Genotype-by-environment interactions for seminal fluid expression and sperm competitive ability

Sperm competition commonly occurs whenever females mate multiply, leading to variation in male paternity success. This can be due to variation in the various traits that might affect sperm competitive ability, which itself depends on both genetic and environmental factors, as well as on genotype-by-environment interactions (GEI). Seminal fluid is a major component of the male ejaculate that is often expected to mediate sperm competition, where different genotypes can differ in their seminal fluid expression as a response to different level of sperm competition (i.e., exhibit GEI). We therefore here focussed on testing for GEI in expression of two recently identified seminal fluid transcripts, suckless-1 and suckless-2, which potentially modulate sperm competitive ability in the simultaneously hermaphroditic flatworm Macrostomum lignano via their effects on manipulating post-mating partner behaviour and ultimately the fate of transferred ejaculates. In addition, we sought to test for GEI in sperm competitive ability, to investigate the relationship between natural variation in the expression of these seminal fluid transcripts generated through GEI and relative paternity success. To do so, we manipulated social group size, which has been shown to successfully alter sperm competition level in M. lignano, in a set of inbred lines (genotypes) and then measured both the expression level of suckless-1 and suckless-2 in focal worms together with their relative paternity success in a standardised sperm competition (P1 & P2) assay. We found GEI for the expression level of suckless-1 and suckless-2, as well as for sperm competitive ability. Moreover, we found a positive relation between the expression of suckless-1 and paternity success. This suggests that natural variation in the expression of this seminal fluid transcript indeed can influence sperm competition outcomes in M. lignano.

Assessing sperm competitive ability of genotypes 180 Since the aim was to estimate GEI effects on seminal fluid expression and sperm competitive 181 ability as a response to sperm competition level in their environment, we first raised genotypes 182 in different social group sizes, namely pairs (group of two worms) and octets (group of eight 183 worms). Following group size treatments, we assessed sperm competitive ability of these 184 genotypes (focals) by conducting double mating trials in which virgin standardized mating 185 partners (recipients) were mated sequentially with a focal worm followed by a (GFP-186 expressing) competitor, that is testing for the defensive sperm competitive ability of focals, P1; 187 or else a competitor followed by a focal, that is their offensive sperm competitive ability, P2. 188 To manipulate social group size, we initially collected ca. 2-3 days old juveniles from main 189 stock cultures of each genotype (ca. 150 per line -F0) and divided them into two glass Petri 190 dishes with ad libitum food to let them grow and lay eggs. Once they started to reproduce, we 191 collected their 2-3 day old offspring (F1) into one Petri dish (for randomization of juveniles 192 collected from two Petri dishes of F0) and then we randomly distributed these F1 offspring into 193 24-well tissue culture plates, including 1 ml of ASW and ad libitum food in each well, to form groups of pairs and octets. Social groups were distributed on plates in a way that balanced for any potential plate position and genotype effects. In total, we formed 40 replicates (each 196 comprising the eight different genotype/group size combinations) for the P1 assay and 41 197 replicates for the P2 assay). In order to avoid the potentially confounding effect of mismatched  Note that all the required offspring to form the social groups of genotypes and GFP competitors 203 were collected within three days to minimise age differences. Thereafter, these offspring were 204 raised for up to eight weeks in their given groups by transferring them to freshly prepared 24-205 well plates every week to prevent accumulation of their newly-hatched offspring.

206
In parallel, DV1 offspring needed for each double mating trialto be used as unmated France per one ml 32‰ ASW) and 7 µl 32‰ ASW with food. Following the colouring step 215 and before the mating trial itself, worms were briefly transferred to fresh 24-well plates without 216 colour solution and food (including only 1ml of 32‰ ASW) for a few minutes for residual 217 colouring to be washed out. In this way, worms were coloured slightly blue, which has no effect on worms' maintenance, fecundity and mating behaviour (Marie-Orleach et al., 2013), but 219 which allows us to easily distinguish them from the focal worm under a stereomicroscope.

220
The double mating trials were initiated approximately eight weeks after the social group size 221 manipulation of the focal and GFP competitorsand the isolation of recipientsbegan. In 222 order to avoid pseudo-replication for both the focal and GFP competitors, one individual worm 223 was picked randomly from each genotype/group size combinations and immediately used for 224 the sperm competition assay for one assay only (and thereafter a subset of these was used for 225 the gene expression measurementssee below). For logistical reasons, we divided the sperm 226 competition assays into blocks performed over 13 days, with identical procedures on each day, 227 and we ensured that recipients, genotypes and competitors used on each day were similarly 228 aged (ranging from 55-62 days old) and randomly assigned.

229
Each day, we first paired a (unmated DV1) recipient (recipient one) for the first mating for two 230 hours with a given individual (either a focal worm from one of the genotype/group size 231 combinations or its GFP competitor, depending of the type of trial). At the end of this two hour 232 period, the given individual was transferred to be immediately paired with a second recipient 233 (recipient two) for a further two hours, while the recipient one was paired with the second 234 individual to mate (either the GFP competitor or a focal worm from one of the genotype/group 235 size combinations, opposite to the first period). Following this, we paired the second individual 236 to mate with recipient two after removing the first individual. The aim of pairing each genotype 237 and its competitor with two recipients sequentially was simply to increase the total number of 238 offspring and thereby the precision of our paternity estimates, considering that individual 239 worms lay relatively few eggs. Immediately after the total four-hour mating period, the focal 240 worms were individually transferred to 1 ml tubes including 25 µl RNALater®, while GFP 241 competitors were paired one-by-one with a separate group of virgin worms (DV1) in 24-well after ca. seven-eight days, we paired it with at least three others to disentangle whether the GFP 244 worm or its partner was the cause of the infertility. We later excluded data where GFP 245 competitors did not achieve any reproductive success when paired with their additional 246 partners.

247
Paternity assessment was done by counting the offspring of recipient one and two, which were 248 isolated after the trials to let them lay eggs, by sorting GFP expressing or non-GFP expressing for the P2 assay were successfully measured for paternity success. The decrease in targeted 254 sample size for P1 and P2 mating trials was due to lost worms during social group size treatment 255 or due to excluded recipients if both (recipient one and two) did not produce offspring. In total, 256 paternities for 2416 and 2291 offspring were assigned in the P1 and P2 assays, respectively. 258 In order to measure the expression of suckless-1 and suckless-2, we picked at random eight 259 samples of each genotype/social group size combination from the worms used to assess P1. We  offspring to total offspring in the P1 assay), for recipient one and recipient two separately. The 299 rationale of fitting separate models for recipient one and two was, first, to be able to evaluate 300 gene expression effects on each recipient separately considering that transcript expression 301 measurements were performed only following matings with recipient two, and second, because 302 we had found an effect of the order of mating with the two recipients on P1 (see Results).

304
As predicted based on our previous results, seminal fluid transcript expression differs 305 significantly between genotypes for both suckless-1 and suckless-2 and both exhibit significant 306 GEI (Table 1). In addition, overall relative expression levels (-∆Ct) were similar between social 307 groups (-3.79 ± 2.60 in pairs and -3.13 ± 2.65 in octets for suckless-1 and 0.31 ± 1.83 in pairs 308 and 0.73 ± 1.40 in octets for suckless-2). Although, the overall mean relative expression level 309 of the two transcripts were apparently quite different (-3.46 for suckless-1 and 0.52 for suckless-310 2), the reaction norms showing differential expression pattern of transcripts across social groups 311 were strikingly similar for all genotypes (Fig. 1), suggesting GEI for expression of these 312 transcripts manifests in a quite coordinated manner.

313
For the sperm competition assay, the initial model comparisons including mating order with 314 recipient one (or two) as a discrete factor, plus genotype, social group size and their two-way 315 interaction, showed that paternity success (scored as P1 and P2) does not exhibit GEI (Table 2). two (Fig. 2), and a significant main effect of mating order shown in this model at least for P2 318 (and a similar, marginally non-significant trend for P1) (Table S1), we (retrospectively) 319 preferred to analyse these data by instead fitting a model including a three-way interaction 320 between genotype, social group size and mating order with the recipient. This model indeed 321 shows that there is a significant three-way interaction, meaning genotypes differ in their relative 322 paternity success depending on both the social group size and mating order ( Table 2 & Table   323 S2). Note that when we fitted models for GEI for each recipient separately, GEI was highly 324 significant for both recipients in the P1 assay (P = 0.002 for recipient one, P = 0.005 for recipient 325 two) and for the first recipient in the P2 assay (P < 0.001 for recipient one, P = 0.17 for recipient 326 two) (Table S3).  (Table S4). We 333 also calculated the correlation between the expression of the two genes, which was marginally 334 significant but weak (r = 0.27, P = 0.037). Based on this, we then dropped suckless-2 from the 335 models and evaluated and visualized only the effect of suckless-1 (Figure 3) for each recipient 336 (GLM for recipient one: z = 3.14, P < 0.001, recipient two: z = 3.96, P < 0.001). To be precise,  First of all, we supported our previous results regarding significant genotypic variation for the 364 relative expression of suckless-1 and suckless-2, beyond which we extended our findings by demonstrating also significant GEI for these transcripts. We already had some evidence of GEI      and three-way interaction (genotype-by-group size-by-mating order with the recipients). The