fadA-fwsR as a model for the evolution of new genes

3.4.3.1 The fadA-fwsR mutation is a model example of the creative power of modularity

The evolution of fwsR provides a real-time illustration of the ability for modular protein domains to fuse together to form a novel protein. Such modularity and the ability for modules to combine together – via fusion mutations or other rearrangements – are important for the diversity of protein families. This is evidenced by the rapidly increasing number of identified protein families with multiple domain architectures, compared to newly recognised domains (Levitt 2009). The promiscuity of domains, combined with the recent ability to mine sequencing data, has resulted in theoretical expectations of the mechanisms that can form new modular gene rearrangements. These arrangements include fusions, exon shuffling, frameshifts, chromosomal recombination and retrotransposition (Long et al. 2003; Kaessmann 2010; Ding et al. 2012; Ranz and Parsch 2012; Bornberg-Bauer and Mar Alba 2013).

Gene fusions provide a unique mechanism for the evolution of new genes. Fusion mutations of neighbouring genes allow the combination of distinct functional protein modules and a concomitant change in expression to result in unique protein functionality and expression. Such unique functionality can extend to novel physiology or behaviour, which may enable a new ecological niche to open, leading to adaptive radiation (Gilbert 1978; Patthy 2003; Ranz and Parsch 2012; Rogers and Hartl 2012). Chimeric genes have been identified, using primarily comparative genomic approaches, across a variety of taxa including zebrafish (Fu et al. 2010), Caenorhabditis elegans (Katju and Lynch 2006), Oryza sativa (Wang et al. 2006), Drosophila (Long and Langley 1993; Nurminsky et al. 1998; Wang et al. 2002; Jones et al. 2005) and mammals (Thomson et al. 2000; Courseaux and Nahon 2001), and in some cases the selective coefficients of the chimeric mutation have been identified (Rogers et al. 2010; Rogers and Hartl 2012). However, the molecular interactions and effects resulting from fusions have rarely been identified, with the notable exceptions of

functional studies of Drosophila genes jingwei (Zhang et al. 2004) and sphinx (Dai et al. 2008).

The fadA-fwsR fusion is one of the few adaptive fusion mutations that have been identified and characterised during the course of an evolution experiment. Comparable recombination events in experimental evolution studies with Escherichia coli have identified adaptive promoter capture events, resulting from tandem amplification events (Blount et al. 2012), or transposon insertions (Stoebel and Dorman 2010). Attempts using synthetic biology have identified the potential interacting components of chimeric proteins and cached the resulting functionalities within greater cellular interactions (Peisajovich et al. 2010). Peisajovich and colleagues identified the potential for artificial combinations of functional protein domains to provide novel interactions. In vivo rearrangements of both spatially determining and enzymatic domains were found in the mating signalling-network of yeast to increase phenotypic variation above gene- duplication controls.

The fadA-fwsR fusion is an example of a naturally generated modular assortment of spatial and enzymatic motifs capable of causing adaptive outcomes. Furthermore, the genetic modules that allow for spatial relocation (from the fadA gene) appear to be remarkably interchangeable with other membrane-locating genetic modules, as seen by alteration of the fwsR gene with the TMDs encoded by mwsR. This supports the theory that modules from separate loci can be easily interchanged, and the resulting interaction of regulatory and catalytic domains may rewire signalling pathways (Pawson and Nash 2003).

3.4.3.2 FadA-fwsR provides direct experimental evidence that translational fusion mutations cause adaptive change via protein relocalisation

The characterisation of the fadA-fwsR fusion provides direct evidence that a newly formed chimeric gene can confer adaptive benefits via the relocalisation of a protein to an alternate subcellular location. Similar fusions may also confer relocalisation to different tissues in multicellular species. Chimeric genes are able to explore the regulatory functionalities of either parent gene and thus expand, narrow or introduce novelty to the targeting of the resulting protein. In doing so, the relocalised protein is able to operate in a new context and is afforded the opportunity to interact with different proteins and thus generate new phenotypes. Most fusion mutations resulting in such

relocations will have deleterious consequences (Lobo et al. 2009; Rosnoblet et al. 2013), however in some cases these mutations will result in adaptive phenotypic change (Nurminsky et al. 1998; Byun-McKay and Geeta 2007).

The fadA-fwsR mutation provides a unique model to investigate the relocalisation of a protein following gene fusion. Such relocalisation, albeit to novel tissues, has been identified as resulting in a spectrum of chimeric genes in Drosophila melanogaster (Rogers and Hartl 2012). However, even in model organisms such as Drosophila, it is difficult to discern the relative effect that relocalisation has on the adaptive sweep of a given chimeric gene, compared to other effects arising from the chimeric event (such as novel interactions or the effects of altered expression). The amenable model in which fadA-fwsR evolves (a model bacterium) has afforded the opportunity to directly assess the phenotypic consequences of relocalisation, as well as exclude other competing hypotheses to explain the adaptive benefit of the chimeric mutation.

Furthermore, the FadA-FwsR fusion has provided evidence that adaptation caused by protein relocalisation can be caused by a single deletion event. This suggests more generally that neighbouring genes may provide ample evolutionary ‘fuel’ – given an appropriate in-frame deletion and redundancy of the constituent genes – in which to cause a change in the location of a protein.

3.4.3.3 Chimeric relocalisation as a novel mechanism behind new genes in the ‘neofunctionalisation model’

The relocalisation of chimeric proteins provides a mechanism of generating new proteins from essential genes following gene duplication. Duplication is considered a main source of new genes and protein functions (Kondrashov et al. 2002). The first theorised mechanism that accounted for how duplications could result in new protein functions was Ohno’s neofunctionalisation model. In this model, duplicated genes are functionally redundant and accumulate deleterious mutations until the extra copy is lost. However, in a small number of cases, mutations can confer a new function in the encoded protein of the extra gene, and selection acts to maintain the new gene (Ohno 1970). Despite the importance of this model, there are few mutational mechanisms – in particular generalizable mutational mechanisms that may apply to more than one locus – that explain how adaptive phenotypes are generated via duplication and divergence.

The relocalisation of proteins into distinct subcellular compartments has been hypothesised as a mechanism to facilitate the diversification of genes following duplication (Byun-McKay and Geeta 2007). This hypothesis has been supported by subsequent research. Approximately 37% of duplicated gene pairs in Saccharomyces cerevisiae encode proteins that are validated as locating to separate cellular compartments (Marques et al. 2008). Of human multi-gene families known to encode proteins predicted to locate to the mitochondria, approximately 64% of families contain a gene predicted to relocate to an alternative subcellular location (Wang et al. 2009). However, point mutations at the 5’ genic region (which can encode the N-terminal target peptide) is the sole mutational mechanism explicitly described that would facilitate relocalisation of duplicated proteins to specific targets (Byun-McKay and Geeta 2007).

FadA-fwsR has provided an insight into an alternative mechanism to nucleotide substitutions to target peptide sequences – one that would enable the relocation of proteins to an alternative subcellular target in a single mutational step. This mechanism is a fusion mutation that reassigns the location of a protein to the subcellular or tissue target of a neighbouring gene (either through the addition of TMDs or signal peptides). This potential alternative mechanism is given plausibility by the finding that approximately 60% of duplicate gene pairs in C. elegans feature structural heterogeneity (Katju and Lynch 2006). Of these heterogeneous pairs, chimeric fusions can be clearly identified in approximately 38% of heterogeneous duplicates. The prevalence of chimeric genes supports a notion that relocalisation driven by chimeric fusions following duplication events may be an important force driving the neofunctionalisation of duplicated genes.

3.4.4 Implications for secondary messenger research and future