Results
First, we examined the expression of each separate SFP gene, and
detected that the expression of LyAcp8b significantly increased
48h after mating (Fig. 1, Table 1). We also found that LyAcp5 andLyAcp8a expression altered significantly after mating, although
post-hoc testing did not show any significant difference between
specific time points (Fig. 1, Table 1). For LyAcp5 , this seems
due to the difference in expression between 48 h and 196 h, and forLyAcp8a (very similar to LyAcp8b) expression increased 48 h after
mating. Two samples showed consistently high expression inLyAcp5 , LyAcp8a and LyAcp8b (#92, #110), but not
in the other genes and we could not find any technical or biological
features explaining this pattern (e.g., RNA extraction date, body size).
In LyAcp8a and LyAcp8b , we detected significant differences
between experimental blocks, but not interaction with Hours after
mating. In contrast, the three remaining SFP genes did not show any
significant change in expression level throughout our monitoring,
suggesting that the production of these SFPs did not increase after
mating (Fig. 1, Table 1). We also like to note that the expression at
196 h after mating is not always low, compared to 3 h after mating (Fig.
1), while by that time the prostate gland is expected to be fully
replenished.
Next, we inspected the overall change in expression across all SFP
genes. To do so, we conducted a PCA to create representative variables
for overall SFP gene expression. PC1 explained 50.0% of the total
variance, and this variable seems to correspond with Hours after mating,
although this is not the case after FDR correction (Fig. S2, Table 2,
Table S2). In contrast, PC2 explained 26.5% of the total variance, and
seemingly explained the difference between SFP genes, again after FDR
correction statistical significance disappeared (Fig. S2, Table 2, S2).
The important, additional insight from PCA is that the expression of SFP
genes after mating differed between separate SFP genes, which is
visualized by the directions of PC loadings (Fig. 2, Table 2).