Altered Odor-Evoked Electrophysiological Responses in the Anterior Piriform Cortex of Conscious APP/PS1 Mice

Abstract

Background:

Olfactory decline is an indicator of early-stage Alzheimer’s disease (AD). Although the anterior piriform cortex (aPC) is an important brain area involved in processing olfactory input, little is known about how its neuronal activity is affected in early-stage AD.

Objective:

To elucidate whether odor-induced electrophysiological responses are altered in the aPC of 3-5-month-old APP/PS1 mice.

Methods:

Using head-fixed multi-channel recording techniques in APP/PS1 AD mouse model to uncover potential aberrance of the aPC neuronal firing and local field potential (LFP) in response to vanillin.

Results:

We show that the firing rate of aPC neurons evoked by vanillin is significantly reduced in conscious APP/PS1 mice. LFP analysis demonstrates reduced low- and high-gamma (γ_low, γ_high) oscillations during both the baseline and odor stimulation periods in APP/PS1 mice. Moreover, according to spike-field coherence (SFC) analysis, APP/PS1 mice show decreased coherence between odor-evoked spikes and γ_low rhythms, while the coherence with γ_high rhythms and the ΔSFC of the oscillations is unaffected. Furthermore, APP/PS1 mice show reduced phase-locking strength in the baseline period, such that there is no difference between baseline and odor-stimulation conditions. This contrasts markedly with wild type mice, where phase-locking strength decreases on stimulation.

Conclusion:

The abnormalities in both the neuronal and oscillatory activities of the aPC may serve as electrophysiological indicators of underlying olfactory decline in early AD.

Keywords

Alzheimer’s disease anterior piriform cortex APP/PS1 firing rate head-fixed recording olfaction oscillations

INTRODUCTION

The piriform cortex, an association cortex, is divided into anterior (aPC) and posterior subregions. The aPC mainly receives afferents from the olfactory bulb (OB) and plays a critical role in olfactory learning and odor coding [1 –3]. Odor input can be divided into two categories, namely active and passive odor stimulation, where the latter is received when the subject is not actively sniffing, but where the odor is transmitted by an airflow [4]. A neuroimaging study by functional magnetic resonance imaging in humans reported that aPC signals increase when odors are actively detected, but decrease during passive odor stimulation [5, 6]. Thus, the multiplex coding patterns in the aPC depend on how the odor is applied.

In rodents, gamma (γ) oscillation activity in the OB is associated with odor perception and identification [7, 8]. Moreover, patients with early-stage Alzheimer’s disease (AD) show impaired olfaction before the appearance of cognitive deficits [9, 10]. We and others have shown that abnormal γ oscillatory activity in the OB correlates with olfactory dysfunction in AD mouse models, such as the abnormalities in reaching the buried food and cookie-finding behavior dependent on their olfactory system [11, 12]. Although there is some controversy about whether γ oscillations in the OB increase or decrease during odor sampling, how the olfactory signal for odor encoding is transmitted from the OB to the PC is well characterized [12, 13]. γ oscillations in the aPC are involved in the encoding of odorant features [14], but little is known about how such γ oscillations might develop and the degree of phase-locking with neuronal spiking during the process of odor sampling under both normal and AD conditions. In particular, whether there is an abnormal neural response to odor in the aPC of conscious AD-model mice has not been investigated.

The present study evaluated neuronal and oscillatory properties in the aPC of APP/PS1 mice at 3–5 months, an age at which they are known to exhibit olfactory deficits [12]. Head-fixed multi-channel recordings were conducted and odor-evoked responses were recorded in conscious mice. We found reduced odor-induced neuronal firing in the aPC of APP/PS1 mice: there were reductions in oscillatory power and SFC in the γ range, as well as abnormal aPC neuronal spiking, which was phase-locked to the low-gamma (γ_low) band during vanillin inhalation. The results suggest that decreased neuronal activity and abnormal phase-locking of aPC neuronal firing with γ_low oscillations might contribute to olfactory dysfunction in young AD mice.

MATERIALS AND METHODS

Animals

Experiments were performed in APP/PS1 transgenic mice (B6C3-Tg(APPswe, PSEN1dE9)85Dbo/Mmjax) and littermate wild-type (WT) control mice (Model Animal Research Center of Nanjing University (Nanjing, China). Mice were housed following the guidelines of the National Institutes of Health on animal care and the ethical guidelines of the Ethics Committee for Animal Research at Guangzhou University. Mice were housed at 23–25°C at 50–60% humidity with a 12-h light/dark cycle. Food and water were freely available. The genotype was confirmed using the standard PCR protocol (genotyping protocol ID, 23370) from The Jackson Laboratory (https://www.jax.org/Protocol?stockNumber=004462&protocolID=23370). The experiments used only male mice to avoid the sex difference in AD-related olfactory decline, such as the influence by gonadal hormones in females [15], and to provide with further mechanism in support of our previous findings which used male mice only [12]. Mice aged 3–5 months from each group were used in the electrophysiological experiments and staining of amyloid plaques, and 13-month-old mice were used as positive control of amyloid plaques staining since AD mice under 5 months of age do not have amyloid plaques [16, 17].

Odor stimulation

Conscious mice were head-fixed with the horizontal bar of a stereotaxic apparatus using the screw to fix its head-plate with 1 cm gap between mouse’s snout and the outlet of odor delivery but was allowed to move on the running wheel (Fig. 1A). Vanillin, which served as the odor stimulation in present study, is commonly used for odor conditioning during olfactory testing and can evoke neuronal activity in the olfactory system [18, 19]. Vanillin was dissolved in distilled water at 1% (w/v) dilution and used as a passively inhaled odor. The odor-stimulation event markers induced by the custom-made air pump were synchronously recorded in the electrophysiological data acquisition system. Initially, a constant clean airflow (room air, no vanillin) at a rate of 1 L/min was given during a 20-s baseline LFP recording, and this was then followed by a 1-s odor stimulation. For each mouse, four trials were made with inter-trial interval of 81 s (20 s baseline + 1 s odor stimulation + 60 s post-odor stimulation) (Fig. 1B). Signals corresponding to a 1-s baseline trace and the 1-s post-odor response in each trial were used for the analysis.

Fig. 1

The vanillin odor-evoked responses of aPC neurons in APP/PS1 and WT mice. A) Diagram showing the experimental setting of in vivo recordings in the aPC of a conscious mouse (left), with details of the multi-channel electrode tip (right). B) Experimental schedule. There were 4 trials separated by 80 s. Each trial was composed of 20 s baseline recording, 1 s odor stimulation, and 60 s post-test period. C-E) Three representative examples of firing rate induced by vanillin odor for APP/PS1 and WT littermates. From left to right, the excitatory response (C), inhibitory response (D), and no response (E); top, raster plots representative of aPC neuronal activity in both WT and APP/PS1 mice; bottom, peri-stimulus time histograms for the firing rate. The light blue shaded area indicates the period of odor stimulation (excitatory response: n = 51 neurons/7 WT mice, n = 40 neurons/7 APP/PS1 mice; inhibitory response: n = 16 neuron/7 WT mice, n = 28 neurons/7 APP/PS1 mice; no response: n = 8 neurons/7 WT mice, n = 7 neurons/7 APP/PS1 mice). F) Cumulative probability of the firing rate of the vanillin odor-evoked excitatory response for the two groups (left), mean firing rate of the excitatory response during the vanillin stimulation period for the two groups (right) (n = 51 neurons/7 WT mice, n = 40 neurons/7 APP/PS1 mice). G) Cumulative probability of vanillin odor-evoked inhibitory responses (left), mean firing rate of the inhibitory response during the vanillin stimulation period (n = 16 neurons/7 WT mice, n = 28 neurons/7 APP/PS1 mice). H) Cumulative probability showing cases of no response evoked by vanillin odor (n = 8 neurons/7 WT mice, n = 7 neurons/7 APP/PS1 mice). I) A recording site labeled by DiI dye in the aPC of a conscious mouse. J) Representative raw traces of aPC neurons in response to vanillin. K, L) Sample waveforms (top) of an aPC neuron in response to vanillin, and respective autocorrelograms (middle) and interspike interval distributions (bottom). The shallowed areas in C-E indicate the SEM. Values represent mean±SEM. ^**p < 0.01 and n.s., not significant.

In vivo electrophysiology

Surgical procedures

Mice were anesthetized by intraperitoneal (i.p.) injection of pentobarbital (Sigma, USA; 80 mg/kg) and then mounted in a stereotaxic apparatus (RWD Life Science, China). Rectal temperature was maintained at 37°C using a temperature controller (Harvard Apparatus, USA). The anesthesia level was maintained by supplemental doses of pentobarbital (20 mg/kg/h) whenever necessary. For head-fixed multi-channel recordings, we cemented a custom-machined aluminum head-plate onto the skull using tissue adhesive (3M, USA). After fitting, the mice were treated with a gel containing lincomycin hydrochloride and lidocaine hydrochloride (anti-inflammation and pain relief drugs) [20]. After returning to home cages, the mice were treated daily with above gel. Over the following week, mice were habituated to the head-fixing set-up and were positioned on a running wheel where they could move the wheel freely. Mice were head-fixed and anesthetized using 1-2% isoflurane (RWD Life Science, China) for surgery. A craniotomy (∼1.5 mm diameter) was drilled on the skull on top of left side aPC (coordinates: bregma, +1.4; lateral, –2.7; electrode insertion depth 3.5 mm) [21]. The surface of the brain was covered with 1% agarose. After the surgery, lidocaine hydrochloride and lincomycin hydrochloride gel were used to relieve pain. Mice were given at least 1 h to recover from isoflurane followed by head-fixed recordings using a multi-channel electrode probe (NeuroNexus, USA; Model A 1×16 probe).

Data acquisition

Recorded signals were acquired, double-blind, in 3-5-month-old mice (n = 7 for each group) via the Cerebustrademark Data Acquisition System (Blackrock Microsystems, USA) sampled at 30 kHz. LFPs were filtered 1–100 Hz and sampled at 2 kHz. Based on Thomson’s regression method, the removal of 50 Hz line noise was accomplished with the “rmlinesc” function in the Chronux toolbox (http://chronux.org/). By doing so the sinusoids in raw signals were removed when power exceeds a criterion in the F distribution [22]. Spiking signals were obtained by the high-pass filtered (≥250 Hz) field potential signals sampled at 30 kHz, and sorted by Offline sorter software (Plexon, USA) using a κ-means method based on waveform features (waveform peak amplitude and energy, the first principal component, etc.).

Power spectrum analysis

The power spectra for each trial of an individual mouse were normalized by calculating the normalized LFP signals (z-scored) as previously described [23]. The power spectrum of the normalized LFP signals (1–100 Hz) was computed in NeuroExplorer, using a multi-taper method with 3–5 tapers [24]. For the time-frequency spectral analysis, the baseline oscillations and 1 s odor stimulation were used with a 0.4 s sliding window. Delta and theta (δ+θ, 1–12 Hz), beta (β, 12–30 Hz), low-gamma (γ_low, 30–70 Hz), and high-gamma (γ_high, 70–100 Hz) rhythms were obtained by an infinite impulse response (IIR) elliptic filter, and the mean power of each specific rhythm was obtained by averaging across the frequency bands. The 1-s LFPs were extracted from pre-and post-odor-stimulus to evaluate the baseline and odor-induced power spectra after normalization. Differences of oscillatory power before and after vanillin were characterized as Δpower, where Δpower = power_response –power_baseline.

Firing rate analysis

The peristimulus time histogram was obtained from a 5-s time window displaying the spike activity between 2 s before and 3 s after onset of the 1 s odor stimulus. Spikes of each trial were extracted over 50 ms bins. To evaluate the statistical significance of the odor-evoked responses, the Wilcoxon matched-pairs signed rank test was applied to determine the 1 s baseline firing rate (1 s prior to odor stimulation) with the 1 s vanillin odor-evoked firing rate across all trials for each neuron recorded. p < 0.05 was taken to show significance for both excitatory and inhibitory responses: for an excitatory response, the odor evoked a higher firing rate compared to baseline. For inhibitory response, the odor evoked a lower firing rate compared to baseline. The mean firing rates for each neuron in both baseline and odor stimulation were averaged to calculate the frequency of spikes.

Spike-field coherence analysis

The spike-field coherence (SFC) of odor induced excitatory response and the filtered LFP (1–100 Hz) during baseline or odor stimulation was calculated by NeuroExplorer with a multi-taper method (TW = 3, K = 5). The SFC was computed as the averaged SFC of each frequency band. The ΔSFC (ΔSFC = SFC_response –SFC_baseline) was computed as the difference between baseline and odor-induced SFC after normalization.

Phase-locking firing analysis

The γ_low range (30–70 Hz) was filtered by using IIR elliptic filter in NeuroExplorer. The amplitude and phase of γ_low oscillations were computed using a Hilbert transform in MATLAB. A Rayleigh’s test (for which the firing rate of an individual neuron during each period comprised at least 50 spikes) was applied to evaluate the significance (p < 0.05) of phase-locking between spike and LFP; where a significant result was obtained, we then referred to “phase-locked” neurons, as previously described [25]. The mean resultant length (MRL), which represents phase-locking strength between the spikes and LFPs in the aPC, was analyzed as reported [26].

Histology

Verification of electrode location

Before the recordings, the multichannel electrode probe tip was painted with DiI dye (Invitrogen, USA; 1μg/μl, dissolved in ethanol). After the DiI dye had dried on the tip, the recording was performed using a multichannel electrode probe as described above. When the recording was complete, we transcardially perfused mice with saline followed by 4% paraformaldehyde (PFA). Then, mouse brains were dissected, fixed with 4% PFA at 4°C overnight and stored in 30% sucrose for three days after which 30μm sections were cut with a freezing microtome (Leica, USA) held at –22°C. The aPC slices were stained with DAPI (Beyotime, China) and recording sites in each section were determined using a Nikon Eclipse Ni fluorescence microscope (Nikon, Japan).

Amyloid plaque staining

Brains were placed in 0.01 M phosphate buffered saline (PBS) containing 30% sucrose at 4°C for three days. Then, 30μm aPC sections were obtained by a freezing microtome (Leica, USA) at –22°C. Sections were rinsed in PBS (3×5 min) and then blocked with 5% bovine serum albumin (BSA) in PBS containing 0.3% Triton X-100 solution (PBS-T) for 60 min at room temperature (RT). Next, sections were incubated overnight with mouse monoclonal anti-Aβ 4G8 (BioLegend, USA) primary antibody in 0.3% PBS-T at 4°C, washed with PBS (3×5 min) and then incubated for 60 min with secondary antibody (Cy3-conjugated goat anti-rabbit IgG; Life Technologies, USA) in 0.3% PBS-T containing 1% BSA at RT. Sections were mounted on glass slides and DAPI containing anti-fade solution was used to add cover slips. Images were captured by a confocal laser scanning microscope (Zeiss, Germany). The staining was performed on three non-consecutive aPC sections from 5-month-old or 13-month-old APP/PS1 and WT littermates; sections from 13-month-old APP/PS1 mice were used as positive controls.

Statistical analysis

Statistical testing was conducted using OriginPro (version 2018; OriginLab, USA) and custom-programmed MATLAB scripts. All data presented as mean±standard error (SEM) unless otherwise indicated. Data that did not conform to a normal distribution were analyzed using the Kolmogorov-Smirnov test (K-S test) for two samples, the Kruskal-Wallis test for multiple comparisons, and the Wilcoxon test for single comparisons. Otherwise, normally distributed data were analyzed by an unpaired t-test for two samples, and by ANOVA tests for multiple comparisons. A chi-squared test was used for categorical data. Points shown in the plots of LFP results are the average of multiple trials for each mouse (Fig. 2), while points in the plots of neuronal firing rate and the relationship between LFP and neuronal spikes are displayed as the average of multiple trials for each neuron (Figs. 1, 3, 4). ^*p < 0.05, ^**p < 0.01, ^***p < 0.001 and ns, not significant.

Fig. 2

Comparison of aPC γ oscillations in APP/PS1 and WT mice. A) Example power spectra of 1-s vanillin odor-evoked γ oscillations in the aPC of both APP/PS1 and WT littermates resulting from analysis of the LFP data after removal of 50-Hz line noise. The red dotted line at time 0 indicates the beginning of odor stimulation. B) Representative oscillatory traces in the aPC of conscious APP/PS1 (red) and WT (black) littermates. The blue dotted line indicates the beginning of odor stimulation. C-F) For both baseline and vanillin odor-evoked response periods, the oscillatory power in the γ_high and γ_low ranges is reduced in APP/PS1 mice, as shown by computing the LFP signals after removal of 50-Hz line noise (n = 7 mice per group). G, H) The changes of Δ oscillatory power in different bands is shown in APP/PS1 mice. The Δ power is the difference of the LFP signals between baseline and response periods after removal of 50-Hz line noise (n = 7 mice per group). The shallowed areas in C, E, and G indicate the SEM. Values represent mean±SEM. ^*p < 0.05, ^**p < 0.01, ^***p < 0.001 and n.s., not significant.

Fig. 3

Comparison of SFC in the γ_low-band in APP/PS1 and WT mice. A) The SFC curve of WT and APP/PS1 mice during the baseline period. Frequency of SFC curve is shown from 1 to 100 Hz (left) and zoomed in on the γ_low oscillation range from 30 to 70 Hz (right). B) In the baseline period, the SFC between spikes and LFPs is reduced in the γ_low oscillation range in APP/PS1 mice (n = 51 neurons/7 WT mice, n = 39 neurons/7 APP/PS1 mice). C) The SFC curve of WT and APP/PS1 mice during the vanillin odor-evoked response period. Frequency of SFC curve is displayed from 1 to 100 Hz (left), and zoomed in on the γ_low oscillation range from 30 to 70 Hz (right). D) In the vanillin odor-evoked response period, the SFC between spikes and LFPs is reduced in the γ_low oscillation range in APP/PS1 mice (n = 51 neurons/7 WT mice, n = 39 neurons/7 APP/PS1 mice). E) The ΔSFC curve between vanillin odor-evoked response and baseline periods of both WT and APP/PS1 mice. Frequency of ΔSFC curve is shown from 1 to 100 Hz (left), and zoomed in on the γ_low oscillation range from 30 to 70 Hz (right). F) No changes in the ΔSFC across all frequency ranges in the aPC of unanesthetized APP/PS1 and WT littermates (n = 51 neurons/7 WT mice, n = 39 neurons/7 APP/PS1 mice). The shallowed areas in A, C, and E indicate the SEM. Values represent mean±SEM. ^***p < 0.001 and n.s., not significant.

Fig. 4

Comparison of the phase-locking of aPC neuronal spiking to field γ_low rhythm in APP/PS1 and WT mice. A) The percentage of phase-locked neurons during baseline and vanillin odor-induced excitatory response periods in APP/PS1 and WT littermates (Baseline: phase-locked, n = 43 neurons/7 WT mice; non-phase-locked, n = 8 neurons/7 WT mice; phase-locked, n = 20 neurons/7 APP/PS1 mice; non-phase-locked, n = 19 neurons/7 APP/PS1 mice. Response: phase-locked, n = 23 neurons/7 WT mice; non-phase-locked, n = 28 neurons/7 WT mice; phase-locked, n = 6 neurons/7 APP/PS1 mice; non-phase-locked, n = 33 neurons/7 APP/PS1 mice). B) Representative phase vector plots in the aPC of conscious APP/PS1 (red) and WT (black) littermates during baseline and vanillin odor-evoked periods. The blue arrow indicates the mean phase vector. C) Mean resultant length (MRL) of phase-locking to γ_low rhythm in APP/PS1 and WT mice during baseline and odor-evoked periods (baseline: n = 46 neurons/7 WT mice, n = 29 neurons/7 APP/PS1 mice; response: n = 32 neurons/7 WT mice, n = 17 neurons/7 APP/PS1 mice). Values represent mean±SEM. ^***p < 0.001 and n.s., not significant.

RESULTS

Decreased vanillin-evoked neuronal firing rate in the aPC of APP/PS1 mice

We have previously shown that impaired olfaction is a marker of early-stage AD mice [12, 27]. However, the mechanisms underlying neural coding in the olfactory network of the aPC as it responds to an odor stimulus, and how these mechanisms might be affected in AD mice, remain unclear. To begin to address this, we used head-fixed multi-channel recordings in mice to evaluate the firing rate in the aPC in response to vanillin stimulation (Fig. 1A, B). By comparing baseline firing with the vanillin-evoked firing rate in aPC neuron, odor-induced neuronal responses could be sorted into excitatory and inhibitory responses as previously described [28]. In both WT and APP/PS1 mice, aPC neurons demonstrated excitatory, inhibitory, or no responses to vanillin stimulation (Fig. 1C–E). The cumulative probability of the firing rate in aPC neurons showing an excitatory response to vanillin revealed that vanillin induced lower firing rates in APP/PS1 mice (Fig. 1F, J–L) (K-S test, D₍₈₉₎ = 0.37, p = 4.7×10^–3). Moreover, we observed similar aPC inhibitory neuronal firing and number of no response units in responding to vanillin between WT and APP/PS1 mice (Fig. 1G, H) (K-S test, inhibitory response: D₍₄₂₎ = 0.18, p = 0.9; no response: D₍₁₃₎ = 0.38, p = 0.51). Verification of recording site was subsequently performed after each experiment (Fig. 1I). These findings suggest decreased excitability of aPC neurons in response to odor stimulation in APP/PS1 mice.

Reduced power of γ oscillations in response to vanillin in APP/PS1 mice

We next focused on γ oscillations in the aPC because these can be evoked by odor inhalation [29] and γ-range abnormalities have been observed in AD brains of humans and rodents [30, 31]. We evaluated aPC neural activity evoked by vanillin odor at the population level, and aPC LFPs in the presence or absence of vanillin were obtained in conscious WT and APP/PS1 mice. LFP power spectrograms filtered for the γ-range band are shown in Fig. 2A, with representative LFP traces shown in Fig. 2B. Spectral analysis highlighted remarkable power decreases in the γ_low (30–70 Hz) and γ_high (70–100 Hz) frequency ranges even for baseline activity in the aPC of APP/PS1 mice (Fig. 2C, D) (unpaired t-test, δ+θ: t₍₁₂₎ = 0.05, p = 0.85; β: t₍₁₂₎ = 0.2, p = 0.85; γ_low: t₍₁₂₎ = 3.46, p = 4.7×10^–3; γ_high: t₍₁₂₎ = 2.31, p = 0.04). We also observed a decrease in the power spectral density (PSD) of both the γ_low and γ_high rhythms in APP/PS1 mice during vanillin stimulation (Fig. 2E, F) (unpaired t–test, δ+θ: t₍₁₂₎ = 0.35, p = 0.73; β: t₍₁₂₎ = 0.28, p = 0.78; γ_low: t₍₁₂₎ = 2.44, p = 0.03; γ_high: t₍₁₂₎ = 2.48, p = 0.03). To determine whether an alteration in PSD might be associated with decreased excitability in APP/PS1 mice, we compared the Δpower, which represents the difference in power between the odor-evoked response and the baseline. Interestingly, we observed a significant decrease in Δpower in the γ_low range, but not the γ_high range, in the aPC of APP/PS1 mice (Fig. 2G, H) (unpaired t-test, δ+θ: t₍₁₂₎ = 0.84, p = 0.42; β: t₍₁₂₎ = 1.15, p = 0.27; γ_low: t₍₁₂₎ = 4.84, p = 4×10^–4; γ_high: t₍₁₂₎ = 1.03, p = 0.32). Thus far, the data suggest that neural oscillatory activity, as well as excitability, is impaired in the aPC of young AD mice.

Weakened coherence between γ_low oscillations and spikes in APP/PS1 mice

SFC is a measure of the synchrony between individual neuronal spiking and bulk network activity (oscillations). Enhanced SFC is an indicator of cognitive and sensory processing [32, 33]. We thus computed the coherence between vanillin-evoked aPC neuronal firing and LFPs. Baseline activity showed a significantly decreased coherence in the aPC of APP/PS1 mice in the γ_low range (Fig. 3A, B) (K-S test, δ+θ: D₍₈₈₎ = 0.13, p = 0.88; β: D₍₈₈₎ = 0.14, p = 0.78; γ_low: D₍₈₈₎ = 0.51, p = 2.5×10^–5; γ_high: D₍₈₈₎ = 0.26, p = 0.11). Similarly, reduced SFC of γ_low rhythms in the aPC of APP/PS1 mice was observed during vanillin stimulation (Fig. 3C, D) (K-S test, δ+θ: D₍₈₈₎ = 0.14, p = 0.81 β: D₍₈₈₎ = 0.14 p = 0.81 γ_low: D₍₈₈₎ = 0.42, p = 8×10^–4; γ_high: D₍₈₈₎ = 0.18, p = 0.5). To determine whether a difference in SFC presents between groups, we next calculated the ΔSFC in the aPC of APP/PS1 and WT mice, and observed no differences in ΔSFC over all frequency ranges tested (Fig. 3E, F) (unpaired t-test, δ+θ: t₍₈₈₎ = 0.58, p = 0.57; β: t₍₈₈₎ = 0.62, p = 0.54; γ_low: t₍₈₈₎ = 0.22, p = 0.83; γ_high: t₍₈₈₎ = 0.39, p = 0.7). The results indicate that, although coherence between aPC neuronal spiking and γ_low oscillations was impaired in APP/PS1 mice under both baseline and odor-sampling conditions (B, DFig. 3), the unchanged ΔSFC may indicate relatively normal coherent activity in APP/PS1 mice.

Impaired phase-locking of aPC neuronal spiking with γ_low rhythm in APP/PS1 mice

The γ oscillations in many brain regions are evoked by sensory stimuli [34], and γ-band oscillatory activity that is phase-locked to neuronal spiking is observed during the odor-sampling period [35]. We therefore evaluated the strength of spikes that were phase-locked to γ_low oscillations by measuring the resultant length of phase distribution between groups for both baseline and vanillin-stimulation situations. The results revealed that APP/PS1 differed significantly from WT mice in the percentage of phase-locked versus non-phase-locked aPC neurons in the presence or absence of vanillin (Fig. 4A) (Chi-squared test, WT versus APP/PS1 in baseline period: χ²_(1,88) = 11.48, p = 7×10^–4; WT versus APP/PS1 in response period: χ²_(1,88) = 8.93, p = 2.8×10^–3). Representative phase vector plots show the aPC phase-locked neurons in both WT and APP/PS1 mice for the baseline and odor sampling periods (Fig. 4B). We found that the MRL value differed significantly between baseline and odor stimulation in WT mice, but there was no difference between the two conditions in APP/PS1 mice (Fig. 4C) (Kruskal-Wallis test with Dunn’s post hoc tests, χ²₍₃₎ = 34.08, WT: baseline versus response, p = 9.2×10^–5; APP/PS1: baseline versus response, p = 1). Moreover, the baseline MRL value was significantly reduced in APP/PS1 compared to WT mice (Fig. 4C) (baseline: WT versus APP/PS1, p = 7.3×10^–5). Thus far, the data suggest that the Δpower of γ_low oscillations is positively correlated with the firing rate of aPC neurons in response to vanillin (Figs. 1F, 2H). Therefore, a smaller Δpower amplitude of γ_low oscillations may underlie the reduction in aPC neuronal firing (Figs. 1F, 2H), leading to decreased strength of aPC phase-locking with γ_low oscillations under baseline conditions in APP/PS1 mice (Fig. 4C), and subsequently unchanged MRL value between baseline and vanillin stimulation (Fig. 4C).

No Aβ plaques in the aPC of 5-month-old APP/PS1 mice

Because a previous study showed that amyloid-beta (Aβ) accumulation in the brain increases electrophysiological alterations in AD transgenic mice [36], we first examined whether or not Aβ plaques contributed to the electrophysiological deficits seen in the aPC of 5-month-old APP/PS1 mice. Using 4G8 staining to detect any amyloid plaques, we did not observe Aβ deposition in the aPC at this age in either APP/PS1 mice or WT littermates (Fig. 5A, B). Our and others’ results suggested that AD mice at 5 months of age did not show Aβ plaques [12, 16], we further used 13-month-old APP/PS1 mice as positive control to test the staining method used in this study. We observed robust Aβ deposition in the aPC of 13-month-old APP/PS1 mice (Fig. 5C, D). These results suggest that alterations in the spike phase-locking of γ_low oscillations may be a useful electrophysiological indicator in the diagnosis of early AD prior to Aβ plaque deposition.

Fig. 5

Comparison of aPC Aβ plaques in APP/PS1 and WT mice. A, B) The level of Aβ plaques in 5-month-old APP/PS1 and WT littermates (n = 3 mice per group). C, D) The level of Aβ plaques in 13-month-old APP/PS1 and WT littermates (n = 3 mice per group). aPC, anterior piriform cortex; ACC, anterior cingulate cortex; MC, motor cortex; SC, somatosensory cortex; GC, gustatory cortex; IC, insular cortex; lot, lateral olfactory tract; OT, olfactory tubercle; MS/vDB, medial septum/vertical limb of the diagonal band; LS, lateral septum; Str, striatum; NAc, nucleus accumbens; ccg, genu of corpus callosum.

DISCUSSION

Olfaction is a key perception and is associated with multiple cognitive behaviors. Olfactory decline has been proposed as an early indicator of AD in both humans and rodent models of AD [27, 37]. Odor-exposure evokes activity in PC neurons, actively or passively, in urethane-anesthetized [38] and conscious [6] mice, suggesting dynamic coding of PC neurons by an odor stimulus. We uncovered, in the present study, alterations in vanillin-induced aPC neuronal firing associated with abnormal γ_low oscillatory activity and coherence between aPC firing and LFPs in 3-5-month-old APP/PS1. Our results and analytical methods are in line with observations in the literature showing that diagnostic and therapeutic targets develop in AD patients before Aβ pathology becomes apparent.

Neuronal spiking reflects communication between neurons and the response to an odor stimulus [39]. In the present study, each mouse was allowed to adapt to the pertaining room conditions, which are not odor-free (i.e., the mice underwent olfactory adaptation), before recording. Normal room air was used to establish a control baseline before and during odor application to stably assess the response to odor. We performed 1 s vanillin stimulation and removed the odor after stimulation using a recovery device. Some residual odor will remain in the nasal cavity of all animals recorded, but the effect of this residue is likely to be similar and minimal across all mice. Consistent with previous reports revealing that odor induces a neuronal response in the aPC [40], we showed here that a vanillin-evoked aPC neuronal response comprised both excitatory and inhibitory firing in WT and APP/PS1 mice. We further showed that aPC neuronal firing is significantly reduced in 3-5-month-old APP/PS1 mice during a vanillin stimulus, indicating an impaired odor-induced aPC neuronal response in early-stage AD mice. Oscillations of neuronal network activity, measured by LFP, represent a temporal pattern of neuronal spiking and underlie the information coding between neural populations [41]. We found that the oscillatory power of γ_low and γ_high oscillations in the aPC is significantly reduced in APP/PS1mice under both baseline and odor-stimulation conditions, suggesting an impairment of aPC γ rhythms in both the absence and presence of odor. The change in oscillatory power represents the efficacy of the stimulus and the strength of the response [42]. We calculated the change between baseline and odor-induced response and found a significantly reduced Δpower of the γ_low oscillations in APP/PS1 mice compared to WT controls (Fig. 2H). Because γ rhythms regulate cortical neuronal spiking [43, 44], we reason that this decreased Δpower of γ_low oscillations may contribute, at least in part, to the abnormal aPC neuronal activity observed in APP/PS1 mice.

SFC, i.e., the coherent activity between spikes and LFP, plays an important role in processing sensation and cognition [24, 45]; SFC in the γ range is involved in decoding sensory stimuli [46]. We observed in the present study, for both the baseline and odor-stimulation periods, decreased SFC in the γ_low-band in APP/PS1 mice, although ΔSFC between WT and APP/PS1 mice remained unchanged (Fig. 3F). These results suggest that the aPC of APP/PS1 mice may retain relatively normal synchronization between neuronal spiking and LFP in response to odor stimulation.

Phase relations between neuronal spiking and oscillations, which display spatiotemporal synchronization, contribute to sensory information processing and coding [47, 48]. Here, we evaluated the phasic relationship of aPC neuronal spiking with LFPs in the presence or absence of odor stimulation and observed a similar phase-locking with γ_low oscillations as indicated by MRL values in APP/PS1 mice in response to vanillin compared to baseline (Fig. 4C). Specifically, WT mice showed reduced MRL values, i.e., weaker phase locking with γ_low rhythms during odor sampling compared to baseline; however, MRL values were unchanged in APP/PS1 mice (Fig. 4C), indicating an inability of the aPC to process odor information in APP/PS1 mice. It has been shown that cortical neuronal spiking is inversely related to the strength of γ-band synchronization during odor sampling [35]. Similarly, in association with increased vanillin-induced neuronal firing, WT mice demonstrated a significant reduction in MRL values, and thus in spike phase-locking and γ_low between baseline and odor-stimulation periods. We thus reason that the unchanged MRL values between baseline and odor stimulation in 3-5 month-old APP/PS1 mice may indicate impaired olfactory information processing.

Olfactory dysfunction is one of the symptoms of AD and is correlated with Aβ pathology in the PC of APP/PS1 mice [49]. Recent studies in mice have found that Aβ impairs long-term potentiation [50] and pyramidal neuronal activity [51] in the hippocampus. In addition, Aβ is associated with high-frequency network oscillatory hyperactivity between the olfactory bulbs and the PC [52], indicating that Aβ pathology might alter the electrophysiological properties of the olfactory system of AD mice. Furthermore, there is a debate on whether amyloid has already been deposited in the olfactory system of young APP/PS1 mice [12, 53], especially before the age of 5 months. Here, we verified that 5-month-old APP/PS1 mice exhibit no amyloid plaques in the aPC by 4G8 (human Aβ-selective antibody) staining, although robust Aβ accumulation can be observed in the aPC of 13-month-old APP/PS1 mice (Fig. 5B, D). This supports the conclusion that the abnormalities in aPC neuronal spiking and γ-oscillatory activities in 3-5-month-old APP/PS1 mice can be observed independently of Aβ accumulation. In addition, aberrant γ oscillations have been observed in AD patients [54] using neural recording implants on the cortical surface that capture electrical signals intracellularly in a single neuron or extracellularly from larger areas of the brain [55]. Our results suggest that clinical assessment of aPC neural activity might facilitate the diagnosis of presymptomatic AD patients, and also underline the conclusion that neuronal dysfunction occurs within the aPC prior to Aβ deposition.

In summary, the present study supports the notion that abnormal odor-induced electrophysiological responses, including aPC neuronal spiking and γ-oscillatory activities, are a likely indicator of impaired olfaction in presymptomatic AD mice. Thus, our study of odor-evoked responses in AD model mice is important not only for improving our understanding of the neurobiological changes involved in AD, but also for highlighting the possible clinical implications of such changes, which provide potential diagnostic and therapeutic targets for the impaired olfaction associated with early-stage AD.

Footnotes

ACKNOWLEDGMENTS

The present study was supported by grants from the National Natural Science Foundation of China (31970915, 32170950, 31771219, 31871170), the Guangdong Natural Science Foundation for Major Cultivation Project (2018B030336001), and the Guangdong Grant ‘Key Technologies for Treatment of Brain Disorders’ (2018B030332001).

Authors’ disclosures available online ().

References

Suzuki

, Bekkers

(2007) Inhibitory interneurons in the piriform cortex. Clin Exp Pharmacol Physiol 34, 1064–1069.

Dalal

, Gupta

, Haddad

(2020) Bilateral and unilateral odor processing and odor perception. Commun Biol 3, 150.

Uchida

, Poo

, Haddad

(2014) Coding and transformations in the olfactory system. Annu Rev Neurosci 37, 363–385.

Frasnelli

, Charbonneau

, Collignon

, Lepore

(2009) Odor localization and sniffing. Chem Senses 34, 139–144.

Gottfried

(2010) Central mechanisms of odour object perception. Nat Rev Neurosci 11, 628–641.

Gire

, Whitesell

, Doucette

, Restrepo

(2013) Information for decision-making and stimulus identification is multiplexed in sensory cortex. Nat Neurosci 16, 991–993.

Lepousez

, Lledo

(2013) Odor discrimination requires proper olfactory fast oscillations in awake mice. Neuron 80, 1010–1024.

David

, Hugues

, Cenier

, Fourcaud-Trocme

, Buonviso

(2009) Specific entrainment of mitral cells during gamma oscillation in the rat olfactory bulb. PLoS Comput Biol 5, e1000551.

, Howard

, Gottfried

(2010) Disruption of odour quality coding in piriform cortex mediates olfactory deficits in Alzheimer’s disease. Brain 133, 2714–2726.

10.

Devanand

, Liu

, Tabert

, Pradhaban

, Cuasay

, Bell

, de Leon

, Doty

, Stern

, Pelton

(2008) Combining early markers strongly predicts conversion from mild cognitive impairment to Alzheimer’s disease. Biol Psychiatry 64, 871–879.

11.

Hernandez-Soto

, Rojas-Garcia

, Pena-Ortega

(2019) Sudden intrabulbar amyloid increase simultaneously disrupts olfactory bulb oscillations and odor detection. Neural Plast 2019, 3424906.

12.

Chen

, Chen

, Huo

, Wang

, Tan

, Misrani

, Jiang

, Chen

, Zhang

, Tabassum

, Wang

, Chen

, Long

, Yang

(2021) Enhancing GABAergic signaling ameliorates aberrant gamma oscillations of olfactory bulb in AD mouse models. Mol Neurodegener 16, 14.

13.

Bolding

, Franks

(2018) Recurrent cortical circuits implement concentration-invariant odor coding. Science 361, eaat6904.

14.

Courtiol

, Buonviso

, Litaudon

(2019) Odorant features differentially modulate beta/gamma oscillatory patterns in anterior versus posterior piriform cortex. Neuroscience 409, 26–34.

15.

Coronas-Samano

, Portillo

, Beltran Campos

, Medina-Aguirre

, Paredes

, Diaz-Cintra

(2014) Deficits in odor-guided behaviors in the transgenic 3xTg-AD female mouse model of Alzheimers disease. Brain Res 1572, 18–25.

16.

Wang

N-Y

, Li

J-N

, Liu

W-L

, Huang

, Li

W-X

, Tan

Y-H

, Liu

, Song

Z-H

, Wang

M-Y

, Xie

, Mao

R-R

, Gan

, Ding

Y-Q

, Zhang

, Shan

B-C

, Chen

L-D

, Zhou

Q-X

, Xu

(2021) Ferulic acid ameliorates Alzheimer’s disease-like pathology and repairs cognitive decline by preventing capillary hypofunction in APP/PS1 Mice. Neurotherapeutics 18, 1064–1080.

17.

Cho

, Lee

, Yang

S-H

, Kim

, Lee

, Kim

, Baek

, Yun

, Kim

, Cho

, Woo

, Kim

(2016) Age-dependent inverse correlations in CSF and plasma amyloid-β(1–42) concentrations prior to amyloid plaque deposition in the brain of 3xTg-AD mice. Sci Rep 6, 20185.

18.

Shakhawat

, Harley

, Yuan

(2014) Arc visualization of odor objects reveals experience-dependent ensemble sharpening, separation, and merging in anterior piriform cortex in adult rat. J Neurosci 34, 10206–10210.

19.

Mukherjee

, Morrison

, Fontaine

, Hou

, Harley

, Yuan

(2014) Unlearning: NMDA receptor-mediated metaplasticity in the anterior piriform cortex following early odor preference training in rats. J Neurosci 34, 5143–5151.

20.

, Liu

, Guo

, Luo

(2019) Long-term fiber photometry for neuroscience studies. Neurosci Bull 35, 425–433.

21.

Paxinos

, Franklin

(2003) The Mouse Brain In Stereo-taxic Coordinates, Academic PressSydney.

22.

Thomson

(1982) Spectrum estimation and harmonic analysis. Proc IEEE 70, 1055–1096.

23.

Chen

, Zhang

, Li

, Zhao

, Ye

, Lin

, Tao

, Rasch

, Zhang

(2017) Distinct inhibitory circuits orchestrate cortical beta and gamma band oscillations. Neuron 96, 1403–1418.e6.

24.

Peng

, Cai

, Hu

, Jiang

, Zhang

, Liu

, Yang

, Long

(2021) Altered corticostriatal synchronization associated with compulsive-like behavior in APP/PS1 mice. Exp Neurol 344, 113805.

25.

Siapas

, Lubenov

, Wilson

(2005) Prefrontal phase locking to hippocampal theta oscillations. Neuron 46, 141–151.

26.

Jacob

, Hahnke

, Nieder

(2018) Structuring of abstract working memory content by fronto-parietal synchrony in primate cortex. Neuron 99, 585–597e585.

27.

Zhang

, Hao

, Jiang

, Feng

, Chen

, Zheng

, Liu

, Zhang

, Long

, Yang

(2018) The mechanisms underlying olfactory deficits in apolipoprotein E-deficient mice: Focus on olfactory epithelium and olfactory bulb. Neurobiol Aging 62, 20–33.

28.

, Liu

, Chen

, Ge

, Lu

, Li

(2020) Excitability of neural activity is enhanced, but neural discrimination of odors is slightly decreased, in the olfactory bulb of fasted mice. Genes (Basel) 11, 433.

29.

Mori

, Manabe

, Narikiyo

, Onisawa

(2013) Olfactory consciousness and gamma oscillation couplings across the olfactory bulb, olfactory cortex, and orbitofrontal cortex. Front Psychol 4, 743.

30.

Iaccarino

, Singer

, Martorell

, Rudenko

, Gao

, Gillingham

, Mathys

, Seo

, Kritskiy

, Abdurrob

, Adaikkan

, Canter

, Rueda

, Brown

, Boyden

, Tsai

(2016) Gamma frequency entrainment attenuates amyloid load and modifies microglia. Nature 540, 230–235.

31.

Herrmann

, Demiralp

(2005) Human EEG gamma oscillations in neuropsychiatric disorders. Clin Neurophysiol 116, 2719–2733.

32.

Rutishauser

, Ross

, Mamelak

, Schuman

(2010) Human memory strength is predicted by theta-frequency phase-locking of single neurons. Nature 464, 903–907.

33.

Fries

, Reynolds

, Rorie

, Desimone

(2001) Modulation of oscillatory neuronal synchronization by selective visual attention. Science 291, 1560–1563.

34.

Bauer

, Oostenveld

, Peeters

, Fries

(2006) Tactile spatial attention enhances gamma-band activity in somatosensory cortex and reduces low-frequency activity in parieto-occipital areas. J Neurosci 26, 490–501.

35.

van Wingerden

, Vinck

, Lankelma

, Pennartz

(2010) Learning-associated gamma-band phase-locking of action-outcome selective neurons in orbitofrontal cortex. J Neurosci 30, 10025–10038.

36.

Ahnaou

, Walsh

, Manyakov

, Youssef

, Drinkenburg

(2019) Early electrophysiological disintegration of hippocampal neural networks in a novel locus coeruleus tau-seeding mouse model of Alzheimer’s disease. Neural Plast 2019, 6981268.

37.

Murphy

(2019) Olfactory and other sensory impairments in Alzheimer disease. Nat Rev Neurol 15, 11–24.

38.

Wilson

(2003) Rapid, experience-induced enhancement in odorant discrimination by anterior piriform cortex neurons. J Neurophysiol 90, 65–72.

39.

Miller

, Cannon

(2019) Combined mechanisms of neural firing rate homeostasis. Biol Cybern 113, 47–59.

40.

Tantirigama

, Huang

, Bekkers

(2017) Spontaneous activity in the piriform cortex extends the dynamic range of cortical odor coding. Proc Natl Acad Sci U S A 114, 2407–2412.

41.

Buzsaki

, Anastassiou

, Koch

(2012) The origin of extracellular fields and currents–EEG, ECoG, LFP and spikes. Nat Rev Neurosci 13, 407–420.

42.

Ray

, Ni

, Maunsell

(2013) Strength of gamma rhythm depends on normalization. PLoS Biol 11, e1001477.

43.

Fernandez-Ruiz

, Oliva

, Soula

, Rocha-Almeida

, Nagy

, Martin-Vazquez

, Buzsaki

(2021) Gamma rhythm communication between entorhinal cortex and dentate gyrus neuronal assemblies. Science 372, eabf3119.

44.

Buzsaki

, Wang

(2012) Mechanisms of gamma oscillations. Annu Rev Neurosci 35, 203–225.

45.

Xiao

, Martinez

, Kulkarni

, Zhang

, Hou

, Rosenberg

, Talay

, Shalot

, Zhou

, Wang

, Chen

(2019) Cortical pain processing in the rat anterior cingulate cortex and primary somatosensory cortex. Front Cell Neurosci 13, 165.

46.

, Gire

, Restrepo

(2015) Upsilon spike-field coherence in a population of olfactory bulb neurons differentiates between odors irrespective of associated outcome. J Neurosci 35, 5808–5822.

47.

Martinez

(2005) Oscillatory synchronization requires precise and balanced feedback inhibition in a model of the insect antennal lobe. Neural Comput 17, 2548–2570.

48.

Bieler

, Sieben

, Cichon

, Schildt

, Roder

, Hanganu-Opatz

(2017) Rate and temporal coding convey multisensory information in primary sensory cortices. eNeuro 4, ENEURO.0037–0017.2017.

49.

Yao

, Hua

, Zhang

, Li

, Qin

(2017) Olfactory dysfunction in the APP/PS1 transgenic mouse model of Alzheimer’s disease: Morphological evaluations from the nose to the brain. Neuropathology 37, 485–494.

50.

, Liu

, Bao

, Wu

, Li

, Zheng

, Anwyl

, Wang

(2014) Rosiglitazone prevents amyloid-beta oligomer-induced impairment of synapse formation and plasticity via increasing dendrite and spine mitochondrial number. J Alzheimers Dis 39, 239–251.

51.

Bahrami

, Asgari

, Hosseinmardi

, Janahmadi

(2019) Peroxisome proliferator-activated receptor (PPAR)-gamma modifies Abeta neurotoxin-induced electrophysiological alterations in rat primary cultured hippocampal neurons. Iran J Pharm Res 18, 1403–1418.

52.

Wesson

, Borkowski

, Landreth

, Nixon

, Levy

, Wilson

(2011) Sensory network dysfunction, behavioral impairments, and their reversibility in an Alzheimer’s beta-amyloidosis mouse model. J Neurosci 31, 15962–15971.

53.

Saiz-Sanchez

, Ubeda-Banon

, De la Rosa-Prieto

, Martinez-Marcos

(2012) Differential expression of interneuron populations and correlation with amyloid-beta deposition in the olfactory cortex of an AbetaPP/PS1 transgenic mouse model of Alzheimer’s disease. J Alzheimers Dis 31, 113–129.

54.

Traikapi

, Konstantinou

(2021) Gamma oscillations in Alzheimer’s disease and their potential therapeutic role. Front Syst Neurosci 15, 782399.

55.

Hashemi Noshahr

, Nabavi

, Sawan

(2020) Multi-channel neural recording implants: A review. Sensors (Basel) 20, 904.

Altered Odor-Evoked Electrophysiological Responses in the Anterior Piriform Cortex of Conscious APP/PS1 Mice

Abstract

Background:

Objective:

Methods:

Results:

Conclusion:

Keywords

INTRODUCTION

MATERIALS AND METHODS

Animals

Odor stimulation

In vivo electrophysiology

Surgical procedures

Data acquisition

Power spectrum analysis

Firing rate analysis

Spike-field coherence analysis

Phase-locking firing analysis

Histology

Verification of electrode location

Amyloid plaque staining

Statistical analysis

RESULTS

Decreased vanillin-evoked neuronal firing rate in the aPC of APP/PS1 mice

Reduced power of γ oscillations in response to vanillin in APP/PS1 mice

Weakened coherence between γ low oscillations and spikes in APP/PS1 mice

Impaired phase-locking of aPC neuronal spiking with γ low rhythm in APP/PS1 mice

No Aβ plaques in the aPC of 5-month-old APP/PS1 mice

DISCUSSION

Footnotes

ACKNOWLEDGMENTS

References

Weakened coherence between γ_low oscillations and spikes in APP/PS1 mice

Impaired phase-locking of aPC neuronal spiking with γ_low rhythm in APP/PS1 mice