Offspring neuroimmune consequences of maternal malnutrition: potential mechanism for behavioral impairments that underlie metabolic and neurodevelopmental disorders

Abstract

Maternal malnutrition significantly increases offspring risk for both metabolic and neurodevelopmental disorders. Animal models of maternal malnutrition have identified behavioral changes in the adult offspring related to executive function and reward processing. Together, these changes in executive and reward-based behaviors likely contribute to the etiology of both metabolic and neurodevelopmental disorders associated with maternal malnutrition. Concomitant with the behavioral effects, maternal malnutrition alters offspring expression of reward-related molecules and inflammatory signals in brain pathways that control executive function and reward. Neuroimmune pathways and microglial interactions in these specific brain circuits, either in early development or later in adulthood, could directly contribute to the maternal malnutrition-induced behavioral phenotypes. Understanding these mechanisms will help advance treatment strategies for metabolic and neurodevelopmental disorders, especially noninvasive dietary supplementation interventions.

1. Introduction

Maternal malnutrition, either overnutrition or undernutrition, significantly increases offspring risk for metabolic dysfunction and metabolic disease (Delahaye et al., 2008; Gardner et al., 2005; Hale et al., 2015; Hales and Barker, 2001; Ralevski and Horvath, 2015; Ravelli and Osmond, 1999; Ravelli et al., 1976; Roseboom et al., 2000; Spencer, 2013a; Sullivan et al., 2015). Disruptions in the endocrine system are clearly associated with this risk since maternal malnutrition affects hypothalamic circuits (Ralevski and Horvath, 2015). However, maternal diet-induced changes in cognitive behaviors that govern food preference and food intake suggest that higher order brain circuits involved in executive control and reward processing also contribute to metabolic programming (Bayol et al., 2007; Bellinger et al., 2004; Conceição et al., 2016; Palou et al., 2010; Rivera et al., 2016). Understanding how maternal diet affects cognitive control of food intake and processing of food-related reward cues is essential to understanding metabolic dysfunction in species with complex eating behaviors (e.g., humans). In addition to increasing metabolic dysfunction in the offspring, maternal malnutrition increases the risk for neurodevelopmental disorders such as autism, attention deficit/hyperactivity disorder, and schizophrenia (Brown and Susser, 2008; Buss et al., 2012; Krakowiak et al., 2012; Mina et al., 2016; Rivera et al., 2015; Rodriguez et al., 2008; Schaefer et al., 1998; Sullivan et al., 2015; Susser et al., 2014). Hallmark features of these disorders include deficits in prefrontal executive control and reward processing (Aron and Poldrack, 2005; Blum et al., 2008; Brisch, 2014; Happé et al., 2006; Kriete and Noelle, 2015; Orellana and Slachevsky, 2013; Previc, 2006; Schachar et al., 1993; Weinberger, 1987). These behavioral features overlap with maternal malnutrition-induced changes in cognitive control of food intake that can predispose offspring to metabolic disorders. We posit that these executive and reward-related behaviors programmed by maternal diet are central to the risk for both metabolic and neurodevelopmental disorders.

Maternal malnutrition can increase offspring exposure to inflammatory factors, which is suggested to play a mechanistic role in neuroendocrine and neurodevelopmental programming (Spencer, 2013b). Microglia, the resident immune cells of the CNS, are extremely active in establishing synaptic networks during the late embryonic and early postnatal period that program brain function into adulthood (Frost and Schafer, 2016; Lenz et al., 2013; McCarthy et al., 2015; Paolicelli et al., 2011; Schafer et al., 2012; Zhan et al., 2014). This microglial-mediated synaptic wiring engages unique neuroinflammatory mechanisms in the retinogeniculate system (Schafer et al., 2012), preoptic area for sex-specific behavior (Lenz et al., 2013; McCarthy and Wright, 2016; McCarthy et al., 2015), and helps establish prefrontal-hippocampal connectivity (Zhan et al., 2014).

It is unknown exactly if or how maternal diet influences microglial-directed neurodevelopmental processes. However, microglia respond to changes in diet (Valdearcos et al., 2014) and maternal malnutrition exposure coincides with a developmental window of increased microglial-mediated brain patterning. Microglia are actively involved in dopaminergic and cortical wiring (Squarzoni et al., 2014), suggesting that these neuroimmune cells may therefore be neuroanatomically situated tocontrol the development of executive and reward-based behaviors through their influences on dopaminergic and cortical tracts. Furthermore, beyond this developmental window, neuroimmune signaling in adulthood can significantly affect neuronal function and modulate some of these same behaviors (Bachtell et al., 2015; Crews et al., 2011; Hutchinson et al., 2012; Northcutt et al., 2015; Vetreno and Crews, 2012; Vichaya et al., 2014), particularly in the context of peripheral inflammation.

Maternal malnutrition influences neuroimmune signaling and reward-related molecular signatures in the offspring (Bilbo and Tsang, 2010; da Silva et al., 2013; Grissom et al., 2016, 2015, Naef et al., 2013, 2011; Silva et al., 2010; Vucetic et al., 2010a, 2010b). Many of these effects are specific to brain regions that control executive function and reward behaviors (Grissom et al., 2016; McKee et al., 2017; Vucetic et al., 2010a, 2010b). Together, we propose that these neuroimmune and reward-related molecular changes share a common mechanism and impart functional changes, whereby neuroimmune signaling programs executive and reward circuits through microglial-neuron interactions in early development and/or later in adulthood. We review the evidence in the field that supports this notion, along with addressing gaps in the literature to inform future studies.

2. The consequences of maternal diet on offspring neurodevelopment and behavior

2.1. Maternal malnutrition risk for neurodevelopmental disorders

Maternal malnutrition significantly elevates children’s risks for neurodevelopmental disorders. This is especially true for neurodevelopmental disorders such as autism spectrum disorders (ASD), attention deficit/hyperactivity disorder (ADHD), and schizophrenia (Brown and Susser, 2008; Buss et al., 2012; Krakowiak et al., 2012; Mina et al., 2016; Rivera et al., 2015; Rodriguez et al., 2008; Schaefer et al., 1998; Sullivan et al., 2015; Susser et al., 2014), that are increasing in prevalence and cause significant debilitation. Children of mothers exposed to nutrient deficiencies or famine are twice as likely to develop schizophrenia (Brown and Susser, 2008; Susser et al., 2014). Alternatively, high maternal body weight can triple the child’s risk for schizophrenia and ASD (Schaefer et al., 1998) and can increase the risk for development of ADHD (Buss et al., 2012). Maternal obesity significantly increases the severity of ADHD and ASD symptoms in children (Buss et al., 2012; Mina et al., 2016; Rodriguez et al., 2008). Maternal conditions associated with poor diet and obesity, such as hypertension and diabetes, also increase the likelihood of a child experiencing developmental delays or being diagnosed with ASD (Krakowiak et al., 2012). These risks associated with maternal nutritional status remain after controlling for other risk factors, such as cigarette smoking, maternal age, or parity (Rodriguez et al., 2008; Schaefer et al., 1998).

ASD, ADHD, and schizophrenia have a high male prevalence (Previc, 2006) and male offspring are also more susceptible to the effects of maternal malnutrition, suggesting that sex affects offspring vulnerability to neurodevelopmental insults (Bhasin et al., 2009; Desai et al., 2005; Palou et al., 2010; Previc, 2006; Ramírez-López et al., 2016; Sugden and Holness, 2002; Whitaker et al., 2012). The mechanism(s) that underline sex differences in the prevalence of neurodevelopmental disorders is currently unknown, and therefore a topic of significant research interest. Potential mechanisms that have been identified and recently reviewed include a role of genetic sex (Arnold, 2004) or steroid hormones (Davies, 2014; Davis & Pfaff, 2014; Markham, 2012). Additionally, basic neuroimmunological sex differences are a likely candidate mechanism that contributes to the differential prevalence of neurodevelopmental disorders (Hanamsagar & Bilbo, 2016; McCarthy, Nugent, & Lenz, 2017), which will be discussed in more detail in Section 3. As a whole, studying the effects of maternal malnutrition on the offspring may shed light on the etiology of ASD, ADHD, and schizophrenia. Specifically, animal models can help unravel the neurodevelopmental mechanisms that directly contribute to these debilitating disorders.

2.2. Animal models of maternal malnutrition

Maternal malnutrition has two broad categories: overnutrition and undernutrition. Malnutrition models are employed in species such as nonhuman primates, sheep, and most commonly rats and mice (Alfaradhi and Ozanne, 2011). Animal models are advantageous because researchers can manipulate maternal nutritional environment explicitly in gestation or lactation, or encompass both critical periods (Alfaradhi and Ozanne, 2011). Additionally, researchers can study the effects of specific macronutrients versus total calories.

2.2.1. Maternal undernutrition models

Maternal undernutrition models include general caloric restriction, low protein, and large litter size (Bertram and Hanson, 2001; Plagemann, 2006; Spencer, 2013a). Caloric restriction does not affect macronutrient composition but allows for varying degrees of undernutrition. This ranges from mild restriction by withholding approximately 15–30% of total calories, moderate restriction by withholding 50% of total calories, or severe restriction by withholding 70% of total calories (Bertram and Hanson, 2001). Mild restriction does not affect pup birth weight (Palou et al., 2010; Ramírez-López et al., 2016), but more moderate caloric restriction causes pups to be born small for gestational age (Delahaye et al., 2008; Desai et al., 2005).

While general caloric restriction may be relevant to extreme cases of famine, deficiencies of specific macronutrients are of issue as well, as a large proportion of the human population is deficient in protein (Morgane et al., 1978). Protein is the most expensive and least available macronutrient in even well-developed countries, causing people to fill their caloric needs with inexpensive and poor quality carbohydrates and fats (Morgane et al., 1978). Therefore, the maternal low protein (LP) model is highly relevant for studying maternal undernutrition (Stocker et al., 2005). Maternal LP diets allow for isocaloric comparisons while reducing the standard 20% protein composition to approximately 8% protein (Stocker et al., 2005; Sugden and Holness, 2002; Whitaker et al., 2012; Zambrano et al., 2006). Offspring from LP dams are small for gestational age and remain small into adulthood (Bhasin et al., 2009; Bieswal et al., 2006; Whitaker et al., 2012).

Adjustment of litter size on the first postnatal day serves as a natural model for studying maternal-offspring nutrition without manipulating dietary content (Fiorotto et al., 1991; Spencer, 2013a; Widdowson and McCance, 1960). As a model for maternal undernutrition, rearing in large litters (dams with more than 15 pups) decreases offspring growth rate. This results from naturally decreased milk supply and changes breast milk composition available to the pups (Fiorotto et al., 1991). Dams with large litters have higher milk protein ratios and their pups are smaller, leaner, and maintain lower growth rates throughout life (Fiorotto et al., 1991; Spencer, 2013a; Widdowson and McCance, 1960). This litter size model for maternal undernutrition is unique in that it exclusively manipulates the offspring postnatal window without changing the diet ingested by the mother.

2.2.2. Maternal overnutrition models

Maternal overnutrition animal models are designed to mimic nutritional patterns in modern Western society (Williams et al., 2014). These models include high fat diet, junk food diet, and small litter size (Bertram and Hanson, 2001; Gugusheff et al., 2015; Plagemann, 2006; Spencer, 2013a). Maternal high fat diet (HFD) is one of the most well-studied overnutrition models, where fat is substituted for carbohydrate content (Sullivan et al., 2011). Control rodent diets generally consist of 7–20% fat while HFD’s range from 20–60% fat (Parente et al., 2008; Reeves et al., 1993; Sullivan et al., 2011; Vucetic et al., 2010a; Williams et al., 2014). Maternal HFD most often produces larger offspring, but occasionally the offspring are born smaller (Sullivan et al., 2011).

An additional model of maternal overconsumption involves providing dams with a cafeteria or junk food (Alfaradhi and Ozanne, 2011; Bayol et al., 2007; Sullivan et al., 2011). These cafeteria/junk food diets include a wide selection of palatable snacks that are high in fat, salt and sugar. Interestingly, dams overconsume carbohydrate and fat at the expense of protein and offspring are born lighter, similar to maternal undernutrition in the form of a LP diet (Bayol et al., 2007). Finally, pups raised in small litters have increased access to a maternal milk supply that differs from that of large or mediumsized litters, representing a model for early postnatal overfeeding (Plagemann 2006, Fiorotto 1991, Widdowson 1960). Together, these maternal malnutrition models produce both shared and dissociable phenotypes in the offspring. This makes them invaluable assets for understanding the mechanistic link between maternal diet and subsequent changes in offspring neuroendocrine function and behaviors that may underlie neurodevelopmental disorders.

2.3. Maternal malnutrition and obesity risk in the offspring

Maternal malnutrition, either in the form of undernutrition or overnutrition, changes levels of neuroendocrine factors during sensitive developmental periods and increases the risk for obesity and metabolic disease (Hales and Barker, 2001; Sullivan et al., 2015). Maternal overnutrition and obesity are associated with increased leptin, glucose, and proinflammatory cytokines that directly transfer to the fetus and can alter fetal-derived neuroendocrine signals to promote obesity in the offspring (Hale et al., 2015; Ralevski and Horvath, 2015; Spencer, 2013a; Sullivan et al., 2015). Paradoxically, maternal undernutrition and protein restriction are also associated with increased leptin and insulin signaling and obesity in the offspring (Delahaye et al., 2008; Gardner et al., 2005; Hales and Barker, 2001; Ralevski and Horvath, 2015; Ravelli and Osmond, 1999; Ravelli et al., 1976; Roseboom et al., 2000). Maternal malnutrition perpetuates neuroendocrine impairment and metabolic syndrome in the offspring. Some effects are immediate but many persist or even emerge in adulthood.

2.3.1. Metabolic effects of maternal undernutrition

Mild caloric restriction does not change pup birth weight but moderate restriction reduces offspring birth weight, impairs the postnatal leptin surge, and decreases circulating leptin and ghrelin in the pups (Delahaye et al., 2008; Desai et al., 2005). Independent of pup birth weight, maternal caloric restriction increases offspring body weight and fat mass in adulthood, particularly in males (Desai et al., 2005; Palou et al., 2010; Ramírez-López et al., 2016). This is accompanied by increased leptin (Desai et al., 2005), glucose intolerance, and insulin resistance in adulthood (Gardner et al., 2005). Interestingly, these adult effects are preceded by increased insulin in early adolescence (Palou et al., 2010). LP offspring have increased insulin sensitivity and glucose tolerance early in life (Stocker et al., 2005; Zambrano et al., 2006). However, aging adult LP offspring later develop insulin resistance, glucose intolerance, increased food intake, greater body weight, and increased adiposity (Bhasin et al., 2009; Stocker et al., 2005; Sugden and Holness, 2002; Whitaker et al., 2012; Zambrano et al., 2006). Similar to maternal caloric restriction, these effects are particularly evident in males (Bhasin et al., 2009; Sugden and Holness, 2002; Whitaker et al., 2012).

In some models, the small birth weight offspring from LP and calorie restricted dams have accelerated catch-up growth, leading to subsequent metabolic dysfunction in adulthood. These animal models are consistent with maternal undernutrition in humans predisposing offspring to obesity, diabetes, and metabolic syndrome later in life (Hales and Barker, 2001; Ravelli and Osmond, 1999; Ravelli et al., 1976). This is in contrast to the effects seen with large litter size. Unlike maternal protein or calorie restriction that increase the risk of metabolic disease states in the offspring, rearing pups in large litters may offer protection from obesity (Patterson et al., 2010). This may be due to the more naturalistic aspects of the model, whereby the breast milk composition may be able to compensate for an increased number of pups (Fiorotto et al., 1991).

2.3.2. Metabolic effects of maternal overnutrition

Interestingly, maternal overnutrition causes similar metabolic phenotypes as the LP and calorie restriction models for undernutrition. Regardless of birth weight, offspring from dams fed HFD develop increased fat mass and body weight, glucose intolerance, insulin resistance, and cumulative signs of metabolic syndrome (Alfaradhi and Ozanne, 2011; Bertram and Hanson, 2001; Sullivan et al., 2011; Williams et al., 2014). Maternal HFD often induces maternal obesity, however the metabolic effects on the offspring can occur independently of maternal metabolic complications (Alfaradhi and Ozanne, 2011; Sullivan et al., 2011). Offspring from dams fed a junk food diet later develop an obesogenic phenotype indicative of metabolic syndrome, similar to those from dams fed LP or HFD (Alfaradhi and Ozanne, 2011). Offspring from small litters develop symptoms of metabolic syndrome, such as increased body weight, hyperphagia, glucose intolerance and hyperinsulinemia (Plagemann et al., 1999). Together, these animal models of maternal malnutrition demonstrate that maternal diet can impair offspring neuroendocrine development and increase the risk for metabolic dysfunction into adulthood, leading to an increased risk for obesity.

2.4. Maternal malnutrition consequences for brain and behavior: reward and PFC function

Obesity and metabolic complications can be treated with proper diet and exercise, however behavioral intervention is a major challenge for treatment efficacy and maintenance of weight loss in human obese patients (Seidell and Halberstadt, 2015). While animals models reliably reproduce the metabolic disturbances seen with human maternal malnutrition, the exact mechanisms that cause these phenotypes are unknown (Sullivan et al., 2011). Interestingly, a constellation of brain and behavioral changes accompany the maternal malnutrition-induced neuroendocrine impairments in the offspring, including behavior shifts that can directly lead to overeating and obesity.

2.4.1. Offspring food preferences (Table 1)

Table 1.

Summary of adult offspring behavioral changes following various maternal malnutrition models. These behavioral shifts impact food preference and intake, suggestive of obesogenic eating habits, loss of executive control, and altered reward processing. Maternal malnutrition also affects executive and reward based behaviors, as indicated by changes in motivation, reward sensitivity, executive control, attention, impulsivity, and social behavior. (M = male; F = female; NHP = nonhuman primates; 5CSRTT = five choice serial reaction time task; ASST = attentional set shifting task)

Behavior	Maternal Undernutrition Model	Maternal Overnutrition Model
	LP	CR	HFD	JUNK FOOD	SM LITTER
Food selection	Rats: M/F increased HFD preference (Bellinger et al., 2004)	Rats: M increased fat preference (Palou et al., 2010) Rats: M increased palatable food intake (Alves et al., 2015)	Mice: M increased sucrose and HFD preference (Vucetic et al., 2010) NHP: Increased juvenile palatable food intake (Rivera et al., 2016)	Rats: increased sugar, fat and salt preference, junk food hyperphagia (Bayol et al., 2007) Rats: M/F increased palatable food consumption (Ong& Muhlhausler, 2011) Rats: M/F decreased opioid antagonist-induced suppression of food intake (Gugusheff et al., 2013)	Rats: M increased HFD preference (Conceicao et al., 2016) Rats: M Increased sweet food intake with acute stress (Portella et al., 2015)
Motivation	Rats: increased motivation once task acquired (da Silva et al., 2013) Rats: increased motivation measured by response speed for food reward (de Melo Martimianoetal., 2015) Mice: M/F increased sign tracking for food reward cue (Grissom et al., 2015)		Rats: M increased motivation for fat pellets (Naef et al., 2011) Mice: M/F decreased motivation for food reward (Grissom et al., 2015)
Reward sensitivity	Mice: Increased locomotor response to cocaine (Vucetic et al., 2010b)		Mice: M decreased locomotor response to amphetamine (Naef et al., 2008)
Executive function	Mice: M/F increased errors in 5CSRTT (Grissom et al., 2015)	Rats: F decreased trials to criterion in ASST reversion (Alves et al., 2015)	Mice: M/F increased errors in 5CSRTT (Grissom et al., 2015)
Attention	Mice: M/F increased inattentive errors in 5CSRTT (Grissom et al., 2015)		Mice: No difference in inattentive errors in 5CSRTT (Grissom et al., 2015)
Impulse control	Mice: No difference in impulsive errors in 5CSRTT (Grissom et al., 2015)		Mice: M increased impulsive errors in 5CSRTT (Grissom et al., 2015)
Social behavior		NHP:M increased aggression (Huber et al., 2015)	Mice: F decreased social interaction (Kang et al., 2014)		Rats: M decreased social play (Carvalho et al., 2016)

Maternal malnutrition consistently increases adult offspring consumption and preference for obesogenic and palatable foods, such as those high in fat and sugar. This is evident across many animal models for maternal malnutrition (see Table 1). In nonhuman primates, maternal HFD and maternal obesity increase offspring preference for high fat and high sugar food (Rivera et al., 2016). In mice, maternal HFD increases male offspring (females not studied) preference for sucrose and HFD in adulthood (Vucetic et al., 2010a). In rats, adult offspring from dams fed junk food diet have increased preference and consumption of foods high in sugar, fat, and salt (Bayol et al., 2007; Ong and Muhlhausler, 2011). When offspring are exposed to junk food after being weaned to chow diet, they are hyperphagic as adults (Bayol et al., 2007). Adult male offspring from small rat litters consume more HFD in a food preference test (Conceição et al., 2016) and increase sweet food consumption with acute stress (Portella et al., 2015). Also in rats, maternal LP increases adult male and female offspring preference for high fat food (Bellinger et al., 2004). Adult male offspring from calorie restricted rat dams have increased fat preference and increased consumption of palatable food, with no differences in the female offspring (Alves et al., 2015; Palou et al., 2010). With respect to maternal malnutrition and sex-specific offspring food preference, males are more often studied or are more sensitive to these effects than females.

2.4.2. Food intake control through executive and reward function

This behavioral shift towards overconsuming highly palatable foods suggests that maternal malnutrition causes deficits in cognitive control of food intake and food-related decision making. Highly palatable foods activate reward centers in the brain, making strong self-control essential for preventing obesogenic eating habits (Ziauddeen et al., 2015). The prefrontal cortex (PFC) is imperative for self-control of food-based decision making and food reward processing (Ziauddeen et al., 2015). For successful self-regulation of food intake, the PFC exerts what is called executive control over impulsive behaviors, while taking into account long term objectives and possible repercussions (Ziauddeen et al., 2015). The PFC sends inputs to the ventral tegmental area (VTA) and the nucleus accumbens (NAc), major dopaminergic centers in the brain that control appetitive and reward behaviors (Ikemoto and Panksepp, 1999). Maternal malnutrition strongly influences offspring food reward processing and impairs executive function, accompanied by brain region-specific changes in the PFC and reward circuits (da Silva et al., 2013; Grissom et al., 2015; Naef et al., 2013, 2011; Rivera et al., 2016; Vucetic et al., 2010a, 2010b). Overall, maternal malnutrition seems to cause dopaminergic hypofunction in reward circuits, which may lead to increased self-administration of rewarding food stimuli to compensate. These changes are evident in PFC dopaminergic projections and also manifest as PFC deficits (Carvalho et al., 2016; Grissom et al., 2015; Kang et al., 2014; Rivera et al., 2016).

2.4.3. Appetitive and reward-related behavioral changes (Table 1)

Maternal malnutrition changes appetitive-based reward processing and motivation (see Table 1). In rats, maternal HFD increases male adult offspring (females not included) operant responding and motivation for fat pellets (Naef et al., 2011). LP rat offspring show initial deficits in learning a food reward task, but have increased motivation for the food reward (da Silva et al., 2013; de Melo Martimiano et al., 2015). In mice, both male and female offspring from HFD dams respond less in fixed ratio stimulus-food reward pairings and are less motivated to work for food reward, indicated by lower breakpoints in a progressive ratio task (Grissom et al., 2015). Also in mice, both male and female offspring from LP dams have increased attention to signs that predict food reward (Grissom et al., 2015). In rats, maternal junk food diet blocks opioid antagonist-induced suppression of food intake (Gugusheff et al., 2013), suggesting that maternal diet also affects appetitive behaviors in the context of drugs that regulate reward systems. Overall, the direction of these changes depends on species or model, but nonetheless the development of these reward-learning appetitive behaviors are sensitive to maternal diet.

2.4.4. Higher order executive function, attention, impulse control, and social behavior (Table 1)

In addition to changes in reward function and reward-related molecules, maternal malnutrition produces distinct deficits in PFC function that are dependent on dietary model (see Table 1). The five choice serial reaction time task (5CSRTT) is a behavioral test designed to assess multiple realms of PFC function, from attention to impulsivity (Robbins, 2002). In mice, both maternal LP and HFD adult male and female offspring have increased errors on the 5CSRTT (Grissom et al., 2015). LP offspring have more errors in the form of response omissions, indicating they are inattentive (Grissom et al., 2015), while HF offspring have increased incorrect trials and males specifically have more premature errors, indicating that they are impulsive (Grissom et al., 2015). Interestingly, in female offspring, maternal caloric restriction actually accelerates learning in a behavioral flexibility task (Alves et al., 2015). This suggests that males offspring are more sensitive to PFC behavioral deficits with maternal malnutrition.

The PFC is also essential for social behavior and the development of social behavior (Grossmann, 2013). Maternal malnutrition affects offspring sociability (see Table 1). In nonhuman primates, maternal calorie restriction increases aggression in male offspring (Huber et al., 2015). In rats, small litter rearing decreases social play behavior in adolescent male offspring (Carvalho et al., 2016). In mice, both male and female offspring from HFD dams have decreased social interaction with an unfamiliar mouse (Grissom and Reyes, 2013; Kang et al., 2014). Males from LP dams display a similar decrease in social interaction and are also hyperactive in general locomotor tests (Fraňková, 1973; Grissom and Reyes, 2013; Kang et al., 2014). This broad scope of changes in PFC-mediated behaviors indicates that PFC development is extremely sensitive to maternal nutritional status.

2.4.5. Molecular changes in executive and reward circuits (summarized in Table 2)

Table 2.

Summary of maternal malnutrition-induced offspring brain changes in regions that control executive function, reward, and food intake. Highlighted effects include dopamine and opioid related molecules. (M = male; F = female; TH = tyrosine hydroxylase; DA = dopamine; DAT = dopamine transporter; D1/2 = dopamine receptor 1/2; MOR = μ-opioid receptor)

Brain Region	Maternal Undernutrition Model	Maternal Overnutrition Model
	LP	CR	HFD	JF	SM LFTTER
PFC	Increased TH expression (Vucetic et al., 2010b) Increased DA, decreased DA turnover (Vucetic et al., 2010b)	F increased TH with palatable food (Alves et al., 2015)	Decreased TH innervation (Rivera et al., 2016) Decreased Dl, D2 (Vucetic et al., 2010a, Rivera et al., 2016) Increased opioid-related genes (Vucetic et al., 2010a)
NAc	Increased neuronal activation to cocaine (Vucetic et al., 2010b) Increased TH expression (Vucetic et al., 2010b) Increased DAT expression (Vucetic et al., 2010b) Increased MOR binding (Thanos et al., 2016) Increased D1 binding (Thanos et al., 2016)	F increased TH with palatable food (Alves et al., 2015) M increased TH (Alves et al., 2015)	Blunted DA response to food reward cue (Naef et al., 2013) Decreased Dl, D2 (Vucetic et al., 2010a) Increased opioid-related genes (Vucetic et al., 2010a) Increased TH, DA, and DOPAC (Naef et al., 2008) Increased Dl binding (Thanos et al., 2016)	Increased MOR (Gugusheff et al., 2013) Increased MOR in adolescence with decrease in adult (Ong& Muhlhausler, 2011) Decreased DAT in adolescence with increase in adult (Ong& Muhlhausler, 2011)	Decreased D2 protein (Portella et al., 2015) Decreased DAT (Conceição et al., 2016)
VTA	Increased TH expression (Vucetic et al., 2010b) Increased DAT expression (Vucetic et al., 2010b) Increased MOR binding (Thanos et al., 2016)		Decreased Dl, D2 (Vucetic et al., 2010a) Decreased D2 (Naef etal., 2011) Increased TH (Naef et al., 2008)	Decreased MOR (Gugusheff et al., 2013)	Increased TH protein (Portellaetal., 2015) Decreased TH with excess HFD intake (Conceição et al., 2016)
HYP	Increased TH expression (Vucetic et al., 2010b) Increased MOR binding (Thanos et al., 2016)	Decreased proenkephalin in fetal PVN (Hawkins et al., 2001)	Increased opioid-related genes (Vucetic et al., 2010a)

In the brain, HFD offspring have blunted NAc dopamine responses to conditioned food reward cues, indicating dopamine hypofunction to anticipatory food cues (Naef et al., 2013). Offspring from maternal HFD have altered dopamine gene expression, indicated by decreased expression of dopamine receptor 1 and 2 (D1, D2) in the NAc and PFC but increased dopamine transporter (DAT) expression in the PFC, VTA, and NAc (Vucetic et al., 2010a). They also have increased opioid-related gene expression in the PFC, NAc, and hypothalamus (Vucetic et al., 2010a). Maternal HFD decreases offspring density of dopaminergic fibers (measured by tyrosine hydroxylase- TH) that project to the PFC (Rivera et al., 2016), consistent with dopamine hypofunction especially in PFC connections. However, HFD offspring have increased TH, dopamine, and D1 binding in the NAc and increased TH in the VTA (Naef et al., 2011, 2008; Thanos et al., 2016). This suggests that maternal HFD may shift offspring reward circuits by dampening PFC executive control and enhancing the salience of reward-related stimuli.

Small litter male offspring (females not included) that overconsume HFD in adulthood have decreased tyrosine hydroxylase (TH) in the VTA and decreased DAT in the NAc (Conceição et al., 2016). Without HFD exposure, small litter offspring have increased TH in the VTA and decreased D2 in the NAc (Conceição et al., 2016; Portella et al., 2015). Adolescent offspring from dams fed a junk food diet have increased μ-opioid receptor and decreased DAT expression in the NAc (Gugusheff et al., 2013; Ong and Muhlhausler, 2011). Interestingly, these changes reverse in adulthood, as adult offspring have decreased expression of NAc μ-opioid receptor and increased expression of NAc DAT (Ong and Muhlhausler, 2011).

Male adult offspring from LP dams have increased locomotion and NAc neuronal activation to cocaine (Vucetic et al., 2010b). These mice have increased TH expression in VTA, PFC, NAc, and hypothalamus (Vucetic et al., 2010b). They also have increased dopamine in the PFC and decreased dopamine turnover, paired with increased dopamine transporter expression in the VTA and NAc (Vucetic et al., 2010b). Furthermore, LP offspring have increased D1 binding in the NAc and increased μ-opioid receptor binding in the NAc, VTA, and hypothalamus (Thanos et al., 2016). Maternal calorie restriction decreases fetal proenkephalin in the hypothalamus (Hawkins et al., 2001) increases TH in the PFC and NAc in adulthood when these offspring are exposed to palatable food (Alves et al., 2015). While the various animal models of maternal malnutrition may produce slightly different patterns of molecular changes, they consistently affect dopaminergic function in the PFC and reward centers. These changes are sensitive to stimuli in adulthood, such as palatable food access, demonstrating that maternal malnutrition affects how the offspring process rewards in adulthood.

Taken together, maternal malnutrition induces molecular changes in executive and reward circuits while affecting PFC-mediated behaviors and reward processing. These behavioral deficits likely underlie the etiology of both neurodevelopmental disorders and obesogenic eating habits. Therefore, investigating the mechanisms through which maternal malnutrition causes these specific behavioral changes may improve treatment options for a broad scope of pervasive and debilitating human disorders.

3. Maternal immune programming of offspring neurodevelopment

3.1. Maternal malnutrition and offspring cytokine exposure

Offspring metabolic and behavioral phenotypes emerge following maternal malnutrition likely through neurobiological programming. Maternal malnutrition alters offspring exposure to metabolic hormones, but also drastically affects the offspring’s exposure to immune-related factors (e.g., cytokines). Immune molecules such as cytokines are active mediators in metabolic signaling pathways and simultaneously affect brain development and behavior. Importantly, inflammatory factors can affect both executive function and reward-related behaviors (Hutchinson et al., 2012; Northcutt et al., 2015; Vetreno and Crews, 2012), making the neuroimmune system a potential candidate for maternal-to-offspring programming.

3.1.1. Maternal transfer of inflammatory molecules (Table 3)

Table 3.

Summary of maternal malnutrition-induced inflammatory changes, with multiple routes to influence the offspring. These effects reach the maternal plasma, breast milk, placenta, and developing fetus. Offspring display peripheral and neuroimmune changes into adulthood.

Tissue	Maternal Undernutrition Model	Maternal Overnutrition Model
	LP	CR	HFD	JF	SM LITTER
Maternal			Increased plasma IL-1β, TNFα, and IL-6 (Reynolds et al., 2015) Increased mammary gland TNFα and IL-6 (Hernandez et al., 2012)	Decreased plasma IL-6 and Cxcl2 (Crew et al., 2016)
Placenta		Decreases prostaglandin E receptor 1 methylation (Chen et al., 2013)	M increased TNFα (Reynolds et al., 2015) NHP: Increased IL-1β and TLR4 (Frias et al., 2011)	M decreased TLR4 (Crew et al., 2016) Increased IL-6 (Crew et al., 2016)
Offspring Peripheral	Increased adult plasma TNFα (Silva et al., 2010) Increased basal and LPS-induced neutrophil iNOS, NO, NF-κB p65 (Silva et al., 2010) Decreased leukocyte trafficking with LPS (Silva et al., 2010)	Sheep: Increased adipose TLR4 on PI, 30, and adult; increased CCR1 in adult (Sharkey et al., 2009)	Increased fetal plasma IL-6 (Sanders et al., 2014) Increased fetal adipose TNFα, CCR2 (Murabayashi et al., 2013) Increased adult LPS-induced plasma IL-1β, IL-6 and CRP (Bilbo & Tsang, 2010)		Decreased P10 plasma TNFα (Argente-Arizón et al., 2016) No differences at weaning on P21 (Kayser et al., 2015) Increased adult adipose TNFα, IL-6, and TNF-R1 (Argente-Arizón et al., 2016; Boullu-Ciocca et al., 2008) Increased adult adipose macrophage numbers with HFD (Kayser et al., 2015)
Offspring Brain	Decreased M adult LPS-induced VTA prostaglandin-related genes (Grissom et al., 2016)		Increased PI hippocampal CD11b and TLR4 expression (Bilbo & Tsang, 2010) F adult increased whole brain IL-1β and TNFα (Kang et al., 2014) F adult increased amygdala lba-1 expression (Kant et al., 2014) Increased adult LPS-induced hippocampal IL-1β, IL-6 and CRP (Bilbo STsang, 2010) Increased M adult LPS-induced PFC TNFα, Cxcl10, Ccl2 (Grissom et al., 2016)

Maternal malnutrition affects inflammatory markers in both the dam and placenta (see Table 3). With the maternal junk food model, dams have decreased plasma IL-6 and chemokine ligand 2 (Cxcl2) levels, male placentas have decreased toll-like receptor 4 (TLR4) expression, and together male and female placentas have increased IL-6 (Crew et al., 2016). Maternal calorie restriction decreases placental prostaglandin E receptor 1 methylation (Chen et al., 2013), which may have implications for intrauterine cytokine interactions (Keelan et al., 2003). In rats, dams fed HFD throughout gestation have increased plasma IL-1β, TNFα, and IL-6 at E18 (Reynolds et al., 2015). This maternal HFD exposure decreases placental weight and increased placental TNFα expression, specifically in male and not female offspring (Reynolds et al., 2015). In nonhuman primates, maternal HFD increases placental IL-1β and TLR4 expression (Frias et al., 2011) In conjunction with increasing plasma IL-6 in the dam, maternal HFD increases plasma IL-6 in the fetus at E17.5 (Sanders et al., 2014). Furthermore, maternal HFD increases TNFα and IL-6 expression in the mammary glands of lactating rats (Hernandez et al., 2012), indicating that cytokines are likely passed to offspring during lactation. There are multiple routes through which peripheral cytokines can affect the brain, including transport at the blood brain barrier (Banks, 2015), signaling at the blood brain barrier via engagement with endothelial and perivascular cells (Serrats et al., 2010), as well as direct action on the vagus nerve (Maier et al., 1998), which signals directly to the CNS. This demonstrates that maternal diet-induced elevations in proinflammatory cytokines have numerous points of entry into the fetal brain, where they can readily affect fetal brain function and development.

3.1.2. Basal inflammatory effects in offspring (Table 3)

Maternal malnutrition increases basal inflammatory markers in newborn offspring (see Table 3), an effect that is evident in fetal adipose tissue, suggesting a role in offspring metabolic function. For example, maternal HFD increases fetal adipose TNFα and chemokine receptor 2 (CCR2) expression (Murabayashi et al., 2013). Maternal caloric restriction increases adipose toll-like receptor 4 (TLR4) expression in sheep offspring on the day of birth, an effect that persists into adulthood (Sharkey et al., 2009). Furthermore, these proinflammatory effects often reach the brain and again persist into adulthood. With maternal HFD, pups on postnatal day 1 have increased expression of microglial marker CD11b in the hippocampus (Bilbo and Tsang, 2010). These pups also have increased hippocampal expression of TLR4 (Bilbo and Tsang, 2010), an innate immune pattern recognition receptor that mediates immune-metabolic crosstalk (Milanski et al., 2009; Saberi et al., 2010). In the LP maternal undernutrition model, offspring have increased plasma TNFα levels in adulthood (Silva et al., 2010). In adult female offspring, maternal HFD increases brain IL-1β and TNFα expression while specifically increasing amygdala expression of microglial marker Iba-1 (Kang et al., 2014).

Perinatal overnutrition through small litter rearing either causes no inflammation early in life or decreases plasma TNFα in P10 offspring of both sexes (Argente-Arizón et al., 2016; Kayser et al., 2015). However, by adulthood, small litter rearing increases TNFα, TNF receptor 1, and IL-6 adipose tissue expression and increases white adipose tissue inflammation in response to HFD feeding (Argente-Arizón et al., 2016; Boullu-ciocca et al., 2008; Kayser et al., 2015). Overnutrition alone without maternal transmission of inflammatory stimuli seems insufficient to induce fetal or postweaning inflammation, but these effects may still develop later in adulthood.

3.1.3. Induced inflammatory responses in offspring (Table 3)

While baseline differences in the immune profiles of offspring from malnourished dams may not be evident, the effects of maternal malnutrition on the offspring’s peripheral and central immune system are often revealed with a subsequent inflammatory challenge, such as with bacterial derived lipopolysaccharide (LPS) (see Table 3). LP offspring have altered peripheral immune responses to LPS, indicated by a reduction in leukocyte trafficking to the peripheral site of LPS injection (Silva et al., 2010). However, these LP offspring have increased basal and LPS-induced blood neutrophil expression of proinflammatory mediators iNOS, NO, and NF-κB p65 (Silva et al., 2010). Adult male and female offspring from HFD-fed dams have exaggerated peripheral and central proinflammatory cytokine responses to an LPS challenge, demonstrated by elevated IL-1β, IL-6 and CRP in plasma and hippocampus (Bilbo and Tsang, 2010). Importantly, these altered offspring immune responses are also observed in brain regions that control executive function and reward processing. In the PFC, maternal HFD but not low protein diet increases adult male offspring LPS-evoked expression of TNFα and proinflammatory chemokines Cxcl10 and Ccl2 (Grissom et al., 2016). In the VTA, maternal low protein diet actually decreases male offspring LPS-evoked gene expression of proinflammatory prostaglandin related genes (Grissom et al., 2016). Therefore, maternal malnutrition may prime the offspring brain in such a way as to alter future responses to inflammatory challenges, such as peripheral viral or bacterial exposure or high fat diet/obesity, that can increase circulation of peripheral cytokines.

3.2. The role of microglia and inflammatory signals in brain development

In addition to responding to peripheral immune stimuli, neuroimmune signaling is crucial for proper brain development. Proinflammatory cytokines serve as signals for brain development, regulating processes such as cellular differentiation, survival, and axonal guidance (Parker-Athill and Tan, 2011). Microglia, the resident immune cell of the CNS, are actively involved in brain development. These immune cells migrate from the embryonic yolk sac to the brain on E9.5, before any resident cells in the CNS start to differentiate (Frost and Schafer, 2016). Once established as the brain’s resident immune cells, microglia actively participate in the development of many different CNS cells types, from neurons to glia and vascular cells (Frost and Schafer, 2016).

3.2.1. The role of microglia in brain development

Microglial signaling is critical for synaptic development in embryonic and early postnatal life, while also controlling maturation of functional brain connectivity and complex behavior into adulthood (Paolicelli et al., 2011; Schafer et al., 2012; Zhan et al., 2014). Interestingly, the role of microglial-mediated development varies by neuroanatomical brain region (Frost and Schafer, 2016). In the retinogeniculate system, micrgolia engulf presynaptic inputs during early postnatal development in an activity-dependent manner via complement receptor 3/complement 3 signaling (Schafer et al., 2012). In the preoptic area, microglia masculinize the brain through prostaglandin E₂ signaling to promote male behaviors in adulthood (Lenz et al., 2013; McCarthy and Wright, 2016). In the hippocampus, microglial fractalkine receptor (Cx3cr1) is necessary for the appropriate development of dendritic spines and excitatory synapses (Paolicelli et al., 2011). Cx3cr1 knockout (KO) mice have increased hippocampal spine density and excitatory synapses paired with decreased or immature synaptic plasticity and excitability, indicating that microglial Cx3cr1 is essential for synaptic pruning (Paolicelli et al., 2011).

3.2.2. The role of microglia in PFC and reward circuit development

Microglia also likely play a major role in the development of prefrontal cortical and reward circuits. During embryonic life, microglia fine-tune dopaminergic axonal wiring and cortical interneuron distribution (Squarzoni et al., 2014). Microglia preferentially organize in dopaminergic tracts at E14.5 and subsequently infiltrate the cortex at E16.5 (Squarzoni et al., 2014). Microglial activity in embryonic life prevents overgrowth of forebrain dopaminergic tracts and dictates appropriate cortical interneuron organization, processes that are sensitive to maternal immune activation by LPS (Squarzoni et al., 2014). The microglial fractalkine receptor mediates PFC development. Cx3cr1 KO mice have decreased PFC functional connectivity with the hippocampus and NAc in adulthood, assessed in vivo by both local field potential coherence and BOLD fMRI (Zhan et al., 2014). This decreased functional connectivity is consistent with neurodevelopmental disorders, such as autism and schizophrenia (Dinstein et al., 2011; Meyer-Lindenberg et al., 2005; Schipul et al., 2011). PFC connectivity dictates social interaction behavior, which is impaired in neurodevelopmental disorders (Zhan et al., 2014). The connectivity deficits in Cx3cr1 KO mice correspond to social deficits, demonstrated by decreased juvenile Cx3cr 1 KO social interaction preference for their mothers or conspecifics (Zhan et al., 2014).

3.2.3. Sex differences in immune-mediated brain development and relevance to neurodevelopmental disorders

Because maternal malnutrition and obesity are both strongly associated with a proinflammatory state, microglia and inflammatory pathways may be a common mechanism linking maternal malnutrition and obesity with the development of offspring metabolic and behavioral phenotypes. Interestingly, masculinization of the brain recruits more inflammatory processes than feminization, demonstrated by males having an increase in neuroinflammatory signals during normal brain development relative to females (McCarthy et al., 2015). As mentioned earlier, male offspring are often more susceptible to the metabolic and behavioral effects of maternal malnutrition (Bhasin et al., 2009; Desai et al., 2005; Palou et al., 2010; Ramírez-López et al., 2016; Sugden and Holness, 2002). The increased dependence of the male brain on immune signals for proper development may predispose males to insults that affect inflammatory status, such as maternal malnutrition. This may contribute to the increased male prevalence in neurodevelopmental disorders with a suspected neuroinflammatory etiology, such as ASD, ADHD, and schizophrenia (Fillman et al., 2013; Monji et al., 2013; Morgan et al., 2012, 2010; Previc, 2006; Suzuki et al., 2013)

3.3. Microglia and neuroimmune signals affecting behavior

Since microglia and neuroimmune signals play such a critical role in brain development, early shifts in these inflammatory pathways induced by maternal malnutrition have the capacity to change cognitive control and behavior into adulthood. These neuroimmune changes may either persist or first emerge in adulthood, but we propose that they likely play a large role in the behavioral phenotypes. As mentioned earlier, executive function and reward processing are key behavioral functions that are compromised in neurodevelopmental and metabolic disorders that are associated with maternal malnutrition. Neuroinflammatory signals and microglia specifically influence brain centers and behaviors linked to executive and reward function.

3.3.1. Similarity between highly palatable foods and drugs of abuse

Research to date has yet to systematically investigate how neuroimmune pathways contribute to the executive and reward-based behavioral consequences of maternal malnutrition. However, maternal malnutrition – especially highly palatable foods – activate similar CNS pathways and have similar programming effects for the offspring as drugs of abuse (Gugusheff et al., 2015). Drugs of abuse or improper diet activate microglia by initiating inflammatory signaling cascades that can alter neuronal transmission (Bachtell et al., 2015). These inflammatory molecules directly regulate executive function and reward-mediated behaviors (Crews et al., 2011; Hutchinson et al., 2012; Northcutt et al., 2015; Vetreno and Crews, 2012).

3.3.2. Microglial signaling cascades in drug-related responses

Drugs of abuse activate the toll-like receptor 4 (TLR4), innate immune system receptors on microglia that initiate a cytokine signaling cascade (Bachtell et al., 2015; Hutchinson et al., 2010). Cocaine increases proinflammatory cytokine expression of IL-1b in the VTA in a TLR4-dependent fashion, and TLR4 activation in the VTA induces a strong dopamine response in the NAc in a similar manner to cocaine (Northcutt et al., 2015). Systemic or intra-VTA antagonism of TLR4 or IL-1b receptors successfully blocks the NAc dopamine response to cocaine (Northcutt et al., 2015). TLR4 antagonism also attenuates the morphine-induced dopamine response (Hutchinson et al., 2012). Behaviorally, TLR4 antagonism or genetic inactivation mutations attenuate conditioned place preference and decrease self-administration with cocaine, morphine, and other opioid analgesics (Hutchinson et al., 2012; Northcutt et al., 2015). Microglial inhibitor minocycline produces similar effects to TLR4 antagonism, suggesting TLR4 signaling within microglia mediates reward behaviors in the context of addictive drugs (Northcutt et al., 2015).

3.3.3. Microglial signaling cascades in executive and reward processes

Drugs of abuse induce TLR4 signaling in reward pathways, but this signaling may also play a role in executive function. Chronic ethanol administration increases TLR4 expression in the medial PFC and this TLR4 expression significantly correlates with deficits in reversal learning (Vetreno and Crews, 2012). This suggests that TLR4 signaling may mediate executive function deficits caused by substance abuse. However, it is unknown whether microglia specifically mediate these effects. The PFC is a key brain region for the behavioral basis addiction but is also exquisitely sensitive to the effects of stress, indicating that PFC-driven responses to stress are also important for executive and reward function (Arnsten, 2009; Goldstein and Volkow, 2012). Chronic stress causes working memory deficits in a PFC-mediated task in conjunction with increased microglial activation in the PFC, effects that are ameliorated by minocycline (Hinwood et al., 2012). Together, this shows that microglia and inflammatory pathways regulate executive and reward function in the context of environmental insults.

3.3.4. Early life programming of neuroimmune signals that affect executive and reward processes

Early life events modify adulthood neuroimmune responses to drugs of abuse, which in turn control drug-seeking behavior. For example, increased care in the form of neonatal handling attenuates the morphine-induced pro-inflammatory cytokine and chemokine response in the NAc and decreases conditioned place preference (CPP) reinstatement (Schwarz et al., 2011). Interestingly, adulthood TLR4 antagonism with ibudilast recapitulates the effects of neonatal handling, suggesting that TLR4 plays a major role in the expression of neuroimmune responses and drug behaviors that are sensitive to early life events (Schwarz et al., 2011). This strongly proposes that early life changes in nutrition (through exposure to maternal malnutrition) may engage similar neuroinflammatory mechanisms to influence adulthood reward and executive behaviors. It is possible that maternal malnutrition may induce microglial and TLR4 activation in the offspring. This TLR4 activation of microglia may in turn influence neuronal signaling to cause the observed executive function and reward processing deficits in HFD offspring. Future studies will investigate this as a potentially causative link.

4. Dietary supplementation

In translating this research to the human population, the most attractive avenues are minimally invasive interventions that can reverse maternal malnutrition-induced deficits. Therefore, nutritional supplementation is an appealing strategy to reverse dietary deficiencies caused by maternal malnutrition. Maternal HFD and obesity cause specific micronutrient deficits in both the mother and fetus (Grant et al., 2011; Sen et al., 2014). The research is promising to show that correcting these deficits either by maternal or fetal supplementation can reverse behavioral and metabolic consequences in the offspring (Carlin et al., 2013; Delaney et al., 2013, 2012; Heerwagen et al., 2013; McKee et al., 2017).

4.1. Omega-3 fatty acid supplementation

In nonhuman primates, maternal HFD decreases levels of omega-3 fatty acids (Grant et al., 2011). Omega-3’s are essential polyunsaturated fatty acids (PUFAs) that are crucial during pregnancy because they are necessary for brain development (Coletta et al., 2010). Of the omega-3’s, docosahexaenoic acid (DHA) and eicosapentaenoic acid (EPA) are the most biologically active and are needed for cell membrane integrity and neurotransmitter signaling (Coletta et al., 2010). Maternal HFD decreases DHA and EPA in maternal plasma, breast milk, and fetal plasma (Grant et al., 2011). As a consequence, omega-6 PUFAs are increased relative to omega-3 PUFAs (Grant et al., 2011). Omega-6 and omega-3 PUFAs are respectively converted into pro-inflammatory and anti-inflammatory signaling molecules (Calder, 2006). Therefore, having excess omega-6’s and deficient omega-3’s increases pro-inflammatory tone (Calder, 2006). In mice, transgenically increasing maternal omega-3’s relative to omega-6’s protects against HFD-induced peripheral and placental inflammation and rescues fetal and adult offspring from the obesogenic metabolic programming caused by maternal HFD (Heerwagen et al., 2013). Future studies should evaluate omega-3 supplementation for reversing the offspring executive and reward deficits caused by maternal malnutrition. This notion is particularly promising because cortical circuits are extremely sensitive to omega-3 levels in development and omega-3’s play a role in PFC-mediated behaviors in both rodents and humans (Larrieu et al., 2014; McNamara et al., 2015, 2010).

4.2. Methyl donor supplementation

In humans, obesity during pregnancy decreases peripheral levels of micronutrients that serve as methyl donors while simultaneously increasing pro-inflammatory cytokines IL-6 and CRP (Sen et al., 2014). In rodents, maternal methyl donor supplementation (MDS) attenuates peripheral inflammation caused by adulthood junk food diet exposure and peripheral inflammation in adult offspring caused by maternal HFD (Delaney et al., 2013, 2012). Specifically, maternal MDS decreases chemokine receptor CCR2 expression in splenic immune cells (Delaney et al., 2013). At the level of the CNS, early life postweaning MDS reverses maternal HFD-induced increases in adulthood PFC chemokine CCL2 expression in females (McKee et al., 2017). Maternal MDS normalizes maternal HFD-induced increases in dopamine and opioid related genes in the VTA, NAc, and PFC in males and females (Carlin et al., 2013). Behaviorally, MDS rescues offspring from maternal HFD-induced increases in fat preference (males) (Carlin et al., 2013) and reverses PFC-related deficits in motivation and performance limits in an operant task to assess executive function (females) (McKee et al., 2017). Together, MDS is promising in reducing maternal HFD-induced inflammation, changes in reward molecules, and deficits in executive function. Mechanistically, this may be due to epigenetic changes in DNA methylation (Carlin et al., 2013; McKee et al., 2017). Epigenetic effects are outside the scope of this review, but interesting research suggests that epigenetic modification of genes on microglia can contribute to early-life programming of reward behaviors (Schwarz et al., 2011).

5. Conclusions and future directions

From the current literature, we know that maternal malnutrition changes offspring executive and reward behaviors into adulthood while simultaneously altering reward-related molecules and neuroimmune signals in brain regions that control these behaviors. These behavioral shifts and molecular changes likely contribute to the etiology of neurodevelopmental and metabolic disorders that are associated with maternal malnutrition. However, more mechanistic studies that determine causal relationships are needed.

5.1. Theoretical model (Figure 1)

Figure 1.

General model for maternal malnutrition changing offspring exposure to inflammatory molecules, either directly through maternal plasma, placenta, or induced de novo in offspring periphery or brain. These resulting neuroimmune changes result in neuron-microglia interactions in executive and reward brain circuits, either early in development or primed to emerge in adulthood. Changes in these brain circuits manifest as adulthood behavioral changes that likely underlie metabolic and neurodevelopmental disorders. Red dashed arrows and red text indicate unknown mechanisms and molecules. Solid black arrows indicate associations with support from the literature. (DA = dopamine; TLR4 = toll like receptor 4; Cx3cr1 = fractalkine; PFC = prefrontal cortex; NAc = nucleus accumbens; HYP = hypothalamus; VTA = ventral tegmental area; ADHD = attention deficit/hyperactivity disorder)

Therefore, we propose a theoretical model whereby maternal malnutrition initiates an immune signaling cascade in the embryonic or early postnatal brain that disrupts the development of executive and reward circuits (see Figure 1). These disruptions are either immediate or primed to emerge in adulthood. Therefore, future targeted research will aim to find maternal malnutrition-induced neuroimmune mechanisms for altering offspring executive and reward behaviors throughout the lifespan. Important lifespan periods include: 1.) maternal-offspring neuroimmune programming of executive and reward circuits in embryonic and early postnatal development 2.) sustained neuroimmune changes responsible for behavioral effects that either persist or emerge in adulthood 3.) neuroimmune and behavioral changes in adolescence (another critical period, but outside the scope of the current review).

5.2. Future directions for the early developmental period

In looking at the embryonic and early postnatal period, it is essential to study developmental neuroimmune processes that are specific to executive and reward brain centers. This will build on existing work from the retinogeniculate system, preoptic area, and hippocampus by including the PFC, NAc, VTA, and even hypothalamus. Neuroimmune pathways in these additional brain regions may be unique or may similarly engage the complement system, prostaglandin E₂ signaling, or microglial fractalkine receptors. Given the sex differences in brain development, especially with neuroinflammatory processes, this work will need to include both males and females. Then the important step is to evaluate how maternal malnutrition influences these pathways and whether disrupting these processes affects behavior into adulthood.

5.3. Future directions for the adulthood period

Many of the behavioral and metabolic consequences of maternal malnutrition do not emerge until adulthood. Therefore, it is possible that maternal malnutrition primes the brain for later in life instead of having immediate effects on embryonic or early postnatal development. To tease this apart, research needs to focus on neuroimmune and behavioral consequences that are either immediate, sustained, or only emerge in adulthood. Behavioral assessment early in life is logistically challenging, but understanding the neuroimmune mechanisms underlying behavioral changes in adulthood will allow us to query whether these changes happen earlier. Furthermore, these changes may not be present at baseline in adult offspring, but may be primed to emerge with subsequent inflammatory or dietary challenges.

5.4. Final remarks

Together, understanding these neuroimmune and behavioral mechanisms of maternal malnutrition could lead to new treatment strategies for neurodevelopmental and metabolic disorders. Dietary supplementation is a promising intervention because it addresses specific nutrient deficits, with the capacity to be simultaneously mechanistic and holistic. The exact kind of maternal malnutrition can vary drastically between individuals and treatment may not be readily available prenatally or may not be necessary until offspring display phenotypes. Therefore, research will shed light on specific deficits caused by maternal malnutrition so that interventions can be individually tailored to provide maximal benefits.

Highlights.

Maternal malnutrition increases offspring risk for metabolic dysfunction and obesity Maternal malnutrition increases offspring risk for neurodevelopmental disorders Animal models of maternal malnutrition reveal executive and reward changes Maternal malnutrition affects offspring peripheral and central immune responses Neuroimmune consequences may contribute to executive and reward changes