Astrocyte-neuronal network interplay is disrupted in Alzheimer's disease mice

Alzheimer's disease (AD) is associated with senile plaques of beta-amyloid (A β ) that affect the function of neurons and astrocytes. Brain activity results from the coordinated function of neurons and astrocytes in astroglial-neuronal networks. However, the effects of A β on astroglial and neuronal network function remains unknown. Simultaneously monitoring astrocyte calcium and electric neuronal activities, we quantified the impact of A β on sensory-evoked cortical activity in a mouse model of AD. At rest, cortical astrocytes displayed spontaneous hyperactivity that was related to A β density. Sensory-evoked astrocyte responsiveness was diminished in AD mice, depending on the density and distance of A β , and the responses showed altered calcium dynamics. Hence, astrocytes were spontaneously hyperactive but hypo-responsive to sensory stimulation. Finally, AD mice showed sensory-evoked electrical cortical hyperresponsiveness associated with altered astrocyte-neuronal network interplay. Our findings suggest dysfunction of astrocyte networks in AD mice may dysregulate cortical electrical activity and contribute to cognitive decline.

We have addressed these issues by simultaneously recording in vivo the spontaneous and sensory-evoked astrocyte calcium activity and electrical neuronal network activity in the primary somatosensory cortex of the AD mouse model APPSwe/PS1dE9 (herein abbreviated APP/PS1) (Borchelt et al., 1997;Delekate et al., 2014;Kuchibhotla et al., 2009).We found that, in basal conditions, cortical astrocytes of APP/PS1 mice displayed calcium hyperactivity at both somas and arborizations.Furthermore, the calcium dynamics of astrocyte responses to sensory stimuli were diminished in APP/PS1 mice, and this hyporesponsiveness depended on the distance of Aβ deposits.Sensoryevoked electrocorticogram (ECoG) gamma activity (30-50 Hz) was enhanced in APP/PS1 mice.Moreover, the relation between neuronal gamma power and astrocytic calcium activities were also altered in APP/PS1 mice, indicating the functional impairment in astrocyteneuronal network interaction in these mice.Taken together, present results indicate that astrocyte calcium activity and astrocyte-neuronal network interactions are altered in the AD mouse model.These modifications are associated with alterations of the sensory-evoked electrical cortical activity, suggesting that dysfunction of astrocyte-neuronal networks contribute to the cognitive decline in Alzheimer's disease.

| Animal use and care
All the procedures for handling and sacrificing animals were approved by the University of Minnesota Institutional Animal Care and Use Committee (IACUC) in compliance with the National Institutes of Health guidelines.We used both female and male animals that were 6-9 months of age, kept on a continuous 12 h light/dark cycle and freely available to food and water.
Transgenic mice that expressed GCaMP7 under the GLT-1 promoter, highly expressed in astrocytes, were crossed with APPSwe/ PS1dE9 mice.Littermates that only expressed GCaMP7 and had no APP mutations were used as controls.

| Calcium imaging
A Leica SP5 upright multiphoton microscope was used to monitor cortical astrocyte structure in red from SR101 staining, astrocyte calcium in green via GCaMP7, and beta-amyloid deposits in blue from methoxy-X04.Images of 512 Â 512 pixels were taken using a 25Â water-dipping objective with a 1.7 digital zoom to monitor 365 Â 365 μm square planes of primary somatosensory cortex in layers 2/3 (100-300 μm below the cortical surface) at 5 Hz sampling rate.

| Electrophysiology recordings
ECoG was recorded from a 0.25 mm tungsten wire atop the cortex, under the glass coverslip, with a reference wire soldered to a screw fastened into the occipital plate.Leads were fed into an A-M Systems model 3000 ACDC differential amplifier, sampled at 10 kHz and band pass filtered between 1 and 3000 Hz, and digitized using an Axon Digidata 1550 A acquisition system connected to a PC.Electrophysiology was monitored using AxoScope software (pClamp; AxoScope version 10.5.0.9).

| Hind paw stimulation
Bipolar electrodes were placed in the left hind paw (contralateral to recordings done in the right primary somatosensory cortex).Square electrical pulses of 0.5 ms duration were delivered at 2 mA and 2 Hz for 20 s.Before stimulation, the preparation was recorded for 30 minutes, and in between each stimulation session a 2-3 min interstimulus was done.

| Calcium imaging processing and analysis
Images were processed and analyzed on PC using custom made MATLAB software (https://www.araquelab.com/code/).Images of simultaneously obtained SR101-stained astrocyte territories were manually outlined to define individual astrocytes in the imaging plane.Next, using Calsee and ClickCell (Lines et al., 2020), individual astrocytes were semi-automatically subdivided into soma and surrounding arborization subcellular regions of interest (ROIs).Traces of astrocyte calcium activity from ROI of somas and arborizations were found by averaging GCaMP7-measured activity from ROIs.Traces were analyzed for spontaneous activity using an event detection algorithm that detected an event if calcium levels went above 2 standard deviations of the average of the entire recording.To assess astrocyte network synchrony, we quantified simultaneous population activity over a 1 s sliding window and averaged the frames with simultaneous activity.For stimulation experiments, onset event detection was determined if calcium amplitude went above 2 standard deviations of the first 5 s of the recordings baseline average.Rise time of calcium responses were determined by how much time passed between the event onset and peak.As decay back to baseline followed an exponential decay, decay tau of astrocyte responses was determined by fitting an exponential to the decay of the curve following the peak.Calsee was also used to outline Aβ plaques to determine the location of astrocyte ROI relative to the edge of the closest plaque.To quantify density measures of Aβ load, the total area of outlined plaques was divided by the total area of the imaging plane.For measures of the percentage of population responding in relation to Aβ plaque distance, astrocytes were grouped together in concentric ringed boundaries around Aβ plaques in 25 μm increments.

| Electrophysiology processing and analysis
Electrocorticogram (ECoG) recordings were processed using custom MATLAB scripts (https://www.araquelab.com/code/)where raw recordings were filtered using a low-pass FIR filter below 100 Hz.Frequency content was extracted from a spectrogram created using a hamming window of 4096 frames in length, converted into decibels and smoothed using Welch's method.Averaging the power across the spectrogram of specific frequency ranges were done to quantify activity in frequency bands for delta (0-4 Hz), theta (4-8 Hz), alpha (8-13 Hz), beta (13-30 Hz), and gamma (30-50 Hz).

| Analysis of astrocyte-neuronal network interaction
The averaged calcium trace from astrocytes in a single imaging plane was plotted versus the normalized gamma power from simultaneously recorded ECoG.Plotting astrocyte population activity versus neuronal network gamma power produces a hysteresis loop described previously (Lines et al., 2020).Each point of the hysteresis loop represents the state of both networks at that moment.Plotting the two network activities in this way highlights the dialogue between these two networks during sensory stimulation.The hysteresis loop created from each individual imaging plane was analyzed to quantify the total area of the loop as well as the angle of the center of mass.

| Statistics
Cellular data from individual imaging planes were grouped together and averaged for use as a single data point.Comparisons between calcium imaging results from GCaMP7 versus APP/PS1xGCaMP7 were done using a two-sample t-test.To investigate the association between astrocyte calcium activity and beta-amyloid density, spontaneous and sensory stimulated responses were determined for each imaging plane and correlated with the density of beta-amyloid in that imaging plane using a t-test of Pearson's correlation.To assess the relationship between plaque distance and astrocyte calcium activity, each imaging plane was subdivided into equidistant rings of area surrounding Aβ plaques and the astrocytes within these rings were grouped together and averaged to be used as a single data point for each imaging plane within that distance from a beta-amyloid plaque and compared with a 1-way ANOVA.Astrocyte pathophysiology and GFAP reactivity levels from distances to Aβ plaques were compared using Spearman correlation.Electrophysiological power analysis was done by comparing spontaneous or sensory evoked levels using two-sampled t-tests between APP/PS1xGCaMP7 and GCaMP7 animals.Statistical tests between hysteresis loop profiles were performed on the total area as well as the center of mass quantities from individual imaging planes and using twosampled t-tests between APP/PS1xGCaMP7 and GCaMP7 animals.

| Code availability
The code used in this study can be found at https://www.araquelab.com/code/ 3 | RESULTS

| Spontaneous and sensory-evoked astrocyte calcium in APP/PS1 mice in vivo
We simultaneously recorded in vivo the spontaneous and sensoryevoked astrocyte calcium activity and electrical neuronal network activity in the primary somatosensory cortex of the AD mouse model  (Borchelt et al., 1997;Delekate et al., 2014;Kuchibhotla et al., 2009).We crossed APP/PS1 mice with transgenic mice expressing the genetically-encoded calcium indicator GCaMP7 under the promoter of the glutamate transporter GLT-1 (Monai et al., 2016) that is highly expressed in cortical astrocytes (Figure S1).Using twophoton microscopy in vivo, we imaged the primary somatosensory cortex of aged mice (6-9 months old), simultaneously monitoring Aβ deposits with the fluorescent blue dye methoxy-X04 (see Section 2) (Klunk et al., 2002), astrocyte morphology with the selective red fluorescent dye sulforhodamine 101 (SR101; see Section 2) (Nimmerjahn et al., 2004) and astrocyte calcium activity with GCaMP7 (Figure 1 c)).While no differences were found in frequency content in theta, alpha or beta ranges, significant alterations were observed in delta and gamma ranges.We observed an increase in cortical delta (0-4 Hz) in APP/PS1 mice compared to control animals (p < .05;n = 6 and 5 mice respectively; Figure 4(d)).In addition, frequency content in the gamma range (30-50 Hz) had a heightened response in APP/PS1 compared to wildtype animals (p < .05;n = 6 and 5 mice, respectively, mice; Figure 4(e)).These results indicate that cortical neuronal network activity responses to sensory stimulation were altered in APP/PS1 mice.Of particular relevance is the alteration of the activity in the gamma range, which is known to play a key role in sensory information processing (Engel & Singer, 2001).
We have previously shown that cortical astrocytes regulate sensory-evoked neuronal network activity (Lines et al., 2020), where the initial cortical gamma activity response to sensory stimulation is followed by a rise in astrocyte calcium that prompts the temporal decline in cortical gamma activity, which is indicative of a regulatory feedback loop that contributes to the cortical sensory adaption (Ang & McMillen, 2013;Wark et al., 2007).Additionally, we have demonstrated cortical astrocyte to neuron signaling dysfunction in APP/PS1 mice in situ (Gomez-Gonzalo et al., 2017).Both control and APP/PS1 mice showed a similar temporal pattern of astrocyte calcium and gamma activity (Figure 4(f)-(g)), but with remarkable differences illustrated by the hysteresis graphs obtained by plotting the sensoryevoked gamma activity vs. the astrocyte calcium, which indicates an alteration of the regulatory feedback loop (Lines et al., 2020) in APP/PS1 mice (Figure 4(h)).To further characterize, the hysteresis graphs, we further analyzed the areas within the hysteresis curves and their center of mass.Since we delivered the same sensory stimuli (2 mA, 2 Hz, 20 s) to both groups of animals (i.e., we provided the same amount of energy to the system), we expected no changes in the areas.Indeed, we found no significant changes in the areas within the hysteresis curves of wildtype and APP/PS1 mice (p = .40;n = 23 and 28 imaging planes, from 6 and 5 mice, respectively).In contrast, the center of mass in APP/PS1 mice was displaced with respect to wildtype mice (angle from origin: 35.8 ± 0.6 in wildtype vs. 38.3± 0.8 in APP/PS1; p < .05),which is consistent with the reduced sensory-evoked astrocyte calcium activity and more "unregulated" sensory-evoked gamma activity in APP/PS1 mice.These results indicate that the astrocyteneuronal network interaction is disturbed in APP/PS1 mice.

| DISCUSSION
Present results demonstrate that spontaneous and sensory-evoked astrocyte function is altered in an amyloidosis AD mouse model.Astrocyte pathophysiology consisted in spontaneous hyperexcitability of astrocyte processes and hypo-responsiveness to sensory inputs, and it was associated with the density and the distance of Aβ plaques.Furthermore, alterations of sensory-evoked cortical gamma activity, which is critical for proper cortical sensory information processing (Engel & Singer, 2001), are associated with alterations of the negative feedback loop provided by astrocytes (Lines et al., 2020), which is evidenced by changes in the hysteresis loop of astrocyte and neuron network activities and which is indicative of astrocyte-neuronal cortical network dysfunctions.Since astrocyte physiology is altered in AD mice and may contribute to cortical neuronal network dysregulation through the dysfunction of the astrocyte-neuronal network interactions, astrocytes may play important roles in cognitive deficits associated with AD and may be potential targets for the treatment of the disease.
By stimulating the hindpaw, we found that sensory-evoked astrocyte calcium response onset and dynamics were sluggish in APP/PS1 mice compared to wildtype.When we related astrocyte network responsiveness to the distance to beta-amyloid we found a decrease in network activity in adjacent astrocytes, in both somas and arborizations.Similar to previous findings (Wyss-Coray & Mucke, 2002), we confirmed that Aβ aggregation produced local reactive astrogliosis that may create a spatial distortion on astrocyte network activity.To our knowledge, this is the first example of evoked responses from reactive astrocytes in vivo.Since astrocytes show hyperactivity related to beta-amyloid density, it is intuitive to hypothesize that evoked astrocyte network activity should also be increased related to Aβ.However, reactive astrocytes have been found to have impairments in G-protein-mediated calcium responses (Chow et al., 2009;Hamby et al., 2012).Additionally, it has been recently demonstrated that local resting astrocyte calcium controls the scale of stimulated astrocyte calcium responses (King et al., 2020).From this, our results suggest that increased spontaneous activity, in agreement with others (Delekate et al., 2014;Kuchibhotla et al., 2009), lead to diminished sensory-evoked responses.
We found that APP/PS1 mice have increased sensory-evoked gamma activity compared to wildtype mice.High frequency gamma activity (30-50 Hz) has been related to cortical sensory information processing (Engel & Singer, 2001) and increases in sensory-evoked gamma power is a pathophysiology in human AD patients and amyloidosis mouse models (van Deursen et al., 2008;van Deursen et al., 2011;Wesson et al., 2011).Present results suggest a relation between Aβ load and gamma-activity.Indeed, Figure 4(e) shows higher sensory-evoked gamma activity in APP/PS1 mice compared to wildtype controls.This idea is consistent with the report of impairments in sensory-evoked gamma power at a time when Aβ deposition occurs and its rescue following beta-amyloid degradation in the TG2576 amyloidosis mouse model (Wesson et al., 2011).Moreover, the astrocyte calcium signal has been found to control the progression of Aβ plaque deposition in APP/PS1 mice (Gomez-Gonzalo et al., 2017), which may contribute to network activity alterations in these mice.A specific range of electrical activity is necessary for proper network functioning (Poil et al., 2012).Too little network activity and information processing may not occur (Singer, 1993).However, too much activity, and networks are unable to distinguish the subtleties in evoked cortical patterns (Studer et al., 2019;Ung et al., 2020).Perceptual difficulties can occur in patients with seizure activity (Grant, 2005), which many AD patients suffer from (Vossel et al., 2013).Taken together, it may be that a major contributor to neuronal network pathophysiology in AD is a loss of astrocytemediated regulation.
We revealed an alteration in astrocyte-neuronal network interplay in the APP/PS1 mouse model through the simultaneous recording of ECoG with two-photon imaging of astrocyte calcium.Calcium signal detected by direct imaging is probably more local than the electrophysiological signal.However, the potential spatial differences may not be huge.Indeed, the astrocyte calcium network activity were monitored in a field of view of 356 Â 356 μm, which is probably a smaller area than the source of the electrophysiological signal but not excessively smaller.
Certainly, we cannot rule out certain disparities between both readouts, and further studies, out of the scope of the present work, should be performed to establish the actual spatial relationship between both signals.
Nevertheless, we believe that the present approach has important value as a first approximation.We have previously shown in situ that astrocyte to neuronal signaling is disrupted in APP/PS1 mice (Gomez-Gonzalo et al., 2017).Additionally, from our previous study in vivo (Lines et al., 2020), we discovered that sensory-evoked cortical gamma power correlates with subsequent astrocyte network responses, which regulate neuronal network activity by reducing heightened gamma power.Present results showing that the relatively reduced sensoryevoked astrocyte calcium signal is associated with an increase of the sensory-evoked gamma activity in APP/PS1 mice is consistent with findings reported in that work, that is, that sensory-evoked astrocyte calcium activity dampened the gamma-activity and that IP 3 R2 À/À transgenic mice with impaired astrocyte calcium signaling had heightened sensory-evoked gamma power.
Astrocyte-neuronal network interaction is visualized in a hysteresis dynamic that demonstrates the presence of a regulatory negative feedback loop.When comparing hysteresis patterns between wildtype and APP/PS1 mice, we observed an impairment in this hysteresis loop.Hysteresis graphs in wildtype and APP/PS1 mice display several differences.Two striking features are conspicuous in Figure 4 (h): (1) the general profiles of the curves, which reflect astrocyteneuron interaction, were altered in APP/PS1 mice compared to wildtype animals; and (2) the prominent gamma activity (the shoulder marked with a red arrow) in APP/PS1 mice compared with the reduction of gamma activity in wildtype mice, indicating the lack of astrocyte-induced gamma activity dampening.The alteration in astrocyte-neuronal network activity fits with the regulatory dynamics described in our earlier study (Lines et al., 2020), where reduced astrocyte activity caused increases in sensory-evoked gamma power, and is similar to those findings where we used a transgenic mouse with ablated astrocyte calcium.Our previous work shows cortical astrocytes regulate sensory-evoked neuronal network activity (Lines et al., 2020), therefore astrocyte pathophysiology related to betaamyloid may contribute to the dysregulation of cortical sensory information processing (Halassa et al., 2007).
In this study we uncovered cortical astrocyte pathophysiology and altered astrocyte-neuronal network interplay.Astrocytes should be considered as a target in future designs of therapeutics for Alzheimer's disease, and may already be indirectly targeted with current treatments.Current therapies for Alzheimer's disease patients are anti-epileptic medications (Cumbo & Ligori, 2010), thereby increasing cortical inhibition.Astrocytes regulate neuronal network activity by reducing cortical activity (Lines et al., 2020;Poskanzer & Yuste, 2016), and perhaps increasing inhibitory signaling as a treatment for AD mimics normal astrocyte regulation.As we further discover important neuromodulatory roles of astrocytes in the central nervous system, we will continue to uncover the astrocytic potential as therapeutic targets in Alzheimer's disease progression.
were extracted and placed in paraformaldehyde overnight and then placed in sucrose.Fixed brains were cut to 40 μm slices on a Thermo Scientific™ HM 450 Sliding Microtome.Next, slices were incubated in Reveal Decloaker (Biocare Medical #RV1000MMRTU) for 30 min in a 78 C water bath and treatment in 88% Formic acid for 10 min.Slices were then stained for glial fibrillary acidic protein (GFAP) (rabbit anti-GFAP; DakoCytomation; 1:4000) and 4G8 (mouse anti-Aβ; Biolegend; 1:500) to image astrocytes and Aβ respectively.Imaging and analysis of immunohistochemistry performed similarly as in physiology experiments.To evaluate astrocyte reactivity, GFAP immunohistochemistry fluorescence brightness was averaged within individual astrocyte somas.The distance between the center of mass of individual somas and the edge of Aβ deposits was associated with each somas GFAP immunohistochemistry fluorescence brightness.

F
I G U R E 1 The APP/PS1 mouse model of Alzheimer's disease has altered spontaneous and sensory-evoked astrocyte Ca 2+ .(a) Scheme to monitor astrocyte calcium activity and fibrillar Aβ in the somatosensory cortex in vivo.(b) GCaMP7 activity in layers 2/3 of the somatosensory cortex.(c) SR101 astrocyte staining (left), methoxy-X04 staining of fibrillar Aβ deposits (center), and a merge (right).Scale bar = 50 μm.(d) Semi-automatically defined ROIs of somas (green) and arborizations (magenta).(e) Pseudocolor Ca 2+ images of basal (left) and during stimulation (right).(f) Astrocyte Ca 2+ traces from arborizations (magenta) and somas (green) during sensory stimulation in e. (g) Raster plot of Ca 2+ responses from ROIs in e during 2 mA, 2 Hz, 20 s stimulation starting at time = 0 s.(h) Raster plots of spontaneous soma (green) and arborization (magenta) activity from GCaMP7 (wildtype) and APP/PS1 x GCaMP7 (APP/PS1) mice.(i) Wildtype (maroon) and APP/PS1 (blue) spontaneous calcium event frequency in arborizations and somas.(j) Average Ca 2+ trace from astrocyte arborizations (left) and somas (right) in wildtype (maroon) and APP/PS1 (blue) mice.(k) Onset of Ca 2+ response following stimulation in arborizations and somas in wildtype (maroon) and APP/PS1 (blue).(l) Rise time of Ca 2+ events in response to sensory stimulation.(m) Exponential tau of decay from Ca 2+ events following stimulation.Mean ± SEM. "***" = p < .001,"**" = p < .01.Two-sample t-test APP/PS1 (a)-(c)), We monitored astrocyte calcium activity in somas and arborizations in resting conditions and in response to electrical stimulation of the hind paw (Figure 1(d) and (e)).Stimulation of the hindpaw increased astrocyte calcium levels in somas and arborizations (Figure 1(f) and (g)) (cf.Ref (Lines et al., 2020)) showing the ability of cortical astrocytes to respond to sensory stimulation.Then, we investigated whether the astrocyte calcium signal was altered in APP/PS1 mice by comparing spontaneous and sensoryevoked astrocyte calcium activity in APP/PS1 (APP/PS1 x GLT-1 GCaMP7) and wildtype control (GLT-1 GCaMP7) mice.In resting basal conditions, the frequency of spontaneous calcium events of somas and arborizations in APP/PS1 mice was higher than in wildtype mice (Figure 1(h) and (i)) (p < .01 in arborizations and p < .001 in somas; n = 42 imaging planes each; 7 mice each).There was no difference in the synchrony of astrocyte somas or arborizations between wildtype and APP/PS1 mice (p = .38 in arborizations and p = .11in somas).Upon somatosensory stimulation (2 mA stimuli at 2 Hz for 20 s), cortical astrocytes in both APP/PS1 and control animals responded with calcium elevations, but with significantly different dynamics (Figure 1 (j)-(m)) (onset and risetime in arborizations p < .001and in somas p < .01;n = 52 imaging planes each; 7 mice each).Indeed, responses in APP/PS1 displayed slower rising and decay phases (Figure 1(k)-(m)) (in arborizations p < .001and in somas p < .01;n = 52 imaging planes each; 7 mice each).These results indicate that astrocytes in APP/PS1 mice are spontaneously hyperexcitable and that their responses to sensory stimulation display slower kinetics.

3. 2 |
Aβ deposits spatiotemporally alter astrocyte physiologyWe next investigated the relation of the astrocyte pathophysiology with the Aβ deposits.We quantified the spontaneous and sensory-F I G U R E 2 Astrocytenetwork pathophysiology is related to Aβ density in APP/PS1 mice.(a) Methoxy-X04 staining of beta-amyloid (left) and merged with SR101 staining of astrocytes (right).Scale bar = 50 μm.(b) Pseudocolor calcium images of basal (left) and stimulation (right).(c) Traces from calcium activity in somas (green) and arborizations (magenta) in b. (d-f) as a-c but for high Aβ density.(g) Spontaneous arbor calcium events (left) and soma calcium events (right) as a function of Aβ density in the imaging plane.(h) Percent of arborizations active (left) and the percent of somas active (right) in response to sensory stimulation related to Aβ density.Student's t-test of Pearson's correlation evoked astrocyte activity in relation to the density of Aβ deposits (i.e., their area relative to the field of view) in individual imaging planes where astrocyte calcium was recorded (Figure 2(a)-(f)).Regarding spontaneous calcium activity, astrocyte arborizations showed a significant increase in calcium event frequency with increased Aβ density, following a distribution of values that could be fitted to a linear regression (Figure 2(g)) (Student's t-test of a Pearson's correlation: p < .05;R 2 = 0.14; n = 42 imaging planes, 7 animals), whereas astrocyte activity in somas was unrelated to Aβ load (Figure 2(g)) (Student's t-test of a Pearson's correlation: p = .44;n = 42 imaging planes, 7 animals), suggesting that subcellular astrocyte calcium alterations, that is, spontaneous hyperactivity of astrocytic processes, are related to the presence of Aβ.We then evaluated whether the sensory-evoked astrocyte network responsiveness was altered in relation with Aβ density.The proportion of responding arborizations and somas decreased as the betaamyloid density increased, according to distributions of values that could be fitted to linear regressions with negative slopes (Student's t-test of a Pearson's correlation: p < .05 and .01,for arborizations and somas, respectively; R 2 = 0.08 and 0.15; n = 52 imaging planes, 7 animals; Figure 2(h)).Taken together, these results suggest that the presence of Aβ deposits is associated with a spontaneous hyperexcitability of astrocyte processes and a hypo-responsiveness to sensory inputs.To further test this idea, we investigated whether the astrocyte calcium signal depended on the distance to Aβ deposits.We quantified the spontaneous and sensory-evoked astrocyte calcium activity in arborizations and somas, and established correlations with the distance to the Aβ plaques (Figure 3(a)-(d)).While no statistical differences were found between the spontaneous calcium activity and the distance to Aβ deposits (Figure 3(e)) (p = .11in arborizations and p = .44 in somas; n = 42 imaging planes, 7 animals), the sensory-evoked responsiveness was significantly different relative to the location of Aβ deposits.Indeed, the proportion of responding arborizations and somas increased as the distance to the Aβ plaque increased (Figure 3(f)) (p < .05 in arborizations and p < .05 in somas; n = 52 imaging planes, 7 animals).Indeed, F I G U R E 3 Astrocyte network pathophysiology in APP/PS1 mice is related to the distance to Aβ deposits.(a) Structural labeling of astrocytes with SR101 (red) and Aβ plaques with methoxy-X04 (blue).Scale bar = 50 μm.(b) ROI defining Aβ plaques (blue), astrocyte somas (green) and arborizations (magenta).(c) Pseudocolor Ca 2+ images of basal (left) and stimulated (right).(d) Astrocyte Ca 2+ traces from near (<100 μm; top) and far (>100 μm; bottom) away from Aβ pathology.(e) Spontaneous Ca 2+ heatmaps and events in astrocyte arborizations (left) and somas (right) related to distance to Aβ deposits.(f) Heatmaps and percentage of arbors (left) and somas (right) active in response to stimulation related to Aβ plaque distance.(g) Immunohistochemical labeling of GFAP (left), Aβ (center) and merge (right) in APP/PS1 mouse.Scale bar = 50 μm.(h) Astrocyte GFAP-labeled reactivity related to distance to Aβ plaque.(i) Spearman correlation between GFAP reactivity vs. percent active arbors (magenta) and somas (green).Mean ± SEM. "***" = p < .001,"**" = p < .01,"*" = p < .05. 1-way ANOVA and t-test of spearman correlation astrogliosis was confirmed to occur surrounding Aβ plaques (Wyss-Coray & Mucke, 2002) (1-way ANOVA; p < .001;n = 8 imaging planes; 2 animals; Figure 3(g)-(h)).Notably, the alterations of sensory-evoked astrocyte responsiveness (Figure 3(f)) correlated with the levels of astrocyte immunoreactivity (Figure 3(h); 1-way ANOVA: p < .001;n = 8 images, 2 animals) (Figure 3(i); Spearman correlation: p < .05 and .01,for arborizations and somas, respectively; n = 9 distances).These results indicate that sensory-evoked astrocyte pathophysiology is affected by the distance of Aβ deposits.