Chronic optogenetic activation of hippocampal CA1 neurons triggers Alzheimer’s disease-like proteomic remodeling

Summary

Neuronal overexcitability triggers synaptic changes, leading to neural hyperactivity, network disruption, and is postulated to trigger neurodegeneration in Alzheimer’s disease (AD). However, the sequence of synaptic changes from excessive activity remains unclear. We employed optogenetics to induce sustained neuronal hyperactivity in the hippocampi of wild-type and AD-like 5xFAD mice. After a month of daily optogenetic stimulation, the proteomic profiles of photoactivated wild-type and 5xFAD mice exhibited remarkable similarity. Proteins involved in translation, protein transport, autophagy, and notably in the AD pathology were upregulated in wild-type mice. Conversely, both glutamatergic and GABAergic synaptic proteins were downregulated. These hippocampal proteomic and signaling alterations in wild-type mice resulted in spatial memory loss and augmented Αβ42 secretion. Collectively, these findings indicate that sustained neuronal hyperactivity alone replicates proteome changes seen in AD-like mutant mice. Therefore, prolonged neuronal hyperactivity may contribute to synaptic transmission disruption, memory deficits and the neurodegenerative process associated with AD.

Graphical abstract

Highlights

• •Chronic CA1 optogenetic stimulation induces Alzheimer’s-like proteomic-wide changes
• •Optogenetically stimulated wild-type mice display spatial memory impairments
• •Stimulated wild-type mice’s proteomes overlap with pre-symptomatic 5xFAD profiles
• •Unique proteomic patterns in 5xFAD mice suggest occluded plasticity upon stimulation

Molecular neuroscience; Cellular neuroscience; Proteomics

Introduction

Aberrant neuronal activity can impair synapse formation and strength resulting in disruption of synaptic plasticity. In turn, altered synaptic plasticity could reorganize synaptic connections ensuing runaway excitation or quiescence and eventually modify local neuronal circuits.¹ For neural circuits to maintain their characteristic firing patterns, neurons within the network undergo homeostatic changes at the cellular level, regulating synthesis and degradation of key synaptic proteins.^2,3 Alterations perturbing these regulations have been linked with the onset of various brain disorders such as epilepsy and autism.⁴

Neuronal hyperexcitability manifests early in Alzheimer’s disease (AD),^5,6,7 leading to cortical and hippocampal hyperactivity^8,9 and under certain conditions to epileptiform activity and seizures in rodents^10,11 and humans.¹² Both amyloid-β (Aβ) and Tau have been found to induce neuronal hyperactivity through distinct cellular mechanisms.^{11,13,14,15,16} Concurrently, neuronal hyperactivity can increase Αβ secretion^17,18 and promote pathological phosphorylation of Tau.¹⁹ These observations suggest that Αβ and Tau can initiate a vicious cycle of neuronal hyperactivity in AD underlying neurodegeneration in disease. While previous studies have used transgenic lines harboring mutated ryanodine receptor 2 to induce neural hyperactivity and replicate aspects of AD pathology,²⁰ it remains unclear whether neuronal hyperactivity alone—within the adult brain and in the absence of familial AD-linked mutations—can induce pathology that mimics neurodegeneration seen in AD.

Optogenetics has emerged as a powerful method to control the activity of genetically defined neurons allowing to study the function of selective brain circuits in physiological but also pathological conditions.^21,22,23 Stabilized step function opsins (SSFOs) also allow for long-lasting activation of neural circuits with just a brief light stimulation.²⁴ Such chronic optogenetic activation is necessary to model neurodegenerative disorders like AD for which neuronal activity is altered for long periods of time. We adopted a similar approach to generate a model of chronic neuronal hyperactivity in the hippocampus of wild-type (WT) and a transgenic mouse line carrying AD-linked mutations and demonstrated that evoked neuronal hyperactivity can disrupt synaptic signaling and facilitate an activity-driven neuropathology similar to AD.

Results

Optogenetic activation of the hippocampus in wild-type and 5xFAD mice using SSFO

To model chronic neuronal hyperactivity in the rodent hippocampus, we unilaterally transduced either SSFO-mCherry or tdTomato under the CaM kinase II promoter (CaMKIIa) into the CA1 of WT or 5xFAD²⁵ male mice (Figures 1A and 1B). We then inserted an optic fiber cannula above the CA1, and blue-light stimulation was delivered for four weeks to activate the transduced neurons (Figure 1A). Based on the light power distribution at the fiber tip (Figure 1B) and the restricted viral expression pattern, we do not anticipate lateral activation of CA3 cell bodies or any activation of dentate gyrus (DG) neurons.

Figure 1.

SSFO optogenetic stimulation activates hippocampal neurons and evokes neuronal hyperactivity

(A) Diagram showing an overview of the proteome experiment following chronic optogenetic stimulation of the hippocampus (HPC).

(B) Representative confocal image showing SSFO expression via its mCherry reporter in a coronal brain section from a mouse infected with AAV-CaMKIIα-SSFO. The inset displays an estimated light power distribution from the implanted optic fiber, with dimensions scaled accordingly. (Scale bars: 1 mm).

(C) Confocal images showing c-Fos immunofluorescence. SSFO expression is restricted to the ipsilateral side of viral infection. Unilateral light stimulation increased c-Fos levels in the ipsilateral CA1 and dentate gyrus (DG), with no observable increase on the contralateral side. (Scale bars: 20 μm; DAPI in blue).

(D) Micro-optrode recording configuration. A 470 nm laser was coupled to an optrode probe for electrical recordings and advanced into the brain.

(E) Example traces recorded in AAV-CaMKIIα-SSFO (top) or AAV-CaMKIIα-tdTomato (bottom) infected mice. 470 nm activation pulse is indicated with a blue bar.

(F) Peristimulus time histogram of the mean cell activity of neurons recorded from mice infected with AAV-CaMKIIα-SSFO (n = 10 cells from N = 3 mice; cyan) or AAV-CaMKIIα-tdTomato (n = 6 cells from N = 4 mice; orange). Inset displays the mean normalized firing rate of neurons recorded from SSFO and tdTom mice during the time bin indicated by the dashed area, which corresponds to the duration of the light pulse delivery. ∗∗∗P < 0.001, Mann-Whitney test.

Data presented as mean ± SEM.

To validate that SSFO activation evoked neuronal activity, we performed both immunostaining and immunoblot analysis for c-Fos in hippocampal sections and lysates, respectively. Coronal sections from mice infected with AAV-CaMKIIα-SSFO-mCherry showed localized expression in CA1 (Figure 1B). Thirty minutes after light stimulation, increased c-Fos expression was observed in CA1, CA3, and DG neurons ipsilateral to the stimulation site, compared to the contralateral side (Figures 1C and S1A). In contrast, mice infected with AAV-CaMKIIα-tdTomato showed no c-Fos expression in either hemisphere (Figure S1). Consistent with these findings, c-Fos levels were significantly higher in hippocampal lysates from SSFO-expressing mice than in tdTomato-expressing controls (Figure S2). We also confirmed SSFO-mCherry and tdTomato expression in the same hippocampal lysates (Figures S3A–S3C). Notably, in WT+SSFO mice, SSFO-mCherry protein significantly correlated with c-Fos expression (Figure S3D), whereas tdTomato levels did not correlate with c-Fos expression in WT+tdTom mice (Figure S3D).

Using extracellular recordings in anesthetized mice we assessed the light-induced changes in the activity of the CA1 region transduced either with the SSFO or tdTomato using a glass micro-optrode we previously designed, which enables dual, co-registered electrophysiological recording and optogenetic stimulation, free of photo-electric artifacts (Figure 1D).^26,27 In mice expressing the SSFO, the multiunit spike rates within the CA1 increased for several minutes after a 2s light pulse as expected (Figures 1E and 1F). Conversely, in tdTomato expressing mice, multiunit spiking activity did not increase following the brief light pulse (Figures 1E and 1F). At the onset of the light stimulation, neurons expressing the SSFO exhibited a significantly higher firing rate compared to neurons expressing tdTomato (Figure 1F). In addition, the elevated cell activity observed in SSFO-expressing neurons remained significantly higher than that of tdTomato neurons throughout the duration following the light pulse (Figure S4). Altogether, these results indicate that the SSFO is transduced in the hippocampus and can be activated by light stimulation to generate neuronal hyperactivity.

Chronic optogenetic stimulation of the hippocampus activates protein translation and phosphorylation but downregulates synaptic proteins in wild-type mice

We sought to identify and quantify alterations in protein levels in response to sustain neuronal hyperactivity in the hippocampus of mice. We analyzed the hippocampi of WT mice expressing SSFO or tdTomato after 4 weeks of daily brief light stimulation (during wakefulness; Figure 1A) using label-free liquid chromatography with tandem mass spectrometry (LC-MS/MS) proteomic analysis. We identified 778 proteins differentially abundant (P value ≤ 0.05) between the two conditions (Figure 2A; Table S1). The chronic optogenetic stimulation in mice expressing SSFO significantly upregulated the expression of 556 proteins and only downregulated the expression of 222 proteins (Figure 2A). Gene ontology (GO) enrichment analysis with GeneCodis²⁸ showed that the abundant proteins were mainly part of the cytoplasmic and mitochondrial cellular components (Figure 2B). Yet, 77 of the proteins identified were part of the synapse (Figure 2B) with approximately 45% appearing downregulated, suggesting that synaptic function is altered by evoked neuronal hyperactivity.

Figure 2.

Chronic optogenetic stimulation elicits synaptic signaling protein downregulation in wild-type mice

(A) Heatmap illustrating the clustering of proteins (P value ≤0.05) in the hippocampus of WT mice expressing SSFO (N = 4) or tdTomato (N = 5) after 4 weeks of daily optogenetic stimulation. The color-coded scale (Z score) denotes the upregulated (red) and downregulated (blue) proteins.

(B–D) Results of the GO enrichment analysis of the top 10 most abundant cellular components (B), biological processes (C) and KEGG pathways (D) in the hippocampus of WT+SSFO mice. In red or blue the fraction of the total proteins upregulated or downregulated, respectively, for each GO-term. The number at the base of each bar denotes the percentage of upregulated proteins associated with each GO-term.

(E) Interactome of functional or physical protein to protein associations of the synaptic proteins found downregulated in WT+SSFO mice. The proteins in orange are associated with glutamatergic synapses while the proteins in green with GABAergic.

(F) Sunburst plots of the GO enrichment analysis of the synaptic compartment (top) and process (bottom) based on the 163 SynGO annotations identified among the list of proteins found altered in the hippocampus of WT+SSFO mice.

(G) Representative immunoblots for the levels of calbindin (Calb1), EAAT1 and EAAT2 in hippocampal lysates from WT + SSFO and WT + tdTom mice.

(H) Quantification of calbindin, EAAT1 and EAAT2 protein levels measured by immunoblotting in WT+SSFO (cyan; N = 12) and WT+tdTom (orange; N = 12) mice after chronic optogenetic stimulation. The total protein (TLP) levels measured on TGX Stain-Free gels is the loading control. Data are analyzed with estimation statistics. Each point represents an individual mouse. The mean difference is dissipated as a black dot and the 95% confidence interval by the ends of the vertical error bars.

See Table S1 for proteomics data and statistical analysis, and Data S1 and S2 for uncropped gels and blots and the respective protein quantification.

Moreover, GO analysis revealed that translation, protein transport, phosphorylation, and autophagy were among the most enriched biological processes found altered by the chronic optogenetic stimulation (Figure 2C). In turn, the AMPK signaling, and ubiquitin-mediated proteolysis were identified among the top 10 most enriched KEGG pathways (Figure 2D). Surprisingly, the AD pathway was also identified, with 31 proteins, all found upregulated in the WT+SSFO mice (Figure 2D).

Out of the 34 downregulated synaptic proteins, interaction network analysis with STRING,²⁹ identified 19 proteins as glutamatergic synapse proteins, while 9 were enriched in GABAergic synapses (Figure 2E). And yet there are several more proteins, key for synaptic transmission, which were found significantly downregulated in WT mice following chronic hippocampal hyperactivity (Figures 2E and S5A). Since the synapse emerged as a highly enriched GO term and multiple synaptic proteins were altered by chronic optogenetic stimulation, we used SynGO³⁰ to further identify synapse-specific GO terms overrepresented among the differentially expressed proteins in WT+SSFO mice. In total, 163 unique SynGO-annotated proteins were found in our dataset, with 5 cellular component terms and 9 biological process terms significantly enriched at a 1% FDR. The pre-synapse was the most prominent cellular compartment, with 17 out of 183 proteins identified (Figure 2F). Among top-level synaptic biological processes (Figure S5B), only synapse organization, transport, and presynaptic processes were significantly enriched in WT+SSFO mice (Figure 2F).

Using immunoblots, we confirmed that calbindin 1 (Calb1) was indeed downregulated in hippocampal lysates from WT+SSFO expressing mice as compared to WT+tdTom mice (Figures 2G and 2H). Additionally, the excitatory amino acid transporter 1 (EAAT1 or Slc1a3) was downregulated, while the excitatory amino acid transporter 2 (EAAT2 or Slc1a2) was unaltered in the WT+SSFO mice (Figures 2G and 2H), consistent with the findings from the proteomic analysis (Figure 2A; Table S1).

Early AD pathology in 5xFAD mouse hippocampus largely mimics the footprint of altered proteostasis caused by chronic optogenetic stimulation

The 5xFAD mice exhibit amyloid deposition accompanied by cognitive deficits recapitulating major features of Alzheimer’s disease amyloid pathology.³¹ Comparison of AD human and rodent brain tissue revealed that the 5xFAD mice exhibit a proteomic signature similar to symptomatic AD.³² Moreover, an independent analysis of the mouse hippocampal proteome focused on the proteomic changes during progression of the amyloid pathology.³³ In turn, we analyzed the proteome changes manifesting in 3-month-old 5xFAD mice, to identify similarities between the pre-symptomatic AD brain and the effect of hippocampal neuronal hyperactivity in WT mice.

When we compared the proteins altered by the mutations and the AD pathology in the hippocampus of 5xFAD mice, we identified 926 proteins as differentially expressed (Figure 3A; Table S2). Most of these proteins were upregulated (Figure 3B). Protein translation and transport, autophagy and phosphorylation were identified among the altered biological processes in the 5xFAD mice (Figure 3C), consistent with our findings in the WT+SSFO mice (Figure 2C). The mitochondria, synapse, and endosomes were among the most enriched cellular components (Figure 3D). The majority of the abundantly expressed synaptic proteins belong to the glutamatergic synapse, with 61% of them found upregulated in 5xFAD mice (Figure 3D).

Figure 3.

Early AD pathology alters translation, phosphorylation and autophagy in the hippocampus of 5xFAD mice

(A) Heatmap illustrating the scaled protein levels in the hippocampus of 3-month-old 5xFAD (N = 4) and wild-type (N = 5) mice overexpressing tdTomato. The color-coded scale (Z score) denotes the upregulated (red) and downregulated (blue) proteins.

(B) Venn diagram showing the number of upregulated (red) and downregulated (blue) proteins in 5xFAD mice. In yellow the number of upregulated proteins found enriched in the AD KEGG pathway.

(C–E) The results of the GO enrichment analysis of the top 10 most abundant biological processes (C), cellular components (D) and KEGG pathways (E) in the hippocampus of 5xFAD mice. The colors in each bar represent the fraction of the total proteins for each GO-term found upregulated (red) or downregulated (blue), while the number at the base of each bar represents the percentage of upregulated proteins for each GO-term.

(F and G) Sunburst plots of the GO enrichment analysis of the synaptic component (F) and process (G) based on the 184 SynGO annotations identified among the list of proteins found altered in the hippocampus of 3-month-old 5xFAD mice.

See Table S2 for detailed proteomics data and statistical analysis.

When we focused our GO analysis specifically on synaptic components, we found high enrichment in the postsynaptic density and endosome (Figure 3F). On the presynaptic side, proteins associated with the synaptic vesicle membrane were prominently enriched (Figure 3F). Regarding synaptic processes, alterations were observed in postsynaptic organization, synaptic vesicle dynamics, and neurotransmitter loading into vesicles (Figure 3G) in 5xFAD mice.

Interestingly, for biological processes such as translation, autophagy, and RNA splicing the overlap of differentially expressed proteins between WT+SSFO and 5xFAD mice exceeded 50% (Figure 4A), with most proteins showing the same valence (Figure 4B). Among synapse-related proteins, the overlap was 45.5% (Figure 4A) with only Arestin 3 (Arr3) displaying opposing valence between the two groups (Figure 4B). Despite the substantial overlap, a considerable number of synaptic proteins were uniquely altered in each group (Figures 4A and S6). Synapse-specific GO analysis identified 41 unique SynGO-annotated proteins in 5xFAD mice, associated with 17 significantly enriched biological processes (Figures S6A and S6B), including postsynaptic density organization, chemical neurotransmission and both pre- and post-synaptic translation. In contrast, the 33 unique SynGo-annotated proteins found in WT+SSFO, were linked to 9 significantly enriched synaptic processes (Figures S6C and S6D), with prominent terms including the presynaptic vesicle cycle and exocytosis, chemical synaptic transmission and regulation of neurotransmitter receptor levels.

Figure 4.

Chronic hippocampal optogenetic activation in wild-type mice induces proteome changes linked to early AD pathogenesis

(A) Venn diagrams illustrating the number of the abundant proteins for each GO-term found in the hippocampus of WT + SSFO and 5xFAD+tdTom mice. The percentages represent the fraction of the WT+SSFO altered proteins commonly found altered in the 5xFAD mice.

(B) Heat maps revealing the fold change in levels for the common proteins of each GO-term in A in the WT+SSFO or 5xFAD+tdTom mice.

(C) The number of common proteins participating in the AD KEGG pathway when comparing the altered proteins in the 5xFAD mice and the proteins altered in the WT+SSFO after the chronic optogenetic stimulation.

(D and E) Interactomes of functional or physical protein to protein associations of the AD KEGG pathway proteins found altered in 5xFAD+tdTom (D) and WT+SSFO (E) mice.

GO analysis further identified 38 proteins implicated in the AD pathway in the presymptomatic 5xFAD mice (Figures 3E and 4C), 19 of which were also abundantly expressed in WT+SSFO mice (Figures 4C–4E), suggesting that chronic optogenetic activation of the hippocampus in WT mice induced signaling cascades involved in the AD pathogenesis. However, among the uniquely altered AD-pathway proteins, we identified multiple 26S proteasome subunits (Psmb5, Psmd1, Psmd8, and Psmd12) exclusively in 5xFAD mice (Figure 4D) suggesting that neurons exhibiting AD-pathology may be under increased proteostatic stress—potentially reflecting an upregulation of the ubiquitin-proteasome system to manage elevated loads of misfolded, damaged, or aggregation-prone proteins like Αβ.^34,35 While this response may initially serve as an adaptive mechanism to maintain proteostasis, it could also indicate proteasomal overload or dysfunction, both considered early hallmarks of AD pathology.^36,37 The additional presence of RELA (NF-κB p65) points to the activation of stress- and inflammation-related transcriptional programs, which may further destabilize proteostasis and increase neuronal vulnerability. Notably, this proteasome response and RELA upregulation are absent in WT+SSFO mice (Figure 4E), suggesting that these neurons may retain proteostatic plasticity—enabling more effective cellular repair—and may be under a comparatively lower burden of misfolded or aggregated Aβ.

Evoked chronic neuronal hyperactivity in the hippocampus of 5xFAD mice downregulates mRNA splicing and protein phosphorylation

We adopted the same approach to induce neuronal hyperactivity in the hippocampus of 5xFAD mice (Figure 1A). The chronic optogenetic stimulation in 5xFAD mice resulted in significantly downregulating the expression levels of 264 proteins while it resulted in upregulating the levels of only 99 proteins (Figure 5A; Table S3). The altered proteins were primarily cytoplasmic, nucleic and part of the spliceosomal complex (Figure 5B). Among the cellular components greatly enriched, the glutamatergic synapse was identified with 29 proteins (Figure 5B), 83% of which were downregulated. Three additional proteins of the GABAergic synapse were found to be downregulated in the 5xFAD mice after the chronic optogenetic stimulation (Figure 5C).

Figure 5.

Chronic optogenetic stimulation downregulates RNA processing and protein phosphorylation in 5xFAD mice

(A) Heatmap revealing the scaled protein levels in the hippocampus of 5xFAD mice expressing SSFO (N = 4) or tdTomato (N = 4) after 4 weeks of daily optogenetic stimulation.

(B) GO enrichment analysis showing the top 10 most abundant biological processes in 5xFAD+SSFO mice following chronic optogenetic stimulation. In red or blue the fraction of the total proteins upregulated or downregulated for each GO-term, respectively. Numbers within each bar represent the percentages of downregulated proteins corresponding to each GO-term.

(C) Interactome of functional or physical protein to protein associations of the synaptic proteins altered in 5xFAD+SSFO mice after optogenetic stimulation. The orange circles represent glutamatergic synapse proteins while the green GABAergic. The proteins written in red are upregulated while the proteins in blue are downregulated.

(D and E) Same as in B but for the top 10 cellular components (D) and KEGG pathways (E) in the hippocampus of 5xFAD+SSFO mice.

(F) Circos plots showing selected biological processes and the proteins found downregulated in 5xFAD+SSFO mice.

(G) Sunburst plots of the GO enrichment analysis of the synaptic component and process based on the 162 SynGO annotations identified among the list of proteins found altered in the hippocampus of 5xFAD+SSFO mice.

See Table S3 for detailed proteomics data and statistical analysis.

Contrary to the WT mice, GO enrichment analysis revealed that neurons in 5xFAD mice responded distinctly to the induced neuronal hyperactivity by downregulating mRNA processing, splicing, and protein phosphorylation (Figure 5D). Consistently, the spliceosome and RNA degradation appeared among the most enriched KEGG pathways in the 5xFAD+SSFO mice (Figure 5E). The purine and pyrimidine metabolism pathways were also affected (Figure 5E). Biological processes such as chemical synaptic transmission and synapse organization appeared altered (Figure 5F) by the chronic optogenetic stimulation in the 5xFAD mice.

Synapse-specific GO analysis further highlighted the distinct effects of neural hyperactivity in WT versus 5xFAD mice. In 5xFAD mice, the postsynaptic density and membrane were primarily affected (Figure 5G), potentially reflecting disrupted synapse organization, as indicated by the downregulation of related GO terms in the 5xFAD+SSFO group. Yet, GO terms related to the presynaptic active zone, synaptic vesicle dynamics, and modulation of chemical synaptic transmission were prominently enriched in 5xFAD+SSFO mice (Figure 5G), but not synaptic signaling terms as revealed in the WT+SSFO mice (Figures 2F and S5B).

To further assess whether 5xFAD mice exhibit a distinct proteomic response to sustained neuronal hyperactivity, we performed clustering analyses based on all significantly altered proteins across the 5xFAD+SSFO, 5xFAD+tdTom, and WT+SSFO groups. Both hierarchical clustering and an unbiased consensus clustering approach—employing bootstrap resampling and the addition of Gaussian white noise—consistently revealed two distinct clusters: one composed exclusively of 5xFAD+SSFO mice, and another grouping together the 5xFAD+tdTom and WT+SSFO mice (Figure S7).

These findings indicate that the 5xFAD mice respond differently to chronic optogenetic stimulation than WT mice, suggesting an occlusion of plasticity mechanisms in 5xFAD. In contrast, the plasticity changes observed in WT+SSFO mice might resemble those seen during early stages of AD pathogenesis—a hypothesis we explored in the following analyses.

Degree of similarity in proteome changes caused by hyperactivity in WT brain versus that in AD-related pathology

Hippocampal hyperactivity has been implicated in memory deterioration and AD pathogenesis.^6,12 To further explore whether chronic optogenetic activation in the hippocampus of WT mice indeed replicates some of the neurodegenerative processes involved in the pathogenesis of AD, we performed a consensus clustering analysis based on our proteomic data. Accordingly, we found that the WT+SSFO mice grouped together with the 5xFAD+tdTom while all the WT+tdTom mice separated from the rest of mice according to the changes in protein levels (Figures 6A and S8A).

Figure 6.

Robust clustering analysis reveals shared signatures between WT + SSFO and 5xFAD+tdTom mice

(A) Heatmap revealing the hierarchical clustering of 5xFAD and WT mice expressing SSFO or tdTomato based on their scaled protein levels in the hippocampus after 4 weeks of daily optogenetic stimulation. Each protein is represented by a color-coded box, with upregulated proteins in red and downregulated proteins in blue. The linkage matrix indicates the probability of individual samples clustering together, with distinct color blocks denoting stable groups. The P value threshold for protein selection was set to the conventional p < 0.05 for this clustering analysis.

(B) The effective stable, nuclear and entropy (E-rank) rank metrics as a function of the P value threshold used to identify proteins included in the 80 independent consensus clustering iterations for the WT+tdTom, WT+SSFO, and 5xFAD+tdTom mice.

(C, E, and G) Heatmaps showing hierarchical clustering of 5xFAD and WT mice expressing SSFO or tdTomato, based on scaled hippocampal protein expression levels after 4 weeks of daily optogenetic stimulation. Protein selection for clustering was performed using different significance thresholds: P ≤ 0.0001 (C), P ≤ 0.023974 (E), and P ≤ 1.0 (G), as indicated above each dendrogram. Distinct colored blocks denote stable clusters. Each protein is represented by a color-coded box, with upregulated proteins in red and downregulated proteins in blue.

(D, F, and H) Consensus clustering matrices generated through 2,000 bootstrap re-samplings and 10,000 permutations, with Gaussian white noise (standard deviation $\sigma =2.0$) added for data perturbation. The matrices correspond to the respective P value thresholds shown above. Color scale indicates the consensus index, representing the frequency with which each pair of mice clustered together across all iterations.

To further assess the robustness of our clustering results with respect to protein selection criteria, we conducted a systematic analysis by varying the ANOVA P value threshold across 80 consensus clustering experiments. For each threshold, proteins with P values below the cutoff were selected, and a consensus clustering matrix was computed based on 2000 bootstrap replicates with added Gaussian noise ($\sigma =2.0$). To evaluate the stability and intrinsic dimensionality of the resulting consensus matrices—particularly under noise—we computed three effective rank measures: stable rank, nuclear rank, and entropy rank (Figure 6B). All three ranks exhibit a plateau across a wide range of intermediate P value thresholds, indicating stable and consistent clustering behavior. At very low P values, the effective ranks—initially divergent—converged rapidly toward 2.0, reflecting the early emergence of two dominant clusters among individual mice. This low-rank structure persisted until higher thresholds, beyond which the effective rank decreases sharply, consistent with the inclusion of noisy or weakly informative proteins (Figure 6B).

To probe cluster compositions more deeply, we used two more stringent thresholds—P ≤ 0.0001, and P = 0.023974—and generated the corresponding heatmaps and consensus matrices (Figures 6C–6F). In both cases, one cluster consistently grouped the WT+SSFO and 5xFAD+tdTom mice together, while all WT+tdTom mice formed a distinct group (Figures 6C–6F). However, including all identified proteins—using more permissive thresholds—clustering performance decreased (Figures 6G and 6H), reflecting increased noise and inclusion of weakly informative proteins. Together, this analysis demonstrated that the clustering results are robust to different statistical thresholds and support the biological relevance of the observed groupings—particularly the proteomic similarity between WT+SSFO and 5xFAD+tdTom mice.

When we further selected the proteins with similar expression levels between WT+SFFO and 5xFAD+tdTom for the GO terms shown in Figure 4B and conducted hierarchical clustering analysis, we observed that the WT+SSFO and 5xFAD+tdTom mice clustered together and separated from the control WT+tdTom mice (Figure S8B). Similar consensus clustering, as before, identified the same two distinct clusters, showing high intra-cluster consensus values close to one and low inter-cluster values (Figure S8C). Moreover, by comparing the proteins altered by stimulation in WT+SSFO mice to the ones altered due to the mutations and AD-related pathology in the 5xFAD mice, we found that those two groups resemble themselves much more than when comparing 5xFAD mice to unstimulated control (only 135 protein changes in common; Figure S8D).

Neuronal hyperactivity in wild-type mice induces spatial memory deficits and augments Αβ42 secretion

The mammalian target of rapamycin complex 1 (mTORC1) is a principal signaling pathway controlling protein translation initiation³⁸ and has been postulated to take part in AD pathology.³⁹ Immunoblotting of hippocampal lysates revealed an increase in the phosphorylation of the mTORC1 downstream translation initiation effector ribosomal protein S6 (rpS6; Figures 7A and 7C), concomitant with an increase in the expression of Rheb (Figures 7A and 7B), the mTORC1 canonical activator⁴⁰ in the WT+SSFO mice as compared to the WT+tdTom mice.

Figure 7.

Chronic hippocampal optogenetic activation induces spatial memory deficits and augments Αβ42 secretion in wild-type mice

(A) Representative immunoblots of hippocampal lysates from WT+SSFO and WT+tdTom mice.

(B) The protein levels of Rheb in WT+SSFO and WT+tdTom mice.

(C) Quantification of ribosomal protein S6 (rpS6) expression (left) and its phosphorylation levels at the Ser240/204 residues.

(D) The latency to the new arm during the 1-h Y-maze spatial memory test for 6-month-old 5xFAD mice (N = 15) vs. their non-transgenic littermates (WT; N = 9) on the left and for WT+SSFO mice (N = 12) vs. WT+tdTom mice (N = 12) on the right.

(E) The concentration of soluble Αβ42 (left) and Αβ40 (right) in the hippocampus of WT+SSFO mice as compared to WT+tdTom. In B, C, D and E, data are analyzed with estimation statistics. Each point represents an individual mouse. The mean difference is dissipated as a dot and the 95% confidence interval by the ends of the vertical error bars.

A detailed summary of statistics is presented in Table S4. Also see Data S3 and S4 for uncropped gels and blots and the respective protein quantification.

To further examine how chronic hippocampal hyperactivity and the synaptic transmission changes (Figure 2F) observed following the optogenetic stimulation impact brain function, we performed behavioral testing of WT and 5xFAD mice. We adopted a modified version of the Y-maze alternation test allowing for the evaluation of spatial memory performance in mice (Figure S9). We found that chronic optogenetic stimulation in the hippocampus of WT+SSFO mice resulted in spatial memory deficits, like the impairment seen here in 6-months-old 5xFAD mice (Figure 7D), or previously described.⁴¹

Finally, since previous observations have linked neuronal and network hyperactivity with an increase in the secretion and deposition of amyloid-β in the brain of transgenic mice carrying several AD-linked mutations,^17,18,42,43 we sought to measure the levels of soluble Αβ40 and Aβ42 in the brain of WT mice following our chronic optogenetic stimulation protocol. In the hippocampus of WT+SSFO mice the levels of soluble secreted Aβ42 were significantly increased as compared to the levels in the brain of WT+tdTom mice (Figure 7E). The levels of Aβ40 showed a trend of increase in the WT+SSFO mice, but this increase was not significantly different from the Αβ40 levels in the WT+tdTom (Figure 7E). Altogether, these results suggest that chronic neuronal hyperactivity is sufficient to induce synaptic transmission disruption and memory loss similar to that observed in the neurodegenerative process in AD, but also augments the amyloidogenic cleavage of APP.

Discussion

We adopted optogenetics and an SSFO to induce chronic neuronal hyperactivity and examine the effect of such manipulation in the hippocampus of young WT and 5xFAD mice. We present evidence of (1) an overt downregulation in the levels of both excitatory and inhibitory synaptic proteins and an activation of translation, phosphorylation and autophagy following neuronal hyperactivity in WT mice; (2) an upregulation of AD-associated proteins in WT+SSFO mice mimicking the AD proteome signature found in the 5xFAD mice; (3) a downregulation in the levels of proteins participating in RNA processing and protein phosphorylation in 5xFAD+SSFO mice; (4) spatial memory impairments in WT+SSFO mice upon light stimulation; and (5) elevated levels of soluble Αβ42 in WT hippocampi following chronic optogenetic stimulation. These findings strongly support the notion that rampant neuronal hyperactivity distorts cellular mechanisms regulating proteostasis similar to that seen in the prodromal phase of AD-like pathology. This implicates overt neuronal hyperactivity as a key driver of proteomic remodeling laying the groundwork for AD pathogenesis.

In various neurological disorders, cognitive decline has been associated with hippocampal hyperactivity and the failure to deactivate brain regions consisting of the default mode network (e.g., posterior cingulate cortex, precuneus, and retrosplenial cortex).^{6,44,45,46,47,48} Reducing neuronal hyperactivity pharmacologically has also been found to improve cognitive performance in humans with mild cognitive decline^49,50 suggesting that neuronal hyperactivity is a cause and not a compensatory mechanism of cognitive decline. But neuronal hyperactivity may have broader implications in disease onset. Apart from contributing to cognitive deficits, neuronal hyperactivity could drive several other pathological aspects related to AD, such as mTORC1 hyperactivity,⁵¹ Αβ release,¹⁸ synaptic loss,^52,53 and neuronal degeneration.⁴⁶

Indirect activation of the hippocampal network by CA1 stimulation

Although our optogenetic stimulation was restricted to CA1 pyramidal neurons, we observed c-Fos expression in upstream hippocampal regions, including CA3 and the dentate gyrus (Figures 1C and S1A). This suggests indirect recruitment of these areas, potentially via polysynaptic pathways involving the entorhinal cortex, disinhibition within local circuits, or back-propagating activity through non-canonical hippocampal loops.^54,55,56 Such circuit-level responses may reflect a broader engagement of the hippocampal network under conditions of sustained CA1 hyperactivity. Given that hippocampal hyperexcitability has been implicated in age-related cognitive decline and AD,^{44,46,57,58,59} future studies are needed to delineate the pathways and mechanisms through which CA1-driven activity propagates within the hippocampus and contributes to network-wide dysfunction.

Self-reinforcing cycle of amyloid-β and neuronal hyperactivity

Previous studies have demonstrated that neuronal stimulation—whether electrical, pharmacological, or optogenetic—can enhance the secretion of Αβ from the presynaptic terminals in various APP transgenic mouse lines.^17,18,42,43 In turn, elevated Αβ levels have been shown to induce neural hyperexcitability and network dysfunction in both WT and APP transgenic mouse lines.^16,60 Similarly, in humans, brain regions with prominent Αβ aggregation often exhibit markedly elevated basal metabolic rates and neuronal activity.^8,42,61 Notably, even soluble monomers and small oligomers of Aβ42 can directly increase membrane excitability and drive network hyperactivity.^16,17,53 Together, these findings support the hypothesis of a self-reinforcing, Αβ-driven cycle of neuronal hyperactivity in AD,¹⁶ aligning with our results showing elevated soluble Αβ42 levels and multifaceted disruption of synaptic transmission in WT animals subjected to chronic neuronal hyperactivity. However, amyloid fragments derived from the rodent App are less prone to aggregation than those from human APP mutations.⁶² This difference may reflect structural variations between human and mouse-derived Αβ fibrils, or it may be due to the limited lifespan of rodents, which constrains the time available for Αβ precipitation. Nonetheless, it is plausible that the mechanism mediating the amyloidogenic APP cleavage is conserved across species and does not depend on APP mutations^62,63,64—a notion further supported by our findings.

A study using chemogenetics to chronically enhance excitability in CA1 parvalbumin (PV) interneurons demonstrated that prolonged interneuron hyperactivity can disrupt hippocampal circuits, leading to compensatory excitatory hyperactivity and impaired memory.⁶⁵ Although initially adaptive, this excitatory hyperactivity can render the system highly vulnerable to even low levels of Aβ, ultimately triggering circuit collapse and cognitive deficits.⁶⁵ Together with our proteomic data showing that chronic excitatory hyperactivity in WT mice induces AD-like molecular changes and Aβ42 secretion, these findings highlight how both excitatory and inhibitory imbalances can independently initiate AD-like pathology. Restoring excitation-inhibition balance may therefore represent a critical therapeutic strategy.

A neural hyperactivity-based paradigm of sporadic AD

The similarity we found in altered proteostasis and spatial memory in overstimulated WT and 5xFAD mice, together with the finding that neuronal hyperactivity in WT mice augments Αβ42 secretion establish chronic optogenetic stimulation in WT animals as a potential paradigm to model sporadic AD-pathogenesis. Numerous transgenic mouse lines have been developed to mimic the genetic variants of AD,⁶⁶ however, reproducing the sporadic forms of the disease, which constitutes over the 97% of all AD cases, remains arduous.⁶⁷ Rodent models, although valuable, are constrained by several limitations that impede their ability to fully recapitulate the complexity and etiology of AD. Additionally, the mechanisms and pathology observed in rodent models may not consistently mirror human AD due to interspecies disparities and the multifaceted nature of the disease.

Proteomic parallels between early neural hyperactivity and AD pathogenesis

A recent multi-omics study comparing the proteomes of brain tissue and cerebrovascular fluid from AD patients at different stages to those of the 5xFAD mice found that only the proteomic signature of 12-month-old 5xFAD mice resembled that of symptomatic AD.³² Specifically, the 12-month-old 5xFAD mice exhibited upregulation of 15 Aβ-related proteins, also identified in symptomatic AD patients, however, they showed unique activation patterns in autophagy, and interferon response while they lack certain human-specific deleterious events such as downregulation of neurotrophic factors.³² Notably, there was no proteome similarity between young 3-month-old 5xFAD mice and late-stage AD humans. Among the Αβ-related proteins, only the Slit homolog 2 protein (SLIT2) was found upregulated at such a young age,³² corroborating our findings in the hippocampus of 3-month-old 5xFAD mice (Table S2). Other studies have utilized proteomics to identify altered biological processes or novel biomarkers of AD, but they exclusively focus on symptomatic transgenic lines, highlighting additional proteins involved in the pathogenesis of familial AD.^33,68 In contrast, in our study we specifically examined the consequences of induced neuronal hyperactivity in young 5xFAD or WT mice, providing insights into the early initiation of AD pathology in both mutated and non-mutated genetic backgrounds. This is particularly valuable given the distinct proteomic signatures at different stages of disease progression, shown in the multi-omics studies comparing the proteomes of AD patients to those of 5xFAD mice.^32,33,68

Several proteomics studies to date provided strong evidence that synaptic proteins levels are significantly downregulated in the brains of AD patients, which likely contribute to the synaptic dysfunction and cognitive decline associated with the disease.^69,70,71 Interestingly, some of the key synaptic proteins identified as downregulated in highly vulnerable brain regions to AD, such as the guanine nucleotide-binding protein G(i), complexin 1, syntaxin 1A and syntaxin 1B,⁶⁹ were also found to be significantly downregulated in our unbiased proteomics analysis of hippocampi in WT+SSFO mice following chronic optogenetic stimulation. Further studies comparing the proteome of brains from rodents or humans experiencing chronic neuronal hyperactivity and early-stage AD patients will allow us to draw direct causal links between sustained neuronal hyperactivity and the onset of the disease.

Loss of proteomic reserve in 5xFAD mice

The distinct proteomic response to chronic optogenetic stimulation in the 5xFAD mice may reflect an occlusion of hyperactivity-induced effects, or a loss of proteostasis reserve. We hypothesize that pre-existing proteostatic stress in 5xFAD mice limits further activation of translation and autophagy pathways, resulting instead in a net downregulation—possibly reflecting homeostatic exhaustion of the proteostasis machinery. This aligns with a stage-dependent trajectory of AD, where early disease (modeled by WT+SSFO) shows upregulation of compensatory proteostatic mechanisms, while later stages (modeled by 5xFAD transgenics) exhibits decompensation and network collapse. Our findings are consistent with prior studies showing that neuronal hyperactivity emerges early in AD, contributing to synaptic and circuit disruption.^6,7,72,73 And yet, while these deficits in circuits arising from neuronal degeneration in AD might appear beyond repair, suppressing excessive neuronal activity early in the progression of the disease may represent a possible therapeutic modality to delay or modify the disease progression and cease the vicious cycle of relentless hyperactivity.

Conclusion

Our study establishes a non-transgenic model for investigating early contributors to sporadic AD, demonstrating that sustained neuronal hyperactivity alone is sufficient to induce proteomic signatures reminiscent of early-stage pathology. This model provides a valuable platform to probe the mechanisms underlying AD initiation and progression. Our findings strongly support the notion that excessive neuronal activity disrupts proteostatic regulation in ways that mirror the prodromal phase of the disease, implicating hyperactivity as a key driver of proteomic remodeling that may lay the groundwork for AD pathogenesis. Moving forward, longitudinal molecular analyses, broader genetic models, and targeted circuit-level manipulations will be essential to clarify the long-term impact and disease relevance of CA1-driven hyperactivity in AD.

Limitations of the study

While our study provides compelling evidence that chronic optogenetic activation of CA1 pyramidal neurons induces AD-like proteomic remodeling and memory deficits, several limitations should be acknowledged. First, the use of broad excitatory stimulation does not distinguish between subpopulations of CA1 neurons, nor does it replicate the precise spatiotemporal patterns of activity associated with early disease states. Second, although indirect activation of DG and CA2/3 neurons was observed, the underlying circuit mechanisms remain speculative and were not directly tested, limiting causal insight into how activity propagates across hippocampal subfields. Third, proteomic changes were assessed at a single time point immediately after the stimulation period; whether these molecular alterations persist long term or reverse after cessation of stimulation remains unknown. Fourth, while the 5xFAD model enabled proteomic and behavioral comparisons, no additional Alzheimer’s disease models—such as tauopathy lines—were included, limiting the generalizability of the findings across disease subtypes.

Resource availability

Lead contact

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Yves De Koninck (yves.dekoninck@neuro.ulaval.ca).

Materials availability

Plasmids for the 2 viruses used in this study are available at Canadian Optogenetics and Vectorology Foundry Core Facility (see key resources table). The wild-type and transgenic 5xFAD mice are available at Jackson Laboratory (see key resources table).

Data and code availability

• •Mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the PRIDE partner repository with the dataset identifier PXD044437 and are publicly available at: https://www.ebi.ac.uk/pride/ as of the date of publication.
• •Software to perform consensus clustering analysis is available at: https://doi.org/10.5281/zenodo.16734504 and is publicly available as of the publication date.
• •Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

Acknowledgments

The authors thank Camille Sugère and Lyane Méthot for assistance with the daily optogenetic stimulation protocol, Drs Efthimios MC Skoulakis and Nicolò Ilacqua for their invaluable comments on the proteomics data and Dr. Arkady Khoutorsky for his guidance with the immunoblots related to mTOR signaling. The authors also thank Dominique Isabel for assistance with RStudio and data visualization.

This work was supported by the grants TR192089 from the Weston Family Foundation, and FDN-159906 from the Canadian Institutes of Health Research (CIHR), and the Canada Research Chair program to Y.D.K. P.D. was supported by grants 06887 and 06492 from the Natural Sciences and Engineering Research Council of Canada. I.K. was partially supported by the Norampac Research grant on Alzheimer’s disease and Related Diseases and the Fondation De La Famille Lemaire research grant on Alzheimer’s disease and Related Diseases through the Faculty of Medicine, Université Laval.

Author contributions

Conceptualization and experimental design, I.K., K.P., and Y.D.K.; surgeries, behavioral testing, immunoblots and ELISA, I.K.; optrode recordings and analysis, I.K. and J.A.; mass spectrometry and proteomics raw data processing, M.S.; proteomics data analysis, I.K., P.D., and K.P.; Pyhton code development and implementation, P.D.; writing—original draft, I.K., K.P., and Y.D.K.; writing—review and editing, I.K., P.D., K.P., and Y.D.K.; funding acquisition, Y.D.K.; supervision, K.P. and Y.D.K.

Declaration of interests

The authors declare no competing interests.

STAR★Methods

Key resources table

REAGENT or RESOURCE	SOURCE	IDENTIFIER
Antibodies
anti-S6 Ribosomal Protein	Cell Signaling	Cat# 2217; RRID: AB_331355
anti-Phospho-S6 Ribosomal Protein (Ser240/244)	Cell Signaling	Cat# 5364; RRID: AB_10694233
anti-Rheb	Cell Signaling	Cat# 13879; RRID: AB_2721022
anti-EAAT1	Abcam	Cat# ab176557
anti-EAAT2	Abcam	Cat# ab205248
anti-c-Fos	Cell Signaling	Cat# 2250; RRID: AB_2247211
anti-Calbindin	Millipore	Cat# ABN2192; RRID: AB_2935805
IRDye® 680RD Goat Anti-Mouse	LI-COR Biosciences	Cat# 926-68070
IRDye® 800CW Goat Anti-Rabbit	LI-COR Biosciences	Cat# 926-68070
Goat Anti-Rabbit IgG (H + L) HRP	Invitrogen	Cat# 32460
Goat anti-Rabbit IgG (H + L) Cross-Adsorbed Secondary Antibody, Alexa Fluor™ 488	Invitrogen	Cat# A-11008
Bacterial and virus strains
AAV2/5-CaMKIIa-hChR2(C128S/D156A)-mCherry	Canadian Optogenetics and Vectorology Foundry Core Facility	N/A
AAV2/8-CaMKIIa-tdTomato	Canadian Optogenetics and Vectorology Foundry Core Facility	N/A
Chemicals, peptides, and recombinant proteins
Mounting Medium with DAPI	Abcam	Cat# ab104139
Iodoacetamide	Acros Organics	Cat# 10346660
SpeedBead Magnetic Carboxylate	Cytiva	Cat# 65152105050250
SpeedBead Magnetic Carboxylate	Cytiva	Cat# 45152105050250
Proteomics grade modified Trypsin/LysC	Promega	Cat# VA1061
Acetonitrile Fisher Chemical™	Fisher Scientific	Cat# A955-500
Formic Acid	Merck Millipore	Cat# 1.00264.1000
4-15% Mini-PROTEAN© TGX Stain-Free™ acrylamide gels	BioRad	Cat# 17000546
Trans-Blot Turbo RTA Mini 0.45 μm LF PVDF Transfer Kit	BioRad	Cat# 1704274
Precision Plus Protein™ Kaleidoscope standards	BioRad	Cat# 1610375
Silk Fibroin	Sigma-Aldrich	Cat# 5154
Critical commercial assays
Human/Rat β Amyloid (40)	Wako	Cat# 294-62501
Human/Rat β Amyloid (42)	Wako	Cat# 290-62601
Deposited data
Proteomics Data	This study	https://ebi.ac.uk.pride/archive/projects/PXD044437
Experimental models: Organisms/strains
B6SJL-Tg(APPSwFlLon, PSEN1∗M146L∗L286V)6799Vas/Mmjax	Jackson Laboratory	RRID: MMRRC_034840-JAX
B6SJLF1/J	Jackson Laboratory	RRID: IMSR_JAX:100012
Software and algorithms
Consensus Clustering Code (Zenodo)	This study	https://doi.org/10.5281/zenodo.16734504
GraphPad Prism	GraphPad Software	https://www.graphpad.com/
Python	Python Software Foundation	https://www.python.org/
Fiji	Schindelin et al.74	https://fiji.sc/
Spike2	Cambridge Electronic Design	https://ced.co.uk/products/spkovan
DIA-NN version	Demichev et al.75	https://github.com/vdemichev/DiaNN
Perseus	Tyanova et al.76	https://maxquant.net/perseus/
SynGO	Koopmans et al.30	https://www.syngoportal.org/
GeneCodis	Garcia-Moreno et al.28	https://genecodis.genyo.es/
STRING	Szklarczyk et al.29	https://string-db.org/
EstimationStats	Ho et al.77	https://www.estimationstats.com/#/
Other
470 nm LED diode	ThorLabs	Cat# M470F3
2.5 mm (FC) Ferrule Patch Cables	ThorLabs	Cat# M81L01
Multimode Hybrid Fiber Optic Patch Cables, 200 μm Core, 0.39NA	ThorLabs	Cat# M75L
Pigtailed 1 × 1 Fiber-optic Rotary Joint	Doric Lenses	N/A
ADAF2 Quick-Release Interconnect for 2.5 mm Ferrules	ThorLabs	N/A
Fiber Optic Cannulae with Ø2.5 mm Ceramic Ferrule, 200 μm Core, 0.39NA	RWD	Cat# R-FOC-F200C-39NA
Laser Diode Fiber Light Source 488/050	Doric Lenses	Cat# LDFLS_488/050
Multimode Fiber Optic Patch Cables	ThorLabs	Cat# M122L
Micro-optrode (550 μm optical core; 0.29 NA; 250 μm hollow core)	LeChasseur et al.27	N/A

Experimental model and study participant details

Animals

Rodent experiments were carried out on male 5xFAD mice,²⁵ and their WT littermates bred and housed in the CERVO Brain Research Center animal facility. The male 5xFAD and female B6SJLF1/J breeders were purchased from the Jackson Laboratory (#34840-JAX and #100012, respectively). For the proteomic analysis, immunoblots, and ELISA quantifications, WT and 5xFAD mice were 12–14 weeks old at sacrifice. For the behavioral experiments, the WT+tdTom (N = 14) and WT+SSFO (N = 16) mice undergoing the chronic optogenetic stimulation were 12–13 weeks old when tested while the cohort of 5xFAD (N = 15) and WT (N = 9) mice used for comparison were 25–28 weeks old at the time of testing. Mice were housed on a 12 h day/night cycle with ad libitum access to food and water. Mice were randomly assigned to experimental groups. All experiments were approved by the committee for animal protection of Université Laval (CPAUL) and followed the guidelines from the Canadian Council for Animal Care.

We excluded female mice from our study to prevent introducing an additional biological variable. This could have obscured the interpretation of the optogenetic manipulations, especially given the marginal sex differences previously described in synaptic protein levels within the rodent hippocampus.^78,79

Viral vectors

The CERVO Canadian Optogenetics and Vectorology Foundry Core Facility (RRID: SCR_016477) manufactured all viruses used in our work. For the optogenetic stimulation protocol we expressed the stabilized step function opsin, hChR2(C128S/D156A), using AAV2/5-CaMKIIa-hChR2(C128S/D156A)-mCherry (7.7E12 GC/mL). To express a reference fluorescence protein in control animals, we used AAV2/8-CaMKIIa-tdTomato (1.3E13 GC/mL).

Method details

Surgical procedures and viral transduction

The tip of the optic fiber (200 μm core, 0.39 NA on 2.5 mm ceramic ferrules) was coated with 100 nL of a 1:1 mix of silk fibroin solution (Sigma 5154-20 ML) and the viral vector AAV2/5-CaMKIIa-hChR2(C128S/D156A)-mCherry (henceforth referred to as AAV2/5-CaMKIIa-SSFO-mCherry) or AAV2/8-CaMKIIa-tdTomato, prior to the surgery, and left to dry for 15 h at 4 °C as previously described.⁸⁰

Stereotaxic surgeries were performed on mice (4-weeks old) maintained on isoflurane anesthesia (1.5%–2% in O2). Briefly, we removed the skin atop the cranium and a ∼500-μm-diameter craniotomy was drilled over the hippocampus at coordinates relative to bregma of −2.30 mm rostro-caudal and −2.00 mm latero-medial. The optic fiber cannula was lowered 1.60 mm into the brain (−1.60 mm dorsoventral) and fixed on the cranium with C&B Metabond (Parkell Inc.). Following the surgery, mice were single-housed, and the chronic optogenetic stimulation experiments started 4–5 weeks later.

For the optrode recordings, the same viral vectors were unilaterally infused into each side of the hippocampal CA1 (rostro-caudal: −2.30 mm, latero-medial: 2.00 mm, dorsoventral: 1.60 mm). The infusion was performed with pulled borosilicate glass capillaries and a NANOLITER2020 injector (World Precision Instruments LLC). The in vivo optrode experiments started a minimum of 4 weeks later.

Optogenetic stimulation

After 4–5 weeks from the optic fiber implantation mice were optically stimulated for 2 s every 24 h (starting at 2-month of age) for 4 weeks with blue light (470 nm, 1 mW). The light was generated by an LED diode (ThorLabs M470F3) and delivered through fiber optics (200 μm core, 0.39 NA) to the implanted fiber optic cannulae. Mice remained awake during stimulation and were not anesthetized, even while connected to optic fiber delivering light.

In vivo extracellular optrode recordings

Simultaneous optical stimulation and electrical extracellular recording in the hippocampal CA1 was performed in WT male mice previously transduced with SSFO or tdTomato using a micro-optrode.²⁷ Briefly, mice were anesthetized with a mix of 100 mg ketamine, 15 mg xylazine and 2.5 acepromazine per kg and placed in a stereotaxic frame on a temperature control pad. A craniotomy was drilled over the hippocampus at coordinates relative to bregma of −2.30 mm rostro-caudal and −2.00 mm latero-medial. A micro-optrode (6–8 μm tip diameter; 550 μm optical core; 0.29 NA; 250 μm hollow core) was filled with 0.5 M potassium acetate solution, connected to a 488 nm laser diode (Doric lenses) through a multimode optical fiber (200 μm optical core; 0.22 NA; ThorLabs) and lowered vertically to the brain. Electrophysiological recordings were initiated at −1.3 mm ventral. A 2s long light pulse was delivered to activate the SSFO. The extracellular electrophysiological signal was amplified (Neurodata IR183, Cygnus technology), filtered (band pass, 300–3,000 Hz, model 440, Brownlee Precision), and digitized. The filtered traces were analyzed with Spike2 (Cambridge Electronic Design). Events were detected and the firing rate of individual neurons was plotted in peristimulus time histograms (PSTH) with 10 s bins. Neurons for which the standard deviation of firing rate for the light pulse bin was 1.5 times superior to the basal firing rate (average of the firing rate during the 60s preceding the pulse of light) were considered activated.

Sample preparation

For the proteomics analysis, brains were excised from anesthetized mice (30% urethane in saline) within 30 min after optogenetic stimulation. Hippocampi ipsilateral to the optic fiber were dissected, collected into sample tubes, immediately flash-frozen in liquid nitrogen, and stored at −80 °C until use. 25 mg of frozen tissue was lysed in 150 μL of freshly made lysis buffer (4% SDS, 0.1 M DTT, 100 mM Tris/HCl; pH 7.6) and homogenized with an electric FisherBrandTM Pellet PestleTM homogenizer (Thermo Fisher Scientific). Homogenates were incubated at 95 °C for 3 min and centrifuged at 16,000 g for 5 min. The supernatants were transferred into sample tubes and stored at −80 °C.

For ELISA and immunoblots, tissue was harvested as before within 30 min after optogenetic stimulation. 25 mg of frozen hippocampus tissue was homogenized in 150 μL of RIPA buffer (50 mM Tris-HCl, 150 mM NaCl, 1% Triton X-100, 0.5% Na deoxycholate, 0.1% SDS; pH 8.0). Homogenates were incubated under constant agitation for 4 h at 4 °C for lysis. Lysates were centrifuged at 17,200 g and 4 °C for 10 min. Supernatants were collected and stored at −30 °C.

Protein mass spectrometry

Protein mass spectrometry analysis was performed in the Proteomics Facility of Biomedical Sciences Research Center Alexander Fleming. Four to five biological replicas for each condition (WT + AAV2/8-CaMKIIa-tdTomato; WT + AAV2/5-CaMKIIa-SSFO-mCherry; 5xFAD + AAV2/8-CaMKIIa-tdTomato; 5xFAD + AAV2/5-CaMKIIa-SSFO-mCherry) were analyzed. Briefly, proteins of the lysed samples were processed according to the sensitive Sp3 protocol.29 The reduced cysteine residues were alkylated in 200 mM iodoacetamide (Acros Organics). 20 μg of beads (1:1 mixture of hydrophilic and hydrophobic SeraMag carboxylate-modified beads; GE Life Sciences) were added to each sample in 50% ethanol. Protein clean-up was performed on a magnetic rack. The beads were washed twice with 80% ethanol followed by one wash with 100% acetonitrile (Fisher Chemical). The beads-captured proteins were digested overnight at 37 °C with 0.5 μg trypsin/LysC mix in 25 mM ammonium bicarbonate under vigorous shaking (1200 rpm, Eppendorf Thermomixer). The supernatants were collected, and the peptides were purified by a modified Sp3 clean-up protocol and finally solubilized in the mobile phase A (0.1% formic acid in water), and sonicated. Peptides concentration was determined through absorbance measurement at 280 nm using a nanodrop instrument.

Peptides were analyzed by a liquid chromatography-tandem mass spectrometry (LS-MS/MS) on a setup consisting of a Dionex Ultimate 3000 nanoRSLC online with a Thermo Q Exactive HF-X Orbitrap mass spectrometer. Peptidic samples were directly injected and separated on a 25 cm-long analytical C18 column (PepSep, 1.9 μm3 beads, 75 μm ID) using a 90 min long run. The full MS was acquired in profile mode using a Q Exactive HF-X Hybrid Quadropole-Orbitrap mass spectrometer operating in the scan range of 375–1400 m/z using 120 K resolving power with an Automatic Gain Control (AGC) of 3 × 106 and a max IT of 60 ms followed by data independent analysis (DIA) using 8 Th windows (39 loops counts) with 15 K resolving power with an AGC of 3 × 105, a max IT of 22 ms and normalized collision energy (NCE) of 26.

Immunoblots

Hippocampal lysates were diluted in Laemmli buffer at a concentration of 10 μg per 20 μL and denaturized for 5 min at 95 °C. Proteins were separated in 4–15% Mini-PROTEAN TGX Stain-Free acrylamide gels (BioRad) and transferred onto 0.45 μm low-fluorescence PVDF membranes (BioRad). Membranes were probed with the rabbit monoclonal anti-EAAT1 (Abcam ab176557), anti-EAAT2 (Abcam ab205248), anti-S6 (Cell Signaling 2217), anti-S6 Ser240/244 (Cell Signaling 5364), anti-Rheb (Cell Signaling 13879), anti-c-Fos (Cell Signaling 2250) or the rabbit polyclonal anti-Calbindin (Millipore ABN2192) in Tris-buffered saline containing 0.2% Tween and 5% bovine serum albumin at 4 °C. All antibodies were used at 1:1000, except for the anti-S6 Ser240/244 which was used at 1:10000. The appropriate anti-mouse or anti-rabbit IgG conjugated with IRDye©800 or IRDye©680, respectively, were used at 1:10000. Proteins were visualized using a ChemiDoc Imaging System (BioRad) and the intensity of protein bands was quantified using Fiji.80 To normalize for sample loading, the intensity of total protein (TLP) measured on the TGX Stain-Free gels was used as previously described.⁸¹

ELISA

Levels of Αβ40 and Aβ42 in hippocampal samples were quantitated by Human/Rat β Amyloid (40) ELISA kit (Wako 294–62501, LOT# WTL5240) and Human/Rat β Amyloid (42) ELISA kit (Wako 290–62601, LOT# WTM4353) respectively. Both ELISAs were performed according to the manufacturer recommendations, and the plates were read at 450 nm using an Eon microplate reader (BioTek).

Immunofluoresence

A Leica Vibratome VT1220S (Leica Microsystems) was used to cut 100 μm coronal sections of paraformaldehyde fixed brain tissue. Sections were rinsed 3 times in 0.1M phosphate-buffered saline with 0.2% Triton X-100 (PBST) for 10 min, blocked for 1 h with 10% Normal Goat Serum (NGS) in PBST and then incubated for 15 h at 4 °C in primary anti-c-Fos (rabbit monoclonal, 1:2000, Cell Signaling 2250) diluted in PBST containing 4% NGS. Sections were washed in PBST and subsequently incubated for 2 h at room temperature in AlexaFluor 488-conjugated goat anti-rabbit (1:500, Invitrogen A11008) diluted in PBST containing 4% NGS. Tissue was mounted on SuperFrost slides (Thermo Fisher Scientific 12-550-15) using a fluorescence mounting medium with DAPI (Abcam ab104139) and cover-slipped.

All confocal images were acquired with a Zeiss LSM710 confocal laser scanning microscope. Acquisitions were 12-bit images, 2048 × 2048 pixels with a pixel dwell time of 3.15 μs. A 40× Plan-Apochromat oil objective (1.4 NA) was used for magnification. Tile scans were acquired at 12-bits, 512 × 512 pixels (each tile) with a pixel dwell time of 0.64 μs and an x5 EC-Plan-Neofluar objective (0.16 NA). Laser power, photomultiplier tube (PMT) settings, filters, dichroic mirrors, scanning speed were kept constant for all acquisitions.

Y-maze spatial memory test

The Y-maze apparatus consisted of 3 identical arms (each 36.2 × 8.25 cm) in a Y configuration (placed at 120° to each other) connected by a center polygonal area. The walls of the arms were transparent so the animals could identify two distinct visual cues placed outside the apparatus to facilitate spatial navigation. During the training phase, the mice were placed individually in the maze and left freely to explore two arms, for 10 min, while the third arm was blocked by a removable door. An hour later, the mice were returned to the maze and allowed to explore all arms. The latency to enter the new (third) arm was measured for each mouse and used to score the spatial memory performance of the mice. Between mice, the maze was cleaned with 70% isopropyl solution to minimize scent cues. Entry into the arm was defined as a mouse placing all four paws on the arm. For the mice receiving the daily optogenetic stimulation, the spatial memory test was performed after the end of the 4 weeks-long stimulation protocol. The day of the test, the mice receive a 2s light pulse before the beginning of the training phase.

Quantification and statistical analysis

Label-free quantification and data analysis

Orbitrap raw data were analyzed in DIA-NN 1.8 (Data-Independent Acquisition by Neural Networks)⁷⁵ against the complete Uniprot Mus musculus proteome (Downloaded April 16, 2021) supplemented with APP, presenilin, mCherry and ChR2. Search parameters were set to allow up to two possible trypsin/P enzyme missed cleavages. A spectra library was generated from the DIA runs and used to reanalyze them. Cysteine carbamidomethylation was set as a fixed modification while N-terminal acetylation and methionine oxidations were set as variable modifications. The match between runs (MBR) feature was used for all the analyses and the output (precursor) was filtered at 0.01 false discovery rate (FDR). The protein inference was performed on the gene level using only proteotypic peptides. The double pass mode of the neural network classifier was also activated. Perseus (version 1.6.15.0)⁷⁶ was used for data processing and statistical analysis. Proteins were subjected to filtering based on valid values setting a threshold of 70% valid values in at least one group. Remaining missing values were imputed separately for each column (based on the normal distribution using width of 0.3 and down shift of 1.8). Two-group comparisons were performed using Dunnett’s t-test, while multiple-group comparisons were analyzed using two-way ANOVA, applying a permutation-based FDR threshold of 0.05. Statistically significant proteins were Z-scored and hierarchically clustered using the Euclidean or cosine distance and the average linkage method for determining cluster distances and then visualized as heat maps.

Gene-Ontology analysis was performed with GeneCodis²⁸ by uploading the list of altered proteins based on the t-test two-sample comparisons. We used SynGO to perform synapse-specific enrichment analysis³⁰ by first converting the gene names listed in Tables S1, S2 and S3 to their respective human SynGO identifiers. We then identified overrepresented synaptic terms by setting a minimum inclusion threshold of five genes/proteins per term and excluding annotations supported solely by proteomics evidence. The interaction network analysis of functional or physical protein to protein associations was performed with STRING (version 11.5).²⁹

Consensus clustering analysis

Comprehensive consensus clustering analysis was performed using the Z-scored hippocampal proteomic profiles. We systematically varied the ANOVA P value threshold for protein inclusion across 80 distinct experiments. For each threshold, proteins with P values below the specified cutoff were retained, and a consensus clustering matrix was generated based on 2,000 bootstrap replicates. In each replicate, the rows (i.e., proteins) were resampled with replacement to simulate experimental variability. Additionally, Gaussian white noise $\left(\sigma =2.0\right)$ was added to each replicate to account for measurement uncertainty. Hierarchical clustering was then applied to each perturbed dataset. For every pair of individuals, the fraction of times they were co-assigned to the same cluster across all bootstrap replicates was recorded to build a consensus (co-association) matrix. This symmetric matrix with values in the range $\left[0,1\right]$ quantifies how consistently individual samples co-cluster under repeated perturbations.

To evaluate the intrinsic dimensionality and structural robustness of the consensus matrices—particularly under the influence of noise—we computed three effective rank measures as previously described⁸¹: stable rank $\sum _{i=1}^r\frac{{\sigma }_i^2}{{\sigma }_1^2}$, nuclear rank $\sum _{i=1}^r\frac{{\sigma }_i}{{\sigma }_1}$, and entropy rank $\mathit{exp}\left[-\sum _{i=1}^r\frac{{\sigma }_i}{\sum _{j=1}^r{\sigma }_j}\mathit{log}\frac{{\sigma }_i}{\sum _{j=1}^r{\sigma }_j}\right]$, where ${\sigma }_1,\dots ,{\sigma }_r$ denote the nonzero singular values of a given consensus matrix, ordered in decreasing magnitude. These measures provide more reliable indicators of matrix structure than conventional rank, here denoted by $r$, which is highly sensitive to perturbations. Effective ranks are especially suited to identifying low-dimensional block structures in near–low-rank matrices, such as those expected from strongly clustered data. An effective rank near $q$, where q≤r by definition, indicates the presence of $q$ dominant clusters, while a sharp decrease in effective rank reflects reduced clustering signal or the inclusion of noisy or weakly informative features.⁸¹

Statistics

For the proteomic experiments, the statistical analysis of the label-free quantification intensities was performed with Perseus (version 1.6.15) using a two-sample t-test with a P value less than 0.05 or an ANOVA for multiple sample tests with a P value lower than of 0.05. The fits for the nonlinear regressions were assessed using a sum-of-squares F test, with a significance level (alpha) set at 0.05. For estimation statistics based on effect size and confidence intervals,⁷⁷ the data were uploaded and analyzed on https://www.estimationstats.com/. The estimation plots show the mean difference between the groups and the 95% confidence interval by the ends of the vertical error bar. Bar graphs were created using GraphPad Prism 9 (GraphPad software) and represent the group’s mean ± the standard error of the mean (SEM). ‘N’ (or ‘n’) indicates the number of mice, unless stated otherwise. Circos plots were drawn using RStudio by uploading the list of proteins enriched in our samples for each biological process selected by the gene-ontology analysis.

Published: August 28, 2025