Introduction

Our modern world is rapidly becoming a hostile environment for many species, reflected in the global decrease in biodiversity. At the forefront of this loss is a concerning decline in insect biomass, attributed to anthropogenic changes in land use, climate change and pollution1. The ecological impacts of declining insect populations are already evident, with parallel declines in populations of insectivorous species and in insect-pollinated plants2. Many of these insect species also perform critical roles in the pollination and natural pest control of agricultural crops and are accordingly an important determinant of food security.

Ironically, the allocation of land for agricultural purposes is considered a major factor driving insect decline and impaired fertility3,4. Many studies attribute impaired insect reproduction to use of pesticides and insecticides in cropping, which can disrupt the reproductive function of insects outside of the target species5. In most cases, these compounds also intentionally act as endocrine disrupting chemicals (EDCs). Known to alter normal endocrine function and reproduction, EDCs are now recognised as a direct cause of in- or subfertility in both vertebrate and invertebrate species6,7,8. In addition to EDCs from pesticides, insects may be exposed to EDC’s present within the crops themselves. The replacement of natural, varied feed sources with uniform crop species increases the widespread dietary exposure of insects to EDCs. Many commercial crop and pasture species, such as clover, soybean and alfalfa, contain phytoestrogens that can act as endocrine disruptors9,10. Phytoestrogens can agonistically or antagonistically interact with both oestrogen receptors11,12 and the arthropod-specific ecdysone receptors13. Phytoestrogens and their metabolic derivatives, which are present in large amounts in plants consumed by wildlife, livestock and throughout the Western diet, are suspected to cause sub-fertility in many species14,15,16,17. A potential contributor to declining insect biomass is the interplay between phytoestrogens and reproduction, which, in other species, can lead to persistent subfertility at a transgenerational level6,18.

Due to the ubiquity of EDCs in the environment, each individual organism has a complex and unique ancestry of EDC exposure. How these complex lineage combinations of EDC exposure contribute to reproductive capacity across generations is not well understood.

The rapid generational turnover within insect populations promotes one of two possible outcomes when faced with an environmental insult. The first is a detrimental population decline due to rapid impact upon many individuals. The second outcome is that the physiological changes elicited by the environmental insult are transgenerational, and that the persistence of these changes results in adaptation at an evolutionary scale19,20. The impact of the exposure to ecologically relevant levels of dietary phytoestrogens on the fertility of insects and the transmission of subfertility across multiple generations by either female or male lineage are not known. Importantly, to our knowledge, no study has comprehensively investigated the effect of different exposure lineages, any cumulative effect if both parents were exposed in adulthood prior to conception and how reproductive compromise manifests within each distinct exposure lineage. This study uses Drosophila melanogaster to model the reproductive impact of adult insect exposure to oestrogenic metabolite equol, a common phytoestrogen derivative in the diets of insects, livestock, humans, and non-domesticated species21,22. As phytoestrogens are shown to be present in both fermented plant material23,24,25 and in pollen26, insects such as D. melanogaster are likely to encounter them in their diet. We use this model to form distinct lineages of either paternal, maternal, or combined maternal-paternal equol exposure to examine reproductive function in both the male and female, across 3 generations. In addition, we demonstrate a novel suite of flow cytometry tools to assess real time, live sperm functionality in D.melanogaster.

Methods

Fly collection and rearing

The experimental population was initially established by breeding 30+ isofemale lines of D. melanogaster collected from Innisfail, Queensland, Australia. Flies were maintained as 2–4 colonies of several thousand flies and were started in 2019 from collections of wild D. melanogaster. The populations were supplemented with wild flies from the same source every few months. Flies were housed under standard conditions (25 °C, 12:12 h light–dark cycle, and 65% relative humidity) and fed a standard Drosophila diet consisting of 1% (w/v) agar, 2% (w/v) brewer’s yeast, 8.5% (w/v), sugar, 6% (w/v) cornmeal and 0.25% (v/v) Nipagin (H5501; Sigma-Aldrich, St Louis, MO, USA).

To generate the parent generation for study inclusion, vials were placed in a population container of several thousand flies to allow females to lay eggs over 24 h. Vials were then removed, and flies were sex-sorted prior to mating to ensure flies were virgins prior to study inclusion. The treatment groups outlined below were developed using isolated virgin flies.

Experimental design

Establishment of relevant concentration of dietary equol

To establish a dosage of the oestrogenic compound equol that would impair fly fertility but not render them completely infertile, flies were exposed to diets containing a final concentration of either 0, 0.5, 5 or 50 μM equol. To account for the ethanol used to dissolve equol, all diets contained 0.5% ethanol (v/v).

Flies were housed in sex-sorted groups of 10 per vial, with a total of 40 vials per diet (20 vials for each sex). Flies were transferred into a fresh vial containing the same diet every 3 days, to ensure feed availability over the study period.

After 10 days of dietary equol exposure, flies were mated (n = 20 vials per dietary treatment group) in the following groups, within dietary group:

  1. a.

    Control female + control male

  2. b.

    Control female + equol-exposed male

  3. c.

    Equol-exposed female + control male

  4. d.

    Equol-exposed female + equol-exposed male

Flies were mated as 2 males and 2 females per vial on standard food. Males were removed after 6 h of mating, and females removed after a further 24 h of egg laying. Reproductive capacity (number of eggs laid per vial, % of eggs developing to adult) was assessed as described in detail below. From these data (Supplementary Fig. 1), it was determined that 5 μM equol was able to reduce both male and female fertility but still allow for the generation of adult F1; 0.5 μM had no significant effect and 50 μM equol caused very few offspring to be produced, inhibiting investigation into the fertility of subsequent generations. A dosage of 5 μM dietary equol was used in the subsequent investigation in fly fertility.

Investigating the transgenerational impact of equol on male fertility in D. melanogaster

Parent generation (P) flies were housed in sex-sorted groups of 10 per vial on either a control or 5 μM equol diet. 20 vials of male and 20 vials of female flies were used to establish the control and exposure lines. Flies were transferred into a fresh vial containing the same diet every 3 days, to ensure feed availability over the study period. For all traits measured in this study, flies were sampled across vials within treatment group, in order to capture variation between vials.

After 10 days on either a control or equol diet, flies were mated (n = 20 vials per dietary treatment group) in the following groups:

  • Control female + control male

  • Control female + equol-exposed male

  • Equol-exposed female + control male

  • Equol-exposed female + equol-exposed male

Newly eclosed flies (F1) were removed and housed into sex-specific groups on standard food as they hatched, to prevent mating. The F1 generation was then bred as described above in the following groups:

  • Control female + control male

  • Control female + male with exposed male parent

  • Control female + male with exposed female parent

  • Control female + male with both parent exposed

  • Female with exposed male parent + control male

  • Female with exposed female parent + control male

  • Female with both parent exposed + control male

Newly eclosed F2 generation flies were removed and housed into sex-specific groups on standard food as they hatched, to prevent mating. The F2 generation was then bred as described above in the following groups:

  • Control female + control male

  • Control female + male with exposed male grandparent

  • Control female + male with exposed female grandparent

  • Control female + male with both grandparents exposed

  • Female with exposed male grandparent + control male

  • Female with exposed female grandparent + control male

  • Female with both grandparents exposed + control male

Lineage treatment groups can be viewed in Supplementary Fig. 2.

Measures of reproductive capacity

For all generations, flies were counted, sexed and day of eclosion recorded as they hatched, which occurred for all groups 9–11 days after females were removed.

A subset of 15 males files, from each group of each generation, were dissected and seminal vesicle size measured. Flies were anaesthetised by cooling vials to approximately 0 °C and then rapidly culled at time of dissection by crushing the head with a dissecting pin. Seminal vesicles were isolated by pulling the abdomen away from the thorax, photographed using a dissecting microscope and area of the seminal vesicles (n = 10 per treatment group) analysed in ImageJ as per previous reports27

qPCR of genes related to reproductive function

Across all generations, a further subset of 40 flies per treatment group were taken at 10 days of age and flash frozen in liquid nitrogen for RNA analysis. Genes with key roles in the processes of D. melanogaster spermatogenesis (Pif1A, tplus3B)28,29 and female germline stem cell survival and oogenesis (eggless)30 were selected following fertility outcomes, in order to investigate the specific reproductive changes observed in the experiment.

RNA isolation

All RT-qPCR work included two technical replicates, three to five independent samples per treatment group for each sex, and 5–8 flies per sample. RNA isolation was performed in accordance with previous D. melanogaster studies31. Total RNA was isolated from each sample using the QIAzol Lysis Reagent as per manufacturer’s instructions (Qiagen, Melbourne, Australia). Reverse transcription was carried out at 42 °C using murine Moloney leukemia virus reverse transcriptase, random hexamers (Promega, Sydney, Australia), and 2 μg of total RNA from each sample. The resulting cDNA was purified using the UltraClean® PCR Clean Up Kit (MoBio Laboratories Inc., Carlsbad, CA, USA) as per manufacturer’s instructions.

Real-time qPCR

Real time qPCR was carried out in duplicate in a CFX384 Touch Real-Time PCR Detection System (Bio-Rad Laboratories Pty Ltd, South Granville, Australia) with a temperature profile of 95 °C for 10 min, followed by 40 cycles of 95 °C for 1 s, an annealing temperature of 60 °C for 15 s, and 72 °C for 5 s to determine relative levels of genes controlling for different reproductive functions, including Pif1A, tplus3B and eggless. The genes Sdha, Act and ef1 were used as reference genes.

Relative mRNA expression levels were quantified against a standard curve of tenfold serial dilutions of PCR product and normalized against the geometric means of the reference genes32. Melt peaks were analysed to ensure that a single PCR product was obtained. Primer pairs and protocols were designed with Primer-BLAST33 to include introns to avoid amplification of contaminating genomic DNA.

Assessment of sperm quality using flow cytometry

All parameters were assessed using a BD FACSCanto flow cytometer (BD Biosciences, NSW, Australia) using previously described methods34,35,36,37,38,39,40,41 modified and optimised for fly spermatozoa (Supplementary information). Sperm were isolated by rapidly dissecting the seminal vesicles out of 10–15 males of the same treatment group in a 20 μl droplet of Grace’s insect culture medium (Sigma-Aldrich, St Louis, MO, USA) on a slide. A coverslip was then placed over the seminal vesicles and gently tapped once to release the sperm. A 200 μl pipette was used to flush the cover slip and slides with Graces insect culture medium to collect sperm, avoiding the remaining sections of seminal vesicles. A total volume of 1 ml from this flushing was collected for use in flow cytometry. This was repeated ten times for each treatment group to form ten replicates for all sperm measures. All reagents and equipment were maintained at 25 °C throughout the experiment.

Briefly, sperm cells were isolated from total events based on forward and side scatter profiles34,35,36,37,38,39,40. The concentration of sperm in sample assessment was between 5 and 10 × 106/mL. Unless otherwise specified, 10,000 sperm cells were recorded per sample for each parameter using BD FACSDiva software (BD Biosciences) and analysed using FlowJo software (Version 10.7, Becton, Dickinson and Company 2020). Aside from acridine orange staining for DNA fragmentation and mitochondrial membrane potential, analysis of all other parameters assessed only the live sperm population, as previously described34,35,36,37,38,39,40.

Viability

Sperm viability and acrosome integrity were examined by dual staining with propidium iodide (PI, final concentration of 4 μM) and fluorescein isothiocyanate peanut agglutinin (FITC-PNA, final concentration 0.4 μg/mL) for 10 min at 25 °C, as adapted from previous protocols40.

Mitochondrial superoxide production

Mitochondrial superoxide production was assessed through co-staining with Mitosox Red (final concentration 2.5 μM) and Sytox green (final concentration 30 nM) for 20 min at 25 °C, as adapted from previous protocols40.

Mitochondrial membrane potential

Ther percentage of spermatozoa with high or low mitochondrial membrane potential were identified through use of JC-10 staining42. An aliquot of sperm cell suspension was gently centrifuged (600 rpm for 5 min), the supernatant removed, and the cells resuspended in 500 μl of 1 × JC-10 loading dye. Spermatozoa were incubated for 20 min at 25 °C, before assessment on the flow cytometer, using a 610/20 and 530/30 nm band pass filter to detect red and green fluorescence respectively. Cells were considered to have high mitochondrial membrane potential if the populations displayed a greater proportion of orange aggregated fluorescence. This is in comparison to the green- fluorescing cell population, which was considered to have low mitochondrial membrane potential.

Membrane lipid disorder

Changes in membrane lipid disorder in the viable sperm population were assessed through dual staining with merocyanine 540 (M540, final concentration 0.8 μM) and Yo-Pro (final concentration 25 nM) for 10 min at 25 °C. M540 fluorescence was detected on a 585/42 nm band-pass filter and Yo-Pro fluorescence detected on 530/30 nm band-pass filter, as adapted from previous protocols40.

Lipid peroxidation

Lipid peroxidation of the sperm membrane was assessed using Bodipy C11 (581/591). Samples were incubated with the Bodipy C11 probe (final concentration 10 μM) for 30 min at 25 °C before being centrifuged for 5 min at 600g. Following the removal of the supernatant, sperm pellets were resuspended in fresh buffer (Graces insect culture medium) and an aliquot counterstained with PI (final concentration 4 μM) for 10 min at 25 °C, as adapted from previous protocols40. An extra sample was incubated with 5 μM hydrogen peroxide to promote higher levels of lipid peroxidation as a positive control.

DNA fragmentation

DNA fragmentation was assessed was per previous protocols40,41 with some adjustments. An aliquot of sperm cell suspension was gently centrifuged (600 rpm for 5 min), the supernatant removed and the cells resuspended in 200 μl with 1× TNE buffer (0.15 M NaCl, 0.01 M Tris HCl, 1 mM disodium EDTA pH 7.4). Samples were then diluted with 400 μl acid detergent solution (0.08 NHCl, 0.15 M NaCl, 0.1% Triton X 100 pH 1.2). Exactly 30 s later, samples were stained with 1.2 mL acridine orange (6 μg/mL). DNA fragmentation was estimated by the relative amount of single stranded and double stranded DNA, indicated by the proportion of sperm demonstrating red fluorescence to total (red and green) fluorescence. As green fluorescence was visualised along the Y-axis and red along the X-axis, cells that exhibited red fluorescence displayed a right-hand shift away from the main population. For each sample, a total of 5000 spermatozoa were recorded.

Sperm protamine deficiency

Samples were assessed for protamine deficiency as previously described43,44, with some adjustments. An aliquot of the sperm cell suspension was gently centrifuged (600 rpm for 5 min) and resuspended in 200 µl of phosphate buffered saline (PBS) with 0.5% Tween. Samples were then washed once in PBS (300g for 10 min) and resuspended in 100ul of McIlvaine’s buffer (17 ml 0.1 mol/l citric acid mixed with 83 ml 0.2 mol/l Na2HPO4 and 10 mmol/l MgCl2, pH 7.0) containing 0.25 mg/ml chromomycin (CMA3). Cells were stained for 20 min.

CMA3 fluorescence from gated cells was obtained through a 528/45 bandpass filter after excitation with a violet laser (405 nm). The degree of protamine deficiency was assessed by recording the median CMA3 fluorescence of the sperm population.

Statistical analysis

To identify differences and any interactions between equol exposure, sex within generation, the sex of the individual fly was combined with the lineage of equol exposure to form a “treatment group”. A one-way ANOVA was performed to compare between lineages within a generation. For the F1 and F2 generations, a generalised linear mixed model was used to investigate any interaction between the sex of the fly and it’s lineage of equol exposure. The generalised linear mixed model was only performed where sex was a relevant factor, which included offspring production and the percentage of male offspring produced. The use of the “summary” function was used to address overdispersion.

Whether differences were assessed via a one-way ANOVA or the Kruskal Wallis test depended on data normality and homogeneity of residual variances. Normality and homogeneity of residual variances were confirmed using the Shapiro–Wilk test and Levene's test, respectively, as well as by visualisation of residual plots. If a one-way ANOVA was used, Tukey's Honestly Significant Difference was used as a post-hoc test to identify differences between each lineage group within a generation. Where Kruskal Wallis was used, Dunn’s test used as a post-hoc test. Data was analysed in R Studio (version 4.3.0)45. Where required, the data were transformed to correct for unequal variances. A log transformation was used to reduce heteroscedasticity of the residuals if necessary. Variables that required transformation are outlined in Supplementary Table 2. If homogeneity of variance could not be achieved by transforming the data, Kruskal Wallis was used. If a log transformation was performed, the results were back transformed and presented as the geometric mean ± 95% confidence intervals. Details of all statistical models can be found in Supplementary Table 2.

Results

Dietary equol reduces the fertility of exposed parents and two subsequent generations

In the parent generation, dietary exposure to equol in either or both parents reduced the number of offspring produced (H = 22.004, DFn = 1, P < 0.001, Fig. 1a). The reduction in fertility persisted in most F1 individuals with parental equol exposure (H = 34.913, DFn = 3, P < 0.001, Fig. 1c) and was evident in ten of the twelve F2 groups with equol exposure in their lineage (H = 43.63, DFn = 6, P < 0.001, Fig. 1f). Equol exposure in either parent, or in combination, resulted in a decreased proportion of male to female offspring produced by P individuals (H = 19.377, DFn = 1, P < 0.001, Fig. 1b). This shift in the male to female ratio was observed in offspring produced from the F1 and F2 generations when equol was included in the diet of the parent generation (Fig. 1d,g).

Figure 1
figure 1

Fertility outcomes and sex ratio of offspring from the P, F1 and F2 generations with differential exposure to dietary equol. Values presented are the mean ± SEM of offspring produced per mating pair (n = 10 per mating combination). Seminal vesicle size is presented from males of F1 and F2 generations with differential lineage of exposure to dietary equol. Values presented are the mean ± SEM. Different superscript represents statistical differences (P < 0.05) within generation. Means are not compared between generations. Refer to Legend below: .

Full size image

There was not an interaction between the sex of the individual and their lineage of equol exposure on the number of offspring produced in the F1 generation (F = 0.157, DFn = 1, P > 0.05). This interaction did occur in the F2 generation (F = 10.435, DFn = 6, P < 0.0001), where daughters from a F1 female with exposed female parent produced more offspring in comparison to their male siblings.

The ratio of male to female offspring produced by the F1 did not demonstrate any interaction between the lineage of equol exposure and the sex of the individual (F = 0.431, DFn = 1, P > 0.05).

In the F2 generation, there was an interaction between the sex of the individual and their lineage of equol exposure on the ratio of males to females produced (F = 22.101, DFn = 6, P = 0.0132). The percentage of male offspring produced was reduced in F2 males, compared to their female siblings, if their father had an exposed male parent, or if their mother had both parents exposed. The percentage of male offspring produced was reduced in F2 females, compared to their male siblings, if the F2 female had a father with an exposed female parent or if their mother had an exposed female parent.

The size of the seminal vesicle sizes of F1 males decreased if either or both parents were exposed to equol, with the smallest seminal vesicle sizes observed if the male parent was exposed to equol (H 22.004, DFn = 3, P < 0.001, Fig. 1e). The size of the seminal vesicle was smaller in all F2 individuals with dietary equol exposure in their lineage than that of flies with no equol exposure in their lineage (H = 34.913, DFn = 6, P < 0.001, Fig. 1h).

Dietary equol negatively alters sperm functionality in exposed males and male descendants of equol-exposed individuals

Direct exposure to dietary equol in the parent generation did not impact sperm viability, protamine deficiency or mitochondrial superoxide production (P > 0.05), but did increase sperm DNA fragmentation (F = 18.438, DFn = 1, P = 0.049, Fig. 2b), membrane lipid disorder (H = 10.556, DFn = 1, P = 0.001, Fig. 3b) and lipid peroxidation (H = 6.412, DFn = 1, P = 0.011, Fig. 3a). Sperm mitochondrial membrane potential tended to be lower in equol-exposed males (H = 3.866, DFn = 1, P = 0.0509, Fig. 4b).

Figure 2
figure 2

The viability, DNA fragmentation and protamine deficiency of spermatozoa from parent, F1 and F2 generation males, assessed via flow cytometry (n = 10 replicates, each replicate consisting of sperm from 10 to 15 males, per treatment group). Parameters presented include the percentage of viable spermatozoa as represented by the % of cells negative for PI; percentage of sperm DNA fragmentation, considered as the ration of red to green fluorescence after staining with acridine orange and percentage of spermatozoa with protamine deficiency, considered as the population of spermatozoa with higher CMA3 fluorescence. Values presented are the mean ± SEM. Different superscript represents statistical differences (P < 0.05). Refer to Legend below: .

Full size image
Figure 3
figure 3

Sperm membrane lipid peroxidation and membrane lipid scrambling in spermatozoa from parent, F1 and F2 generation males, assessed via flow cytometry (n = 10 replicates, each replicate consisting of sperm from 10 to 15 males, per treatment group). Parameters presented include the percentage of viable spermatozoa with lipid scrambling in the sperm membrane, considered as the M540 positive/YoPro negative population and percentage of viable spermatozoa with lipid peroxidation of membrane lipids, considered as the Bodipy C11 581/591 positive/PI negative population. Values presented are the mean ± SEM. Different superscript represents statistical differences (P < 0.05). Refer to Legend below: .

Full size image
Figure 4
figure 4

Sperm mitochondrial superoxide production and mitochondrial membrane potential in spermatozoa from parent, F1 and F2 generation males, assessed via flow cytometry (n = 10 replicates, each replicate consisting of sperm from 10 to 15 males, per treatment group). Parameters presented include percentage of viable spermatozoa with high mitochondrial superoxide production, considered as the Mitosox positive/Syber Green negative population and the percentage of spermatozoa with high mitochondrial membrane potential, represented by the population exhibiting a shift toward increased JC-10 fluorescence. Values presented are the mean ± SEM. Different superscript represents statistical differences (P < 0.05). Refer to Legend below: .

Full size image

Males of the F1 generation had reduced sperm viability if one or both parents were exposed to equol (H = 23.003, DFn = 3, P < 0.001, Fig. 2d). F1 males with an exposed female parent or both parents exposed produced spermatozoa with higher protamine deficiency than control males of the same generation (F = 4.169, DFn = 3, P = 0.006, Fig. 2f). F1 Males produced from a mating where either the male or female parent had been exposed to equol had increased sperm membrane lipid disorder (H = 21.550, DFn = 3, P < 0.001, Fig. 3d), lipid peroxidation (H = 31.998, DFn = 3, P < 0.001, Fig. 3c) and mitochondrial membrane potential (H = 13. 868, DFn = 3, P = 0.003, Fig. 4d). F1 Males with both parents exposed to equol did not differ from non-exposed controls in any parameters outside of sperm viability and DNA fragmentation (P > 0.05, Figs. 3 and 4).

Sperm function in F2 males was differentially impacted by lineage and P exposure to equol (Figs. 2, 3 and 4). Sperm membrane lipid disorder was reduced in all F2 males with equol exposure in their lineage (H = 40.579, DFn = 6, P < 0.001, Fig. 3f). Sperm DNA fragmentation was increased in all F2 males with equol exposure in their lineage (F = 2.406, DFn = 6, P = 0.004), with the exception of F2 males produced from the mating combination of a control female with a male with both parents exposed (Fig. 2h). Compared to F2 control males, sperm lipid peroxidation was also lower in F2 males produced from the mating combination of a control female with a male with both parents exposed (H = 33.266, DFn = 6, P < 0.001, Fig. 3e). Lipid peroxidation was increased in F2 males if their female parent (F1) had either an exposed female parent, or both parents exposed (H = 33.266, DFn = 6, P < 0.001, Fig. 3e). The spermatozoa of F2 males were protamine deficient compared to control males if the males were produced from a female with an exposed male or female parent crossed with a control male (H = 43.63, DFn = 6, P = 0.02, Fig. 2i).

Expression of genes related to reproductive function is reduced in the equol-exposed male parent, but not in subsequent generations

Dietary exposure to equol reduced the expression of Pif1A (F = 6.92, DFn = 1, P = 0.007, Fig. 5b) and TplusB (F = 19.46, DFn = 1, P = 0.009, Fig. 5c). in male flies of the parent generation. These changes did not persist beyond the exposed parent generation (Fig. 5e,f,h,i), aside from an increase in Pif1A expression in F1 males with equol-exposed fathers (F = 11.05, DFn = 3, P = 0.0003, Fig. 5e), no significant changes in expression of these genes were observed in males of the F1 and F2 generations. In the F2 generation, daughters of (i) an F1 female with an exposed female parent, (ii) F1 male with both parents exposed and (iii) an F1 female with both parents exposed had higher relative expression of eggless compared to control F1 females (F = 4.447, DFn = 4, P = 0.01, Fig. 5g). The relative expression of eggless was not different between groups in the P and F1 generations (P > 0.05, Fig. 5a,d).

Figure 5
figure 5

The relative expression of genes eggless, Pif1A and TplusB in the P, F1 and F2 generations. Values for eggless are presented only for female flies and Pif1A and TplusB only for male flies. The mating combination describes the parentage of the female (eggless) or male (Pif1A, TplusB) offspring in the F1 and F2 generations. Not all mating combinations are presented for the F2 generation due to low numbers of offspring available for all measures. Values presented are the mean ± SEM. Different superscript represents statistical differences (P < 0.05) within generation. Means are not compared between generations. Refer to Legend below: .

Full size image

Discussion

In mammals, evidence suggests that endocrine disrupting chemicals are a key environmental stressor driving the compromise of reproductive function at a transgenerational level. The present study demonstrates that this phenomenon is also observed in D. melanogaster, and that sub-fertility persists even after the introduction of non-exposed individuals into the breeding population. In this insect model, dietary exposure to oestrogenic EDC equol negatively impacts both male and female fertility two generations beyond the exposed individual, depending upon the specific lineage contribution of equol exposure. This result supports that phytoestrogens, which are naturally occurring dietary EDCs in wildlife, humans and livestock, not only compromise reproductive function in insects via direct exposure, but can promote epigenetic modifications leading to transgenerational decreases in reproductive capacity. The observed impacts of equol exposure on fertility are particularly evident in male descendants. We show that the functionality of mature spermatozoa is impacted in transgenerational manner, and that the profile of sperm functionality manifests differently depending upon where dietary equol exposure occurred within an individual’s ancestry.

Direct exposure to dietary equol clearly reduces offspring production in both male and female D. melanogaster, suggesting that dietary phytoestrogens could contribute to the decline in insect populations. Interestingly, the ability of insecticides to act as EDCs and promote a transgenerational disruption of fertility is well studied in mammals46, but there is less documentation on the reproductive outcomes of non-target insect species. Unsurprisingly, the insect growth regulators used in common pesticides reduce fecundity and lifespan in non-target invertebrates, as well as inducing hypermethylation in the DNA of both exposed individuals47 and non-exposed offspring48. Despite the lack of insect-specific studies, mammalian models broadly support that phytoestrogens and other steroid-like EDCs can impair fertility across generations. Phytoestrogens have been shown to directly impair both male and female reproductive function in sheep49,50, rodents51,52,53, rhinoceros17 and humans54 at least partially through interactions with oestrogen receptors55,56. As the D. melanogaster oestrogen-related receptor (ERR) is analogous to the mammalian oestrogen receptor57,58,59, it is possible that equol is similarly able to interact with ERR to promote reproductive dysfunction.

While it has long been suspected that epigenetic reprogramming is responsible for the transgenerational impacts of EDCs on mammalian fertility, it is unclear how these changes manifest, and whether reproductive dysfunction persists in lineages diluted by breeding with non-exposed individuals. Though insects have a rapid generational turnover compared to most mammals, the transgenerational effects of EDCs on insect reproduction are not well documented or characterised. In the present study, a reduction in offspring production mostly persists in the F1 generation, with F1 males bearing the greatest cost to reproductive output, along with F1 females with exposed male parents. By the F2 generation, we find that both male and female fertility is compromised, with the exception of F2 females generated from an F1 male with both parents exposed and F2 females generated from F1 females with an exposed male parent. These two lineage patterns are also the only exposure groups that do not demonstrate a bias toward the survival of female offspring. Regardless of their lineage, all male offspring of the F1 and F2 generation present with reduced seminal vesicle size, a morphological change commonly linked with reduced fertility60. This supports evidence in mammals that oestrogenic EDCs mediate permanent alterations to male reproductive tract morphology that can impair fertility61,62,63. Our results suggest that both male and female DD. melanogaster with an ancestry of equol exposure are at risk of reproductive dysfunction, with the poorest fertility outcomes in the F1 and F2 generations observed in males with equol exposure in their lineage.

During the development of offspring, there are instances where one sex may be biologically prioritised over the other when faced with an environmental stressor. There is an extra layer of complexity when the environmental stressor is also an endocrine disruptor, as these compounds can directly interfere with normal endocrine signalling required for sex differentiation64,65. Our study supports previous evidence that oestrogenic EDCs promote a female survival bias in two generations offspring from exposed individuals. The skew toward female offspring following ECD exposure has been previously observed in the F3 generation in mice, where female mice of the parent generation were exposed to steroid-like EDC di(2-ethylhexyl) phthalate66. An enhanced ratio of female offspring has also been reported in developing tadpoles exposed to a glyphosate salt67 and in catfish larvae exposed to a mixture of phytoestrogens68. In each of these cases however, exposure occurred during a developmentally sensitive period and the promotion of female offspring was attributed to altered endocrine signalling by the steroid-like properties of the EDCs. In the present study using mature D. melanogaster, it is possible that the oestrogen-like structure of equol allows this molecule to interact ERR, causing downstream signalling events that interfere with sexual differentiation of subsequent generations. The ERR has crucial roles in male Drosophila fertility and sexual differentiation, with ERR knock out models resulting in males with abnormal sperm morphology, lowered sperm production and comprised sperm mitochondrial function58,59. The oestrogenic EDC BPA increases ERR expression in directly exposed male Drosophila, however whether this change in ERR expression impacts future generations was not explored69. Phytoestrogens are known to interact with other hormonal processes in insects, including both agonistic and antagonistic interaction with ecdysone receptors (EcR)13. A steroid-like hormone, ecdysone directly modifies the expression of genes controlling sexual differentiation in several insect species70,71. An alteration in ERR or EcR expression and the subsequent steroid-like hormone dysfunction could partially explain the increased female offspring production from exposed individuals, along with reduced sperm function and fertility in males directly exposed to equol. However, a change in either ERR or EcR expression does not clearly account for the degree of subfertility and skewed sex ratio observed for a further two generations. Another possible pathway via which this sex bias occurs following equol exposure may be through sex allocation of offspring by female Drosophila72. Previous studies suggest that when paired with—or at risk of producing- “low quality” males, female Drosophila are able to skew the ratio of male to female offspring produced, favouring the production of females72. Given that the present study observed that all males from equol-exposed lineages had lower reproductive fitness, maternal sex allocation may have also contributed to the female bias in offspring of the P, F1 and F2 generations.

Our study supports a growing body of evidence that EDC’s promote changes to the male germline, which may manifest as reduced reproductive function. The transgenerational impacts upon sperm functionality observed in equol-exposed Drosophila lineages corroborate with previous rodent work, whereby exposing male mice to BPA impairs spermatogenesis, sperm quality and hyper-methylates sperm DNA in both the directly exposed male and in two generations of his male descendants73. Previous studies have identified sub-fertile male phenotypes in generations with exposure to EDCs in their lineage73,74, but the molecular function of mature spermatozoa driving the observed subfertility has not been comprehensively characterised. In the present study, direct exposure to equol disrupted normal sperm function, with exposed males demonstrating increased sperm membrane lipid peroxidation and DNA fragmentation, alongside reduced fertility. These changes are similar to those observed in adult mammals exposed to phytoestrogens50,53,75,76. Across these studies, it is concluded that EDC’s impact the reproductive function of adult males by both directly interfering with the sperm plasma membrane76, by impairing sperm nuclear DNA decondensation77 and by disrupting certain stages of spermatogenesis50,53,75. Here, we show that equol mediates abnormal sperm function in directly exposed male D. melanogaster via a combination of these pathways.

While it is suspected that EDCs cause multigenerational impacts on male fertility via epigenetic pathways, it is not clear how these epigenetic modifications translate to sperm function and male fertility. Regardless of whether the male or female parent was exposed, we find that dietary equol had detrimental effects upon sperm quality and function in F1 males. Males with an exposed parent produced spermatozoa with reduced viability, lipid peroxidation, membrane lipid disorder mitochondrial membrane potential, and, if the female parent was exposed, greater protamine deficiency. Taken together, it appears that these cells exhibit a greater degree of metabolic stress, potentially indicating apoptotic-like changes. This was not entirely the case if an F1 male had both parents exposed to equol, whereby sperm still exhibited reduced viability and increased protamine deficiency but did not have the same indications of metabolic exertion. It is possible that the double route of exposure decreases sperm quality via a different mechanism, or that these cells simply die more rapidly, before cellular events leading to cell death are captured. By the F2 generation, the most detrimental changes to sperm function, including high sperm DNA fragmentation, protamine deficiency, poor viability and increased lipid peroxidation, were predominantly in lineages of female exposure. This is likely because males of the P and F1 generation presented with more severe impairments of sperm function, and were therefore less likely to successfully reproduce. Protamine deficiency in the sperm nucleus similarly indicates that this defect is carried predominantly through the female lineage, likely because sperm that are protamine deficient are less likely to successfully fertilise. As oestrogen receptors are known to regulate chromatin remodelling in the developing sperm cell78, it is possible that exposure to oestrogenic EDCs could change the expression of these receptors, which may persist across generations. This may partly explain the differences observed in protamine deficiency in males with equol exposure in their lineage. Future work should delve further into both oestrogen receptor expression following ancestral EDC exposure and the epigenetic modifications regulated by by oestrogen receptors79.

Of note, we observe that exposure to equol in the P generation downregulated several genes known to be important for male reproductive function in Drosophila, yet these changes did not persist in subsequent generations. Genes Pif1a and Tplus3B are known to promote normal spermatogenesis in Drosophila28,29 and are both down-regulated in males exposed to equol, but not their offspring. In the female, the gene eggless is responsible for the survival and differentiation of germline cells, and later in mature Drosophila oogenesis80. While we expected a downregulation of eggless expression in female Drosophila exposed to equol alongside reduced fertility, though poor fertility persists in the offspring of exposed individuals, the abnormal expression of eggless does not mirror this. This suggests that while EDC’s such as equol can reduce fertility in the exposed individual partly through modification of gene expression and the associated downstream pathways, these specific genetic changes are not heritable and is not the primary mechanism via which transgenerational subfertility occurs. As discussed above, future avenues to explore in insects may include changes to the expression of ERR or EcR, which were not included in the present study but, in hindsight, are likely related to the observed results. Previous studies in other species suggest a variety of genes may be differentially expressed following both direct and ancestral EDC exposure. In zebrafish, males exposed to EDC 2,3,7,8-Tetrachlorodibenzo-p-dioxin (TCDD) and two generations of their male descendants demonstrate a multitude of differentially expressed genes within the testes81. These include genes encoding for proteins responsible for normal spermatogenic function, synthesis of steroid hormones, lipid metabolism and metabolic disease81. A transgenerational down-regulation of genes involved in steroid hormone synthesis has also been demonstrated in atrazine-exposed male rats without any associated decrease in sperm motility82. Broadly, it appears that the impact upon gene expression, and how this translates to reproductive function, varies depending upon both the EDC, the species exposed and whether exposure initially occurred during a developmentally sensitive period, such as in utero.

We find that a common dietary endocrine disruptor, equol, could contribute to the decline in insect populations by reducing both male and female fertility Though the negative impacts of agriculture on insect biomass have previously been attributed to pesticide use, we here present evidence that a common compound within crops can impair reproduction in a transgenerational manner in an insect model. This suggests that the homogenisation of insect diet could be an equally critical factor driving population dynamics. Here we show that the inclusion of a common phytoestrogen metabolite, equol, into the diet disrupts reproductive function across three generations. Further research is needed to determine whether reproductive function is rescued after extended generational turnover, or if EDC-mediated reproductive defects represent a permanent increase to subfertility within a population.