Adolescent Ethanol Exposure Disrupts the Mature Phenotype of Surviving Neuroprogenitors in Female Mice

ABSTRACT

Rodent models of human adolescent binge drinking persistently reduce adult neurogenesis and induce innate immune signaling cascades, suggesting long‐lasting dysfunction in the hippocampal neurogenic niche. Nevertheless, many neuroprogenitors survive, and the impact of adolescent ethanol on the mature phenotype of these surviving neurons is unknown, resulting in an unstudied cellular population that could impact hippocampal function. To label a cohort of immature neurons, we used doublecortin‐CreERT2 transgenic mice crossed with Rosa26‐CAG‐tdTomato reporter mice, resulting in tamoxifen‐induced labeling of newborn neurons. Female offspring underwent adolescent intermittent ethanol (AIE) (5 g/kg ethanol or water by gavage, 2‐day‐on/2‐day‐off from postnatal day 29–58 ± 1), followed by a multi‐month ethanol‐free period. AIE increased dendritic branch width and cellular migratory distance, increased hilar axonal varicosities, reduced dendritic spine density, and shifted action potential generation with escalating current injections in tdTomato+ cells. Although tdTomato+ expression was also decreased in AIE relative to CON mice, there was no significant change in the number of tdTomato+ somata. AIE also reduced microglial territory in the polymorphic and molecular layers despite no change in cell number. Furthermore, microglia entwined in the dendritic trees of surviving adolescent neuroprogenitors exhibited increased sphericity, potentially indicative of a shift towards a more reactive morphological phenotype. Surviving neuroprogenitors exhibited no changes in labeling of the vesicular acetylcholine transporter, despite robust cholinergic innervation proximal to the mature dendritic tree that reduced rapidly with increasing distance from dendritic arbors. These results suggest that AIE causes lasting changes in the architectural and physiological phenotype of surviving adolescent neuroprogenitors, which may be related to disrupted neuronal‐microglial dynamics and have critical consequences for both hippocampal circuitry and function.

1. Introduction

Alcohol is ubiquitous in society: its use is culturally sanctioned, socially encouraged, often begins in early adolescence, and persists throughout the adult lifespan (Petit et al. 2013). However, binge (4–5+ drinks in 2 h) drinking during adolescence is particularly deleterious, hijacking plasticity of the adolescent brain (Arain et al. 2013) and causing disruptions in normative brain maturation and adult cognitive‐behavioral function (Crews et al. 2019). These lasting molecular and cognitive‐behavioral repercussions have been documented from both human and rodent studies. In humans, adolescent binge drinking is associated with compromised white matter integrity (Bava et al. 2009, 2010; McQueeny et al. 2009) and decreased temporal lobe volume relative to non‐drinkers, emphasizing sensitivity of the temporal lobe to adolescent‐ethanol induced damage (Lees et al. 2020; Squeglia et al. 2014). Converging findings from rodent models of adolescent intermittent ethanol exposure (AIE), a model of adolescent weekend/holiday binge drinking, reveal that AIE‐driven alterations in hippocampal molecular integrity are linked to functional deficits in learning, memory, and behavioral flexibility (for reviews see Crews et al. 2019; Macht et al. 2022). Notably, unlike adult binge ethanol exposure, the molecular and behavioral consequences of adolescent ethanol persist despite extended ethanol‐free recovery periods (Crews et al. 2019; Macht, Elchert, and Crews 2020). These converging lines of evidence suggest that the adolescent brain—especially its hippocampal circuits—is uniquely vulnerable to the enduring impact of binge ethanol exposure.

In addition to its critical role in learning and memory, the hippocampal formation also holds the distinction as one of the two neurogenic regions in the adult mammalian brain, originally identified by Altman and Das in 1965 (Altman and Das 1965), where newborn neurons functionally integrate throughout the lifespan (for reviews see Gould et al. 2000; Hastings and Gould 1999; Kempermann 2015; Kempermann et al. 2015; van Praag et al. 2002; Zhao et al. 2008). This makes hippocampal neurogenesis a unique timestamp for experience‐dependent circuitry development. Acute and chronic ethanol robustly inhibit hippocampal neurogenesis across all developmental stages (Anderson et al. 2012; Broadwater et al. 2014; Crews et al. 2006; Ehlers et al. 2013; Geil et al. 2014; He et al. 2005; W. Liu and Crews 2017; Nixon and Crews 2002; Nwachukwu et al. 2022). Adolescent rodents are particularly sensitive to these effects, with 5 g/kg acute binge ethanol significantly reducing bromodeoxyuridine‐labeled cells 5 h after exposure (Crews et al. 2006). However, the duration of these effects varies by age of exposure and age of evaluation (for review see Macht et al. 2022). For example, intermittent binge ethanol exposure in adulthood results in acute but not persistent deficits in neurogenesis (Broadwater et al. 2014) in males, suggesting that the adolescent brain is uniquely more vulnerable than the adult brain to the persistence of these effects (Crews et al. 2006). In fact, a lasting loss of adult hippocampal neurogenesis is one of the most consistent observations after AIE exposure in rodents (Broadwater et al. 2014; Ehlers et al. 2013; W. Liu and Crews 2017; Macht, Crews, and Vetreno 2020; Macht et al. 2021, 2023; Nwachukwu et al. 2022; Reitz et al. 2021; Swartzwelder et al. 2019; Vetreno and Crews 2015). Moreover, AIE‐loss and restoration of adult neurogenesis are associated with corresponding loss and restoration of performance in many cognitive‐behavioral tasks dependent on hippocampal integrity (Macht, Elchert, and Crews 2020; Macht et al. 2023; Vetreno et al. 2018), making the hippocampal neurogenic niche a key region of interest in linking ethanol‐induced molecular changes to behavioral outcomes (Wooden et al. 2021).

AIE‐induced loss of adult hippocampal neurogenesis is associated with a persistent induction of innate immune signaling cascades, indicating persistent shifts in the neurogenic microenvironment, likely mediated in part by the primary immunocompetent cells in the brain: microglia. Microglia colonize the hippocampus early in neonatal development with morphological shifts from an amoeboid to ramified structure during adolescence (Schwarz et al. 2012). These immunocompetent cells are also primary sculptors of synaptic architecture (Paolicelli et al. 2011), and they are sensitive to acute ethanol with increased dystrophy evident in adolescent and adult males after a 2‐day binge (Marshall et al. 2020). This suggests that AIE shifts in microglial phenotype could not only influence the surrounding microenvironment of the neurogenic niche, but phenotypic changes in microglia could have important implications for the maturation of granule neuron architecture.

However, an important caveat is that while AIE causes a persistent decline in adult neurogenesis (Crews et al. 2006; Ehlers et al. 2013; Macht et al. 2021, 2023; Reitz et al. 2021; Vetreno et al. 2018; Vetreno and Crews 2015), the vast majority of newborn neurons survive AIE. Presumably, these surviving newborn neurons incorporate into functional neural circuits and influence neural network function. Despite this, no studies have examined whether AIE impacts the mature phenotype of these surviving adolescent hippocampal newborn granule neurons. To address this knowledge gap, we used a transgenic mouse line of doublecortin‐CreER^T2 (DCX‐CreER^T2) (Cheng et al. 2011), which was crossed with Rosa26‐CAG‐tdTomato reporter mice to label newborn neurons at temporally specific developmental windows. Because these neurons permanently express tdTomato protein following tamoxifen‐induced recombination, this model enables us to fate‐map neurons that were newborn during adolescent ethanol exposure and then assess their mature phenotype much later in adulthood. We tested the hypothesis that AIE would disrupt the mature morphological and physiological characteristics of immature granule neurons that survived AIE. As microglia are the primary innate immune effectors in the brain and have critical interactions with newborn neuron maturation, we further tested the hypothesis that microglia intertwined in dendritic trees of labeled neurons would exhibit a more reactive phenotype. As several studies have illustrated the connection between the integrity of cholinergic innervation to the hippocampus, innate immune signaling, and adult neurogenesis (Frinchi et al. 2019; Li et al. 2019; Macht et al. 2021; Mohapel et al. 2005; Swartzwelder et al. 2019), we investigated whether AIE disrupted vesicular acetylcholine transporter (VAChT) labeling proximal to tdTomato+ mature granule cell dendrites. Notably, this study exclusively examined females, addressing a longstanding underrepresentation of females in preclinical models of adolescent ethanol exposure. Enduring ethanol‐driven deficits in females warrant more focused investigation given higher post‐pandemic alcohol use and binge drinking in high‐school females relative to males (Hoots 2023) and evidence that alcohol consumption is disproportionately associated with worse long‐term outcomes in females (Patrick et al. 2024). Collectively, these findings indicate that AIE disrupts the phenotype of surviving adolescent neuroprogenitors, providing novel insight into the long‐term consequences of adolescent binge ethanol exposure on hippocampal neurocircuitry.

2. Materials and Methods

2.1. Animals

In order to fate‐map neurons that were in an immature developmental state during adolescence, these experiments used DCX‐CreER^T2 transgenic mice generously provided by Dr. Eric Schnell at Oregon Health & Science University and originally developed by Dr. Zhi‐Qi Xiong, Institute of Neuroscience, Shanghai, China (Cheng et al. 2011). These mice allowed for selective labeling of newborn neurons in the dentate gyrus with tamoxifen‐induced fluorescent proteins. These mice were bred in‐house with female Cg‐Gt(ROSA)26Sortm14(CAG‐tdTomato)Hze/J (Ai14) mice from The Jackson Laboratory, resulting in a heterozygous DCX‐CreER^T2/tdTomato cross (Villasana et al. 2015). For a timeline of all experimental procedures, see Figure 1.

FIGURE 1.

Timeline of experimental procedures. To characterize the mature phenotype of surviving newborn neurons, DCX‐CreER^T2/tdTomato transgenic mice were used to fate‐map adolescent immature neurons. To model patterns of human adolescent binge drinking, the Neurobiology of Adolescent Drinking in Adulthood (NADIA) consortium developed an adolescent intermittent ethanol (AIE) rodent model with a 2‐day on/2‐day off binge gavage cycle of either water or 5 g/kg ethanol delivered by intragastric intubation from PND 29–58 (±1). Consistent with other studies, female DCX‐CreER^T2/tdTomato transgenic mice exhibit BECs of 305.58 ± 19 mg/dL 1‐h post gavage. Two experiments were performed to assess both the morphological (Experiment 1) and physiological (Experiment 2) characteristics of hippocampal granule neurons that were undergoing critical maturation during AIE exposure.

All animals used in experiments were bred in‐house at the University of North Carolina at Chapel Hill, with birth assigned as postnatal day (PND) 0. The mice were housed in a vivarium maintained at a stable 20°C, with a 12‐h light–dark cycle (lights on at 7:00 am), and with ad libitum access to food and water. The experiments conducted were approved by the Institutional Animal Care and Use Committee at the University of North Carolina at Chapel Hill and followed the established standards by NIH regulations for rodent research. These studies consisted of two sets of experiments: (Experiment 1) morphological assessment (N = 24; CON = 12, AIE = 12), and (Experiment 2) electrophysiological assessment (N = 22; CON = 11, AIE = 11); each experiment consisted of a separate cohort of mice.

2.2. Adolescent Intermittent Ethanol (AIE) Exposure Paradigm (Experiments 1 and 2)

Female DCX‐CreER^T2/tdTomato mice were randomly assigned to either the condition of a water‐treated control (CON) or adolescent intermittent ethanol exposure (AIE) in a split‐litter design. From PND 29–58 (±1), the mice were subjected to a 2‐day on/2‐day off oral gavage of either water (CON) or 5 g/kg ethanol (AIE) for a total of 16 doses. Blood ethanol content was assessed from a separate subgroup (n = 12) of DCX‐CreER^T2/tdTomato mice 1 h after the final ethanol administration as previously described (Coleman et al. 2011) using a GL5 Analox Analyzer (GL5, Analox Instruments). BECs of AIE‐treated mice were consistent with prior studies, with readings in the expected range of ethanol exposure: 305.58 ± 19 mg/dL. Following the end of the adolescent ethanol exposure period, the mice underwent an extended ethanol‐free period.

2.3. Tamoxifen‐Induced Recombination (Experiments 1 and 2)

In DCX‐CreER^T2/tdTomato mice, tamoxifen administration induces Cre recombinase expression via the DCX promoter driving CreER^T2 expression. This triggers the removal of the STOP codon upstream of the tdTomato gene, allowing fluorescent tdTomato expression in newborn neurons in the dentate gyrus. These newborn neurons continue to express the fluorescent protein beyond the period when they initially expressed DCX, allowing fate‐mapping of immature neurons from discrete developmental windows (Cheng et al. 2011; Villasana et al. 2015).

2.3.1. Experiment 1 (IHC)

To induce sporadic labeling of immature neurons for assessment of morphometric determinations of mature neuronal architecture, 60 mg/kg tamoxifen (Sigma Aldrich, #50‐165‐6821) in corn oil was administered intraperitoneally to both the CON and AIE groups on PND 47. This specific injection day was chosen because doublecortin is transiently expressed for the first few weeks after cell proliferation (Plümpe et al. 2006). Therefore, any labeling induced by tamoxifen on PND 47 should capture cells that were either newly born during adolescence or immature during adolescence. Animals were additionally administered 50 mg/kg/day 5‐ethynyl‐2′‐deoxyuridine (EdU) intraperitoneally (Sigma Aldrich, 900584‐50MG), which was prepared in sterile water to label proliferating cells throughout the AIE paradigm (PND 29–58).

2.3.2. Experiment 2 (Electrophysiology)

To achieve more robust labeling of adolescent immature neurons for electrophysiological experiments, 60 mg/kg tamoxifen (Sigma Aldrich, #50‐165‐6821) was administered intraperitoneally to both the CON and AIE groups on PND 43–47, with the first tamoxifen injection beginning the day following the eighth gavage session of either ethanol or water. For all experiments, tamoxifen was prepared fresh weekly in corn oil and stored at 4°C.

2.4. Euthanasia for IHC (Experiment 1) and Electrophysiology (Experiment 2)

For all experiments, assessments were performed after a multi‐month ethanol‐free period as the majority of morphometric and physiological properties plateau and stabilize between 4 and 6 weeks of cellular age (for review see Toni and Schinder 2016).

2.4.1. Experiment 1

All mice were deeply anesthetized with isoflurane gas and then transcardially perfused with 0.1 M phosphate‐buffered saline (PBS) for 2 min followed by 4% paraformaldehyde/PBS (PFA). Brains were harvested and stored in PFA for 24 h, after which they were transferred to a 30% sucrose/0.1 M PBS solution. Brains were cut into 100 μm sections using a Thermo Scientific HM450 sliding microtome and then stored in cryoprotectant at −20°C until further assessment.

2.4.2. Experiment 2

All mice were allowed to mature to at least PND 100, after which they were anesthetized with 100 mg/kg ketamine and 10 mg/kg xylazine, followed by transcardial perfusion with ice‐cold, 95% O₂/5% CO₂‐gassed HEPES‐buffered low‐sodium (NMDG) artificial cerebrospinal fluid solution: 92 mM N‐methyl‐D‐glucamine (NMDG), 20 mM HEPES, 30 mM NaHCO₃, 25 mM D‐glucose, 5 mM N‐acetyl‐L‐cysteine, 5 mM ascorbic acid, 3 mM sodium pyruvate, 2.5 mM KCl, 2 mM thiourea, 1.2 mM KH²PO₄, 10 mM MgSO₄, and 0.5 mM CaCl₂, adjusted to pH 7.4 with HCl, for whole‐cell patch‐clamp slice electrophysiology. The whole brain was removed and blocked onto 4% agar blocks on the slicing chuck of a Campden Instruments 7000smz‐2 vibrating microtome (Campden Instruments, Lafayette, IN). Coronal slices containing the hippocampal dorsal dentate gyrus were sliced (250 μm) in the same HEPES‐buffered NMDG solution and then recovered for 10 min at 35°C in the same solution. Slices were then transferred to a HEPES‐buffered saline for at least 1 h at room temperature: 92 mM NaCl, 20 mM HEPES, 30 mM NaHCO₃, 25 mM D‐glucose, 5 mM N‐acetyl‐L‐cysteine, 5 mM ascorbic acid, 3 mM sodium pyruvate, 2.5 mM KCl, 2 mM thiourea, 1.2 mM KH2PO₄, 2 mM MgSO₄, 2 mM CaCl₂, adjusted to pH 7.4 with NaOH.

2.5. Immunohistochemistry (Experiment 1)

Six dorsal hippocampal sections were selected from each animal with a minimum of a 1.2 mm intersection interval and representative sections spanning the dorsal hippocampus. Brain sections were washed 5 × 5 min in PBS followed by antigen retrieval in Citrabuffer (BioGenex, HK086‐9K) for a 60‐min incubation at 70°C. After another 3 × 5‐min wash in PBS, sections were incubated with one of the following primary antibodies: 1:500 rabbit anti‐Iba1 (Wako, 019‐19741), 1:200 mouse anti‐VAChT (Millipore, ABN100), or 1:1000 rabbit anti‐DSRed (TaKaRa, 632496). They remained in the primary antibody solution for 48–72 h. Afterward, the tissue was washed three times with PBS. The sections were then incubated with 1:1000 secondary antibody Alexa Fluor donkey anti‐mouse 488 (#A32787) or Alexa Fluor donkey anti‐rabbit 594 (R37119) for 90 min in the dark. After washing with PBS 5 × 5‐min, the sections were mounted and coverslipped with ProLong Diamond Antifade mounting medium. A Click‐iT Plus EdU Cell Proliferation Kit for Imaging, Alexa Fluor 488 (#C10637) assay was also performed according to the manufacturer's instructions to validate colocalization with tdTomato+ cells (Figure S1).

2.6. Unbiased Stereological Quantification of tdTomato+ Cell Number, Dendritic Arbor Width, Migratory Distance, Axonal Varicosities, Iba1+ Territory, and Iba1+ Cell Number (Experiment 1)

For analysis of tdTomato+ cell number, dendritic arbor width, axonal varicosities, microglial (Iba1+) territory and cell number, and EdU labeling, images were taken using a Nikon Eclipse Ni‐E Upright Microscope System. All analyses were performed by a blinded observer and quantified using a modified version of unbiased stereological analysis (as described in Macht et al. (2023)).

2.6.1. tdTomato+ Neuron Histological Analysis

To quantify the number of adolescent‐immature neurons that survive AIE, visible tdTomato+ soma within the granule cell layer were counted on the superior and inferior blades of the dentate gyrus from four different sections of the dorsal hippocampus per animal. Dendritic width was measured as the linear distance from the two most distal dendritic branch points with similar lengths. Only cells with clearly discernible dendritic structures from neighboring cells were analyzed. Apical dendritic arbor widths were measured across an average of 6 cells per hippocampal section with 4 representative sections per animal; each data point therefore represents an average of 24 cells per animal. For migratory distance, the shortest distance of the labeled tdTomato+ neuron to the subgranular zone was measured; this distance was assessed for every visible tdTomato+ soma within each section with 4 sections analyzed per animal. Radial migratory distances were averaged across all sections, resulting in a single radial migratory metric for each animal.

Granule neurons are known for their en passant varicosities—that is, engorged bead‐like structures that run along their axons (see Figure S2). Axonal varicosities can form synaptic boutons with GABAergic interneurons and excitatory mossy cells in the polymorphic layer (also known as the hilus) (Acsády et al. 1998), although in some pathological instances they can also be indicative of axonal injury (Sun et al. 2022). Number of putative varicosities in the polymorphic (hilar) and CA3 regions of the hippocampus was assessed using Nikon NIS‐Elements software. Quantification focused on enlarged puncta or swellings that were visibly associated with tdTomato+ axonal processes within the expected mossy fiber‐associated trajectory. Thresholding parameters were optimized per image by a blind observer to isolate larger varicosity‐like structures while minimizing inclusion of thinner axonal segments and diffuse background fluorescence. Varicosity counts were normalized to regional area and the number of tdTomato+ labeled cells within each section. Because the transgenic strategy utilized in the present study also appears to label a subset of off‐target neurons in CA3, these measurements should be interpreted conservatively as tdTomato+ axonal varicosity labeling rather than definitive mossy fiber bouton quantification.

2.6.2. Iba1+ Microglial Analysis

To assess microglial (Iba1+) cell numbers, territory, and fluorescent overlap with tdTomato+ granule neurons, hippocampal regions were divided into the molecular layer, which includes the dendrites of granule neurons, the granule cell layer, which is predominantly comprised of granule cell soma, and the polymorphic layer, which includes the axonal varicosities of granule cells. These layers were each assessed for Iba1+ immunoreactivity in order to systematically examine microglial cell number and territory across the various hippocampal regions that granule cells span. As microglia play critical roles in synaptic sculpting, we further assessed microglial proximity to the dendrites of granule neurons that were undergoing critical maturation during adolescence by quantifying the territory occupied by both Iba1+ and tdTomato+ fluorescence. This was defined as an overlay of Iba1+ immunofluorescence with tdTomato+ immunofluorescence in given photomicrographs and termed “proximity” in graphs. Microglial‐granule neuron proximity was corrected for variations in tdTomato+ territory within each section to ensure that changes in dendritic contact were not driven by gross changes in tdTomato+ cell number and the respective territory that the branches covered.

2.7. Confocal Microscopy for Dendritic Complexity Index and Sholl Analysis for Adolescent Immature Neuron Morphometric Assessment (Experiment 1)

3D images of tdTomato+ labeled neurons were captured at 20× magnification using the Zeiss LSM 980 Confocal Microscope at the UNC Neuroscience Microscopy Core. Four neurons from the granule cell layer of the dorsal dentate gyrus superior blade were selected for Sholl analysis with Imaris 10.0.1. Any neurons that were sheared due to insufficient section thickness were not included in the study. Imaris version 10.0.1 was utilized with the AI filament tracer to track dendritic structure. Sholl analysis was run on each neuron using Sholl spheres with 10 μm incremental increases in radii (Figure S2), with the origin point of the spheres beginning at the edge of the soma to control for variable soma width. Each data point represents an average of all four cells per animal, resulting in one data point per animal.

A global dendritic complexity index (DCI) was also quantified for reconstructed neurons (Reagan et al. 2021). This global index of cellular complexity takes into account the following measurements for each neuron: total dendritic length (μm), number of primary dendrites (i.e., dendrites emerging directly from the cell soma), number of terminal tips, and branch order for each terminal tip. The formula is as follows: DCI = (Σ branch tip orders + number of branch tips) × (total dendritic length/total number of primary dendrites). As with the Sholl analysis, DCI was extracted from 4 cells per animal with a single final datapoint per animal.

2.8. Microglial Confocal Imaging and Imaris Volumetric Assessments (Experiment 1)

Microglia that exhibit a classically reactive phenotype retract their branches and adopt a morphology characterized by decreased complexity and increased sphericity (Green and Rowe 2024). To assess whether AIE produces more spherical/reactive microglial morphologies in microglia that are intertwined with the dendritic trees of newborn neurons that survived AIE, we performed a sphericity assessment on microglia using Imaris volumetric analyses. For volumetric assessments of microglia, imaging of microglial cells specifically intertwined with tdTomato+ neurons in the dorsal dentate gyrus of the mouse tissue, a Zeiss LSM 980 confocal microscope was utilized with a 63× oil‐immersion objective. Two channels were used in the acquisition of these images. The imaging was performed with the Airyscan MLPX TRK line switch enabled to optimize multiple signal detection. A Z‐stack acquisition was employed to capture the three‐dimensional images of the tissue, and tile scanning was utilized as needed, depending on image size. Post‐acquisition, the images were processed using the Airyscan software to apply background subtraction and enhance signal. Five individual microglial images along with one full neuronal image were taken per mouse.

Volumetric analysis of the microglial cells was performed using Imaris 10.2.0 software (Figure S3). Raw confocal images acquired from the Zeiss LSM 980 were initially converted from CZI format to the Imaris‐compatible format (IMS) using the Imaris File Converter software. The files were then loaded into Imaris for further analysis. The “Surface” function within the Imaris software was used to generate 3D reconstructions of individual microglial cells. Surface area‐based volumetric and sphericity data were collected for each image generated, with a total of five microglia assessed for each mouse. The reconstruction was based on the intensity threshold of the images to allow for comprehensive analysis of cell morphology.

2.9. Analysis of Dendritic Spine Density on Adolescent‐Maturing Neurons (Experiment 1)

Granule neurons exhibit robust spines along their dendritic structures where they receive most of their synaptic inputs, making spine density a critical index of granule neuron connectivity (Zhao 2006). To assess whether AIE adversely impacts the spine density in adulthood in neurons undergoing critical periods of development across adolescence, we performed spine analysis and quantification using Neurolucida 360 software. Four images per animal were used, and the analysis was completed by an experimenter who was blinded to all conditions. Each image captured one specific dendritic region 125–150 μm away from the soma. Sholl concentric circles were applied to the image to determine the distance of the dendritic segments to ensure consistent regional analysis across all samples. Neurolucida's tracing tools were used to manually trace the segment of the dendritic branch to allow for 3D reconstruction of the neurons. The software's specific spine analysis function marked the individual spines along the segment. The spine count and length of the dendritic segment were measured to standardize the number of spines per micron of dendrite to provide a normalized measure of spine density (Figure S4).

2.10. Colocalization of VAChT Onto the Dendritic Structure of Surviving Adolescent‐Maturing Granule Neurons (Experiment 1)

Vesicular acetylcholine transporter (VAChT) is a protein integral to cholinergic neurotransmission, responsible for packaging acetylcholine into synaptic vesicles for release into the synaptic cleft, and loss of cholinergic signaling can have profound consequences on survivability of newborn neurons (Cooper‐Kuhn et al. 2004). Moreover, prior studies find that VAChT+ immunoreactivity is reduced on newborn (doublecortin+) neurons in adulthood after AIE (Macht et al. 2023). However, the spatial distribution of VAChT on dendritic arbors of surviving granule neurons is unknown. In this study, we investigated the colocalization of VAChT expression with tdTomato‐positive cells to determine whether neurons that survive AIE also exhibit loss of cholinergic connectivity.

High‐resolution imaging of immunolabeled brain sections was performed using a Zeiss LSM 980 confocal microscope with optimized settings for detecting tdTomato and VAChT fluorescence. Z‐stack images were acquired using a 40× objective to ensure accurate 3D reconstruction of cellular structures. The resulting images were analyzed using Imaris software version 10.2. TdTomato‐positive regions were first reconstructed into 3D surfaces using the Imaris “Surface” function, with thresholding adjusted to define the tdTomato‐expressing cellular structures accurately. Following this, VAChT‐positive regions were segmented as separate volumetric structures using a similar process. The spatial relationship between VAChT and tdTomato‐positive surfaces was assessed through proximity analysis using the Imaris “Distance to Surface” tool. Distances were categorized into three bins: 0, 0–0.25, and 0.25–0.50 μm. This approach enabled precise analysis of VAChT distribution relative to tdTomato‐labeled neuronal populations.

2.11. Whole Cell Patch‐Clamp Slice Electrophysiology (Experiment 2)

When ready for recordings, slices were transferred to a heated glass‐bottom chamber on a Zeiss FS‐2 microscope and viewed using TRITC‐filtered illumination. Slices were superfused with aCSF recording solution: 134 mM NaCl, 3 mM KCl, 2 mM CaCl₂, 1.3 mM MgSO₄ 10 mM D‐glucose, 3 mM myo‐inositol, 1.25 mM KH₂PO₄, 2 mM sodium pyruvate, 0.4 mM sodium ascorbate, and 25 mM NaHCO₃, pH 7.4, gassed with 95% O₂/5% CO₂ at 34°C. Action potentials were recorded from tdTomato+ labeled granule neurons in whole‐cell current clamp with an ~6‐9 MΩ recording pipette (Sutter Borosilicate Glass, #BF150‐86‐10) with internal solution: 130 mM K‐gluconate, 4 mM NaCl, 0.2 mM EGTA, 10 mM HEPES, 4 mM Mg₂ATP, 0.3 mM Tris‐GTP, and 10 mM Tris‐phosphocreatine, pH adjusted to 7.2 with KOH. A small amount of TMR Biocytin was added to the internal solution to aid visibility. After data collection was complete, cells were ruptured with positive pressure through the recording pipette. Patched cells were verified as tdTomato+ by observing the rapid loss in fluorescence following cell rupture. Any cells where red fluorescence was not rapidly lost following cell rupture were deemed non‐tdTomato+ and excluded from analysis. Patching was performed at 40× magnification. All recordings were made with a MultiClamp 700B amplifier (Molecular Devices), digitized with a National Instruments NI‐6251 16‐bit A‐D card at 400 kHz and downsampled to 100 kHz to reduce high‐frequency noise. Data acquisition was controlled by open‐source ACQ4 (available at www.acq4.org). For each cell, responses to current injections (1 s duration) with 10 pA escalating square current steps (range: −200 or −20 to +200 pA) were recorded, along with the resting membrane potential. Spike threshold was determined from the voltage at which the rising slope of the first evoked spike at the threshold current exceeded 20 mV/ms. Afterhyperpolarization was measured as the nadir of the voltage following this spike, relative to the spike threshold. Input resistance was measured as the maximum slope of the steady‐state current–voltage relationship for negative current steps below rest, as extracted from a cubic polynomial fit to the current–voltage relationship. Measurements were performed with the analysis package “ephys” (www.github.com/pbmanis/ephys). Recordings for analysis included 45 cells across 16 mice, with an average of 2.8 cells recorded per mouse.

2.12. Statistical Analysis

2.12.1. Quantification of Histological Endpoints

The number of tdTomato+ cells, surviving adolescent‐immature (tdTomato+) neuronal territory, dendritic complexity index, dendritic arbor width, migratory distance, varicosity counts, dendritic spine density, EdU counts, microglial territory and cell number, and microglial sphericity were analyzed using GraphPad Prism 10 and SPSS with a two‐tailed t‐test. Sholl analysis data, VAChT colocalization, and action potential frequency (Hz) were analyzed using repeated measures ANOVA. Simple main effects follow‐up analyses were run with Bonferroni corrections when relevant. For all assessments, missing data were determined to be at random and not imputed. For all statistical measurements α = 0.05.

2.12.2. Quantification of Electrophysiological Endpoints

For electrophysiology, the resting membrane potential, membrane time constant (τ), and input resistance were all analyzed using an ANCOVA with the animal's age at assessment used as a covariate in addition to the independent variable of group (CON, AIE). Resting membrane potential was corrected for an estimated electrode junction potential of −12 mV. While age of assessment did not vary across the groups, this allowed for any potential variations in physiological cellular parameters due to animal age to be controlled for. The mean age at assessment was P137 ± 33.8 (mean ± standard deviation).

The frequency‐current (F‐I) curve was analyzed using a mixed multilevel generalized linear regression model with the nested variables of animal ID and cell ID; repeated current injections were run as a within‐subjects comparison, and the dependent target variable was identified as the number of action potentials with a Poisson distribution and log‐link function. This distribution profile was selected to account for the “count” aspect of the number of action potentials in any given run and the non‐normal distribution of residuals. Data were linearly transformed as the rounded integer of the mean number of action potentials across cumulative runs plus one in order to avoid violations of zero or non‐integer data. Fixed effects that were assessed included a dummy code of group (CON = 0, AIE = 1), age, and both the linear and curvilinear relationship with pA of current, identified as current and current squared. Random effects included the intercept of animal ID. For the mixed model, degrees of freedom were calculated using the Welch‐Satterthwaite method to account for inequality of variances. For all assessments, missing data were determined to be at random and not imputed. For all statistical measurements α = 0.05.

3. Results

3.1. AIE Reduces tdTomato+ Labeling in Adolescent‐Immature Neurons (Experiment 1)

Tamoxifen (PND 48) treatment of control mice, followed by 50 days of maturation into adulthood, resulted in tdTomato+ mature progenitors, some with soma deep in the dentate gyrus, extensive dendritic arborization, and axonal varicosities. The number of tdTomato+ cell somata did not significantly differ between CON and AIE mice, t(22) = 1.82, p = 0.08 (Figure 2A). However, when assessed for total tdTomato+ immunoreactive area (% area) across the granule cell layer, the molecular layer, and the polymorphic layer of the dorsal dentate gyrus, which includes the soma, dendritic arbors, and axons, respectively, AIE‐treated adult mice exhibited a significant decrease in immunoreactivity across all regions, t(22) = 2.077, p = 0.049; t(22) = 2.139, p = 0.044; t(21) = 2.118, p = 0.046 (Figure 2D). AIE also did not impact EdU+ labeling of adolescent progenitors that survived into adulthood, t(20) = 1.146, p = 0.27, (CON = 1006 ± 79.6; AIE = 915 ± 74.8) (Figure 2B). Because EdU labeling captures proliferating cells broadly and is not restricted to the tdTomato‐labeled neuronal populations, these measures may reflect aspects of both hippocampal neurogenesis and gliogenesis. Together, these findings indicate that AIE persistently reduces immunoreactive area of tdTomato+ labeling across multiple anatomical compartments, despite no significant differences in the number of tdTomato+ cell soma under the present experimental conditions.

FIGURE 2.

AIE reduces tdTomato+ expression in adulthood but does not significantly impact tdTomato+ cell soma number. (A) Mice exhibit a nonsignificant (41%) decrease in the number of surviving adolescent‐immature (tdTomato+ cells) after AIE in the dentate gyrus in adulthood, t(22) = 1.82, p = 0.08. (B) EdU labeling was not significantly impacted by AIE, t(20) = 1.15, p = 0.27. (C) Representative diagram of the three regions analyzed for tdTomato+ territory coverage: Granule cell layer (red), molecular layer (yellow), polymorphic layer (pink). While granule cell bodies are localized to the granule cell layer of the hippocampus, dendrites from granule cells exhibit diffuse projections throughout the molecular layer. In addition, the axons of granule cells extend along the mossy fiber pathway through the polymorphic layer. (D) AIE significantly reduced the territory covered by adolescent‐immature (tdTomato+) granule cells in adulthood relative to controls across the granule cell, t(22) = 2.077, p = 0.049, molecular, t(22) = 2.139, p = 0.044, and polymorphic layer of the dentate gyrus, t(21) = 2.118, p = 0.046. (E) Representative hippocampal photomicrographs of a CON (top) and AIE (bottom) mouse. Note that tdTomato+ labeling robustly expresses across the cellular soma, dendritic arbors, and axonal varicosities. Graphs represent the mean ± SEM.

3.2. Adolescent‐Immature (tdTomato+) Neurons Exhibited Increased Dendritic Width and Migratory Distance After AIE in Adulthood Relative to Controls (Experiment 1)

AIE did not impact the overarching dendritic complexity index of surviving newborn neurons, t(21) = 0.92, p = 0.37 (Figure 3A). However, some subtle architectural differences emerged with Sholl analysis: AIE mice exhibited a shift in the number of branch points with increasing distance from the edge of the soma, F(28, 616) = 1.38, p = 0.09. These effects were most evident at the medial and distal neuronal branch points where AIE produced subtle but spatially specific alterations in dendritic complexity—increasing dendritic complexity at intermediate (130 and 140 μm) distances but decreasing complexity at more distal locations (240 μm) from the soma, p's = 0.05, 0.04, and 0.02 (Figure S2). More dramatically, surviving adolescent‐immature neurons from AIE‐exposed mice exhibited a dendritic width an average of 15 μm (10%) wider than that of controls, t(22) = 3.121, p = 0.005 (Figure 3B,E), and an increase in the radial migratory distance from the subgranular zone, t(20) = 3.114, p = 0.005 (Figure 3C,D). This suggests that surviving newborn granule neurons cover greater territory in the dorsal dentate gyrus and are located further from the subgranular zone.

FIGURE 3.

Dendritic arbors of surviving adolescent‐immature neurons cover more territory and exhibit increased radial migratory distance. (A) AIE does not impact the overall dendritic complexity index of surviving newborn neurons, p > 0.05. (B) Neurons that were undergoing early maturation during AIE but survive into adulthood exhibit an approximate 10% increase in dendritic arbor width, suggesting that while fewer neurons survive AIE, the dendritic branches of those neurons cover more territory, t(22) = 3.12, p = 0.005. (C) As newborn granule neurons mature, they migrate away from the subgranular neurogenic zone through the granular cell layer, known as the radial migratory distance. (D) AIE significantly increased the mean radial migratory distance of adolescent‐immature neurons in adulthood, t(20) = 3.114, p = 0.005. (E) Example photomicrographs of adolescent‐immature neurons' dendritic arbors in adulthood. The left two images are examples of control neurons. The right two images are examples of neurons from AIE‐treated mice. Graphs represent the mean ± SEM.

3.3. Adolescent‐Immature (tdTomato+) Neurons Exhibited Decreased Spine Density and Increased Frequency of tdTomato+ Varicosities in AIE Relative to Controls in Adulthood (Experiment 1)

Immature neurons extend their dendritic arbors, increase axonal innervation, and exhibit maturation of spine density across approximately 8 weeks post‐birth. Neurons that were immature during AIE were tracked into maturity in adulthood, where polymorphic axonal varicosities and molecular spine density were assessed (Figure 4A). The number of spines from 125 to 150 μm was quantified and standardized to the length of the individual segment across that distance from the soma. AIE exposure significantly decreased spine density across this region, t(22) = 2.810, p = 0.010 (Figure 4B,C). In the polymorphic layer, tdTomato+ axonal varicosities approximately doubled when normalized to tdTomato+ cell numbers, t(22) = 2.923, p = 0.008 (Figure 4D). Conversely, tdTomato+ varicosities in CA3 were not impacted by adolescent exposure when normalized to tdTomato+ cell somata, t(22) = 1.726, p = 0.098. Together, these findings indicate that AIE impacts the structural features of neurons that were immature during adolescence, including reducing dendritic spine density and increasing tdTomato+ axonal varicosity labeling in some but not all mossy fiber‐associated regions.

FIGURE 4.

AIE decreases surviving adolescent newborn neuron spine density and increases tdTomato+ axonal varicosities in the polymorphic layer of the dentate but not CA3. (A) The dendrites of granule neurons exhibit dense synaptic spines, reflecting their innevation from various regions. In addition, the axons of granule neurons exhibit substantial varicosities, also known as en passant synapses, which predominantly form synapses on inhibitory interneurons in the polymorphic layer of the dentate gyrus (PoDG) and in CA3. (B) Spines were assessed along a section of the intermediate portion of the neuron, approximately 135 μm from the cell soma. Results indicate that AIE caused a significant reduction in spine density on surviving adolescent‐immature (tdTomato+) neurons in adulthood relative to controls, t(22) = 2.810, p = 0.0102. (C) Example photomicrographs (top) of CON (left) and AIE (right) spines. Bottom: 3D rendering of spines from Imaris Software. (D) Number of varicosities labeled within a section were normalized to the number of tdTomato+ cell soma across both polymorphic (PoDG) and (E) CA3 regions, resulting in an estimate of the number of varicosities per tdTomato+ labeled soma. Results indicate that AIE increased the number of axonal varicosities in the polymorphic layer of surviving granule cells, t(22) = 2.923, p = 0.0079; tdTomato+ varicosities were not increased in CA3 t(22) = 1.726, p = 0.098. (F) Photomicrographs of example CON and AIE varicosities in the polymorphic layer of the dentate gyrus, taken at 100× magnification. Varicosities represent the tdTomato+ swellings along the axon shaft. Graphs represent the mean ± SEM.

3.4. AIE Reduces Territory Covered by Iba1+ Microglia in the Dorsal Hippocampus in Adulthood but Increases Contact Onto Surviving Adolescent‐Immature (tdTomato+) Neurons (Experiment 1)

AIE causes persistent induction of innate immune signaling cascades in the hippocampus that can adversely impact hippocampal neurogenesis (for review see Macht, Elchert, and Crews 2020). Not only are microglia the primary effectors of the innate immune response, but they are also involved in neurocircuitry maturation and likely contribute to synaptic pruning of neuroprogenitors (Paolicelli et al. 2011). Therefore, microglial territory (Iba1+ immunoreactivity) was assessed across the granule cell layer, molecular layer, and polymorphic layer of the dorsal dentate gyrus (Figure 5A). Despite no change in Iba1+ cell number in any layer, molecular: t(22) = 0.7827, p = 0.4421, polymorphic: t(22) = 1.642, p = 0.1148, granule cell: t(22) = 0.830, p = 0.4149, there was a significant reduction in Iba1 + immunoreactivity in the molecular and polymorphic but not the granule cell layer, t(22) = 2.517, p = 0.0196, t(22) = 2.932, p = 0.0077, t(22) = 1.251, p = 0.2241, respectively. One possibility is that microglia have shifted to a reactive phenotype. This has historically been associated with a retraction of ramifications and an increase in microglial sphericity (Bosch and Kierdorf 2022). Therefore, we sampled sphericity from microglia specifically intertwined within the dendritic arbors of adolescent‐immature (tdTomato+) neurons and found that AIE significantly increased microglial sphericity relative to controls, t(22) = 2.562, p = 0.0178 (Figure 5B,C). Of note, this increase in sphericity was independent of any overarching changes in microglial cell volume, suggesting a change in the shape but not the size of the microglia associated with tdTomato+ dendrites, t(22) = 1.060, p = 0.3007.

FIGURE 5.

AIE‐microglia exhibit more reactive phenotypes in adulthood than controls coupled with increased relative contact onto dendritic spines. (A) Iba1+ cells cover less territory across the molecular and polymorphic layers in the absence of any changes in cell number, suggesting morphological shifts in cellular territory (t(22) = 2.517, p = 0.0196, t(22) = 2.932, p = 0.0077, t(22) = 1.251, p = 0.2241). (B) Microglia entwined in the dendritic arbors of adolescent‐immature (tdTomato+) neurons exhibit increased sphericity after AIE relative to controls (t(22) = 2.562, p = 0.0178). (C) Example volumetric assessments of microglia from controls (CON) and AIE‐treated animals. Note that microglia from controls (gray) exhibited greater ramification of processes whereas microglia from AIE‐treated mice (red) exhibit a more reactive morphological phenotype. Graphs represent the mean ± SEM. Scale bars for AIE and CON microglia are each set to 40 μm.

Microglia also have critical roles in synaptic pruning (Paolicelli et al. 2011); as such, changes in spine density may be related to shifts in microglial function. Therefore, we assessed the average amount of Iba1+ microglia proximal to tdTomato+ labeling with assessments performed across the granule cell layer, molecular layer, and polymorphic layer of the dorsal dentate gyrus. Results indicated that AIE increased microglial contact onto adolescent‐immature (tdTomato+) neurons across the granule cell layer, t(22) = 3.550, p = 0.0018 and molecular layer, t(20) = 4.091, p = 0.006, but not the polymorphic layer, t(21) = 1.779, p = 0.0898.

3.5. Adolescent‐Immature Granule Neurons (tdTomato+) That Survive AIE Do Not Exhibit Any Changes in Cholinergic Innervation Relative to Controls (Experiment 1)

Acetylcholine inhibits microglial proinflammatory activation and promotes neurogenesis, prompting a determination of cholinergic innervation. Therefore, we determined whether AIE disrupted hippocampal dentate cholinergic innervation (VAChT) of surviving adolescent‐immature neurons (tdTomato+) in adulthood (Figure 6A,B). Interestingly, in both controls and AIE, cholinergic innervation was greater with closer proximity to the cellular architecture (i.e., dendritic arbors and soma) and decreased as distance increased from surviving granule neurons (tdTomato+), F(2, 44) = 73.937, p < 0.001 (Figure 6C). In fact, both control and AIE‐exposed mice exhibited equivalent reductions in cholinergic innervation with each 0.25 μm increase in distance from the surviving neuron, p's < 0.001. Surprisingly, AIE did not alter VAChT/tdTomato proximal colocalization in adulthood.

FIGURE 6.

Surviving adolescent granule neurons exhibit a high degree of colocalization with cholinergic terminals which attenuates with increasing distance from dendritic branches. (A) Confocal image of adolescent‐immature (tdTomato+) neuron in adulthood (red) with VAChT+ immunoreactivity (green). (B) Imaris rendering of VAChT puncta with increasing disance from adolescent immature (tdTomato+) neurons. Puncta were separated into separate categories. Complete colocalization (0 μm from tdTomato+ labeling, blue), and then less than 0.25 μm distance (fuscia), and 0.25–0.5 μm distance (turquoise). Puncta between 0.5 and 1 μm from neurons are identified here in red but were excluded from analysis. (C) A significant proportion of VAChT cholinergic puncta colocalized with surviving adolescent‐immature (tdTomato+) neurons, and as distance from surviving neuron's cellular architecture increased, cholinergic innervation decreased for each 0.25 μm step, F(2, 44) = 73.937, p < 0.0001, p's < 0.001.

3.6. Surviving Adolescent‐Immature Neurons in AIE‐Treated Mice Exhibit Shifts in Firing Capacity With Increasing Current Injections Later in Adulthood AIE Relative to Controls (Experiment 2)

To assess whether AIE causes changes in the physiological parameters of surviving adolescent‐immature neurons, electrophysiology was performed in adulthood on surviving granule neurons (Figure 7A). While age of the animal at euthanasia was controlled for across each assessment, these values were randomly distributed and there were no group differences between AIE and CON on age of assessment, t(35) = −0.432, p = 0.67. There was a trend for labeled granule neurons from AIE animals to exhibit slightly depolarized resting membrane potentials relative to controls, F(1, 32) = 3.67, p = 0.065. The mean resting membrane potential from labeled granule neurons was −71.93 ± 1.21 mV (AIE) and −75.83 ± 1.80 mV (CON). In addition, results indicated that AIE caused a significant shift in the membrane time constant (τ), F(1, 33) = 5.14, p = 0.03 with AIE‐treated cells (14.56 ms) exhibiting a higher τ than CON‐treated cells (10.43 ms), suggesting that surviving AIE‐adolescent‐immature (labeled) neurons may be slower to return to baseline following stimulation than labeled granule neurons from controls. One τ value in cells from CON mice was a statistical outlier but was retained here for transparency; neither its inclusion nor exclusion impacted significance of results. Input resistance was not significantly different between groups, F(1, 24) = 2.56, p = 0.12. Collectively, these results indicate that AIE causes adolescent‐immature granule neurons to exhibit a shift in passive membrane properties when they reach their mature state (Figure 7B).

FIGURE 7.

Surviving granule neurons have altered passive and active physiological properties after AIE. (A) Example photomicrograph of tdTomato+ neuron with recording electrode. (B) Passive membrane properties from recorded cells: Resting membrane potential, the membrane time constant (τ), and input resistance (left to right). AIE significantly increased τ, p = 0.03. (C) Active membrane properties of neurons, including the threshold for action potential generation and the relative change in afterhyperpolarization. (D) There was a significant interaction between escalating current injection and AIE treatment on induced action potentials in tdTomato+ labeled granule neurons, p < 0.001. Granule neurons from AIE‐exposed mice exhibited a leftward shift in firing capacity with increases in action potentials generated at lower currents but reductions in action potential frequency at high rates of stimulation. (E) Example traces with a 200 pA current stimulation from tdTomato+ neurons in adulthood from control (left) and AIE (right) mice. Data are reported as mean ± SEM.

With escalating current stimulation, labeled granule neurons from AIE‐treated animals also exhibited changes in physiological responsivity. In particular, labeled granule neurons from AIE‐treated animals had greater relative afterhyperpolarization compared to controls, F(1, 27) = 11.56, p = 0.002 (Figure 7C). When the age of the animal was controlled for, there was a trend for AIE‐exposed animals to exhibit a more depolarized action potential threshold (−50.4 ± 1.2 mV) relative to controls (−54.6 ± 2.4 mV), F(1, 27) = 3.25, p = 0.08. Notably, older animals did exhibit a significant decrease in the action potential threshold relative to younger animals, F(1, 27) = 5.078, p = 0.03. Collectively, these parameters suggest that AIE impacts the dynamic responsive properties of mature granule neurons that were immature during adolescent ethanol exposure, which could influence patterns of action potential generation (Figure 7C).

In support of these observations, a mixed multilevel generalized linear model was used to assess the linear and curvilinear relationship between group and current on the number of action potentials generated per second (Hz) in labeled granule neurons. Although there were no group differences at any individual current step, there was a significant curvilinear relationship between current and action potential such that the association begins with a significant linear positive association, B = 0.01274, t(372) = 22.66, p < 0.001, that then concaves downwards, B = −0.00016, t(349) = −17.63, p < 0.001 (Figure 7D,E). Furthermore, the linear and curvilinear association between current and action potential rate was qualified by a significant interaction with ethanol exposure, B = 0.00217, t(372) = 2.87, p = 0.004 and B = 0.00005, t(349) = 4.38, p < 0.001, respectively. These results indicate that the strength of the positive relationship was stronger for CON than AIE‐treated mice, indicating that AIE increased action potential rate at lower current injections but cells began to fail to fire sooner than cells from CON‐treated animals. These results highlight that AIE impacts the firing properties of surviving newborn neurons by affecting the scaled relationship between current‐driven stimulation and cell firing capacity. Age did not significantly impact action potential generatation in this model, t(12) = 0.043, p = 0.967. Random variance around the intercept was significant, τ = 0.23, Z = 2.41, p = 0.016, indicating that multilevel modeling was appropriate.

4. Discussion

AIE‐loss of hippocampal neurogenesis persists into adulthood despite sustained ethanol‐free recovery periods, with reductions in doublecortin+ labeling frequently reported in AIE‐treated adults (Broadwater et al. 2014; W. Liu and Crews 2017; Nwachukwu et al. 2022; Reitz et al. 2021; Vetreno et al. 2018; Vore et al. 2021) coupled with increases in cell death markers in doublecortin+ neurons across both males and females (Macht et al. 2023), suggesting AIE produces long‐term disruption of cell survival in the hippocampal neurogenic niche. Despite these consistent findings, many cells survive AIE, whereupon they presumably incorporate into existing neural circuits to influence function. The current study examined the phenotype of this surviving population of newborn neurons following AIE. Of note, this study focused exclusively on females as prior studies find similar reductions in hippocampal neurogenesis in both males and females after AIE (Macht et al. 2023; Nwachukwu et al. 2022; Reitz et al. 2021). We report here that neurons that were immature during adolescence and survive to maturity in adulthood exhibit persistent alterations in mature phenotype after AIE. Morphologically, AIE produces subtle increases in the dendritic complexity at intermediate distances from the cell soma in addition to increasing the width of dendritic arbors by 10% and decreasing spine density in that same region. The number of tdTomato+ polymorphic axonal varicosities per surviving granule neuron was increased after AIE, suggesting persistent changes to both input and output of surviving neurons. Physiologically, newborn neurons that survive AIE also exhibit subtle shifts in firing capacity and membrane properties. AIE‐induced shifts in both dendritic and axonal innervation coupled with alterations in the generation of action potentials may have significant consequences for the function of the hippocampal circuit. We further report that AIE increased microglial sphericity and enhanced overlap with dendritic arbors of adolescent‐maturing neurons, suggesting that altered microglial engagement may contribute to persistent changes in surviving neuronal morphology and connectivity. These studies add to previous findings linking AIE reductions in hippocampal neurogenesis to disruptions in innate immune function and add additional insight into AIE‐disruption of hippocampal circuitry, which may underlie previously observed deficits in adult cognitive‐behavioral function (Crews et al. 2019).

4.1. AIE Disrupts the Adult Architecture of Adolescent‐Developing Hippocampal Granule Neurons

Though some isolated findings suggest that a small subset of newborn neurons in the hippocampus can develop into GABAergic basket cells (Liu et al. 2003), the vast majority of cells generated from adult hippocampal neurogenesis become a single neuronal phenotype: the hippocampal granule neuron (for review see Kempermann et al. 2015). These granule neurons are glutamatergic in nature, densely populate the granule cell layer of the hippocampal formation, and are one of the only cellular populations that exhibit continued proliferation and functional network integration into adulthood (i.e., hippocampal neurogenesis). The architecture of the mature hippocampal granule neuron is critical to hippocampal function (Llorens‐Martín et al. 2016), and both loss of hippocampal granule neurons and disruption of mature granule cell architecture are common features in models of neurological disease or injury such as epilepsy (Murphy et al. 2012), traumatic brain injury (Carlson et al. 2014; Villasana et al. 2015), stroke (Niv et al. 2012), and entorhinal lesion (Perederiy et al. 2013). We report here that in female mice, granule neurons that were immature during adolescence but survive AIE exhibit subtle but long‐lasting changes in their cellular architecture.

Although AIE decreased the mean number of tdTomato+ soma by 41% relative to controls, this reduction was statistically non‐significant. Tamoxifen‐driven doublecortin‐tdTomato labeling in this transgenic model identifies a restricted subpopulation of immature neurons rather than the broader population captured by conventional doublecortin immunolabeling. The resulting variability may have obscured doublecortin loss that has been historically consistent in the AIE field (Broadwater et al. 2014; Macht et al. 2023; Nixon and Crews 2002; Reitz et al. 2021; Vetreno and Crews 2015). Nevertheless, our findings indicate that even outside global indices of progenitor loss, AIE subtly alters the architectural and physiological phenotype of surviving neurons. These surviving neurons from AIE‐treated mice also covered more territory, evidenced by a 10% increase in the breadth of their dendritic arbors without changing overall complexity indices. These results contrast with findings from a month‐long adult liquid ethanol diet in adulthood that reduces the dendritic complexity of DCX+ immature neurons in male rats (He et al. 2005). However, increasing age also decreases complexity of dendritic arborization of granule neurons during immature but not mature developmental states (Trinchero et al. 2017), suggesting that age slows the morphological maturation of adult‐born neurons. While doublecortin labels immature neuroprogenitors, DCX‐CreER^T2/tdTomato+ labeling fated‐mapped immature neurons over months of development and were assessed during timeframes which should result in a mature cellular phenotype. Collectively, these findings suggest that AIE may shift the maturation of newborn granule neurons towards an aging‐like maturational trajectory, where diminished complexity during immature developmental stages is less apparent when neurons reach maturity.

Though one might speculate that increased dendritic span is a compensatory reaction to fewer surviving granule cells, models of traumatic brain injury produce similar increases in dendritic width with the notable contrast of increased neurogenesis (Villasana et al. 2015), suggesting that changes in dendritic width may be independent of the number of surviving cells. Interestingly, this same model of traumatic brain injury produced a similar shift in the granule cell architecture as AIE. Although the global index of dendritic complexity did not differ between AIE and controls, there was some evidence for subtle increases in dendritic complexity at proximal but decreased complexity at distal branch points. Despite this subtlety, the morphological similarity is striking and suggests common aberrations in physiological features in the cellular microenvironment (e.g., proinflammatory signaling factors) may mediate these similar shifts in mature structural domains. One common mechanism worth consideration is innate immune induction of toll‐like receptor 4 (TLR4). AIE increases TLR4 signaling cascades in a variety of brain regions (Vetreno and Crews 2012), including the hippocampus (Vetreno and Crews 2015), as does traumatic brain injury (Jiang et al. 2018). Moreover, Karelina et al. (2017) found that binge ethanol exposure in adulthood exacerbates adolescent brain injury trauma effects on hippocampal neurogenesis by exacerbating TLR4‐proinflammatory signaling cascades (Karelina et al. 2017). TLR4 is expressed on adult neuroprogenitors (Rolls et al. 2007), and TLR4 knockout impairs neurite outgrowth in vitro (Ustinova et al. 2013). Conversely, the TLR4‐agonist lipopolysaccharide decreases arborization of adult‐born granule neurons (Rodríguez‐Moreno et al. 2024; Valero et al. 2014), suggesting that TLR4 signaling may impact maturing cellular architecture in potentially complex ways that need to be further investigated. Furthermore, we find that immature granule neurons that survive AIE exhibit increased radial migratory distance from the subgranular zone—an observation that has also been linked to TLR4‐driven activation of innate immune signaling pathways (Belarbi et al. 2012).

Adolescent‐immature neurons that survive into adulthood exhibit further architectural remodeling, characterized by reduced dendritic spine density and increased axonal varicosities within the polymorphic layer, consistent with potentially altered synaptic input and output. Granule neurons exhibit spines that manifest as a reflection of diverse synaptic input from the entorhinal cortex, hilar mossy cells, commissural dentate projections, CA3 pyramidal back projections, and septo‐hippocampal glutamatergic projections (Amaral et al. 2007; Toni and Schinder 2016), with variations in total spine number reflective not only of experimental differences but also of species differences, strain differences, sex differences, animal and cellular age differences, and location of measurement both within the dendritic architecture and across the subregions of the molecular layer of the hippocampus. We find that AIE reduces the spine density of adolescent‐immature granule neurons once matured into adulthood—an observation that parallels similar findings of reduced spine density on adult hippocampal granule neurons after AIE in male rats (Mulholland et al. 2018). Interestingly, donepezil reversed loss of spine density following AIE (Mulholland et al. 2018), suggesting that cholinergic alterations may underlie this change. Although we did not characterize spine type, Mulholland et al. (2018) determined that this granule neuron loss of spine density after AIE was predominantly the result of a loss of long and mushroom spines. This suggests that AIE‐loss of dendritic spines does not appear to be unique to adolescent‐immature neurons. However, whether granule neurons that were undergoing critical periods of maturation during AIE treatment exhibit a similar plasticity of recovery is unknown. For example, while AIE causes a loss of cholinergic input onto immature granule neurons (Macht et al. 2023), the same does not appear to be true for the surviving neuroprogenitors seen here as we find they do not exhibit differences in cholinergic input relative to controls.

In addition to CA3 targets, hippocampal granule neurons exhibit a robust expression of axonal varicosities which form synaptic associations with GABAergic interneurons, excitatory hilar mossy cells (Acsády et al. 1998), and in some cases can be indicative of neuronal damage or degeneration (Gu 2021; Sun et al. 2022). However, the vast majority of mossy fiber varicosities are thought to be synapses associated with inhibitory interneurons (Acsády et al. 1998); these inhibitory interneurons provide modulatory feedback onto granule cells, contributing to the quiescent nature of these cells by reducing their excitability and preventing excessive firing patterns (for review see Piatti et al. 2013). An increase in axonal varicosities could indicate enhanced inhibitory feedback onto surviving adolescent‐immature granule neurons after AIE in females. However, as axonal varicosities can also reflect axonal damage instead of synapses (Gu 2021); future studies will need to further examine the nature of this change to axonal architecture.

In contrast to our observations in the granule cell layer, AIE results in exaggerated long‐term potentiation in CA1 (Risher et al. 2015; Swartzwelder et al. 2017). As two critical parts of the trisynaptic circuit, AIE‐induced shifts in granule‐cell action potential generation coupled with enhanced CA1 long‐term potentiation could have important consequences for the collective functionality of this circuit in learning and memory‐related deficits after AIE. For example, as granule neurons are often considered the gatekeeper of the hippocampus and the first stop in the trisynaptic circuit, shifts in stimulus‐driven action potential generation in granule neurons could impact the initiation, strength, and specificity of circuit plasticity necessary for behavioral and cellular flexibility. In fact, LPS‐driven hippocampal neuroinflammation reduces adult‐born granule neuron excitability in response to behaviorally‐evoked stimuli, evidenced by reduced expression of the immediate early gene Arc specifically in adult‐born hippocampal granule neurons (Belarbi et al. 2012). This reduction in Arc expression in adult‐born neurons after LPS was positively correlated with activated microglia, further suggesting that AIE‐induction of microglial reactive phenotypes may be an important consideration in adult‐born granule neuron excitability. In support of these physiological observations on hippocampal circuitry, AIE produces protracted adverse behavioral flexibility consequences which also would require a high degree of cellular plasticity in learning and memory circuits, including deficits in reversal learning (Macht, Elchert, and Crews 2020). This loss of behavioral flexibility is thought to be a key characteristic of AIE (for review see Crews et al. 2019) and is a risk factor for later development of substance use disorders. As such, further examination into the relationship of cellular dynamics within the hippocampal circuit after AIE and the relation of these dynamics to behavioral dysfunction is a critical future investigation.

Cholinergic innervation is critical for newborn neuron survival (Mohapel et al. 2005), and IgG saporin‐based cholinergic lesions cause several adverse effects in the adult hippocampus, including loss of adult hippocampal neurogenesis (Mohapel et al. 2005) and increased microglial activation (Dobryakova et al. 2019). Moreover, several studies have reported that integrity of the basal forebrain cholinergic system is disrupted after AIE (Reitz et al. 2021; Vetreno and Crews 2018), and that rescuing deficits in the integrity of the central cholinergic system after AIE reduces proinflammatory cascades, rescues hippocampal neurogenesis in rodents, and recovers AIE‐induced cognitive‐behavioral deficits (Hall and Savage 2016; Macht et al. 2021, 2023; Swartzwelder et al. 2019; Vetreno et al. 2018). We report here that VAChT cholinergic terminals are highly colocalized with the dendritic structures of surviving newborn neurons, with decreasing innervation occurring as distance from dendritic arbors increases. This could suggest that either a high degree of cholinergic innervation favors cell survival, or that surviving newborn neurons actively recruit innervation by cholinergic networks. These observations further support the integral role of cholinergic innervation in granule cell neuron survival, and that the role of cholinergic deficits after AIE in disrupted hippocampal circuitry may exhibit some nuances that need further investigation.

4.2. Hippocampal Microglia Intertwined Within the Dendritic Structure of Surviving Granule Neurons Exhibit Increased Morphological Sphericity After AIE

Microglia, the primary immune effectors in the central nervous system, play key roles in both healthy and aberrant synaptic development (Paolicelli et al. 2011), and AIE‐induced changes in microglial function may underlie some of the structural consequences in hippocampal circuits, including loss of spine density. Therefore, we assessed microglial morphological phenotypes both across the dorsal hippocampal formation and specifically within the dendritic arbors of surviving adolescent‐immature neurons. Interestingly, AIE‐exposed female mice exhibited more spherical microglia when entwined within the dendritic architecture of surviving neurons, indicative of a reactive phenotype. This observation adds to a growing body of evidence that microglial dysfunction is a key feature of both AIE and postmortem AUD. For example, findings from postmortem brains from individuals with AUD suggest long‐term alcohol exposure in humans can lead to microglial dysfunction (Henriques et al. 2018) characterized by an increase in reactive microglial markers in various regions of brain tissue (He and Crews 2008). Emerging findings from preclinical studies also suggest that adolescent binge ethanol exposure disrupts microglial phenotypes (Melbourne et al. 2019). For example, AIE causes hippocampal microglia to exhibit increases in cell body size coupled with thicker processes (McClain et al. 2011), suggesting a shift in microglia to a reactive primed state (Marshall et al. 2016); we see similar alterations in microglial morphology in our findings, reflected via increased microglial sphericity in adult females after AIE. Historically, this change in phenotype has been associated with functional changes in reactivity; however, the functional consequences of these AIE‐driven morphological changes will require further study with additional functional endpoints. Additionally, the AIE model has been shown to produce transient increases in the microglial Iba1+ immunoreactive pixel density during intoxication in the frontal cortex, amygdala, and hippocampus, followed by long‐term reductions in Iba1+ pixel density (Sanchez‐Alavez et al. 2019); for reviews, see (Crews et al. 2016, 2019). While that study did not differentiate pixel density from number of microglial cells, the observation nevertheless parallels findings reported here: we report long‐term reductions in Iba1+ immunoreactivity across various regions of the dorsal hippocampus—particularly both the molecular and polymorphic layers, in the absence of any changes in microglial cell number. This suggests that while the number of microglial cells remains unchanged in the hippocampus after AIE, the morphological phenotype is different. In support of this observation, hippocampal microglia exhibit a dystrophic phenotype in adult males after four‐day binge ethanol exposure during adolescence (Marshall et al. 2020). Our findings extend these observations to indicate increased microglial sphericity following AIE in females, suggesting that AIE may cause long‐lasting changes in microglial pathology, with the hippocampus being particularly vulnerable to lasting effects with potential consequences for hippocampal circuitry and for innate immune signaling. Moreover, our findings further suggest that these changes in microglial phenotype may have distinct consequences for the shaping of neural networks, which requires further investigation.

4.3. Limitations, Conclusions, and Future Directions

This study investigated the long‐term impact of repeated binge ethanol exposure across adolescence on the mature phenotype of hippocampal granule neurons that were immature during adolescence and survived ethanol exposure. Future studies need to link these observations directly to behavioral deficits and establish whether observations on morphological changes in microglia contribute to synaptic changes evidenced in granule neurons. Moreover, while several studies find that loss of adult hippocampal neurogenesis is reversible after AIE, whether similar interventions can recover phenotypic abnormalities in surviving granule neurons that have incorporated into existing circuits remains unknown. Further investigation is also required to assess (1) whether AIE‐driven changes in microglial sphericity correspond with functional deficits, and (2) whether changes in spine density reflect physiological changes in synaptic function, including the generation of excitatory postsynaptic potentials, or changes in network connections within the dentate. In addition, it is unclear whether observed varicosity changes reflect increased varicosity density along individual axons or overall increased axonal innervation, thereby resulting in a greater number of visible varicosities. Although this study focused exclusively on effects in females—and thus extension to males remains an important future area of investigation—it nevertheless provides key insights that even if newborn neurons survive AIE, they exhibit morphological and physiological alterations that could have a significant impact on hippocampal circuitry and function in adulthood, potentially contributing to long‐lasting cognitive‐behavioral deficits.

Author Contributions

Victoria Macht and Fulton Crews: conceptualization. Victoria Macht, Michael Kasten, Paul Manis, Kaitlyn Campbell, Madison McDowell: methodology. Victoria Macht, Kaitlyn Campbell, Madison McDowell, Hunter Kelley, Eymani Alston, Michael Kasten: data curation. Victoria Macht, Paul Manis: formal analysis. Victoria Macht and Kaitlyn Campbell: writing – original draft preparation. Victoria Macht, Kaitlyn Campbell, Fulton Crews, Michael Kasten, Paul Manis, Eymani Alston, Hunter Kelley, Madison McDowell: writing – review and editing. Fulton Crews and Michael Kasten: supervision. Victoria Macht and Fulton Crews: funding acquisition. All authors have read and agreed to the published version of the manuscript.

Funding

This research was supported by grants from the National Institutes of Health (NIH), National Institute on Alcohol Abuse and Alcoholism (NIAAA; 10.13039/100000027): AA030089 (VM), and the Neurobiology of Adolescent Drinking in Adulthood (NADIA) AA020023 (FC) and AA020024 (FC).

Disclosure

AI was not used in the generation of this manuscript in any aspect, except for limited assistance with syntax recommendations provided by Chat GPT. All scientific content, analysis, and interpretation were conducted by the authors.

Ethics Statement

The animal study protocol was approved by the Institutional Review Board of the University of North Carolina at Chapel Hill (#s 23‐123, 23‐148).

Conflicts of Interest

The authors declare no conflicts of interest.

Supporting information

Figure S1: Example of EdU colocalization with adolescent‐immature (tdTomato+) neurons in adulthood. Tdtomato+ neurons, labeled with tamoxifen on PND 47, are in red. EdU (green) labels proliferating cells, and many tdTomato+ neurons exhibited colocalization with EdU labeling (yellow), suggesting these neurons were born during adolescence. Some tdTomato+ neurons did not exhibit colocalization with EdU, suggesting that a subset of neurons was born prior to the adolescent period.

Figure S2: Example rendering of axonal varicosities and Sholl spheres with tdTomato + neuron. (A) Hippocampal granule neurons exhibit bead‐like structures along their axonal processes known as varicosities. Varicosities in the polymorphic layer were selected using a thresholding method by a blind observer and normalized to the number of tdTomato+ neurons. (B) Imaris software created spheres with an expanding 10 μm radius, radiating from the cell soma. A 2D example of this 3D assessment is pictured here on one labeled neuron. Dendritic tracings rendered in yellow. Concentric rings (white) were drawn for illustrative purposes only at 10 μm radii. (C) Although a summary metric of dendritic complexity does not yield significant differences in cellular architecture, a Sholl analysis does reveal some small differences in complexity with increasing radii from the cell's soma. Specifically, AIE increases dendritic complexity at intermediate branch points but generates a loss of dendritic complexity at distal branch points, suggesting shifts in the architecture of surviving adolescent‐immature (tdTomato+) hippocampal granule neurons, p's = 0.05, 0.04, 0.02.

Figure S3: Example of Imaris volumetric assessment of Iba1+ microglia. Volumetric assessment on Iba1+ microglia was specifically formed on cells intertwined within the dendritic arbors of tdTomato+ (adolescent‐maturing) neurons. Iba1 labeling is rendered in green. Imaris 3D rendering is in purple.

Figure S4: Example rendering of dendritic spines for assessment and photomicrograph of axonal varicosities. 120–150 μm from the cell soma was identified (A) and spine density was assessed by a blind observer (B) using Neurolucida 360. (C) Example volumetric fillings of spines are provided from Imaris Software for visualization.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.