This web page was produced as an assignment for an undergraduate course at Davidson College.

   Gene duplication can impart fragility, not robustness, in the yeast protein interaction network

Summary & Opinion:

The objective of this study was to investigate whether gene duplications may protect cells from genetic perturbations, a widely speculated functional outcome of gene duplication, and to elucidate what molecular mechanisms might serve to confer such an increase in robustness. Both ohnologs (products of whole-genome duplications) and small-scale duplicates (from the duplication of small portions of the genome) are major forces driving evolution and various models have been hypothesized surrounding the long-term evolutionary fate of duplicated genes or “paralogs.” (Magadum et al., 2013) The authors of this study analyzed protein interactions in yeast cells with either wild-type or single-paralog deletion backgrounds in order to determine the functional relationships between paralogs and their in vivo behavior. They concluded that in many cases, paralog pairs require the presence of one another to maintain their interactions and thus may increase mutational fragility, rather than resilience.

This paper characterized and substantiated an additional possible outcome of gene duplications, with negative implications for evolutionary fitness. Although a number of the sample sizes examined were relatively limited, I nonetheless believe that its observations represent an important expansion in understanding of gene duplications and their implications for evolutionary fitness. While single nucleotide mutations were previously considered to be the main molecular mechanism of evolution and determinant of fitness, this paper further emphasizes the importance of the physiological context in which mutations occur in determining their impact. Rather than simply providing a buffer against negative mutations in one copy of a duplicated gene, or allowing increased specialization of a paralog, the authors provide strong evidence that co-dependent paralogs evolved from self-interacting proteins can make particular interaction networks more susceptible to disruption.


Figure 1: The possible functional outcomes of gene duplication and paralog deletion, as well as their experimentally observed in vivo behavior.

To investigate the relative impacts of paralog deletions, the researchers performed protein fragment complementation assays, in which proteins are linked to complementary moieties of the enzyme dihydrofolate reductase (DHFR) and expressed at endogenous levels in living cells. (Remy et al., 2007) Because DHFR plays an essential role in the synthesis of certain amino acid precursors, the interaction of two DHFR-linked proteins is necessary to facilitate colony growth on a restrictive medium. As the growth rate of colonies reflects the quantity of protein complex formed, the assay can be used to measure PPI intensity changes in various genotypic contexts.

The possible fates of duplicated genes relative to one another, excluding pseudo-geneticization of one paralog, are depicted in panel A.  Panel B illustrates how protein interactions of paralogs might diverge over time, as well as the PPI repercussions of paralog deletion in cases of their functional dependency and compensation of protein interactions. Panel C shows the interaction scores, determined from colony size, in WT versus paralog deletion backgrounds. The correlation of scores between backgrounds indicates that the global PPI networks were preserved in cases of deletion, but significantly increased and decreased scores for specific colonies indicate instances of PPI compensation and deletion, respectively. The PPI scores for all interactions mediated by either member of a given paralog pair are presented along horizontal lines in panel D, which exhibits that pairs often demonstrate either compensation or dependency, but rarely both.



Figure 2: Examination of whether a shift in binding equilibrium, attributable either to the increased abundance of the remaining paralog after deletion or a disparity in affinity for specific interactions between paralogs in WT backgrounds, was the primary mechanism of compensation. Protein abundance was measured using flow cytometry of green fluorescent protein (GFP)-tagged proteins in paralog deletion strains.

As shown in panel A, only four of twelve examined proteins exhibited a low magnitude increase in abundance meeting the threshold of significance in the deletion background compared to the WT background, suggesting differential affinity for specific reactions as the primary source of compensation among this examined set of proteins.  To confirm that paralogs can manually exclude their partners, low- and high-copy-number plasmids were used to manually overexpress paralogs. Overexpression resulted in a significant shift in the relative balance of some protein interactions but not others, as shown in panels C and D, respectively. These results supported the hypothesis that paralogs can mutually exclude one another in certain cases of interaction with partner proteins. Panel E details that compensated PPIs are more likely to be decreased by the compensating paralog than non compensated interactions.  


Figure 3: Characterization of the relationship between interactions of paralogs with one another and their functional dependency by computational analysis.

Heterodimers, paralogs that interact with one another, were determined more likely to exhibit dependency than non-interacting paralogs, as plotted in panel A. Panel B indicates that the independent paralog is typically more abundant than its dependent partner in cases of asymmetric dependency. The results of western blotting and flow cytometry using GFP-tagged proteins, show in panel C, indicated that the abundance of the dependent parlor further decreases when its independent partner is deleted. The x-axis represents the abundance of individual paralogs in the WT background and the y-axis their abundance in the deletion background, with the western blots superimposed over the plot. Notably, the deletion of heteromer-forming small-scale duplicates was seen to reduce fitness more than that of both singletons and ohnologs, as illustrated in panel D. This substantiates the authors' assertion that gene duplication, in the case heteromer SSDs, can actually increase the fragility of protein interaction networks, as their deletion is on average more deleterious relative to that of singletons and ohnologs.