Effects of Developmental Nicotine Exposure on Frontal Cortical GABA-to-Non-GABA Neuron Ratio and Novelty-Seeking Behavior

Abstract

Cigarette smoking during pregnancy is a major public health concern, resulting in detrimental health effects in the mother and her offspring. The adverse behavioral consequences for children include increased risk for attention deficit hyperactivity disorder, working memory deficits, epilepsy, novelty-seeking, and risk-taking behaviors. Some of these behavioral conditions are consistent with an imbalance in frontal cortical excitatory (glutamate) and inhibitory (GABA) neurotransmitter signaling. We used a GAD67-GFP knock-in mouse model to examine if developmental nicotine exposure alters frontal cortical GABA neuron numbers, GABA-to-non-GABA neuron ratio and behavioral phenotypes. Female mice were exposed to nicotine (100 or 200 μg/mL) in drinking water beginning 3 weeks prior to breeding and until 3 weeks postpartum. Male and female offspring were examined beginning at 60 days of age. The nicotine exposure produced dose-dependent decreases in GABA-to-non-GABA neuron ratios in the prefrontal and medial prefrontal cortices without perturbing the intrinsic differences in cortical thickness and laminar distribution of GABA or non-GABA neurons between these regions. A significant increase in exploratory behavior and a shift toward “approach” in the approach–avoidance paradigm were also observed. Thus, developmental nicotine exposure shifts the cortical excitation–inhibition balance toward excitation and produces behavioral changes consistent with novelty-seeking behavior.

Introduction

The prevalence of nicotine use during pregnancy continues to be a major public health concern worldwide. For instance, in the United States of America, 7.2% of pregnant women smoke cigarettes, and the prevalence increases to almost 16% during the initial postpartum period, presumably while many mothers are nursing (Tong et al. 2013; Martin et al. 2018). In addition, the rising popularity of e-cigarettes, which may deliver even higher levels of nicotine than traditional cigarettes (CDC-DHHS 2016; Mirbolouk et al. 2018), is a relatively recent and growing concern. Nicotine use during pregnancy increases the risk for adverse health effects for the child. For example, developmental nicotine exposure increases the risk for attention deficit hyperactivity disorder (ADHD), conduct disorder, aggression, developmental delays, working memory deficits, learning disabilities, depression, epilepsy, as well as novelty-seeking and risk-taking behavior for children and adolescents [for review (Wickstrom 2007; Pagani 2014)]. Preclinical studies are consistent with many of these clinical observations and show that developmental nicotine exposure produces hyperactivity, attention deficit, working memory deficit, impaired emotional and social behaviors, and impaired prepulse inhibition [for review (Heath and Picciotto 2009; Pagani 2014; Zhang et al. 2018)].

The behavioral impairments produced by the developmental nicotine exposure may have a basis in long-term changes in multiple neurotransmitter signaling mechanisms [for review (Dwyer et al. 2009; Heath and Picciotto 2009; Pagani 2014)]. Preclinical studies have focused on the effects of developmental nicotine exposure on nicotinic acetylcholine receptor (nAChR) signaling mechanisms (Slotkin et al. 1987, 2002, 2007; Nordberg et al. 1991; Miao et al. 1998; Buck et al. 2019), as the nAChR appears early in brain development (Zoli et al. 1995; Broide et al. 1996; Conroy and Berg 1998; Tribollet et al. 2004). Moreover, the nAChR is expressed by proliferating progenitor cells and newborn neurons in the embryonic and early postnatal brain (Atluri et al. 2001; Schneider et al. 2002) suggesting that developmental nicotine exposure has the potential to influence multiple developmental processes including neurogenesis, neuronal migration, survival, and maturation.

Although preclinical models of developmental nicotine exposure support the clinical findings, the increased novelty-seeking behavior in children and adolescents exposed to nicotine in the prenatal period [for review (Wickstrom 2007; Pagani 2014)] has not been fully supported by the preclinical data. Similarly, a neurotransmitter basis for these behavioral phenotypes has not been demonstrated. In the present study, we examined the balance between inhibition and excitation in the frontal cortex by analyzing the ratio between GABA (inhibitory) and non-GABA (presumptive excitatory, glutamatergic) neurons in a mouse model of developmental nicotine exposure. We then correlated the data with behaviors consistent with novelty-seeking behavior. For this, we used a GAD67-GFP knock-in mouse model in which GABA neurons express the green fluorescent protein (GFP) (Tamamaki et al. 2003) and therefore can be distinguished from the non-GABA neurons of the frontal cortex based on GFP fluorescence. We analyzed two behaviors that correlate with novelty-seeking in humans, namely, approach–avoidance behavior in an elevated plus maze (EPM) and exploratory behavior in a novel environment (Lynn and Brown 2009; Wingo et al. 2016). Other behavioral parameters analyzed included spontaneous locomotor activity and working memory, which are known to be influenced by developmental nicotine exposure (Zhu et al. 2012, 2017; Zhang et al. 2018). In addition, we analyzed cortical thickness and laminar distribution of GABA and non-GABA neurons in two regions of the frontal cortex to determine if the core cytoarchitectonic differences between frontal cortical regions were altered.

We found a significant deficit in cortical GABA-to-non-GABA neuron ratio, a shift in approach–avoidance balance in favor of the approach behavior and an increase in exploratory behavior in a novel environment. These findings suggest that developmental nicotine exposure alters the frontal cortical inhibition–excitation balance in favor of excitation and produces behavioral phenotypes consistent with novelty-seeking behaviors that are reported in children and adolescents prenatally exposed to nicotine. These behavioral changes occurred without significant alterations in the intrinsic differences between frontal cortical areas in the thickness or the laminar distribution of GABA and non-GABA neurons.

Materials and Methods

Animals

Swiss Webster (SW; Charles River Laboratories) and GAD67-GFP knock-in mice (GAD67-GFP) (Tamamaki et al. 2003) in the SW background (Brown et al. 2008; Chen et al. 2010; McCarthy and Bhide 2012) were used. The mice were housed in a temperature- and humidity-controlled environment on a 12 h light:dark cycle with an ad libitum supply of food and water. All experimental procedures were in full compliance with our institutional and NIH guidelines for the care and use of laboratory animals. Female wild-type SW mice were bred with hemizygous GAD67-GFP males to produce wild type and hemizygous GAD67-GFP offspring. This mating paradigm does not produce homozygous GAD67-GFP offspring, which are reported to show abnormal phenotypes (Tamamaki et al. 2003). The day of birth was designated postnatal day 0 (P0), and all litters were standardized to contain 8–12 offspring. Each offspring was examined under GFsP-5 goggles (Biological Laboratory Equipment Maintenance and Service Ltd) for the presence of GFP fluorescence in the brain and spinal cord within 2 days of birth when the GFP fluorescence is visible through the intact skin. GAD67-GFP offspring (i.e., GFP⁺ offspring) were tattooed on the back right paw, whereas wild-type mice were tattooed on the back left paw for permanent identification; all offspring were marked to control for the tattoo procedure. GFP⁺ and GFP⁻ offspring from a given litter were housed together in the same cage with their mother until weaning. Offspring were weaned around P21, and males and females were housed separately in groups of 2–4 per cage until the end of the study. GFP⁺ and GFP⁻ mice of the same sex were housed together throughout the study.

Developmental Nicotine Exposure

Female wild-type SW mice were placed on plain drinking water or drinking water containing either 100 or 200 μg/mL nicotine (Sigma Chemical Co., Catalog #: N3876) beginning 3 weeks prior to mating. The drinking water treatment continued during mating, throughout the prenatal period and during the 3-week postnatal period (Fig. 1A). Nicotine exposure of the dams began 3 weeks before mating for consistency with our previous studies (Zhu et al. 2012, 2014, 2017; Zhang et al. 2018). This nicotine exposure paradigm facilitates development of nicotine tolerance by the dams by the time of mating and ensures elevated serum nicotine levels at conception (Pauly et al. 2004). The treatment paradigm produced offspring with the following three types of developmental exposures: plain drinking water (WATER), 100 μg/mL nicotine (Nic100), and 200 μg/mL nicotine (Nic200). All offspring were placed on plain drinking water at the time of weaning.

Figure 1.

Developmental nicotine exposure mouse model. Wild-type SW female mice were placed on plain drinking water (WATER) or drinking water containing either 100 or 200 μg/mL nicotine (Nic100 and Nic200, respectively). Following 3 weeks of exposure, the females were mated with drug-naive GAD67-GFP males. The drinking water treatment of the females continued throughout pregnancy and during the 3-week postpartum period until the offspring were weaned on approximately P21. Thus, the offspring in the Nic100 and Nic200 groups were exposed to nicotine during the prenatal and early postnatal periods. Developmental milestones of the offspring from each treatment group were recorded during the postnatal period. At the time of weaning, offspring from all three groups were placed on plain drinking water. From P60–P90 (adult), behavioral analyses were performed and brains were collected for histological analyses (A). GFP⁺ male and female offspring from all three treatment groups (WATER, Nic100, and Nic200) were used for histological analyses. For each experimental group, we used 6–7 male and female mice derived from 3–5 litters (B). Cell counts were performed in rostral and caudal sections of the mPFC and PFC of which were based on anatomical landmarks (C). Cresyl violet-stained representative sections from a non-nicotine-exposed brain show cell packing density, which was used as the criterion for identification of cortical layers within the mPFC and PFC (D). GFP⁺ and GFP⁻ male and female offspring from WATER and Nic200 treatment groups were used for behavioral analyses. For these analyses, 8–19 male and female mice from 6–11 litters were used in each group (E).

Litter Metrics and Developmental Milestones

Breeding success, gestational length, litter size, and litter weight were recorded on P0. Offspring body weights were recorded also on P4, P7, P14, P21, and P80. The postnatal day of external ear detachment, appearance of fur, and eye opening were also recorded.

Plasma Cotinine Content

Trunk blood was collected from dams and P21 offspring from the WATER, Nic100, and Nic200 groups in anti-coagulant 2,2′,2′′,2′′′-(Ethane-1,2-diyldinitrilo)tetraacetic acid (EDTA)-coated tubes. The samples were centrifuged for 20 minutes at 4°C at 2000 rpm. The plasma was collected and stored at −80°C. Cotinine content of the plasma was assayed by enzyme-linked immunosorbent assay (ELISA) (Immunalysis, Catalog #: 217-0096).

Tissue Processing and NeuN Immunohistochemistry

On P60, GFP⁺ male and female offspring from the three treatment groups (WATER, Nic100, and Nic200) (Fig. 1B) were anesthetized with ketamine (125 mg/kg i.p.) and xylazine (25 mg/kg i.p.), and perfused via the heart with 4% formaldehyde in 0.1 M phosphate buffer. The brains were dissected from the skull and placed in the same fixative solution for 24 h at 4°C. Next, brains were cryoprotected in 15% and 30% sucrose solutions, frozen on dry ice, and stored at −80°C. Brains were then sectioned in the coronal plane at 50 μm thickness, and the sections were collected and placed serially into a 24-well plate containing a cryoprotectant solution (30% sucrose, 1% w/v polyvinylpyrrolidone, 30% ethylene glycol in 0.1 M phosphate buffer) before being stored at −20°C. For each brain, one column of the 24-well plate was randomly chosen. From each column, two sections [one rostral and one caudal, 600 μm apart, corresponding to stereotaxic coordinates bregma = +1.34 mm and bregma +0.74 mm; Figures 20 and 25, respectively, from Paxinos and Franklin (2001)] were selected for immunohistochemical processing (Fig. 1C). These sections were chosen because they contained the prefrontal and medial prefrontal cortices, which represent the regions of interest for this study. Sections were washed with 0.01 M phosphate-buffered saline (PBS) incubated with 1% bovine serum albumin (BSA), 10% normal donkey serum, and 0.3% Triton X-100 in PBS for 2 h at room temperature to reduce nonspecific antibody binding. Sections were incubated with anti-NeuN (1:200; mouse monoclonal, EMD Millipore MAB377) or anti-GABA (1:100; rabbit polyclonal, Millipore Sigma, ABN131) primary antibody in 1% BSA, 1% normal donkey serum, and 0.3% Triton X-100 PBS solution at 4°C overnight. The next day, sections were washed with PBS and then incubated in either Cy3 AffiniPure Donkey Anti-Mouse IgG secondary antibody for NeuN labeling (1:200; Jackson ImmunoResearch, Catalog #: 715-165-150) or Cy3 AffiniPure Donkey Anti-Rabbit IgG secondary antibody for GABA labeling (1:200; Jackson ImmunoResearch, Catalog #: 711-165-152) in 1% BSA, 1% normal donkey serum, and 0.3% Triton X-100 PBS solution for 1 h at room temperature. Sections were washed and mounted on Superfrost Plus slides with VECTASHIELD mounting medium.

Analysis of GFP⁺/NeuN⁺ (GABA) and GFP⁻/NeuN⁺ (Non-GABA) Cells in the Prefrontal Cortex and Medial Prefrontal Cortex

Following NeuN immunohistochemistry, sections were examined using an Olympus FluoView FV1000 laser confocal microscope. Images encompassing the full radial extent of the prefrontal cortex (PFC) and medial prefrontal cortex (mPFC) (i.e., from the white matter to the pial surface) were collected using a 10× objective at 2.5 μm thickness. Uniform immunohistochemical labeling for NeuN throughout the section thickness was confirmed by examination of the labeling throughout the z-axis of the section. Images of the PFC and mPFC were then cropped to a 100-μm-wide area. The radial dimension (determined by the distance between the white matter of the corpus callosum and the dorsal boundary of layer 2 beneath the pial surface) was variable from section to section due to the intrinsic variation among sections. A “bin” method of analysis (McCarthy and Bhide 2012) was used to calculate the number of cells within a counting grid that was superimposed on the digital images of the sections (Fig. 3A,B). Briefly, a counting grid consisting of 10 equally spaced bins (numbered from 1 to 10) was superimposed digitally on the images of the histological sections using Photoshop software such that bin #1 was placed at the dorsal edge of layer 2 and bin #10 was placed at the dorsal border of the white matter of the corpus callosum. Each “bin” was 100 μm in width and represented 10% of the total radial thickness of the section. GFP⁻NeuN⁺ and GFP⁺/NeuN⁺ cells were manually counted in each bin, and the counts were expressed as number of cells per square micrometer (Fig. 3A,B). The cortical layers were identified based on cell packing density, and layer boundaries were superimposed on the image of the section so that each bin could be assigned to a layer (Fig. 1D).

Figure 3.

Numerical density of GFP⁺ NeuN⁺ (GABA neurons) and GFP⁻/NeuN⁺ (non-GABA) cells. Representative images of histological sections in the coronal plane from postnatal day 60 GAD67-GFP mice processed for NeuN immunohistochemistry and viewed under a confocal microscope to show the PFC (A) and the mPFC (B). Presumptive GABA neurons are GFP⁺ and NeuN⁺ (yellow), and non-GABA neurons are GFP⁻ and NeuN⁺ (red). A digital image of a counting grid (divided into 10 equally spaced “bins”) was superimposed onto the image of the histological section (A, B). GFP⁺ NeuN⁺ and GFP⁻/NeuN⁺ cells were counted in each bin. The layer boundaries were established based on cell packing density and each “bin” was assigned to the appropriate layer (A, B). A higher magnification view of the image illustrating the resolution of GFP⁺NeuN⁺ cells (GABA neurons; yellow) and GFP⁻NeuN⁺ cells (non-GABA; red) (C). Overlap between GFP fluorescence (D) and GABA immunohistochemical labeling (E) confirms the fidelity of GFP as a marker of GABA neurons (D–F). The numerical densities of GABA (G) and non-GABA (H) neurons were quantified, and the GABA-to-non-GABA ratio was calculated (I) in offspring from dams exposed to plain drinking water (WATER), 100 μg/mL nicotine (Nic100), and 200 μg/mL nicotine (Nic200). For each group, the data are shown separately for cortical layers II–III, V, and VI. ^*P < 0.05; ^**P < 0.01.

Behavioral Analyses

The behavioral analyses were also performed on adult mice beginning at approximately P90. The mice were handled by the experimenter for 2–3 min/day for 3 days prior to the beginning of the analyses. Mice were habituated to the testing room for 30 minutes before the analyses commenced. The handling, habituation, and behavioral testing occurred during the lights-off period, under red lights for ambient illumination (with the exception of locomotor testing). Both male and female GFP⁺ and GFP⁻ mice from the WATER and Nic200 groups were used for the behavioral analyses (Fig. 1E). Each mouse went through a total of three behavioral tests with each test conducted 1–2 weeks apart. The behavioral tests were counterbalanced such that all groups were represented on each day of testing (e.g., male/female, GFP⁻/GFP⁺, Water/Nic200). The sequence of the behavioral tests was as follows: locomotor activity, EPM, and finally the Y-maze. An overhead video camera was used to record the behavioral tests with the exception of the locomotor activity test during which the activity of the mouse was monitored by infrared sensors. An individual “blinded” to the identity of the experimental groups scored the behavioral parameters.

Locomotor Activity

Mice were placed individually into testing cages equipped with photobeam motion sensors (Photobeam Activity System; San Diego Instruments) (Zhu et al. 2012, 2017; McCarthy et al. 2018; Zhang et al. 2018). Sensors were positioned along the x and y planes, and each instance of consecutive beam breaks in adjacent photobeams (positioned 5.4 cm apart) was scored as an ambulatory event. Testing began at 17:00 h and continued until 7:00 h the following morning. The period of analysis included a 2-h lights-on period followed by a 12-h lights-off period. The initial 1 h of the lights-on period represented exploratory behavior in a novel environment, whereas activity in the 12-h lights-off period represented spontaneous locomotor activity during the “active” period.

Elevated Plus Maze

A maze consisting of two open arms (50 cm × 10 cm) and two closed arms (50 cm × 10 cm with 40-cm high walls) was used (Med Associates, Inc.) to measure approach–avoidance and exploratory behaviors (Montgomery 1955). In addition, head dips, a measure of exploratory behavior (Rodgers et al. 1997), were also analyzed. The mouse was placed at the center of the maze while facing an open arm and was allowed to freely explore the maze for a 5-minute period (Zhang et al. 2018). The percentage of time spent in the open arms and the total number of head dips (i.e., head dips over the open arms regardless of whether the body was in the closed or open arm) were recorded.

Y-Maze

A custom-built Y-maze (each arm = 35 cm long × 6 cm wide × 10 cm high) was used to assess spatial working memory (Zhu et al. 2017; McCarthy et al. 2018; Zhang et al. 2018). Unique visual cues were placed on the outer surface of the walls of each arm as well as on the walls of the testing room to facilitate spatial orientation. The mouse was placed at the center of the “Y” and allowed to freely explore the maze for 6 minutes. An arm entry was considered to have occurred when all four limbs of the mouse entered an arm. A “spontaneous alternation” was defined as a sequence of three consecutive nonrepeating arm entries (e.g., ABC, BCA, CAB but not ABB, CCB, BAA, etc.). The percentage of spontaneous alternations was calculated using the formula: number of correct spontaneous alternations/(number of entries-2) × 100. The total number of arm entries was also recorded.

Statistical Analyses

The body weight (growth curve) and water intake data were analyzed using a mixed model repeated measures analysis, whereas the chi-squared test was used to determine breeding success differences among treatment groups. Developmental milestones, behavioral data, cell numbers, and ratios were analyzed using a general linear statistical model to evaluate between-subject and within-subject factors. Plasma cotinine levels were analyzed using one-sample and independent samples t-tests. Between-subject factors were treatment (WATER, Nic100, and Nic200), sex (female and male), and, in the case of the behavioral data, the GFP transgene (GFP⁺ and GFP⁻). The within-subject factors analyzed for the cell count data were cortical regions (PFC and mPFC), position (rostral and caudal), and cortical layer (II–III, V, and VI). Statistically significant (P < 0.05) findings from omnibus tests were followed up using multivariate contrast tests to analyze between-subject and within-subject effects and their interactions (i.e., interactions among treatment, sex, the GFP transgene, cortical region, rostro-caudal position, and cortical layer). A one-way analysis of variance was used to analyze between-subject effects and main effects of treatment, sex, and the GFP transgene. For the multivariate tests, probability distribution was analyzed using the Wilks’ lambda distribution (i.e., P and F values). Marginal means and pairwise comparisons were used to directly compare statistically significant main effects (of treatment, sex, GFP, etc.) and statistically significant interactions (for a between-subject factor combined with any of the within-subjects factors). For post-hoc analyses of the main effect of treatment (which had three groups), we applied Fisher’s least significant difference test. We controlled for type I error in the post hoc analysis of those significant interactions using the linear step-up procedure (Benjamini and Hochberg 1995). Outliers were identified using the median ± 1.5 interquartile ranges (IQR) criterion (Tukey 1977) for each treatment group. Once identified, all outliers greater than the median + 1.5 × IQR were “brought to the fence” by assignment of the value equal to the median + 1.5 × IQR, whereas all outliers smaller than the median—1.5 × IQR were “brought to the fence” by assignment of the value equal to the median—1.5 × IQR. The analyses were performed using SPSS statistical software (IBM). The number of outliers “brought to the fence” for each analyses are listed in Supplementary Table 1.

For water intake, developmental milestones, and body weights, we used a litter-based design where each litter was considered an “n.” For the cell counts and behavioral analyses, each individual mouse was considered an “n.” We tested for the effect of litter using a hierarchical linear model with litter as a factor for each of the analyses. Of the 47 analyses that were performed, after controlling for type I error using the linear step-up procedure, none showed a significant effect of litter (P > 0.05). The number of individual subjects as well as the number of litters from which the individual subjects were derived for each analysis are shown in Figure 1B,E and are also stated in the Results section.

Results

Drinking Water Intake

We measured the water intake for all three treatment groups: WATER (n = 4), Nic100 (n = 4), and Nic200 (n = 4) in nonpregnant female mice during the initial 3-week period prior to mating. We found a significant main effect of treatment (F_{(2, 9)} = 8.48, P < 0.01) (mean [mL/day] ± SEM: WATER: 9.72 ± 0.81; Nic100: 7.03 ± 0.81; Nic200: 5.03 ± 0.81). Pairwise analysis revealed that the Nic100 and Nic200 groups consumed significantly lower volume of drinking water compared to the WATER group (WATER vs. Nic100: P < 0.05; t = 2.35, df = 9.00; WATER vs. Nic200: P < 0.01; t = 4.10, df = 9.00). There were no significant differences in water consumption between Nic100 and Nic200 groups (Nic100 vs. Nic200: P > 0.05; t = 1.75, df = 9.00).

For the prenatal and postnatal periods, we analyzed the water intake for the WATER (n = 11) and Nic200 (n = 10) groups. We did not find a significant main effect of treatment (F_{(1, 15.98)} = 4.16, P > 0.05) (mean [mL/day] ± SEM: WATER: 14.96 ± 0.55; Nic200: 13.35 ± 0.56) but did find a significant main effect of time such that, as expected, the water intake increased over time during the prenatal and postnatal periods (F_{(4, 12.43)} = 81.75, P < 0.001) (prenatal week 1 [PR1]: mean ± SEM: 9.92 ± 0.87; prenatal week 2 [PR2]: 9.52 ± 0.39; prenatal week 3 [PR3]: 10.15 ± 0.49; postnatal week 1 [PO1]: 16.92 ± 1.04; postnatal week 2 [PO2]: 24.26 ± 1.01). We did not find a significant treatment by time interaction (Fig. 2).

Figure 2.

Drinking water consumption by the female mice in the plain drinking water group (WATER) and in the Nic200 group was measured for each of the 3 weeks during pregnancy (prenatal) and for each of the 2 weeks during the postpartum period (postnatal).

Plasma Cotinine Levels

We analyzed plasma cotinine levels in the WATER (n = 4), Nic100 (n = 4), and Nic200 (n = 4) groups at the end of the initial 3-week nicotine exposure period (i.e., prior to breeding) using a one-sample t-test to determine whether cotinine levels were significantly different from 0. As expected, the plasma cotinine level in the WATER group was not significantly different from 0. In contrast, the cotinine levels in the Nic100 and Nic200 treatment groups were significantly different from 0 (mean [ng/mL] ± SEM: Nic100: 63.10 ± 11.69, P < 0.05; t = 5.40, df = 3; Nic200: 82.95 ± 24.27, P < 0.05; t = 3.42, df = 3). An independent samples t-test showed that the Nic100 and Nic200 treatment groups were not significantly different from each other (P > 0.05; t = −0.74, df = 6).

Plasma cotinine levels were assessed in the dams and offspring from the WATER (dams: n = 7; pups: n = 7) and Nic200 (dams: n = 9; pups: n = 4) groups at the time of weaning the offspring around P21. One-sample t-test showed that the plasma cotinine levels in the dams and the offspring in the WATER group were not significantly different from 0. In contrast, cotinine levels in the dams and offspring in the Nic200 group were significantly different from 0 (mean [ng/mL] ± SEM: Nic200 dam: 132.44 ± 3.64, P < 0.001; t = 36.40, df = 8; Nic200 P21 pups: 128.18 ± 8.81, P < 0.01; t = −14.09, df = 3). An independent samples t-test showed that there was no significant difference in cotinine levels between the dams and the P21 offspring in the Nic200 group (P > 0.05; t = 1.07, df = 11).

Breeding Success, Gestational Length, Litter Size at Birth, Body Weights, and Developmental Milestones

We did not observe differences among the WATER (n = 16), Nic100 (n = 7), and Nic200 (n = 15) groups with respect to breeding success (i.e., incidence of vaginal plugs; χ²_{(2, N = 38)} = 1.10, P > 0.05). Similarly, we did not observe significant differences among the groups with regard to gestational length (F_(2,31) = 0.13, P > 0.05) (WATER: n = 12; Nic100: n = 7; and Nic200: n = 13) or litter size (F_{(2, 33)} = 1.57, P > 0.05) (WATER: n = 14; Nic100: n = 7; and Nic200: n = 13) using a general linear model, consistent with findings from our earlier studies of developmental nicotine exposure (Zhu et al. 2012, 2014, 2017; Zhang et al. 2018).

Body weights of the offspring at P0, P4, P7, P14, and P21 were analyzed using a mixed-model repeated measures analysis (WATER: n = 13; Nic100: n = 5; Nic200: n = 9). There was no significant main effect of treatment on body weight (F_{(2, 23.60)} = 0.27, P > 0.05), whereas there was a significant main effect of age (F_{(1, 23.63)} = 1033.38, P < 0.001), as expected. The treatment × time interaction was also significant (F_{(2, 23.75)} = 3.47, P < 0.05) (Table 1). Post hoc analyses revealed that the body weight for the Nic100 group was significantly higher at P7, P14, and P21 compared to the WATER group: P7 (P < 0.05; t = 2.10, df = 24.62), P14 (P < 0.05; t = 2.31, df = 24.52), and P21 (P < 0.05; t = 2.39, df = 24.41).

Table 1.

Gestational Length, Litter Size, Body Weights, and Developmental Milestones at Birth

	Water	Nic100	Nic200
Gestational length
# days	19.41 ± 0.21	19.29 ± 0.28	19.46 ± 0.20
Litter size on P0
# Pups	11.79 ± 0.52	10.86 ± 0.76	10.39 ± 0.54
Body weight (g)
P0	1.66 ± 0.04	1.67 ± 0.07	1.62 ± 0.05
P4	3.20 ± 0.10	3.55 ± 0.16	3.34 ± 0.12
P7	4.36 ± 0.16	4.96 ± 0.25a	4.62 ± 0.18
P14	7.05 ± 0.28	8.26 ± 0.44a	7.63 ± 0.33
P21	9.75 ± 0.41	11.56 ± 0.64a	10.63 ± 0.47
Developmental milestones (# days)
External ear detachment	3.77 ± 0.19	3.50 ± 0.27	4.11 ± 0.23
Appearance of fur	4.00 ± 0.26	3.50 ± 0.45	4.30 ± 0.29
Eye opening	14.17 ± 0.28	13.80 ± 0.40	13.56 ± 0.32

Mean ± SEM values for the length of the gestational period, litter size on P0, and body weights of the offspring on P0, P4, P7, P14, and P21 from the plain drinking water (WATER) and drinking water containing either 100 μg/mL (Nic100) or 200 μg/mL (Nic200) groups. The postnatal day (mean ± SEM) for developmental milestones such as external ear detachment, the appearance of fur, and opening of eyes were also recorded.

^aWATER vs. Nic100, P < 0.05.

We also analyzed other developmental milestones using a general linear model and did not find a significant main effect of treatment on time of external ear detachment (F_{(2, 27)} = 1.65, P > 0.05) (WATER: n = 13; Nic100: n = 6; and Nic200: n = 9), time of appearance of fur (F_{(2, 26)} = 1.17, P > 0.05) (WATER: n = 13; Nic100: n = 4; and Nic200: n = 10), or time of eye opening (F_{(2, 25)} = 1.21 P > 0.05) (WATER: n = 12; Nic100: n = 5; and Nic200: n = 9) (Table 1).

GFP⁺NeuN⁺ Cells (GABA Neurons)

The numerical densities of GFP⁺NeuN⁺ cells (GABA neurons) were calculated for WATER (n = 13; male: 7; female: 6), Nic100 (n = 12; male: 6; female: 6), and Nic200 (n = 12; male: 6; female: 6) treatment groups using a general linear model. Analysis of the effects of between-subject factors showed no significant main effects of treatment (F_{(2, 30)} = 3.03, P > 0.05) (mean ± SEM: WATER: 915.08 ± 63.80; Nic100: 1017.03 ± 69.44; and Nic200: 782.31 ± 66.21) (Fig. 3G) or sex (F_{(1, 30)} = 0.15, P > 0.05) (females: 919.55 ± 55.83; males: 890.06 ± 52.75). The treatment × sex interaction was also not significant (Supplementary Table 2).

Analysis of the effects of within-subject factors showed a significant main effect of cortical area -(F_{(1, 30)} = 72.34, P < 0.001) (PFC: 749.35 ± 25.01; mPFC: 1060.26 ± 54.71). There was also a significant main effect of rostral–caudal position (F_{(1, 30)} = 7.10, P < 0.05) (mean ± SEM: rostral: 855.64 ± 41.50; caudal: 953.97 ± 43.69) and cortical layers (F_{(2, 29)} = 69.45, P < 0.001) (layers II–III: 1097.72 ± 47.55; layer V: 881.66 ± 38.59; layer VI: 735.03 ± 36.19). The three layer groups were significantly different from one another such that the GABA neuron numerical density in layers II–III was higher compared to layer V (P < 0.001; t = 9.31, df = 29) and layer VI (P < 0.001; t = 11.94, df = 29), and the numerical density in layer V was significantly higher compared to layer VI (P < 0.001; t = 6.95, df = 29) (Supplementary Table 3).

The between-subject and within-subject interactions were analyzed using multivariate tests. We did not find significant treatment × cortical area, treatment × rostral–caudal position, or treatment × layer interactions (Fig. 3G). We did, however, find a significant treatment × cortical area × rostral–caudal position interaction (F_{(2, 30)} = 4.01, P < 0.05), a significant treatment × rostral–caudal position × layer × sex interaction (F_{(4, 58)} = 5.92, P < 0.001), and a significant treatment × cortical area × layer interaction (F_{(4, 58)} = 2.84, P < 0.05). However, post hoc analyses did not reveal significant differences for any of the pairwise comparisons.

GFP⁻NeuN⁺ Cells (Non-GABA Neurons)

The numerical densities of GFP⁻NeuN⁺ cells (non-GABA neurons) were calculated for WATER (n = 13; male: 7; female: 6), Nic100 (n = 12; male: 6; female: 6), and Nic200 (n = 12; male: 6; female: 6) treatment groups. Analysis of the effects of between-subject factors did not show significant main effects of treatment (F_{(2, 31)} = 1.41, P > 0.05) (mean ± SEM: WATER: 6752.46 ± 345.07; Nic100: 6796.44 ± 358.10; Nic200: 7503.91 ± 358.10) (Fig. 3H) or sex (F_{(1, 31)} = 1.17, P > 0.05) (mean ± SEM: females: 6797.07 ± 292.39; males: 7238.14 ± 285.34). The treatment × sex interaction was also not significant (Supplementary Table 2).

Similar to the GABA neurons, analysis of the effects of within-subject factors showed significant main effects of cortical area (F_{(1, 31)} = 114.45, P < 0.001) (PFC: 5986.79 ± 172.60; mPFC: 8048.42 ± 268.76) and of layers (F_{(2, 30)} = 195.07, P < 0.001) (mean ± SEM: layers II–III: 7426.22 ± 233.79; layer V: 6309.47 ± 204.36; layer VI: 7317.12 ± 195.35) but not of rostral–caudal position (F_{(1, 31)} = 0.33, P > 0.05) (mean ± SEM: rostral section: 7083.68 ± 239.97; caudal section: 6951.53 ± 229.73). Post hoc pairwise comparisons revealed significant differences among layers such that the non-GABA neuron numerical density in layer V was reduced compared to layers II–III and layer VI (mean ± SEM: layer V vs. layers II–III: P < 0.001; t = −15.19, df = 31; layer V vs. layer VI: P < 0.001; t = 11.30, df = 31) (Supplementary Table 3).

The between-subject and within-subject interactions were analyzed using multivariate tests. We did not find significant treatment × cortical area, treatment × rostral–caudal position, or treatment × layer interactions. We did, however, find a significant treatment × cortical area × rostral–caudal region interaction (F_{(2, 31)} = 4.27, P < 0.05). However, post hoc pairwise comparisons did not reveal significant differences.

GABA-to-Non-GABA Neuron Ratio

The GABA-to-non-GABA neuron ratios were calculated for WATER (n = 13; male: 7; female: 6), Nic100 (n = 12; male: 6; female: 6), and Nic200 (n = 12; male: 6; female: 6) treatment groups. Analysis of the effects of between-subject factors showed a significant main effect of treatment (F_{(2, 30)} = 5.884, P < 0.01) (mean ± SEM: WATER: 0.141 ± 0.010; Nic100: 0.161 ± 0.011; Nic200: 0.108 ± 0.011) (Fig. 3I) but not of sex (F_{(1, 30)} = 0.147, P > 0.05) (mean ± SEM: females: 0.139 ± 0.009; males: 0.134 ± 0.009). The treatment × sex interaction was also not significant. Pairwise comparisons revealed that the Nic200 treatment group had a significantly lower GABA-to-non-GABA neuron ratio compared to the WATER and Nic100 treatment groups (WATER vs. Nic200: P < 0.05; t = −2.133, df = 30; Nic100 vs. Nic200: P < 0.01; t = −3.313, df = 30) (Supplementary Table 2).

Analysis of the effects of within-subject factors showed no significant main effects of cortical area (F_{(1, 30)} = 2.701, P > 0.05) (mean ± SEM: PFC: 0.132 ± 0.006; mPFC: 0.141 ± 0.008) or rostral–caudal position (F_{(1, 30)} = 4.026, P > 0.05) (rostral: 0.127 ± 0.007; caudal: 0.147 ± 0.009). There was, however, a significant main effect of layer (F_{(2, 29)} = 46.508, P < 0.001) (mean ± SEM: layers II–III: 0.159 ± 0.008; layer V: 0.147 ± 0.007; layer VI: 0.105 ± 0.006). Post hoc pairwise comparisons revealed significant differences among cortical layers such that layers II–III had a significantly higher ratio than layers V and VI and that layer V had a significantly higher ratio than layer VI (mean ± SEM: layers II–III vs. layer V: P < 0.05; t = 2.200, df = 30; layers II–III vs. layer VI: P < 0.001; t = 9.000, df = 30; layer V vs. layer VI: P < 0.001; t = 8.600, df = 30) (Supplementary Table 3).

The between-subject and within-subject interactions were then analyzed using multivariate tests. We found a significant treatment × cortical area interaction (F_{(2, 30)} = 3.533, P < 0.05). Pairwise comparisons indicated that the GABA-to-non-GABA neuron ratio in the Nic200 treatment group is significantly reduced in the mPFC compared to Nic100 group (P < 0.01; t = −3.595; df = 30) (mean ± SEM: Nic100 mPFC: 0.176 ± 0.014; Nic200 mPFC: 0.106 ± 0.014). We did not find significant treatment × rostral–caudal position or treatment × layer interactions. In addition, there was a significant treatment × cortical area × rostral–caudal position interaction (F_{(2, 30)} = 6.114, P < 0.01). Pairwise comparisons revealed that in caudal regions of the mPFC, the Nic100 group had a significantly higher ratio compared to both WATER (P < 0.01; t = −2.885, df = 30) and Nic200 groups (P < 0.001; t = 3.704, df = 30) (mean ± SEM: WATER caudal mPFC: 0.137 ± 0.018; Nic100 caudal mPFC: 0.212 ± 0.019; Nic200 mPFC: 0.112 ± 0.019). Finally, we found a significant treatment × rostral–caudal position × layer × sex interaction (F_{(4, 58)} = 4.423, P < 0.01). However, pairwise comparisons did not reveal significant differences.

Cortical Thickness

The cortical thickness was measured in histological sections of the PFC and mPFC from the WATER (n = 13; male: 7; female: 6), Nic100 (n = 12; male: 6; female: 6), and Nic200 (n = 12; male: 6; female: 6) treatment groups, and the data were analyzed using a general linear model. Analysis of the effects of between-subject factors showed no significant main effects of treatment (F_{(2, 31)} = 0.693, P > 0.05) (mean (mm) ± SEM: WATER: 0.862 ± 0.012; Nic100: 0.881 ± 0.012; Nic200: 0.875 ± 0.012) or sex (F_{(1, 31)} = 0.158, P > 0.05) (mean ± SEM: females: 0.875 ± 0.010; males: 0.870 ± 0.010). The treatment × sex interaction was also not significant. Analysis of the effects of within-subject factors showed a significant main effect of cortical area (F_{(1, 31)} = 2447.349, P < 0.001) (mean ± SEM: PFC: 1.085 ± 0.010; mPFC: 0.660 ± 0.007) and rostro–caudal position (F_{(1, 31)} = 275.444, P < 0.001) (Mean ± SEM: rostral: 0.912 ± 0.007; caudal: 0.833 ± 0.008).

The between-subject and within-subject interactions were then analyzed using multivariate tests. We found a significant interaction between cortical area and rostral–caudal position (F_{(1, 31)} = 13.608, P < 0.01) (mean ± SEM: PFC rostral: 1.115 ± 0.008; mPFC rostral: 0.709 ± 0.008; PFC caudal: 1.056 ± 0.012; mPFC caudal: 0.610 ± 0.006). Pairwise comparisons revealed that in both the PFC and mPFC, there was a significant effect of rostral–caudal position such that the cortical thickness was significantly greater in the rostral position (PFC rostral vs. caudal: P < 0.01; t = 7.375; df = 31; mPFC rostral vs. caudal: P < 0.01; t = 16.500, df = 31). We did not find significant treatment × cortical area, treatment × rostral–caudal position, or treatment × cortical area × rostral–caudal position interactions.

Behavioral Analyses

We performed the behavioral studies described below in WATER and Nic200 groups. We did not include the Nic100 group because there were no significant changes in this group in either the numbers of GABA or non-GABA cells, or in the GABA to no-GABA neuron ratio (Fig. 3; Supplementary Table 2). Therefore, the probability of detecting behavioral effects of the nicotine exposure, if any, would be higher in the Nic200 group than in the Nic100 group. Based on this rationale, we included only the Nic200 group in the behavioral studies to conserve effort and minimize animal use.

Locomotor Activity

Locomotor activity data were analyzed separately for the initial 1 h upon placement in the testing cage (exploratory behavior in a novel environment) and for the subsequent 12 h (spontaneous locomotor activity) using a general linear model for WATER (n = 50; male GFP⁻: 17; male GFP⁺: 8; female GFP⁻: 17; female GFP⁺: 8) and Nic200 (n = 49; male GFP⁻: 13; male GFP⁺: 11; female GFP⁻: 15; female GFP⁺: 10) treatment groups. During the initial 1-h exploration period, we observed a significant main effect of treatment (F_{(1, 98)} = 4.17, P < 0.05) (mean ± SEM: WATER: 1093.55 ± 68.73; Nic200: 1287.47 ± 65.56) (Fig. 4A), sex (F_{(1, 98)} = 5.12, P < 0.05) (mean ± SEM: females: 1297.92 ± 67.10; males: 1083.10 ± 67.22) as well as the GFP transgene (F_{(1, 98)} = 7.02, P < 0.05) (mean ± SEM: GAD67-GFP⁻: 1316.32 ± 57.93; GAD67-GFP⁺: 1064.69 ± 75.26) on exploratory activity. The treatment × sex, treatment × GFP, or treatment × sex × GFP interactions were not significant.

Figure 4.

Behavioral analyses. Locomotor activity (A, B), approach–avoidance, and exploratory behavior in an EPM (C, D, E) and spatial working memory in a Y-maze (F, G) were analyzed in male and female (GFP⁺ and GFP⁻) offspring from dams exposed to plain drinking water (WATER) or 200 μg/mL nicotine (Nic200) in the drinking water. Locomotor activity for the initial 1 h (A; exploratory behavior) and the subsequent the 12 h (B; lights-off; spontaneous locomotor activity) were analyzed. For the EPM, time spent in the open arms as a percentage of the total time in the maze (C),total arm entries (D), and the total number of head dips (E) were analyzed. For the Y-maze, percent spontaneous alternations (F) and total arm entries (G) were analyzed. ^*P < 0.05; ^aWATER female vs. Nic200 female, P < 0.05.

For spontaneous locomotor activity during the 12-h lights-off period, we did not find a significant main effect of treatment (F_{(1, 98)} = 0.09, P > 0.05) (mean ± SEM: WATER: 4733.09 ± 196.51; Nic200: 4653.22 ± 187.44) (Fig. 4B). However, there were significant main effects of sex (F_{(1, 98)} = 32.64, P < 0.001) (mean ± SEM: females: 5468.90 ± 191.87; males: 3917.41 ± 192.19) and the GFP transgene (F_{(1, 98)} = 8.87, P < 0.01) (mean ± SEM: GAD67-GFP⁻: 5097.44 ± 165.65; GAD67-GFP⁺: 4288.87 ± 215.20). The treatment × sex and treatment × GFP interactions were not significant. The treatment × sex × GFP interaction was significant (F_{(2, 98)} = 3.26, P < 0.05). However, follow-up post hoc analyses did not reveal significant pairwise comparisons.

Approach–Avoidance Behavior

The approach–avoidance behavior was measured using the EPM for WATER (n = 56–58; male GFP⁻: 19; male GFP⁺: 11; female GFP⁻: 18–19; female GFP⁺: 8–9) and Nic200 (n = 53–55; male GFP⁻: 14; male GFP⁺: 13; female GFP⁻: 18–19; female GFP⁺: 8–9) treatment groups. We examined the percent time spent in the open arms, the number of total arm entries, and the number of total head dips. For the percent time spent in the open arms of the EPM, we found significant main effects of treatment (F_{(1, 112)} = 5.87, P < 0.05) (mean ± SEM: WATER: 28.51 ± 1.80; Nic200: 34.70 ± 1.82) (Fig. 4C), sex (F_{(1, 112)} = 8.95, P < 0.01) (mean ± SEM: females: 27.78 ± 1.86; males: 35.43 ± 1.76), and the GFP transgene (F_{(1, 112)} = 6.77, P < 0.05) (mean ± SEM: GAD67-GFP⁻: 34.93 ± 1.56; GAD67-GFP⁺: 28.28 ± 2.03). The treatment × sex, treatment × GFP, or treatment × sex × GFP interactions were not significant.

For total arm entries, we did not find a significant main effect of treatment (F_{(1, 112)} = 2.78, P > 0.05) (mean ± SEM: WATER: 20.22 ± 0.65; Nic200: 21.75 ± 0.65) (Fig. 4D), sex(F_{(1, 112)} = 0.12, P > 0.05) (mean ± SEM: females: 21.14 ± 0.67; males: 20.83 ± 0.63), or the GFP transgene (F_{(1, 112)} = 0.26, P > 0.05) (mean ± SEM: GAD67-GFP⁻: 21.22 ± 0.59; GAD67-GFP⁺: 20.75 ± 0.73). In addition, there was no significant treatment × sex interaction, treatment × GFP interaction, or treatment × sex × GFP interaction.

For the total number of head dips, we did not find significant main effects of treatment (F_{(1, 108)} = 3.20, P > 0.05) (mean ± SEM: WATER: 13.08 ± 1.04; Nic200: 15.72 ± 1.05) (Fig. 4E), sex (F_{(1, 108)} = 0.022, P > 0.05) (mean ± SEM: females: 14.29 ± 1.10; males: 14.51 ± 0.99), or the GFP transgene (F_{(1, 108)} = 0.07, P > 0.05) (mean ± SEM: GAD67-GFP⁻: 14.59 ± 0.88; GAD67-GFP⁺: 14.20 ± 1.18). However, there was a significant treatment × sex interaction (F_{(1, 108)} = 7.92, P < 0.01). Pairwise comparisons for the significant treatment × sex interaction revealed that females in the Nic200 treatment group had significantly greater number of head dips compared to WATER females (mean ± SEM: WATER: 10.90 ± 1.55; Nic200: 17.68 ± 1.55; P < 0.01; t = 3.10, df = 101). Males in the Nic200 treatment group did not significantly differ from WATER males (mean ± SEM: WATER: 15.26 ± 1.38; Nic200: 13.75 ± 1.41; P > 0.05; t = −0.77, df = 101).

Spatial Working Memory

Spatial working memory was analyzed using the Y-maze for WATER (n = 57; male GFP⁻: 18; male GFP⁺: 11; female GFP⁻: 19; female GFP⁺: 9) and Nic200 (n = 56; male GFP⁻: 14; male GFP⁺: 13; female GFP⁻: 19; female GFP⁺: 10) treatment groups. We examined percent spontaneous alternations and the number of total arm entries. We did not find significant main effects of treatment (F_{(1, 112)} = 1.56, P > 0.05) (mean ± SEM: WATER: 61.26 ± 1.20; Nic200: 59.17 ± 1.18) (Fig. 4F), sex (F_{(1, 112)} = 2.48, P > 0.05) (mean ± SEM: females: 58.89 ± 1.21; males: 61.54 ± 1.17), or the GFP transgene (F_{(1, 112)} = 2.26, P > 0.05) (mean ± SEM: GAD67-GFP⁻: 58.95 ± 1.04; GAD67-GFP⁺: 61.48 ± 1.32) on spontaneous alternations in the Y-maze. Nor was there a significant interaction for treatment × sex, treatment × GFP, or treatment × sex × GFP.

Total number of arm entries in the Y-maze did not show significant main effects of treatment (F_{(1, 112)} = 0.002, P > 0.05) (mean ± SEM: WATER: 42.69 ± 1.27; Nic200: 42.77 ± 1.26) (Fig. 4G), sex (F_{(1, 112)} = 1.46, P > 0.05) (mean ± SEM: females: 43.82 ± 1.29; males: 41.65 ± 1.24), or the GFP transgene (F_{(1, 112)} = 2.22, P > 0.05) (mean ± SEM: GAD67-GFP⁻: 44.07 ± 1.10; GAD67-GFP⁺: 41.40 ± 1.41). Nor was there a significant interaction for treatment × sex, treatment × GFP, or treatment × sex × GFP.

Discussion

Our data show that developmental nicotine exposure produces significant deficits in frontal cortical GABA-to-non-GABA neuron ratio and significant increases in exploratory behavior in a novel environment and a shift in the approach–avoidance behavior toward approach behavior. The effects are dose and sex dependent. We suggest that the reduction in GABA-to-non-GABA neuron ratio is evidence of a reduction in the overall inhibitory tone in the frontal cortex. Consistent with this suggestion, we found a reduction in GABA neuron numbers (approximately 14%), although the difference was not statistically significant (P = 0.06). The non-GABA neuron numbers were unaffected. The behavioral changes are consistent with novelty-seeking behaviors reported in prenatally nicotine-exposed human subjects (Cornelius and Day 2000). Thus, our findings suggest a correlation among developmental nicotine exposure, reduced frontal cortical inhibitory tone, and increased novelty-seeking behavior. These behavioral changes occurred without significant alterations in the intrinsic region-specific cortical cytoarchitectonic features such as cortical thickness or laminar distribution of neurons that distinguish the PFC from mPFC.

The present findings may be consistent with clinical reports that developmental nicotine exposure is associated with reduced cortical inhibitory tone. For example, cigarette smoking during pregnancy increases the risk for ADHD in the offspring, and memantine, an NMDA receptor antagonist that reduces the excitatory tone, produces therapeutic benefit in ADHD patients (Surman et al. 2013). Other reports showed that developmental nicotine exposure increases the risk for seizures (Al-Hachim and Mohmood 1985; Cassano et al. 1990; Berg et al. 1995; Sidenvall et al. 2001; Rong et al. 2014) and seizures are thought to result from reduced inhibitory tone (Kobayashi et al. 2003).

In the present study, the nicotine-exposed mice spent longer periods of time in the open arms of the EPM, suggesting impaired approach–avoidance balance and increased novelty-seeking behavior. An earlier study using a developmental nicotine exposure paradigm similar to ours also reported increased time spent in the open arm of the elevated zero maze, another method to assay approach–avoidance behavior (Buck et al. 2019). In addition, the nicotine-exposed mice in our study showed increased locomotor activity during the initial 1 h of locomotor activity assay, which represents increased exploration of a novel environment. Collectively, these behavioral data indicate increased novelty-seeking phenotypes (Cloninger 1986), which share neurobiological mechanisms with risk-taking and drug-seeking behaviors (Kelley et al. 2004; Hansson et al. 2012; Wang et al. 2015). Novelty-seeking and risk-taking behaviors reliably predict increased risk for drug abuse and compulsive drug-taking (Piazza et al. 1990; Zuckerman 1990; Dellu et al. 1996; Hayton et al. 2012). In addition, developmental nicotine exposure increases the risk for drug addiction, including nicotine self-administration later in life (Cornelius and Day 2000; Hellström-Lindahl and Nordberg 2002; Slotkin et al. 2005; Levin et al. 2006; Button et al. 2007; Cornelius and Day 2009; Chistyakov et al. 2010).

Studies examining risk assessment or cost–benefit analysis in human subjects use tests such as Balloon Analog Risk test, Iowa Gambling Task, and probabilistic or delayed discounting. These tests have been modified for use in rodents successfully. However, these tasks rely heavily on the concept of loss or mitigation of reward. In the case of rodents, the risk–benefit assessment may be ethologically more relevant when there is a tradeoff between danger or threat (from predators) and the drive to explore the environment in search of food. In the EPM and the novel environment used in the locomotor activity assay present study, the mouse is exposed to such intrinsic “conflict.” Other studies using rodent models confirm the use of the EPM as a measure of novelty-seeking behavior in rodents (Montgomery 1955; Macrì et al. 2002; Lynn and Brown 2009). Thus, the behavioral methods used here appear to be relevant to evaluate novelty-seeking behavior in rodents. Therefore, we suggest that the present study not only extends the preclinical and clinical findings of increased novelty-seeking as a behavioral consequence of developmental nicotine exposure but also offers a cellular and neurochemical basis for the consequence.

We used a GAD67-GFP transgenic reporter mouse to facilitate analysis of GABA and non-GABA neuron numbers. Although previous studies had confirmed the specificity of GFP labeling and lack of off-target behavioral effects in this mouse model (Tamamaki et al. 2003), we ascertained the fidelity of the GFP labeling by using GABA immunohistochemistry. We found that every GABA⁺ neuron was also GFP⁺ (Fig. 3D–F), consistent with the findings from our previous study that had used the same reporter mouse to examine the effects of prenatal cocaine exposure (McCarthy and Bhide 2012).

Our experimental design permitted quantification of the independent effects of the GFP transgene on behavioral phenotypes. We found that although the GFP transgene did not produce significant effects on working memory, it did produce significant main effects on exploratory behavior and approach–avoidance behavior. To our knowledge, this is the first report of behavioral changes in heterozygous GAD67-GFP mice. Homozygous mice are prone to seizures and are not viable (Tamamaki et al. 2003). In addition, the GABA content of the forebrain is reduced from birth until 7 weeks of age in the homozygous mice. It is possible that GABA content is altered in the heterozygous mice as well and that the reduced GABA may contribute to the behavioral phenotypes observed here in the heterozygous mice. In fact, changes in GABA function would be consistent with changes in exploratory behavior and approach–avoidance behavior observed in the heterozygous mice in the present study. However, a major finding from the present study is that the interaction between GFP transgene and nicotine exposure was not significant for any of the phenotypes. In other words, the developmental nicotine exposure did not affect the GFP⁺ and GFP⁻ (i.e., wild type) mice differently.

We found a dose-dependent effect of the nicotine exposure on the GABA-to-non-GABA neuron ratio. Consistent with this finding, the drinking water consumption and plasma cotinine levels suggested that the nicotine exposure in the Nic100 group was lower compared to the Nic200 group. The dose-dependent effects on the GABA-to-non-GABA neuron ratio likely reflect differences in activation thresholds for different nAChR subtypes during development (Cohen et al. 2005; Giniatullin et al. 2005). Incidentally, the cotinine concentrations in the present study are equivalent to those reported in other studies that used nicotine exposure paradigms similar to ours (Pauly et al. 2004).

We also found that the effects of developmental nicotine exposure on some of the behavioral phenotypes were sex dependent. For example, although novel environment-induced exploratory behavior was not significantly different between male and female mice in the control (WATER) group, female but not male mice in the Nic200 group had significant increase in this behavior compared to their counterparts in the control group. Similarly, female but not male mice in the Nic200 group had significantly greater total number of head dips compared to their counterparts in the control (WATER) group. These findings are not surprising, as sex-dependent effects of developmental nicotine exposure have been reported previously (Slotkin et al. 2007; Hall et al. 2016) [for review (Zhang et al. 2018)].

Another factor that could affect the phenotypes in the nicotine-exposed groups is the potential impact of any nicotine-induced changes in mother–infant interactions during the preweaning period. Mother–infant interactions can play a significant role in shaping the behavioral phenotypes of the offspring in the short and long (Champagne and Meaney 2006; Faure et al. 2019). We did not perform detailed analysis of such interactions. However, nicotine exposure did not produce significant main effects on developmental milestones suggesting that variability in mother–infant interactions may not have been a significant factor in our study. However, a detailed and systematic analysis of mother–infant behavioral interactions would be necessary to draw definitive conclusions.

The effect of developmental nicotine exposure on GABA-to-non-GABA neuron ratio may reflect changes in multiple developmental events such as neurogenesis, neuronal migration, or survival. Nicotine exposure during the prenatal period impairs proliferation of progenitor cells in the ventricular and subventricular zones of the presumptive cerebral cortex (Aoyama et al. 2016). In contrast to non-GABA neurons that undergo proliferation within the presumptive cerebral cortex, GABA neurons of the rodent cerebral cortex originate in the ganglionic eminences of the basal forebrain (Anderson et al. 1997, 2001; Marín and Rubenstein 2001; Jiménez et al. 2002; Petanjek et al. 2008) and migrate to the developing cerebral cortex during the prenatal period (De Carlos et al. 1996; Anderson et al. 1997; Tamamaki et al. 1997; Letinic et al. 2002). Interestingly, carbon monoxide, ethanol, and cocaine have been shown to influence GABA neuron migration in the mouse embryo (Cuzon et al. 2008; McCarthy et al. 2011; Trentini et al. 2016). Whether nicotine also influences neurogenesis in the basal forebrain or migration of the GABA neurons from the basal to the dorsal forebrain, however, is not yet known, although GABA neurons in the embryonic forebrain express nAChRs creating a substrate for nicotine’s action (Liu et al. 2006).

Across the three groups of mice (WATER, Nic100, and Nic200), the numerical density of GABA and non-GABA neurons was greater in the mPFC compared to the PFC and greater in layer II compared to layers V and VI in both mPFC and PFC. These observations indicate that although the nicotine exposure produced significant changes in GABA-to-non-GABA neuron ratio, it did not alter the core intrinsic differences in cytoarchitecture and organization between the PFC and mPFC. In addition, the intrinsic differences in cortical thickness between mPFC and PFC or along the rostro-caudal axis within mPFC or PFC were preserved despite the developmental nicotine exposure. Collectively, these data suggest that developmental nicotine exposure in our mouse model did not alter the core cytoarchitecture of the frontal cortex and that the behavioral changes are likely due to changes in function rather than cytoarchitecture.

In summary, our data show that developmental nicotine exposure produces long-term changes in brain structure and function. Specifically, we show that the nicotine exposure reduces cortical GABA-to-non-GABA neuron ratio, which suggests reduced inhibitory tone in the cortex. In addition, nicotine exposure increases approach–avoidance and exploratory behaviors, which indicate increased novelty-seeking and may foreshadow risk-taking and drug-seeking behaviors. Our findings are consistent with clinical and other preclinical studies, which collectively highlight developmental nicotine exposure as a continuing and serious public health concern.

Funding

Jim and Betty Ann Rodgers Chair Fund, Florida State University College of Medicine (grant number F00662).