Perinatal Nutrition and Programmed Risk for Neuropsychiatric Disorders: A Focus on Animal Models

Abstract

Maternal nutrition is critically important for fetal development. Recent human studies demonstrate a strong connection between diet during pregnancy and offspring risk for neuropsychiatric disorders including depression, anxiety, and attention-deficit/hyperactivity disorder. Animal models have emerged as a crucial tool for understanding maternal nutrition’s contribution to prenatal programming and the later development of neuropsychiatric disorders. This review highlights preclinical studies examining how maternal consumption of the three macronutrients (protein, fats, and carbohydrates) influence offspring negative valence behaviors relevant to neuropsychiatric disorders. We highlight the translational aspects of animal models, and so examine exposure periods that mirror the neurodevelopmental stages of human gestation. Due to our emphasis on programmed changes in neurobehavioral development, studies that continue diet exposure until assessment in adulthood are not discussed. The presented research provides a strong foundation of preclinical evidence of nutritional programming of neurobehavioral impairments. Alterations in risk assessment and response were observed alongside neurodevelopmental impairments related to neurogenesis, synaptogenesis, and synaptic plasticity. To date, the large majority of studies utilized rodent models and the field could benefit from additional study of large animal models. Additional future directions are discussed, including the need for additional studies examining how sex as a biological variable affects the contribution of maternal nutrition to prenatal programming.

Introduction

Neuropsychiatric disorders are a major global health concern. They are highly prevalent and lack effective prevention and treatment strategies, consequently imposing enormous societal costs (1-5). The origins and mechanisms of neuropsychiatric disorders are of great consequence to the public and scientific community, with findings from epidemiological and animal studies indicating that early life conditions greatly contribute to risk of mental disorders (6, 7). The fetal programming hypothesis posits that prenatal environmental factors influence long-term neuropsychiatric outcomes by altering epigenetic control of neural processes or disrupting neural function during critical periods of development (7, 8). The prenatal environment is significantly influenced by nutrient availability, and nutritional programming specifically investigates the residual impact of fetal nutrient imbalance (9). The mother supplies offspring nutrition during gestation and lactation, with an excess or deficiency in most nutrients impacting fetal neurodevelopment (10). The placenta facilitates maternofetal nutrient exchange in utero, buffering glucose fluctuations, storing lipids, and producing a few essential amino acids (11) (Figure 1). In addition to diet, maternal metabolic conditions (e.g. maternal obesity, diabetes) alter nutrient balance and placental function (10, 12); clinical and animal models implicate these conditions in increased offspring risk of neuropsychiatric disorders (13-15). Maternal nutrition and metabolic state are highly inter-related and associated with diet, making their contributions to offspring neuropsychiatric impairments difficult to differentiate.

Figure 1. Maternal Diet and Fetal Macronutrient Availability.

Food sources rich in protein, carbohydrates, and fats increase the presence of amino acids, glucose, and fatty acids in maternal circulation. These nutrients enter fetal circulation via different methods of placental transport. Amino acids provided by dietary protein require active transport across the placenta, and the placenta can produce select amino acids like glutamate into the fetal circulatory system. The transport of glucose (derived from sugars and grains) across the placenta is facilitated by glucose transport proteins, causing changes in fetal blood glucose concentrations that closely mirror maternal levels. Fatty acids derived from triglycerides present in dietary fats like butter and oil can freely diffuse across the placental boundary and fatty acid transport proteins provide additional energy-dependent transfer. While many fatty acids directly enter fetal circulation, the placenta can convert the fatty acids back to triglycerides for storage in lipid droplets. After entering fetal circulation, macronutrients traverse the blood-brain-barrier via similar mechanisms to placental transport. As described throughout this review, nutritional programming results in altered neural function and development (depicted in gray). Discussed mechanisms include elevated neuroinflammation, dendritic atrophy and instability, and delayed glial maturation resulting in reduced myelination. Many of the reported studies investigated altered expression, production, and function of cell signaling molecules and receptors. Inadequate fetal nutrition is associated with changes in neurotrophin (e.g. brain-derived neurotrophic factor signaling), neurotransmitter (e.g. dopamine), and neuromodulator systems (e.g. endocannabinoid signaling).

Epidemiologic studies exploring nutritional programming are complicated by substantial variation in nutrition and limitations in modifying the diet of pregnant mothers and newborns. Animal models allow precise control of diet content and the ability to limit manipulation to critical periods of development. Notably, developmental ontogeny varies depending on the model utilized; neurodevelopmental processes that occur during late gestation in humans take place postnatally in rodents (16, 17) (Figure 2). It is therefore necessary that translational efforts consider differences in nutrient transfer and demand between the intrauterine and extrauterine environments. Adopting an endophenotype approach to neuropsychiatric research, this review focuses on evidence from animal models demonstrating nutritional programming of offspring negative valence behaviors (18, 19). In humans, the negative valence domain encompasses fear and sadness and, while adaptive in many contexts, negative valence behaviors are dysregulated in multiple psychopathologies (20, 21). Although not equivalent to the complex spectrum of behavioral endophenotypes in humans, animal models can reliably investigate aspects of the negative valence system via behavioral assays. Just as excessive fear and anxiety can impair a person’s ability to evaluate and respond to a stressful or threatening situation, an animal’s behavioral response can reveal altered appraisal and avoidance of a threatening stimulus, giving insight to potentially dysregulated defensive response (22-24) (Figure S1). In rodents, changes in thigmotaxis (the tendency to avoid open, exposed areas) and passive coping behaviors can reliably indicate altered threat response, similar to anxiety in humans (25-29). Importantly, major components of negative valence networks are well-defined and conserved among mammals. Investigation of the corticolimbic system, involved in risk assessment and response, gives insight to neurobiological aspects of neuropsychiatric disorders in humans (30).

Figure 2. Neurodevelopmental Ontogeny and Nutrient Requirements.

Nutritional requirements for developing offspring change in accordance with physiological demand. During in utero development, glucose provided by circulating maternal glucose is the main source of energy for the developing fetus. Availability of energy and amino acids determine the rate of protein synthesis, and protein accretion is critical to sustaining fetal growth in early and late gestation. During advanced gestation demands for omega-3 (ω-3) and omega-6 (ω-6) essential fatty acids increase, and the fetus is capable of some de novo fatty acid synthesis to assist in lipid accumulation. Immediately following birth, blood glucose drops rapidly as the continual source of maternal glucose via the placenta is replaced with nutrition provided by nursing. Lipids account for a substantial source of energy in the early postnatal stage as fat accumulation continues. The importance of these nutritional changes across development is conserved across animal models, however the timing varies according to developmental rate. Presented are paired neurodevelopmental timelines for humans (above, in gestational weeks post-conception; solid bars) and rats (below, in gestational days post-conception and postnatal days after birth; striped bars) (similar in mice). Different brain regions have asynchronous development; therefore, the general stages of brain development are presented in this image. Discordant schedules of neurodevelopment are evident between the two species. In particular, processes that occur during late gestation in humans take place postnatally in rats. Weaning in rodents is triggered naturally by sexual maturity around three weeks of age, marking the beginning of adolescence/adulthood. It is important to note that different rat and mouse species have innately varied timelines, and for some of the animal models discussed in this review weaning occurred at the natural point for that species, which may be a number of days before or after postnatal day 21.

We focus on negative valences behaviors and neural outcomes from animal models relevant to global dietary patterns and seek to investigate programming influence of maternal diet independent from metabolic disorders. This review highlights the translational aspects of animal models, and so examines exposure periods that mirror the neurodevelopmental stages of human gestation (16, 17). Studies are included that examine diet manipulation during gestation (conception to birth), lactation (birth to weaning), or both combined, defined here as perinatal exposure. Due to our emphasis on programmed changes in neurobehavioral development, all studies provided diet intervention (switched to control diet) by weaning. The correction of nutrient balance prior to neurobehavioral assessment in adulthood allows for the examination of long-term changes in offspring behavior. Tables are provided to supply key details for the presented behavioral studies. We concentrate on neurobehavioral differences resulting from maternal dietary manipulation of the three macronutrients: protein, fats, and carbohydrates.

Protein

Proteins are macronutrients comprised of amino acid subunits that are required for growth and the production of cellular receptors, transporters, and signaling molecules like neurotransmitters (31). Amino acids require active transport across the placenta and blood brain barrier. While many are produced endogenously, nine essential amino acids must be obtained from dietary sources (11). The placenta prioritizes resources to sustain fetal growth; however, animal models demonstrate that placental amino acid transport is reduced at the detriment of fetal amino acids when protein restriction limits maternal amino acid availability (32-37). Protein restriction is highly relevant as millions of people worldwide consume dietary staples that are poor sources of amino acids, both in terms of quality and quantity (38). While the effects of gestational protein restriction on overall growth are well-documented, limited animal studies investigated the effect of maternal protein restriction on offspring negative valence behaviors (Table 1). The few animal models examining this relationship have focused on casein. Unlike the sources typically available in protein deficient human diets, casein is a high-quality protein that provides all essential amino acids. As a rat model demonstrated that protein quality alone significantly affects offspring development (39), this discrepancy is important to note when translating findings to the human condition. Nevertheless, preclinical research provides important insights regarding the effect of inadequate dietary protein during development on offspring behavioral regulation.

Table 1.

Protein

Source	Model	Strain/ Species	Subject Sex	Diet Manipulation	Diet Exposure Period	Behavioral Assays	Testing Age	Significant Outcomes
Reyes-Castro 2012 (44)	Rat	Wistar	Male	10% protein (control 20%)	Gestation, Lactation, Perinatal	Open field test, Elevated plus maze	10 weeks	Gestational, lactational, and perinatal restriction impaired risk assessment (EPM).
Reyes-Castro 2012 (43)	Rat	Wistar	Female	10% protein (control 20%)	Gestation, Lactation, Perinatal	Open field test, Elevated plus maze, Operant conditioning & Progressive ratio	12-22 weeks	Perinatal and gestational restriction increased thigmotaxis (EPM, OFT), gestational restriction decreased motivation (PR).
Crossland 2017 (41)	Mouse	C57BL6/J	Male	8% protein (control 20%)	Pre-conception perinatal	Open field test, Elevated plus maze, Light/dark transition, Forced swim test, Tail suspension test, Fear conditioning	8-12 weeks	Increased thigmotaxis (EPM)
Belluscio 2014 (40)	Mouse	CF-1	Male and female	9% protein (control 20%)	Perinatal	Open field test, Elevated plus maze, Tail suspension test, Cage escape	3-9 weeks	Decreased motivation (CE), increased thigmotaxis and decreased rearing and head dipping (OFT, EPM). Increased behavioral despair in females only (TST).
Pillay 2016 (42)	Mouse	African Striped Mouse	Male and female	10% protein (control 19%)	Pre-conception perinatal	Open field test	10-11 weeks	Increased thigmotaxis (OFT)

Diet manipulation values presented as percent of g/kg. Abbreviations: EPM=Elevated Plus Maze, OFT=Open Field Test, PR=Progressive Ratio, CE=Cage Escape, TST=Tail Suspension Test

Perinatal protein restriction beginning at conception consistently increased threat aversion and reduced passive coping behaviors in adult mouse offspring; females further exhibited increased behavioral despair (40). When perinatal protein restriction began prior to conception, mouse offspring displayed less profound behavioral alterations, elevating only thigmotaxic response (41, 42). These findings suggest that initiating protein restriction weeks prior to conception may allow maternal acclimation to the diet, potentially lessening the impact on the fetus. Additional studies in rats found that perinatal and gestational restriction similarly enhanced stress sensitivity, with females demonstrating increased risk aversion and frustrative non-reward (43). In contrast, males exhibited decreased threat avoidance with any exposure to maternal protein deficiency (44). Overall, perinatal protein restriction increased stress sensitivity and fear response in males and females, and a single study in rats suggests sex-specificity. These studies further noted increased behavioral despair in females only, consistent with higher rates of clinical depression in women (45). However, few studies examined depression-like responses in both male and female offspring, and the field would benefit from additional research to validate this finding.

Preclinical behavioral outcomes are supported by evidence that maternal protein restriction is detrimental to offspring neural functions related to the negative valence system. Two models investigated the effects of perinatal protein restriction on neurotransmitter systems important in behavioral regulation, finding that mouse offspring displayed a hyperactive dopaminergic system attributed to hypomethylation (46) and rat offspring displayed desensitization of serotonergic receptors (47). While global alterations were observed in rat offspring, including decreased brain weight and protein levels, hippocampal neurogenesis appears to be particularly disrupted by gestational or perinatal protein deficiency (48, 49). These disturbances are highly relevant to negative valence behaviors; the hippocampus is a crucial part of the limbic system and fear circuitry, and impaired hippocampal neurogenesis is a potential contributor to neuropsychiatric impairment in humans (50-52). Perinatal and gestational protein restricted rat offspring displayed evidence of impaired hippocampal development, exhibiting reduced brain-derived neurotrophic factor (BDNF) levels, brain volume, and neuron population (48, 49). Neurotrophic growth factors like BDNF and insulin-like growth factor (IGF) are important for healthy neurodevelopment (53, 54) and are decreased in mice and rats exposed to maternal protein deficiency (34, 35, 48). Preclinical evidence suggests that decreased hippocampal BDNF could reflect elevated stress hormones due to protein restriction (43, 44, 55, 56), as altered BDNF in this region is specifically associated with prenatal stress-induced methylation changes (57, 58).

Demonstrating the complexity of nutritional programming, nutrient changes due to protein restriction are not limited to amino acids. Gestational protein restriction decreased maternal lipid availability in rats, lowering fetal brain fatty acid levels and potentially contributing to the long-lasting reductions in myelin produced by early lactation protein restriction (59, 60). Clinical studies show that myelin deficits are associated with neuropsychiatric disorders, and both clinical and preclinical evidence supports the importance of sufficient brain fatty acids for neurobehavioral health (60, 61). Clearly, altering maternal protein content triggers multiple compensatory changes as the body attempts to optimize both maternal and fetal health. In addition to the aforementioned associations with offspring neural health, maternal fatty acids, steroid hormones, and growth factors regulate placental amino acid transport, alluding to the complex interrelation of mechanistic components (32). Future research expanding on the presented findings should consider these potential mechanisms and additionally investigate the impact of protein quality on offspring neurobehavioral health. Importantly, casein is the standard protein utilized in laboratory animal chow, and so is consistent with other models of dietary manipulation, such as Western-style diets.

Western-Style Diets

Considerable attention has been given to the neurobehavioral impacts of developmental exposure to highly-palatable dietary patterns. These diets are calorically dense, provide increased calories from fat, and incorporate sugar as a noteworthy source of carbohydrates. When consumed consistently, this dietary pattern, referred to as Western-style diet (WSD), produces metabolic impairments including obesity, disrupted glucose and insulin homeostasis, and altered metabolic hormones. The WSD and resulting metabolic disorders have considerable global prevalence and reviews of clinical and preclinical research demonstrate that each contribute to increased risk of offspring neuropsychiatric disorders (13-15, 62-64). Small animal models of WSD-induced obesity report alterations in maternal oocyte quality and placental function that independently influence fetal nutrient availability and neurodevelopment (65, 66). However, findings from chronic maternal WSD models are rarely able to distinguish between the effects of diet and metabolic state. A recent study in nonhuman primates showed these unique influences as perinatal WSD exposure and maternal obesity, but not maternal insulin resistance, differentially impaired offspring behavioral regulation (67). While maternal WSD models are fairly common, there is limited literature investigating the long-term effects on neurobehavioral development in the absence of maternal obesity.

To examine diet-induced changes without metabolic impairments, we focus on acute models of maternal WSD consumption: those that begin diet exposure a maximum of two weeks pre-conception and do not produce differences in maternal body weight before conception (Table 2). A rat model concluded that WSD during lactation decreased risk aversion and elevated exploratory activity at weaning, consistent with disinhibition (68). In adulthood, rat offspring exposed to lactation WSD likewise demonstrated impaired risk assessment in males, but conversely increased inhibition in males and females (69). A similar disconnect between avoidance and inhibition resulted from a perinatal WSD model in Oldfield mice, with females exhibiting increased freezing behaviors (70). Unlike typical laboratory strains, both control and WSD offspring exhibited an atypical preference for exposed areas, complicating the interpretation. Another rat model investigating sustained threat response found that WSD during lactation shortened threat evasion but did not increase immobility, suggesting altered risk aversion but not conclusively behavioral despair (71). While mouse offspring with gestational WSD exposure exhibited similarly reduced risk aversion (72), investigations of gestational and perinatal WSD exposure did not alter fear response in rats (69, 73).

Table 2.

Western-style diet.

Source	Model	Strain/ Species	Subject Sex	Diet Manipulation	Diet Exposure Period	Behavioral Assays	Testing Age	Significant Outcomes
Wright 2011(69)	Rat	Wistar	Male and female	Chow and cafeteria diet: DAC 9.53g fat (2.70g control), 5.71g sucrose (control DAC 2.70g fat, 1.75g sucrose)	Gestation, Lactation, Perinatal	Open field test, Elevated plus maze	10 weeks	Lactation WSD decreased risk aversion in males (EPM, OFT). Lactation WSD decreased activity and passive coping behaviors in males and females (OFT, EPM).
Janthakhin 2017 (73)	Rat	Wistar	Male	Chow: 45% energy lard fat, 17.5% energy sucrose (control 0% lard, 0% sucrose)	Perinatal	Open field test	3-5 months	No differences.
Speight 2017 (68)	Rat	Wistar	Male and female	Chow and cafeteria diet: DAC 11.63g fat, 5.95g sucrose (control DAC 3.325g fat, 1.97g sucrose)	Lactation	Open field test, Elevated plus maze, Home-cage activity	3 weeks	Decreased risk aversion and increased activity and rearing (OFT, EPM).
Giriko 2013 (71)	Rat	Wistar	Male	Chow: 18% ration lard fat, 2% ration sucrose (control 0% lard, 0% sucrose)	Lactation	Forced swim test, Foot-shock, Open field test (activity)	8-14 weeks	Decreased climbing and swimming (FST) and increased aggressive response (foot-shock).
Johnson 2017 (70)	Mouse	Oldfield	Male and female	Chow: 15% ration lard fat, 20% ration sugar (control 0% lard, 10% sugar)	Pre-conception perinatal	Elevated plus maze, Voluntary wheel running, Home-cage activity	12 weeks	All animals had increased number of entries into open arms compared to closed arms. Increased immobility in females (EPM). Decreased head dipping but increased rearing in males (EPM). Decreased activity (home-cage).
Ribeiro 2018 (72)	Mouse	Swiss	Male and Female	Chow and cafeteria diet	Gestation	Light-dark transition test, Open field test (activity)	4 weeks	Decreased risk aversion in males and females, exaggerated in males (LDT).

Cafeteria diets are provided in addition to nutritionally complete chow and vary depending on the model, but typically are an assortment of candy and chips. Nutritional or energy intake not available from Wright 2011 so reported averages were taken from a different publication from the same group (135). Ribeiro 2018 (72) did not report nutritional values comparable to other cafeteria diet models but additional information regarding component products’ energy and nutritional content is available. Open field tests that did not consider zone differences were used to assess activity only, not threat response. DAC=daily average consumption, calculated experimentally, OFT=Open field test, EPM=Elevated plus maze, FST=Forced swim test, LDT= Light-dark transition

These results indicate that early developmental exposure to WSD, particularly during lactation in rat models, impairs risk assessment and modulates later-life stress sensitivity. The importance of this exposure window could be due to the changing nutrition requirements of neonates. Mother’s milk is extremely lipid dense, mostly in saturated fats like those elevated in WSD, suggesting that nursing offspring are more susceptible to maternal WSD effects (74). Additionally, during the early postnatal period rodents undergo important neurodevelopmental processes that, if disrupted, could be responsible for the observed perturbations in behavior (75). Rat and mouse offspring provided evidence that perinatal or lactation WSD exposure resulted in dendritic atrophy and spine instability in the amydgala, hippocampus, and prefrontal cortex (73, 76, 77), with abnormal dendritic environments implicated in various psychopathologies (78). A porcine model also demonstrated hippocampal disturbances, as perinatal WSD reduced hippocampal volume and altered neurogenic mechanisms (79, 80). The dopamine system contributes to attentional and impulse control, and several studies of perinatal WSD in rats demonstrate persistent impairments in dopamine transmission (81-83). The impairments in fear and anxiety circuits observed in animal models provide strong evidence that maternal WSD exposure disrupts neurobehavioral development in a manner highly translatable to human neuropsychiatric disorders.

There are a number of physiological pathways by which maternal WSD alters the course of offspring neurobehavioral development, including inducing neuroinflammatory response and altering the microbiotic environment (77, 79). Major confounding factors of the current WSD literature are the limited models investigating programmed WSD effects independent from maternal metabolic state and the considerable disparity in diet formulation. While this is true with any investigation of diet-derived outcomes, it is particularly pronounced in WSD models, as the experimental manipulation is designed to emulate a multi-faceted dietary pattern and not a single targeted factor. In fact, the three main aspects of the WSD (caloric density, increased fat, and increased sugar) have each been individually associated with altered offspring neurobehavioral outcomes. The following sections discuss how specific alterations in maternal carbohydrate and fat sources each independently alter offspring neurobehavioral development.

Carbohydrates

The WSD is associated with increased consumption of simple carbohydrates and sugars, typically in the form of sucrose or fructose-derived sweeteners. Sucrose is a dimer of glucose and fructose, both becoming freely available in the blood following a meal. Glucose has enhanced importance during gestation; the fetus derives 80% of its energy from glucose, with fetal blood glucose mirroring maternal fluctuations (11). Unlike glucose, fructose is not regulated by insulin and produces unique metabolic consequences as it is slowly converted to glucose by the liver. Despite the body of metabolic programming research investigating the impact of gestational diabetes on offspring risk of neuropsychiatric disorder (84), there is limited mechanistic insight regarding how glucose/insulin homeostasis or fructose influences offspring neurobehavioral development, particularly in the absence of metabolic disorders. Maintaining our focus on maternal nutrition, we identified models of moderate sugar intake without gestational diabetes. Importantly, these studies did not alter maternal weight gain or induce diabetes in offspring. To date, only one preclinical source investigated the influence of maternal sugar intake on offspring behavioral programming (Table S1).

Choi et al. (85) examined the effect of added sugar during gestation on behavior of male mice. Gestational sucrose exposure impaired risk assessment, induced hyperactivity, and decreased spontaneous alternation behavior (suggesting inattention or behavioral inflexibility). Aberrant attentive control and impulsivity were associated with altered striatal dopamine transport and receptor expression, despite normal dopaminergic neuron density. Changes in striatal dopamine function are believed to be key to ADHD pathology; although striatal dopamine transporter activity in humans is associated with trait impulsivity (86), but not conclusively with ADHD (87). Nonetheless, the outcomes from Choi et al. suggest maternal sugar consumption may contribute to the pathophysiology of dopamine-related neurobehavioral abnormalities.

Despite limited behavioral research, the few neural studies evaluating maternal sugar consumption indicate several mechanisms of impaired hippocampal neurogenesis. One group found that gestational exposure to sucrose-sweetened beverages accelerated neurodegeneration, with decreased central IGF levels in sucrose-exposed rat offspring suggesting impaired neuroprotection contributed to hippocampal atrophy (88, 89). Increased maternal plasma glucose could contribute to the uninhibited apoptosis, as preclinical models of gestational diabetes demonstrated that maternal glucose and insulin levels directly influence fetal plasma IGF and neural IGF expression in neonates (90, 91). The involvement of neurotrophic factors is supported by a rat model of elevated maternal fructose consumption, with increased histone modification suppressing BDNF production (92). Evidence of fructose-induced epigenetic modification is supported by a study in adult mice demonstrating that fructose consumption alters transcript abundance and other epigenetic controls, including DNA methylation (93). Yet another group found that perinatal fructose exposure modulated expression of several hippocampal neurosteroidgenic enzymes in rat offspring, consistent with preclinical evidence that glucocorticoids contribute to hippocampal atrophy and associated neurobehavioral impairments (94-96).

These perinatal models of increased sugar exposure indicate the importance of further fetal programming research. Despite the limited studies and differences in sugar type, disturbances in a variety of mechanistic pathways disrupted cell signaling and neurogenesis. Further study is needed concerning potential differences between maternal sucrose or fructose consumption, as each results in unique patterns of maternal glucose and insulin response, impacting placental function (97). Additionally, it is unclear how glucose and fructose differentially affect the brain, or even if fructose can cross the blood-brain-barrier. Recent clinical evidence suggests that peripheral levels of glucose determine central fructose concentrations (98), and that fructose could influence neural function by altering cerebral blood flow (99). Other aspects of carbohydrate intake, such as carbohydrate complexity and glycemic index, should be addressed in follow-up studies. Although current research is limited, preliminary evidence clearly suggests that maternal carbohydrate intake impacts offspring neurodevelopment.

Fatty Acids

The WSD is characterized by a high percentage of saturated fats (prevalent in most animal products) rather than unsaturated fats common in plants and fish. Dietary fat is a triglyceride: a macronutrient composed of three fatty acid chains that cannot traverse the placental boundary unless broken down into component fatty acids (11). Maternal fatty acid consumption determines fetal availability, and animal research indicates that diets low in protein or high in sugar alter fetal fatty acid levels (59, 97). These changes directly influence offspring brain fatty acid profiles, impacting neurodevelopment; fatty acids are utilized in the brain for myelin synthesis, membrane components, cellular signaling, and energy (100, 101). Altered fatty acid profiles are implicated in neuropsychiatric disorders (102), with specific poly-unsaturated fatty acids (PUFA) particularly significant to fetal neurodevelopment (103). Termed essential fatty acids because they must be derived from the diet, omega-3 (ω-3) and omega-6 (ω-6) PUFAs compete for access to the enzymatic pathway that produces long-chain products utilized throughout the brain (Figure 3) (104). While long-chain ω -3 and ω -6 molecules are both crucial to neural function, the ω-3 end-product has an enhanced role in neurodevelopment (104, 105). The ratio of ω-6/ω-3 is critical for brain development, and minor dietary changes in essential fatty acids can dramatically affect cerebral lipid profiles and neural function (106, 107). In excess, maternal PUFAs are associated with similar neurobehavioral phenotypes to WSD: mouse and rat offspring demonstrated impaired risk assessment and decreased hippocampal neurogenesis and synaptic transmission (108-110). Considering that current dietary practices reflect a ω-6/ω-3 ratio of 15/1 (significantly skewed in comparison to the 1/1 ratio maintained in a hunter-gatherer diet) (111), the influence of maternal fatty acids has important ramifications for offspring neurobehavioral development.

Figure 3. Essential Fatty Acid Balance.

Essential fatty acids are poly-unsaturated fatty acids (PUFA) that can only be obtained from dietary sources. These PUFA have one of the double bonds in their hydrocarbon chain located on the third or sixth carbon from the end, omega-3 (ω-3) or omega-6 (ω-6) PUFA respectively. The most basic essential fatty acids are linoleic acid (18:2 (18 carbon chain, two double bonds); ω-6) and α-linolenic acid (18:3; ω-3). Once consumed, commonly from plant-based oils, these molecules can be endogenously modified by a step-wise pathway of desaturase enzymes which convert them into long-chain molecules: arachidonic acid (20:4; ω-6) and docosahexaenoic acid (22:6; ω-3). Linoleic and α-linolenic acids compete for access to these enzymes, meaning excess of either contributes to a relative deficiency of the opposing long-chain product. Arachidonic and docosahexaenoic acids have unique neurodevelopmental functions and can also be obtained directly from diet. PUFAs can cross both the placenta and the blood-brain-barrier via gradient-dependent diffusion and active transport. The relative abundance of these nutrients in fetal blood and in cerebrospinal fluid suggest that essential fatty acids can be preferentially transported in a selective order: docosahexaenoic acid> α-linolenic acid>linoleic acid>arachidonic acid. The exact controls and implications of this transport selectivity are uncertain; however, it is clear that dietary maternal fatty acid imbalance has important ramifications for fetal essential fatty acid availability and neural function.

In an effort to simulate how essential fatty acids are altered in human diets, animal models swap fat sources with moderate levels of ω-3 PUFAs (like canola or flaxseed oils) for low ω-3, high ω-6 fat sources (like safflower or sunflower oils), producing experimental diets that are equally calorie and lipid dense (Table 3). Animal studies commonly examine a high ω-6/ω-3 ratio that results in a relative shortage of ω-3 PUFA, reflecting the trend of ω-3 deficient foods in current dietary patterns. Adult mice with perinatal exposure to high maternal ω-6/ω-3 consistently exhibited increased thigmotaxic, risk aversion behaviors (112-114). Rat offspring also displayed elevated anxiety-like behaviors, and further showed that high maternal ω-6/ω-3 exaggerated physiological stress response and increased behavioral despair (115). Although these studies demonstrated increased brain ω-6/ω-3 ratio during fetal and early postnatal development, the observed behavioral phenotypes in adulthood were not accompanied by long-term changes in brain essential fatty acid availability (113-115).

Table 3.

Fatty Acids.

Source	Model	Strain/ Species	Subject Sex	Diet Manipulation	Diet Exposure Period	Behavioral Assays	Testing Age	Significant Outcomes
High ω-6/ω-3 Models
Jones 2013 (112)	Mouse	C57BL/6J	Male and Female	Safflower oil, 51.3/1 (control: soybean oil, 6.9:1)	Pre-conception perinatal	Elevated plus maze, Open field test (activity)	8-9 weeks	Increased risk aversion (EPM).
Sakayori 2016a (113)	Mouse	C57BL/6N	Male and Female	Safflower oil, 74.4/0.3 (control: canola oil, 2.2/1)	Pre-conception perinatal	Open field test, Elevated plus maze	13-15 weeks	Increased thigmotaxis and risk aversion (OFT, EPM).
Sakayori 2016b (114)	Mouse	C57BL/6N	Male and Female	Safflower oil, 74.4/0.3 (control: canola oil, 2.2/1)	Pre-conception perinatal	Open field test, Elevated plus maze, Forced swim test	13-15 weeks	Increased thigmotaxis (OFT, EPM). Further increased risk aversion in males (OFT). Increased activity in females (EPM).
Chen 2013 (115)	Rat	Sprague- Dawley	Male	Sunflower oil, 61/0 (control: sunflower plus fish oil, 2.6/1)	Perinatal	Elevated plus maze, Forced swim test	10 weeks	Increased thigmotaxis (EPM) and behavioral despair (FST).
Supplement Models
Roversi 2016 (134)	Rat	Wistar	Male and female	3 g/kg daily gavage: Hydrogenated vegetable fat, water	Pre-conception perinatal	Elevated plus maze	6-7 weeks	Decreased open arm time and head dipping (EPM).
Pase 2017 (133)	Rat	Wistar	Male	3 g/kg daily gavage: Hydrogenated vegetable fat, soybean/fish oil mix (control)	Gestation, Lactation	Novel object recognition, Y-maze	12 weeks	Any HVF reduced novelty preference, with lactation period HVF showed long-term novelty aversion (NOR). Any HVF decreased spontaneous alternations (Y-maze).
Ferraz 2008 (130)	Rat	Wistar	Male	3 g/kg daily gavage: Coconut fat, fish oil, no supplement (control)	Pre-conception perinatal	Open field test, Elevated plus maze, Forced swim test	15 weeks	Fish oil decreased immobility time (FST).

High ω-6/ω-3 models elevate ω-6 at the expense of ω-3, producing relative ω-3 deficiency. Diet manipulations indicate the source of fat used to generate the experimental ω-6/ω-3 ratios listed. These sources used different oils to modify the ratio except for Chen 2013, which supplemented the deficient diet with fish oil to alter essential fatty acid ratio. Supplement models provided animals with nutritionally complete chow, and thus did not examine ω-3 deficiency. Supplements were isocaloric and normolipidic except for water controls, which led for Pase 2017 to use a combination soybean/fish oil supplement as a control. Abbreviations: HVF=Hydrogenated vegetable fat, OFT= Open field test; EPM= Elevated plus maze, FST= Forced swim test.

Despite normal brain lipid profiles, perinatal fatty acid exposure impacts long-term neural function. Studies investigating the effect of essential fatty acid rehabilitation during different developmental stages found that altered maternal ω-6/ω-3 ratio through lactation impaired dopamine and serotonin release, reduced myelin yield, and delayed brain growth in adult mouse and rat offspring (116-119). Evidence from the perinatal period demonstrates that these neural processes, as well as hippocampal development and microglia activation, are already disrupted before weaning (120-125). Pre-weaning examinations additionally show ω-6 PUFA alterations impair hippocampal neuroplasticity via altered endocannabinoid signaling and glucocorticoid inhibition (126, 127). This literature suggests that the timing of altered brain lipids during critical periods of development could contribute to neurobehavioral impairments later in life. However, a second interpretation is possible: the observed neural impairments in adulthood are not due to programmed changes in brain development, but rather the length of time between essential fatty acid rehabilitation and neurobehavioral assessment. Brain ω-3 PUFA levels take about eight weeks to normalize after diet rehabilitation (128), and the lack of sustained neural impairment in early interventions is potentially due to the extended recuperation period (116, 129). Although there is a wealth of potential mechanisms for fatty acid programming, few behavioral or neural outcomes have been examined more than eight weeks after diet intervention, highlighting an important future direction for fatty acid and nutritional programming research.

Supportive evidence of PUFA programming is provided by models of fatty acid supplementation to nutritionally complete chow. Perinatal exposure to fish oil supplement decreased behavioral despair in adult rat offspring (130, 131), reflecting clinical interest in associations between depression and ω-3 PUFAs (132). Non-essential fatty acids also present strong evidence of lasting neurobehavioral impairment. Maternal supplement with hydrogenated vegetable fats, like those found in margarine, induced stress sensitivity and behavioral inflexibility, decreased hippocampal plasticity factors, and increased neuroinflammation at the detriment of cellular function in adult rat offspring (133, 134). Strikingly, rats exposed to hydrogenated fat during gestation or lactation exhibited decreased hippocampal ω-3 PUFA nine weeks after diet intervention (133). The long-term depletion of ω-3 PUFA with maternal hydrogenated fat exposure could be due to interference caused by the presence of trans fats in offspring neural lipid profile. Trans fats are essentially nonexistent in natural food sources, and it is possible that developing brains have an impaired ability to accommodate this unusual lipid form. By extension, it follows that the influence of fatty acid availability on offspring behavioral programming is moderated by the capacity of the developing brain to efficiently optimize neural lipid content. The observed effects of maternal ω-3 deficiency on offspring negative valence behaviors could be compounded when altered fatty acid availability is accompanied by protein deficiency, calorie density, and increased sugar.

Conclusion

Current animal literature supports the programming effect of maternal nutrition on offspring negative valence behaviors. Maternal protein deficiency and fatty acid manipulation exaggerated fear response, with exposure to sustained threat inducing behavioral despair in a potentially sex-dependent manner. Alternatively, WSD exposure during the lactation period impaired risk assessment and response. Furthermore, a model of elevated gestational sugar impaired attention and impulse control. These behavioral alterations are supported by long-term disruptions in neural processes associated with neuropsychiatric disorders. Perinatal nutritional programming resulted in persistent impairments in synaptic plasticity and neurotransmitter systems as well as neurogenic, apoptotic, and brain growth anomalies. Reported outcomes were observed after macronutrient supply was normalized, suggesting that nutrition during critical periods of perinatal development contributes to programming of offspring neurobehavioral impairment.

The strong foundation of literature supports continued investigation of nutritional programming mechanisms. To date, the manner by which maternal macronutrients trigger neurobehavioral abnormalities is under-studied, though evidence from each macronutrient model implicates altered placental function. Importantly, the placenta is regulated by many factors modified by diet, including nutrient availability, maternal stress response, inflammation, and offspring growth factors. Initiating diet manipulation immediately prior to conception or cross-fostering offspring are strengths of animal models and useful in limiting potentially confounding metabolic effects. Current limitations of animal models, including the inconsistency of diet formulations and use of physiologically irrelevant nutritional values, can be improved to enhance translatability. To date, a significant portion of animal studies have exclusively investigated the hippocampus. Examining other brain regions will help generate a more holistic understanding of observed neurobehavioral impairments and increase relevance to human psychopathology. These translational efforts can be further enhanced by increased use of animal models with more complex behavioral phenotypes and similar developmental ontogeny and neuroanatomy to humans. While the presented literature included a near-balanced mix of male and female offspring, sex was often not considered in statistical analysis and future studies should investigate the extensive contributions of sex to neurobehavioral outcomes. Altogether, the presented literature has paved the way for focused, future research to identify the contribution of maternal nutrition to offspring neuropsychiatric risk.