Inferring the transcriptional landscape of bovine skeletal muscle by integrating co-expression networks

Abstract

Background: Despite modern technologies and novel computational approaches, decoding causal transcriptional regulation remains challenging. This is particularly true for less well studied organisms and when only gene expression data is available. In muscle a small number of well characterised transcription factors are proposed to regulate development. Therefore, muscle appears to be a tractable system for proposing new computational approaches.

Methodology/principal findings: Here we report a simple algorithm that asks "which transcriptional regulator has the highest average absolute co-expression correlation to the genes in a co-expression module?" It correctly infers a number of known causal regulators of fundamental biological processes, including cell cycle activity (E2F1), glycolysis (HLF), mitochondrial transcription (TFB2M), adipogenesis (PIAS1), neuronal development (TLX3), immune function (IRF1) and vasculogenesis (SOX17), within a skeletal muscle context. However, none of the canonical pro-myogenic transcription factors (MYOD1, MYOG, MYF5, MYF6 and MEF2C) were linked to muscle structural gene expression modules. Co-expression values were computed using developing bovine muscle from 60 days post conception (early foetal) to 30 months post natal (adulthood) for two breeds of cattle, in addition to a nutritional comparison with a third breed. A number of transcriptional landscapes were constructed and integrated into an always correlated landscape. One notable feature was a 'metabolic axis' formed from glycolysis genes at one end, nuclear-encoded mitochondrial protein genes at the other, and centrally tethered by mitochondrially-encoded mitochondrial protein genes.

Conclusions/significance: The new module-to-regulator algorithm complements our recently described Regulatory Impact Factor analysis. Together with a simple examination of a co-expression module's contents, these three gene expression approaches are starting to illuminate the in vivo transcriptional regulation of skeletal muscle development.

Figure 1. The profiles of MYOD1 and MYOG across the 6 transcriptional landscapes.

Their significant correlation in each of the 6 instances explains their inclusion in the Always Correlated landscape.

Figure 2. The frequency distributions of all correlation coefficients in each of the six transcriptional landscapes (black) plus those deemed significant by PCIT (red).

Figure 3. The Always Correlated transcriptional landscape.

Networks were visualised using the organic algorithm of Cytoscape . A) Node size was mapped to average transcript abundance, edge colour was mapped to the sign of the correlation in the “Overall” landscape and node colour was mapped to Gene Ontology process. Node shape was mapped as follows: TFs (triangles), signalling molecules (squares) and chromatin remodelers (diamonds). All other genes (i.e. non-regulators) were mapped as ovals. B) Node size was mapped to number of connections. C) The transcription landscape built from connections with correlation coefficients >0.99.

Figure 4. The expression profiles of mammalian muscle over development.

Representatives from each of the main functional modules are shown: Immune, nuclear and mitochondrially-encoded mitochondrial genes (A); Extra-cellular matrix, fat and glycolysis gene transcription (B); Vasculature, fast and slow twitch muscle (C); and Cell cycle, ribosome and neuron gene transcription (D).

Figure 5. The connectivity of all genes in the Always Correlated transcriptional landscape versus the transcriptional regulators.

Figure 6. The expression profiles of the neuron module genes across the Overall landscape (i.e. the 10 Piedmontese and 10 Wagyu development time points, plus the starvation-realimentation experiment).

The expression profile of the neurogenesis TF TLX3 is also shown, which did not make the module by PCIT but was ranked top by the downstream (nerve) module-to-regulator analysis.

Figure 7. Range in expression level of genes versus frequency.

Distribution of genes in postnatal Piedmontese and Wagyu samples in red and in all Piedmontese and Wagyu samples in blue. Including pre-natal as well as post-natal muscle stages increases the exploration of parametric expression space. An increase in the frequency of genes experiencing moderate-high changes in expression level reduces the formation of spurious edges in the computed co-expression networks.