Prenatal Choline Supplementation Ameliorates the Long-Term Neurobehavioral Effects of Fetal-Neonatal Iron Deficiency in Rats^,,

Abstract

Background: Gestational iron deficiency in humans and rodents produces long-term deficits in cognitive and socioemotional function and alters expression of plasticity genes in the hippocampus that persist despite iron treatment. Prenatal choline supplementation improves cognitive function in other rodent models of developmental insults.

Objective: The objective of this study was to determine whether prenatal choline supplementation prevents the long-term effects of fetal-neonatal iron deficiency on cognitive and social behaviors and hippocampal gene expression.

Methods: Pregnant rat dams were administered an iron-deficient (2–6 g/kg iron) or iron-sufficient (IS) (200 g/kg iron) diet from embryonic day (E) 3 to postnatal day (P) 7 with or without choline supplementation (5 g/kg choline chloride, E11–18). Novel object recognition (NOR) in the test vs. acquisition phase, social approach (SA), and hippocampal mRNA expression were compared at P65 in 4 male adult offspring groups: formerly iron deficient (FID), FID with choline supplementation (FID-C), IS, and IS with choline supplementation.

Results: Relative to the intact NOR in IS rats (acquisition: 47.9%, test: 60.2%, P < 0.005), FID adult rats had impaired recognition memory at the 6-h delay (acquisition: 51.4%, test: 55.1%, NS), accompanied by a 15% reduction in hippocampal expression of brain-derived neurotrophic factor (Bdnf) (P < 0.05) and myelin basic protein (Mbp) (P < 0.05). Prenatal choline supplementation in FID rats restored NOR (acquisition: 48.8%, test: 64.4%, P < 0.0005) and increased hippocampal gene expression (FID-C vs. FID group: Bdnf, Mbp, P < 0.01). SA was also reduced in FID rats (P < 0.05 vs. IS rats) but was only marginally improved by prenatal choline supplementation.

Conclusions: Deficits in recognition memory, but not social behavior, resulting from gestational iron deficiency are attenuated by prenatal choline supplementation, potentially through preservation of hippocampal Bdnf and Mbp expression. Prenatal choline supplementation may be a promising adjunct treatment for fetal-neonatal iron deficiency.

Introduction

Iron deficiency is the most pervasive early-life nutrient deficiency, affecting over half of all pregnancies globally (1). Iron deficiency during gestation and early postnatal life can have adverse effects on the developing brain, evidenced by both acute and long-lasting deficits in cognitive function and socioemotional behaviors (2). Iron-deficient (ID)⁹ infants score lower on indices of mental development (3, 4) and exhibit impairments in recognition (5) and spatial memory (6). In addition, ID infants and formerly iron-deficient (FID) children show blunted emotional reactions and have been described as wary of strangers and less likely to engage caregivers (7–9). The neurocognitive and socioemotional impairments of early life iron deficiency persist into adulthood despite treatment and resolution of iron deficiency during infancy, resulting in reduced potential for educational and occupational achievement and higher risk of psychopathologies such as anxiety, depression, and schizophrenia (2, 6, 10, 11).

The acute and long-term learning and memory deficits found in humans have been replicated in rodent models of fetal-neonatal iron deficiency. Adult FID rats exhibit poor spatial memory in the Morris water maze and win-shift tasks (12–14) and impaired fear learning (15, 16). The neural mechanisms of fetal-neonatal iron deficiency include general impairments in monoamine metabolism (17) and myelination (18, 19), along with long-term disruptions in multiple aspects of hippocampal development and function, including synaptic plasticity (15, 20, 21), dendritic morphology (20, 22, 23), and gene expression (23–25). Lack of neuronal iron appears to be critical for the behavioral and morphologic changes following fetal-neonatal iron deficiency as indicated by impaired spatial memory and abnormal hippocampal neuronal structure in adult animals from 2 nonanemic genetic mouse lines in which iron is depleted specifically in hippocampal neurons during late gestation (23, 26).

The effects of early-life iron deficiency on socioemotional behaviors have also been examined in rodent models. Consistent with the increased wariness and neophobia reported in ID infants (8), fetal-neonatal iron deficiency results in elevated anxiety-like behavior in adult rats (13, 27, 28), which may be improved by postnatal iron repletion (27). However, the effects of early life iron deficiency on rodent social approach behaviors are understudied.

The finding that long-term behavioral abnormalities continue to be present in FID humans and rodents despite early treatment indicates the need to identify adjunct treatments for early life iron deficiency. One attractive candidate is maternal choline supplementation. Choline is an essential nutrient that is critically involved in early brain development (29) and is readily found in multiple food sources. Choline supplementation during specific pre- and postnatal time windows reverses some of the cognitive deficits observed in rodent models of developmental disorders including fetal alcohol syndrome (30, 31) and Down syndrome (32) and can protect adult animals from the memory impairments associated with brain injury (33), seizures (34), and age-related dementia (35). The improvement in memory resulting from choline supplementation is accompanied by increased hippocampal neurogenesis (36), reduced thresholds for long-term potentiation (37), increased dendritic arborization (38), and increased expression of neurotrophic factors such as brain-derived neurotrophic factor (BDNF) (36).

The present study tested whether prenatal choline supplementation provided during one of its critical periods, embryonic day (E) 11–18 (35, 39), improves or prevents cognitive and socioemotional behavioral deficits in adulthood by using novel object recognition (NOR) and social approach (SA) tasks. The NOR task is ideally suited as a measure of memory in FID animals because of its current use as an assay of recognition memory in ID infants (5) and the dependence of recognition memory on hippocampal integrity (40, 41), which is disrupted by iron deficiency. Furthermore, it was recently shown that iron deficiency starting at gestation leads to impaired NOR accompanied by poor hippocampal neuronal maturation (42). We sought to use the broad measure of SA toward an unfamiliar conspecific based on observations of withdrawal and blunted affect toward strangers in ID infants. We also assessed whether choline supplementation protects against the long-term changes in the hippocampal expression of plasticity-associated genes Bdnf-IV and -VI, myelin basic protein (Mbp), and ApoE. Expression of these genes is altered by early life iron deficiency (18, 19, 24, 43) and is known to influence hippocampal function and learning and memory (44–50).

Materials and Methods

Rats and diet.

Timed-pregnant Sprague Dawley dams (Charles River) consumed either an ID (2–6 g/kg iron; Harlan-Teklad TD# 110137, see Supplemental Table 1 for diet composition) or iron-sufficient (IS) (200 g/kg iron; TD# 110138) fortified diet (Harlan-Teklad) ad libitum from E3 to postnatal day (P) 7. At day 7, all rats were fed the IS diet for the rest of the experiment. This model induces a 50% reduction in brain iron concentration in P10 pups (51–53) but restores normal hematologic status by P28 (52) and brain iron before P56 (51, 54). Half of the dams fed the IS or ID diet were administered dietary choline supplementation [5 g/kg choline chloride supplemented (55, 56), IS diet: TD# 110140; ID diet: TD# 110139] from E11 to E18, whereas the remaining dams were administered a diet with standard choline content (1.1 g/kg). Thus, dams and their litters were randomly assigned to 1 of 4 groups based on maternal diet and labeled based on their adult status (Fig. 1): FID (without choline supplementation), FID with choline supplementation (FID-C), always IS (without choline supplementation), and always IS with choline supplementation (IS-C). Within 24 h of birth, litters were culled to a total of 8 pups containing 6 males and 2 females when possible. For litters with fewer than 6 males, additional females were added to keep the total litter size at 8 pups. Pups were weaned at P21 into same-sex groups of 4–6 pups, which were further separated into groups of 2–3 during adulthood. Rats began behavioral testing or were killed via overdose of Pentobarbitol (100 mg/kg; Fatal Plus, Vortech) for hippocampal dissection as adults on P65. An additional subset of rats from the FID, FID-C, and IS diet groups was killed at P15 or P65 for measurements of hematocrit concentrations, body weight, and brain weight. Because rat pups from the FID and FID-C groups were still brain ID at P15 (20, 52), they are referred to at that time point as ID and ID with choline supplementation (ID-C), respectively. All experiments were approved by the University of Minnesota Institutional Animal Care and Use Committee.

FIGURE 1.

Timeline of ID and choline-supplemented diets administered to pregnant rat dams and the experimental procedures for male offspring. E, embryonic day; FID, formerly iron deficient; FID-C, formerly iron deficient with choline supplementation; ID, iron deficient; IS, iron sufficient; IS-C, iron sufficient with choline supplementation; P, postnatal day.

Hematology.

P15 rat pups from the ID, ID-C, and IS diet groups were weighed and then killed via rapid decapitation for blood and brain collection. Hematocrit concentrations (percentage of red blood cells in whole blood) were measured from trunk blood, and brains were extracted and weighed. This procedure was repeated for separate groups of FID, FID-C, and IS adult rats at P65.

Behavior.

The NOR procedure was modified from the task described by Clark et al. (40). Prior to testing, rats were handled for 2 d followed by 2 d of habituation to the test chamber (black ABS plastics, 50 cm × 50 cm × 50 cm), which contained Velcro strips to attach objects. Rats were habituated to the chamber for 10 min/d, and on day 2 a pair of identical objects not used for testing were placed in the chamber to habituate rats to the presentation of novel objects within the chamber and reduce the potential neophobia of FID rats (13).

Testing for NOR consisted of an acquisition phase and a test phase. Each phase followed a similar procedure: rats were rehabituated to the test chamber for 1 min then removed, at which point objects were secured in the chamber and rats were returned. During the acquisition phase, 2 identical (familiar) objects were placed in the chamber and exploration of the objects was video recorded for 3 min. After a delay of 1 or 6 h, during which the rats were returned to their home cage and housed in the test room, rats were returned to the chambers for the test phase. For the test phase, the familiar object from the acquisition phase was placed at 1 location within the chamber, a novel object was placed at the other location, and exploration of the objects was recorded for 5 min. Between tests, chambers were wiped down with 70% ethanol and objects were cleaned with diluted bleach. Three different pairs of novel/familiar objects were used for testing (ranging in size from ∼102 to 1048 cm³) for a total of 6 objects, which varied in color, material, texture, or shape. Objects included small plastic or rubber toys, plastic cups and bowls, a metal wrench with a rubber handle, and a plastic toilet paper holder. Similar-sized objects were used for novel/familiar object pairs, and different copies of the same objects were used for acquisition and testing.

All rats were tested for NOR at both the 1- and 6-h delays, and the 1- and 6-h tests were separated by at least 24 h and conducted by using a different familiar/novel object pair, similar to previous applications of the NOR task (40, 57). Test order, novel/familiar object pair, and novel object location during testing were all counterbalanced across each diet group. A location bias was not observed during the acquisition periods because rats from all diet conditions explored objects equally regardless of location within the chamber (data not shown). Raters unaware of the experimental conditions scored object investigation for the entire 3-min video from the acquisition phase but only the first 2 min of the test phase by using computer-assisted software (Button box; Behavioral Research Solutions). The rating period for testing was chosen based on previous observations of a rapid decline in investigation of the novel object (58). Analysis of the subset of videos scored by both raters confirmed a high inter-rater reliability (R_p = 0.94, n = 21, P < 0.0001). Object memory was defined as a significant increase in novel object preference during the test phase compared with preference for the object in the same location during the acquisition phase. For all groups at both delay periods, the object preference during the acquisition phase was equivalent to 50%.

The SA task is based on the original design for mice used by Nadler et al. (59) with modifications of the task for rats (60, 61). SA testing was conducted in the same chambers used for NOR at least 1 wk after completion of the final NOR test. Two identical wire-mesh enclosures (26.7 cm × 13 cm × 30.5 cm) were placed on opposite sides of the chamber, and test rats were placed into the chamber to habituate for 5 min. Following habituation, test rats were returned to the home cage, and the subject rat and a novel object distinct from any of the objects used for the NOR task were placed in different enclosures. Subject rats were unfamiliar age-matched rats from the same diet group. Test rats were returned to the chamber, and exploration of the enclosures containing the novel object and the unfamiliar rat was video recorded for 5 min. Rats were only tested once, and subject rats were never used as test animals. Chambers and enclosures were thoroughly cleaned with 70% ethanol between tests. Two raters unaware of the experimental conditions scored all videos for investigation of each enclosure, and the mean score was used for subsequent statistical analysis.

Hippocampal extraction and real-time qPCR.

Procedures used for hippocampal extraction, RNA isolation, cDNA synthesis, and real-time qPCR were carried out according to the Minimum Information for Publication of Quantitative Real-Time PCR Experiments guidelines (62) and have been described previously (23, 24, 26, 63). Briefly, P65 rats were killed via rapid decapitation and brains were extracted and dissected on an ice-cold metal block wetted with PBS. Both hippocampi were removed, flash frozen in liquid nitrogen, and stored at −80°C. RNA from the right hippocampus was isolated by using the RNAqueous total RNA isolation kit (Ambion). Hippocampal tissue was lysed in 800-μL lysis buffer; 400 μL of the RNA lysate was further processed according to the manufacturer’s instructions. RNA concentration and purity was assessed by using a NanoDrop-1000 spectrophotometer. One microgram of total RNA was used to generate cDNA (High Capacity RNA-to-cDNA kit; Applied Biosystems) following the manufacturer’s instructions. The cDNA was diluted to 10% of the final volume with ddH2O prior to real-time qPCR analysis. Each real-time qPCR reaction consisted of 5.0-μL qPCR FastStart Universal Probe Mastermix (Roche), 4.5-μL cDNA, and 0.5 μL of probes for Bdnf-IV (Rn 01484928_m1), Bdnf-VI (Rn2531967), Mbp (Rn00566745), or ApoE (Rn00593680). β-actin (Actb) was used as an internal control (Rn 01412977). Reactions were run in a 96-well plate in a MX3000P thermocycler (Stratagene).

Statistical analysis.

Statistical analyses were performed by using JMP (SAS Institute). All data were analyzed by using an initial ANOVA followed where appropriate by post hoc tests with significance set at P < 0.05. Hematocrits and brain and body weights from P15 and P65 rats were analyzed by using a 1-factor ANOVA followed by Tukey’s HSD post hoc tests. NOR was analyzed by using a separate 2 × 4 ANOVA (test phase × prenatal diet) for each delay period (1 or 6 h) followed by planned orthogonal contrasts between percentage preference during the acquisition phase object preferences. Because the relevant comparisons between test and acquisition preferences within the same treatment group make up a small proportion of the total possible comparisons, planned contrasts were used rather than post hoc tests to reduce the probability of producing type I and II errors (64). SA was assessed by using a 2 × 4 ANOVA (object vs. social × prenatal diet) followed by Tukey’s HSD post hoc tests. Gene expression was analyzed by ANOVA and a post hoc t test for group differences by using Prism GraphPad.

Results

Prenatal choline supplementation does not affect growth or hematologic status in P15 offspring.

ID pups, regardless of choline supplementation, had reduced hematocrit concentrations and body weights relative to IS controls at P15 (Table 1), with lower brain weights in the ID pups and a trend toward lower brain weights in the ID-C pups (P = 0.1 vs. IS pups). There were no differences in hematocrit concentrations, brain weights, or body weights among groups at P65 (Table 1).

TABLE 1.

Hematocrit concentrations and brain and body weights of P15 pups and P65 adult rats following gestational iron deficiency with or without prenatal choline supplementation¹

	P15			P65
Measurements	IS	ID	ID-C	IS	FID	FID-C
Hematocrit, % RBC	32.3 ± 1.2 (10)	24.9 ± 0.9* (9)	24.0 ± 1.0* (10)	46.8 ± 0.8 (6)	48.8 ± 1.2 (6)	47.7 ± 0.7 (6)
Brain weight, g	1.49 ± 0.02 (6)	1.37 ± 0.02** (6)	1.41 ± 0.02 (6)	2.2 ± 0.03 (6)	2.3 ± 0.11 (6)	2.15 ± 0.06 (6)
Body weight, g	39.9 ± 0.7 (10)	31.5 ± 0.3* (7)	35.3 ± 1.5** (10)	417.9 ± 14.9 (6)	409.8 ± 13.2 (6)	416.0 ± 6.9 (6)

¹Values are means ± SEMs; sample sizes are listed in parentheses. Significant difference vs. IS control group: *P < 0.001, **P < 0.01. FID, formerly iron deficient; FID-C, formerly iron deficient with choline supplementation; ID, iron deficient; ID-C, iron deficient with choline supplementation; IS, iron sufficient; P, postnatal day.

Prenatal choline supplementation improves cognitive function in FID rats.

Following a 1-h delay, all groups had a significant exploration bias toward the location of the novel object during testing relative to acquisition (Fig. 2A; P < 0.0001), which is indicative of object memory retention. The degree of bias toward the novel object was of a similar magnitude in all diet treatment groups. After a 6-h delay, a similar exploration bias toward the novel object was observed (P < 0.0001); however, this effect was dependent on prenatal diet treatment (Fig. 2B; P < 0.05). IS rats exhibited a novelty bias regardless of prenatal choline supplementation, as demonstrated by a significant difference between testing and acquisition novel object preference (IS rats: P < 0.005; IS-C rats: P < 0.0001). A novelty bias was absent in FID rats at the 6-h delay, indicating a lack of object memory. However, novelty preference was preserved by prenatal choline supplementation (FID-C group, test preference vs. acquisition, P < 0.0005). Total object investigation during the acquisition phase was similar among all groups (Supplemental Fig. 1), suggesting that recognition memory deficits in FID rats were not due to reduced object exposure during the acquisition phase.

FIGURE 2.

Recognition memory of adult rats following gestational iron deficiency with or without prenatal choline supplementation. Panels show preference for the novel object during the test and acquisition phases following a 1-h (A) or 6-h (B) delay. Values are means ± SEMs, n = 17–29. Significant difference from acquisition preference: *P < 0.05, **P < 0.01, ***P < 0.001. Significant effects: test phase (A); test phase, prenatal diet × test phase (B). FID, formerly iron deficient; FID-C, formerly iron deficient with choline supplementation; IS, iron sufficient; IS-C, iron sufficient with choline supplementation.

Prenatal choline supplementation does not affect social behavior of FID rats.

All groups explored the enclosure containing a social partner significantly more than the enclosure containing a novel object (Fig. 3; P < 0.0001; mean investigation time: social partner = 136.3 s, object = 24.5 s). Investigation of the object and social enclosures were differentially affected by prenatal diet (P < 0.0005). Although object exploration was similar for each prenatal diet group, social exploration time was lower in FID rats than in IS controls (P < 0.05). Prenatal choline supplementation did not significantly alter investigation of the social enclosure for either FID-C or IS-C groups relative to FID and IS groups, respectively.

FIGURE 3.

Social and novel object approach of adult rats following gestational iron deficiency with or without prenatal choline supplementation. Values are means ± SEMs, n = 11–15. Bars without common letters differ, P < 0.05. Significant effects: enclosure type, prenatal diet × enclosure type. FID, formerly iron deficient; FID-C, formerly iron deficient with choline supplementation; IS, iron sufficient; IS-C, iron sufficient with choline supplementation.

Prenatal choline supplementation reverses changes in synaptic plasticity genes in hippocampi of FID rats.

Consistent with previous findings (25, 63), hippocampal expression of Bdnf-IV and -VI was reduced in P65 FID rats relative to IS controls (Fig. 4A, B; Bdnf-IV: P < 0.05; Bdnf-VI: P < 0.05). Prenatal choline supplementation increased expression of Bdnf-IV (P < 0.01) and caused a nonsignificant increase in expression of Bdnf-VI (P = 0.08) in FID-C adult rats compared with FID rats. Unexpectedly, prenatal choline supplementation resulted in lower Bdnf-VI expression in the IS-C group relative to the IS group (P < 0.05).

FIGURE 4.

Hippocampal mRNA expression of P65 adult rats following gestational iron deficiency with or without prenatal choline supplementation. Panels show fold change from IS expression values of Bdnf-IV (A), Bdnf-VI (B), Mbp (C), and ApoE (D). Values are means ± SEMs, n = 4–6 for each gene. Within each panel, bars without common letters differ, P < 0.05. Bdnf, brain-derived neurotrophic factor; FID, formerly iron deficient; FID-C, formerly iron deficient with choline supplementation; IS, iron sufficient; IS-C, iron sufficient with choline supplementation; Mbp, myelin basic protein; P, postnatal day.

Mbp mRNA expression was lower and ApoE mRNA expression was higher in the FID group than in the IS group (Fig. 4C, D; FID vs. IS group: Mbp, P < 0.05 and ApoE, P < 0.05), findings that are consistent with previous reports (19, 42). Prenatal choline supplementation increased hippocampal expression of Mbp in FID rats (FID vs. FID-C group: P < 0.01) but did not affect ApoE expression.

Discussion

Early life iron deficiency leads to cognitive and socioemotional deficits that persist despite iron therapy (2, 10, 65). Because iron therapy alone fails to restore complete brain health, there is a need to identify evidence-based adjunctive therapies that temper or alleviate the long-term effects of early life iron deficiency. The present study modeled these deficits in a developmentally appropriate rat model of fetal neonatal iron deficiency and identified prenatal choline supplementation as a potential adjunctive treatment for early life iron deficiency. Choline supplementation improved cognitive function and normalized expression of genes important in hippocampal synaptic plasticity in adult rats that were ID during the fetal-neonatal period. Nevertheless, social behavior was largely unaffected by prenatal choline supplementation, suggesting a domain-specific effect rather than a global improvement in brain function.

The present study demonstrates that fetal-neonatal iron deficiency results in reduced SA and impaired recognition memory in adulthood. We used the NOR task to assess recognition memory, a measure with good face validity to the memory test used to measure recognition memory following fetal-neonatal iron deficiency (66). As seen in humans (67) and mice (26), fetal-neonatal iron deficiency resulted in a graded degree of learning and memory impairment in the FID group as a function of the difficulty of the task, with poorer performance at the more challenging delay of 6 h. Likewise, our use of the SA task to uncover the abnormal socioemotional behavior in the rat model of gestational iron deficiency is consistent with the observation of a lack of engagement and increased wariness during social interactions in FID children (7).

The reversal of the recognition memory deficit in the FID-C group adds to the growing body of evidence for the beneficial effects of timed prenatal choline supplementation on adult neuroplasticity and cognitive function in rodent models of genetic and early environmental insults (30, 35, 36, 68). The improvement in NOR may be due to the effects of choline supplementation on hippocampal development, a structure that is critical for intact recognition memory (40, 41). Choline supplementation during embryonic development (E11–E18) coincides with proliferation of hippocampal neurons (69, 70), and was shown to alter hippocampal structure (38) and facilitate the expression of synaptic plasticity genes within the hippocampus (36).

Consistent with choline-induced changes in hippocampal gene expression, prenatal choline supplementation reversed the long-term changes in hippocampal gene expression resulting from fetal-neonatal iron deficiency. In line with previous findings (63), concentrations of Bdnf-IV and -VI were reduced in the hippocampi of FID rats compared with controls, but this downregulation was reversed in the FID-C group. BDNF has long been thought to be an important factor for long-term memory (49), and a hippocampal-specific deletion of BDNF disrupts NOR in mice (44). Thus, downregulation of Bdnf-IV and -VI in the hippocampi of FID rats and normalization following choline supplementation may be related to the observation of a NOR deficit in FID rats but not the FID-C group. In a similar manner, hippocampal expression of Mbp and ApoE was altered in FID rats relative to IS controls, and Mbp was restored to normal concentrations by prenatal choline supplementation. Hippocampal protein concentrations of MBP and APOE have previously been shown to be disrupted by early life iron deficiency (18, 19, 43), and both proteins appear to be important for normal myelination (45, 50) and cognitive performance (46–48). Indeed, hypomyelination was reported in animal models of early life iron deficiency (18, 71), and the altered conduction velocity in the brains of FID children (as measured by evoked potentials) is consistent with abnormal myelination (72). Taken together, altered expression of Mbp and ApoE in the hippocampi of FID rats may help to explain our finding of impaired NOR in FID rats and previous observations of abnormal myelination following fetal-neonatal iron deficiency. Similar to Bdnf-IV and -VI, normalization of Mbp in response to choline supplementation may be related to improved recognition memory in FID-C rats.

Although choline has long been recognized as a substrate crucial for central nervous system development (68, 73), it remains unclear how supplementation during early development can reverse the changes in expression of Bdnf and other genes that result from fetal-neonatal iron deficiency. Choline supplementation does not appear to mitigate the effects of iron deficiency by improving the iron status of rat pups exposed to an ID diet, as demonstrated by the similar hematocrit concentrations of P15 ID-C and ID rat pups. Alternatively, choline supplementation may lead to long-term changes in gene expression through epigenetic regulation of gene promoter regions (73, 74), such as was shown to occur for BDNF following early life stress (75). As a methyl group donor (76), choline can influence methylation of cytosine residues at CpG islands in promoter regions, which alters the accessibility of genes and ultimately influences expression (77). Epigenetic alterations such as methylation are stable and when they occur early in development can lead to changes in gene expression lasting into adulthood (78).

The finding that maternal choline supplementation during mid-to-late gestation facilitates the recovery of recognition memory, but not social behavior, in the FID group suggests a differential effect of choline supplementation on developing neural systems. Based on evidence from the mouse model of hippocampal-specific iron deficiency, depletion of iron from the hippocampus and the resulting disruptions in neuronal morphology and gene expression are sufficient for impairments in spatial memory (26), suggesting that the memory deficits observed following a dietary model of iron deficiency may largely arise from hippocampal-specific changes. However, the effects of iron deficiency on socioemotional behaviors, such as heightened anxiety and neophobia (13, 27, 28), have been linked to disruptions in dopamine signaling within the ventral striatum (2, 79). Indeed, the mouse model of hippocampal-specific iron depletion exhibited a reduction in anxiety-like behavior relative to controls, suggesting that the increased anxiety observed in FID rats occurs independently of changes within the hippocampus (21). Therefore, considering the strong evidence that prenatal choline supplementation can alter hippocampal development (see above), it is possible the deficits in hippocampal-dependent learning and memory resulting from early life iron deficiency may be particularly sensitive to choline supplementation. By contrast, changes in socioemotional behaviors including a reduction in social approach, a relatively broad measure of sociability that is sensitive to heightened anxiety (80), may be less responsive to choline supplementation. Assessment of nuanced changes in social behavior within models of developmental insults such as gestational iron deficiency may benefit from the use of more sophisticated behavioral tasks (81). Future studies could employ measures of social behavior including conditioned social reward (82, 83) or reciprocal social play (84) to further characterize the social deficit resulting from fetal-neonatal iron deficiency.

The present study supports the possibility of using maternal choline supplementation as an adjunctive treatment to iron supplementation during pregnancy for populations at risk of early life iron deficiency. Choline is found in a variety of widely available foods including eggs, poultry, soybeans, and wheat germ (73), allowing for potential supplementation of dietary choline as a preventative measure for long-term cognitive effects in populations at risk of fetal-neonatal iron deficiency. The positive effects of choline supplementation may be limited to neurocognitive outcomes because the impaired socioemotional behavior observed in FID rats was not corrected by choline supplementation. Preclinical studies have revealed additional therapeutic windows for choline supplementation during postnatal development (30). Although it is unclear whether choline supplementation during the postnatal period could similarly reverse the cognitive impairments associated with fetal-neonatal iron deficiency, the positive findings in this study encourage the testing of choline supplementation after gestation. Similar findings using postnatal choline supplementation would allow for treatment after a diagnosis of iron deficiency in the newborn.