Subtle modulation of ongoing calcium dynamics in astrocytic microdomains by sensory inputs

Abstract

Astrocytes communicate with neurons through their processes. In vitro experiments have demonstrated that astrocytic processes exhibit calcium activity both spontaneously and in response to external stimuli; however, it has not been fully determined whether and how astrocytic subcellular domains respond to sensory input in vivo. We visualized the calcium signals in astrocytes in the primary visual cortex of awake, head-fixed mice. Bias-free analyses of two-photon imaging data revealed that calcium activity prevailed in astrocytic subcellular domains, was coordinated with variable spot-like patterns, and was dominantly spontaneous. Indeed, visual stimuli did not affect the frequency of calcium domain activity, but it increased the domain size, whereas tetrodotoxin reduced the sizes of spontaneous calcium domains and abolished their visual responses. The "evoked" domain activity exhibited no apparent orientation tuning and was distributed unevenly within the cell, constituting multiple active hotspots that were often also recruited in spontaneous activity. The hotspots existed dominantly in the somata and endfeet of astrocytes. Thus, the patterns of astrocytic calcium dynamics are intrinsically constrained and are subject to minor but significant modulation by sensory input.

Keywords: Astrocyte; endfoot; map; orientation selectivity; visual cortex.

Figure 1

Subcellular calcium activity in astrocytes of awake mice. (A), Nissl-stained coronal section of an Mlc1-tTA::tetO-YC-Nano50 mouse. (B), Two-photon image of three YC-Nano50-expressing astrocytes in layer 2/3 of the primary visual cortex. (C) YFP (top) and CFP (bottom) fluorescence images of the astrocyte boxed in B. (D), Representative time-lapse images of the YFP/CFP ratio (R) of the astrocyte shown in C. (E), Schematic diagram for the detection of subcellular calcium domains. A contiguous aggregate of adjacent pixels with correlation coefficients (r) higher than 0.9 was defined as a “domain” (red).

Figure 2

Visual responses of calcium signals in single astrocytes. (A), Representative map of 31 activated domains in a single astrocyte (top). Different colors indicate individual domains. The raw traces of their fluorescence intensities are superimposed on the thick-lined average trace (bottom). The dashed line indicates the onset time of an upward-drifting grating stimulus. (B), Activated domains observed in a single astrocyte during nine sessions, each of which contained gratings with eight directions (arrows) and a gray screen (blank). Black areas indicate domains. (C), The average number of activated domains in a single cell per trial as a function of the sessions. Error bars indicate the SDs of 36 cells from nine videos. *P = 0.037, **P = 0.008, Scheffe’s test.

Figure 3

Visual stimulation induces larger calcium domains. (A), The mean number of calcium domains in a single cell per trial (left) and the mean size of individual domains (right) are compared between spontaneously emerging domains (Spont) and visually activated domains (Stim). Error bars are the SDs of 39 cells. Each gray line indicates a single astrocyte. **P = 0.009, paired t-test. (B), Percentage rank orders of the sizes of 5236 (Stimulation) and 720 (Spontaneous) domains. (C), Same as A, but for tetrodotoxin-treated visual cortex. n = 21 astrocytes.

Figure 4

Nonsignificant orientation tuning of astrocytic calcium signals at the whole-cell level. (A), Maps of activated domains stacked over six sessions for each direction in two representative astrocytes (left). Scale bar=20 μm. The right polar plots for these two astrocytes indicate the number of activated domains per trial (No.), the total area of activated domains per trial (Area), and the mean size of individual domains (Size) for gratings with eight directions. (B–D), The cumulative distributions of OSIs and DSIs for the number of activated domains (B), the total area of activated domains (C), and the mean size of individual activated domains (D) were compared between real datasets (red) and their 1000 trial-shuffled surrogates (blue). The D and P values were determined using the Kolmogorov–Smirnov test. n = 39 astrocytes.

Figure 5

Nonsignificant orientation tuning of astrocytic calcium signals at the calcium domain level. (A), Map of pixels that responded at least twice to a single direction in real datasets (left) and 1000 trial-shuffled surrogates (right). Colors indicate the preferred directions to which each pixel responded most frequently. (B), Mean numbers of responsive pixels in 37 real datasets and their pixel-shuffled surrogates. Gray lines indicate single astrocytes, and error bars represent the SD. (C), Mean ± SD of OSIs (left) and DSIs (right) of the responsive pixels in 23 real datasets and their surrogate data.

Figure 6

Hotspots for astrocytic calcium signals. (A), Heat maps of activated domains stacked across all 48 trials during six sessions in real data (left) and domain-shuffled surrogates (right). (B), The Z scores calculated from the response probabilities of real and surrogate datasets in A are shown for each pixel in a pseudocolored scale. (C), The cumulative distribution of the Z scores of individual pixels in each astrocyte (gray) is superimposed on their average (red), indicating that a few pixels exhibited extremely high-Z values. (D), The mean geometric energy of the spatial distribution of the Z score within individual cells is compared between 37 real datasets and their 1,000 domain-shuffled surrogates. Red dots represent astrocytes with significantly high-geometric energy, indicating that highly responsive pixels were spatially clustered.

Figure 7

Visual responses resemble spontaneous activity patterns. (A), Representative Z-score maps in activated domains (Stimulation) and spontaneous domains (Spontaneous). (B), Spatial correlations of the Z scores in all pixels between Stimulation and Spontaneous conditions. Red dots indicate astrocytes with significantly high correlation coefficients (Pearson product–moment correlation coefficient test). (C), Same as Fig.6D, but for spontaneous domains that occurred in tetrodotoxin-treated visual cortex. n = 19 astrocytes.

Figure 8

Subcellular localization of calcium hotspots. (A), Representative two-photon image of an astrocyte (YFP, cyan) and FITC-dextran-injected blood vessels (FITC, red) in the layer 2/3 of the primary visual cortex (left). The cell was divided into three regions, soma, process, and endfoot (right). (B–D), Pixels with Z scores in the top 1% (B), 5% (C), and 10% (D) in their entire distributions were defined as “hotpixels”. The total areas (left) and the ratios (right) of hotpixels to the total area in the three zones are plotted. Error bars are SEMs of 33 cells (soma), 22 cells (endfoot), and 38 cells (process) from 10 videos. *P < 0.05, **P < 0.01, Tukey’s test after one-way ANOVA.