Progeny effects of rotenone exposure depend on parental toxicity

Abstract

Parental exposure to toxicants can affect progeny health. However, laboratory studies often employ exposures that result in loading of pollutants to gametes or toxic effects to parents, which could indirectly affect germ cell or gamete health. Here, we took advantage of the biology of Caenorhabditis elegans to carry out a study in which we minimized the potential for maternal loading of toxicants, and used an exposure paradigm that either did (high concentration) or did not (low concentration) significantly impact the health of the P0 generation. We hypothesized that parental exposure to mitochondrial toxicants during germ cell and gamete development, at levels not causing P0 toxicity, would result in altered mitochondria and organismal health in offspring. In the P0 generation, a high rotenone concentration altered growth, mitochondrial respiration, gene expression, induction of the mitochondrial unfolded protein response, and susceptibility to dopaminergic neurodegeneration induced by a chemical rechallenge later in life. However, we observed minor or no effects in P0 at a low concentration. In high-exposure F1 offspring, we observed altered embryo size, larval developmental stage distribution, spare respiratory capacity, heat shock protein expression, and dopaminergic neurodegeneration after a secondary rotenone challenge. The only effects observed in the F1 offspring of the low exposure were a 1.7% decrease in egg size (size later in development was normal), and moderate evidence of a slightly increased sensitivity to heat shock protein expression and dopaminergic neurodegeneration caused by a secondary later-in-life rotenone exposure. We recommend that parental toxicity be carefully assessed to contextualize offspring outcomes.

Graphical abstract

Exposure to toxicants or other stressors prior to fertilization comprises an important category of the various “windows of vulnerability” posited to contribute to the developmental origins of health and disease (Heindel 2019). However, such exposures are less well-studied experimentally than exposures that occur after fertilization. Of studies that have been published in which the experimental exposure ended prior to fertilization, many examine the effects of exposures that result in loading of a pollutant into eggs or sperm (Yue et al. 2020; Kim et al. 2021; Annunziato et al. 2022; Marin et al. 2024), or that compromise maternal health to the point of potentially limiting provision of nutrients to the offspring or altering the gestational environment (Ma et al. 2019; Yue et al. 2020; Kim et al. 2021). However, there is reason to suspect that exposures that do not cause significant loading of pollutants or compromised parental health could also affect offspring. Reasons include evidence of the effects of prior exposure on epigenetic patterning in sperm that are unlikely to transfer significant amounts of toxicant (Stermer et al. 2019; Schrott et al. 2020) and observations that exposure effects can persist for three or more generations in the absence of continued exposure (Camacho et al. 2018).

Mitochondria are an important target of pollutants, toxins, and drugs (Meyer et al. 2013). High-throughput screens have shown that 1% to 15% of tested chemicals primarily affect mitochondria (Attene-Ramos et al. 2013; Wills et al. 2013; Attene-Ramos et al. 2015; Datta et al. 2016), with one screen reporting ∼30% (Wills et al. 2015). Parental exposure to mitochondrial toxicants can result in effects in offspring (Divi et al. 2010; Ditzel et al. 2016; Kishimoto et al. 2017; Lozoya et al. 2020; Kozal et al. 2023), as can early life starvation (Jobson et al. 2015) and caloric restriction (Hibshman et al. 2016), which also affect mitochondrial function. However, we are not aware of studies designed to specifically test the impacts of mitochondrial toxicity that occurs prior to fertilization on offspring. This is an important knowledge gap because the basic biology of mitochondria in germ cells and gametes suggests the potential for significant sensitivity at this stage. Mitochondria undergo biogenesis and major functional changes during germ cell proliferation and development (Stewart and Larsson 2014; Bahety et al. 2024), and epigenetic reprogramming occurs at the same time (Dahl et al. 2016; Zhang et al. 2016).

Energy metabolism is largely glycolytic in germ cells, oocytes, and sperm, but there is evidence that oxidative phosphorylation is also important in oocytes (Ramalho-Santos et al. 2009; Van Blerkom 2011) and sperm (Ruiz-Pesini et al. 2007). This indicates that toxicant impacts on mitochondrial respiration could be functionally important. Furthermore, mitochondria are not quiescent in these cells. Mitochondria (and mitochondrial DNA, or mtDNA) undergo dramatic expansion during germ cell and gamete development (Jansen 2000; Motta et al. 2000; Shoubridge and Wai 2007). Reactive oxygen species (ROS) play important signaling roles in male and female gametes (Ramalho-Santos et al. 2009; Van Blerkom 2011). These ROS signaling roles may be susceptible to interference by exposures that result in elevated ROS, as occurs with many mitochondrial toxicants, including those we studied here. There is strong evidence that mitochondrial function is critical in both oocyte (Babayev and Seli 2015) and sperm (Pena et al. 2009) for successful reproduction, and mitochondrial-mediated apoptosis affects oocyte survival (Tiwari et al. 2015).

There is also evidence that chemical exposures of germ cells and gametes may lead to altered epigenetic patterns (Perera and Herbstman 2011; Pacchierotti and Spano 2015; Wu et al. 2015). Epigenetic programming in germ cells and gametes may be particularly susceptible to environmental influences (Ly et al. 2015) because of the dynamic nature of the process of replacing histone proteins throughout specific portions of the genome with histone variants and the fluctuation in the histone modifications that occur during this period (Verdikt et al. 2023). These epigenetic patterns are important for regulating transcriptional activity in the parental gametes (Arico et al. 2011) and in preparing the germ cells for fertilization (Samson et al. 2014). Environmental exposures could conceivably shift the fidelity of this process, skewing the normal patterns in a manner that is carried forward into the next generation and affects phenotype. Finally, mitochondrial dysfunction (Weinhouse 2017) or redox stress (Weinhouse 2021) that often results from mitochondrial dysfunction, can both affect epigenetic patterning.

We hypothesized that mitotoxicant exposure during germ cell and gamete development, at levels not causing frank organismal toxicity in the parental generation, would impact mitochondrial and/or epigenetic processes in germ cells or gametes prior to fertilization. Furthermore, we proposed that this programming, along with associated mitochondrial alterations, would persist into adulthood and influence the phenotype of the offspring. To test this hypothesis, we used the model organism Caenorhabditis elegans, a powerful model for environmental toxicology (Leung et al. 2008) and environmental epigenetics (Kelly 2014; Weinhouse et al. 2018; Filipowicz and Allard 2025). C. elegans is typically a hermaphrodite with a small percentage (∼0.1% of the population) of males and a rapid and highly predictable developmental pattern. This allowed us to carry out exposures during early germ cell proliferation and sperm development, but prior to the production of eggs and fertilization, as detailed in “Establishment of exposure timing and concentrations” section. Another advantage of C. elegans’ reproductive biology for this purpose is that although eggs are fertilized internally, they are laid in under 3 h, around the 30-cell stage. This helped us to minimize potential impacts of exposure-mediated changes to maternal physiology which could alter the gestational environment, indirectly changing offspring phenotype.

We previously found that C. elegans exposed during larval development to a relatively high concentration of rotenone, an inhibitor of Complex I of the electron transport chain, resulted in growth delay, impaired mitochondrial bioenergetics, and altered susceptibility to pathogens in the parental (P0) generation (Mello et al. 2022). Rotenone is best known in the context of human health for its association with Parkinson’s disease (Tanner et al. 2011). Here, we report on additional effects of rotenone exposure on the parental generation, including a lower-concentration exposure. We also tested for effects of those two parental exposure concentrations on the offspring F1 generation. We used exposure concentrations carefully selected to either cause or not cause significant stress to the P0 generation. In some cases, we also tested the effects of parental exposure to the Complex III inhibitors Antimycin A and pyraclostrobin, and used a secondary later-in-life “rechallenge” to test for latent changes that might only manifest upon further stress.

Materials and methods

C. elegans strain and maintenance

N2 Bristol and SJ4100 (zcIs13[hsp-6p::GFP]) strains were obtained from the Caenorhabditis Genetics Center (CGC). BY200[vtIs1(dat-1p::GFP; rol-6)] was obtained from Michael Aschner. Large populations of worms were generated for egg isolation (described below) by culturing them on K agar plates (Williams and Dusenbery 1988) with Escherichia coli OP50 at 20°C unless otherwise noted. Liquid culture occurred in complete “K+” medium, which is K medium (Williams and Dusenbery 1988) plus 5 mg/l cholesterol, 3 mM MgSO₄, and 3 mM CaCl₂.

Experimental design

The P0 generation was exposed to rotenone or other chemicals as previously described (Mello et al. 2022). Briefly, gravid adult C. elegans were treated with a sodium hypochlorite solution to obtain embryos (Lewis and Fleming 1995). Embryos were placed in liquid solution containing HB101 bacteria (OD₆₀₀ = 2), K+ medium, and DMSO vehicle or chemical (rotenone, antimycin A, or pyraclostrobin) exposure, at a density of 5 embryos/100 µl unless otherwise stated. We chose this concentration of HB101 (OD₆₀₀ = 2) because it optimized larval growth at 108 h from sodium hypochlorite treatment, while avoiding egg retention that occurs at higher concentrations of bacteria (Hibshman et al. 2016) (Fig. S1A). Nematodes were maintained in these conditions until they were mid-L4 larval stage as determined by vulval development. The L4 stage is the final pre-adult larval stage. This allowed for exposure to occur throughout development but prior to oogenesis (Hubbard and Greenstein 2005) and therefore prior to fertilization. P0 worms were then washed and placed onto K agar plates with OP50 bacteria and without chemicals for 48 h, to allow worms to depurate chemicals and begin egg laying. After the washout period, worms were again treated with hypochlorite solution to obtain F1 embryos, which were placed in the same liquid conditions as the P0 with only HB101 and K+ medium (no rechallenge treatment or DMSO) or with DMSO/chemical in the case of the “rechallenge” treatments. Worms remained in liquid conditions until they reached the mid-L4 larval stage, then washed at least three times to remove bacteria/chemicals by centrifuging at 2200 rcf for 1 min and removing supernatant. Figure 1A presents this experimental design schematically.

Fig. 1.

Effects of rotenone exposure on larval size of exposed generation, and egg and L4 size of offspring. A) Experimental design schematic. Created in BioRender Mello, D. (2026). B) Length of P0 worms at the L4 stage (defined based on vulval development), corresponding to 52 for DMSO-only and 0.03 µM rotenone, and 74 h for 0.5 µM rotenone. C) Size of F1 eggs after parental exposures. D) Length of F1 worms at the L4 stage (52 h post-egg isolation). Bars and error bars represent mean ± standard error of the mean. For B and D, there were 9 to 11 biological replicates with 34 to 541 individuals per replicate; points represent the average for each replicate. For C, there were 3 to 6 biological replicates with 17 to 54 embryos per replicate; points represent each embryo. * P < 0.05, 1-way ANOVA followed by Dunnett’s test for multiple comparisons.

Chemical exposures

Early-life exposures

P0 worms were exposed in K+ medium to 0.25% DMSO (control), 0.03 µM, or 0.5 µM of rotenone (CAS Number: 83-79-4; R8875; Sigma) in final DMSO concentrations of 0.25%, from eggs until the L4 larval stage. Rotenone doses were chosen based on initial growth inhibition range-finding experiments, as described in Results. Gravid adults were treated with hypochlorite solution to obtain synchronized embryos. 1 mM rotenone stocks were made in 100% DMSO and stored in the −80°C freezer for no more than one month before being replaced. Rotenone exposures were carried out in glass containers to minimize possible sorption on plastic. The half-life of rotenone in C. elegans has not been reported, but in water it is as short as 12 h up to a few days (Finlayson et al. 2014 and references therein) and likely to be much shorter given metabolic breakdown by worms and their bacterial food, as well as potential sorption processes. Therefore, to maintain the rotenone concentration throughout the exposure, re-exposure occurred every 24 h, which might have caused some variability in DMSO. Without taking into account any breakdown or other loss, our final DMSO concentrations would have been 0.125% after the first addition (0 h), 0.25% after the second (24 h), and 0.38% after the third (48 h; only for the highest 0.5 µM rotenone exposure). DMSO concentrations in all developmental exposures were below 0.5%, which does not affect developmental rate, physiological rates, or lifespan (AlOkda and Van Raamsdonk 2022).

During P0 generation exposures, the highest concentration (0.5 µM) caused a significant growth delay. To ensure that any difference observed was the result of rotenone exposures, rather than rotenone-mediated developmental delay, worms exposed to the highest concentration remained in liquid culture with rotenone longer, until they reached the L4 larval stage, assessed as described below. To reach the L4 larval stage, worms in control and 0.03 µM solutions were exposed for 52 h, while worms in 0.5 µM rotenone were exposed for 74 h. Initial staging experiments showed that this resulted in the worms in all groups reaching a median L4.6 stage of development. The roughly mid-point stage of L4 development was chosen to be well-removed from the L3/L4 molt, to avoid the large molt-associated changes that occur in many endpoints, including size and oxygen consumption.

Antimycin A (CAS Number: 1397-94-0; A8674; Sigma) and pyraclostrobin (CAS Number: 175013-18-0; 33696; Sigma) exposures were identical to rotenone exposures except that there was no re-exposure: both were added only once with 0.25% DMSO, immediately after egg isolation.

Rechallenge exposures

F1 eggs were “rechallenged” with developmental exposure to 0.5 µM rotenone, 0.75 µM Antimycin A, or 50 µM pyraclostrobin for 52 h, also in K+ medium as described above.

For the neuron scoring, a special later-life rechallenge protocol was performed. Four days after P0 and F1 worms reached the L4 stage (day 4 adults), P0 and F1 worms were exposed to a high concentration of rotenone (25 µM) previously identified as causing neurodegeneration in adults (Jessica Hartman and Joel Meyer, personal communication) or 1% DMSO for 48 h in 96-well microplates. 1% DMSO is an acceptable concentration for adult toxicity studies (Boyd et al. 2012).

Worm size measurements

Embryo size was assessed by measuring the area, which was calculated using microscopy and a micrometer as previously described (Hibshman et al. 2016). Worm length and volume were measured after rinsing worms with K medium to remove all bacteria and rotenone. Rinsed worms were then plated on unseeded 10 cm K agar plates without peptone. We assessed the size of larvae and adults with both length and volume measurements. These are generally correlated (eg Yang et al. 2012) but one can increase disproportionately to the other. For instance, some chemical exposures result in mild length decreases but large volume decreases, indicating “thin” worms, whereas others greatly decrease length but have less effect on volume, which often corresponds to increased lipid deposition. Therefore, we chose to collect both measurements, presenting some in the main figures and some in the Supplementary Material. For all lifestages, images were taken with a Keyence BZ-X710 microscope and images were analyzed using the Wormsizer plugin in FIJI (Moore et al. 2013).

Developmental staging

Since we observed a significant delay with the highest rotenone dose (0.5 µM), we wanted to accurately stage-match worms to ensure that measured parameters did not appear to be different simply because the worms were at different developmental stages (eg mitochondrial respiration changes dramatically with stage: Mello et al. (2024)). To do this with high precision, we used the vulval development scoring system developed by Mok et al. (2015) with a few alterations. Briefly, worms were removed from bacteria and rotenone exposures and paralyzed with 50 mM sodium azide. This allowed for vulval morphology to be examined using the Keyence BZ-X710 microscope with a 40× objective. Based on the nine developmental stages determined by Mok et al., worms were assigned a stage on a scale from 4.0 to 4.9.

ATP quantification

Once nematodes reached the L4 stage and were rinsed thoroughly to remove bacteria and rotenone, 100 worms were aliquoted in 100 µl of K medium into an Eppendorf tube, flash frozen in liquid nitrogen, and stored at −80°C. To measure ATP levels, frozen samples were placed on a 95°C heat block for 15 min, placed on ice for 5 min, and centrifuged for 10 min at 14,800 g at 4°C. Samples were diluted by adding 100 µl of DI water to 20 µl of supernatant and run in duplicate. ATP was measured using the CellTiter-Glo Luminescent Cell Viability Assay (Promega G7572) as described (Bailey et al. 2016). ATP levels were normalized based on worm volume.

DNA copy number

We measured mitochondrial and nuclear copy number in P0 and F1 nematodes at the L4 stage using real-time quantitative PCR as described (Leuthner et al. 2021). Briefly, six nematodes per treatment were picked into PCR tubes, in triplicate, with 90 µl of lysis buffer and stored at −80°C until analysis. Samples were lysed at 65°C for 1 h using a thermocycler, and 2 µl of the lysate was used as template with Power SYBR Green PCR Master Mix (ThermoFisher Scientific, Waltham, Massachusetts) for real-time PCR. Ct values were converted to genome copy number using a plasmid DNA-based standard curve.

Oxygen consumption

Exposed and control P0 worms and their progeny were harvested at the L4 stage and transferred into 15 ml centrifuge tubes and washed to remove bacteria (and rotenone in the case of P0s). Respiration measurements were performed according to Mello et al. (2024). Once worms were sufficiently washed, 75 worms were pipetted in K medium into a 24-well plate with a final volume of 525 µl. Basal respiration was measured first, followed by an injection of FCCP for a final concentration of 25 µM or DCCD for a final concentration of 40 µM. Lastly, a final injection of sodium azide with a final concentration of 80 mM was used to measure non-mitochondrial respiration. After measurement was complete, worms were manually counted in each well for normalization.

RNA sequencing experimental design

In order to obtain approximately 750,000 synchronized eggs for all treatments for RNA sequencing, synchronized eggs were first obtained from gravid adults grown on K agar plates with OP50 bacteria through sodium hypochlorite treatment. Approximately 90,000 eggs were grown in liquid culture using HB101 bacteria (OD₆₀₀ = 4) and K+ medium at a density of 1 egg/µl. After 96 h, the gravid adults were treated with sodium hypochlorite to obtain approximately 900,000 synchronized eggs for the P0 generation. The P0 eggs were cultured in 50 ml Erlenmeyer flasks with the HB101 bacteria (OD₆₀₀ = 2) and K+ medium at a density of 1 egg/µl. The flasks were then treated with either 0.25% DMSO (control), 0.03 µM rotenone, or 0.5 µM rotenone and grown until the L4 larval stage.

Once the worms reached the L4 larval stage, a subsample was taken from each treatment for RNA sequencing and transferred to a 15-ml centrifuge tube and rinsed to remove bacteria and rotenone through a series of 1-min centrifugation steps at 2,200 RCF and rinsing with K medium. Samples for RNA sequencing were then transferred in 100 µl of K medium to Eppendorf tubes and stored at −80°C.

An aliquot of P0 worms was transferred from the liquid culture to obtain the F1 generation. P0 worms were rinsed thoroughly from bacteria and rotenone and placed in liquid culture with HB101 bacteria (OD₆₀₀ = 2) and K+ medium for 48 h to depurate. After 48 h, synchronized F1s were obtained through sodium hypochlorite treatment. The F1s were cultured in the same conditions as the P0 except no DMSO or rotenone were added.

RNA sequencing

RNA sequencing and analysis methodology were previously described (Mello et al. 2022). A portion of the transcriptomic data, specifically that of the P0 generation 0.5 μM-exposed worms, was published in Mello et al. (2022).

Mitochondrial unfolded protein response quantification

To measure mitochondrial unfolded protein response, the C. elegans reporter strain SJ4100 was cultured as described above and at 52 (P0 0.25% DMSO or 0.03 µM, all groups in the F1 generation, including the rechallenge experiment) or 74 h (P0 0.5 µM rotenone group) transferred to 96-well microplates (μClear, Greiner Bio-One) containing 100 µl of K-medium. Nematodes were paralyzed by adding 60 mM (final well concentration) sodium azide until paralyzed before imaging using the Keyence BZ-X700 microscope. Fluorescence was quantified using Fiji/ImageJ by manually outlining each worm with the polygon selection tool. Mean gray values are reported after correcting for background (area outside the worm).

Neuron scoring

Worms were collected at day 6 of adulthood (i.e. 6 d post-L4) in a 15 ml conical tube and washed three times with K medium. Worms were transferred to a glass slide with a 2% agarose pad and paralyzed with 5 to 10 µl of 100 mM sodium azide (Sigma-Aldrich). The glass slide was mounted for fluorescence imaging on a Keyence BZ-X710 microscope equipped with a BZ-X700E light source. Z-stacks were acquired for the head of 20 individual worms using a 40× objective with 200 ms of exposure. Maximum projections of these Z-stacks were generated, and each dendrite of the cephalic dopaminergic neurons was scored on a scale of 0 to 6 as previously described (Bijwadia et al. 2021): 0—no damage, 1—irregularities (kinks), 2—less than 5 blebs, 3—5 to 10 blebs, 4—more than 10 blebs or minimal number of breaks, 5—multiple breaks or 25 to 75% dendrite loss, and 6—more than 75% dendrite loss. Scorings were registered per dendrite (up to 4 dendrites per worm) and per worm, identifiably grouping sets of dendrites to individual worms.

Statistics

For most data, we used analysis of variance (ANOVA) followed by Dunnett’s or Tukey’s tests for multiple comparisons. For comparisons of the larval stage, we used a linear mixed effects model. For analysis of neurodegeneration, we used an order-restricted Bayesian ordinal mixed effects model. Details of statistical analyses, sample sizes, and statistical results are described in the figure legends or manuscript text. All statistical analyses were done in RStudio (Rstudio Team 2025).

Results

Establishment of exposure timing and concentrations

Our goal was to test the effects on offspring of parental exposures that were terminated prior to fertilization. We began by defining the exposure timeline and concentrations to minimize any post-fertilization chemical exposure and to cause either minimal or significant growth delay in the parental generation, in order to test whether any effects in progeny would depend on the presence of generalized toxicity in the parental generation. The C. elegans germ line is established in the early embryo, when the first primordial germ cell is specified. During gastrulation, this cell divides once (Sulston et al. 1983). The next division does not occur until halfway through the first larval stage (L1), and then the number of germ cells increases exponentially for the first two larval stages, with no meiotic divisions until the fourth (L4) stage. Most individual C. elegans are hermaphrodites, with a low (∼1/1,000) spontaneous occurrence of males. In hermaphrodites, a limited number of male gametes are produced during the final larval (L4) stage, followed by female gametes, which are produced beginning in young adulthood and continuing for several days (Hubbard and Greenstein 2005). Fertilization occurs in adults. Therefore, to target germ cells throughout proliferation and into the development of gametes, but prior to fertilization, we carried out exposures from the L1 to the L4 larval stages in what we refer to as the P0 generation. We then waited 48 h to permit depuration of most rotenone, in order to minimize any direct post-fertilization exposure of the F1 generation to rotenone (see “Chemical exposures” section), and we collected eggs in order to measure outcomes in the F1 offspring (Fig. 1A). We identified two concentrations of rotenone to test in this paradigm: 0.03 and 0.5 µM. These led to roughly 10% and 50% decreases in growth over 52 h in the P0 generation (Fig. S1B). We previously reported that in this exposure paradigm, these concentrations did not affect the number of offspring [(Supplementary Figure 1 in Mello et al. (2022)]. We chose these concentrations to include a concentration that had a minimal effect on gross organismal development (0.03 µM), such that any effects observed on offspring would be unlikely to be attributable to frank toxicity in parents, and a concentration that had a large effect on parental growth (0.5 µM). We reasoned that the high concentration exposure would facilitate comparisons to previous studies that have often shown significant maternal toxicity, and increase the likelihood of observing an effect, such that if we failed to observe any effects on offspring, that observation would be more convincingly negative. Because the 50% delay in growth at the high concentration also delayed egg laying, we carried out rotenone exposures in P0 not for a fixed amount of time post-egg isolation, but until the time when most worms had reached the L4 stage as defined by vulval development (Mok et al. 2015), which was 74 h after eggs were isolated. The rotenone exposure led to slightly smaller P0 L4 animals even when stage-matching them as closely as possible to controls (Fig. 1B; volume measurements in Fig. S1C). Perfect stage-matching was not possible because at 0.5 µM rotenone, the distribution of developmental stages increased (Fig. S2). Therefore, if exposure occurred longer, we would have risked having exposure occur after fertilization in a subset of more-developed individuals. As described in “Chemical exposures” section, the time required for development to the L4 stage was very similar for all F1 offspring groups, so 52 h post-egg isolation was used to define mid-L4.

We measured most outcomes in both the F1 generation and the P0 generation. Results in the F1 generation served to test for parental exposure effects, while P0 generation results served both to ascertain the effects of the exposures on the parents, which might affect offspring, and as a basis for interpreting the magnitude of any effects observed in offspring.

Low and high-concentration rotenone exposure caused small, statistically significant changes in F1 embryo size and larval stage distribution

Growth inhibition is a common effect of toxic exposures. In C. elegans, mitochondrial dysfunction specifically inhibits larval development (Tsang et al. 2001), and poor environmental conditions in adults can reduce egg size (Hibshman et al. 2016). After parental exposure, F1 embryos were slightly smaller at both rotenone concentrations (1.7% and 5.7% at the 0.03 and 0.5 µM parental exposure concentrations; Fig. 1C). However, at 52 h post-hatch, F1 offspring of 0.03 µM-exposed P0 worms were the same size (Fig. 1D, Fig. S1C) and showed the same larval stage distribution as control offspring (Fig. S2), while F1 offspring of 0.5 µM-exposed worms were the same size (Fig. 1D, Fig. S1C) but showed an altered larval stage distribution (Fig. S2).

High-concentration rotenone exposure reduced F1 generation spare respiratory capacity

Rotenone is a relatively specific inhibitor of Complex I of the electron chain, and we hypothesized that parental exposure could lead to altered mitochondrial function including metabolic adaptations (Gonzalez-Hunt et al. 2021) or changes to mitochondrial DNA copy number (Meyer et al. 2017) in offspring. We observed no rotenone-related differences in steady-state ATP levels or mitochondrial DNA copy number in either the P0 or F1 generation (Fig. 2A and B). As expected, given rotenone’s mechanism of action, we did observe decreased total basal, mitochondrial basal, and ATP-linked respiration in the P0 generation (Fig. 2C). We did not detect changes to these parameters in the F1 generation, but we observed reduced spare respiratory capacity in the highest rotenone dose compared to the controls (Fig. 2C).

Fig. 2.

ATP levels, mitochondrial: nuclear DNA ratios, and oxygen consumption rates in rotenone-exposed P0 worms and their progeny. A) Whole body ATP measurements, n = 5 to 6 biological replicates/group; points represent the average for each replicate. B) Ratio of mitochondrial DNA to nuclear DNA copy number; n = 5 to 8 biological replicates; points represent the average for each replicate. C) Oxygen consumption measurements; 2 to 9 biological replicates; points represent technical replicates. Bars and error bars represent mean ± standard error of the mean. An initial 3-way ANOVA showed statistically significant interaction between treatment level, generation, and OCR measurement (total basal, mitochondrial basal, etc.). Subsequently, we analyzed the effect of rotenone treatment on each form of OCR for each generation separately. * P < 0.05, 1-way ANOVA followed by Dunnett’s test for multiple comparisons.

No detectable effect of parental rotenone exposure on progeny transcriptomics

We considered that more subtle changes resulting from parental rotenone exposure might be detectable using transcriptomic analyses. We carried out RNASeq experiments on control, 0.03 µM-exposed, and 0.5 µM-exposed P0 generation and F1 generation worms collected at the L4 stage. We detected a strong transcriptomic signal in the P0 generation 0.5 µM-exposed worms, with 180 differentially expressed genes (FDR < 0.05), which was previously published (Mello et al. 2022). Remarkably, however, we detected zero differentially expressed genes in the 0.03 µM-exposed P0 worms compared to DMSO-only P0 worms (FDR = 1 for every gene), F1 offspring of 0.03 µM-exposed compared to F1 offspring of DMSO-only P0 worms (FDR = 1 for every gene), or F1 offspring of 0.5 µM-exposed compared to F1 offspring of DMSO-only P0 worms (FDR >0.6 for every gene) (Data File S1). Plots of fold-change versus mRNA counts illustrate the same point (Fig. 3A and B and Fig. S3A and B). Principal components analysis (Fig. 3C) also shows that only the 0.5 µM-exposed P0 groups cluster separately from other groups, and that separation is driven by PC2, which only explains 11% of the observed variance.

Fig. 3.

Parental rotenone exposure does not alter the transcriptome in offspring. A) 180 genes (out of 14,426) were differentially expressed (FDR < 0.05) in P0 L4s after exposure to 0.5 µM rotenone compared to DMSO, but none (FDR = 1 in all cases) were observed after 0.03 µM rotenone. Log fold-change versus log counts per million. B) No genes were differentially expressed in F1 L4s after parental exposure to 0.03 or 0.5 µM rotenone compared to DMSO. C). There is no clustering detectable for gene expression in F1 L4s after parental exposure to 0.03 or 0.5 µM rotenone compared to DMSO. Only P0 L4s exposed to 0.5 µM rotenone show any separation, but 95% confidence intervals overlap most other groups, even for P0 L4s exposed to 0.5 µM. D) This PCA plot shows the same PCA analysis as panel C, but experimental replicates are identified, and each sample is labeled with the hours per hatch of that sample as calculated by RAPToR. PC1 appears to reflect hours from hatch as estimated by RAPToR.

We wondered what might explain the 47% of variance captured in PC1, since it was not treatment- or generation-related. When we looked closely at the individual replicates, we found that PC1 appears to be driven largely by experimental replicate, with replicates one and two on the left, three and four on the right (Fig. 3D). This separation among replicates appears to be attributable at least in part by slight differences in the average developmental stage at which each replicate was harvested, despite our extensive efforts to create a highly precise staging protocol. We measured size (length and volume) and larval developmental stage for all replicates. Invariant somatic cell development in this species, combined with existing very high-resolution RNASeq developmental datasets, allows the average age of worm populations to be calculated from RNASeq datasets using the RAPToR tool (Bulteau and Francesconi 2022). We found that PC1 reflected hours from hatch as estimated by RAPToR (Fig. 3D), with replicates one and two on average larger and having a more advanced distribution of developmental stages. The same was true of worm size (Fig. S4A) and developmental stage (Fig. S4B).

In this subset of experiments, across all four replicates, the average size (Fig. S5A and 5B), stage based on vulval development (Fig. S5C), and age based on transcriptomics (Fig. S5D) indicated that F1 offspring of 0.5 µM-exposed P0 were smaller and younger than offspring of control and 0.03 µM-exposed P0s.

Effects of parental exposure on sensitivity to secondary mitotoxicant challenge in F1s

We next tested the possibility that offspring of exposed parents might harbor latent changes that would be uncovered upon stressor challenge. The stressor challenges that we used in F1 offspring were exposure to high concentrations of rotenone or the Complex III inhibitors Antimycin A and pyraclostrobin. We measured F1 sensitivity in the form of three mitochondrial stress-induced phenotypes previously characterized in C. elegans: growth delay (Tsang et al. 2001), induction of the mitochondrial unfolded protein response (mtUPR) (Nargund et al. 2012), and dopaminergic neurodegeneration (Morton et al. 2025). To be able to carry out roughly equitoxic exposures using Antimycin A and pyraclostrobin (i.e. concentrations that would inhibit growth very mildly or very strongly, as we did with rotenone), we first characterized growth inhibition by pyraclostrobin and Antimycin A after 52 h exposure following egg isolation (Fig. S6A and B). We found that 0.088 and 0.75 µM Antimycin A inhibited growth by roughly 10% and 50% at 52 h, comparable to 0.03 and 0.5 µM rotenone. 50 µM pyraclostrobin caused a 10% decrease, but due to limited solubility, we were unable to achieve a 50% decrease with pyraclostrobin. Because the high concentration of Antimycin A led to significant growth delay, we again stage-synchronized to permit comparable development by allowing an extra 16 h to develop; this again led to a widening of variability in larval stages (Fig. S6C). We found that parental exposure to 0.088, 0.75 µM Antimycin A, or 50 µM pyraclostrobin without subsequent challenge had either undetectable or very small effects on offspring size at 52 h (Fig. S6D), as we previously observed with rotenone (Fig. 1D, Fig. S5).

In our first rechallenge experiment, we measured the size at 52 h of F1 offspring in the context of secondary rechallenge during development at the higher concentration of each chemical (0.5 µM rotenone, 0.75 µM Antimycin A, and 50 µM pyraclostrobin). Parental exposures had no detectable effect on the degree of growth inhibition caused by the rechallenge (Fig. 4A).

Fig. 4.

Minor effects of parental mitotoxicant exposure on growth and mtUPR response upon re-challenge in F1 progeny. A) Parental exposure to rotenone, antimycin A, and pyraclostrobin had no effect on the degree of growth inhibition in progeny caused by rechallenged exposure to 0.5 µM rotenone, 0.75 µM Antimycin A, and 50 µM pyraclostrobin. n = 3 to 4 biological replicates; points represent average of each replicate. B) hsp-6p::GFP fluorescence in P0 worms and their progeny at the L4 stage and with F1 progeny rechallenged with 0.5 µM rotenone. n = 2 to 4 biological replicates; points represent each worm. Bars and error bars represent mean ± standard error of the mean. * P < 0.05, 1-way ANOVA followed by Dunnett’s test for multiple comparisons.

Next, we examined the effect of rotenone P0 exposure and rotenone F1 rechallenge on mtUPR induction by using worms containing a hsp-6p::GFP reporter construct. Rotenone induced a robust mtUPR induction in the P0 generation (Fig. 4B, P0 Not Rechallenged), but not in offspring, although a small but statistically significant decrease in GFP fluorescence was observed in the offspring from the higher exposure group (Fig. 4B, F1 Not Rechallenged). Next, we tested whether parental exposure would alter hsp-6p::GFP induction upon re-exposure to the highest dose of rotenone (0.5 µM) in offspring. We saw a small, statistically significant increase in GFP induction by 0.03 µM but not 0.5 µM rotenone in progeny (Fig. 4B, Rechallenged-Rotenone).

Finally, we tested dopaminergic neurodegeneration on day 6 of adulthood. By this time, worms are past their peak reproductive phase and may start showing signs of aging. The exposure and imaging timeline is shown schematically in Fig. S7A. First, we exposed the P0 generation to rotenone through the L4 stage, as in all previous experiments, except that the 0.5 µM exposure was begun 20 h ahead of time so that they would reach the L4 stage at the same time as the other groups. As before (Fig. 1), this resulted in smaller L4s in the P0 but not F1 generations (Fig. S7B). Day 4 adults were either exposed or not to a late-life high-concentration rotenone rechallenge (25 µM), and neuronal integrity was assessed 48 h later, on day 6 of adulthood. Representative images of dopaminergic neurons with and without damage are shown in Fig. 5A. We did not detect significant dopaminergic neurodegeneration in the P0 generation of worms submitted to the developmental exposure only (Fig. 5B, P0 not rechallenged). Nonetheless, the later-life 25 µM rotenone rechallenge did lead to neurodegeneration in the worms that had been developmentally exposed to rotenone (Fig. 5B, P0 rechallenged), but not in the worms that had not been exposed during development. Thus, although the developmental exposure did not by itself lead to dopaminergic neurodegeneration, it did sensitize those worms to a later-life rechallenge. We did not observe dopaminergic neurodegeneration in the F1 generation as a result of parental-only P0 developmental exposure (Fig. 5B, F1 not rechallenged), but did observe neurodegeneration when we exposed those worms to 25 µM rotenone later in life (Fig. 5B, F1 rechallenged), with higher levels of neurodegeneration in the offspring of both 0.03 and 0.5 µM rotenone-exposed groups compared to non-rechallenged F1s. Compared to offspring of control P0s, we found strong evidence of higher levels of neurodegeneration in the offspring of 0.5 µM rotenone-exposed adults, and moderate evidence of higher levels of neurodegeneration in the offspring of 0.03 µM adults. Finally, interestingly, we observed significant interindividual variability in response, with some worms showing significant damage and some showing no damage in the highest exposure groups; furthermore, worms that had one damaged neuron were likely to have additional damaged neurons. We assessed these effects statistically by constructing an order-restricted Bayesian ordinal mixed effects model. We found that the posterior probabilities that there was no difference between the control and 0.03 µM exposure, no difference between the control and 0.5 µM exposure, and no difference between the 0.03 µM exposure and 0.5 µM exposure in the re-challenged P0 group were approximately 0.302, <0.001, and 0.036, respectively. Similarly, these probabilities were approximately 0.020, <0.001, and 0.304, respectively, in the re-challenged F1 group. In the groups that were not rechallenged, we found no meaningful difference between the control and 0.5 µM group for generation P0 with this probability at 0.436. To assess whether there were dopaminergic neurodegenerative differences in worms after being rechallenged, we constructed 95% credible intervals for the differences between the control and rechallenge groups at each exposure level within in each generation. We found that the interval at the 0.5 µM level in the P0 generation and the interval at the 0.03 µM level in the F1 generation did not contain 0, suggesting a difference between these groups. However, this was not observed at the other levels. Additionally, the posterior median of the intra-class correlation coefficient of the worm effect was approximately 0.67 [95% posterior credible interval (0.61, 0.72)], indicating that a moderate proportion of the total variability in the dopaminergic neurodegeneration is due to dependence among neurons within the same worm.

Fig. 5.

Neurodegeneration with and without re-challenge in progeny after parental exposure to rotenone. A) Representative images of neuron scoring. + indicates kinks, * blebs, and ^ breaks within the neuron. B) No dopaminergic neurodegeneration is observed as a result of developmental (through L4) rotenone exposure in the P0 generation. Exposure to 25 µM rotenone in adults does not cause detectable neurodegeneration unless they were previously exposed to rotenone during development. No dopaminergic neurodegeneration is observed as a result of parental rotenone exposure in the F1 generation. Exposure to 25 µM rotenone in adults led to increased neurodegeneration in the offspring of the 0.5 µM-exposed adults, compared to the offspring of unexposed P0 generation adults. There was moderate evidence of increased neurodegeneration in the offspring of the 0.03 µM-exposed adults after rechallenge, compared to the offspring of unexposed P0 generation adults after rechallenge. Additional comparisons and statistical analysis are provided in the text.

Discussion

In this study, we found that parental exposure to rotenone ending before fertilization resulted in a small number of subtle effects in offspring. The same was true of the more limited outcomes we examined after parental exposure to antimycin A and pyraclostrobin. Of the effects that we did observe in F1s (embryo size, altered larval stage distribution at 52 h, spare respiratory capacity, hsp-6p::GFP expression in non-rechallenged F1s, and dopaminergic neurodegeneration after rechallenge), most were statistically significant only in the offspring of the higher-concentration rotenone exposure. Because that concentration resulted in a significant delay in P0 growth and smaller egg size, it is possible that those outcomes could be caused by poorer health in exposed P0 generation adults, potentially via production of less well-provisioned eggs or other mechanisms (Hibshman et al. 2016; Perez et al. 2017). The three phenotypes that were statistically different in offspring of the 0.03 µM rotenone exposure, which had mild if any effects on the P0 generation, were a 1.7% decrease in egg size, a slightly increased hsp-6::GFP induction upon rechallenge, and, possibly, an increased sensitivity to secondary (“rechallenge”) rotenone-mediated neurodegeneration.

The reduced size in F1 eggs from 0.03 µM-exposed P0s did not persist until L4, suggesting that this degree of egg size decrease was one from which the worms were able to recover without noticeable effects on most of the other phenotypes that we assessed, including size at L4. However, the effects of parental rotenone exposure on offspring dopaminergic neurodegeneration are particularly intriguing. F1 offspring were sensitized to rechallenge exposure at both concentrations, perhaps reflecting the fact that dopaminergic neurons are especially vulnerable to alterations in mitochondrial function (Sherer et al. 2002). The results were also intriguing because this was our only cell type-specific assessment. It is possible that other cell type-specific phenotypes occurred but were not observable because our other measurements were made at the whole-organism level. For example, if oxygen consumption were affected only in a few cells, we might not detect this with whole-worm respirometry.

Even in the case of the F1 offspring of the 0.5 µM-exposed P0s, in most cases we saw either no effects or relatively mild effects. This raises the possibility that some of those differences were the result of random chance, despite relatively large sample sizes. The lack of any altered RNA levels in F1 offspring (from either parental exposure concentration) is a compelling negative outcome, particularly in the context of the “Transcriptomic Point of Departure” approach that suggests that exposures that occur at concentrations that do not affect transcriptomics are not toxic (Thomas et al. 2011; Johnson et al. 2022). A caveat to comparing the RNASeq outcome to other outcomes is that we used a higher worm density for the RNASeq experiments than other experiments, to obtain sufficient biological material.

There are a number of papers reporting effects of pre-fertilization exposures (Yue et al. 2020; Kim et al. 2021; Oluwayiose et al. 2021; Annunziato et al. 2022; Schrott et al. 2022; Marin et al. 2024; Li et al. 2025), and all described more and larger effects than we saw. What might explain why we saw so few exposure effects on offspring? One possibility is that other researchers have obtained but not published negative results. Another is experimental design differences. For example, because even non-chemical maternal stress alters nutrient provisioning and offspring phenotype in C. elegans (Jordan et al. 2019), we intentionally created a design that included an exposure group (0.03 µM rotenone) in which there was almost no detectable effect in the parent generation, and we tried to minimize any maternal loading of rotenone into the eggs. In the published comparator literature that we identified, either maternal health effects were not assessed, or there were clear effects on maternal health. For example, Yue et al. (2020) identified a decrease in brood size, as well as altered loading of nutrients into eggs. Kim et al. (2021) observed both decreased maternal survival and decreased egg laying. Ma et al. (2019), who employed a design that appeared similar to ours, including the use of one of the mitochondrial toxicants that we used, Antimycin A, saw very robust effects of maternal exposure on offspring. However, they used a concentration of antimycin A that induced significant developmental delay in the P0 generation, and did not describe what measures, if any, were taken to synchronize the exposed and control P0 generation individuals. Another consideration is that in many publications, the design was such that maternal loading of the pollutant is highly likely, especially given concurrent or nearly concurrent exposure and egg collection timelines. Vitellogenin contains both lipophilic and hydrophilic domains and serves as a vector for maternal transfer of contaminants (Monteverdi and Di Giulio 2000). Indeed, significant maternal contaminant loading has frequently been demonstrated in similar studies (Yue et al. 2020; Kim et al. 2021; Annunziato et al. 2022; Marin et al. 2024). It is also possible that delays in parental development and delays or deficiencies in gamete production may not have been considered in all studies. This is important because parental age and health affect the quality of sperm and eggs, such that differences in offspring may result simply from delayed parental development. For example, large phenotypic variations in offspring of isogenic C. elegans raised in nearly uniform conditions are largely the result of maternal age (Perez et al. 2017). We were careful to ensure that offspring were collected from nearly-identically aged adults, but in most papers, if such measures were taken, they are not described. Finally, we note that there are also studies suggesting that epigenetic alterations caused in sperm by high-fat diet exposure and cannabis use are diminished or absent if those exposures cease well in advance of fertilization (Chen et al. 2016; Schrott et al. 2021; Tomar et al. 2024).

The literature on transgenerational effects of toxicant exposure is much larger than the literature on pre-fertilization exposure effects on offspring, and it may also be informative. If studies include enough generations, then whatever chemical exposure occurred was likely pre-fertilization in the later generations, with any offspring phenotypes transmitted somehow through germ cells, gametes, or influences of the mother’s physiology on the zygote. From this perspective, it is interesting that a recent thorough examination of transgenerational effects of four stressors in four different species of nematodes failed to find any transgenerational effects, despite very robust effects in the first generation of offspring (exposures were not pre-fertilization, increasing the likelihood of effects on F1) (Burton et al. 2021). We recently reported a lack of transgenerational effects of arsenite exposure, despite strong effects in the exposed generation (Hershberger et al. 2026). One report that identified effects in offspring of a pre-fertilization exposure, albeit with maternal loading of contaminant (Li et al. 2025), found no effect in the next generation (F2; not technically a transgenerational study). A large-scale review of reported transgenerational effects of stressors in many species including people found that it was not possible to draw firm conclusions about how widespread and robust transgenerational effects are, based on the existing literature (Walker et al. 2018).

Our study has a number of limitations. We did not measure rotenone concentrations in offspring, and so cannot unequivocally rule out the possibility of some maternal transfer of rotenone. However, as discussed in “Chemical exposures” section, given our experimental design and what is known about rotenone’s half-life, we think that significant transfer would be unlikely. We were also unable to precisely and entirely eliminate any developmental stage differences between exposure groups, especially in the high (0.5 µM rotenone) exposure group. Both limitations, however, would be more important as caveats in the context of phenotypes that we did observe in offspring; since we observed very few, especially at the low exposure concentration, those caveats are of less concern. Nonetheless, it is conceivable that the larger effects we did see at the higher exposure concentration resulted in whole or part from maternal transfer of rotenone. Finally, we cannot rule out the possibility that phenotypes other than those we examined, especially including cell type-specific outcomes, were altered in F1 offspring.

Overall, our results do not robustly support our hypothesis that mitotoxicant exposure during germ cell development, at levels not causing frank organismal toxicity or time courses resulting in maternal pollutant loading, would alter mitochondrial function-related phenotypic changes in the offspring.

Author contributions

Conceptualization: Joel N. Meyer, L. Ryan Baugh, Susan K. Murphy, Danielle F. Mello; Data curation: Kinsey Fisher, Rojin Chitrakar, Javier Huayta; Formal analysis: Christina M. Bergemann, Danielle F. Mello, Kinsey Fisher, Rick Presman, Amy H. Herring; Funding acquisition: Joel N. Meyer, L. Ryan Baugh, Susan K. Murphy; Investigation: Christina M. Bergemann, Danielle F. Mello, Shefali R. Bijwadia, Javier Huayta, Ian T. Ryde, Zhiqing Huang, Rojin Chitrakar; Project administration: Joel N. Meyer, L. Ryan Baugh, Susan K. Murphy; Resources: Joel N. Meyer, L. Ryan Baugh, Susan K. Murphy; Supervision: Joel N. Meyer, L. Ryan Baugh, Susan K. Murphy; Visualization Christina M. Bergemann, Danielle F. Mello, Shefali R. Bijwadia, Javier Huayta, Kinsey Fisher: Writing—original draft: Joel N. Meyer, Christina M. Bergemann; Writing—review and editing: Joel N. Meyer, Christina M. Bergemann, Danielle F. Mello, L. Ryan Baugh, Susan K. Murphy.

Supplementary material

Supplementary material is available at Toxicological Sciences online.

Funding

This work was funded by the National Institute of Health (R01ES028218, P42ES010356). Some strains were provided by the Caenorhabditis Genetics Center, which was funded by NIH Office of Research Infrastructure Programs (P40 OD010440).

Conflicts of interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this article.

Data availability

RNASeq data are available from GEO (accessions GSE195584 and GSE292477) with DEGs available in Data File S1. Other data are available at the Duke Data Repository: Https://doi.org/10.7924/r4c82jx2p.