Nasonia vitripennis lines and rearing

Nasonia vitripennis is a globally distributed parasitoid of blowfly pupae used broadly in genetics, behavioral, and ecological research for decades [35]. It has haplodiploid reproduction; unmated females produce exclusively male offspring from haploid eggs, and mated females produce daughters from fertilized eggs in addition to a proportion of sons. All lines were maintained under conditions with 25°, 16:8 LD cycle, ~55% RH, with two-week generation cycles on Calliphora sp. pupae hosts. The Whiting polyploid line (WPL) was acquired from the J.H. Werren laboratory (University of Rochester, Rochester, New York, USA) and was maintained in our laboratory for 20 years. This strain is cured of Wolbachia endosymbiotic bacteria. It carries two complementary recessive eye-color mutations, scarlet (st) and oyster (oy). Virgin triploid females produce red-eyed haploid and diploid purple (wildtype)-eyed males. The diploid purple-eyed males are crossed to females of the red eye mutant line scarlet, which are used to recoup triploid females for another breeding cycle (see detailed breeding cycle scheme in [36, 39]). The genetically variable HVRx strain was acquired from B.A. Pannebakker laboratory (University of Wageningen, Wageningen, The Netherlands). HVRx was founded by mixing several Dutch field lines and genetic variation is maintained by re-hosting four mass culture tubes every generation and mixing hosts several days after oviposition [43]. It has not been cured of Wolbachia. This line was used to generate the neopolyploid tKDL line. Untreated mated females from this line were also used to generate control (no descent from injection) individuals for assays. All crosses were done intraline to their WPL or HVRx background.

Generation of the neopolyploid tra KD line (tKDL)

The tra KD line (tKDL) was created following the ds tra RNA synthesis and female injection protocol of [39]. Maternal knockdown of tra results in sex reversal of diploid eggs, i.e. diploid males rather than females. Polyploids (diploidized males) and non-polyploid males (from unfertilized haploid eggs) thus occur in the same generation. To account for descent from the injected females as an independent factor with possible effect on phenotype, assays were conducted considering control non-polyploids from the untreated HVRx population, non-polyploids of the tKDL, and (neo)polyploids of the tKDL. To generate the tKDL line, HVRx virgin females (N = 200) were injected as white stage pupae with 4μg/μl ds tra mixed with red dye (to confirm injection) with a FemtoJet microinjector (the F0 generation). They were mated with control haploid males as adults. They were given three hosts, and produced tKDL diploid males from fertilized eggs and tKDL haploid males from unfertilized eggs (the F1 generation). When diploid males, which produce diploid sperm, are mated with diploid females, their female offspring will be triploid. Every generation was outcrossed to the untreated HVRx population to maintain genetic variability (~20 founders or foundresses and five offspring each generation). The fate of the tKDL neopolyploid line was followed to generation F5. S1 Fig indicates individuals used, the crosses to obtain them, and which assays were conducted for each generation. As there were no means to visually distinguish diploid and haploid males in the tKDL they were typed for ploidy using a process combining flow cytometry of heads and fecundity assessment of daughters (diploid daughters will have higher fecundity, reflecting a haploid father, and triploid daughters will have lower fecundity, reflecting a diploid males). The full ploidy-typing process is described in S2 Fig.

Male mate competitions

Male mate competition experiments were conducted between pairs of tKDL diploid males and haploid males. One experiment competed the diploid with an unrelated haploid male from the untreated HVRx strain. Another experiment competed diploid and haploid brothers from the same tKDL mother. tKDL diploid males were also compared to WPL diploid males. Two types of competition experiments were conducted, single competitions were for one virgin female, and multiple competitions were for 10 virgin females. These females were from the HVRx population, which as a genetically variable background is expected to exhibit strong monandry [44–46] (females mate once and all their female offspring descend from a single father). Male competition experiments were conducted by placing males and females in a single tube for 24 hours. Females were then removed and given three hosts. For the tKDL haploid and tKDL diploid competitions, the possibility of using two haploids or two diploids could not be guaranteed a priori. Therefore, post-competition, males were individually mated to single HVRx females, to generate daughters for the daughter-typing component of the ploidy-typing process, and subsequently stored at -20°C for the flow cytometry step (S2 Fig). For all competitions, the male mate of female was determined by typing the ploidy of daughters (diploid daughters indicated a haploid father, triploid females a diploid father) (described in S2 Fig). For the tKDL diploid and WPL diploid competitions, daughters were instead hosted as virgins. Those that had male offspring with red eyes indicated that their mother mated with the WPL male, and those that did not have male offspring with red eyes had mothers that mated with the tKDL male. Females from competitions that produced all male offspring were presumed to have not mated and were excluded from analyses, as were the tKDL competitions that were found to have used two males of the same ploidy.

Female fecundity and offspring ploidy distribution

To bypass any knockdown effects and to increase sample size, we examined tKDL F4 generation instead of the F2 generation (S2 Data). We scored progeny size for virgin and mated diploid control females, F4 tKDL diploid and triploid females, and WPL triploid females to assess the influence of polyploid background on fecundity. For the mated females, we also assessed progeny sex ratio. In the case of the triploid females, progeny ploidy distribution was also scored, as eggs of triploid females can be haploid or diploid. Females of the mated assays were given 24 hours to mate with a control haploid HVRx male. All females were then given three hosts. All progeny were counted for each female 14 days later, and in the case of the mated females, sexed. For the virgin and mated triploid F4 tKDL triploid females, entire families were scored for ploidy using flow cytometry (excluding larvae, which cannot be typed for ploidy with flow cytometry) (S1 Fig).

Wing cells and cell reduction mechanisms

To investigate the possible existence of cell reduction mechanisms in Nasonia polyploids (and as a case study for cell reduction in invertebrate and insect polyploids), we examined wing cells for both sexes across all ploidy levels for both the WPL and the tKDL. We counted setae (wing hairs), as each hair corresponds to single cell (following [47]) (S3 Data). The adult right forewing was removed for each individual and mounted on a glass slide using clear nail polish. Using the Motic Images Plus 2.0ML program, wings were imaged with a Moticam 2000 camera attached to a Car Zeiss SV6 microscope. In Photoshop CS6 (64 bit) a 0.25 mm² square subsection below the cross-vein was consistently sampled for each specimen (Fig 2), and setae counted with the count tool. For the WPL, females of the scarlet line used for maintaining the line represented control diploid females. For the tKDL diploid male measurement, ≤20% of individuals might have actually been haploid (based on typical male production of mated N. vitripennis females [48]) because haploids could not actually be sorted from this class. However, significant differences among groups could be attributed to the majority presence of diploids. For each group, N = 5 randomly selected individuals were also scored for setae count across the whole right forewing using the count tool in Photoshop.

Gene dosage of housekeeping genes Ak3 and ef1α

To assess the effect of ploidy and sex on gene expression, we quantified the absolute expression of two housekeeping genes Adenylate kinase 3 (Ak3) and elongation factor 1 alpha (ef1α) for all backgrounds (N = 5 each). We followed the protocol of Dalla Benetta et al., [49] with the following modifications. In brief, RNA was extracted from heads and abdomens separately. Each wasp’s head was individually placed into a 1.5 ml Eppendorf tube. The same was done for abdomens. Thoraxes were discarded due to likely endopolyploidy [50]. Samples were immediately frozen at -80° C, extracted using TriZol (Invitrogen, Carlsbad, CA, USA), and cDNA conversion performed according to manufacturer’s instructions. Five technical replicates for each sample type were performed to control for pipetting error. qPCRs were run for Ak3 and ef1α (primers in S1 Table) on the Applied Biosystems 7300 system. Reactions were first performed at a 5x dilution (Fig 2, S1 Data) in the event of large amount of RNAi producing asymptotic readings instead of accurately measuring the absolute expression levels. Analyses were then also run for 50x dilutions to test for consistency (S2 Data). The qPCR reactions were run with 3 min of activation phase at 95°C, followed by 35 cycles of: 15s at 95°C, 30 s at 56°C, and 30 s at 72°C. Note that in typical qPCR reactions these two genes are used as reference for RNA quantity, but in this assay we use the amplification amount as a direct measure for gene expression level as function of ploidy and sex.

Statistical analyses

All data are reported as means with standard deviation. Statistics were done in SPSS Statistics 25 [51] and R [52]. No datasets met normality or homogeneity of variance (Shapiro-Wilks Brown-Forsythe tests), so Kruskal-Wallis rank sum tests were used to compare groups. Post-hoc Dunn’s multiple comparison tests with a Bonferroni correction were performed to test for differences between paired groups. Binomial tests were used to assess mate competitions with null hypotheses that competing male types had equal mate competition ability in every competition. An additional general linear mixed model was used to determine how likely a female was to mate with each male type in each competition type (S1 Table). The qPCR results were analyzed using the LinReg PCR software, with absolute expression normalized to the program’s calculation for N₀ concentration [53, 54]. A general linear model was also used to analyze these results with extraction set, and individual set as random effects; and ploidy, sex, and background (control, tKDL, or WPL) set as fixed effects. A Satterthwaite approximation and estimation of robust variance was used to account for low sample size and non-normal distribution.

Male mate competition

Like most parasitoid wasps, Nasonia have female mate choice and females typically only mate once [44, 45]. In other hymenopteran studies [55], including that on WPL [36], diploid male mate competition ability is equal to haploids. To compare our lines to this existing trend in polyploid mate competition research, we conducted competitions between diploid males of the WPL and tKDL against haploid males of an untreated control line, and haploid-diploid brother pairs of the WPL and tKDL. We also competed diploid males of WPL against diploid males of tKDL. We did this for competitions for single females and multiple (N = 10) females. In the single female mate competition trials the successful male was scored as the one that sired offspring, and in the multiple female trials, the successful was the male that sired offspring with more females overall. Competitions were performed a priori without knowing the ploidy of the tkDL males, which was checked after using flow cytometry (S1 Fig). The tKDL diploid males were consistently outcompeted by haploids of both the control background of the untreated HVRx population, and their own haploid brothers from the same tKDL background (Table 1). In competitions for the single female haploid males, the tKDL diploid male only won 9 out of 33 trials (P = 0.014) when competing with a control haploid male, and 7 out of 21 competitions against a haploid tKDL brother (P = 0.095, binomial test). For multiple females, the tKDL diploid male only won 1 out of 22 competitions against a control haploid male (P<0.001, binomial test), and 6 out of 28 competitions against a haploid tKDL brother (P = 0.095, binomial test).

Download:

• PNG
  larger image
• TIFF
  original image

Table 1. Male mate competitions.

Males competition pairs were given either a single or multiple (N = 10) virgin female(s), and winners were determined through offspring scoring of daughter fecundity or flow cytometry. The winning male type (more trials won) is marked with an asterisk (*). P-values are for binomial tests.

https://doi.org/10.1371/journal.pone.0288278.t001

As these experiments were done in the F1 generation immediately following RNAi knockdown, we considered that RNAi carry over effects could be influencing results. So eliminate this possibility and to test for heritable effect, we repeated these experiments in the F5 generation. Flow cytometry indicated that many trials had to be discarded because a haploid tKDL male rather than a necessary diploid tKDL male was used. However, even for the resultant F5 smaller sample size, a pattern of tKDL diploid male inferiority against haploids held. For competitions for a single female, the F5 tKDL male only won one out of four single competitions against a control haploid male (P = 0.313, binomial test), and four out of 14 single competitions against a F5 tKDL haploid brother (P = 0.180, binomial test). In competitions for multiple females, the F5 diploid male only won two out of ten competitions against a control haploid male (P = 0.001), and one out of ten competitions with a F5 haploid tKDL brother (P = 0.022). The low preference of females for the tKDL diploid male is reflected in a general linear mixed model analysis (GLMM), with females being 2.3–3.3 times less likely to mate with diploid males versus haploid males in the F1 generation, and 2.4–22.6 times less likely to mate with diploid males than haploid males in the F5 generation (Table 2).

Download:

• PNG
  larger image
• TIFF
  original image

Table 2. General linearized mixed models (GLMM) to test whether females were more likely to mate with one type of male in mate competition experiments.

Trial number was set as a random effect. A multinomial logistic regression with a generalized logit link and a Satterthwaite approximation and estimation of robust variance were used to account for low sample size and non-normal distributions. The exponential coefficient gives the number of times females are more likely to mate with the superior mate competitor type (the haploid in every competition, or tKDL in the diploid-diploid competitions). Significant results are indicated with an asterisk (*).

https://doi.org/10.1371/journal.pone.0288278.t002

To specifically test for differences between polyploidization background (neopolyploid and generated tKDL versus long-existing and spontaneous WPL), we competed F1 tKDL diploid males against WPL diploid males (Table 1). Despite the tKDL diploid males performing poorly against haploids, whereas WPL diploid males had equal mate competition ability to haploids in a previous study [36], the F1 tKDL diploid males outperformed the WPL diploid males. In competitions for a single female, the F1 tKDL diploid male won 11 out of fifteen competitions against the WPL diploid male (P = 0.118, binomial test), and in competitions for multiple females, the F1 tKDL male won nine out of 13 multiple competitions (P = 0.267, binomial test). Although sample sizes were small and so not significant, tKDL males were 4.5–6.6 times more likely to mate with females than WPL diploids (Table 1). In addition, fewer females mated in the tKDL diploid-WPL diploid multiple mate competitions (on average, 4.3 females) compared to the tKDL haploid-tKDL diploid competitions (7.70–9.25 females) (P<0.05, Kruskal-Wallis test and Dunn’s test). In short, tKDL diploid males appear to be inferior competitors against haploids, but are superior competitors to WPL diploid males, and in general diploid males are less able to induce mating acceptance from females than haploids.

Female fecundity

The high fecundity of N. vitripennis polyploid males is known from the WPL [37], with WPL diploid males having the same number of offspring as haploids with a single mate [36]. We assessed the progeny size of virgin and mated WPL triploid females, and tKDL diploid and triploid females, under standard culturing conditions. We further measured the progeny sex ratio of mated females, as females vary in the number of eggs they fertilize based on various environmental and genetic factors [45–49]. We also assessed the proportion of progeny that were polyploid for virgin and mated females for both WPL and tKDL. We did this to investigate the degree by which polyploidy impacts N. vitripennis fecundity, which is a major challenge to the establishment of polyploid lineages. These data are summarized in Table 3.

Download:

• PNG
  larger image
• TIFF
  original image

Table 3. Female fertility.

Progeny size is reported for all females. Progeny sex ratio is reported for mated females. Progeny polyploid proportion is reported for triploid tKDL females. All data are means ± standard deviation. NA indicates that data was not collected. Relevant significant pairwise differences are discussed in the text.

https://doi.org/10.1371/journal.pone.0288278.t003

For WPL triploids, the mean progeny size was low, 3.64 ± 3.32 for virgins (N = 49) and 2.34 ± 2.43 for mated females (N = 50). This conforms to expectations of high aneuploidy. In contrast, diploid scarlet females used to maintain the line produce 60–90 offspring. The mean progeny size of tKDL triploids was higher than WPL triploids for both tKDL triploid virgins (12.40 ± 4.94, N = 50) and mated females (20.77 ± 8.49, N = 47). As expected, tKDL triploid females had lower virgin progeny sizes than control diploids (97.73 ± 19.59, N = 45) and tKDL diploids (69.18± 18.00, N = 49). The tKDL triploid females also had lower mated progeny sizes than the control diploids (59.34 ± 20.70, N = 47) and tKDL diploids (74.78 ± 18.24, N = 49). All WPL triploid versus tKDL triploid differences progeny size differences were significant, as were all tKDL triploid differences with the control and tKDL diploids (P<0.001 Kruskal-Wallis test, P<0.05 Dunn’s test). These results reflect a tKDL triploid progeny size 3.4 times (virgin) and 8.9 times (mated) larger than that of WPL triploids.

A typical Nasonia progeny sex ratio (male offspring/total offspring) of about 25% male [56] was consistent among the mated tKDL triploids (0.214 ± 0.174, N = 47), control diploids (0.233 ± 0.178, N = 47), and the tKDL diploids (0.264 ± 0.217, N = 49) (P = 0.448, Kruskal-Wallis test). In contrast, mated WPL triploids (N = 50) skew heavily towards male offspring, with a mean progeny sex ratio of 0.89 ± 0.26, which is significantly higher than the tKDL triploid female (P<0.001, Dunn’s test). This indicates that the tKDL triploids not only produce more viable euploid eggs than the WPL, but also that these eggs are more amenable to fertilization and female production.

Whole broods of virgin (N = 31) and mated tKDL (N = 47) triploid females were typed for ploidy using flow cytometry to determine if this line has higher polyploid production than WPL. The mean polyploid percentage (polyploid offspring /total offspring) was 31.9 ± 22.4%, and 36.6 ± 1.91% for virgin and mated females respectively. This is not different from the 25% polyploid proportion reported for WPL [37, 57], but lower than the 50% proportion expected from random segregation.

Cell size and reduction

Setae count for subsections and whole wings of WPL and tKDL indicated a difference in cell number and size for polyploid backgrounds (Table 4). For WPL, for both the sampled subsections and whole wings, there was a clear progression of fewer, larger cells for higher level ploidy, and females having fewer and larger cells than males (Fig 1B). Haploid WPL males (subsection: 189.19 ± 38.22, N = 47; whole wing: 506.4 ± 67.12, N = 5) averaged significantly more cells than diploid WPL males (subsection: 157.20 ± 38.22, N = 25; whole wing: 394.4 ± 45.55, N = 5). WPL Females had fewer and larger cells than males for subsections, but more for whole wings, as female wings are larger. Between WPL females, diploid scarlet (WPL) females had more cells (subsection: 46.71 ± 5.98; N = 34; whole wing; 902.2 ± 65.44, N = 5) than triploid WPL females (subsection: 35.63 ± 5.28, N = 32; whole wing: 709.6 ± 59.6, N = 5).

Download:

• PNG
  larger image
• TIFF
  original image

Fig 1. Cell reduction exists in the long-established Whiting polyploid line (WPL) but not the neopolyploid transformer knockdown line (tKDL).

(A) 0.25 mm² subsamples of the right forewing were taken using the single cross vein as a landmark to ensure consistency (specimen pictured is a 3N WPL female). As each seta (wing hair) corresponds to a single cell, greater distance between setae indicates larger cells. (B) In the Whiting polyploid line, cells become progressively larger with ploidy level, and as wings are similarly sized within sex for polyploids and non-polyploids, this suggests cell number reduction mechanisms at work (C) In contrast, in the tKDL knockdown line, cell size remains the same across sex and ploidy levels, suggesting that cell number reduction mechanisms only evolve over time, or that tra has a role in cell and body size regulation that has been lost in these specimens. Dotted lines for (B) and (C) indicate separation of ploidy levels. For quantitative values, see Table 4.

https://doi.org/10.1371/journal.pone.0288278.g001

Download:

• PNG
  larger image
• TIFF
  original image

Table 4. Subsection and whole wing setae count.

Pairwise comparisons are Mann-Whitney U tests, multiple comparisons are Kruskel-Wallis tests. An asterisk (*) indicates the group that has significantly more cells.

https://doi.org/10.1371/journal.pone.0288278.t004

In contrast, there were no clear progression of cell number or size differences for ploidy levels or sex for tKDL (Fig 1C). For the subsection sampling, haploid control males (69.73 ± 16.83, N = 87) had significantly fewer cells than the tKDL diploid (95.93 ± 15.39 N = 79), the opposite of what would be expected in the case of polyploid cell reduction. However, there was no significant difference between haploid control (580.4 ± 29.29, N = 5) males and tKDL diploid males (485.8 ± 77.10) for whole wing setae count. Again, females had fewer subsection cells males and more cells overall for larger whole wings. However, there were no significant differences among the control diploid, tKDL diploid, and triploid tKDL females for subsections (control diploid females: 54.38 ± 5.72, N = 20; tKDL diploid females: 49.25 ± 5.41, N = 21; tKDL triploid females: 50.82 ± 6.33, N = 55) or whole wings (control diploid females: 1417.4 ± 85.69, N = 5; tKDL diploid females: 1361.2 ± 97.22, N = 5; tKDL triploid females: 1257.8 ± 72.46, N = 5). (P-values for all comparisons are reported in Table 4.) It should be noted there is large overlap in values among the groups (as is reflected by the large standard deviations), indicating that even in WPL there is no clear segregation in setae count between polyploids and non-polyploids. These data suggest cell size increase and cell number reduction in one polyploid background (WPL) but not for another (tKDL).

Gene expression levels

Generally, the absolute expression of the housekeeping genes Ak3 and ef1α for a 5x dilution of cDNA (Fig 2) was lower in males than females. This applied to both the head and abdomen, and for both the WPL and the tKDL backgrounds (Fig 2). Results for a 50x cDNA dilution (S2 Data) replicated all these patterns, demonstrating that the 5x dilution results were not biased by upper limits of expression detection. There was no pattern of an increase in expression correlating to increased ploidy level. For Ak3 and ef1α head expression, males of almost all backgrounds (WPL haploid, WPL diploid, control haploid and tKDL haploid males) resembled each other strongly. However, tKDL diploid male head expression was female-like (like the diploid and triploid females). As these males arose from diverted female development through tra knockdown, it is possible they are deficient in male character relative to typical haploid males. Overall, there was more variation in female head expression, which tended towards being higher than the males. However, for Ak3, in diploid control, tKDL diploid, diploid (scarlet) WPL, and triploid WPL females the expression was as low as the haploid WPL male’s (Fig 2A). A clearer separation of lower male values versus higher female values was observed for the head expression of ef1α (Fig 2B). Abdominal Ak3 and ef1α expression was consistently low for WPL haploid, WPL diploid, tKDL haploid, and tKDL diploid males (Fig 2C), and high but variable for control diploid, WPL diploid (scarlet), WPL triploid, tKDL diploid, and tKDL triploid females (Fig 2D). These data show that in two polyploid backgrounds, gene expression does not scale to ploidy, but rather is dependent on sex.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 2. qPCR results for absolute gene expression of housekeeping genes Ak3 and ef1α.

Absolute expression and standard error for three replicates for 5x dilution of cDNA for (A) Ak3 in heads (B) ef1α in heads (C) Ak3 in abdomens and (D) ef1α in abdomens. Blue indicates male background, and red indicates female background. Control indicates non-injected individuals of the strain used to generate the transformer knockdown line (tDKL). If the prevalent hypotheses (expression scales to ploidy and sex-linked dosage mechanisms do not exist in Hymenoptera) were true expression would simply be 1x, 2x, and 3x for haploids, diploids, and triploids, respectively. Instead, expression is consistent for 1n and 2n males (with the exception for the head expression of the 2n tra KD males, possibly because these individuals would have normally developed into females without deactivation of the feminization process through tra knockdown). These data suggest that sex-linked expression conservation may be a mechanism to modulate dosage changes from polyploidization. (These patterns are also consistent for a 50x dilution).

https://doi.org/10.1371/journal.pone.0288278.g002

Male-female mating interactions differ between and among polyploid backgrounds

In our study, neopolyploid males were poor competitors against haploids for female mates (Tables 1 and 2). This was observed for both the F1 and the F5 generation, indicating a heritable phenotype. Our results diverge from the near-consensus of all other hymenopteran studies [55] (including WPL [36] that diploid males have equal mating success to haploid counterparts. The tKDL diploid males’ lesser ability to compete for females adds complexity to the current hymenopteran “diploid male vortex” narrative. For Hymenoptera in general it has been assumed diploid males are not inferior at mating success, so accelerating extinction for species with sterile diploid males, particularly those that arise from inbreeding as in species with complementary sex determination (CSD) [31–33, 60–62]. For CSD, hemizygotes for a csd locus or loci are haploid males and heterozygotes are diploid females, but homozygotes are typically reproductively impaired males. It is unclear what proportions of Hymenoptera are CSD versus non-CSD species. Here, we provide an example (for any hymenopteran) that diploid males of different backgrounds of a single species can be highly contrasting in their success with females. Additionally, despite performing poorly against haploids of their own background, the tKDL diploid males were superior mate competitors against the WPL diploids. Therefore, no general effects of polyploidy can be inferred, without considering various initial polyploidy events. Consequently, polyploidization cannot be considered as a single event but is dependent on the duplicated genome characteristics. Apparently, not all genomes are equally suited to survive diploidization.

The cause for disparities in tKDL and WPL diploid male mate competition ability are currently unknown. Although more study is needed, tKDL diploid males are not deficient in courtship or copulatory behaviors, and females do not reject them when they have no other mate choice. The fecundity of N. vitripennis diploid males is also similar to haploid males for both WPL [36] and tKDL [60, 63], so female preference for haploids does not seem to be based on reproductive potential. The tKDL diploid males also have no obvious physical impairment (as in the only other known case of hymenopteran diploid male mating failure, in which the male was too large for the female [64]). Interestingly, in the multiple competitions, male rejection was higher when only diploids were present, with female mating rate being lower in the tKDL diploid and WPL diploid competitions compared to the tKDL diploid versus haploid competitions. This suggests that even if they themselves have impaired attractiveness, diploid males may benefit from ambient haploid cues increasing female receptivity, whether behavioral, mechanical or, chemosensory. This could also partially account for the WPL diploids’ success against their own haploids [36].

Cell reduction varies between polyploid backgrounds

The WPL but not the tKDL appears to have a sex-linked cell reduction mechanism. There are fewer, larger cells in higher-level ploidy individuals (Fig 1, Table 4). Consistent with most hymenopterans (A. Thiel, pers. comm), polyploids and non-polyploids do not differ much from each other in overall body size for tKDL or WPL [63]. Therefore, it does not seem that a cell reduction mechanism is required to avoid polyploid gigantism, making its function for WPL unclear. It could simply be that cell reduction is a later stage polyploid phenotype, and so is present in long-maintained WPL but not neopolyploid tKDL. Whereas polyploid body size and its relationship to cell number and size has been sparsely studied outside of vertebrates [4, 68, 69], this is the first example of cell reduction varying drastically within an animal polyploid.

It should be noted that these inferences are clouded by tra having a known role in body size regulation (in Drosophila [70, 71]). The absence of cell reduction in tKDL could be an artefactual result of deactivated tra, and not a difference in polyploid phenotype between polyploid backgrounds per se. General corruption of N. vitripennis tra, whose best-known role is sex determination through activation of the feminization pathway, is apparent from the observation of occasional gynandromorph offspring from the dsRNA injected female. As other genes that can be targeted to create N. vitripennis neopolyploids [41, 42] do not have a known role in body size, additional study of knock down lines and longer-term study of neopolyploids will elucidate how common cell reduction mechanisms are in polyploids, and whether they evolve over time. Ideally, more tissues should be sampled to investigate cell reduction in Nasonia polyploids. Unfortunately, this is complicated by some tissues having endopolyploid (multinucleated) cells; for example, nearly all hymenopterans have male endopolyploid thoracic tissues [50]. These cells may be larger, preventing accurate comparison between non-polyploid and polyploid individuals. Cell-counting accuracy of 3D organs is also difficult due to incomplete marker penetrance or disassociation of individual cells. A good candidate for developing protocols for further cell reduction investigation are the brain cells, which consistently and accurately reflect the individual’s ploidy [72]. A reference atlas for the Nasonia brain has recently been published [73], so it should be possible to i.e. use exact neuron count to more fully assess cell reduction in Nasonia polyploids.

Gene dosage is controlled by sex and does not scale to ploidy

How a neopolyploid organism copes with expression changes following a sudden increase in genomic material is one of the biggest questions of polyploid biology [5, 39, 74]. An intuitive assumption would be that gene transcript number scales directly with ploidy to maintain balanced gene networks (e.g. 1x haploid, 2x diploid, 3x triploid) [5]. However, there has been little evidence to support this [75, 76]. Additionally, sexual dosage compensation mechanisms vary between organismal groups (e.g. mammals versus birds versus insects[77]) and it is not clear whether dosage compensation mechanisms exist in haplodiploids to accommodate disparate ploidy levels inherent to the sexes, despite their lack of sex chromosomes [2, 50, 78]. We examined WPL and tKDL to ask 1) is absolute gene dosage directly related to ploidy 2) do hymenopterans have sexual dosage compensation 3) if they do, are they disrupted by polyploidization and 4) do dosage effects vary across polyploid backgrounds?

In our study we observed in both WPL and tKDL a strong pattern of sex-linked dosage conserved against polyploidy (Fig 2, S2 Data), which has not been previously observed in any other hymenopteran species. That we studied absolute expression of two housekeeping genes tentatively implies a general mechanism across genes without sex-specific function. Interestingly, throughout Hymenoptera (except for basal-most Xylidae), endopolyploidy (diploidy) is high in body tissues of haploid males, purportedly to match the metabolic and mechanical abilities of diploid females [50]. This would imply that the haploid genome cannot just double its expression. Yet, in our study, male diploidy does not increase expression (in the head or abdomen), calling into question if they differ fundamentally from (diploid) body cells, and if so, how and why.

Although there have been a few studies on diploid male ants [79, 80], this study focuses on polyploid hymenopteran expression to consider triploid females. We uncovered consistent sex-conserved dosage for both the long established and the neopolyploid backgrounds despite their many contrasting polyploid phenotypes, suggesting an ingrained mechanism that may act as a buffer against the genomic shock of ploidy changes. It may be a factor in how there are additional forms of aberrant ploidy in Nasonia such as haploid [79] and tetraploid females [37] that are viable and not seriously impaired outside of reproductive function. Although more in-depth studies are needed, particularly across whole transcriptomes and with rigorous sampling standardization [75], our results point at the existence of sex-linked mechanisms that buffer against would-be dosage altering effects of polyploidization.

Phenotypic variation bridges the polyploid hop and Nasonia vitripennis as an experimental animal polyploid model

We synthesize our results to introduce the idea that different polyploidization events can reflect extreme phenotypic variation in polyploid phenotypes, with some allowing the polyploid hop to be bridged (Fig 3). In an evolutionary context, there is a gradient of outcomes. Highly deleterious pathways resulting in an “evolutionary dead end” are the most common. However, it has long been hypothesized without experiment-driven proof that there are pathways that have fewer impacts on fecundity and cause fewer problems of epigenetic disruption to allow establishment of polyploid lineages [5, 68]. Although our study concerns a system with haplodiploidy (~20% of animal species [80]), it lends possible insight in the prevalence of WGDs in the ancestry of many eukaryotes.

Download:

• PNG
  larger image
• TIFF
  original image

Fig 3. Outcomes of hypothetical alternate polyploidization pathways.

An organism can become neopolyploid through multiple pathways representing a gradient of least likely to result in polyploid lineage establishment (an evolutionary dead end) to most to likely (the hypothetical optimal polyploidization pathway). However, one polyploid pathway can simultaneously result in some neopolyploid phenotypes that are more conducive to survival and success, and others that are not. Both WPL and tKDL serve as examples, but for different traits.

https://doi.org/10.1371/journal.pone.0288278.g003

Polyploidization occurs more often in invertebrates than vertebrates [5, 68, 81], but animal polyploidy research has been heavily biased towards fish and amphibians [18, 23, 69]. This is possibly due to the popularity of some of these vertebrate species as models, but this ironically translates to a striking dearth of knowledge of polyploid evolution in invertebrates, the taxa for which it is most common. Furthermore, just as there has been taxonomic bias in favor of plants versus animals in polyploid research due to easier means of study and a misconception about relative evolutionary and applied importance [18, 69], there has been a parallel situation in hymenopteran polyploid research. Most of this work has been done in CSD species, possibly because polyploidy is readily inducible (i.e. through inbreeding) and is readily detected. However, non-CSD parasitoid wasps comprise most of the commercially important hymenopterans [82] and the majority of hymenopteran diversity (including the Nasonia’s superfamily Chalcidoidea, which by itself may account for 6% of all animal diversity on Earth [74, 83].

The types of polyploid questions investigated in other animal systems are also perhaps illustrative of the specific advantages and limitations of neopolyploid research in Nasonia. For example, both the fish genus Carrasius and the frog lab model Xenopus also exhibit a range of ploidies. Carrasius specifically has several modes of sexual and asexual reproduction and varying degrees of fecundity, and work has been done to link these factors to ploidy [84]. Xenopus has the widest range of ploidy levels in tetrapods (diploid to dodecaploid), originating from species hybridization events (allopolyploidy). It has been used for example to study the evolution of subgenomes over evolutionary time [85]. Unlike Crassius and Xenopus, Nasonia polyploidy has not been recorded in nature, so it cannot be used to study real-world evolution of polyploid population dynamics (although artificial lab populations are potentially possible). Furthermore, there is no evidence that Nasonia allopolyploids are possible, although hybrid offspring between the four species are commonly produced in studies [e.g. 86, 87]. These Nasonia hybrids do, however, present an interesting possibility of combining hybridization with knockdown polyploidization, to study variation in dosage and dominance effects from different variations on which species contributes the polyploidized genome. Given its well annotated genome and the possibility of incorporating eye markers to easily track polyploid and nonpolypoid individuals (as in the WPL), N. vitripennis is uniquely well positioned for lab studies on neoautopolyploid phenotypes and gene expression.

Three of the challenges that successful autopolyploid lineages must overcome in their polyploid hop are meiotic problems, cellular changes, and gene dosage issues [28], but animal experiments demonstrating how this can be achieved have been limited because polyploids typically die in the first generation. We have shown that in a single species, there can be contrasting polyploid phenotypes for male fitness, meiosis, and cell reduction, but also mechanisms that are more consistent, such as sex-linked gene dosage that is resistant to ploidy change. Additional insight becomes possible with more N. vitripennis neopolyploid lines, which can easily be induced through knockdown of transformer [40], transformer-2 [42] and wasp overruler of feminization [41] to produce founder diploid males and assay their phenotypes. Surveying the full range of polyploid biology for the long-established WPL, the neopolyploid single gene knockdown lines, and possibly induced WGD lines (e.g. adapting the methods of [30]) would further refine what constitutes a polyploidization pathway best suited to bridge the polyploid hop. The ability to create various long-term polyploid lineages, in conjunction with many existing resources for advanced genetic study of N. vitripennis, e.g. fully sequenced genome and transcriptome [35, 88], RNAi for gene knockdowns [89], and CRISPR for gene knockouts [90], mean that N. vitripennis can be developed into a comprehensive experimental model for animal polyploidy, a need in polyploid research that has been repeatedly called-for [2, 18, 28, 58, 59, 69].