Advertisement
Browse Subject Areas
?

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Open Access

Peer-reviewed

Research Article

APOE4 and doxorubicin impair inhibitory interneuron function and homeostatic regulation in the entorhinal cortex

  • Nancy Luo,

    Roles Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Writing – original draft

    Affiliations Interdisciplinary Program in Neuroscience, Georgetown University, Washington D.C., United States of America, Department of Neuroscience, Georgetown University, Washington D.C., District of Columbia, United States of America

    ⨯
  • Harshul Pandit,

    Roles Investigation, Writing – review & editing

    Affiliation Department of Neuroscience, Georgetown University, Washington D.C., District of Columbia, United States of America

    ⨯
  • Shreya Kalra,

    Roles Investigation, Writing – review & editing

    Affiliation Department of Neuroscience, Georgetown University, Washington D.C., District of Columbia, United States of America

    ⨯
  • Erin Tran,

    Roles Investigation

    Affiliation Department of Neuroscience, Georgetown University, Washington D.C., District of Columbia, United States of America

    ⨯
  • Jeanne Mandelblatt,

    Roles Conceptualization, Investigation, Resources, Writing – review & editing

    Affiliation Department of Oncology, Georgetown Lombardi Institute for Cancer and Aging Research (I-CARE), Lombardi Comprehensive Cancer Center, Georgetown University, Washington D.C., United States of America

    ⨯
  • Stefano Vicini,

    Roles Conceptualization, Formal analysis, Investigation, Methodology, Resources, Writing – review & editing

    Affiliation Department of Pharmacology and Physiology, Georgetown University, Washington D.C., United States of America

    ⨯
  • G. William Rebeck

    Roles Conceptualization, Data curation, Funding acquisition, Methodology, Project administration, Supervision, Writing – review & editing

    gwr2@georgetown.edu

    Affiliations Interdisciplinary Program in Neuroscience, Georgetown University, Washington D.C., United States of America, Department of Neuroscience, Georgetown University, Washington D.C., District of Columbia, United States of America

    ⨯

Abstract

APOE4 is a risk factor for several disease states associated with cognitive impairment, including Alzheimer’s disease and cancer-chemotherapy induced cognitive impairment. Using mouse knock-in models of human APOE alleles, we examined the effects of APOE genotype and chemotherapy on the ex vivo electrophysiological characteristics of excitatory and inhibitory neurons in the entorhinal cortex (EC). We found that APOE4 is associated with a significantly higher excitatory/inhibitory ratio (0.33 ± 0.04) in the layer 2/3 pyramidal cells of the entorhinal cortex compared to APOE3 (0.19 ± 0.04). We crossed APOE mice to mice with parvalbumin (PV) interneurons tagged with tdTomato, allowing us to measure effects specifically on this inhibitory cell type. For EC pyramidal neurons, the chemotherapeutic agent doxorubicin caused increases in the amplitudes of both spontaneous excitatory and inhibitory post-synaptic currents, with significant responses (***p < 0.001; **p < 0.01 respectively) in APOE3 brains. For EC PV neurons, APOE4 genotype was associated with significantly lower firing rates at injections of high currents (**p < 0.01), but rates were unaffected by doxorubicin. Doxorubicin doubled the percentage of PV cells that showed inactivation block in APOE3 brains (25% to 52%) but had no effect on APOE4 brains (50% to 54%). This ex vivo study suggests that APOE4 impairs homeostatic synaptic transmission in pyramidal cells under control conditions and causes a lack of responsiveness to a stressor (doxorubicin treatment) in PV cells.

Citation: Luo N, Pandit H, Kalra S, Tran E, Mandelblatt J, Vicini S, et al. (2026) APOE4 and doxorubicin impair inhibitory interneuron function and homeostatic regulation in the entorhinal cortex. PLoS One 21(3): e0343276. https://doi.org/10.1371/journal.pone.0343276

Editor: Ioannis Liampas, University of Thessaly Faculty of Medicine: Panepistemio Thessalias Tmema Iatrikes, GREECE

Received: November 21, 2025; Accepted: February 3, 2026; Published: March 26, 2026

Copyright: © 2026 Luo et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.

Data Availability: The data underlying the results presented in the study are available through the Guidelines International network, https://doi.org/10.12751/g-node.qybv42.

Funding: This work was supported by NIH grants R01AG071745 (GWR), F30AG082448 (NL), and R35CA283926 (JM), and by the Alzheimer’s Association grant 24-1310437 (HP). Dr. Mandelblatt was also supported by the Frank M. Ewing Foundation Endowed Chair of Hematology and Oncology. Funding support was also received from the Georgetown University Medical Center. The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Competing interests: No authors have competing interests.

Introduction

The APOE4 allele is the strongest genetic risk factor for sporadic Alzheimer’s disease (AD), present in about 25% of the US population [1]. APOE4 is also implicated in other neurocognitive disorders, suggesting that APOE4 has broad effects on cognition not limited to AD pathogenesis [1]. One example is cancer chemotherapy-related cognitive impairment (CRCI), where APOE4 is linked to worse cognitive outcomes in breast cancer survivors who had adjuvant chemotherapy treatment compared to APOE3-matched survivors [2–4]. Preclinical models support these findings: APOE4 mice and rats treated with different chemotherapeutic agents exhibited impaired learning and memory as well increased signs of neuroinflammation compared to treated APOE3 mice [5–9]. These studies suggest that APOE genotype differentially affects the brain’s ability to respond to stressors (e.g., amyloid plaques, peripheral chemotherapy), resulting in APOE4 individuals being more susceptible to cognitive deficits compared to their APOE3 counterparts. With the population of cancer survivors estimated to increase by 44% by 2040 [10], combined with the high prevalence of APOE4 in the general population, it is important to understand how the APOE4 allele predisposes individuals to cognitive decline in the setting of CNS stressors.

A brain region susceptible early in AD pathogenesis and critical for cognition is the entorhinal cortex (EC). Its specific vulnerability to pathological development may be influenced by several factors: its anatomical location, higher metabolic demand, early network dysfunction, increased oxidative stress, and elevated neuroinflammatory profile [11,12]. The EC has rich connectivity with neocortical regions, the thalamus, and limbic structures, and thus is well positioned within the temporal lobe to control the information flow necessary for cognition [13]. At a circuit level, cognition relies on the maintenance of an excitatory-inhibitory (E/I) balance; perturbations cause reduced brain flexibility, working memory deficits, and impaired decision-making [14,15]. Neuronal connectivity, both structural and functional, is disrupted under several conditions, including in cancer patients treated with chemotherapy [16]. E/I dysregulation is apparent in AD, occurring at early disease stages, perhaps even before measurable amyloid accumulation [17]. E/I imbalance progressively worsens in later stages of the disease, alongside increased network decoupling driven primarily by impairment in inhibitory, rather than excitatory, connectivity [17,18]. This impairment is supported by resting-state magnetoencephalography, which showed that E/I imbalance and hyperexcitation were associated with worse cognitive performance in AD dementia [19].

Given the critical effect of E/I regulation on cognition, we examined this property in human APOE knock-in mice in our analysis of the effects of APOE4 on brain functions. APOE4 affects normal brain activity [20], including alterations of cortical hypometabolism [21] and hyperactivity [22,23]. We used an ex vivo model, which allowed us to precisely measure electrophysiological variables in both excitatory and inhibitory circuits of the EC related to normal cognition. In addition, we examined its effects in the presence of the acute stressor doxorubicin, a chemotherapeutic drug linked with CRCI particularly in breast cancer survivors [24]. Although direct toxic effects have been observed for some chemotherapies [25], doxorubicin does not cross the blood brain barrier [8] and thus may be linked to indirect mechanisms such as peripheral inflammation [25], disruption of the blood brain barrier [26] or alterations to the gut biome [27]. Doxorubicin may also have gross effects on the hippocampus through inhibiting or altering neurogenesis [28–30], but little has been done to test whether it has effects on neuronal functions [31].

We used whole-cell patch clamp experiments in ex vivo brain slices to interrogate the functional differences in excitatory and inhibitory neurons. Of the inhibitory interneuron subtypes, we focused on the fast-spiking parvalbumin (PV) interneurons due to their abundance in the EC, unique physiological functions, and importance in AD pathogenesis [32–34]. Evidence across AD brains and AD models suggests that dysregulation of these neurons occur at different stages of pathogenesis [35], including reduced firing in the cortex before the formation of plaques [36]; these effects may be related to changes in perineuronal net coverage with APOE4 [37] and with aging [38]. We found that inhibitory tone in APOE4 was disrupted due to differences in inhibitory PV interneuron firing rate, leading to an altered E/I balance. When doxorubicin was administered intraperitoneally to APOE4 mice, EC excitatory neurons did not respond with changes in synaptic transmission characteristics while APOE3 pyramidal neurons displayed increased excitatory and inhibitory spontaneous postsynaptic currents. Together, our work demonstrates that APOE4 genotype reduces circuit flexibility in response to stress in the EC.

Materials and methods

Mice

The experiments performed were in accordance with the guidelines and regulations of Public Health Service Policy on Humane Care and Use of Laboratory Animals, and comply with guidelines set forth in the Guide for the Care and Use of Laboratory Animals. They were approved by the Georgetown University Institutional Animal Care and Use Committee (protocol # 2016−1322). All methods are reported in accordance with ARRIVE guidelines. For study of the general effects of APOE genotype on neuronal excitation and inhibition, we used APOE-Targeted Replacement C57BL/6J mice homozygous for human APOE3 or APOE4 [39]. For study of the specific effects of PV-positive inhibitory cells, we crossed APOE knock-in mice commercially available through Jackson Laboratories (JAX: 029018; 027894) [40] with knock-in tdTomato Ai14 (B6.Cg-Gt(ROSA)26Sor tm14(CAG-tdTomato)Hze/J; JAX:007914) and PV-cre reporter (B6;129P2-Pvalstm1(cre)Arbr/J; JAX:008069) mice. The mice generated were homozygous for human APOE and heterozygous for tdTomato and PV-cre; the tdTomato tag allowed for identification of PV-positive cells in patch clamp analyses. Mice were housed on a 12:12 light-dark cycle with ad libitum access to food and water. For electrophysiology experiments, all mice were 1.5 to 2 months old. For analyzing the effects of aging on PV cell density, we used the same PV-tagged APOE knock-in mice but aged to 20 months. Since we were modeling chemotherapy used in treatment of breast cancer, only females were used. A schematic of the general workflow is shown in Fig 1.

thumbnail
Download:
Fig 1. Experimental design.

APOE mice (6 to 8 weeks old) were euthanized, and brains perfused in preparation of electrophysiology experiments. Horizontal brain slices were prepared, equilibrated in 95% O2, 5% CO2 and used for whole cell patch clamping; recorded cells were biocytin-filled. Tissue was then fixed and stained for the excitatory neurons or PV-positive interneurons studied, which were imaged by confocal microscopy.

https://doi.org/10.1371/journal.pone.0343276.g001

Genotyping.

Mice were ear tagged with a numerical metal tag on one ear, while the non-tagged ear was used to collect tissue for genotyping. The Kaneka Easy DNA extraction kit was used to extract DNA from the samples and a PCR master mix (ThermoFisher DreamTaq 2X) was used for target gene amplification. TdTomato reporter Ai14 primers used: (mutant forward: CTG TTC CTG TAC GGC ATG G, reverse: GGC ATT AAA GCA GCG TAT CC), (WT forward: AAG GGA GCT GCA GTG GAG TA, reverse: CCG AAA ATC TGT GGG AAG TC). Human APOE primers were: (APOE4 forward: AGG AGG TTG AGG TGA GGA TG, APOE4/APOE3 reverse: AAT TTT TCC CTC CGC AGA CT, APOE3 forward: ACA GCT GCT CAG GGC TAT TG). Parvalbumin primers used: (mutant forward: AAA TGC TTC TGT CCG TTT GC, reverse: ATG TTT AGC TGG CCC AAA TG), (WT forward: CAG AGC AGG CAT GGT GAC TA, reverse: AGT ACC AAG CAG GCA GGA GA). PCR products were loaded onto a 5% (weight/volume) agar gel with ethidium bromide and electrophoresed at 120 V for 45 minutes.

Doxorubicin and control injections.

Doxorubicin hydrochloride was dissolved in ultrapure DMSO to generate a 20 mM stock solution. Female littermates were individually weighed and distributed across treatment groups when possible. The total volume per injection was 200 uL. Doxorubicin-treated animals received a single injection containing 170 uL 1X PBS and 30 uL DMSO to achieve a 10 mg/kg dosage. Control animals received identical injections without any doxorubicin added. Mice were injected intraperitoneally at the lower left quadrant of the abdomen, returned to their home cage with littermates, and monitored for 48 hours for adverse side effects during the elimination period of the drug. Mice in the experiments related to doxorubicin treatment were handled more frequently, including moving between housing zones due to the use of hazardous materials. Mice were left to recover for one week prior to experimental use.

Electrophysiology slice preparation

Mice were anesthetized and euthanized with unmetered isoflurane and transcardially perfused with ice-cold high sucrose dissection buffer (87 mM NaCl, 2.5 mM KCl, 7 mM MgSO4,1.25 mM NaH2PO4, 75 mM sucrose, 25 mM dextrose, 0.5 mM CaCl2, 25 mM NaHCO3) bubbled with carbogen (95% O2, 5% CO2). 275 µm thick horizontal brain slices were obtained on a Vibratome Series 3000. The entorhinal cortex was visually determined based on its location relative to the anatomical shape of the hippocampus. Four to five hemi-brain slices per animal were incubated for 30 minutes at 30°C in artificial cerebrospinal fluid (ACSF; 124 mM NaCl, 4.5 mM KCl, 1 mM MgCl2, 1.2 mM NaH2PO4, 10 mM Glucose, 2 mM CaCl2, 26 mM NaHCO3, 300–310 mOsm/kg). Slices were then recovered at room temperature for a minimum of 30 minutes prior to patch clamp recordings. Recordings were routinely done across several slices from the same mouse; this approach allowed larger numbers of neurons to be used to address the research questions.

Whole-cell patch clamp recordings

Slices were organized into three groups based on the aspects of the EC. Slices were considered “early” if they occurred before the presence of the stria terminalis, “middle” if they contained the stria terminalis, and “late” if primarily caudate/putamen was observed. Cell recordings were only done in “middle” and “late” slices to minimize the chance of recording from non-entorhinal areas. Medial and lateral EC were not differentiated during whole-cell recordings. Layer 2/3 of the EC was determined as roughly half the cortical width. Recordings were performed at ~30°C with 3–5 MΩ glass pipette electrodes filled with varying internal solutions based on the experimental parameters. Slices were placed in a unidirectional ACSF perfusion chamber to prevent recycling of cell waste. 2X brightfield images were taken at the end of each whole cell recording to ensure that the electrode was positioned in the correct layer and region. Data were collected using the MultiClamp 700B amplifier, Bessel filtered at 2 kHz (voltage clamp) or 10 kHz (current clamp) and digitized at 10 kHz using Digidata 1440A and pCLAMP 11 (Molecular Devices).

Pyramidal spontaneous postsynaptic current recordings

A cesium-based internal solution (120 mM CsMeSO3, 5 mM NaCl, 10 mM tetraethylammonium TEA-Cl, 10 mM HEPES, 4 mM (Na)ATP, 0.3 mM (Na)GTP, 4 mM QX314, 280–300 mOsm/kg, pH 7.2–7.4) with 0.5% (w/v) biocytin was used for spontaneous postsynaptic current recordings. Spontaneous excitatory postsynaptic currents (sEPSCs) were measured in voltage-clamp at a holding potential of −60 mV, while inhibitory postsynaptic currents (sIPSCs) were measured at a holding potential of 0 mV within the same cell. A 10 mV current step was given before the switch from −60 to 0 mV to allow measures of capacitance, series and input resistance.

Pyramidal miniature postsynaptic current recordings

Miniature (m) EPSCs and IPSCs followed the same experimental paradigm as in the spontaneous postsynaptic current recordings, with the addition of a Y-tube placed near the patch pipette; Y-tubing in patch slices allowed for drug delivery at the recording site [41]. Cells were initially held at −60 mV with ACSF flowing through the Y-tubing to ensure stable responses. After 30 seconds, 1 µM TTX was applied via Y-tubing to inhibit voltage gated sodium channels, allowing recording of mEPSCs for 3–5 minutes. A mixture of 1 µM TTX, 5 µM NBQX, and 5 µM CPP was then applied to block AMPA and NMDA-mediated currents prior to giving a 10 mV seal test and changing the holding potential to 0 mV to record mIPSCs for an additional 3–5 minutes. Finally, 10 µM gabazine was applied to confirm that the mIPSCs recorded were indeed GABAA receptor mediated.

Whole-cell patch clamp recordings: Spontaneous and miniature events

Both spontaneous and miniature events were taken from a portion of the entire experimental recording to provide an estimate of excitatory and inhibitory synaptic currents’ amplitude and frequency measures per each cell. A stable recording segment was exported for analysis during −60 mV and 0 mV tracings to provide a representative sample of sEPSCs and sIPSCs, respectively. For miniature recordings, the mEPSC sample was taken 30 seconds after the application of TTX onto the patched cell via Y-tubing. The mIPSC sample was taken after the application of TTX, APV, and NBQX and switch to 0 mV holding potential. Each representative sample for sEPSCs, sIPSCs, mEPSCs, or mIPSCs was analyzed using Easy Electrophysiology v2.4.0. Recordings were Bessel filtered at 1 KHz and the ‘event-threshold’ feature set at 2.5 pA was used as a first pass to detect the largest postsynaptic currents followed by manual selection of the smaller peaks identified using the template matching algorithm in the Easy Electrophysiology Software (https://www.easyelectrophysiology.com/features). Individually sorted events were visually validated, and artifacts discarded. Extra-large sIPSCs occurred in some cells possibly due to upstates (reflected in Fig 1); they were also excluded from the analyses. Cumulative distribution plots were created in GraphPad Prism v10 with a bin width of 0.05.

Whole-cell patch clamp recordings: E/I ratio

The E/I ratio per cell was calculated in Clampfit v11.1 using the same representative traces from the analysis of sEPSCs and sIPSCs. These recordings were split into 500 ms segments and the total charge per segment was calculated as the area between the trace and baseline. The average charge of all sEPSC segments was divided by the average charge of all sIPSCs segments to determine the E/I ratio for that individual cell.

Intrinsic excitability of PV cells

Layer 2/3 entorhinal PV cells were identified using tdTomato fluorescence (554 nm excitation) and recorded from with glass patch electrodes filled with a K+-based internal solution (130 mM K Gluconate, 5 mM KCl, 0.6 mM EGTA, 2 mM MgCl2, 2 mM Na2ATP, 0.3 mM (Na)GTP, and 10 mM HEPES, 280–300 mOsm/kg, pH 7.2–7.4) containing 0.5% (w/v) biocytin. Depolarizing current was injected into the cell to reach a membrane potential of −70 mV prior to recording action potential trains. Alternating 20 pA increments of depolarizing and hyperpolarizing pulses (500 ms each) were used to measure passive and active properties of PV cells. Current-clamp recordings were stopped when either the injection reached 600 pA or the cell displayed inactivation following maximum steady-state firing rate.

PV cell active and passive properties

All PV cell properties were measured in Clampfit v11.1. Firing frequency was based on PV cells that did not display inactivation block, since the presence of inactivation made it difficult to accurately quantify the number of spikes per 500 ms current injection. The maximum steady-state firing rate was defined as the highest frequency observed in each PV cell typically occurring in the plateau at current injections > 350 pA. Cells were deemed inactivated if at any point during the strongest current steps there were signs of inhibited fast-spiking trains under sustained depolarization. The voltage changes with hyperpolarization due to Ikr were estimated by plotting hyperpolarizing current steps (0 to −600 pA) against voltage, subtracting the slope of the current-voltage line at lower currents (−20 to −60 pA) from the slope of the line at higher currents (−300 to −400 pA), and dividing by the slope of the current-voltage line at lower currents. See S2 Table for other detailed passive and active property measurements.

Immunohistochemistry in electrophysiology slices

Outside-out patches were made after electrophysiological recordings to preserve the biocytin (0.5%) within the cell. Brain slices were fixed in 4% paraformaldehyde/4% sucrose/PBS overnight, then transferred into 1X PBS containing 0.05% sodium azide. Slices were permeabilized with 0.5% Triton X-100 and treated with fluorescein-avidin antibody (1:500) for 2 hours followed by a PBS rinse to visualize the morphology of the recorded cells.

Parvalbumin cell density: Slice preparation

Heterozygous tdTom/PV-Cre mice on APOE3 or APOE4 backgrounds were euthanized via carbon dioxide inhalation and transcardially perfused with ice-cold 1X PBS until the liver became pale. Brains were extracted and placed in 4% paraformaldehyde/4% sucrose at 4°C for 48 hours. A sucrose gradient of 10%, 20%, and 30% was used to gradually dehydrate the brains over the course of one week, after which the brains were flash frozen in cold 2-methlybutane for long-term storage at −80 °C until slicing. Brain hemispheres were cut on a microtome at 30-micron thickness in the horizontal orientation. Free floating slices were stored in cryoprotectant at 4 °C until mounting on slides with Fluoromount.

Image acquisition: Confocal microscopy

Free floating electrophysiological slices were imaged with a Thorlabs Inc. resonance laser scanning confocal (Sterling, VA, United States) equipped with an argon laser (λ = 488 nm) integrated on a Nikon Eclipse FN1 upright microscope. For the biocytin-filled cells, confocal z-stacks were acquired with a 20X Nikon objective at a step size of 0.5 μm and resolution of 0.962 µm/pixel. The confocal z-stacks were analyzed by Image J to define cell morphology and confirm cell type.

Image acquisition: Widefield microscopy

Mounted PV cell density sections were imaged with a Mica microscope (Leica Microsystems) containing both widefield and confocal capabilities. Due to the relatively large size of PV somas, widefield microscopy with the tdTomato filter (λ = 594 nm) was used to obtain cell density counts for the EC. A tiling approach paired with z-stacks at a step size of 2 µm was used to generate images of the entire EC and parts of the hippocampus for depth identification. Tiled z-stacks were then processed using the ‘maximum projection’ feature on the Mica software to create a 2D projection of the maximum fluorescence of the section. These images were saved and uploaded to ImageJ for analysis.

Image analysis: Parvalbumin cell density

20X tiled widefield images were imported into ImageJ and ROIs were manually drawn for layers 2, 3 and 4/5 (combined). Each layer was analyzed by first excluding the image outside of the ROI then thresholding the ROI to yield the highest cell body count with minimal noise. A binarized mask was created, processed with ‘open’, and filtered at a medium radius of 10 pixels to remove small particles. The ‘watershed’ function was used to separate conjoined cells. Particles were analyzed with the size greater than or equal to 30 μm2 and the circularity set from 0.30–1.00. The individual cell ROIs were overlayed to the original image to ensure accurate markings drawn around each PV cell soma. Any ROIs that were missed or incorrectly drawn were fixed manually. Cell density was calculated by dividing the total number of cells by the area of the layer ROI (µm2), then converting to mm3 using the thickness of each slice (30 µm).

Statistics

Data were analyzed with the Mann-Whitney U test with Bonferroni correction, 2-way ANOVA with post-hoc Tukey’s HSD, Kolmogorov Smirnov test, or 3-way ANOVA with post-hoc Tukey’s HSD as indicated in the figure legends. Statistical tests were determined based on the number of independent variables in the data set as well as the data distribution. Normality was tested using the Shapiro-Wilk test. If data did not pass normality on any of these tests, an appropriate nonparametric analysis was used. Sample sizes are displayed in figure legends as (number of cells, number of animals). Lines and error bars in the dot plot histograms across all figures reflect the mean and SEM.

Results

APOE4 increases pyramidal cell hyperexcitability in the entorhinal cortex

We used young (6–8-week-old) human APOE-targeted replacement mice as a model of sporadic AD risk devoid of pathological hallmarks [39]. Spontaneous excitatory and inhibitory postsynaptic currents (sEPSCs, sIPSCs) were recorded using patch clamp techniques from layer 2/3 pyramidal cells in the EC, which receive input from various cortical regions and project their axons into the dentate gyrus and the CA subfields of the hippocampus (Fig 2A). Patched cells were filled with 0.5% biocytin and stained with streptavidin-fluorescein post-hoc to confirm pyramidal cell morphology (Fig 2B), an important step given that another primary excitatory cell type, the stellate cell, is also present in the EC [42]. We found that the E/I ratio was significantly elevated by 74% in APOE4 compared to APOE3 EC pyramidal cells (Fig 2C). To parse out excitatory versus inhibitory changes that could contribute to an elevated E/I ratio, sEPSC and sIPSC amplitude and frequency were compared. The frequency of sIPSCs was significantly lower in APOE4 compared to APOE3, along with a small reduction in sEPSC frequency (Fig 2D and 2F). No differences in sEPSC or sIPSC amplitudes were observed (Fig 2E and 2G).

thumbnail
Download:
Fig 2. APOE4 pyramidal cells in layer 2/3 of the EC are hyperexcitable.

A. Schematic of the EC circuitry in the horizontal slice orientation showing layer 2/3 pyramidal cells projecting their axons into the hippocampal formation. B. Representative image of a 0.05% biocytin-filled pyramidal cell from tissue fixed and stained with streptavidin after patch clamp recording. Scale bar = 70 μm. C.-G: Analysis of APOE3 (22 cells, 4 mice) and APOE4 (23 cells, 4 mice). C. E/I ratio of APOE3 (0.19 ± 0.04) and APOE4 (0.33 ± 0.04) layer 2/3 pyramidal cells. Circles represent individual cells. D. sIPSC frequency represented as a dot plot (left) and cumulative probability distribution (right). Representative traces are displayed below graphs. APOE3 (7.62 ± 0.68), APOE4 (3.97 ± 0.48). E. sIPSC amplitude represented as a dot plot (left) and cumulative probability distribution (right). Representative average amplitude is overlaid below graphs. APOE3 (10.65 ± 0.88), APOE4 (10.41 ± 0.76). F. sEPSC frequency dot plot (left) and cumulative probability distribution (right) in APOE3 (8.87 ± 0.80) and APOE4 (6.61 ± 0.79) pyramidal cells. G. sEPSC amplitude in APOE3 and APOE4 pyramidal cells. APOE3 (5.55 ± 0.48), APOE4 (5.39 ± 0.47). H.-I. Analysis of APOE3 (16 cells, 3 mice) and APOE4 (35 cells, 6 mice). H. mIPSC amplitude, frequency, and half-width in APOE3 and APOE4 pyramidal cells. Example traces are shown below graphs. mIPSC amplitude: APOE3 (9.27 ± 0.67), APOE4 (5.41 ± 0.43). mIPSC frequency: APOE3 (5.75 ± 0.57), APOE4 (5.37 ± 0.47). mIPSC half-width: APOE3 (15.54 ± 0.87), APOE4 (17.05 ± 0.98). I. mEPSC amplitude, frequency, and half-width in APOE3 and APOE4 pyramidal cells. mEPSC amplitude: APOE3 (4.86 ± 0.45), APOE4 (4.32 ± 0.33). mEPSC frequency: APOE3 (5.73 ± 0.68), APOE4 (7.48 ± 0.93). mEPSC half-width: APOE3 (10.24 ± 0.90), APOE4 (11.42 ± 0.50). All mice analyzed were female. All statistical significance tested using the Mann-Whitney U test. Data are represented as (mean ± SEM). * p < 0.05, **** p < 0.0001.

https://doi.org/10.1371/journal.pone.0343276.g002

Because spontaneous postsynaptic currents are influenced by network activity and synaptic connectivity, we tested the miniature (m) postsynaptic currents. This measure tested if the changes seen in sIPSC frequency were primarily driven by a decrease in inhibitory neuron activity or in the number of inhibitory synapses. No differences in mIPSC frequency or half-width were noted, suggesting that the number of inhibitory terminals, release probability, and receptor kinetics were not affected by APOE genotype. However, APOE4 mIPSC amplitude was significantly lower than APOE3 mIPSC amplitude by 42% (Fig 2H), similar to published findings of 20-month-old APOE4 mice [23]. These results indicate that APOE4 brains have deficits in inhibitory regulation within the EC, at both the microcircuit and synaptic levels. While we had observed a modest decrease in sEPSC frequency in APOE4 brains compared to APOE3 brains (consistent with the reduction in cortical neuron spine density in APOE4 mice [43]), no differences in mEPSC amplitude, frequency or half-width were seen (Fig 2F and 2I). These data support the conclusion that inhibitory dysfunction may be an early marker of elevated risks associated with APOE4 brains.

APOE4 parvalbumin interneurons have reduced maximum steady-state firing

We assessed parvalbumin (PV)-positive interneurons as a possible source of inhibitory dysfunction due to their abundance in the EC and their synapse localization directly onto pyramidal cell somas and proximal dendrites [44]. PV cells have a unique fast-spiking phenotype that differentiates them from other inhibitory interneuron types in the cortex and enables them to tightly control inhibition within microcircuits [45]. We crossed a knock-in mouse line expressing the tdTomato Ai14 reporter and PV-cre with newer lines of human APOE3 or APOE4 knock-in mice [39,46,47]. This mouse model allowed us to identify inhibitory PV interneurons in the EC using red fluorescence, enabling analysis of APOE genotype effects specifically in these cells.

To test for differences in PV cell density as a potential source of reduced inhibition in APOE4, we counted the PV interneurons in layers 2, 3 and 4/5 of the EC (Fig 3A). No differences in layers 2 or 3 were found between genotypes in both young (6-week-old) and aged (20-month-old) animals, although some difference in PV cell density was observed in layer 4/5 in young APOE mice (Fig 3B and 3C). We recorded sEPSCs to determine if inhibitory PV interneuron excitatory input could account for the difference in inhibitory tone between APOE3 and APOE4. Both sEPSC amplitude and frequency were unaffected by APOE genotype, indicating that any difference in PV interneuron function is more intrinsic and unlikely driven by upstream circuit changes (Fig 3D).

thumbnail
Download:
Fig 3. APOE4 PV interneurons show deficits in firing frequency.

A. Representative images from tdTom+/-;PVCre+/-;APOE+/+ transgenic mice. Example ROI layer tracings in the EC. Scale bar = 1000 μm. B. PV cell density across EC layers 2, 3, and 4/5 (combined) in young (6-weeks-old). Circles represent individual brain slices. Layer 2young: APOE3 (5640 ± 294), APOE4 (5878 ± 470). Layer 3 young: APOE3 (5814 ± 319), APOE4 (4857 ± 438). Layer 4/5 young: APOE3 (5471 ± 373), APOE4 (3896 ± 460). APOE3young (23 slices, 5 female mice), APOE4young (19 slices, 5 female mice). C. PV cell density in aged (20-months-old) APOE animals. Layer 2aged: APOE3 (5981 ± 260), APOE4 (6311 ± 408. Layer 3aged: APOE3 (6606 ± 346), APOE4 (6560 ± 414). Layer 4/5aged: APOE3 (6170 ± 435), APOE4 (6392 ± 567). APOE3aged (78 slices, 3 males, 2 females). APOE4aged (43 slices, 3 males, 2 females). 2-way ANOVA with post hoc multiple comparisons test. D. APOE3 and APOE4 PV sEPSC frequency (left) and amplitude (right) in layer 2/3 of the EC. sEPSC frequency APOE3 (17.1 ± 1.66), APOE4 (12.5 ± 1.04). sEPSC amplitude APOE3 (9.0 ± 1.11), APOE4 (8.1 ± 0.64). APOE3 (19 cells, 3 female mice). APOE4 (23 cells, 3 female mice). Mann-Whitney U test. E. PV firing frequency across increasing current injections. APOE3: (16 cells, 3 female mice). APOE4: n = (20 cells, 3 female mice). Kolmogorov-Smirnov test. F. Example current-clamp recordings of PV interneuron maximum steady-state firing frequency per genotype. G. Dot plot of maximum steady-state action potential frequency excluding cells with inactivation block. APOE3 (164 ± 8.5), APOE4 (108 ± 7.6). APOE3 (18 cells, 3 female mice), APOE4 (20 cells, 3 female mice). Mann-Whitney U test. H. Inward rectification at hyperpolarizing current steps in non-injected APOE3 (0.39 ± 0.032) and APOE4 (0.31 ± 0.022) mice. APOE3 (18 cells, 3 female mice), APOE4 (24 cells, 3 female mice). Mann-Whitney U test. I. Distribution of total recorded PV interneurons that displayed inactivation block during current-clamp injections in APOE3 (11%) and APOE4 (25%). White numbers indicate number of cells. APOE3 (18 cells, 3 female mice), APOE4 (24 cells, 3 female mice). ** p < 0.01, ****p < 0.0001.

https://doi.org/10.1371/journal.pone.0343276.g003

We assessed PV interneuron function using current clamp injections of hyperpolarizing and depolarizing incremental steps (20 pA) to look at active and passive properties associated with firing frequency. The firing frequency of PV interneurons (excluding those that displayed inactivation block) was similar between APOE3 and APOE4 at lower current injections (< 300 pA); however, striking differences in firing frequency were evident at higher injections (> 350 pA) (Fig 3E and 3F). Maximum steady-state firing frequency, defined as the greatest spike frequency occurring within the plateau of the current injection-response curve, was significantly lower in the APOE4 group, providing a possible explanation for the impaired inhibitory tone seen in layer 2/3 EC pyramidal cells (Fig 3G). We also measured inward-rectification mediated by potassium channels (Ikr) in PV interneurons given that stronger inward-rectification hyperpolarizes the resting membrane potential and decreases input resistance, making it harder for neurons to elicit action potentials. We did not observe any difference in the Ikr ratio between APOE3 and APOE4 cells, which was in line with the lack of change in rheobase, resting membrane potential, and input resistance (Fig 3H, S1 Table). Qualitative analysis of sustained PV action potential firing showed a larger proportion of PV interneurons inactivating at higher depolarizing current steps in the APOE4 cohort (25%) compared to APOE3 cohort (11%) (Fig 3I). These findings suggest that inhibitory dysfunction in APOE4 EC is rooted in abnormalities within the PV neurons themselves. This conclusion is in line with the hypothesis that APOE4 causes selective vulnerability in PV positive GABAergic interneurons [48].

Doxorubicin increases synaptic transmission in APOE3 but not APOE4 pyramidal cells

To understand how the EC circuit differentially responds to an acute stressor in APOE3 and APOE4 brains, we treated young female APOE3 and APOE4 mice with a single intraperitoneal injection of the chemotherapeutic agent doxorubicin (10 mg/kg) or vehicle (DMSO) and analyzed changes in synaptic activity in the EC layer 2/3 pyramidal cells (Fig 4A). Under these conditions, the overall E/I ratio was not significantly affected by APOE genotype, although it showed the same directionality of the effect in Fig 2 but in APOE knock-in mice from a different source [40]. The E/I was not affected by doxorubicin treatment or by the number of days post-injection (Fig 4B, S1 Fig). We analyzed spontaneous postsynaptic currents to gain a detailed picture of excitatory and inhibitory changes. sIPSC frequency did not change for the most part, although some significance was detected between the doxorubicin-treated groups (Fig 4C). sIPSC amplitude was significantly increased in the APOE3 doxorubicin group compared to its control group, while differences were not observed in the treated APOE4 brains (Fig 4D left). This effect of doxorubicin was noticeable in the sIPSC amplitude cumulative distribution plot, where APOE3 doxorubicin showed a strong rightward shift of the curve compared to APOE3 control (Fig 4D right). Interestingly, we found that the excitatory component of the E/I ratio followed a similar response to that of the inhibitory component: sEPSC frequency was unaffected across genotype and treatment groups, but sEPSC amplitude was significantly increased by doxorubicin in the APOE3, but not APOE4, group (Fig 4E and 4F). Thus, APOE3 brains responded to the stress of doxorubicin by increasing excitatory and inhibitory neurotransmission into pyramidal cells, while the APOE4 brains did not.

thumbnail
Download:
Fig 4. Doxorubicin increases excitatory and inhibitory synaptic transmission in APOE3 but not APOE4 pyramidal cells.

A. Timeline of doxorubicin and vehicle (control) injections. B.-F. Analysis of APOE3 CTRL (13 cells, 2 mice), APOE3 DOX (28 cells, 4 mice), APOE4 CTRL (21 cells, 4 mice), and APOE4 DOX (23 cells, 5 mice). B. Layer 2/3 EC pyramidal cell E/I ratios in APOE3 and APOE4 control and doxorubicin treated groups. Circles represent individual cells. APOE3 CTRL (0.22 ± 0.04), APOE3 DOX (0.34 ± 0.05), APOE4 CTRL (0.29 ± 0.04), APOE4 DOX (0.36 ± 0.05). C. sIPSC frequency dot plot (left), cumulative probability distribution (middle), and examples APOE4 CTRL and DOX traces (right). APOE3 CTRL (9.43 ± 1.03), APOE3 DOX (7.07 ± 0.67), APOE4 CTRL (8.88 ± 1.01), APOE4 DOX (10.88 ± 1.31). D. sIPSC amplitude dot plot (left), cumulative probability distribution (middle), and example APOE3 CTRL and DOX traces (right). APOE3 CTRL (7.57 ± 0.87), APOE3 DOX (13.07 ± 1.00), APOE4 CTRL (9.70 ± 0.65), APOE4 DOX (10.80 ± 1.06). E. sEPSC frequency dot plot (left), cumulative probability distribution (middle), and example APOE4 CTRL and DOX traces (right). APOE3 CTRL (9.44 ± 0.77), APOE3 DOX (9.95 ± 0.91), APOE4 CTRL (12.48 ± 1.22), APOE4 DOX (13.38 ± 1.31). All mice were female. F. sEPSC amplitude dot plot (left), cumulative probability distribution (middle), and examples of APOE3 CTRL and DOX traces (right). APOE3 CTRL (4.13 ± 0.44), APOE3 DOX (6.27 ± 0.32), APOE4 CTRL (4.67 ± 0.36), APOE4 DOX (4.88 ± 0.35). All statistics were analyzed using 2-way ANOVA with post hoc multiple comparisons tests. * p < 0.05, ** p < 0.01, *** p < 0.001.

https://doi.org/10.1371/journal.pone.0343276.g004

Doxorubicin induces inactivation block in APOE3 but not APOE4 PV interneurons

Our prior results showed no effect of APOE4 genotype on PV interneuron density, but instead on the maximum steady-state firing rate and inactivation block. Here, we tested if doxorubicin had similar effects on PV interneurons (Fig 5A). sEPSC frequency and amplitude revealed no significant differences across group comparisons, suggesting that changes in excitatory input did not underlie any functional changes between vehicle- and doxorubicin-injected APOE3 and APOE4 female mice (Fig 5B and 5C). When we compared the firing rate of interneurons that did not display inactivation block, we saw that doxorubicin had no effect within APOE genotype groups (Fig 5D and 5E). This was most prominently seen at higher current injections (> 300 pA), while at lower current injections there was more variability and differences in rheobase; these differences overall remained statistically nonsignificant. Across all distributions, there was a significant difference between APOE3 and APOE4 controls and between APOE3 and APOE4 doxorubicin-treated groups (Fig 5F). Quantification of the maximum steady-state firing rate between groups showed that APOE3 vehicle-treated mice had a significantly higher firing frequency compared to APOE4 vehicle-treated mice, a finding consistent with our prior results from non-injected APOE3 vs APOE4 female mice (Fig 5G and 3G). Inward rectification was not significantly different in APOE3 PV interneurons in doxorubicin-treated mice compared to vehicle controls (Fig 5H).

thumbnail
Download:
Fig 5. Doxorubicin induces inactivation block in APOE3 PV interneurons, but no additional effects in APOE4.

A. Image of PV-positive cells (in red), with one that was analyzed by patch clamp and filled with biocytin (in yellow). B.-C. Analysis of APOE3 CTRL (19 cells, 4 mice). APOE3 DOX (18 cells, 3 mice). APOE4 CTRL (n = 22 cells, 3 mice). APOE4 DOX (15 cells, 3 mice). B. PV sEPSC frequency in APOE3 and APOE4 control and doxorubicin groups. APOE3 CTRL (15.53 ± 1.34), APOE3 DOX (13.81 ± 0.87), APOE4 CTRL (16.61 ± 1.60), APOE4 DOX (15.03 ± 2.05). 2-way ANOVA with post hoc multiple comparisons test. C. PV sEPSC amplitude in APOE3 and APOE4 control and doxorubicin groups. APOE3 CTRL (10.97 ± 0.81), APOE3 DOX (9.63 ± 1.15), APOE4 CTRL (8.87 ± 0.60), APOE4 DOX (9.59 ± 1.19). APOE3 CTRL (19 cells, 4 mice), APOE3 DOX (18 cells, 3 mice), and APOE4 CTRL (22 cells, 3 mice), APOE4 DOX (17 cells, 3 mice). 2-way ANOVA with post hoc multiple comparisons test. D. Cumulative distribution of only APOE3 control and doxorubicin-treated cells. Kolmogorov-Smirnov test used for significance test between APOE3 CTRL and APOE3 DOX. E. Cumulative distribution of only APOE4 control and doxorubicin-treated cells. Kolmogorov-Smirnov test used for significance test between APOE4 CTRL and APOE4 DOX. F.-G. Analysis of APOE3 CTRL (17 cells, 4 mice). APOE3 DOX (11 cells, 3 mice), APOE4 CTRL (17 cells, 3 mice), and APOE4 DOX (10 cells, 3 mice). F. Combined cumulative distribution plot showing firing frequency across increasing current injections (excluding inactivated interneurons) in APOE3 and APOE4 control and doxorubicin-treated animals. Kolmogorov-Smirnov test for significance between (APOE3 DOX vs. APOE4 DOX) and (APOE3 CTRL vs. APOE4 CTRL). G. Quantification of maximum action potential firing frequency in APOE3 and APOE4 doxorubicin and control groups. APOE3 CTRL (158 ± 11), APOE3 DOX (158 ± 13), APOE4 CTRL (111 ± 8), APOE4 DOX (119 ± 10). 2-way ANOVA with post hoc multiple comparisons test. H. Comparison of inward-rectification measure in control and doxorubicin-injected APOE3 and APOE4 mice. APOE3 DOX (0.41 ± 0.03), APOE4 DOX (0.41 ± 0.03), APOE3 CTRL (0.37 ± 0.03), APOE4 CTRL (0.35 ± 0.04). APOE3 DOX (18 cells, 3 mice), APOE4 DOX (24 cells, 3 mice). APOE3 CTRL (19 cells, 3 mice), APOE4 CTRL (25 cells, 3 mice). 2-way ANOVA with post hoc multiple comparison test. I. Distribution of APOE3 (top) and APOE4 (bottom) PV interneuron recordings showing inactivation block following control or doxorubicin injection.

https://doi.org/10.1371/journal.pone.0343276.g005

In all, doxorubicin did not affect PV interneuron maximum steady-state firing frequency or excitatory input. Under basal conditions in this experiment, there was a higher percentage of inactivated PV interneurons than in Fig 3, but with the same effect of APOE genotype (APOE3, 11% to 25%; APOE4, 25% to 50%). There was a further strong effect on PV interneuron inactivation block due to doxorubicin: treatment doubled the percentage of PV interneurons that inactivated in APOE3 (25% to 52%) without producing an effect in APOE4 (50% to 54%) (Fig 5I). Together, the doxorubicin-induced increases in inactivation support a reduction in inhibitory PV interneuron output in APOE3 but not APOE4 under normal physiological conditions. These data complement the previous electrophysiological findings in Fig 4 showing a homeostatic increase in inhibitory synaptic transmission in APOE3, but not APOE4, pyramidal cells.

Discussion

We found that APOE genotype affects the electrophysiological properties of pyramidal and PV interneurons in the EC at baseline and in response to an acute stressor (the chemotherapeutic agent doxorubicin). Our results led us to propose a model on which APOE3 genotype, but not APOE4 genotype, affords pyramidal cells the flexibility to modulate excitatory and inhibitory synaptic transmission. This modulation would compensate for stress-induced changes in PV interneuron inhibition and ultimately preserve the local circuit E/I balance to maintain the information flow necessary for cognition (Fig 6).

thumbnail
Download:
Fig 6. Proposed model of APOE and chemotherapy effects on pyramidal cell excitation.

APOE4 reduces PV interneuron firing rate and pyramidal cell GABAergic receptors, collectively impairing inhibitory tone in the EC and elevating E/I balance. When doxorubicin is introduced peripherally, it reduces the amount of inhibition APOE3 PV interneurons can contribute to the EC circuit, leading to a compensatory increase in pyramidal cell sIPSC and sEPSC amplitude that maintains the circuit E/I balance. Doxorubicin does not affect APOE4 PV cells, perhaps due to a ceiling effect where the circuitry can only physiologically tolerate a maximum level of PV cell inactivation. Doxorubicin also does not elicit any effect in APOE4 pyramidal cells, suggesting that the increase in excitatory and inhibitory synaptic transmission seen in APOE3 pyramidal cells may be involved in preserving cognition after chemotherapy treatment.

https://doi.org/10.1371/journal.pone.0343276.g006

The increase in sEPSC and sIPSC amplitude in ‘normal’ risk APOE3 may be reflective of homeostatic synaptic plasticity, a feedback mechanism where the levels or function of glutamatergic or GABAergic receptors at postsynaptic sites are adjusted to maintain stable neuronal activity. In the case of reduced inhibitory function (as was observed in APOE3 PV interneurons following doxorubicin treatment), GABAergic (GABAA) and glutamatergic (AMPA) receptor levels would respond in the direction that preserves E/I balance. Loss of inhibitory tone can either increase GABAA receptors or decrease GABAA or AMPA receptor plasticity in response to inhibitory blockades [49,50]. How GABAA and AMPA receptors coordinate their directional response to inhibitory changes within a circuit is less clear. In our study, we observed unidirectional increases in both GABA and AMPA transmission in APOE3 doxorubicin-treated mice, similar to findings of increased inhibitory and excitatory quantal amplitude following application of a GABAA receptor blocker [51]. We propose that our electrophysiological findings align with homeostatic synaptic plasticity mechanisms. Miniature synaptic current recordings would be required to test whether the changes in spontaneous postsynaptic current amplitude are due to increased receptor trafficking to the cell surface as opposed to alternative presynaptic factors (e.g., an increase in the number of presynaptic terminals activated or larger readily releasable vesicular pools).

Although neurons in APOE4 brains did not display the same receptor flexibility as those in APOE3 brains, the E/I balance overall remained the same after doxorubicin, raising a question about the importance of synaptic changes when the overall circuit output is unaltered. These results could simply reflect the degree of efficient information coding occurring within each circuit [52]. Increasing glutamatergic and GABAergic transmission during development can improve the signal-to-noise ratio and neuronal firing precision [53,54]. In the case of synaptic upscaling (a neuroplasticity response to persistent network inhibition), increasing both excitatory and inhibitory postsynaptic receptors allows circuits to remain sensitive to new information [55].

A major factor in the modulation of local E/I balance is the fast-spiking PV inhibitory neuron, whose contribution to the refinement and control of local inhibition is apparent from monocular deprivation during the critical period [56]. PV interneurons provide rapid feedforward and feedback inhibition to pyramidal cells, allowing them to regulate the spiking timing of pyramidal cells and enhance the network synchrony of gamma oscillations necessary for information processing, memory, and cognition [57]. These cells are densely connected to nearby pyramidal cell somas, giving them the unique ability to quickly control ensembles of excitatory neurons in a manner that maximizes information processing and cognitive performance [58,59]. A range of deficits in PV interneurons have been reported in pathological AD mouse models, including hyperexcitability, decreased cell density, and reduced firing rates [33,60,61]. Although little is known about its role in PV-specific functioning, APOE4 is broadly associated with inhibitory network dysfunction leading to brain hyperexcitability [23,48,62]. Changes to this inhibitory neuron subtype informs how stressors like APOE4 and doxorubicin affect the E/I balance in associated memory regions, contributing to cognitive impairment.

The role PV interneurons play in responding to CNS stressors is not well known. Factors such as brain region, age, sex, model, stressor type, and duration all influence PV interneurons responses, leading to conflicting evidence of what a ‘normal’ response to stress is within a microcircuit [63]. We found that the innate stress of the APOE4 allele in young female mice produced a striking decrease in PV interneuron firing frequency and an increase in the proportion of PV interneurons with inactivation block at baseline (Fig 3). The reduction in PV interneuron activity observed with the APOE4 genotype at baseline differs from reports of increased PV activity in other brain areas following chronic stress [64,65]. However, similar deficits in PV firing have been shown in young mouse models of AD, highlighting the nuance in stress-induced PV interneuron functioning [66,67]. Interestingly, doxorubicin treatment did not affect either firing frequency or PV inactivation in APOE4 brains, suggesting that there exists a ceiling effect at which APOE4 brains can no longer meet the functional requirements of the circuitry to preserve normal cognition (Fig 5I). Effects of APOE genotype on brain structure and activity are observed early in development [68]; likewise, cancer and its treatment in childhood can often impair cognition in adult survivors [69]. Our results demonstrate that even the subtle effects of the genetic risk factor APOE4, which codes for a single amino acid difference from the APOE3 allele, is sufficient to induce PV interneuron dysfunction early in life [70].

Differences in our treatment paradigms have led us to be conservative in our comparisons of the findings based solely on APOE genotype differences (Figs 2 and 3) with the findings that included doxorubicin and vehicle treatments (Figs 4 and 5). In particular, data in Fig 2 was collected with one APOE knock-in model [39], while that in Fig 4 was collected in another [40]. We used different models given the ready availability of a newly generated APOE knock-in model that presumably exhibited less genetic drift compared to those generated nearly 30 years ago [39]. We previously reported that these two APOE knock-in models share APOE4-related phenotypes of reduced apoE protein expression, reduced dendritic spines, and response to anti-inflammatory treatment [71]. The measured E/I ratio was higher in APOE4 pyramidal cells compared to APOE3 pyramidal cells in both models (Fig 2: 0.33 vs. 0.19; Fig 4: 0.29 vs. 0.22), but only statistically significant in Fig 2. Despite possible differences in these models, the elevated E/I ratio in APOE4 brains reproduces an earlier study [23] and is consistent with increased susceptibility of these APOE4 mice to seizures [72,73].

The work presented in this study has several major strengths and weaknesses. Strengths include the investigation of electrophysiological parameters in young APOE3 and APOE4 mice and the use of a model that combines genetic risk (APOE4) with environmental risk (doxorubicin) to study neuronal electrophysiology. We also assessed specific neuron subtypes, the pyramidal cell and the PV interneuron, providing cell-specific characterization of the EC microcircuitry. However, ex vivo measures do not always reflect more complicated in vivo systems which maintain more synaptic connections and are not affected by variables associated with euthanasia and maintenance of tissue health. Another limitation is the comparison of the effects of APOE genotype across conditions (i.e., Figs 2 and 3 compared to Figs 4 and 5), since the experiments were conducted in APOE knock-in mice derived from two sources and required different handling of mice due to exposure to doxorubicin and removal of hazardous waste. While these models behaved similarly in several ways, the magnitude of the APOE genotype effects were different. Our choice of CRCI model was limited in that there was no tumors or tumor resections in the mice. Furthermore, we used a single injection of only one chemotherapeutic agent, which does not model well the multiple rounds of treatment with chemotherapy combinations to target cancer types. Future research directions would be aimed at better understanding how these multi-modal approaches affect APOE4 circuit resiliency in the EC.

Conclusion

Overall, our work demonstrates that APOE4 and chemotherapy, risk factors for debilitating cognitive impairments in many cancer survivors, may promote early inhibitory deficits and impaired compensatory synaptic plasticity as early signs of brain dysfunction.

Supporting information

S1 Table. Passive and active properties of APOE3 and APOE4 PV interneurons.

https://doi.org/10.1371/journal.pone.0343276.s001

(DOCX)

S1 Figure. E/I ratio does not meaningfully correlate with days post-injection.

The number of days after doxorubicin injection before electrophysiological measures for individual mice is shown in the X axis. The values of the E/I ratios in EC neurons are shown on the Y axis (APOE4 CTRL (r = 0.41), APOE4 DOX (r = −0.10), APOE3 CTRL (r = 0.31), APOE3 DOX (r = −0.45)). The correlation was tested by Spearman’s rank correlation coefficient.

https://doi.org/10.1371/journal.pone.0343276.s002

(DOCX)

S2 Table. Summary of quantification of action potential-related parameters.

https://doi.org/10.1371/journal.pone.0343276.s003

(DOCX)

References

  1. 1. Jackson RJ, Hyman BT, Serrano-Pozo A. Multifaceted roles of APOE in Alzheimer disease. Nat Rev Neurol. 2024;20(8):457–74. pmid:38906999
  2. 2. Ahles TA, Saykin AJ, Noll WW, Furstenberg CT, Guerin S, Cole B, et al. The relationship of APOE genotype to neuropsychological performance in long-term cancer survivors treated with standard dose chemotherapy. Psychooncology. 2003;12(6):612–9. pmid:12923801
  3. 3. Mandelblatt JS, Small BJ, Luta G, Hurria A, Jim H, McDonald BC, et al. Cancer-Related Cognitive Outcomes Among Older Breast Cancer Survivors in the Thinking and Living With Cancer Study. J Clin Oncol. 2018;36(32):JCO1800140. pmid:30281396
  4. 4. Harrison RA, Rao V, Kesler SR. The association of genetic polymorphisms with neuroconnectivity in breast cancer patients. Sci Rep. 2021;11(1):6169. pmid:33731765
  5. 5. Demby T, Gross PS, Mandelblatt J, Huang JK, Rebeck GW. The chemotherapeutic agent doxorubicin induces brain senescence, with modulation by APOE genotype. Exp Neurol. 2024;371:114609. pmid:37944881
  6. 6. Demby TC, Rodriguez O, McCarthy CW, Lee Y-C, Albanese C, Mandelblatt J, et al. A mouse model of chemotherapy-related cognitive impairments integrating the risk factors of aging and APOE4 genotype. Behav Brain Res. 2020;384:112534. pmid:32027870
  7. 7. Patel C, Diglio F, Durham BH, Cole PD. Influence of ApoE genotype on doxorubicin-induced cognitive impairment in juvenile rats. Sci Rep. 2025;15(1):44620. pmid:41436539
  8. 8. Du J, Zhang A, Li J, Liu X, Wu S, Wang B, et al. Doxorubicin-Induced Cognitive Impairment: The Mechanistic Insights. Front Oncol. 2021;11:673340. pmid:34055643
  9. 9. Fowler CG, Tarantino MI, Gillett AE, Muñiz V, Anderson MA, Bonner RL, et al. Using a murine model to explore the impact of sex and APOE4 on cisplatin-induced cognitive impairment. Brain Behav Immun. 2025;129:1–14. pmid:40374126
  10. 10. Tonorezos E, Devasia T, Mariotto AB, Mollica MA, Gallicchio L, Green P, et al. Prevalence of cancer survivors in the United States. J Natl Cancer Inst. 2024;116(11):1784–90. pmid:39002121
  11. 11. Karimani F, Asgari Taei A, Abolghasemi-Dehaghani M-R, Safari M-S, Dargahi L. Impairment of entorhinal cortex network activity in Alzheimer’s disease. Front Aging Neurosci. 2024;16:1402573. pmid:38882526
  12. 12. Olajide OJ, Suvanto ME, Chapman CA. Molecular mechanisms of neurodegeneration in the entorhinal cortex that underlie its selective vulnerability during the pathogenesis of Alzheimer’s disease. Biol Open. 2021;10(1):bio056796. pmid:33495355
  13. 13. Hernández-Frausto M, Vivar C. Entorhinal cortex-hippocampal circuit connectivity in health and disease. Front Hum Neurosci. 2024;18:1448791. pmid:39372192
  14. 14. Lam NH, Borduqui T, Hallak J, Roque A, Anticevic A, Krystal JH, et al. Effects of Altered Excitation-Inhibition Balance on Decision Making in a Cortical Circuit Model. J Neurosci. 2022;42(6):1035–53. pmid:34887320
  15. 15. Wang Y, Meng W, Liu Z, An Q, Hu X. Cognitive impairment in psychiatric diseases: Biomarkers of diagnosis, treatment, and prevention. Front Cell Neurosci. 2022;16:1046692. pmid:36406755
  16. 16. Leskinen S, Alsalek S, Galvez R, Ononogbu-Uche FC, Shah HA, Vojnic M, et al. Chemotherapy-Related Cognitive Impairment and Changes in Neural Network Dynamics: A Systematic Review. Neurology. 2025;104(2):e210130. pmid:39804577
  17. 17. Burns AP, Duran MI, Fortel I, Lazarov O, Zhan L, Bendlin B, et al. Excitation–inhibition homeostasis in Alzheimer’s disease: a selective multiscale review of mechanisms, sex differences, and therapeutic opportunities. Neuropsychopharmacol. 2026.
  18. 18. Li G, Hsu L-M, Wu Y, Bozoki AC, Shih Y-YI, Yap P-T. Revealing excitation-inhibition imbalance in Alzheimer’s disease using multiscale neural model inversion of resting-state functional MRI. Commun Med (Lond). 2025;5(1):17. pmid:39814858
  19. 19. van Nifterick AM, Mulder D, Duineveld DJ, Diachenko M, Scheltens P, Stam CJ, et al. Resting-state oscillations reveal disturbed excitation-inhibition ratio in Alzheimer’s disease patients. Sci Rep. 2023;13(1):7419. pmid:37150756
  20. 20. Wishart HA, Saykin AJ, Rabin LA, Santulli RB, Flashman LA, Guerin SJ, et al. Increased brain activation during working memory in cognitively intact adults with the APOE epsilon4 allele. Am J Psychiatry. 2006;163(9):1603–10. pmid:16946187
  21. 21. Johnson LA. APOE and metabolic dysfunction in Alzheimer’s disease. Int Rev Neurobiol. 2020;154:131–51. pmid:32739002
  22. 22. Uzun C, Li Y, Liu S. APOE ε4 disrupts neuronal and network-level function in the anterior olfactory nucleus: Influence of age and sex. Alzheimers Dement. 2025;21(10):e70865. pmid:41157904
  23. 23. Nuriel T, Angulo SL, Khan U, Ashok A, Chen Q, Figueroa HY, et al. Neuronal hyperactivity due to loss of inhibitory tone in APOE4 mice lacking Alzheimer’s disease-like pathology. Nat Commun. 2017;8(1):1464. pmid:29133888
  24. 24. Whittaker AL, George RP, O’Malley L. Prevalence of cognitive impairment following chemotherapy treatment for breast cancer: a systematic review and meta-analysis. Sci Rep. 2022;12(1):2135. pmid:35136066
  25. 25. Chughtai S, Doyle D, Tata S, Ram D, Oymagil I. Chemotherapy-induced cognitive impairment: Mechanisms, emerging biomarkers, and therapeutic interventions. Biochem Biophys Res Commun. 2025;779:152456. pmid:40779978
  26. 26. Patel C, Glytsou C, Jang M-H, Cole PD. The effects of doxorubicin on blood-brain barrier integrity in hCMEC/D3. Neurotoxicology. 2025;111:103355. pmid:41265652
  27. 27. Cronin B, Kandel S, McElroy T, Syed S, Swinton C, Corley C, et al. Changes in gut, microbiome, and cognition after doxorubicin, cyclophosphamide, and paclitaxel chemotherapy treatment. Sci Rep. 2026;16(1):3820. pmid:41519984
  28. 28. Sekeres MJ, Bradley-Garcia M, Martinez-Canabal A, Winocur G. Chemotherapy-Induced Cognitive Impairment and Hippocampal Neurogenesis: A Review of Physiological Mechanisms and Interventions. Int J Mol Sci. 2021;22(23):12697. pmid:34884513
  29. 29. Usmani MT, Krattli RP Jr, El-Khatib SM, Le ACD, Smith SM, Baulch JE, et al. BDNF Augmentation Using Riluzole Reverses Doxorubicin-Induced Decline in Cognitive Function and Neurogenesis. Neurotherapeutics. 2023;20(3):838–52. pmid:36720792
  30. 30. Ng CAS, Pedus M, Lee MF, Kromer LF, Mandelblatt J, Rebeck GW. The chemotherapy agent doxorubicin induces CNS expression of Ascl1, a regulator of adult neurogenesis and differentiation. Sci Rep. 2025;15(1):9725. pmid:40118985
  31. 31. Budamagunta V, Kumar A, Rani A, Manohar Sindhu S, Yang Y, Zhou D, et al. Senolytic treatment alleviates doxorubicin-induced chemobrain. Aging Cell. 2024;23(2):e14037. pmid:38225896
  32. 32. Bjerke IE, Yates SC, Laja A, Witter MP, Puchades MA, Bjaalie JG, et al. Densities and numbers of calbindin and parvalbumin positive neurons across the rat and mouse brain. iScience. 2020;24(1):101906. pmid:33385111
  33. 33. Hijazi S, Smit AB, van Kesteren RE. Fast-spiking parvalbumin-positive interneurons in brain physiology and Alzheimer’s disease. Mol Psychiatry. 2023;28(12):4954–67. pmid:37419975
  34. 34. Wilbers R, Galakhova AA, Driessens SLW, Heistek TS, Metodieva VD, Hagemann J, et al. Structural and functional specializations of human fast-spiking neurons support fast cortical signaling. Sci Adv. 2023;9(41):eadf0708. pmid:37824618
  35. 35. Smeralda CL, Pandit S, Turrini S, Reilly J, Palmisano A, Sprugnoli G, et al. The role of parvalbumin interneuron dysfunction across neurodegenerative dementias. Ageing Res Rev. 2024;101:102509. pmid:39306248
  36. 36. Papanikolaou A, Graykowski D, Lee BI, Yang M, Ellingford R, Zünkler J, et al. Selectively vulnerable deep cortical layer 5/6 fast-spiking interneurons in Alzheimer’s disease models in vivo. Neuron. 2025;113(14):2265–79.e7. pmid:40345184
  37. 37. Deasy S, Amontree M, Colon Z, Thorland E, Modi K, Hummel K. Increased levels of HAPLN2, which anchors dense extracellular matrix, in the hippocampus of APOE4 targeted replacement mice. bioRxiv. 2025.
  38. 38. Colon ZA, Chan SC, Maguire-Zeiss KA. Age-related inflammatory changes and perineuronal net dynamics: implications for aging. J Neuroinflammation. 2025;22(1):274. pmid:41254697
  39. 39. Sullivan PM, Mezdour H, Aratani Y, Knouff C, Najib J, Reddick RL, et al. Targeted replacement of the mouse apolipoprotein E gene with the common human APOE3 allele enhances diet-induced hypercholesterolemia and atherosclerosis. J Biol Chem. 1997;272(29):17972–80. pmid:9218423
  40. 40. Foley KE, Hewes AA, Garceau DT, Kotredes KP, Carter GW, Sasner M, et al. The APOEε3/ε4 Genotype Drives Distinct Gene Signatures in the Cortex of Young Mice. Front Aging Neurosci. 2022;14:838436. pmid:35370604
  41. 41. Hevers W, Lüddens H. Pharmacological heterogeneity of gamma-aminobutyric acid receptors during development suggests distinct classes of rat cerebellar granule cells in situ. Neuropharmacology. 2002;42(1):34–47. pmid:11750914
  42. 42. Rowland DC, Obenhaus HA, Skytøen ER, Zhang Q, Kentros CG, Moser EI, et al. Functional properties of stellate cells in medial entorhinal cortex layer II. Elife. 2018;7:e36664. pmid:30215597
  43. 43. Dumanis SB, Tesoriero JA, Babus LW, Nguyen MT, Trotter JH, Ladu MJ, et al. ApoE4 decreases spine density and dendritic complexity in cortical neurons in vivo. J Neurosci. 2009;29(48):15317–22. pmid:19955384
  44. 44. Micheva KD, Kiraly M, Perez MM, Madison DV. Extensive Structural Remodeling of the Axonal Arbors of Parvalbumin Basket Cells during Development in Mouse Neocortex. J Neurosci. 2021;41(45):9326–39. pmid:34583957
  45. 45. Hu H, Gan J, Jonas P. Interneurons. Fast-spiking, parvalbumin⁺ GABAergic interneurons: from cellular design to microcircuit function. Science. 2014;345(6196):1255263. pmid:25082707
  46. 46. Hippenmeyer S, Vrieseling E, Sigrist M, Portmann T, Laengle C, Ladle DR, et al. A developmental switch in the response of DRG neurons to ETS transcription factor signaling. PLoS Biol. 2005;3(5):e159. pmid:15836427
  47. 47. Madisen L, Zwingman TA, Sunkin SM, Oh SW, Zariwala HA, Gu H, et al. A robust and high-throughput Cre reporting and characterization system for the whole mouse brain. Nat Neurosci. 2010;13(1):133–40. pmid:20023653
  48. 48. Najm R, Jones EA, Huang Y. Apolipoprotein E4, inhibitory network dysfunction, and Alzheimer’s disease. Mol Neurodegener. 2019;14(1):24. pmid:31186040
  49. 49. Gonzalez-Islas C, Sabra Z, Fong M-F, Yilmam P, Au Yong N, Engisch K, et al. GABAergic synaptic scaling is triggered by changes in spiking activity rather than AMPA receptor activation. Elife. 2024;12:RP87753. pmid:38941139
  50. 50. Rannals MD, Kapur J. Homeostatic strengthening of inhibitory synapses is mediated by the accumulation of GABA(A) receptors. J Neurosci. 2011;31(48):17701–12. pmid:22131430
  51. 51. Wilhelm JC, Wenner P. GABAA transmission is a critical step in the process of triggering homeostatic increases in quantal amplitude. Proc Natl Acad Sci U S A. 2008;105(32):11412–7. pmid:18678897
  52. 52. Chen L, Li X, Tjia M, Thapliyal S. Homeostatic plasticity and excitation-inhibition balance: The good, the bad, and the ugly. Curr Opin Neurobiol. 2022;75:102553. pmid:35594578
  53. 53. McKeon SD, Perica MI, Calabro FJ, Foran W, Hetherington H, Moon C-H, et al. Prefrontal excitation/inhibition balance supports adolescent enhancements in circuit signal to noise ratio. Prog Neurobiol. 2024;243:102695. pmid:39622336
  54. 54. Wlodarczyk AI, Xu C, Song I, Doronin M, Wu Y-W, Walker MC, et al. Tonic GABAA conductance decreases membrane time constant and increases EPSP-spike precision in hippocampal pyramidal neurons. Front Neural Circuits. 2013;7:205. pmid:24399937
  55. 55. Hofmann M, Mader P. Synaptic Scaling-An Artificial Neural Network Regularization Inspired by Nature. IEEE Trans Neural Netw Learn Syst. 2022;33(7):3094–108. pmid:33502984
  56. 56. Huang Y, Liu Z, Zhan Z, Zhang X, Gao L, Wang M, et al. Interactions between excitatory neurons and parvalbumin interneurons in V1 underlie neural mechanisms of amblyopia and visual stimulation treatment. Commun Biol. 2024;7(1):1564. pmid:39587348
  57. 57. Cernotova D, Hruzova K, Levcik D, Svoboda J, Stuchlik A. Linking social cognition, parvalbumin interneurons, and oxytocin in Alzheimer’s disease: an update. J Alzheimers Dis. 2023;96(3):861–75.
  58. 58. Booker SA, Vida I. Morphological diversity and connectivity of hippocampal interneurons. Cell Tissue Res. 2018;373(3):619–41. pmid:30084021
  59. 59. Packer AM, Yuste R. Dense, unspecific connectivity of neocortical parvalbumin-positive interneurons: a canonical microcircuit for inhibition? J Neurosci. 2011;31(37):13260–71. pmid:21917809
  60. 60. Ali F, Baringer SL, Neal A, Choi EY, Kwan AC. Parvalbumin-Positive Neuron Loss and Amyloid-β Deposits in the Frontal Cortex of Alzheimer’s Disease-Related Mice. J Alzheimers Dis. 2019;72(4):1323–39. pmid:31743995
  61. 61. Caccavano A, Bozzelli PL, Forcelli PA, Pak DTS, Wu J-Y, Conant K, et al. Inhibitory Parvalbumin Basket Cell Activity is Selectively Reduced during Hippocampal Sharp Wave Ripples in a Mouse Model of Familial Alzheimer’s Disease. J Neurosci. 2020;40(26):5116–36. pmid:32439703
  62. 62. Filippini N, MacIntosh BJ, Hough MG, Goodwin GM, Frisoni GB, Smith SM, et al. Distinct patterns of brain activity in young carriers of the APOE-epsilon4 allele. Proc Natl Acad Sci U S A. 2009;106(17):7209–14. pmid:19357304
  63. 63. Perlman G, Tanti A, Mechawar N. Parvalbumin interneuron alterations in stress-related mood disorders: A systematic review. Neurobiol Stress. 2021;15:100380. pmid:34557569
  64. 64. Bhatti DL, Medrihan L, Chen MX, Jin J, McCabe KA, Wang W, et al. Molecular and Cellular Adaptations in Hippocampal Parvalbumin Neurons Mediate Behavioral Responses to Chronic Social Stress. Front Mol Neurosci. 2022;15:898851. pmid:35813065
  65. 65. Page CE, Shepard R, Heslin K, Coutellier L. Prefrontal parvalbumin cells are sensitive to stress and mediate anxiety-related behaviors in female mice. Sci Rep. 2019;9(1):19772. pmid:31875035
  66. 66. Olah VJ, Goettemoeller AM, Rayaprolu S, Dammer EB, Seyfried NT, Rangaraju S. Biophysical K(v)3 channel alterations dampen excitability of cortical PV interneurons and contribute to network hyperexcitability in early Alzheimer’s. Elife. 2022;11.
  67. 67. Petrache AL, Rajulawalla A, Shi A, Wetzel A, Saito T, Saido TC, et al. Aberrant Excitatory-Inhibitory Synaptic Mechanisms in Entorhinal Cortex Microcircuits During the Pathogenesis of Alzheimer’s Disease. Cereb Cortex. 2019;29(4):1834–50. pmid:30766992
  68. 68. Meyer-Acosta KK, Diaz-Guerra E, Varma P, Aruk A, Mirsadeghi S, Perez AM, et al. APOE4 impacts cortical neurodevelopment and alters network formation in human brain organoids. bioRxiv. 2024.
  69. 69. Phillips NS, Stratton KL, Williams AM, Ahles T, Ness KK, Cohen HJ, et al. Late-onset Cognitive Impairment and Modifiable Risk Factors in Adult Childhood Cancer Survivors. JAMA Netw Open. 2023;6(5):e2316077. pmid:37256617
  70. 70. Frieden C, Garai K. Structural differences between apoE3 and apoE4 may be useful in developing therapeutic agents for Alzheimer’s disease. Proc Natl Acad Sci U S A. 2012;109(23):8913–8.
  71. 71. Sepulveda J, Luo N, Nelson M, Ng CAS, Rebeck GW. Independent APOE4 knock-in mouse models display reduced brain APOE protein, altered neuroinflammation, and simplification of dendritic spines. J Neurochem. 2022;163(3):247–59. pmid:35838553
  72. 72. Hunter JM, Cirrito JR, Restivo JL, Kinley RD, Sullivan PM, Holtzman DM, et al. Emergence of a seizure phenotype in aged apolipoprotein epsilon 4 targeted replacement mice. Brain Res. 2012;1467:120–32. pmid:22682924
  73. 73. Lamoureux L, Marottoli FM, Tseng KY, Tai LM. APOE4 Promotes Tonic-Clonic Seizures, an Effect Modified by Familial Alzheimer’s Disease Mutations. Front Cell Dev Biol. 2021;9:656521. pmid:33796539