3.7 Mechanisms of opsin expansion
We determined gene conversion, which could be used to explain opsin gene expansion. Gene conversion is any process that causes a segment of DNA to be copied onto another segment of DNA, and plays an important role in evolution (Guttman & Dykhuizen 1994). We found that there were two gene conversions of the SWS2 gene in C. undulatus , SWS2 a vs. SWS2 d (314 bp, P = 0.016), and SWS2 b vs.SWS2 c (79 bp, P = 0.044), one gene conversion ofLWS1 gene, LWS1 a vs. LWS1 d (70 bp, P = 0.047). Two gene conversions occurred in the Rh2 gene,Rh2 a vs. Rh2 b (96 bp, P = 0.0006) and Rh2 d vs. Rh2 e (197 bp, P = 0.0001), suggesting opsin expansion by gene conversion. We also tested genes with gene conversions by sliding windows using the rate of non-synonymous to synonymous substitutions (ꞷ). The ꞷ value was much less than 1.0 for Rh2 a vs. Rh2 b, and Rh2 d vs. Rh2 e (Fig. 8A), SWS2 a vs. SWS2 d (Fig. S6A), SWS2 b vs. SWS2 c (Fig. S6B), and LWS1 a vs. LWS1 d (Fig. S6C), indicating that the opsin gene underwent purifying selection (ꞷ < 1.0). To uncover positive selection sites in the opsin gene, we used PAML to find five positive selection codon sites with a probability greater than 95% in the Rh2 gene (Fig. 8A), but no site was found in the LWS1or SWS2 genes (Table 3). Our results suggest that opsin gene conversions occurred during post-speciation of C. undulatus at the evolutionary level.
We then determined whether the sudden increase in opsin copies inC. undulatus is the result of an increased rate of local gene expansion events rather than entire genome duplication. In this case, duplicates should share a flanking sequence (Lagman et al. 2013), we determined the flank sequence of the local gene where gene conversion occurred in SWS2 , LWS1 , and Rh2 genes. The identifications of the flank sequences were very low, and no more than 48% (Table 4). Our results demonstrated that, after the 3R (Lagmanet al. 2013), gene conversions have contributed to the number of opsin genes. Besides, the retrotransposed duplicates also gave rise to opsin gene copies. Based on our results, it is difficult to interpret opsin gene expansion based on a single factor. It has been reported that the genomic environment, such as genomic architecture, can affect opsin gene conversion (Sandkam et al. 2017).
The activity of transposons can shape genomic architecture (Mat Razaliet al. 2019), and the C. undulatus genome showed a high content of transposons (39.88% of the entire genome); therefore, we determined the transposon content of a 100 kb window along chromosome 3, owing to opsin gene expansion mainly occurring in this chromosome. We found that the number of transposons in the LWS1 -SWS2window (111) was lower than that in the adjacent windows (up 131 and down 130) along the negative strand (Fig. 8B), and the number was significantly different between them (P = 0.04). For theRh2 gene, the number of transposons in the Rh2 windows (82) was significantly lower than that in the adjacent windows (up 102 and down 111) (Fig. 8B), with a significant difference (P = 0.02 for up and 0.002 for down). The average means of transposons per window is 113 in the negative strand and 116 in the positive strand. The average means of transposons per gene window was 107 in the negative strand and 111 in the positive strand. Transposons play important roles in genome plasticity to adaptive behavior in evolution (Liu et al. 2020; Robert et al. 2008). It is reasonable to believe that transposons may be ascribed to opsin expansion.