Predicting metabolic preferences through transcriptomics: a data-driven approach to align metabolic signatures with gene expression profiles

Complex organisms such as mammals have a sophisticated metabolic network to meet energy demand under varying conditions. This network, which includes the exchange of metabolites between organs, is absent in <i>ex vivo</i> model systems like cell culture or isolated organ perfusion. These systems therefore require external management of metabolic substrates; since failure to meet the specific metabolic requirements will lead to cellular stress, non-physiological behaviour and in turn limited translatability, it should be ensured that model systems exhibit <i>ex vivo</i> metabolism that recapitulates <i>in vivo</i> processes. To better support but also assess tissue and cell metabolism under <i>ex vivo</i> conditions, it is thus crucial to be knowledgeable of their specific <i>in vivo</i> metabolic preferences. As <i>in vivo</i> organ- and cell-specific metabolic preferences are only partially characterised, a surrogate marker of metabolism is required that can easily be measured in both <i>in vivo</i> and <i>ex vivo</i> isolated organ or cell culture systems. In an attempt to identify surrogate predictive markers of metabolism that could be easily measured in <i>ex vivo</i> model systems, we investigated the extent to which organ-specific metabolite consumption and production patterns (referred to as "metabolic signatures") from available arteriovenous flux data align with organ-specific metabolic gene expression patterns. Whilst different tissues displayed distinctive patterns in the consumption and production of metabolites, these did not directly correspond to expression of known metabolic genes. These findings are indicative of the complexity of mammalian metabolism.