Mitochondria are membrane-enclosed organelles found in the cytoplasm of most eukaryotic cells. They are composed of two lipid membranes (the inner and the outer membranes) and have their own genome. Mitochondria are key elements for the cell functioning. They are the main source of energy (ATP production) and are involved in a range of cellular processes such as cell differentiation, apoptosis, signaling or cell cycle control. An increasing number of diseases are related to mitochondrial dysfunctions including, among others, Friedreich's ataxia, Leber's neuropathy, myopathies and neurodegenerative disorders such as Parkinson's, Alzheimer's and Huntington's diseases.

In yeast Saccharomyces cerevisiae, the construction of a functional organelle requires, in addition to the lipid membranes, the correct assembly of about 800 proteins. These proteins are encoded both by the nuclear genome and the mitochondrial genome. Regulatory mechanisms are required to precisely coordinate in time the expression of the two genomes (see the figure on the left).

Also, mRNAs produced in the nucleus can have different fates in the cytoplasm. Some are directly translated into proteins, whereas others are first addressed to mitochondria, before being translated into proteins. These mRNAs are called MLR for 'Mitochondrial Localized mRNA'. The MLR phenomenon is a spatial regulation, i. e. mRNAs have to be localized at the right place in the cell (near mitochondria).

Mitochondrial biogenesis requires time and space regulation processes (see the figure below). Several studies report spatio-temporal regulations in yeast Saccharomyces cerevisiae. In Saint-Georges et al. (2008), the authors described three classes of nuclear mRNAs encoding mitochondrial proteins differing in their site of translation. Class I and class II mRNAs are found near mitochondria, whereas class III mRNAs are translated on free cytoplamic polysomes. The subcellular localization of class I mRNAs is dependant on the activity of the RNA binding protein Puf3p, whereas class II mRNAs are Puf3p independant.

In Tu et al. (2005), the authors used a yeast system with synchronous properties and performed microarray experiments. They described a Yeast Metabolic Cycle (YMC) in association with a periodicity in the expression of the genome. They observed the existence of similar temporal expression patterns in functionally connected groups of genes. In particular, most of the genes associated with mitochondria appeared to be expressed with exceptionally robust periodicity, consistent with the variations in the amount of dissolved oxygen in the medium of synchronized cell cultures.

In Lelandais et al. (2009), the authors developed the EDPM algorithm to analyze in more details the Tu et al. microarray dataset. They identified functional modules which, expressed sequentially and periodically, contribute to the complex and intricate mitochondrial architecture.

The principle of the EDPM algorithm consists in breaking each gene expression pattern (measured with microarray experiments) down into a mixture of 15 model profiles (see the figure below), which are time-delayed mathematical functions mimicking ideal expression oscillations. EDPM allows the calculation of W-values that quantify the contribution of each model profile to the observed gene expression measurements. As all the model patterns differ only in term of the time interval between patterns, the W-values calculated with EDPM indicate the time phase during the Yeast Metabolic Cycle at which the gene is strongly expressed.

Comparison between the MLR classes defined in Saint-Georges et al. and the EDPM phases defined in Lelandais et al. identified a substantial overlap between phase A and class I genes. This observation implies a coordination between mRNA oscillations and site of the translation in the cytoplasm. It requires both transcriptional and post-transcriptional regulations.

To analyze the balance between transcriptional and post-transcriptional regulation processes, Garcia-Martinez et al. (2004) defined a r coefficient to estimate the correlation between values of trascription rate (TR) and mRNA levels. They could observe that whereas TR is the main step for expression regulation for many genes (positive r coefficient), some groups of genes seem to increase their mRNA level even having a decrease in TR (negative r coefficient). Early-expressed nucleo-mitochondrial genes (assembly factors, translation machinery proteins) belong to this post-transcriptionnaly regulated gene group, whereas late expressed genes (amino acids synthesis) exhibit a positive r coefficient.

626 genes involved in mitochondrial biogenesis were analyzed with the EDPM algorithm and classified into 6 clusters according to their W-values (see the figure below). These clusters were named A to F and comprised distinct subclasses of genes:

• That are periodically expressed;
• Whose mRNA quantities peak in different time windows of the Yeast Metabolic Cycle

The mRNAs coding for mitochondrial proteins peak at different times during the Yeast Metabolic Cycle. The first mRNAs to appear are those for genes whose function is associated with the translation machinery (or regulation) and assembly factors (phase A), followed by those involved in the synthesis of the respiratory chain structural proteins (phase B) and finally mRNAs coding for enzymes involved in amino-acid biosynthesis are more abundant in phase F.