Functional Connectivity and Metabolic Alterations in Medial Prefrontal Cortex in a Rat Model of Fetal Alcohol Spectrum Disorder: A Resting-State Functional Magnetic Resonance Imaging and in vivo Proton Magnetic Resonance Spectroscopy Study

Fetal alcohol spectrum disorder (FASD) is an umbrella term encompassing a range of outcomes that can occur following prenatal exposure to alcohol. The prevalence of FASD may be as high as 1–5% in school children in the US and some Western European countries [1, 2]. People with a FASD may present with deficits in executive functions, such as cognitive flexibility, set shifting, or working memory, which can lead to lifelong difficulties in adapting to and interacting in society [3]. Many aspects of disrupted executive function have been successfully modeled in animal studies [4-6]. Interestingly, we found sex differences in tests of working memory and cognitive flexibility, with performance of ethanol-exposed (Eth) male rats worse than that of Eth females, although both sexes showed worse performance than their respective control (Ctr) groups [6].

The human dorsolateral prefrontal cortex (PFC) and its rodent analogue, the medial PFC (mPFC), are considered central to executive functions [7, 8]. Prenatal alcohol exposure is known to interfere with the development of the frontal lobe. Dysmorphology of the frontal lobe, such as alterations in frontal volume and cortical thickness, has been reported in children with prenatal alcohol exposure [9-11]. In animal models, prenatal ethanol exposure leads to a significant reduction in the number of neurons in the mPFC [12] and an increase in the number of GABAergic interneurons, suggesting an alteration in the excitatory/inhibitory balance which may affect functioning [13, 14].

Cooperative activity between the frontal cortices and other brain regions such as the basal ganglia is required for adaptive response selection and response execution [15-17]. A critical component of the frontobasal ganglia pathway is the frontostriatal circuit, which exhibits task-driven changes in activity during executive function tests. A functional magnetic resonance imaging (fMRI) study examining children exposed to alcohol prenatally reported altered frontostriatal functional connectivity in functionally distinct corticostriatal loops during a working memory task compared to nonexposed children [18]; alcohol-exposed children exhibited increased functional connectivity between the frontal cortex and the putamen, a subregion of the striatum critical for organization of movement. This was coupled with hypoconnectivity between the frontal cortex and the dorsal caudate, a subregion of the striatum involved in executive function. The thalamus, especially the mediodorsal nucleus which has reciprocal connections with the frontal lobe, is another region contributing to executive function [19, 20]. A study of the thalamo-cortico-thalamic loop revealed alterations in the thalamocortical circuits, especially at axon terminal fields, in adult rats with prenatal alcohol exposure [21]. Others report reduced basilar dendritic complexity in pyramidal neurons in layer II/III of rat mPFC after developmental exposure to alcohol [22]. Layer III pyramidal neurons play a major role in integrating and organizing inputs from the thalamus [23]; thus, dendritic alterations in the PFC indicate a possible disruption of thalamofrontal cortex functional interaction. Together, these results point to a dysregulation between frontal cortices and motor, sensory, and cognitive circuits that likely contributes to poor executive function in subjects exposed to alcohol prenatally. Therefore, investigation into the effect of prenatal ethanol exposure on the functional connectivity of frontal cortices may provide meaningful insight into which component of the interactions between frontal lobe and other regions could be vulnerable.

Non-task-driven functional connectivity can be measured by resting-state fMRI (rsfMRI). It relies on the correlation of the spontaneous low-frequency fluctuations in blood oxygen level-dependent (BOLD) signal between brain regions. A strong temporal correlation in fluctuations of the BOLD signal is interpreted as existence of a strong functional connectivity between two disparate brain regions [24, 25].

In vivo proton magnetic resonance spectroscopy (¹H-MRS) is a noninvasive neuroimaging technique that measures concentrations of specific neurochemicals from localized brain regions. In humans, three neurochemicals are often analyzed; N-acetylaspartate (NAA), a marker of neuronal integrity, choline (Cho), a measure of cell turnover, and creatine, a measure of energy stores [26]. Typically, the ratios among these neurochemicals are reported. Studies in children with FASD using ¹H-MRS do not reach consensus, in part because different brain regions are studied, but also because the age of the children varies, as does the magnitude and timing of alcohol exposure. That said, the majority of studies report that at least one of the ratios is altered in one or more brain regions. In frontal cortical regions, Cho is lower in the white matter in the frontal/parietal region [27], NAA/creatine and NAA/Cho were found to be lower in adolescents and young adults with FASD [28], whereas another study reports these to be unchanged in this region in children [29]. Overall, the results suggest that changes in the frontal lobe may relate mainly to glial cells rather than to neurons.

In this study, we used a rat FASD model to assess the resting-state functional interactions involving mPFC by evaluating the functional connectivity between the mPFC (containing the prelimbic and anterior cingulate cortex) and other regions across the brain in young adult rats. We also applied ¹H-MRS to determine the expression of brain metabolites in mPFC. Since Eth rats show changes in mPFC neuron composition and alterations in executive functions that utilize the mPFC, we hypothesized that prenatal alcohol exposure alters the functional connectivity and biochemical profile of mPFC. We predicted that these effects would be sex-dependent and worse in males, as they show greater deficits in cognitive tests than females.

Timed pregnant Long-Evans rats (Harlan, now Envigo, Frederick, MD, USA) were housed in an Association for Assessment and Accreditation of Laboratory Animal Care International-accredited facility. The facility was controlled for temperature (22°C) and humidity (40–45%) and kept on a 12/12 h light/dark cycle (lights on at 7 a.m.). Rats were given ad libitum access to an ethanol-containing liquid diet (L10251A; Research Diets, New Brunswick, NJ, USA; 2.1% v/v ethanol on gestational days 6–7 [G6–G7]; 4.27% ethanol on G8–G10; 6.36% on G11–G20). This model typically produces blood alcohol concentrations of approximately 120 mg/dL [30, 31]. Ctr animals received an equivalent volume of a non-ethanol-containing liquid diet or ad libitum access to laboratory chow. All animals were returned to lab chow on G21 and had ad libitum access to water throughout.

Young adult offspring underwent imaging on one day between postnatal day 66 (P66) and P101. In the current study, we evaluated frontostriatal functional connectivity and metabolic alter-ations in male rats (14 Ctr and 12 Eth). We also incorporated sex as a variable in experimental designs (12 Ctr females, 6 Eth females) to investigate whether prenatal ethanol exposure affected the male and female brain differently.

Animals were placed in a Bruker Biospec 7.0-Tesla 30-cm horizontal bore scanner (Bruker Biospin MRI GmbH, Germany) equipped with a BGA12S gradient system and interfaced to a Bru-ker Paravision 5.1 console. A Bruker 72-mm linear volume coil was used as the transmitter and a Bruker ¹H four-channel surface coil array was used as the receiver. Anesthesia was induced with 2% isoflurane followed by intramuscular bolus administration of the α2-agonist dexmedetomidine (0.03 mg/kg). Light anesthesia was maintained using 0.25–0.5% isoflurane in oxygen-enriched air with continuous infusion of dexmedetomidine (0.015 mg/kg/h) during data acquisition. A magnetic resonance-compatible small-animal monitoring and gating system (SA Instruments, Inc., New York, NY, USA) was used to monitor respiration rate and body temperature. Body temperature was maintained at 35–37°C using warm water bath circulation. Ear pins and bite bars were used to minimize head motion.

A three-slice (axial, mid-sagittal, and coronal) scout image using rapid acquisition with fast low-angle shot [32, 33] was used to orient the rat brain. A fast shimming procedure (FASTMAP) was used to improve the B₀ homogeneity covering the brain [34]. Both proton density and T₂-weighted images were obtained for anatomical reference using a two-dimensional rapid acquisition with relaxation enhancement sequence covering the entire brain [35]. Imaging was performed over a 3.5-cm field of view in the coronal plane with an in-plane resolution of 137 μm using 22 slices at 1 mm thickness and two averages, at an effective echo time (TE) of 18.94 ms for the proton density-weighted images and an effective TE of 56.82 ms for the T₂-weighted images. The repeat time (TR) was 3,500 ms.

rsfMRI was acquired by matching the anatomical images using a single-shot, gradient-echo-planar imaging sequence (TR/TE = 1,000/14.7 ms) with a 3.5-cm field of view and an in-plane resolution of 547 μm² using 22 slices at 1 mm thickness. Six hundred repetitions were taken, resulting in a total scanning time of around 10 min for each rsfMRI run. Three such rsfMRI runs were acquired on each animal per imaging session.

¹H-MRS data were obtained from a voxel (3.5 × 3 × 2.5 mm³) that covered the mPFC (Bregma 4.5 to 2.0, mainly cingulate cortex area 1 and prelimbic cortex [36]; Fig. 1a). Prior to acquiring the spectra, adjustments of all first- and second-order shims over the voxel of interest were accomplished with the FASTMAP procedure. A short-TE point-resolved spectroscopy pulse sequence (TR/TE = 2,500/10 ms, NA = 600) was used for MRS data acquisition [37]. The unsuppressed water signal from the prescribed voxel was used as a reference for determining the specific metabolite concentrations. The total duration of the whole imaging experiment was approximately 2.5 h.

Body weight data were analyzed with two-way ANOVA to assess the effects of ethanol, sex, and ethanol-sex interaction. The simple effect of ethanol in males and females was estimated by a follow-up partitioned analysis with sex difference controlled [38]. A subset of animals (3 males and 3 females) were randomly selected from the Eth groups for investigation of brain volume alterations. Ctr animals were selected to match Eth animals for age and sex. Brain volume analysis was performed using Medical Image Processing, Analysis, and Visualization (https://mipav.cit.nih.gov/). The anatomical images of each animal were skull-stripped manually. Brain volume was estimated by adding the slice-by-slice volume of the 11 slices that were consistently acquired across animals as shown in Figure 1b. Body weight, brain volume, and brain volume/body weight ratio data were analyzed with two-way ANOVA to assess the effects of ethanol, sex, and ethanol-sex interaction. In addition, a partitioned analysis was preformed to estimate the simple effects of ethanol in males and females, respectively. The simple effect of sex was estimated in Ctr animals to assess the baseline difference between males and females. The significance level was set at p < 0.05.

All rsfMRI image preprocessing and processing was conducted using SPM12 (http://www.fil.ion.ucl.ac.uk/spm/) and AFNI (http://afni.nimh.nih.gov/afni). The processing pipeline included slice timing correction, motion correction, alignment to a brain anatomical template, orthogonalization of motion-derived parameters, spike censoring, band-pass filtering, and smoothing. Specifically, the first ten volumes were discarded from each rsfMRI run to minimize any steady-state effects. Slice timing correction and motion correction was performed in SPM12 with the 11th slice being the reference slice. rsfMRI images of each animal were aligned to a rat brain stereotaxic template [39] with its anatomical image as reference. This step, which uses fourth-order B-spline interpolation, matched the rsfMRI images from animals with different brain volume and shape to a standard rat brain template [40]. Then the images underwent signal detrending to remove signal intensity drift, spike (outlier volume) censoring, orthogonalization of six motion-derived parameters, as well as signals from cerebrospinal fluid, band-pass filtering (0.008–0.15 Hz), and smoothing (FWHM = 0.8 mm) with 3 dB and pass function in AFNI.

For the functional connectivity analysis, the regionally averaged BOLD time courses for each rsfMRI run of individual animals were extracted from the mPFC (containing the prelimbic and anterior cingulate cortex to match with MRS voxel) that was manually defined based on Paxinos and Watson [36] (Fig. 1c). The extracted time course was correlated with other voxels across the whole brain. The correlation map was transformed to a z-score connectivity map using Fisher’s transformation. Connectivity maps of the three rsfMRI runs of each animal were averaged and subjected to a two-sample t test between Ctr males and females to assess baseline sex difference, or between the Ctr and the Eth group separately for male and female animals to assess the effect of ethanol on each sex. 3dFWHMx was used to calculate the spatial smoothness variance which was then used as input in 3dClustSim to estimate the required minimum cluster size to maintain a 5% type 1 error rate [41]. A cluster in the dorsal striatum (dStr) showed significant alterations in functional connectivity with mPFC; therefore, follow-up analyses of connectivity between mPFC and dStr were performed. The regionally averaged BOLD time courses of mPFC and dStr (left or right side, as shown in Fig. 1c) of each animal were correlated using Pearson’s correlation and transformed to a connectivity z-score using Fisher’s transformation. Two-way ANOVAs were performed to assess the effects of ethanol or sex on the connectivity between mPFC and either left or right dStr. A partitioned analysis was performed to estimate the simple effect of ethanol in males and females, respectively, and the simple effect of sex was estimated in Ctr animals to assess the baseline difference between males and females.

¹H-MRS data were fitted using the LCModel package (version 6.3-0G; LCModel Inc., Oakville, ON, Canada) [42]. The criteria to select the reliable metabolite concentrations were based on the Cramer-Rao lower bounds; a value ≤20% was considered an acceptable level of quantification reliability to study the rat brain [43]. Therefore, only metabolites with standard deviations ≤20% were included for further analysis. All concentrations were expressed as mean ± standard error of the mean. Metabolite concentrations were analyzed with two-way ANOVA to assess the effects of ethanol, sex, and ethanol-sex interaction. False discovery rate correction was performed to control the type 1 error rate, and both the uncorrected p values and the corrected probabilities (q values) are reported. In addition, a partitioned analysis was performed to estimate the simple effect of ethanol in males and females, respectively. The simple effect of sex was estimated in Ctr animals to assess the baseline difference between males and females. To understand the relationship between metabolite expression in mPFC and functional connectivity that are significantly affected by ethanol exposure, correlations were tested using Pearson’s correlation test. The significance level was set at p < 0.05. All statistical analyses were performed using SAS software, University Edition (SAS Institute Inc., Cary, NC, USA).

A two-way ANOVA (sex and prenatal group) identified a sex difference in body weight (F_{(1, 40)} = 102.91, p < 0.001), with males being heavier than females. Body weight was not significantly affected by ethanol exposure (F_{(1, 40)} = 1.87, p = 0.179), and there was no ethanol-sex interaction on body weight (F_{(1, 40)} = 0.74, p = 0.396). Although Eth animals tended to be lighter, there were no significant differences between Eth and Ctr animals in males or females. In males, Ctr animals weighed 421 ± 18 g compared with 385 ± 15 g in Eth animals (p = 0.080). In females, Ctr animals weighed 243 ± 8 g compared with 235 ± 13 g in Eth animals (p = 0.748).

Data were also analyzed separately for the subset of animals used for estimation of whole brain volume. In this dataset, there was a main effect of sex on body weight of selected animals; females had significantly lower body weight than males (F_{(1, 8)} = 18.50, p = 0.003). This effect of sex was significant in both Ctr animals (p = 0.007) and Eth animals (p = 0.035). Body weight was not significantly affected by ethanol exposure (F_{(1, 8)} = 0.21, p = 0.662). There was no ethanol-sex interaction on body weight (F_{(1, 8)} = 0.53, p = 0.488) (Fig. 2a).

The brain volume data revealed a significant ethanol-sex interaction (F_{(1, 8)} = 6.87, p = 0.031) and significant effects of ethanol and sex (ethanol: F_{(1, 8)} = 8.55, p = 0.019; sex: F_{(1, 8)} = 11.75, p = 0.009). In Ctr animals, males had significantly larger brain volume than females (p = 0.003). The brain volume was significantly lower in male Eth rats compared to male Ctr rats (p = 0.004), but the brain volume of female Eth rats remained unchanged compared to female Ctr rats (p = 0.835) (Fig. 2b).

After normalizing the brain volume to body weight, the ethanol-sex interaction and the effect of ethanol on brain volume disappeared (ethanol: F_{(1, 8)} = 0.02, p = 0.892; ethanol-sex interaction: F_{(1, 8)} = 0.04, p = 0.848). The analysis identified a significant sex effect on this ratio; females had a higher brain volume/body weight ratio than males (F_{(1, 8)} = 18.27, p = 0.003). This was significant in both Ctr animals (p = 0.013) and Eth animals (p = 0.020) (Fig. 2c). Since brain images were matched to a rat brain template, the brain volume difference between males and females or between Ctr and Eth animals was not a factor in rsfMRI analysis.

The functional connectivity between the mPFC and the whole brain demonstrated a significant difference between Ctr males and Ctr females (Fig. 3a). The functional connectivity between the mPFC and the anterior cingulate cortex, retrosplenial cortex, and hippocampus was significantly lower in Ctr males compared to Ctr females at the level of p < 0.0025 and α < 0.05.

A cluster of voxels (>7 voxels) with significant reduction in connectivity with the mPFC was detected in the dStr of male Eth rats compared to male Ctr rats at the level of p < 0.0025 and α < 0.05 (Fig. 3b); this was most apparent in the connectivity between the mPFC and the left side of the dStr (see below). No differences in mPFC connectivity were observed in female Eth rats compared to female Ctr rats. The mean and standard error for functional connectivity between the mPFC and the dStr is shown in Table 1.

A two-way ANOVA revealed a significant effect of ethanol (F_{(1, 40)} = 5.72, p = 0.022) and a significant ethanol-sex interaction (F_{(1, 40)} = 5.99, p = 0.019) on functional connectivity between the mPFC and the left dStr. However, there was no significant main effect of sex (F_{(1, 40)} = 1.31, p = 0.260). There were also no significant effects on functional connectivity between the mPFC and the right dStr (ethanol: F_{(1, 40)} = 0.50, p = 0.483; sex: F_{(1, 40)} = 0.63, p = 0.432; ethanol-sex interaction: F_{(1, 40)} = 2.17, p = 0.148). A simple effect of ethanol on functional connectivity between the mPFC and the left dStr was apparent in males (p < 0.001) but not females (p = 0.972). A simple effect of sex on this functional connectivity was detected in Ctr animals (p = 0.006) but not in Eth animals (p = 0.412).

As there was a significant difference in the brain volume in Eth males compared with Ctr males, we tested the effect of spatial normalization by analyzing the data of male animals without spatial normalization. We calculated the functional connectivity between the mPFC and the left and right side of the dStr of Ctr and Eth males and performed a two-sample t test to assess between-group difference. The results were similar to those described above, i.e., mPFC-left dStr connectivity was significantly lower in Eth males compared to Ctr males (p = 0.007). The difference was not significant for mPFC-right dStr connectivity (p = 0.117). This suggests that spatial normalization in animals with different brain volumes did not affect the outcomes of the statistical comparisons.

Representative in vivo high-resolution ¹H-MRS localized on the mPFC in a male Eth rat is shown in Figure 1a. Levels of metabolites with standard deviations ≤20% of the mean are shown in Table 2. Two-way ANOVA (sex and prenatal group as factors) revealed significant differences (p < 0.05) in metabolites. However, only the main effect of sex on glucose (Glc) level survived false discovery rate correction.

Glc. The two-way ANOVA identified significant effects of sex (F_{(1, 40)} = 9.41, p = 0.004, q = 0.033) and prenatal group (F_{(1, 40)} = 4.64, p = 0.037, q = 0.241) on Glc levels, but there was no significant ethanol-sex interaction (F_{(1, 40)} = 0.78, p = 0.382, q = 0.732). The concentration of Glc was higher in males than females and lower in Eth than Ctr animals. The simple effect test confirmed that male Ctr animals had significantly higher Glc levels than female ones (p = 0.003). This sex difference was not significant in Eth animals (p = 0.173). Prenatal ethanol exposure significantly reduced Glc levels in males (p = 0.019) but not females (p = 0.424) (Fig. 4a).

Phosphorylcholine (PCh). The PCh level was significantly lower in Eth animals (F_{(1, 40)} = 4.75, p = 0.035, q = 0.241). There was no main effect of sex (F_{(1, 40)} = 2.16, p = 0.150, q = 0.290) or ethanol-sex interaction on PCh (F_{(1, 40)} = 1.72, p = 0.197, q = 0.569). The simple effect of ethanol was apparent only in females; PCh was lower in Eth than Ctr females (p = 0.032). There was no difference in males (p = 0.487) (Fig. 4b).

Gamma-Aminobutyric Acid (GABA) and Glutamate (Glu). GABA and Glu were not affected by ethanol, sex, or ethanol-sex interaction (GABA: ethanol F_{(1, 40)} = 0.06, p = 0.802, q = 0.898, sex F_{(1, 40)} = 1.51, p = 0.226, q = 0.326, ethanol-sex interaction F_{(1, 40)} = 0.46, p = 0.502, q = 0.732; Glu: ethanol F_{(1, 40)} = 0.11, p = 0.744, q = 0.898, sex F_{(1, 40)} = 0.41, p = 0.525, q = 0.525, ethanol-sex interaction F_{(1, 40)} = 1.70, p = 0.200, q = 0.569).

GABA/Glu Ratio. Two-way ANOVA identified a significant effect of sex (F_{(1, 40)} = 5.46, p = 0.025, q = 0.108) and ethanol-sex interaction (F_{(1, 40)} = 6.34, p = 0.016, q = 0.208) on the GABA/Glu ratio. The ratio was not significantly different between Ctr males and females (p = 0.885) but was significantly higher in Eth females compared to Eth males (p = 0.004). A significant effect of ethanol was observed in females; ethanol exposure increased the GABA/Glu ratio (p = 0.032).

Correlations between the significantly changed metabolites and functional connectivity between the mPFC and the dStr were performed within the sex that showed the metabolite change. Specifically, in male animals, we tested the correlations between mPFC-dStr functional connectivity and Glc. In female animals, we tested the correlations between mPFC-dStr functional connectivity and PCh or GABA/Glu ratio. No significant correlations were found (all p values > 0.05).

Our findings suggest that the effects of prenatal ethanol exposure on brain size, functional connectivity, and neurometabolite expression are sex-dependent. Although only assessed in a subset of animals, male rats exposed to ethanol prenatally showed a significantly smaller brain volume. This effect was not seen in females. Reduction in overall brain size or volumes of various brain cortical and subcortical areas has been previously reported following prenatal alcohol exposure [9, 44-48]. However, in this study the apparent effect on brain volume disappeared after these data had been normalized to body weight. Thus, it is possible that the brain volume change reflects the smaller size of Eth animals.

In this study, male but not female rats showed ethanol-induced changes in functional connectivity between the mPFC and the dStr, suggesting that this circuit in males may be particularly vulnerable to prenatal alcohol exposure. Alcohol-induced alterations in activity in the circuit between frontal cortex and dStr are seen in humans during executive function tasks such as response inhibition [49] and working memory [18]. Here, for the first time, we show that in animals, functional connectivity differences in this circuit also exist at rest. The sex-dependent effect of prenatal alcohol exposure on brain functional connectivity between two regions critical to executive function is consistent with findings in rat behavior. Both male and female rats prenatally exposed to ethanol exhibit spatial working memory deficits, but males show deficits in working memory [e.g., 6, 50]. Thus, sex differences in functional connectivity may affect behavior differently between the sexes and may be dependent on the difficulty of task demands.

Our findings are consistent with previous reports that Eth rats have lower immediate early gene expression, suggestive of lower metabolic activity, in the PFC and the caudate nucleus [50]. However, in another rsfMRI study with rats exposed to a lower concentration of ethanol prenatally [51], both sexes exhibited reduced coordinated activity in cortical, hippocampal, and cerebellar regions that were functionally connected in control rats, although functional connectivity was more globally altered in ethanol-exposed males than females.

Our results revealed that functional connectivity between the mPFC and the striatum was only altered in the left side of the dStr, suggesting that the effect of ethanol exposure can be asymmetric. Hemispheric asymmetry and regional volume asymmetry have been reported in humans prenatally exposed to ethanol [52, 53]. A survey study that summarized previous reports on subjects with drug exposure demonstrated more reported effects on brain structures in the right than left hemisphere after prenatal ethanol exposure [52]. Another study on caudate volume revealed enlarged left caudate relative to right caudate of prenatally exposed subjects when compared to healthy control subjects [53]. Here, we provide evidence of an asymmetric effect of ethanol on functional connectivity in an animal model.

This study did not assess the vasculature condition in the brain, which may change after prenatal alcohol exposure [54]. Although we cannot exclude the possibility that changes in vasculature development may drive differences between male and female rats, others have reported no significant alteration in average blood perfusion in the frontal cortex of prenatal ethanol-exposed rats during resting state [51].

Our current study also showed that prenatal ethanol exposure affects neurochemical expression in the mPFC differently in male and female offspring, with reduced Glc in males and PCh in females and an increased GABA/Glu ratio in females. Glc is critical for the normal development and function of the brain [e.g., 55]. Besides being a major energy source, Glc plays an important role in processes including neurotransmitter synthesis and regulation of cerebral blood flow. Impaired Glc transport and uptake as well as alterations in metabolites have been reported in prenatal ethanol-exposed rats, humans, as well as in cell culture studies [56-59]. In control animals, Glc levels were higher in males and ethanol exposure lowered these levels. The current study provides in vivo evidence of Glc imbalance in prenatal Eth rats.

PCh is the main metabolite that contributes to the Cho resonance peak. It is the precursor metabolite of Cho in the glycine, serine, and threonine metabolism pathways (KEGG, map00260; http://www.genome.jp/kegg/pathway.html) and an intermediate between Cho and cytidine-diphosphate choline in the glycerophospholipid metabolism pathway (KEGG, map00564). In the brain, it is involved in pathways of phospholipid synthesis and degradation, and thus may reflect membrane synthesis and degradation [60, 61]. Lower Cho levels in the brain have been associated with neurological disorders such as multiple sclerosis and Alzheimer’s disease [62, 63]. Cho supplementation has been shown to improve memory, visual stimulus processing, and cognitive deficits in rats and humans after prenatal alcohol exposure [5, 6, 64, 65]. In Ctr animals, levels of PCh are similar in males and females, but Eth females show significantly less. The low PCh level found in the PFC of prenatally ethanol-exposed female rats may provide indirect evidence of involvement of the cholinergic system in FASD and/or changes in membrane turnover.

The GABA/Glu ratio is an indicator of inhibitory/excitatory balance. In this study, there was no significant alteration of GABA or Glu levels in either sex, but there was an increase in the GABA/Glu ratio in females that resulted from a nonsignificant increase in GABA and decrease in Glu levels. Others have shown that prenatal ethanol exposure affects the balance between the GABAergic and glutamatergic systems. This is seen as an increased number of GABAergic interneurons in the mPFC and an increased ratio of inhibitory/excitatory spontaneous postsynaptic currents in this region [13]. Such changes may contribute to behavior phenotypes including working memory, cognitive flexibility, and social behaviors [see 66, 67 for review].

To determine whether changes in metabolites might be related to changes in functional connectivity, we performed correlation analyses. None of the significantly altered metabolites described above showed any correlation with functional connectivity between the mPFC and the dStr. This could mean that the metabolite changes are not related to connectivity; however, the metabolites were measured in the PFC, whereas functional connectivity depends on the temporal relationship between blood oxygen level changes in the two regions assayed, so the outcome(s) could be driven by changes in the mPFC, the striatum, or both.

Other works describe anatomical and/or functional effects of prenatal ethanol exposure on the PFC and the striatum. The PFC shows increased numbers of parvalbumin-positive cells and an increased inhibitory/excitatory balance in Eth animals [14], and this effect could contribute to the change in the GABA/Glu ratio seen here. Prenatal ethanol exposure also alters the dopaminergic system in the striatum, including a reduced number of spontaneously active dopaminergic neurons in the dStr [68], an increase in dopaminergic activity [69, 70], and alter-ations in receptor activity [71, 72]. Other reports show somewhat global decreases in 2-deoxyglucose uptake in brains of ethanol-exposed rats [e.g., 73]. Some of these studies do not report which sex was examined, and others only included a single sex, the exception being the study by Sobrian et al. [72], which did not identify differences in behavioral responses to dopamine D1 receptor agonists or antagonists between the sexes.

In addition to the sex-dependent effect of ethanol, the present study revealed sex differences between Ctr male and female animals. As expected, female animals had lower body weight and brain volume compared to male animals. However, when normalized to body weight, the brain volume/body weight ratio was significantly higher in female animals than male animals. Similar effects have been reported for brain weight/body weight ratios, the ratio being higher in females than males [e.g., 74]. The functional connectivity between the mPFC and the anterior cingulate cortex, retrosplenial cortex, and hippocampus was significantly lower in male than female animals. This was consistent with previous studies on humans which reported higher overall anatomical and functional cortical connectivity in females than males [75, 76]. In addition, higher Glc levels were detected in the mPFCs of male rats compared to those of female rats. In previous studies on mPFC development, female rats showed more pruning during adolescence, which resulted in fewer neurons, dendrites, and synapses in the female than the male mPFC [77-79]. This may be associated with a higher energy demand in males which may, in turn, underlie the higher Glc level seen in this study. Taken together with previous studies, sex difference in the PFC may contribute to the sex-dependent effects of ethanol exposure on functional connectivity and neurochemicals. Our findings strongly suggest the need for including sex as a factor on the effects of prenatal ethanol exposure in future studies.

In the present study, we report that the effects of prenatal alcohol exposure on brain volume, frontostriatal connectivity, and metabolite levels in the mPFC are sex-dependent, with disrupted frontostriatal functional connectivity and lower Glc levels in Eth male rats, but lower PCh levels and a higher GABA/Glu ratio in female rats. These sex-dependent effects of prenatal alcohol exposure on the mPFC underscore the necessity for considering sex as a variable when studying outcomes and deciphering underlying mechanisms of executive problems after prenatal alcohol exposure.

The authors thank Marie Hanscom for her assistance in generating the animals. They would also like to thank the Core for Translational Research in Imaging @ Maryland, part of the University of Maryland School of Medicine Center for Innovative Biomedical Resources, Baltimore, MD.

All animal work was approved by the University of Maryland Institutional Animal Care and Use Committee and performed in accordance with NIH guidelines.

The authors have no conflicts of interest to declare.

S.M. Mooney receives funding from the NIH (NIAAA AA024980, AA022413, AA017823, AA026109).

S. Tang analyzed the rsfMRI data and wrote the draft of the manuscript. S. Xu and W. Zhu performed and processed the imaging. S. Xu analyzed the MRS data. S. Xu, J. Waddell, R.P. Gullapalli, and S.M. Mooney edited the manuscript. J. Waddell and S.M. Mooney conceived and designed the study. All authors read and approved the manuscript.