In about half of human cases, the causes of PTB remain unknown. Here, we provide evidence that gestational stress across generations of timed-pregnant rats has downstream effects on endocrine, metabolic and behavioural manifestations of PTB, and leads to shortened gestational length. Developmental trajectories across all generations of offspring were affected as early as P7. In terms of molecular mechanisms, stress in the parental F0 generation had minor effects on regulatory miRNA pathways in brain, uterus and placenta. By contrast, a history of stress in the F2 generation was associated with drastic changes in somatic tissue miRNA profiles and altered expression of genes that have been linked to PTB in humans. Notably, genuine transgenerational programming of developmental trajectories was observed in the F3 generation, in which gestational stress was imposed on the great-grandmaternal generation and was inherited to affect the developing embryo. These findings suggest that the mechanisms involved in the timing of parturition and associated behavioural and physiological signatures can be programmed through the maternal lineage.
A main finding of the present study is that gestational length is influenced by prenatal stress rather than by gestational stress. The impact of prenatal stress on phenotype in the present study was illustrated by delayed developmental milestones in the F1 generation, with an even stronger impact in subsequent F2 and F3 generations. Prenatal stress has been shown to program fetal brain development, HPA axis function and mental health [33–35]. Thus, early experiences can prime physiological and immunological processes that may lead to variations in gestational length [36,37] and susceptibility to altered glucose metabolism, such as type 2 diabetes  in adulthood. The timing and severity of the stressor is critical in that stress in early pregnancy may have greater effects on health outcomes than stress experienced in the last trimester . The present study induced stress from gestational days 12 to 18, representing a period thought to cover a large extent of the human second trimester [40,41] and in rats has been shown to be particularly susceptible to environmental influences, inflammatory processes and stress . While the present study controlled for adverse effects of stress, other factors, such as anesthetic administration, may still have affected the present outcomes . Altogether, it is possible that gestational and inter-generational programming of HPA axis responses may sensitize the response to environmental adversity thus resulting in gradually shortened gestation across generations and further reduction in the multigenerationally stressed cohorts (SSN and SSS).
Beyond fetal endocrine programming by an altered gestational endocrine milieu, maternal distress during pregnancy may also critically affect offspring brain development and physiology through variation in maternal behaviours [21,22,44]. The present data show that prenatal stress alters patterns of early postpartum maternal behaviours, which may be predictive of altered maternal care and stress coping at later times. The first hour after completed parturition may represent a critical transition phase in which the characteristic patterns of late antepartum behaviours, including tail chasing activity and nest building, convert into maternal care of the offspring . It has been demonstrated that the early postnatal environment, such as variations in maternal care, determines developmental and epigenetic outcomes [45,46]. Patterns of altered maternal behaviour may transmit to subsequent generations [21,47]. The contribution of endocrine and behavioural influences to generational programming is complex and likely reciprocally regulated by the epigenome.
In line with previous findings of stress-induced alteration in brain miRNA profiles  and according to the present observations of altered maternal behaviour, multigenerational stress in the F2-SSS group had prominent effects on miRNA expression patterns in the frontal cortex. Interestingly, F2-SSS dams showed upregulated miR-23b, which regulates oligodendrocyte development and myelination . miR-200 family members, including the downregulated miR-200a, are predicted to target genes that regulate synaptic function, neurodevelopment and neuronal survival . Stress also downregulated miRNAs that possess potential roles in the pathogenesis of psychiatric diseases, such as miR-96 , miR-182 and miR-183 . Furthermore, stress-induced downregulation concerned miR-429, which potentially influences development by altering cell proliferation and apoptosis . It is important to note that many neurodegenerative and psychiatric disorders share a pathology involving miRNA regulation [22,54,55] and that these miRNAs may in turn regulate central stress responses . Though not determined in this study, it is likely that these miRNA changes are not limited to the prefrontal cortex and uterus, thus indicating a potential intersection linking psychological stress to altered gestational length.
Across all generations, mechanisms of prenatal stress to modulate gestational length may include modulation of the complex pro-inflammatory state leading to PTB . Furthermore, stress may affect levels of hormones and neuropeptides, including prolactin, progesterone and oxytocin, which are involved in maintenance of pregnancy and timing of delivery . Increased fetal HPA axis activity may induce prostaglandin production by fetal membranes and decidua leading to uterine activation . In addition, stress may stimulate cytokines, which regulate the activity of placental 11-beta-hydroxysteroid dehydrogenase  to elevate PTB risk. These endocrine regulations have led to the notion that PTB risk may have roots in childhood . The present data confirm this notion and provide possible mechanistic links to epigenetic regulation of gene expression related to PTB risk.
Including the downregulated miR-200b, the miR-200 family may exert peripheral effects to control uterine quiescence and contractility during pregnancy and labour . Interestingly, miR-200b/200c/429 are induced at term labour in mice and humans and miR-200b/200c/429 are upregulated in mouse models of preterm labour . This group of miRNAs may largely interact with the endocrine cascade involved in pregnancy maintenance and termination, including progesterone and oxytocin . Moreover, miR-451 is expressed in the uterus [58,59] and regulated by estrogen and progesterone .
Target genes of the miR-200 family include three particular genes, Stat5b, Zeb1 and Zeb2, all involved in pregnancy maintenance . In the uterus, all three were downregulated by multigenerational stress in the F1 generation. Effects on Zeb2 expression were transmitted to the F2 generation. These findings concur with the reduction in gestational length. Accordingly, a decrease in Stat5b expression was linked to reduced progesterone activity and the initiation of labor, in particular in preterm birth . Furthermore, ZEB1 serves as transcription factor to inhibit the miR-200 family, thus enhancing Stat5b expression . As the myometrium transitions to term or preterm labor, reduced progesterone activity decreases ZEB1 and ZEB2 levels via a feed-forward mechanism [18,19], thus regulating the timing of parturition. The upregulation of uterine miR-200b may be causative for the suppression of Stat5b and ZEB1 and ZEB2; however, they may also reflect low postpartum progesterone levels due to timing of tissue sampling in the present study. Although the direction of these and the placental miR-181a changes are opposed to the downregulation found in human preterm birth , their differential expression across generations coincides with shortened gestational length and indicates a causal or, at least, predictive signature of preterm birth.
A role for genuine epigenetic inheritance of stress response is suggested by the present findings concerning the F3 generation. While context-dependent programming may have mainly determined the F1 and F2 phenotype, programming of the germ-line became evident by altered development in the F3 generation . Indeed, the most dramatic impact of prenatal stress on developmental trajectories was found in the F3 generation. Only in the F3 generation did the offspring display low body weight already on P1, which was associated with reduced growth trajectories and a drastic sensorimotor behaviour deficit. Since these phenotypic changes persisted to the F3-SNNN generation in the absence of direct somatic exposure, they are arguably mediated by genuine transgenerational programming of the female germline [16,17,60]. This suggests transgenerational epigenetic inheritance whereby the epigenetic modifications may have been passed on via the gametes that have escaped reprogramming [16,61,62]. Thus, the study of transgenerational programming of epigenetic signatures may provide a unique opportunity to identify predictive biomarkers and future therapeutic targets to promote maternal and child health.