Maternal P7C3-A20 Treatment Protects Offspring from Neuropsychiatric Sequelae of Prenatal Stress

Abstract

Aims: Impaired embryonic cortical interneuron development from prenatal stress is linked to adult neuropsychiatric impairment, stemming in part from excessive generation of reactive oxygen species in the developing embryo. Unfortunately, there are no preventive medicines that mitigate the risk of prenatal stress to the embryo, as the underlying pathophysiologic mechanisms are poorly understood. Our goal was to interrogate the molecular basis of prenatal stress-mediated damage to the embryonic brain to identify a neuroprotective strategy.

Results: Chronic prenatal stress in mice dysregulated nicotinamide adenine dinucleotide (NAD⁺) synthesis enzymes and cortical interneuron development in the embryonic brain, leading to axonal degeneration in the hippocampus, cognitive deficits, and depression-like behavior in adulthood. Offspring were protected from these deleterious effects by concurrent maternal administration of the NAD⁺-modulating agent P7C3-A20, which crossed the placenta to access the embryonic brain. Prenatal stress also produced axonal degeneration in the adult corpus callosum, which was not prevented by maternal P7C3-A20.

Innovation: Prenatal stress dysregulates gene expression of NAD⁺-synthesis machinery and GABAergic interneuron development in the embryonic brain, which is associated with adult cognitive impairment and depression-like behavior. We establish a maternally directed treatment that protects offspring from these effects of prenatal stress.

Conclusion: NAD⁺-synthesis machinery and GABAergic interneuron development are critical to proper embryonic brain development underlying postnatal neuropsychiatric functioning, and these systems are highly susceptible to prenatal stress. Pharmacologic stabilization of NAD⁺ in the stressed embryonic brain may provide a neuroprotective strategy that preserves normal embryonic development and protects offspring from neuropsychiatric impairment. Antioxid. Redox Signal. 35, 511–530.

Introduction

Human and animal studies show that prenatal stress can harm neurodevelopment and lead to lifelong neuropsychiatric complications in offspring, including cognitive dysfunction (27, 46, 57). These stressors include, for example, major depression, war, exposure to violence, natural disaster, or death or major illness of a loved one (8, 35, 37, 40, 54, 66). Unfortunately, the underlying mechanisms are poorly understood, and consequently there is a lack of effective means to protect the developing fetus.

Prenatal stress is frequently modeled in mice through repetitive physical restraint of the pregnant mother, which increases generation of reactive oxygen species (ROS) in the offspring (9, 51). Oxidative stress contributes to neuropsychiatric disorders, including attention-deficit hyperactivity disorder, autism spectrum disorder (ASD), depression, and schizophrenia (16, 18, 23). We have also shown that increased ROS in the prenatally stressed embryonic brain disrupt development of inhibitory cortical gamma-aminobutyric acid (GABA) containing interneurons, with negative consequences persisting into adulthood (9, 43). As with oxidative stress, dysfunction of cortical and hippocampal GABAergic interneuron systems is also implicated in the same neuropsychiatric disorders (15, 17, 19, 39, 42).

The subtype of parvalbumin (PV)-expressing GABAergic interneurons is especially vulnerable to redox stress due to high baseline activity (61), and PV interneuron deficits are tightly associated with neuropsychiatric disorders, including schizophrenia, Alzheimer's disease, ASD, bipolar disorder, and aging-related cognitive decline (39, 43). These neurons manifest complex developmental pathways required for normal brain circuitry, and this complexity is subject to multiple points of compromise that can preclude normal brain development. This class of neurons is also more susceptible to dysfunction across the life span due to unusually high-energy requirements and their strong excitatory drive (59). Thus, substantial evidence suggests that increased ROS production from prenatal stress disrupts normal embryonic GABAergic interneuron development, which impairs neuropsychiatric functioning extending into adulthood.

Innovation

We show for the first time that prenatal stress induces aberration in nicotinamide adenine dinucleotide (NAD⁺)-synthesis machinery in the embryonic brain, which is associated with disrupted interneuron development and neuropsychiatric deficits in adult offspring. We also show that maternal administration of the NAD⁺-stabilizing agent P7C3-A20 concurrent with prenatal stress protects offspring from these deleterious outcomes.

With this risk to offspring neurodevelopment in mind, there is a clear unmet need for a neuroprotective treatment that mitigates the effects of prenatal oxidative stress on the developing embryonic brain. However, developing cells also require a certain amount of basal ROS generation (6, 14, 58), and we have shown that excessive depletion of ROS can also be deleterious to the developing brain (9).

Fundamentally, the brain's ability to buffer ROS in the appropriate physiologic range is critical to proper development, and much of this is related to nicotinamide adenine dinucleotide (NAD⁺) metabolism (47). For example, a high NAD⁺/NADH ratio protects neurons from oxidative stress, whereas a low NAD⁺/NADH ratio promotes ROS production (71). Furthermore, neuronal health is impacted by varying expression levels of NAD⁺-synthesis machinery (55, 72). We therefore hypothesized that embryonic brain NAD⁺-producing mechanisms might be perturbed by prenatal stress, and if so, that this might provide a new approach to therapeutic intervention. Specifically, we hypothesized that P7C3-A20, a neuroprotective aminopropyl carbazole that preserves normal cellular levels of NAD⁺ under conditions of stress, might stabilize NAD⁺ levels and related processes in the prenatally stressed developing brain.

As summarized in Figure 1, we found that chronic maternal prenatal stress disrupted normal expression of NAD⁺ synthesis enzymes in the embryonic brain and perturbed cortical interneuron development, leading to cognitive deficits and depression-like behavior in adulthood. P7C3 neuroprotective molecules preserve normal NAD⁺ levels in the face of otherwise overwhelming toxic insult, and promote cell survival in a variety of neurodegenerative conditions in rodents and nonhuman primates (4, 5, 20, 21, 33, 38, 41, 45, 50, 55, 56, 65, 67, 69, 74). The potent P7C3 analog, P7C3-A20, also rescues depressive- and anxiety-like phenotypes in mice (5, 70).

FIG. 1.

Summary graphic illustration. Maternal stress during pregnancy impairs embryonic expression of NAD⁺-synthesis genes and cortical interneuron development, leading to lifelong neuropsychiatric complications. These deleterious effects of prenatal stress are prevented by concurrent maternal treatment with P7C3-A20. NAD⁺, nicotinamide adenine dinucleotide. Color images are available online.

We show here that concurrent administration of maternal P7C3-A20 during chronic prenatal stress provides a stabilizing effect on NAD⁺ synthesis enzyme expression and GABAergic interneuron development, which protects prenatally stressed offspring from developing cognitive deficits and depression-like behavior in adulthood.

Results

Prenatal stress disrupts NAD⁺-synthesis machinery in the embryonic brain, which is rescued by concurrent maternal P7C3-A20 treatment

To investigate the effects of chronic prenatal stress, we exposed pregnant mothers to 45 min sessions of restraint stress three times daily throughout pregnancy, beginning on embryonic day 5 (E5) (Fig. 2). No sex differences in the data for embryonic day 18 (E18) were identified, so data were pooled for these measures. We found no prenatal stress-induced changes in expression of the antioxidant and oxidative stress-sensitive genes that we previously examined in the embryonic brain after prenatal stress (10): glutathione peroxidase 1 (Gpx1), nuclear factor erythroid 2-related factor 2 (Nrf2), catalase (Cat), tropinone reductase 1 (Tr1), sestrin 1 (Sesn1), sestrin 2 (Sesn2), or sestrin 3 (Sesn3) (Supplementary Fig. S1).

FIG. 2.

Experimental timelines. (A) Long timeline for adult behavior and collection. (B) Short timeline for embryonic tissue collection. (C) Molecular structure of the P7C3 analog used in this study, P7C3-A20.

However, we did discover striking changes in embryonic brain expression of genes related to NAD⁺-synthesis machinery. Specifically, prenatal stress decreased expression of nicotinamide nucleotide adenylyltransferase 2 (Nmnat2) (Fig. 3B) (a priori two-tailed t-test, t = 2.213, df = 12, p = 0.0470) and nicotinamide phosphoribosyltransferase (Nampt) (Fig. 3D) (a priori two-tailed t-test, t = 2.508, df = 12, p = 0.0275), and increased expression of nicotinamide nucleotide adenylyltransferase 3 (Nmnat3) (Fig. 3C) (trend, a priori two-tailed t-test, t = 2.109, df = 14, p = 0.0534).

FIG. 3.

NAD⁺-related gene expression and NAD⁺/NADH ratio in E18 forebrain relative to Gapdh. (A)Nmnat1 expression is increased by maternal P7C3-A20 treatment (main effect of P7C3-A20, F[1,28] = 5.004, *p = 0.0334). (B) Prenatal stress decreases Nmnat2 expression (two-tailed Student's t-test, t = 2.213, df = 12, *p = 0.0470). (C) Trend increase of Nmnat3 expression by stress (two-tailed Student's t-test, t = 2.109, df = 14, p = 0.0534). (D) Nampt is decreased by prenatal stress (two-tailed Student's t-test, t = 2.508, df = 12, *p = 0.0275). (E) NAD⁺/NADH ratio is not changed by prenatal stress nor by maternal P7C3-A20 (n.s.). Nampt, nicotinamide phosphoribosyltransferase; Nmnat1, nicotinamide nucleotide adenylyltransferase 1; Nmnat2, nicotinamide nucleotide adenylyltransferase 2; Nmnat3, nicotinamide nucleotide adenylyltransferase 3; n.s., not significant.

After establishing in pregnant mothers that orally administered P7C3-A20 crossed the placenta and entered the embryonic brain at biologically relevant levels (Supplementary Fig. S2), we assessed its effects on expression of these three enzymes. In each case, concurrent maternal administration of P7C3-A20 throughout prenatal stress fully normalized expression of these genes in offspring brain, supported by trend interactions between stress and P7C3-A20 (Fig. 3A–C) (Nmnat2: trend, F_interaction[1,26] = 3.729, p = 0.0644; Nampt: trend, F_interaction[1,26] = 3.848, p = 0.0606; Nmnat3: trend, F_interaction[1,28] = 3.319, p = 0.0782). Nicotinamide nucleotide adenylyltransferase 1 (Nmnat1) expression was not affected by prenatal stress but was increased by maternal P7C3-A20 treatment (Fig. 3D) (F_P7C3-A20[1,28] = 5.004, p = 0.0334).

Interestingly, the ratio of NAD⁺/NADH in E18 forebrain was normal regardless of prenatal stress and P7C3-A20 treatment (n.s., Fig. 3E). As with prenatal stress, P7C3-A20 did not alter expression of any ROS-related genes (Supplementary Fig. S1). We also assessed gene expression in the adult brain of the NAD⁺-synthesizing enzymes that were perturbed at E18, and all were normalized except Nmnat2, which was increased by prenatal stress in female offspring hippocampus (Supplementary Fig. S3).

Prenatal stress disrupts GABAergic interneuron development, which is rescued by concurrent maternal P7C3-A20 treatment

We have previously reported that prenatal stress impairs GABAergic interneuron development in the embryonic forebrain (9, 25, 43, 62). Thus, we investigated here whether prenatal stress additionally alters forebrain transcription of genes that are preferentially expressed in developing interneurons (3). Gene expression of the calcium binding protein Pv, specific to the fast spiking interneurons that are prominently implicated in neuropsychiatric disorders (59), was trend decreased by prenatal stress and normalized by P7C3-A20 (Fig. 4A) (trend, a priori two-tailed t-test, t = 2.079, df = 14, p = 0.0565). Furthermore, the ratio of Pv/glutamate decarboxylase 1 (Gad1) expression was significantly decreased by prenatal stress and restored by P7C3-A20 (Fig. 4B) (a priori two-tailed t-test, t = 2.146, df = 14, p = 0.0499).

FIG. 4.

Interneuron-related gene expression in E18 forebrain relative to Gapdh. (A)Pv expression is trend decreased by prenatal stress (two-tailed Student's t-test, t = 2.079, df = 14, p = 0.0565). (B) The ratio of Pv to Gad1 expression is decreased by prenatal stress (two-tailed Student's t-test, t = 2.146, df = 14, *p = 0.0499). (C) Neto1 expression is decreased by prenatal stress (two-tailed Student's t-test, t = 2.591, df = 12, *p = 0.0236). (D) Cxcr4 expression is trend decreased by prenatal stress (two-tailed Student's t-test, t = 2.115, df = 15, p = 0.0516), and rescued by P7C3-A20 (interaction of stress × P7C3-A20 via two-way ANOVA, F[1,31] = 6.784, *p = 0.0140). (E) Som expression is increased by maternal P7C3-A20 treatment (main effect of P7C3-A20 via two-way ANOVA, F[1,26] = 4.956, *p = 0.0349). ANOVA, analysis of variance; Cxcr4, C-X-C chemokine receptor type 4; Gad1, glutamate decarboxylase 1; Neto1, neuropilin and tolloid like 1; Som, somatostatin.

Neuropilin and tolloid like 1 (Neto1), a regulator of synaptic input onto hippocampal interneurons, was also decreased by prenatal stress and restored by P7C3-A20 (a priori two-tailed t-test, t = 2.591, df = 12, p = 0.0236, Fig. 4C). Expression of C-X-C chemokine receptor type 4 (Cxcr4), which mediates interneuron migration from the ganglionic eminence into the cortex, was also decrea-sed by prenatal stress (trend, a priori two-tailed t-test, t = 2.115, df = 15, p = 0.0516) and restored by P7C3-A20 (F_interaction[1,31] = 6.784, p = 0.0140, Fig. 4D). Another calcium binding protein, somatostatin (Som), was not affected by prenatal stress but was increased by maternal P7C3-A20 (F_P7C3-A20[1,26] = 4.956, p = 0.0349, Fig. 4E).

Expression of several other genes relevant to interneuron development, such as GABA receptors, GABA-producing enzymes, and GABA-specific ion channels, remained unchanged by prenatal stress or maternal treatment with P7C3-A20, including neurexophilin 1 (Nxph1), atypical chemokine receptor 3 (Ackr3), gamma-aminobutyric acid type A receptor subunit alpha5 (Gabra5), vesicular inhibitory amino acid transporter (Slc32a1), Gad1, glutamate decarboxylase 2 (Gad2), potassium chloride cotransporter 2 (Kcc2), basolateral Na-K-Cl symporter (Nkcc1), GABA transporter 1 (Gat1), glutamate ionotropic receptor AMPA type subunit 1 (Gria1), and GABA type A receptor subunit gamma2 (Gabrg2) (Supplementary Fig. S4). Despite this, the aberrations we found in forebrain gene expression indicated that prenatal stress-mediated neurodevelopmental disruption was ameliorated by concurrent maternal P7C3-A20 treatment.

Prenatal stress induces sex-specific aberrations in GABAergic systems in the medial prefrontal cortex of adult offspring, which are partially rescued by concurrent maternal P7C3-A20 treatment

To probe whether GABAergic systems remained perturbed in adulthood after chronic prenatal stress, we first measured GAD67⁺ (encoded by the Gad1 gene) and PV⁺ cell cortical and hippocampal density through unbiased stereology, analyzing male and female offspring separately. We also assessed the expression of genes central to the maturation and function of cortical and hippocampal inhibitory interneuron function.

As we observed previously in the medial prefrontal cortex (mPFC) of male offspring (43), prenatal stress caused a trend increase of PV⁺ cell density (Fig. 5A) (F_stress[1,17] = 3.669, p = 0.07224). Also similar to previous results, prenatal stress did not change total GAD67⁺ cell density (Fig. 5B), and only a trend increase of PV⁺/GAD67⁺ cell ratio was observed (Fig. 5C) (F_stress[1,17] = 3.474, p = 0.0797). Moreover, these measures were not altered by maternal P7C3-A20. No effect of either stress or P7C3-A20 was observed for mPFC volume (Fig. 5D).

FIG. 5.

Prenatal stress and maternal P7C3-A20 cause sex-specific changes in GAD67⁺ and PV⁺ cell density in mPFC. (A) Trend increase in PV⁺ cell density in male mPFC caused by stress (main effect of stress, two-way ANOVA, F[1,17] = 3.669, p = 0.0724). (B) No differences in GAD67⁺ cell density (n.s.). (C) Trend increase in ratio of PV⁺/GAD67⁺ cells in male mPFC caused by stress (main effect of stress, two-way ANOVA, F[1,17] = 3.474, p = 0.0797). (D) No effect on mPFC volume in males (n.s.). (E) P7C3-A20 caused an increase in PV⁺ cell density in females (main effect of P7C3-A20, two-way ANOVA, F[1,20] = 5.306, *p = 0.0321; post hoc Tukey's multiple comparisons test, PS Veh vs. PS P7C3 ^&p = 0.049). (F) In females, prenatal stress caused a decrease in GAD67⁺ cell density (post hoc Tukey's multiple comparisons test, Ctrl Veh vs. PS Veh ^&p = 0.0289), normalized by P7C3-A20 (interaction of stress × P7C3-A20, two-way ANOVA, F[1,20] = 10.62, **p = 0.0039). (G) Females: PV⁺/GAD67⁺ ratio is not affected by stress or P7C3-A20 (n.s.). (H) Prenatal stress increased mPFC volume in females (main effect of stress, two-way ANOVA, F[1,20] = 6.758, *p = 0.0171). mPFC, medial prefrontal cortex; PV, parvalbumin.

While mPFC PV⁺ cell density in female offspring was not affected by prenatal stress, it was increased by P7C3-A20, an effect mainly driven by P7C3-A20 affecting prenatally stressed females (Fig. 5E) (F_P7C3-A20[1,20] = 5.306, p = 0.0321; post hoc Tukey's multiple comparisons test, PS Veh vs. PS P7C3 p = 0.049). Also in contrast to males, prenatal stress decreased female total GAD67⁺ cell density (post hoc Tukey's multiple comparisons test, Ctrl Veh vs. Ps Veh p = 0.0289), which was rescued by concurrent maternal P7C3-A20 treatment (Fig. 5F) (F_interaction[1,20] = 10.62, p = 0.0039). As with males, the PV⁺/GAD67⁺ cell ratio in female mPFC was not affected by prenatal stress or maternal P7C3-A20 treatment (Fig. 5G). Interestingly, prenatal stress increased mPFC volume in both vehicle and P7C3-A20-treated female offspring (F_stress[1,20] = 6.758, p = 0.0171, Fig. 5H).

With respect to gene expression, prenatal stress increased Pv transcript in the mPFC in male but not female offspring (a priori two-tailed t-test, t = 2.591, df = 14, p = 0.0214, Fig. 6A, D), and the aberration in males was rescued by maternal P7C3-A20 treatment (Fig. 6A). While mPFC Gad1 expression was unaffected by prenatal stress or maternal P7C3-A20 in males (Fig. 6B), the ratio of Pv/Gad1 expression in males was increased by prenatal stress and rescued by P7C3-A20 (Fig. 6C) (F_stress[1,26] = 5.814, p = 0.0233; post hoc Tukey's multiple comparisons test, Ctrl Veh vs. PS Veh p = 0.0305). No changes in Pv or Gad1 expression, or their ratio, were observed in females (n.s., Fig. 6E, F). Finally, in both males and females, no changes in other genes related to interneurons, neuropeptide Y (Npy), Som, or brain-derived neurotrophic factor (Bdnf) were observed in the mPFC as a function of either prenatal stress or maternal P7C3-A20 (Supplementary Fig. S5).

FIG. 6.

Prenatal stress and maternal P7C3-A20 cause sex-specific changes in inhibitory interneuron-related gene expression in the mPFC relative to Gapdh. (A) Prenatal stress caused an increase in Pv expression in male mPFC (two-tailed Student's t-test, t = 2.591, df = 14, *p = 0.0214), which was rescued by P7C3-A20 (n.s.). (B) Gad1 expression was not changed in male mPFC (n.s.). (C) Prenatal stress caused an increase in Pv/Gad1 ratio (main effect of stress, two-way ANOVA, F[1,26] = 5.814, *p = 0.0233; post hoc Tukey's multiple comparisons test, Ctrl Veh vs. PS Veh ^&p = 0.0388), which was rescued by P7C3-A20 (n.s.). In females, there was no change in Pv (D) or Gad1 (E) expression, and no change in Pv/Gad1 ratio (F).

Prenatal stress induces sex-specific aberrations in GABAergic systems in the hippocampus of adult offspring, which are partially rescued by concurrent maternal P7C3-A20 treatment

Next, we investigated the same measures of interneuron populations and gene expression in the hippocampus of male and female offspring. In males, our stereological measures revealed a main effect of maternal P7C3-A20 to increase PV⁺ cell density in hippocampus, with no effect of prenatal stress (Fig. 7A) (F_P7C3-A20[1,15] = 7.064, p = 0.0160; post hoc Tukey's multiple comparisons test, Ctrl Veh vs. PS P7C3 p = 0.0266). Also in males, P7C3-A20 independently increased GAD67⁺ cell density in the hippocampus, with this measure unaffected by prenatal stress alone (Fig. 7B) (F_P7C3-A20[1,15] = 18.08, p = 0.0007). Similar to the mPFC, prenatal stress increased PV⁺/GAD67⁺ cell ratio in the male hippocampus, which was not rescued by P7C3-A20 (Fig. 7C) (F_stress[1,15] = 7.572, p = 0.0148; post hoc Tukey's multiple comparisons test, Ctrl Veh vs. PS Veh p = 0.0450).

FIG. 7.

Prenatal stress and maternal P7C3-A20 cause sex-specific changes in GAD67⁺ and PV⁺ cell density in the hippocampus in males only. (A) In male offspring hippocampus, maternal P7C3-A20 causes increased PV⁺ cell density (main effect of P7C3-A20, F[1,15] = 7.064, *p = 0.0160; post hoc Tukey's multiple comparisons test, Ctrl Veh vs. PS P7C3 ^&p = 0.0266). (B) Maternal P7C3-A20 also causes increased GAD67⁺ cell density in hippocampus (main effect of P7C3-A20, F[1,15] = 18.08, ***p = 0.0007). (C) PV⁺/GAD67⁺ cell ratio was increased by prenatal stress in male hippocampus (main effect of stress, F[1,15] = 7.572, *p = 0.0148; post hoc Tukey's multiple comparisons test, Ctrl Veh vs. PS Veh ^&p = 0.0450). Male hippocampal volume was unchanged (D). In females, there were no changes in hippocampal PV⁺ density, GAD67⁺ density, PV⁺/GAD67⁺ ratio, or volume (E–H).

In female offspring, there were no effects of prenatal stress on stereological counts in hippocampus (Fig. 7E–G), and hippocampal volume was not affected in either sex (Fig. 7D, H).

With respect to interneuron gene expression, prenatal stress increased Pv expression in male offspring hippocampus (trend, a priori two-tailed t-test, t = 2.113, df = 15, p = 0.0518), which was normalized by P7C3-A20 as in male mPFC (Fig. 8A) (F_interaction[1,28] = 4.813, p = 0.0367). However, a prenatal stress-induced decrease in Gad1 expression in male hippocampus was not rescued by P7C3-A20 (Fig. 8B) (F_stress[1,30] = 7.436, p = 0.0106; post hoc Tukey's multiple comparisons test, Ctrl Veh vs. PS Veh p = 0.0148, Ctrl Veh vs. PS P7C3, p = 0.0402). Importantly, prenatal stress increased the Pv/Gad1 expression ratio in male hippocampus (a priori two-tailed t-test, t = 2.675, df = 15, p = 0.0173), which was fully rescued by maternal P7C3-A20, as in male mPFC (Fig. 8C) (F_interaction[1,28] = 4.795, p = 0.0370).

FIG. 8.

Prenatal stress and maternal P7C3-A20 cause sex-specific changes in inhibitory interneuron-related gene expression in the hippocampus relative to Gapdh. (A) Pv expression in male hippocampus is trend increased by prenatal stress (two-tailed Student's t-test, t = 2.113, df = 15, p = 0.0518) and rescued by maternal P7C3-A20 (interaction of stress × P7C3-A20 via two-way ANOVA, F[1,28] = 4.813, *p = 0.0367). (B) Gad1 expression in male hippocampus is decreased by prenatal stress (main effect of stress via two-way ANOVA, F[1,30] = 7.436, *p = 0.0106; post hoc Tukey's multiple comparisons test, Ctrl Veh vs. PS Veh, ^&p = 0.0148, Ctrl Veh vs. PS P7C3, ^&p = 0.0402). (C) Ratio of Pv to Gad1 expression in male hippocampus is increased by prenatal stress (two-tailed Student's t-test, t = 2.675, df = 15, *p = 0.0173), and rescued by P7C3-A20 (interaction of stress × P7C3-A20 via two-way ANOVA, F[1,28] = 4.795, *p = 0.0370). (D–F) neither prenatal stress nor maternal P7C3-A20 treatment changed Pv, Gad1, or Pv/Gad1 ratio expression in female hippocampus.

Similar to mPFC, hippocampal interneuron gene expression in female offspring was unaffected by either stress or maternal P7C3-A20 treatment (Fig. 8E–H). Finally, as was seen in the mPFC, no changes in Npy, Som, or Bdnf expression were observed in hippocampus of male or female offspring as a function of prenatal stress or maternal P7C3-A20 treatment (Supplementary Fig. S6).

Prenatal stress impairs adult offspring cognition, which is rescued by concurrent maternal P7C3-A20 treatment

We also assessed whether adult cognitive behavior was impaired in our model of prenatal stress. Learning and memory were assessed in adult offspring by two assays beginning at 8 weeks of age: the accelerating rotarod and fear conditioning (FC).

On the rotarod task, all groups showed similar learning over 10 trials across 2 days of training (Supplementary Fig. S7). However, prenatal stress significantly impaired 2-week memory of this learning in both male and female offspring (Fig. 9A) (a priori two-tailed t-test, t = 1.961, df = 99, p = 0.0527). Furthermore, maternal P7C3-A20 treatment during stress protected offspring from this memory deficit (main effect of P7C3-A20 via two-way analysis of variance [ANOVA], F_P7C3-A20[1,197] = 4.106, post hoc Tukey's multiple comparisons test, PS Veh vs. PS P7C3, p = 0.079). On the FC task, prenatal stress reduced freeze time in both male and female offspring (Fig. 9B), indicating a deficit in fear memory behavior. We observed an interaction between stress and maternal P7C3-A20 treatment (F_interaction[1,194] = 10.58, p = 0.0013), demonstrating rescue of FC by maternal P7C3-A20.

FIG. 9.

Prenatal stress and maternal P7C3-A20 treatment affect adult offspring learning and depression-like behaviors. (A) Rotarod: deficit caused by prenatal stress (trend, two-tailed Student's t-test, t = 1.961, df = 99, p = 0.0527), and main effect of treatment (F[1,197] = 4.106, *p = 0.0441). (B) Fear conditioning: deficit caused by prenatal stress (two-tailed Student's t-test, t = 2.437, df = 97, *p = 0.0167), interaction between stress and maternal P7C3-A20 treatment (F[1,194] = 10.58, **p = 0.0013). (C) Tail suspension test: main effect of prenatal stress (F[1,202] = 5.466, *p = 0.0204) and maternal P7C3-A20 treatment (F[1,202] = 7.005, **p = 0.0088) on total immobility. (D) Main effect of maternal P7C3-A20 treatment (F[1,202] = 3.977, *p = 0.0475) on latency to immobility.

Importantly, offspring locomotor activity was not changed by stress or P7C3-A20 in female offspring (Supplementary Fig. S8B). While activity level in males was not altered by stress or P7C3-A20 alone, a significant interaction effect demonstrated that only P7C3-A20 treated males differed by prenatal stress exposure (Supplementary Fig. S8A). Center time in the open-field test (OFT) was not changed in offspring of either sex (Supplementary Fig. S8C, D).

Prenatal stress elicits depression-like activity in adult offspring, which is rescued by concurrent maternal P7C3-A20 treatment

We also evaluated adult offspring in the tail suspension test (TS) model of depression-like behavior. Across both sexes, prenatally stressed adult offspring showed increased immobility on TS (main effect of prenatal stress via two-way ANOVA) (F_stress[1,202] = 5.466, p = 0.0204, Fig. 9C). Maternal P7C3-A20 resulted overall in decreased immobility, such that the prenatally stressed offspring that were exposed to P7C3-A20 were similar to controls (main effect of maternal P7C3-A20 treatment) (F_P7C3-A20[1,202] = 7.005, p = 0.0088, Fig. 9C). P7C3-A20 generally increased latency to immobility, which was not affected by prenatal stress (main effect of maternal P7C3-A20 treatment) (F_P7C3-A20[1,202] = 3.977, p = 0.0475, Fig. 9D).

Prenatal stress elicits elevated axonal degeneration in adult offspring, which is partially rescued by concurrent maternal P7C3-A20 treatment

In adult injury models, we have previously shown that axonal degeneration can be quantified by measuring the magnitude of silver staining by automated optical densitometry, and that degeneration in the hippocampus is associated with impaired cognition (67, 68, 74, 75). We applied this technique to assess whether this might occur in adult offspring as a function of prenatal stress.

Interestingly, there was a significant increase in axonal degeneration in the adult whole brain (Fig. 10A) (F_stress[1,40] = 4.277, p = 0.0451). Broken down into subregions, we found a significant increase in axonal degeneration in hippocampal CA1, which was ameliorated by an independent effect of maternal P7C3-A20 (Fig. 10B) (F_stress[1,39] = 4.874, p = 0.0332; F_P7C3-A20[1,39] = 7.302, p = 0.0101). In adult corpus callosum, there was strong increase in axonal degeneration as a function of prenatal stress, but there was not a significant effect of concurrent maternal P7C3-A20 (Fig. 10C) (F_stress[1,40] = 30.78, p < 0.0001; post hoc Tukey's multiple comparisons test, Ctrl Veh vs. PS Veh p = 0.0266, Ctrl P7C3 vs. PS P7C3 p < 0.0001). No differences in axonal degeneration were noted in cerebral cortex, hippocampal CA3, or dentate gyrus (Fig. 10D–F).

FIG. 10.

Prenatal stress causes axon degeneration in adult offspringwhitematter. Silver staining showed that prenatal stress caused an increase in axonal degeneration in whole brain of adult offspring (A) (main effect of stress, F[1,40] = 4.277, *p = 0.0451). In CA1, prenatal stress caused an increase in axonal degeneration, and maternal P7C3-A20 administration caused a decrease in axonal degeneration independently of one another (B) (main effect of stress, F[1,39] = 4.874, *p = 0.0332; main effect of P7C3-A20, F[1,39] = 7.302, *p = 0.0101). In corpus callosum, prenatal stress caused an increase in axonal degeneration that was not rescued by P7C3-A20 (C) (main effect of stress, F[1,40] = 30.78, ****p < 0.0001; Tukey's post hoc multiple comparisons test, Ctrl Veh vs. PS Veh ^&p = 0.0266, Ctrl P7C3 vs. PS P7C3 ^&&&&p < 0.0001). There were no changes to axonal degeneration observed in cerebral cortex, CA3 hippocampus, or the dentate gyrus of the hippocampus (D–F). Color images are available online.

Discussion

While others have reported prenatal stress-induced memory impairments and depression-like behavior in rodent offspring of both sexes (32, 49, 53), we are the first to report an association of these effects with alterations in NAD⁺-synthesis machinery, as well as efficacy of a maternally administered neuroprotective pharmacologic intervention related to NAD⁺. We also report the novel finding of prenatal stress-induced changes in inhibitory neurons of offspring postnatal prefrontal cortex and hippocampus, and amelioration of these changes by maternal P7C3-A20. The common thread in these findings is the potential of harnessing NAD⁺ metabolism during brain development for preserving normal neuropsychiatric outcome in adulthood.

Here, we show that expression levels of Nmnat1, Nmnat2, and Nmnat3 are changed by prenatal stress and/or maternally administered P7C3-A20. In general, Nmnat1 is predominantly responsible for nuclear NAD⁺ production, Nmnat2 for cytoplasmic NAD⁺, and Nmnat3 for mitochondrial NAD⁺ (13). How this applies specifically to developing and differentiated neurons in the brain, however, is not as well characterized. The most well-understood brain function of the three is Nmnat2, which maintains axon integrity (28). The expression of Nmnat2 is regulated by CRE binding protein (CREB) phosphorylation and calcium/calmodulin-dependent protein kinase IV (CaMKIV), indicating overall Ca⁺⁺, and thus activity, dependence of Nmnat2 upregulation in adult mouse brain (2). However, how Nmnat enzymes are regulated during brain development is yet to be explored.

Here, we found that Nmnat2 expression at E18 is decreased by prenatal stress (Fig. 3B), and axon degeneration is coordinately increased in the hippocampal CA1 region and corpus callosum in adult offspring that had been subjected to prenatal stress (Fig. 10B, C). Although Nmnat2 expression was normalized in adult offspring subjected to prenatal stress (Supplementary Fig. S3), persistent axon degeneration at this stage could be related to developmental defects put into play by prenatal-stress-induced deficient embryologic Nmnat2 expression. We note that concurrent maternal administration of P7C3-A20 during prenatal stress restored normal expression of embryologic Nmnat2, but how this impacted the magnitude of adult axonal neurodegeneration is unclear, as axon integrity was restored in the CA1 but not the corpus callosum. Future work will focus on the role of embryologic expression of Nmnat2 in adult axonal integrity, as well as putative avenues for therapeutic treatment across the life span.

Unfortunately, little is known about the role of NAD⁺ metabolism during neurodevelopment. Recent evidence shows that prenatal maternal NAD⁺ treatment can prevent alterations to GABAergic neurodevelopment arising from forebrain angiogenic deficits (64), and postnatal maternal treatment with an NAD⁺ precursor, nicotinamide riboside, promotes offspring forebrain synaptic pruning and learning (24, 64). Disrupted NAD⁺ and NADH metabolism has also been implicated in the pathology of ASD, further suggesting a neurodevelopmental role (11). Furthermore, it is becoming clear that sirtuins, NAD⁺-dependent enzymes, play an indispensable role in developmental processes such as neuronal migration, dendritic arborization, and axonal elongation (31, 34). Thus, perturbations to NAD⁺ and its related processes appear to be a key mechanism by which offspring neurodevelopment and behavior can be disrupted and/or protected.

Prenatal stress has previously been shown to cause deficits in motor learning on the rotarod (48). Interestingly, here we found no deficit in motor learning (Supplementary Fig. S7) but did find a deficit in long-term motor memory (Fig. 9A). P7C3-A20 has previously been shown to rescue motor memory/retention on the rotarod in adult mice with motor degeneration phenotypes (65). Together with our data, this suggests that P7C3-A20 has a role in protecting the complex neural circuitry underlying motor memory.

Other types of memory deficits and depression-like behavior in prenatally stressed offspring have also been previously linked to mitochondrial dysfunction in the adult offspring hippocampus (26, 29). Given the important role of NAD⁺ metabolism in mitochondria, this converges with our findings that stabilization of NAD⁺ metabolism during development rescues offspring from adult memory deficits and depression-like behavior. However, our findings show that the expression of NAD⁺-producing enzymes in adulthood was not different after prenatal stress, suggesting that NAD⁺ dysregulation in the embryo, upstream of these postnatal changes, may be critical for later effects. Based on our findings, prenatal stress and pharmacologic stabilization of NAD⁺ have later effects on medial prefrontal cortical and hippocampal interneurons, as well as memory function.

For inhibitory neuron developmental assessments, male and female offspring had similar outcomes in embryonic forebrain, but distinct changes in adult brain. For example, Neto1, a gene involved in synaptic input to hippocampal interneurons (52), and the Pv/Gad1 ratio, both of which may reflect the maturation of interneurons, were downregulated in embryonic brain by prenatal stress and normalized by maternal P7C3-A20, regardless of sex. This suggests that neuronal inhibitory systems may be developmentally disrupted by prenatal stress and corrected by concurrent administration of maternal P7C3-A20 in similar ways across sex.

Interneurons continued to be disrupted in adult offspring, but this differed by sex. In female offspring, stress affected forebrain interneurons only at the protein level and only in mPFC, which was rescued by P7C3-A20. In male offspring, forebrain interneurons were disrupted at the gene expression across mPFC and hippocampus, particularly in an increased Pv/Gad1 gene expression ratio. These effects were rescued by P7C3-A20. Trend effects in males at the protein level were not rescued by P7C3-A20. Interestingly, aside from stress effects, in hippocampus we found an effect of P7C3-A20 on PV⁺ and GAD67GFP⁺ cell density only in males, suggesting a limited but potentially important effect of P7C3-A20 in males that warrants future investigation.

In general, our findings suggest that interneuron changes due to prenatal stress and resulting aberrations in NAD⁺-synthesis machinery diverge postnatally and show more susceptibility that can also be corrected in males. Some common changes across males and females in both interneuron development and behavior suggest that these are interdependent.

Interestingly, we also observed that prenatal stress elicited axon degeneration in hippocampal CA1 and the corpus callosum of adult offspring. This effect in hippocampal CA1 was normalized by P7C3, similar to other hippocampal findings in offspring. In contrast to the other deleterious outcomes of prenatal stress, corpus callosum effects were not rescued by maternal P7C3-A20. Corpus callosum is known to be especially sensitive to genetic and environmental stressors, and it was recently reported that prenatal exposure of mice to the concentrated ambient fine particulate matter model of air pollution yielded offspring with reduced corpus callosum size (22). Aberrant development of the corpus callosum is well known to be associated with a variety of neuropsychiatric impairments, including cognitive dysfunction, social impairment, obsessive or compulsive behavior, and psychosis.

While the underlying basis for impaired formation of the corpus callosum in our model of prenatal stress requires further investigation, it may be independent of the NAD⁺-related mechanisms that we have shown are linked to interneuron development and adult behavior. In future studies, it will be important to follow the offspring into later adulthood to determine whether new behavioral deficits arise as a function of axonal degeneration. For some of these such as found in hippocampal CA1, maternal P7C3-A20 may provide amelioration. Treatment during adulthood may also be of therapeutic benefit, as therapeutic efficacy for arresting corpus callosum axonal degeneration with this agent after both acute and chronic injuries has previously been demonstrated (67, 74).

We also acknowledge other limitations of our study. We did not measure NAD⁺ or its related enzymes before E18 when there may be dynamic changes, as is common in many components of development. Future work could reveal altered trajectories of NAD⁺ metabolism in embryonic brain acutely after the onset of prenatal stress and then subsequently during gestation. Another limitation was the use of oral gavage, which could result in animals experiencing mild stress. We minimized this with a well-trained and highly skilled animal handler performing these procedures, and this variable was controlled in the sense that all animals received the same procedure.

Our work presented here lays the foundation for several important next steps. First, the prenatal environment must be further explored to more deeply understand the biochemical and cellular mechanisms that are disrupted by prenatal stress, and how maternal P7C3-A20 administration impacts those mechanisms. For example, our laboratory has found that prenatal stress increases oxidative stress in the embryonic brain, and that oxidative stress disrupts the migration of GABAergic progenitors (9). An agent that increases NAD⁺ in times of cellular stress that drive increased energy demand, thereby buffering against oxidative stress, may have the ability to restore GABAergic migration to its normal trajectory in the context of prenatal stress.

However, although stress disrupted the expression of enzymes involved in NAD⁺ production in the embryonic brain, the net NAD⁺/NADH ratio at the completion of prenatal stress was unaffected. Furthermore, expression of antioxidant and oxidative stress-response genes was not altered by chronic stress in the E18 brain (Supplementary Fig. S4). Thus, further investigation is required to determine whether oxidative stress does indeed play a role in the acute phase of prenatal stress and is compensated chronically.

Future work should also focus on establishing the mechanistic role of NAD⁺ in the embryonic brain for determining impacts on inhibitory neurons and behavioral outcomes. A potential mechanism for identifying this may be the use of the slow Wallerian degeneration (WldS) mouse strain, which could isolate embryonic from maternal NAD⁺ metabolism. The WldS mutation causes a chimeric triplicate mutation of the E4 ubiquitination factor Ube4b and the Nmnat1 coding region (Ube4b/Nmnat). In brain tissue of animals expressing WldS, Nmnat1 enzymatic activity is increased fourfold, dramatically increasing the rate at which required NAD⁺ can be synthesized (44). We have previously used this mouse strain to demonstrate the critical role of NAD⁺ in preventing axon degeneration after injury to the adult brain (75).

Another important future direction will be investigation of the role of the placenta in NAD⁺-related changes caused by prenatal stress. The placenta is involved in many mechanisms that are also linked to the effect of maternal stress on the embryonic brain, including oxidative stress, inflammation, and monoaminergic signaling (57). One example related to known NAD⁺ function is the NAD⁺ dependence of placental 11β-hydroxysteroid dehydrogenase (11β-HSD2) activity, which buffers glucocorticoids by converting active cortisol to inactive cortisone (1, 7, 12, 63, 73). Thus, disturbances in NAD⁺ metabolism in the placenta may be directly relevant to how stress impacts the embryonic brain.

Finally, P7C3 compounds have previously been shown to rescue chronic stress-induced phenotypes in adult rodent models (69, 70). Future work could investigate the impact of P7C3-A20 on behaviors of adult pregnant females after experiencing chronic prenatal stress.

In sum, we found that maternal prenatal stress led to impaired NAD⁺-synthesis enzyme expression, aberrant GABAergic interneuron development, and adult neuropsychiatric impairment in the form of cognitive dysfunction and depression-like behavior. Importantly, concurrent maternal administration of the NAD⁺-stabilizing agent P7C3-A20 during prenatal stress rescued these effects. Although much work remains to progress our findings to treatment in patients, including the challenging determination of a maternal peripheral biomarker of activity that reflects therapeutic levels of P7C3 compounds in the embryonic brain, our results provide proof of principle in a preclinical model for the efficacy of new neuroprotective approach to preserving normal embryonic brain development and adult neuropsychiatric functioning after prenatal stress.

Methods

Mice

All animals were housed in accordance with the University of Iowa Institutional Animal Care and Use Committee (IACUC) policies. All mice were housed in cages on a 12 h light/dark cycle with free access to food and water. For all experiments, GAD67-GFP^+/− knock-in mice were used and bred on a CD1 background for the purpose of measuring GAD67⁺ cell density via unbiased stereology. Breeding-naïve GAD67-GFP^−/− female mice were bred with GAD67-GFP^+/− males. For timed pregnancies, observation of a vaginal plug established embryonic day 0 (E0).

Prenatally stressed dams were singly housed from E0 onward, and nonstressed dams were cohoused with at least one other cagemate. When adult offspring were used, litters were culled to approximately four males and four females to control for differences in litter size and composition. Offspring were weaned at postnatal day 21 (P21) and housed in cages of three to five littermates of the same sex.

Prenatal stress and P7C3-A20 administration

Four conditions were used: (1) nonstressed control with vehicle (Ctrl Veh), (2) nonstressed control with P7C3-A20 (Ctrl P7C3), (3) prenatally stressed with vehicle (PS Veh), and (4) prenatally stressed with P7C3-A20 (PS P7C3). Animal conditions were pseudorandomly assigned, and there were six to eight successful pregnancies per group.

Stress procedure

Beginning on E5 and continuing through E18, dams were given restraint stress for 45-min sessions three times daily (∼0900, 1200, and 1500) in accordance with previous work (10, 30, 43). Our group has found restraint stress to be a reliable method of inducing neurodevelopmental changes in offspring without causing injury to the gestating dam or offspring. To reduce circadian habituation to stress, the exact time stress sessions took place was varied day to day. At least 2 h of time undisturbed in the home cage was given between stress sessions. During stress sessions, dams were restrained in a clear Plexiglas tube and placed under a 60-watt equivalent CFL bulb.

P7C3-A20 administration

P7C3-A20 or equivalent dose of vehicle (dimethyl sulfoxide [2.5% v/v], Kolliphor [10% v/v], dextrose [5% w/v, 87.5% v/v], pH 7.0–7.2) was administered via oral gavage at a dose of 10 mg/kg twice daily. Formulation was as previously described (56). If a stress session followed a dose of P7C3-A20/vehicle, we waited ∼30 min between administration and the start of stress.

P7C3-A20 measurement

After E5-E18 P7C3-A20 administration, dams were deeply anesthetized with ketamine/xylazine cocktail and euthanized via decapitation. Trunk blood was collected from the dams, and anticoagulant and plasma were isolated by centrifugation for 10 min at 9600 g. Brains were harvested from both dams and fetuses, washed gently with phosphate buffered saline (PBS) to remove any surface-adhering blood, weighed, and snap frozen. Samples were stored at −80°C until analysis. Brain tissue was added to BeadBug™ tubes containing 3 mm Zirconium beads and homogenized in a 3 × volume of PBS (w/v) using a BeadBug microtube homogenizer set at 280 RPM for 2 min. Standards were prepared by spiking blank mouse plasma or brain homogenate with P7C3-A20.

Compound was extracted from plasma or brain homogenate samples or standards by passage over a Phenomenex (Torrence, CA) Phree Phospholipid Removal 1 mL tube to remove phospholipids as well as proteins. In brief, plasma or brain homogenate was loaded onto the column, and a volume of acetonitrile containing 0.13% formic acid and 6.7 ng/mL n-benzylbenzamide internal standard (IS) equal to eightfold the volume of applied plasma was added. The mixture was pipetted gently twice to mix. A vacuum was applied and the flow-through collected. An additional 8 × volume of acetonitrile containing formic acid and IS was applied to wash the column, and the flow-through added to the original material.

Compound levels were monitored by liquid chromatography with tandem mass spectrometry with a mass spectrometer in multiple reaction monitoring mode by following the precursor to fragment ion transition 507.0 to 204.1 (positive mode; M + H⁺). Reverse-phase chromatography with a C18 column was utilized under the following gradient conditions: Buffer A, 0.1% formic acid in water; Buffer B, 0.1% formic acid in methanol, 0–1 at 0 min 5% B, 1.0 to 1.5 min gradient to 100% B, 1.5 to 3.0 min 100% B, 3.0 to 3.15 min gradient to 5% B, 3.15 to 4.5 min 5% B. A value of threefold above the signal obtained in the blank plasma was designated the limit of detection (LOD). The limit of quantitation was defined as the lowest concentration at which back calculation yielded a concentration within 20% of the theoretical value and above the LOD signal.

The concentration of P7C3-A20 in the dam brains was adjusted to remove compound present in the vasculature of the brain (0.03 mL blood/g tissue) (36). This adjustment was not possible for fetal brains as blood concentrations were not determined for the fetuses, but as it represented only a modest 6–7% reduction in adult brain concentrations, it is not anticipated to significantly impact fetal brain concentrations.

Behavior assays

Behavior assays were performed on adult offspring starting at 8 weeks of age and completed by 15 weeks of age. Adult offspring performed all tests during the light cycle with ∼1 day of rest between different tests in the order listed below. Mice were habituated to the testing room each day for ∼30 min.

Open-field test

Mice were placed at the center of the OFT apparatus (40 × 39 × 30 cm) and allowed to roam freely for 30 min. Mice were monitored via an overhead camera. After testing, videos were analyzed via AnyMaze software for measurement of variables, including time in center, time on perimeter, and distance traveled. Two OFT sessions were conducted on consecutive days (OFT1 and OFT2). Locomotor activity was assessed as the total distance traveled (m) in the full 30 min of OFT2 to avoid novelty effects on OFT1.

Tail suspension test

A piece of tape ∼5 cm long was attached to the mouse's tail and adhered to a free-standing horizontal bar elevated 45 cm from the table surface. Mice were allowed to hang suspended by the tail for 6 min while being monitored via a camera mounted onto the table. Upon completion, mice were gently lifted and detached from the apparatus. The tape was carefully removed from the tail upon return to the home cage. Videos were hand-scored for time immobile over the entire 6 min as well as latency to immobility. More time spent immobile is indicative of a “depressive-like” phenotype.

Rotarod

To evaluate procedural memory, a 2-day rotarod training paradigm with a single-day probe was used. Training took place on two consecutive days, and the probe took place 2 weeks after the second training day. Mice were tested for 5 trials/day, with ∼30 min of rest between trials.

Mice were placed on a horizontal, rotating cylinder (rod) suspended 28 cm above a metal plate, below which a sensor recorded each mouse's latency to fall. The speed of the rotarod began at 4 rotations per min (RPM) and accelerated to 80 RPM over the course of 4 min. Mice that had not fallen by 5 min had a latency to fall time of 5 min recorded. Initial learning was calculated as the change in latency to fall from the beginning of training to peak learning at the end of the two training days (average of first two training trials divided by average of last two training trials), and learning retention was calculated as the change in latency to fall from peak learning to the probe presented 2 weeks later (average of last two training trials divided by average of five probe trials).

Fear conditioning

Contextual FC was performed using a 2-day paradigm. On day 1, mice were placed into individual arenas with a grated floor, and allowed to explore the arena for 2 min and 28 s. After 2 min and 28 s elapsed, mice received a 2 s 1.5 mA shock. After the shock ended, mice remained in their arenas for an additional 30 s. After a total of 3 min, the trial was ended, and mice were removed and returned to their home cages. On day 2, 24 h after the day 1 training, mice were placed into the same arenas in which they were trained and allowed to explore for 5 min. After 5 min had elapsed, the mice were removed and returned to their home cages. FreezeScan software was used to monitor and measure freezing behavior.

Immunohistochemistry

GAD67-GFP^+/− offspring were identified using polymerase chain reaction (PCR) genotyping, and were used for immunohistochemistry and fluorescence microscopy. Adult offspring were euthanized at 17 weeks of age, 2 weeks after conclusion of behavior experiments to avoid effects of acute stress from behavior assays. Animals were deeply anesthetized with ketamine-xylazine cocktail. They were then transcardially perfused first with 1 × PBS followed by 4% paraformaldehyde. Tissue was then postfixed in 4% paraformaldehyde for 24 h and transferred to 20% sucrose for ∼16 h. The tissue was then frozen and embedded in optimal cutting temperature for cryosectioning. Tissue was sectioned coronally at 50 μm, and then incubated in 10% goat serum/PBS⁺⁺ blocking solution (PBS with 0.025% Triton X-100, 0.0125% Tween 20) at room temperature for ∼1 h.

For PV subtype analysis, sections were incubated with 5% neutral horse serum/PBS⁺⁺ and primary antibodies, anti-PV (1:1000, SAB4200545; Sigma Aldrich, St. Louis, MO) and anti-GFP (1:1000, AB13970; Abcam). Sections were then incubated with Alexa dye-conjugated secondary antibodies (1:500–1000; Molecular Probes) in 5% goat serum/PBS⁺⁺. Tissue was slide mounted with DAPI-containing mounting medium for nuclear staining (#H-1200; Vector Laboratories).

Fluorescence microscopy and stereology

Stereological counting in adult offspring mPFC and hippocampus was performed as previously described on every 10th brain serial section (43). For hippocampal counts, CA1 and CA3 were analyzed separately and assessed for regional differences. When none were found, data were pooled between regions for each individual. In brief, a Zeiss Axiocam, equipped with motorized stage and digital camera, coupled to a computer with StereoInvestigator software (Microbrightfield, Colchester, Vermont), was used to count GFP-GAD67⁺-expressing cells and PV⁺-expressing cells. A grid size of 300 × 300 and a 150 × 150 × 25 μm counting frame was used for all regions. Between two to three sections of mPFC and five to seven sections of hippocampus were counted for each animal.

Cell densities were calculated as the total estimated counts divided by the total calculated volume, and PV⁺/GAD67⁺ ratio was the number of cells expressing both PV and GAD67 divided by the total population of GAD67⁺ cells.

Silver staining

Silver staining was performed in free-floating sections using the FD NeuroSilver Kit following the manufacturer's protocol (PK301; FD NeuroTechnologies, Columbia, MD). Images were acquired using the Zeiss Axio Scan.Z1, keeping the light intensity and exposure time constant. ImageJ version 1.42 software was used to analyze silver staining (black staining), and quantification was performed using the plugin of the color deconvolution method described by Ruifrok and Johnston (60).

Gene expression

For E18 brain, dams were deeply anesthetized with ketamine-xylazine cocktail, euthanized, and embryos were collected. For quantitative polymerase chain reaction gene expression assays, embryonic forebrain was dissected out and flash frozen in dry ice. For adult offspring, animals were deeply anesthetized with ketamine-xylazine cocktail before euthanizing. Animals were then decapitated, and the brain was quickly removed. The tissue was sectioned at 1 mm thickness using a brain slicer matrix, transferred to slides, and flash frozen in dry ice. A tissue 2 mm diameter punch was used to remove regions of interest from frozen sections stored on slides.

For all tissue samples, mRNA isolation was performed using the RNeasy Plus Mini Kit (Quiagen, Valencia, CA) per the manufacturer's protocol. Total RNA concentrations were measured using a spectrophotometer (Nanodrop; Thermo Scientific). Gene expression was measured in triplicate (ViiA 7 Real-Time PCR System; Thermofisher), and relative expression values were calculated from average cycle threshold (CT) values normalized to GAPDH (2^−ΔCT) and to control group average value. Primers and sequences are found in Supplementary Table S1.

NAD⁺ measurement

NAD⁺ was measured in offspring in E18 forebrain and adult frontal cortex. The BioVision NAD⁺/NADH Quantitation Colorimetric Kit (San Francisco, CA) was used per the manufacturer's protocol to measure total NAD⁺/NADH (NADt) and NADH for each sample. Absorbance was measured at 450nm using an accuSkan GO spectrophotometer (Thermo Fisher Scientific), and NADt and NADH concentrations were derived from standard curve values. To calculate NAD⁺, NADH was subtracted from NADt.

Data analysis and sampling

All data were first checked via three-way ANOVA for a main effect of sex. If sex differences were not found, the data were pooled across sexes. If sex differences were present, males and females were analyzed independently.

A priori two-tailed t-tests were also performed to compare PS Veh and Ctrl Veh groups. Two-way ANOVA was used to compare Ctrl Veh, PS Veh, Ctrl P7C3, and PS P7C3 groups, with post hoc pairwise comparisons when appropriate. Tukey's multiple comparisons test was used to test post hoc pairwise comparisons and correct for multiple comparisons. Relevant results are presented in the text, and all results are presented in Supplementary Table 2. For each measure, rescues were defined as the lack of a significant difference between the Ctrl Veh group and the PS P7C3 group when a significant effect of stress was present and/or by significant interactions of stress and P7C3. We regarded p < 0.05 as significant for t-tests, ANOVAs, and post hoc pairwise comparisons.

Abbreviations Used

Ackr3

atypical chemokine receptor 3

ANOVA

analysis of variance

ASD

autism spectrum disorder

Bdnf

brain-derived neurotrophic factor

Cat

catalase

CT

cycle threshold

Cxcr4

C-X-C chemokine receptor type 4

E0

embryonic day 0

E5

embryonic day 5

E18

embryonic day 18

FC

fear conditioning

GABA

gamma-aminobutyric acid

Gabra5

gamma-aminobutyric acid type A receptor subunit alpha5

Gabrg2

GABA type A receptor subunit gamma2

Gad1

glutamate decarboxylase 1

Gad2

glutamate decarboxylase 2

Gat1

GABA transporter 1

Gpx1

glutathione peroxidase 1

Gria1

glutamate ionotropic receptor AMPA type subunit 1

IS

internal standard

Kcc2

potassium chloride cotransporter 2

LOD

limit of detection

mPFC

medial prefrontal cortex

n.s.

not significant

NAD⁺

nicotinamide adenine dinucleotide

Nampt

nicotinamide phosphoribosyltransferase

Neto1

neuropilin and tolloid like 1

Nkcc1

basolateral Na-K-Cl symporter

Nmnat1

nicotinamide nucleotide adenylyltransferase 1

Nmnat2

nicotinamide nucleotide adenylyltransferase 2

Nmnat3

nicotinamide nucleotide adenylyltransferase 3

Npy

neuropeptide Y

Nrf2

 nuclear factor erythroid 2-related factor 2

Nxph1

neurexophilin 1

OFT

open-field test

PBS

phosphate buffered saline

PCR

polymerase chain reaction

PV

parvalbumin

ROS

reactive oxygen species

RPM

rotations per min

Sesn1

sestrin 1

Sesn2

sestrin 2

Sesn3

sestrin 3

Slc32a1

vesicular inhibitory amino acid transporter

Som

somatostatin

Tr1

tropinone reductase 1

TS

tail suspension test

WldS

Wallerian degeneration

Authors' Contributions

R.S., A.A.P., and H.E.S. designed experiments, interpreted data, and wrote the article. P.S., L.N., A.L., N.S.W., C.J.C.-P., and E.V.-R. conducted experiments and data analysis. All authors reviewed and approved the article.

Author Disclosure Statement

R.S., P.S., L.N., A.L., N.S.W., K.P.K., C.J.C.-P., and E.V.-R. have no conflicts to disclose. A.A.P. is an inventor on patents related to P7C3. H.E.S. is an inventor on a patent related to NAD⁺ precursors.

Funding Information

A.A.P. was supported by a grant from the Brockman Foundation. A.A.P. was also supported by the Elizabeth Ring Mather & William Gwinn Mather Fund, S. Livingston Samuel Mather Trust, G.R. Lincoln Family Foundation, Wick Foundation, Gordon & Evie Safran, the Leonard Krieger Fund of the Cleveland Foundation, the Maxine and Lester Stoller Parkinson's Research Fund, the Louis Stokes VA Medical Center resources and facilities, and Project 19PABH134580006-AHA/Allen Initiative in Brain Health and Cognitive Impairment.

H.E.S. and R.S. were supported by a Junior Research Program of Excellence awarded to H.E.S. from the Roy J. Carver Charitable Trust and research grants from the Nellie Ball Trust to H.E.S. H.E.S. was also supported by NIH grant R01 MH122485-01 and by a Career Development Award from the University of Iowa Environmental Health Science Research Center (P30 ES005605). R.S. was supported by the University of Iowa Graduate Post-Comprehensive Research Fellowship and the Ballard-Seashore Dissertation Fellowship.

Supplementary Material

Supplementary Table S1

Supplementary Table S2

Supplementary Figure S1

Supplementary Figure S2

Supplementary Figure S3

Supplementary Figure S4

Supplementary Figure S5

Supplementary Figure S6

Supplementary Figure S7

Supplementary Figure S8