Using results from statistical analysis of the degree and eigenvector centrality from the proteins involved in the PDH complex, the biochemistry students were able to design experiments to observe the mechanisms of pyruvate dehydrogenase deficiency.
Using experimental methods such as Yeast-2 Hybrid, CRISPR, Western Blot and gene knockouts, we were able to observe both the causes and possible treatments for PDCD, with yeast, human cells and mice as models. PDCD is a poorly-researched metabolic disease so by identifying protein-protein interactions using the Louvain Community Finding Algorithm, we were able to see how our target proteins PDA1, PDB1, LAT1 and LPD1 interacted with the pyruvate dehydrogenase complex. Using the yeast homologs LAT1, LIP5, AIM22, and NFU1, we designed experiments in yeast models to mimic interactions within the complex, while a human cell line was used to observe the effect of mutations in the LIAS protein and its functional relevance. Yeast and mouse models were used to observe the effect of a lipoic acid diet on the pyruvate dehydrogenase complex and PDCD.
While we tested protein interactions using Yeast-2-Hybrid, pull down assays and co-IP, we could have used other effective methods, such as experiments involving cross-linking analysis. This is an approach that stabilises the interactions through linkages. With this type of analysis, other approaches such as cell lysis or mass spectrometry can be used to analyse protein-protein interactions (Overview of Protein–Protein Interaction Analysis - US n.d.). Other approaches include gel chromatography, Luminescent Oxygen Channelling Assay (LOCI) and Circular Dichroism Spectroscopy (CD) (Miura 2018).
​There were also a few other approaches that we could have taken to explore PDCD treatments. Apart from lipoic acid, we could have also observed dichloroacetate (DCA) as a treatment for PDCD. DCA is a drug primarily used for mitochondrial diseases, and alters the phosphorylation stability in the PDH complex (Stacpoole et al. 2008). Studies have shown that combined DCA and gene therapy treatment could be beneficial in decreasing the symptoms of PDCD (Stacpoole et al. 2008). This could have been interesting to research more in-depth and provides potential directions for future studies. Another treatment that we could have investigated is phenyl-butyrate, which increases PDC activity through increasing de-phosphorylation and inhibiting kinases activity (Ferriero et al. 2014). Hence, this drug also has the potential to be an effective therapy for PDCD. However, studies have shown phenyl-butyrate is responsive to only a select amount of proteins, which suggests that the scope of this treatment may be limited (Ferriero et al. 2014).
A limitation of the mathematical analysis is the use of an unweighted graph to model the interactions between each protein. Although the STRING database provides weightings for each node, only the proteins considered significant were kept and weightings were removed. Certain protein-protein interactions may have a heavier weighting than others and thus the results from the community finding algorithms including Louvain, Leiden and Spectral clustering maybe a poor representation of the biological network. Applying other algorithms to the weighted network could provide novel information to better understand the role of different yeast proteins in PDH deficiency (Liu et al. 2014).