Abstract
In many species, environmental stimuli can affect the germline and contribute to phenotypic changes in the offspring, without altering the genetic code1–5. So far, little is known about which biological signals can link exposure to germ cells. Using a mouse model of postnatal trauma with transgenerational effects, we show that exposure alters lipid-based metabolites in blood of males and their non-exposed offspring. Comparable alterations are validated in serum and saliva of orphan children exposed to trauma. Peroxisome proliferator-activated receptor (PPAR) is identified as mediating the effects of metabolites alterations. Mimicking PPAR activation with a dual PPARα/γ agonist in vivo induces changes in the sperm transcriptome similarly to trauma, and reproduces metabolic phenotypes in the offspring. Injecting serum collected from adult males exposed to postnatal trauma into controls recapitulates metabolic phenotypes in the offspring. These results suggest conserved effects of early life adversity on blood metabolites, and causally involve paternal blood factors and PPAR nuclear receptor in phenotype heritability.
Main
We postulate that circulating factors in blood can be carriers of signals between the environment and germ cells and can mediate the effects of exposure to postnatal trauma. Blood metabolites are proposed to be likely candidates because they include several classes of potent signalling molecules such as hormones, lipids, organic acids and antioxidants in mammals that can be dynamically regulated by physiological states. Several metabolites have been previously implicated in epigenetic mechanisms of genome regulation6–8. To determine if blood metabolites are affected by environmental exposure, we first examined blood from mice exposed to traumatic stress during early postnatal life and their offspring by unbiased high-throughput time-of-flight mass spectrometry (TOF-MS). We used an established transgenerational mouse model based on unpredictable maternal separation combined with unpredictable maternal stress (MSUS, Fig. 1a), which exhibits metabolic and behavioural symptoms that are transmitted to the offspring up to 4 generations3, 9–11. The results showed that polyunsaturated fatty acid (PUFA) metabolism, in particular metabolites involved in α-linolenic and linoleic acid (ALA/LA), and arachidonic acid (AA) pathways, are significantly upregulated by MSUS. In contrast, bile acid biosynthesis and to a lesser extent, steroidogenesis, were downregulated (Fig. 2, full table in Extended Data Fig. 1). Altered steroidogenesis corroborates previous findings in the MSUS model that the mineralocorticoid receptor (MR) and its steroidogenic ligand, aldosterone, are downregulated (Extended Data Fig. 2) and that pharmacological blockade of MR mimics MSUS effects9. Remarkably, except for AA metabolism, these pathways were also altered in plasma of the adult offspring of MSUS males (Fig. 2).
To relate these findings in a mouse model to trauma conditions in human, we assembled a cohort of children (6-12 years old girls and boys) from an SOS Children Village in Lahore, Pakistan, who have lost their father and were separated from their mother (paternal loss and maternal separation, PLMS). Control children were schoolmates not exposed to trauma and living with both parents. PLMS and control groups were matched for gender, age and body mass (Extended Data Fig. 3b-d). For this type of study, a Pakistani population is advantageous because consanguinity is high in Pakistan12, and thus significantly reduces genetic heterogeneity. PLMS and control children were analysed by psychometrics, and both blood and saliva were collected. PLMS children had increased depressive symptoms compared to controls (Extended Data Fig. 3a), consistent with depressive-like behaviours observed in MSUS mice3. Further, their serum metabolites showed significant positive enrichment for AA metabolism and modest negative enrichment for bile acid biosynthesis compared to controls (Fig. 2). In saliva both ALA/LA and AA metabolism, and steroidogenesis were also altered (Fig. 2; Extended Data Fig. 4), indicating alterations across body fluids and partial concordance of metabolomic alterations by trauma exposure in mouse and human.
Fatty acids, especially PUFAs, and their metabolites modulate metabolism, inflammation and cognitive functions, and act as potent ligands of peroxisome proliferator-activated receptors (PPAR). PPAR is a class of nuclear receptors that form transcription factor complexes with retinoid X receptor (RXR) to regulate gene expression and chromatin structure, and which can interact with epigenetic modifying enzymes13, 14. Bile acids and steroid metabolites are ligands for farsenoid X receptor (FXR) and liver X receptor (LXR), which belong to the same family of nuclear receptors as PPAR and RXR, and that can interact15. We examined whether serum can activate PPAR and tested germ cells to assess a potential link with effects in the offspring. We exposed spermatogonial stem cell-like cells (GC-1 spg cells) to culture medium enriched with 10% serum collected from either control or MSUS adult males. Prior to exposure, GC-1 spg cells were transfected with a plasmid expressing luciferase under the control of a PPAR response element (PPRE). Luciferase luminescence was significantly increased in cells exposed to MSUS serum compared to control serum (Extended Data Fig. 5a), suggesting PPAR activation by MSUS serum factors.
Previous studies have implicated PPAR and other nuclear receptors in the effects of environmental exposure and phenotype transmission4, 5, 16–18, but did not test their causal involvement. We assessed causality between PPAR and effects in the offspring and the functional influence of PPAR-mediated pathways in vivo by conducting a series of experiments. First, we measured the expression of PPARγ, an isoform enriched in sperm19, by quantitative PCR. We observed that PPARγ is upregulated in sperm of adult MSUS males (Extended Data Fig. 5b). Second, we examined ligand-dependent PPAR activation in adult tissues using transcription factor binding assays. We found that binding of PPARγ to its consensus sequence is increased by MSUS, in particular in white adipose tissue where PPARγ is abundant and regulates adipocyte differentiation20 (Extended Data Fig. 5c), indicating increased ligand activation. Then, we also examined PPARα targets in liver, a tissue with high PPARα activity21. We observed that several targets are differentially expressed, suggesting PPARα activation (Extended Data Fig. 5d). Lastly, we assessed whether PPAR activation is causally linked to phenotype transmission by mimicking it in adult control males via chronic intraperitoneal (i.p.) injection of the dual PPARα/γ agonist tesaglitazar (10 µg/kg) (Fig. 1b). After a delay of 46 days to allow a full spermatogenesis cycle and eliminate transient effects, males were bred with control females to generate an offspring. When adult, the offspring of tesaglitazar-injected males had significantly lower body weight (Fig. 3a), despite it being higher at PND8 (Extended Data Fig. 6), and reduced glucose level during a glucose tolerance test (GTT) (Fig. 2b) compared to the offspring of males injected with a vehicle control. These adult phenotypes were similar to those observed in the offspring of MSUS males (Fig. 4a and 22), suggesting that PPAR activation can mimic the effects of MSUS across generations.
Since sperm RNA has been causally involved in the transmission of the effects of MSUS to the offspring 22, 23, we examined whether transcriptional changes can be detected in sperm of tesaglitazar-injected males. Deep sequencing of sperm RNA revealed differential RNA expression in tesaglitazar-injected compared to vehicle-injected males, in particular a global dysregulation of transposable elements (TEs) (Extended Data Fig. 7-8). These results are consistent with previous observations in liver of tesaglitazar-treated mice24 and in sperm of males exposed to MSUS23. To directly compare TEs in tesaglitazar-injected and MSUS sperm datasets, we split RNA by type and separated TEs from mRNAs/lincRNAs (long intergenic non-coding RNAs). This revealed a significant fold change correlation of common differentially expressed TEs, including several long terminal repeats (LTRs) between tesaglitazar-injected and MSUS sperm (Fig. 3c; Extended Data Fig. 7a). A modest fold change correlation of mRNAs/lincRNAs with several lncRNAs was also noted, in particular with lncRNAs that are the most enriched across datasets (Fig. 3d; Extended Data Fig. 7b). Further, in sperm from tesaglitazar-injected males, several genes were altered (FDR < 0.05), for instance genes involving the mitochondrial respiratory chain complex (Extended Data Fig. 9), consistent with a role for PPAR in mitochondrial metabolism25. Together, these data suggest a link between PPAR pathways in the periphery and sperm, which overlaps with the effects of MSUS. The lasting effects of tesaglitazar are not likely due to secondary metabolic alterations that persist until the time of breeding since there was no difference in plasma metabolites in treated males after 46 days (Extended Data Fig. 10).
Finally to assess the causal link between circulating factors and transmission of phenotypes, we collected blood from 4-month old MSUS and control males, prepared serum and chronically injected 90 µl intravenously (i.v.) in control adult males (Fig. 1c). Following 4 weeks of treatment, males were bred with control females to generate offspring that were phenotyped when adult. The offspring of males injected with MSUS serum trended towards reduced weight (Fig. 4a) and had significantly lower blood glucose upon stress (Fig. 4c). These symptoms were similar to those observed in MSUS offspring (Fig. 4a-b and 22). No difference in blood glucose was observed in MSUS serum-injected males during GTT (Extended Data Fig. 11). Together, these results suggest that factors in blood are sufficient to induce the transmission of effects of early trauma to the offspring, and thus, that blood can transfer signals related to previous life experiences to the germline.
Slightly different phenotypic outcomes were observed after injection of tesaglitazar or MSUS serum, suggesting that additional circulating factors likely contribute to changes in sperm cells. This is expected since many blood components in addition to PUFAs are altered by MSUS and may not be reproduced by tesaglitazar alone. We assessed some of the possible additional factors and examined RNA in serum of adult MSUS mice. RNA has been implicated in epigenetic inheritance22, 26, 27 and circulating miRNAs can communicate with tissues outside their site of origin28, thus RNA in blood could also contribute to the effects of MSUS. We found that small RNAs identified by deep sequencing are not significantly altered in MSUS serum after correcting for multiple comparisons (Extended Data File 1), even if individual miRNAs had previously been found to be altered by qPCR22. This suggests that miRNAs in blood likely do not contribute. However, the possibility that more complex signaling interactions involving RNA, such as exosomal RNA uptake29, cannot be excluded. Further to RNA, we also examined proteins by conducting a proteomics screen in serum. The results did not reveal any robust changes in circulating proteins by MSUS (Extended Data Fig. 12a), except for a downregulation of C-reactive protein (CRP) (Extended Data Fig. 12b). This suggests a possible downstream effect of PPAR since CRP is a marker of inflammation that can be negatively regulated by PPAR activity30.
Germ cells are the carrier of biological heredity that pass information from parent to progeny via the genome and epigenome31. Because they are sensitive to environmental factors especially in early life32, germ cells are subjected to alterations by exposure and if these alterations are present at the time of conception, they may be transferred to the offspring. The present results provide evidence that nuclear receptors are involved in the transfer of signals from paternal experiences to the offspring and identify PPAR as causally responsible. They show that PPAR can be altered in adult sperm by postnatal trauma, and that mimicking PPAR activation can lead to phenotype transmission associated with transcriptional changes in sperm. These results provide a PPAR-mediated molecular link between the environment and the germline, supporting the previously known association of PPAR with the transmission of diet-induced metabolic phenotypes4, 5, 33. They also reveal PPAR as a transmission mediator of the effects of traumatic experiences, which is unexpected and highlights a previously unknown function of PPAR. This suggests that PPAR is a key signalling component of the heritability of parental experiences. Since transcription factors can confer a poised transcriptional state to gametes and influence the developmental trajectory of zygotes34, PPAR activation in MSUS sperm may contribute to the differential gene expression previously observed in the offspring at the zygotic stage23. We also provide evidence that serum can recapitulate some of the effects of exposure in the offspring, pointing to the influence of blood components on the germline4, 35–37. This further questions the already challenged Weismann barrier theory, which posits that signals cannot pass from soma to germline38, 39, by showing that signals can indeed be transferred from blood to the germline. In the further analyses of the mechanisms involved in such information transfer, the use of high-throughput metabolomic profiling in other body fluids like seminal fluid40 and applied to other inter- or transgenerational models may help identify additional molecular pathways involved in epigenetic inheritance. Finally, these findings raise important questions about heredity and evolution, and extend the notion of transmission mechanisms to communicating factors from circulation.
Funding
We thank the University Zürich, the ETH Zürich, the Swiss Science National Foundation (31003A-135715), ETH grants (ETH-10 15-2 and ETH-17 13-2), Novartis Foundation (16B097), Roche Postdoctoral Fellowship Program (ID233), Cancer Research UK (C13474/A18583), and the Wellcome Trust (104640/Z/14/Z, 092096/Z10/Z). Katharina Gapp was supported by an early and advanced PostDoc mobility fellowship from the Swiss National Science Foundation.
Author contributions
GvS, KG and IMM conceived and designed the study. GvS and IMM wrote the manuscript. GvS conducted MSUS treatments together with FM, collected and prepared plasma for metabolomic and proteomic analyses and tissue from MSUS and tesaglitazar-injected mice, performed molecular analyses of tissues, performed tesaglitazar and vehicle injections, organized breedings for tesaglitazar-and vehicle-injected mice and phenotyped offspring together with FM, and purified RNA from tesaglitazar-injected sperm used for sequencing. KG collected and prepared serum for injections and phenotyped serum-injected offspring and prepared RNA libraries collected from MSUS and control serum and sperm from tesaglitazar- and vehicle-injected males. AJ collected serum and saliva samples from children at the SOS village in Lahore, Pakistan, and performed all related data measurements including analysis of CES-DC results, performed the CRP ELISA and assisted with writing sections of the manuscript. PLG performed bioinformatic analysis, helped to prepare figures, assisted with statistical analyses, and provided key insight into manuscript development. FM organized animal housing and breeding logistics in Zürich, tracked animal welfare and performed phenotyping with GvS. DKT performed and assisted PLG with bioinformatics analyses. NZ measured metabolites in plasma, serum and saliva and analysed the data. NG, AE and KT helped with molecular analyses. IMM and EM provided essential conceptual support throughout the project and raised funds to finance the project.
Competing interests
The authors declare no competing or conflicting interests regarding the contents of this manuscript.
Data and materials availability
Repository accession numbers will be available at publication or by request through the corresponding author. All other data, including code data, are available in the main text or the supplementary materials, or available from the corresponding author upon reasonable request.
Acknowledgments
We thank Irina Lazar-Contes and Martin Roszkowski for assisting with MSUS breedings, Silvia Schelbert for taking care of the animal license and lab organization in Zürich, Lukas von Ziegler, Paolo Nanni and Peter Gehrig for support with proteomics sample preparation and analysis, Johannes Bohacek for advice and Yvonne Zipfel for animal care in Zürich. We thank Paul Green for help with serum injections, and Pawel Zielekinski for help with general animal care in Cambridge. We thank Chris Lelliot for conceptual support and recommendations for tesaglitazar injections, Darren Logan, Wayo Matsushima and Tomas diDomenico for advice on early bioinformatics analysis. We are highly grateful to the administration of the SOS Children’s Village, Pakistan, to Saba Faisal, Mrs. Rubina Asghar Ali, Almas Butt and Sajida Makhdoom at The Educators school, Lahore, Pakistan for allowing the assessment of PLMS and control children respectively, Anooshay Abid and Mehr Shafique at Lahore University of Management Sciences for technical help, Omar Chughtai at Chughtai Laboratories in Lahore for assistance with blood collection, and Safeeullah Chaudhry and Shaper Mirza at Lahore University of Management Sciences for organizational support.