disruption of amino acid
homeostasis (nucleotide interconversions) and branched chain amino acid
metabolism.

Fig. 2. Enriched subsystems
across multiple tissues by (A) downregulated tissue specific aging genes and
(B) upregulated tissue specific aging genes.

References

Lopez-Otin, C.,
Blasco, M.A., Partridge, L. and Kroemer, G. (2013). The Hallmarks of Aging. Cell,
153, 1194-1217.

Simon, C.J., Dong,
X., Vijg, J., et al. (2015). Genetic evidence for common pathways in human
age-related diseases. Aging Cell, 14, 809-817.

Barzilai, N.,
Huffman, D.M., Muzumdar, R.H. et al. (2012). The critical role of metabolic
pathways in aging. Diabetes, 61, 1315-1322.

Thiele, I.,
Swainston, N., Fleming, R. and Palsson, B. (2013) A community-driven global
reconstruction of human metabolism. Nature Biotechnology, 31, 419-425.

Mele, M., Ferreira,
P., Reverter, F. and Guigo, R. (2015). The human transcriptome across tissues
and individuals. Science, 348, 660-665.

Houtkooper, R.H.,
Argmann, C., Houten, S. and Auwerx, J. (2011). The metabolic footprint of aging
in mice.  Scientific Reports, 1, 134.