CHAPTER 3: Results
3.3. Higher transposon expression is weakly correlated to higher copy
we confirmed the results of the bioinformatics analysis, with higher copy in Poly Sb males compared to both Mono SB and Poly SB (P <0.05, ANOVA and post-hoc Tukey test, Supplementary Figure 4B).
Together these results show the reliability of our CNV analysis: 14 out of 17 genes (82%) showed significant copy number difference through qPCR or ddPCR.
3.3. Higher transposon expression is weakly correlated to
higher copy number
Previous studies have found TEs with differential expression (DE) between SB and Sb-carrying individuals (Nipitwattanaphon et al., 2013) and we expected these TEs to show corresponding copy number differences. To this end, we assessed the CNV status for 13 TEs with known DE between SB and Sb-carrying individuals (Figure 3.3).
Surprisingly, eleven of the 13 TEs showed no copy number variation among genotypes in both DNA qPCR and CNV analysis (nine showed no copy number variation among genotypes, and two showed marginally significant copy number variation). Only a DNA Harbinger and a piggyBac element were upregulated in both SB/Sb queens and workers, and both showed higher copy number in Sb. These results revealed that higher TE expression does not necessarily correspond to greater copy number.
These TEs were expected to show corresponding copy number differences, however, We also examined the converse case: whether TEs with greater copy number in Sb have correspondingly higher expression in SB/Sb individuals (Figure 3.4). We correlated CNV TEs with gene expression and found a positive linear relationship for DNA TEs. When excluding one outlier datapoint, only the DE analysis (but not the absolute expression analyses) showed a linear correlation with copy number (p = 0.01, R2 = 0.342, Figure 3.4a). In all cases, low R2 values showed that the correlation between copy number and expression is not straightforward. We found no positive correlation when considering LTR and LINE elements. The observed weak correlation between the TE copy number and their expression suggests that more factors (e.g., age of TE in genome) need to be considered to explain the difference in TE copy number between SB and Sb.
3.4. Enzymes involved in cuticular hydrocarbon (CHC) synthesis are highly duplicated in Sb and over-expressed in SB/Sb queens
We tested if “highly represented” gene categories (≥3 CNV genes) were over-represented among CNV genes. Eleven non-TE gene functions were highly represented, and of these, five were over-represented: genes encoding FASs, FPPSs, histones, toll-like receptors, and OBPs (Table 3.1, P <0.05, Fisher’s exact test with Bonferroni correction).
FASs and two of the highly represented gene families (cytochrome P450s and reductases) are possibly involved in the synthesis of CHCs (Figure 3).
Fatty acid synthases are responsible for incorporation of malonyl-CoA or methylmalonyl-CoA into an acetyl-CoA, thus forming a medium-chain fatty acyl-CoA. Desaturases and elongases add double bonds and elongate the carbon chain, thus regulating number and position of double bonds, and chain length, respectively. The resulting long-chain acyl-CoA is reduced to aldehyde by a reductase enzyme. Here we report the total number of reductases with copy number variation between SB and Sb, however, some of them may be responsible for reduction of other molecules than long-chain acyl-CoA. Specifically, three reductases duplicated in Sb were fatty acyl-CoA reductases, while the others were ambiguous. Finally, the hydrocarbon is produced from the aldehyde by a decarboxylase (which is thought to be a cytochrome P450 enzyme). As for reductases, we report the number of all CNV cytochrome P450.
However, only the cytochrome P450 CYP4 family is responsible for fatty acid metabolism (Kirischian & Wilson, 2012). In particular, an enzyme of the CYP4G family has been shown to produce CHCs in D. melanogaster (Qiu et al., 2012), but among our CNV cytochrome P450 genes we only found enzymes from the families CYP4A, CYP4C, and CYP6.
Because fire ant queen acceptance or rejection by workers is based on an odor cue (Eliyahu, Ross, Haight, Keller, & Liebig, 2011; Trible & Ross, 2015), one or more of these duplicated genes could plausibly affect queen hydrocarbon profiles and consequently worker behavior, contributing to the social form differentiation.
Given higher copy in Sb of these three gene families (25 genes) associated with CHCs, we next looked for additional genes putatively involved in the synthesis of CHCs with greater copy number in Sb and found two; an acyl–coenzyme A desaturase Δ11 and an elongase of very long chain fatty acid (Figure 2.5), for a total of 27. These two enzymes control the presence of double bonds (i.e., saturated or unsaturated hydrocarbons) and the length of the CHCs, respectively (Fang et al., 2009;
Finck, Berdan, Mayer, Ronacher, & Geiselhardt, 2016; Helmkampf et al., 2015).
We compared the log2 ratio of Sb:SB copy number with DE between SB/Sb and SB/SB virgin queens for the 27 CHC genes with greater copy number in Sb (Figure 3.6a). Only 20 passed the low expression threshold, and of these 17 (63%) were significantly over-expressed in SB/Sb (p <
0.05, FDR) and only one in SB/SB (p = 0.01, FDR). Taking all CNV genes into consideration, only 22.6% of non-TEs with greater copy number in Sb were over-expressed in SB/Sb queens. There was no positive correlation between copy number and DE (p = 0.93, F-test, linear regression).
A similar analysis using absolute expression revealed no positive correlation between copy number and absolute gene expression in SB/Sb virgin queens (p = 0.07, F-test, linear regression) (Figure 3.6b) and SB/Sb reproductive queens (p = 0.17, F-test, linear regression) (Figure 3.6c). All
having greater expression than the average of a set of conserved hymenoptera genes (BUSCO, Benchmarking Universal Single-Copy Orthologs, n =1,998). The desaturase gene was the most highly expressed gene in both SB/Sb virgin and reproductive queens.