Chronic Gq activation of ventral hippocampal neurons and astrocytes differentially affects memory and behavior

Network dysfunction is implicated in numerous diseases and psychiatric disorders, and the hippocampus serves as a common origin for these abnormalities. To test the hypothesis that chronic modulation of neurons and astrocytes induces impairments in cognition, we activated the hM3D(Gq) pathway in CaMKII+ neurons or GFAP+ astrocytes within the ventral hippocampus across 3, 6 and 9 months. CaMKII-hM3Dq activation impaired fear extinction at 3 months and acquisition at 9 months. Both CaMKII-hM3Dq manipulation and aging had differential effects on anxiety and social interaction. GFAP-hM3Dq activation impacted fear memory at 6 and 9 months. GFAP-hM3Dq activation impacted anxiety in the open field only at the earliest time point. CaMKII-hM3Dq activation modified the number of microglia, while GFAP-hM3Dq activation impacted microglial morphological characteristics, but neither affected these measures in astrocytes. Overall, our study elucidates how distinct cell types can modify behavior through network dysfunction, while adding a more direct role for glia in modulating behavior. Highlights CaMKII- and GFAP-Gq activation impacted memory, anxiety, and social behaviors. Novel environment exploration was affected by CaMKII- and GFAP-Gq activation. CaMKII-Gq modified microglial number, while GFAP-Gq affected microglial morphology. Neither cell manipulation affected astrocytic number or morphology.

Full slice images of RFP amplified hM3Dq expression (Figure 1B,D) were captured along with zoomed in a 300 x 300 tile (micrometer) z-stack at 20x magnification. 3-4 different slices were imaged for each animal for averaging, for each brain region of interest. Images of NeuN, GFAP, and Iba-1 were captured in a 300 x 300 single-tile (micrometers) z-stack at 20x magnification. 3-4 mice per group, with 18 single-tile ROIs covering the vHPC were used for cell counts of NeuN, GFAP and Iba-1. Morphological analysis of microglia (Iba1) and astrocytes (GFAP) was performed using 3DMorph, a MATLAB-based tool that analyzes glial morphology from 3-dimensional imaging data (York et. al., 2018). For cFos counts, 3 mice per group with 4 images (1182 x 1756 micrometers) each covering the vCA1 pyramidal cell layer 20x magnification were used. Each pyramidal cell layer ROI was hand-traced in ImageJ and cFos counts were normalized to the number of DAPI (%cFos/DAPI). NeuN, GFAP, Iba-1, DAPI and cFos counts were performed using Ilastik, a machine-learning-based image analysis tool (Berg et. al., 2019).
For Gq receptor fluorescence intensity measures (Supplemental Figure 7), sagittal slices of vHPC for CaMKII-and GFAP-hM3Dq groups were imaged for mCherry expression, (no immunohistochemistry was used to amplify the signal). Here, we imaged DAPI and Gq-mCherry expression for 3 mice with 3 vHPC ROI's each (2938 x 3513 micrometers), across groups (GFAP/CaMKII-hM3Dq) x time points (3,6,9 months). These images were taken using a 10x objective over the entire vHPC, with the ROI maintained across images, slices and animals. All imaging parameters were identically maintained across all animals and slices for fluorescence intensity comparison. Fluorescence intensity measures were conducted in ImageJ using two methods to ensure consistency. Method 1: select vHPC ROI (entire image taken), Analyze, Set Measurements, Select 'mean gray value', Analyze, Measure. The mean gray value was used as our measure, as all images were the same area and all settings were maintained, so this should be standardized across all mice and groups, unlike 'integrated density' that takes different areas into account. Method 2: select 3 ROIs in the fluorescent region and 3 ROIs in the dark 'background' of each image at random, then use the discussed steps to gather 'mean gray value', 'area' and 'integrated density' values for each small ROI. These values were averaged for each image, then averaged again for each mouse for a final 'background' and 'fluorescent' mean gray, area and integrated density value. Using these metrics, we calculated the average corrected total cell fluorescence (CTCF) using integrated density of the 'fluorescent' region -(area of the 'fluorescent' ROI x mean gray fluorescence of 'background' region).

2.7) Statistical Methods
Data was analyzed using GraphPad Prism. All data were tested for normality using Shapiro-Wilk and Kolmogorov-Smirnov tests. Outliers were removed prior to statistical analysis using the ROUT method recommended by GraphPad Prism, which uses identification from nonlinear regression. We chose a ROUT coefficient Q value of 10% (False Discovery Rate), making the threshold for outliers less-strict and allowing for an increase in power for outlier detection. To analyze differences between groups and across timepoints, we used: Two-way ANOVAs (betweensubject factor: Group; within-subject factor: Timepoint). Alpha was set to p<0.05. Post-hoc analyses were run using Tukey's-multiple comparisons test with a 95% CI. To analyze differences between groups and across time within a single session we used: Two-way repeated measures (RM) ANOVAs (between-subject factor: Group; within-subject factor: Time). Alpha level was set to 0.05. Post-hoc analyses were run using Šídák's multiple comparisons test with a 95% CI. To analyze differences between groups, we used: Independent t-tests [CaMKII-hM3Dq vs. mCherry; GFAP-hM3Dq vs. mCherry]. Alpha was set to 0.05. AAV5-GFAP-hM3Dq-mCherry AddGene 50487 -AAV5

3.1) Induction of network dysfunction in neurons or astrocytes across time.
Network dysfunction is an early hallmark of neurodegenerative diseases such as Alzheimer's Disease (AD) both in humans and rodent models (Lauterborn et. al., 2021;Vico Varela et. al., 2019). However, it is unknown whether chronic activation of astrocytes or neurons in the vHPC may induce behavioral and histological changes related to neurodegeneration in the brains of wild type mice. To address this question, we used chemogenetic methods to chronically activate the hM3D(Gq) pathway in either CaMKII+ neurons or GFAP+ astrocytes in the vCA1 region of the HPC. For the neuronal groups, wild type mice were injected bilaterally with AAV5-CaMKII-hM3D(Gq)-mCherry or control vector AAV5-CaMKII-mCherry to selectively express excitatory DREADD receptors in vCA1 ( Figure 1A-B). For the astrocyte groups, C57BL/6 wild type mice were injected bilaterally with AAV9-GFAP-hM3D(Gq)-mCherry or control vector AAV5-GFAP-mCherry to selectively express DREADD receptors in astrocytes within vCA1 as well ( Figure 1C-D). Our viruses were robustly expressed in each cell type, as indicated by co-staining with RFP/GFAP+ and RFP/NeuN+ for DREADD receptors and their unique cell markers ( Figure 1B,D). After 4 weeks of recovery post-surgery, a water-soluble DREADD ligand, deschloroclozapine dihydrochloride (DCZ) was administered via the animal's daily water to chronically activate the hM3D(Gq) receptors of neurons or astrocytes ( Figure 1E) (Nagai et. al., 2020). DCZ was selected for its high potency, selectivity and affinity for hM3D(Gq) receptors compared to previously used DREADD agonists, such as clozapine-N-ozide (CNO) and compound 21 (C21) (Nagai et. al., 2021). Even at low doses (0.001-0.1 mg/kg), DCZ enhances neuronal activity when bound to hM3Dq receptors expressed in the mouse and monkey brain (Nagai et. al., 2021;Nentwig et. al., 2022). Chronic activation of these DREADD-receptors via the intraperitoneal (I.P.) route may cause pain or unnecessary stress for the animal, even with adequate habituation (Mimura et. al., 2021). To circumvent this issue, recent studies have employed administration of CNO with water or food (Padilla et. al., 2017;Zhan et. al., 2019;Fernandez et. al., 2018). DCZ has been used recently in this manner for voluntary oral (P.O.) administration, and these experiments have demonstrated no significant differences between the I.P. and P.O. routes on behavior at the same dose of 0.5mg/kg (Ferrari et. al., 2022). Limitations of this method include decreased control over timing and dose of administration (Mimura et. al., 2021), but this voluntary P.O. administration was deemed superior for chronic stimulation of these receptors for our experiments here. Given this, mice underwent chronic activation for 3, 6 or 9 months, for a total of 12 groups: 3 Month GFAP mCherry and 3 Month GFAP hM3Dq; 3 Month CaMKII mCherry and 3 Month CaMKII hM3Dq; 6 Month GFAP mCherry and 6 Month GFAP hM3Dq; 6 Month CaMKII mCherry and 6 Month CaMKII hM3Dq; 9 Month GFAP mCherry and 9 Month GFAP hM3Dq; 9 Month CaMKII mCherry and and 9 Month CaMKII hM3Dq ( Figure 1E). At each group's designated timepoint, mice were subjected to a battery of behavioral tasks including open field, zero maze, social interaction, y-maze, contextual fear conditioning (CFC), recall and extinction ( Figure 1E). Brain tissue was obtained following behavioral testing and was used to perform various histological assessments. Finally, to measure whether our manipulations were killing neurons within the vCA1, a confounding variable, we quantified a neuron-specific nuclear marker, NeuN (Figure 1F-I; Supplemental Figure 3A-F). Analysis of CaMKII-hM3Dq and -mCherry groups revealed no interaction between time point and group for the number of NeuN+ cells, nor did each factor alone contribute to a change in NeuN+ cells ( Figure 1F-G)(Two-way ANOVA; Interaction F(2,12)=0.5069, p=0.6147; Timepoint: F (2, 12) = 0.5594, p=0.5858; Group: F (1, 12) = 0.01321, p=0.9104). However, while analysis of the GFAP groups revealed no interaction between time point and group for the number of NeuN+ cells, there was an impact of group alone (Two-way ANOVA; Interaction: F (2, 12) = 0.1286, p=0.8805; Timepoint: F (2, 12) = 0.8680, p=0.4446; Group: F (1, 12) = 5.227, p=0.0412)( Figure 1H-I). This suggests that our manipulation of the GFAP-Gq pathway is only mildly affecting the number of NeuN+ cells in a time-independent manner, possibly indicating that there is an initial 'insult' or loss of neurons that does not worsen over time with our chronic Gq activation.
To confirm the successful acute activation of our Gq-DREADD receptors, we performed intraperitoneal injection of deschloroclozapine (DCZ) in separate cohorts of CaMKII-hM3Dq, CaMKII-mCherry, GFAP-hM3Dq and GFAP-mCherry mice and sacrificed them 90 minutes later to stain for peak cFos protein levels. Additionally, we sacrificed wild-type mice without the administration of DCZ to understand baseline cFos levels in vHPC and provide insight into the inherent change in cFos due to I.P. injection alone or with the addition of the ligand in the brain. Importantly, recent evidence has shown that DCZ and saline administration did not show any significant differences in cFos levels in rats lacking DREADD receptor expression, suggesting that DCZ alone does not impact cFos levels (Nentwig et. al., 2022). We found that across CaMKII-hM3Dq+DCZ, CaMKII-mCherry+DCZ, and -DCZ groups, there was a significant difference in mean cFos levels (One-way ANOVA; F(2,6)= 137.4, p<0.0001). Specifically, multiple comparisons revealed that DCZ administration significantly increased cFos levels between CaMKII-mCherry and CaMKII-hM3Dq + DCZ groups, indicating an increase in neuronal activity in vHPC due to presence of the ligand (Tukey's: p<0.0001) (Supplemental Figure 6A-B, E-F). Additionally, the -DCZ control was not significantly different than the CaMKII-mCherry + DCZ group, suggesting that DCZ alone is not impacting cFos levels in the absence of active receptor (Tukey's: p =0.6732)(Supplemental Figure 6A-B, E-F). This neuronal result is consistent with previous literature showing that hM3Dq activation of neurons increases neuronal activity at low doses of DCZ (Nagai et. al., 2021).
Additionally, GFAP-hM3Dq+DCZ, GFAP-mCherry+DCZ, and the -DCZ control group had significant differences in mean cFos levels (One-way ANOVA; F(2,6) = 5.907, p=0.0382). Specifically, post-hoc multiple comparisons revealed that GFAP-hM3Dq+DCZ trended towards a decrease in cFos levels in vCA1 of the HPC (Tukey's: p=0.0578)(Supplemental Figure 6C-E, G). Additionally, it is notable that there was no significant pairwise difference between the GFAP-hM3Dq+DCZ and -DCZ control group (Tukey's: p=0.9986)(Supplemental Figure 6C-E, G). But we observed a trend towards a difference in cFos levels across the GFAP-mCherry+DCZ and -DCZ control mice (Tukey's: p=0.0543)(Supplemental Figure 6C -E, G). This result appears to suggest that DCZ administration in the GFAP-mCherry+DCZ group is driving a small increase in cFos levels and our GFAP-hM3Dq+DCZ is not being activated by the ligand, maintaining similar levels to 'baseline' -DCZ cFos. However, we speculate that the I.P. injection alone may be driving this apparent increase in the GFAP-mCherry+DCZ group as this region of the brain processes stress and negative emotional valence, while our -DCZ mice did not receive an I.P. injection. If this were the case, performing the same 'baseline' experiment with the addition of saline I.P. instead of -DCZ alone would likely show similar levels of cFos to the GFAP-mCherry+DCZ group. Thus, we would still observe a modest decrease in cFos in the GFAP-hM3Dq+DCZ compared to GFAP-mCherry+DCZ control group in our chronic experiment. The small decrease in cFos that we observe in the astrocytic hM3Dq group may seem counterintuitive, but it has become clear in recent studies that modulation of astrocytes and neurons with chemogenetics results in differential effects on cFos expression and behavior. For instance, Gi-and Gq-activation in astrocytes stimulates the release of extracellular glutamate and increases neuronal excitability, suggesting that inhibitory signaling may be a unique property to neurons (Durkee et. al., 2019). Another study has shown that Gq-and Gi-pathways in astrocytes drive synaptic potentiation in hippocampal CA1 via Ca2+-dependent and Ca2+-independent mechanisms, respectively (Van Den Herrewegen et. al., 2021). This suggests that our astrocytic Gq manipulation strategy may be inducing unique cellular outcomes compared to neuronal Gq activation, and future studies can seek to tease out the differential mechanisms underlying each.
There are clear limitations to only measuring cFos with acute administration of DCZ, as receptor expression over these long time points may diminish. To provide evidence that we are maintaining receptor expression, we quantified fluorescence intensity of hM3Dq receptor expression in vHPC across the experimental groups. We speculate that cFos levels measured at the chronic 3, 6 and 9 month time point may show similar cFos levels across groups, despite any changes in regional activity, as the brain may attempt to restore homeostasis with increases in cellular activity over time. This would render any cFos levels at each time point potentially inconclusive read-outs of continual activation of the DREADD receptors. Future experiments will be needed to confirm the electrophysiological responses in the vHPC across chronic manipulations to confirm at the most basic level any decline in receptor functionality. Overall, we show that our acute administration of the DCZ ligand activated the Gq pathway successfully and induced cellular changes in the brain area of interest with acute administration. Additionally, we show the maintenance of Gq-mCherry expression across the 3, 6 and 9 month time points to better understand if the DREADD receptor remains present and potentially active across long time periods (Supplemental Figure 7). Measurements for CaMKII-hM3Dq mice across the three time points did not reveal any significant differences in mean gray value or corrected total cell fluorescence (CTCF) across time ([Mean gray: One-way ANOVA; F (2, 6) = 0.6852, p=0.5395][CTCF: One-way ANOVA; F (2, 6) = 0.6155, p=0.5713)(Supplemental Figure 7A-C). For the GFAP-hM3Dq mice, we did not observe any differences in mean gray or CTCF across any time point ([Mean gray: One-way ANOVA; F (2, 6) = 0.7896m p=0.4961][CTCF: One-way ANOVA; F (2, 6) = 0.7541, p=0.5103])(Supplemental Figure 7D-F).
Together, these results provide evidence that chronic administration of DCZ does not produce differences in the number of NeuN+ cells in CaMKII-hM3Dq and mCherry groups, but there is an effect of our manipulation across the GFAP-hM3Dq and mCherry groups. Additionally, acute administration of DCZ results in differential cellular changes in vCA1, with an increase in cFos in the CaMKII-hM3Dq group and a decrease in cFos in the GFAP-hM3Dq group compared to their mCherry controls. Finally, maintenance of receptor expression across all time points provides evidence that there may be successful activation of DREADD receptors even at long time periods after surgery, but this is not a deterministic measure of success within the confines of our experiment.
For extinction day 3, there was no interaction between time point and group, but there was an effect of time point on average freezing levels (Two-way ANOVA; Interaction: F (2, 60) = 2.147, p=0.1257; Timepoint: F (2, 60) = 7.099, p=0.0017; Group: F (1, 60) = 0.006334, p=0.9368) ( Figure 2E; iv). Within session at 3 months, there was no significant interaction between time bin and group (Two-way ANOVA RM; Interaction: F(29,580) = 0. When observing behavior across days, CaMKII-hM3Dq and mCherry groups had no interaction between day [fear conditioning, recall, extinction days 1-3] and group [CaMKII-hM3Dq and mCherry] at the 3, 6 and 9 month time points. However, at the 3 and 9 month time points, there were individual effects of both day and group, suggesting that our manipulation of the Gq pathway was impacting average freezing levels. Additionally, at the 6 month time point, there was an effect of day only, suggesting that our experimental manipulation was not having an impact at this time point ( In summary, our CaMKII-hM3Dq manipulation significantly decreased freezing levels during CFC compared to CaMKII-mCherry controls at the 9 month time point. During CFC, we also observed an increase in fear with aging in the 3 vs. 9 month CaMKII-mCherry groups. For recall, we observed a change in average freezing due to aging alone. For extinction day 1, we observed effects of aging across the 3 vs. 9 month and 3 vs. 6 month CaMKII-mCherry groups. Most notably, we observed a significant increase in average freezing during extinction in CaMKII-hM3Dq compared to control mice at the 3 month time point. For extinction day 2, there were individual impacts of aging and our manipulation for average freezing levels. Specifically, we observed an increase in freezing across 3 vs. 6 month CaMKII-mCherry mice. Finally, for extinction day 3 we only observed an impact of aging on freezing behavior. Our findings that Gq activation increases freezing at the 3 month time point during extinction may indicate that our intervention increases anxiety and/or impairs extinction mechanisms. Additionally, a decrease in freezing at the 9 month time point during CFC may indicate a decrease in anxiety or impairments in memory encoding. Overall, our neuronal-targeted Gq manipulation produces differential effects on contextual fear acquisition, recall and extinction behaviors, suggesting that normal functioning of excitatory neurons in vHPC is necessary for proper fear acquisition and maintenance.

3.3) Chronic Gq activation of GFAP+ astrocytes mildly impacts freezing behavior during contextual fear acquisition and maintenance.
Next, to test the hypothesis that chronic astrocytic Gq activation is sufficient to induce behavioral deficits across all timepoints, we subjected our mice to CFC, recall, and three days of extinction ( Figure 1E). For CFC, there was a significant interaction between group and time point for average freezing, suggesting that our manipulation and aging are interacting to impact memory acquisition (Two-way ANOVA; Interaction: F(2,56) = 3.24, p=0.0466; Timepoint: F (2, 56) = 0.6979, p=0.05019; Group: F (1, 56) = 0.8864, p=0.3505)( Figure 3A; iv). Post hoc multiple comparisons did not reveal significant pairwise differences between groups ( Figure 3A; iv). Within the session, mice at 3 months showed no interaction between time bin and group. On the other hand, each factor alone had an effect on freezing level during CFC (Two-way ANOVA RM; Interaction: F (5, 115) = 1.201, p=0.3133; Time: F (5, 115) = 135.1, p<0.0001; Group: F (1, 23) = 8.509, p=0.0078; Mouse: F (23, 115) = 2.421, p=0.0011)( Figure 3A; i). At 6 months, there was no interaction between time bin and group, but there was an effect of time bin alone that is expected as mice acquire fear successfully within session (Two-way ANOVA RM; Interaction: F ( We observed no significant interaction between time point and group for extinction day 1, only an effect of time point alone, suggesting that only aging contributed to the variance in freezing behavior (Two-way ANOVA; Interaction: F (2, 58) = 3.041, p=0.0555; Timepoint: F (2, 58) = 5.841, p=0.0049; Group: F (1, 58) = 1.637, p=0.2059)( Figure 3C; iv). Within session at 3 months, there was no interaction between time bin and group, however each factor contributed independently to freezing levels during extinction (Two-way ANOVA RM; Interaction: F (29, 690) = 0.6285, p=0.9370; Time: F (29, 690) = 2.529, p<0.0001; Group: F (1, 690) = 63.51, p<0.0001)( Figure 3C; i). At 6 and 9 months, there was no interaction between time bin and group, but our manipulation alone contributed significantly to freezing levels within session (Two-way ANOVA; For extinction day 3 there was an interaction between group and time point for average freezing levels, suggesting that age and our manipulation impacted behavior (Two-way ANOVA; Interaction: F (2, 58) = 4.106, p=0.0215; Timepoint: F (2, 58) = 0.8808, p=0.4199; Group: F (1, 58) = 2.624, p=0.1107)( Figure 3E, iv). Post hoc multiple comparisons did not reveal significant differences in means across groups or time points. At 3 months, there was no significant interaction between time bin and group within session, but time bin had an effect on the change in freezing level (Two-way ANOVA RM; Interaction: F (29, 667) = 1.078, p=0.3573; Time: F (6.706, 154.2) = 2.136, p=0.0455; Group: F (1, 23) = 1.622, p=0.2155; Mouse: F (23, 667) = 16.12, p<0.0001)( Figure 3E; i). However, at 6 months there was an interaction between group and time bin on freezing levels within session, suggesting that our manipulation had an effect in the GFAP-hM3Dq group ( When measuring across days, GFAP-hM3Dq and mCherry groups had no significant interaction between day and group at the 3 and 6 month time points. However, there was an effect of day alone, suggesting that our manipulation was not having any effect on learning (Mixed-effects model (REML) Figure 3H). Further post hoc analysis revealed significant differences in freezing between groups at the 9 month time point for the recall day (Sidak's: p=0.0142).
In summary, we observed significant individual impacts of our manipulation and aging on average freezing during CFC. Only the 3 month time point showed an impact of our manipulation on within-session freezing levels during CFC. For recall, there was a significant impact of both aging and our manipulation on average freezing levels. Only the 9 month time point showed an impact of our manipulation on within-session freezing levels during contextual recall. For extinction day 1, there was only an effect of aging on average freezing levels. All of the time points showed an impact of our manipulation on within-session freezing levels during extinction day 1. For extinction day 2, there was no impact of aging or our manipulation on average freezing levels. Only the 9 month time point demonstrated an effect of the group on freezing levels within-session for extinction day 2. Finally, for extinction day 3 there was an impact of both group and aging on average freezing levels. At 6 and 9 months withinsession, there was a difference in freezing levels that could be described by group manipulation.

3.4) Chronic Gq activation of neurons induces behavioral changes in locomotor and anxiety-related behaviors.
The open field and elevated zero maze are tasks that are classically used to measure locomotor activity and anxiety-like behaviors (Seibenhener & Wooten, 2015;Tucker & McCabe, 2017). To test our hypothesis that chronic activation of the Gq pathway in neurons will exhibit disrupted anxiety-like behaviors, we subjected the mice to these tasks on consecutive days ( Figure 1E). For open field, CaMKII-hM3Dq and mCherry mice had no significant interaction between time point and group in the total distance traveled, time spent in center, and number of entries to the center ( Figure 4A Figure 4B). Further, there was no significant interaction between time point and group for center mean visit time (Two-way ANOVA; Interaction: F (2, 57) = 1.168, p=0.3182; Timepoint: F (2, 57) = 1.972, p=0.1486; Group: F (1, 57) = 2.985, p=0.08950)( Figure 4E). This suggests that the combination of aging and the manipulation of the CaMKII-Gq pathway drives significant differences in anxiety-related behaviors.
In the elevated zero maze, there was no significant interaction between time point and group for total distance traveled, mean speed,  Figure 4J). Most interestingly, our manipulation of the CaMKII-Gq pathway induced a significant increase in the mean visit time to the open area of the zero maze at the 9 month time point when comparing CaMKII-hM3Dq and mCherry groups (Tukey's: p=0.0148)( Figure 4J). Finally, we observe normal aging changes in distance traveled, mean speed, body entries and head entries to the open area, indicating an increase in anxiety with age (Yanai & Endo, 2021). Thus overall, CaMKII Gq activation had a combinatory effect on anxiety-related behaviors with age.

3.5) Chronic Gq activation of astrocytes induces changes in locomotion and anxiety-related behaviors, as evidenced by the open field and zero maze tests.
We next tested the hypothesis that chronic modulation of the Gq pathway in astrocytes would induce changes in anxiety-related behaviors ( Figure 1E). For open field, GFAP-hM3Dq and mCherry mice showed no significant interaction between time point and group in the total distance traveled and mean speed.

3.6) Chronic Gq activation impacts social behaviors in CaMKII-and GFAP-hM3Dq groups.
To assess how social behaviors are impacted by our manipulation, we performed a social interaction test by placing the subject mouse in an open arena with a cup containing a male mouse or an empty cup for a 10 minute session. For CaMKII-hM3Dq and mCherry mice, there was no interaction between time point and group for total distance traveled, head time spent at the empty cup, or the number of head entries into the empty cup area. However, these metrics were all impacted by an effect of time point alone, suggesting that aging is contributing to these behavioral changes (  Figure 1D-F). Interestingly, for the number of head entries into the mouse cup (i.e. social target), there was a significant interaction between time point and group (Two-way ANOVA; Interaction: F (2, 59) = 6.145, p=0.0038; Timepoint: F (2, 59) = 35.11, p<0.0001; Group: F (1, 59) = 0.4773, p=0.4924)(Supplemental Figure 1A). Post hoc analysis revealed significant decreases in the number of head entries into the mouse cup between 3 vs. 9 month CaMKII-mCherry (Tukey's: p<0.0001), 6 vs. 9 month CaMKII-mCherry (Tukey's: p<0.0001), 3 vs. 9 month CaMKII-hM3Dq (Tukey's: p=0.0002), and 6 vs. 9 month CaMKII-hM3Dq (Tukey's: p=0.0176) groups (Supplemental Figure 1A). Most notably, there was a significant decrease in the number of head entries into the mouse cup at the 6 month time point between our CaMKII-hM3Dq and mCherry groups (Tukey's: p=0.0391)(Supplemental Figure 1A). Finally, for the total amount of head time and mean visit time to the mouse cup, as well as the mean visit time to the empty cup, there was no significant interaction between time point and group (Two-way Figure 1B-C, G).
Overall, we found a significant decrease in the number of entries into the mouse cup area within the 6 month time point, suggesting that these mice may have decreased interest in the social target with our manipulation. CaMKII-hM3Dq and mCherry mice displayed no significant differences in distance traveled, time spent at the empty cup and head entries into the empty cup zone, but aging did have an impact. Head entries into the mouse cup had a significant impact of aging and our manipulation, with a decrease in the number of entries across the 3 vs. 9 month and 6 vs. 9 month time points for both groups. These results are consistent with effects of aging, such as general decreased interest in social targets (Shoji et al, 2016;Oizumi et al, 2019;Yanai & Endo, 2021).
For GFAP-hM3Dq and mCherry mice, there was no significant interaction between time point and group for total distance traveled, mean speed, head time at the mouse cup (i.e. social target) or empty cup (i.e. control), head entries into the mouse cup or empty cup zones, or mean visit time to the mouse cup or empty cup during social interaction. However, we did observe significant effects of aging across all mouse cup metrics, but not in the empty cup metrics  Figure 2A-G). Overall, we observed effects of aging on measures of locomotion (i.e. distance traveled and mean speed) and engagement levels with the social target (i.e. mouse cup), but not in the control target (i.e. empty cup) metrics.

3.7) Novel environment exploration is affected by CaMKII-and GFAP-hM3Dq activation.
When exploring a novel environment, rodents generally investigate novel objects and spaces. To assess novel context exploration and quantify putative cognitive deficits, we performed an elevated y-maze test. Our assessment of novel environment exploration using the y-maze can also be used to assess short-term spatial working memory, as mice are required to remember the most recent arm they explored and navigate to a 'novel' arm instead. After introduction into the center of the y-maze, mice freely explore each of three arms (labeled A, B, C). The behavioral output of this task is measured by the percentage of spontaneous alternations and percentage of re-entries into the same arm. A spontaneous alternation is defined as three consecutive entrances into novel arms without repeats (i.e. ABC, CBA). A re-entry is defined as entering the same arm twice in a row (i.e. AA, BB, CC). For chronic activation of CaMKII-specific Gq pathway, there was a significant interaction between group and time point for the percentage of spontaneous alterations (Two-way ANOVA; Interaction: F (2, 60) = 6.655, p=0.0025; Timepoint: F (2, 60) = 12.65, p<0.0001; Group: F (1, 60) = 0.9491, p=0.3339)(Supplemental Figure 1H). Post hoc analyses revealed that there were significant pairwise differences between the 3 vs. 6 month CaMKII-hM3Dq (Tukey's: p<0.0001) and 3 vs. 9 month CaMKII-hM3Dq groups (Tukey's: p=0.0002)(Supplemental Figure  1H). This suggests that our manipulation of CaMKII-specific Gq pathways and aging are interacting to drive this specific effect in y-maze behavior. Additionally, there was no interaction between group and time point for the number of re-entries in the CaMKII-hM3Dq and control groups. However, there was an independent effect of time point in this behavioral metric, suggesting that aging alone is contributing to a change in y-maze behavior (Twoway ANOVA; Interaction: F (2, 60) = 1.893, p=0.1596; Timepoint: F (2, 60) = 3.878, p=0.0261; Group: F (1, 60) = 0.2529, p=0.6169)(Supplemental Figure 1I).
For Gq activation of GFAP+ astrocytes, we observed no significant interaction between group and time point for percentage of spontaneous alternations or number of re-entries. However, there were main effects of time point and group for re-entries, suggesting that aging and our manipulation are contributing to a behavioral change in the y-maze  Figure  2H-I). Overall, CaMKII-Gq and GFAP-Gq activation compounded with aging had differential effects on y-maze behaviors.

3.9) GFAP-hM3Dq activation impacted microglial morphology, but CaMKII-hM3Dq did not.
To further analyze the impact of our CaMKII or GFAP Gq pathway activation, we performed morphological analysis of microglia in vHPC. For CaMKII-hM3Dq and mCherry groups, there was no significant interaction between time point and group for microglial cell territory volume, cell volume, ramification index or average branch length (Two-way ANOVA; [Cell territory volume: Interaction: F (2, 12) = 0.  Figure 4K-N). Overall, Gq manipulation of CaMKII+ neurons did not have an effect on microglial morphology and only aging had an impact.

3.10) Gq activation of GFAP+ or CaMKII+ cells in vHPC did not significantly affect astrocytic morphology, only aging had a mild impact in the GFAP group.
To further analyze the impact of our CaMKII or GFAP Gq pathway activation on glial cells, we performed morphological analysis of astrocytes in vHPC. For CaMKII-hM3Dq and mCherry groups, there were no significant interactions between time point and group for astrocytic cell territory volume, total cell volume, ramification index, or average branch length (Two-way ANOVA; [Cell territory volume: Interaction: F (2, 12) = 0.1182, p=0.  Figure 6K-N). Interestingly, there was an effect of group for minimum branch length, suggesting that our manipulation of the neuronal Gq pathway was affecting this characteristic of astrocytes independent of time point for this metric. Overall, Gq activation in CaMKII neurons only affects astrocytic minimum branch length, and no other metrics related to morphological changes.
For GFAP-hM3Dq and mCherry groups, there were no significant interactions between time point and group for astrocytic cell territory volume, total cell volume, ramification index, or average branch length. However, there was an effect of time point alone for both ramification index and average branch length, suggesting an impact of normal aging, but no effect of our Gq manipulation (Two-way ANOVA; [Cell territory volume: Interaction: Our findings demonstrate that chronic activation of the Gq pathway in neurons and astrocytes differentially affects fear memory, social interaction, exploration, and anxiety-related behaviors, as well as markers of cellular stress. Together, our results add to burgeoning literature demonstrating that chemogenetic, optogenetic and pharmacological manipulation of neurons and astrocytes across brain regions induces behavioral enhancements or impairments (Nagai et. al., 2019;Adamsky et. al., 2018;Martin-Fernandez et. al., 2017;Xiao et. al., 2020;Lei et. al., 2022;Shelkar et. al., 2021;Padilla-Coreano et. al., 2017;Deisseroth et. al., 2014;Jimenez et. al., 2018;Jennings et. al., 2013;Stuber et. al., 2012). Although these studies begin to address the role of these cell types in a variety of behaviors, there is a gap in our understanding of what prolonged or "chronic" manipulations do to network functioning as it relates to behavioral output. Our study aimed to address both of these gaps with chronic manipulation of the Gq pathway within vCA1.
vHPC is known to process contextual fear memory via communication with the PFC and amygdala, sending information about contextual representations and emotional valence. In this study, we find that CaMKII-hM3Dq activation in the vHPC decreased fear acquisition at 9 months and increased fear during extinction at 3 months. Recent literature has shown that stimulation of the vHPC at different frequencies can differentially impact freezing levels. For example, 20Hz stimulation decreases freezing levels via inhibition of basal amygdala (BA) excitatory responses (Graham et. al., 2021). Our stimulation of neurons at the 9 month time point may be impacting freezing levels during acquisition in a similar way via disruption of this monosynaptic fear pathway between vHPC and the amygdala. In this same study, lower frequency stimulation (i.e. 4Hz) showed a trend towards increased freezing during extinction sessions, providing a potential mechanistic understanding of our findings at 3 months (Graham et. al., 2021). An alternative explanation for our findings is derived from studies showing that lesions of the vHPC induce contextual freezing impairments by enhancing locomotor activity. If we disrupt this circuit, we may be enhancing locomotor activity and thus, decreasing freezing during acquisition without any real effects on memory processing itself (Richmond et. al., 1999). Another paper shows that vHPC double-projecting neurons to the PFC and amygdala are preferentially recruited during contextual fear memory acquisition (Kim & Cho, 2017). Our manipulation could be disrupting this contextual information transfer to both brain regions during acquisition and disturbing the synchronization that is necessary for updating the memory across extinction. For astrocytes, we find that GFAP-hM3Dq activation impacted CFC, recall and extinction at the 6 and 9 month time points. This is in line with recent literature supporting the importance of astrocytes in the successful acquisition and maintenance of memories in the hippocampus. For example, optogenetic activation of hippocampal astrocytes with channelrhodopsin (ChR2) during fear memory consolidation decreases fear and related anxiety behaviors in mice. Additionally, photostimulation of these cells rescues the decrease in distance traveled and time spent in the center of the open field that was produced by fear conditioning . Another study has shown that chemogenetic Gq activation or optogenetic stimulation of astrocytes in the dorsal hippocampus before or during fear conditioning enhances memory acquisition and subsequent recall via NMDA dependent LTP in dCA1 via D-serine release (Adamsky et. al., 2018). Our chronic Gq manipulation in the ventral hippocampus may be uniquely impacting memory in a manner that is mechanistically different from the described acute, dorsal hippocampal manipulations. These findings broadly suggest that different patterns of stimulation with opto-or chemogenetics in the hippocampus may induce differential effects on astrocytic signaling with neighboring neurons via unique release of gliotransmitters or be dependent on the brain region of interest.
Further, we find that both CaMKII-hM3Dq activation and aging exert a combinatory effect on anxietyrelated behaviors across the open field and elevated zero maze. This is in line with previous work demonstrating that dentate gyrus (DG) stimulation of granule cells in the vHPC exerts anxiolytic effects, thus suppressing innate anxiety levels (Khierbeck et. al., 2013). Other work has shown that vCA1 is enriched with 'anxiety cells' that are activated by anxiogenic environments via projections to the lateral hypothalamus (LH) and are necessary for avoidance behaviors (Jimenez et. al., 2018). In addition, direct input from vHPC to the prefrontal cortex (PFC) is necessary for anxiety-related neural activity and inhibition of this projection decreases anxiety behaviors (Padilla-Coreano et. al., 2016;Ciocchi et. al., 2015). Together, this work allows us to understand how activation of neurons in this anxiety 'hub' through our manipulation could produce changes in anxiety. On the other hand, GFAP-hM3Dq activation decreased anxiety-related behaviors at the 3 month time point in the open field, but only showed impacts of aging in the zero maze. In light of recent literature, astrocytic activation with ChR2 increases calcium signaling intracellularly and reduces anxiety-like behaviors, thus suggesting that astrocytes play a role in overcoming or reducing anxiety levels (Cho et. al., 2022). Consistent with our results at the 3 month time point, optogenetic stimulation of astrocytes increases speed, distance and time spent in the center of the open field (Cho et. al., 2022). Another study showed that there is an interaction between anxiety and fear behaviors with astrocytic activation in the dorsal hippocampus, with activation decreasing anxiety-related behaviors in mice postfear conditioning . In the context of Khierbeck et. al., 2013 discussed above, we believe that astrocytic stimulation and release of adenosine triphosphate (ATP) increases excitatory synaptic transmission in the ventral DG granule cells, together contributing to a suppression of innate anxiety (Cho et. al., 2022).
In the present study, we observed a decrease in social engagement in the CaMKII-hM3Dq mice at the 6 month time point that received chronic activation. GFAP-hM3Dq mice displayed only effects of aging on locomotion and social engagement. The findings of our neuronal activation are consistent with previous literature showing that vCA1 pyramidal cells projecting to the PFC to regulate mouse social behaviors , as well as chemogenetic excitation of vCA1-PFC projecting cells impairing social memory (Philips et. al., 2019). Gq activation within the hippocampus may modulate neuronal outputs to these brain regions and impact social behaviors. Interestingly, novel environment exploration in the y-maze was impacted by a combination of chronic Gq activation and aging in both neuronal and astrocytic groups. Relatedly, recent findings suggest that stimulation of hippocampal astrocytes increases exploratory behaviors in anxiety-related tasks, making it difficult to fully disentangle a drive to explore a novel environment from innate anxiety levels (Cho et. al., 2022).
Glial cells, such as microglia and astrocytes exhibit unique changes morphologically, transcriptionally and functionally with aging, disease and cellular stress. In the dysfunctional human brain, glial cells become more vulnerable and may fail in their roles of neuronal protection, synaptic regulation and maintenance of homeostasis, gliotransmission and blood-brain barrier support. In the present study, we observe that CaMKII-hM3Dq mice have changes in microglial, but not astrocytic cell number in vHPC, while GFAP-hM3Dq mice showed no differences in glial cell number. In CaMKII-and GFAP-hM3Dq groups, we did not observe any changes in glial cell number with aging. This is consistent with literature showing that astrocytic staining does not show differences in the CA1 of aging mice (Grosche et. al., 2013). For microglia, studies have shown evidence that Iba1+ cell numbers in the hippocampus of old and young rats did not differ with aging (VanGuilder et. al., 2011). We speculate that changes in the number of microglia in vHPC from our neuronal manipulation may be due to efforts of the brain to reduce excitotoxicity induced by Gq activation over time.
In terms of morphological changes, CaMKII-hM3Dq activation did not have an impact, but aging affected the number of branch points and endpoints of microglia. GFAP-hM3Dq activation and aging combined to impact microglial cell volume, ramification index and minimum branch length. Studies have shown that older mice have a shortening and decrease in the complexity of microglial processes and branching structure (Hefendehl et. al., 2014). For astrocytes, CaMKII-hM3Dq activation minimally affected minimum branch length, and there was no effect of aging. Finally, GFAP-hM3Dq activation did not impact astrocytic characteristics, but aging had an effect on ramification index and the average branch lengths. Other work has provided evidence that astrocytes change morphology with aging (Ferrer et. al., 2017), modifying their GFAP+ surface area and volume in the hippocampus (Rodriguez et. al., 2014). A notable limitation of the quantification of glial cell number and morphology lies in the chosen cellular markers that were stained for each cell type. For example, GFAP expression does not allow for detection of fine astrocytic processes, providing conflicting results for the changes in area and complexity of these cells in the hippocampus with aging (Rodriguez et. al., 2014;Cerbai et. al., 2012).
Surprisingly, chronic manipulation of the Gq pathway in neurons and astrocytes may abolish some agedependent changes in task performance that we observe in our control groups. For example, we observe an agerelated increase in freezing behavior during CFC in our CaMKII-mCherry group that is abolished by our CaMKII-hM3Dq manipulation at the 9 month time point. Our manipulation may be working against these age-dependent changes by maintaining heightened levels of hippocampal activity, increased plasticity or activity-dependent increases in neurogenesis. Future research will be necessary to understand the underlying mechanisms of this agerelated rescue with Gq activation and this may inform therapeutic interventions in an aging population that may rely on a mild increase in network activity within the hippocampus and/or neighboring brain regions.
On this note, our study provides evidence that cell-type specific targeting induces differential effects on behavioral outcome, which may help inform subsequent treatments for disorders of the brain. For instance, while deep brain stimulation (DBS) and transcranial magnetic stimulation (TMS) are often effective in the treatment of psychiatric disorders, the cell-types directly affected through each perturbation and their outcomes remain ripe for exploration. Along similar lines, recent literature has suggested that clinical DBS-like, high-frequency stimulation of human astrocytes promotes changes in gene expression that are relevant to extracellular matrix formation, likely aiding in synapse development and induction of neuronal plasticity (Jang et. al., 2019). Another example of this therapeutic potential is demonstrated by entorhinal cortex (EC) deep brain stimulation in 6 month old mice having the capacity to rescue memory deficits in a mouse model of neurodegeneration (Xia et. al., 2017). Further understanding of how these techniques may be non-discriminately targeting all cells within a brain region vs. specific cell-types to exert their effects on network functioning and improvement in symptomatology would improve their efficacy. Our work suggests that targeting astrocytes or neurons within the vHPC may be an effective means to differentially modulate fear, anxiety, exploratory, and social behaviors.
Chronic manipulation of Gq pathways in neurons and astrocytes across time has inherent limitations. DREADD-mediated cell activation has not been as thoroughly investigated with chronic use, as it has been in acute administration studies. As such, it is possible that across 9 months, the efficacy of the ligand binding decreases over time. However, and promisingly, we observe persistence in the receptor expression, as indicated by the robust expression of hM3Dq-mCherry across 3, 6 and 9 month time points in both cell types. Indeed, even if the DREADD-mediated manipulation was only robust until the 3 month time point, we are still inducing pronounced acute insults that impact behavior and cellular markers across all time points. Another important limitation of this study is the use of only male mice across all time points. In humans, although there is an increase in risk for women to develop neurodegenerative disease, males are more likely to develop severe cognitive and behavioral phenotypes. Specifically, males demonstrate increased aggression, shorter lifespan, increased severity of cognitive decline and earlier onset of Parkinson's-associated dementia (Podcasy & Epperson, 2016). We chose to start with males and investigate the impact of both neuronal and astrocytic manipulation on behavior and cellular markers, but underscore the importance of work that will be needed to investigate the impact of chronic Gq pathway activation in females across cell types, together providing vital information on how male and female brains process network dysfunction differentially and offer insight into sex-specific therapeutic interventions for these devastating diseases and disorders that may result from this activity imbalance.
Finally, future research may investigate the in vivo physiological response using electrophysiology in vCA1 across time points as these cells are chronically activated. The impact of this would be two-fold: understanding how cellular activity changes of neurons in vCA1 change with manipulation across time, and confirming physiological response of these cells to the manipulation with a more concrete read-out than behavior. This would additionally allow for investigation of how cellular activity is related to the different behavioral changes (e.g. early changes in anxiety-related behaviors vs. later changes in fear memory). Overall, our data suggests that chronic manipulation of the Gq pathway in neurons and astrocytes across multiple time points impacts behavior. The results presented here provide valuable insights into the differential effects of chemogenetic manipulations of multiple cell types in a single brain region, as well as the specific effects of network disruption in the vHPC on cellular and behavioral phenotypes.

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

Figure 1. Chronic activation of Gq pathways in CaMKII+ neurons and GFAP+ astrocytes in vCA1.
(A) Viral strategy schematic for the neuronal groups. The AAV5-CaMKII-hM3D(Gq)-mCherry or control vector AAV5-CaMKII-mCherry was bilaterally injected into the vCA1 region of wild type mice. (B) Representative images of RFP/NeuN+ costaining (red and green, respectively) and DAPI+ cells (blue). (C) Viral strategy schematic for the astrocyte groups. The AAV9-GFAP-hM3D(Gq)-mCherry or control vector AAV5-GFAP-mCherry was bilaterally injected into the vCA1 region of wild type mice. (D) Representative images of RFP/GFAP+ co-staining (red and green, respectively) and DAPI+ cells (blue). (E) Schematic representation of chronically activating neuron or astrocyte Gq receptors through administration of the water-soluble DREADD ligand deschloroclozapine dihydrochloride (DCZ) for either 3, 6, or 9 months. Mice underwent a battery of behavioral tests at each end point (F-I) Histological assessment of NeuN+ cells to determine whether our manipulations were killing vCA1 neurons in the CaMKII groups (F and G) and/or GFAP groups (H and J). NeuN counts were assessed with a two-way analysis of variance (ANOVA) with time point and group as factors. Error bars indicate SEM. p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001, ns = not significant. Per group: n= 3 mice x 18 tiles (region of interest (ROI): vCA1) each were quantified for statistical analysis of NeuN counts. Scale bars indicate 50 micrometers.