A spatial threshold for astrocyte calcium surge Authors: Justin Lines, Andres Baraibar, Carmen Nanclares, Eduardo D Martin, Juan Aguilar, Paulo Kofuji, Marta Navarrete, Alfonso Araque, Klaus-Armin Nave, Justin Lines, Andres Baraibar, Carmen Nanclares, Eduardo Martin, Juan Aguilar, Paulo Kofuji, Marta Navarrete, Alfonso Araque Published: 2024 ABSTRACT Astrocytes are active cells involved in brain function through the bidirectional communication with neurons, in which astrocyte calcium plays a crucial role. Synaptically evoked calcium increases can be localized to independent subcellular domains or expand to the entire cell, i.e., calcium surge. Because a single astrocyte may contact ~100,000 synapses, the control of the intracellular calcium signal propagation may have relevant consequences on brain function. Yet, the properties governing the spatial dynamics of astrocyte calcium remains poorly defined. Imaging subcellular responses of cortical astrocytes to sensory stimulation in mice, we show that sensory-evoked astrocyte calcium responses originated and remained localized in domains of the astrocytic arborization, but eventually propagated to the entire cell if a spatial threshold of >23% of the arborization being activated was surpassed. Using Itpr2-/- mice, we found that type-2 IP3 receptors were necessary for the generation of astrocyte calcium surge. We finally show using in situ electrophysiological recordings that the spatial threshold of the astrocyte calcium signal consequently determined the gliotransmitter release. Present results reveal a fundamental property of astrocyte physiology, i.e., a spatial threshold for astrocyte calcium propagation, which depends on astrocyte intrinsic properties and governs astrocyte integration of local synaptic activity and subsequent neuromodulation. Introduction Accumulating evidence indicates that astrocytes play active roles in synaptic function and neural information processing by exchanging signals with neurons. They respond to synaptic activity with intracellular calcium elevations, which stimulate the release of gliotransmitters that regulate neuronal and synaptic function (Araque et al., 2014; Perea et al., 2009). Hence, the astrocyte calcium signal is a crucial signaling event in the bidirectional communication between neurons and astrocytes. Moreover, astrocyte calcium manipulations have been shown to regulate neuronal network function (Ahmadpour et al., 2024; Lee et al., 2014; Lines et al., 2020; Mederos et al., 2021; Miguel-Quesada et al., 2023; Perea et al., 2016; Poskanzer and Yuste, 2016) and animal behavior (Adamsky et al., 2018; Corkrum et al., 2020; Kofuji and Araque, 2021; Martin-Fernandez et al., 2017; Monai et al., 2016; Nagai et al., 2021; Oliveira et al., 2015), and disruption of astrocyte calcium has been proposed to contribute to brain diseases (Delekate et al., 2014; Jiang et al., 2016; Kuchibhotla et al., 2009; Lines et al., 2022; Nanclares et al., 2023; Tian et al., 2005; Yu et al., 2018). Astrocyte calcium variations represent a complex signal that exists over a wide range of spatial and temporal scales (Ahrens et al., 2024; Bazargani and Attwell, 2016; Rusakov, 2015; Semyanov et al., 2020; Shigetomi et al., 2016; Volterra et al., 2014). Spatially, the astrocyte calcium signal may occur at discrete subcellular regions, termed domains, in the astrocyte arborization, or may encompass large portions of the cell or even the entire astrocyte (Bazargani and Attwell, 2016; Rusakov, 2015; Semyanov et al., 2020; Shigetomi et al., 2016; Volterra et al., 2014). Subcellular astrocyte calcium events have been recently proposed to be involved in the representation of spatiotemporal maps (Curreli et al., 2022; Doron et al., 2022; Serra et al., 2022) and to contribute to long-term information storage (Curreli et al., 2022; Doron et al., 2022; Georgiou et al., 2022; Vignoli et al., 2021). Calcium activity in astrocytes is believed to originate in domains within astrocytic processes that contact and bidirectly communicate with nearby synapses termed microdomains (Arizono et al., 2020; Bindocci et al., 2017; Di Castro et al., 2011; Grosche et al., 1999; Khakh and Sofroniew, 2015; Panatier et al., 2011). While calcium transients in discrete domains have been found to be independent events within astrocyte arborizations (Chen et al., 2020; Stobart et al., 2018; Ung et al., 2021), they can also occur in concert (Agarwal et al., 2017; Georgiou et al., 2022; Otsu et al., 2015; Shigetomi et al., 2013) and eventually expand to a larger cellular region, including the astrocytic soma, a phenomenon termed astrocyte calcium surge (Hirase et al., 2004). Moreover, the regulation of the amplitude and spatial extension of the astrocyte calcium signal by the coincident activity of different synaptic inputs has been proposed to endow astrocytes with integrative properties for synaptic information processing (Durkee and Araque, 2019a; Fedotova et al., 2023; Lines, 2025; Perea and Araque, 2005; Rupprecht et al., 2024). Because a single astrocyte may contact ~100,000 synapses (Bushong et al., 2002), the integrative properties of a single astrocyte and the control of the intracellular calcium signal propagation may have relevant consequences by regulating the spatial range of astrocyte influence on synaptic terminals (Fellin et al., 2004; Gordleeva et al., 2019; Perea and Araque, 2005). While there have been recent works describing molecular underpinnings of microdomain calcium transients (Diaz et al., 2019; Ma and Freeman, 2020; Montagna et al., 2019), the underlying processes governing the connection between the two subcellular activity states – independent or concerted events – and the spatial extension of the intracellular calcium signal remain unknown. To address these issues, we have monitored sensory-evoked astrocyte calcium activity in the mouse primary somatosensory cortex in vivo, combining astrocyte structural imaging data and subcellular imaging analysis. Here, we use an unbiased and semi-automatic algorithm to perform high-throughput analysis across a large number of astrocytes (~1000) to discover a subcellular property. We have found that astrocyte calcium responses originate in the surrounding arborizations and propagate to the soma if over 23% of the surrounding arborization is activated. If the astrocyte calcium spatial threshold is overcome, this spurs a surge of calcium into the surrounding arborization. Using transgenic Itpr2-/- mice, we found that the activation of type-2 IP3 receptors is necessary for the generation of astrocyte calcium surge. Patch-clamp recordings of neurons near activated astrocytes showed an increase in slow inward currents (SICs), detailing an output of astrocyte calcium surge. Using a combination of structural and functional two-photon imaging of astrocyte activity, we define a fundamental property of astrocyte calcium physiology, i.e., a spatial threshold for astrocyte calcium propagation. Results Imaging astrocyte structure and function simultaneously in vivo We simultaneously monitored calcium activity in identified sulforhodamine 101 (SR101)-labeled astrocytes in the primary somatosensory cortex using two-photon microscopy in vivo. We used transgenic mice expressing GCaMP6f in astrocytes, generated as described in the Methods (Bindocci et al., 2017; Lines et al., 2020), to monitor sensory-evoked intracellular astrocyte calcium dynamics in combination with SR101 labeling to monitor astrocyte morphology (Nimmerjahn et al., 2004; Figure 1A and B). Regions of interest (ROIs) were computationally determined from SR101-positive structural imaging (Bindocci et al., 2017) by outlining individual astrocytes and performing semi-automatic segmentation into somas and arborizations (Figure 1C; see Figure 1—figure supplement 1 for an in-depth description of segmentation). Next, subcellular quantification of soma and arborization calcium signals from individual astrocytes was evaluated in response to peripheral electrical stimulation of the hindpaw (2 mA at 2 Hz for 20 s; Figure 1D). Following the segmentation of an individual astrocyte into soma and arborization, the astrocyte arborization was further discretized into a grid of maximally sized 4.3 × 4.3 µm2 square regions of interest, which we define as astrocyte domains (Figure 1E and F; Agarwal et al., 2017; Di Castro et al., 2011; Grosche et al., 1999; Shigetomi et al., 2013). The concept of domain to define all subcellular domains in the astrocyte arborization should not be confused with the concept of microdomain, that usually refers to the distal subcellular domains in contact with synapses. Thus, we were able to quantify the sensory-evoked calcium responses in individual domains, as well as in the arborization and soma (Figure 1G). Imaging astrocyte structure and function simultaneously in vivo. (A) Scheme of in vivo preparation to image astrocyte Ca2+ and structure. (B) Sulforhodamine 101 (SR101)-stained astrocyte structure, GCaMP6 to monitor astrocyte Ca2+ signal, and merge. Scale bar = 50 µm. (C) Regions of interest (ROIs) from SR101-stained structure of somas (blue) and arborizations (red). (D) Ca2+ traces from B from somas (blue) and arborizations (red). Scale = F/Fo, 10 s. (E) SR101-stained astrocyte (left), ROIs outlining soma and arborization (center) and ROIs defining the soma and domains (right). Scale bar = 10 µm. (F) Pseudocolor Ca2+ image during basal (left) and hindpaw electrical stimulation (right). (G) Ca2+ traces from F from domains (salmon), arborization (red), and soma (blue). Scale = F/Fo, 10 s. Figure 1—figure supplement 1.Semi-automatic method for the segmentation of astrocyte morphology.(A) Sulforhodamine 101 (SR101)-stained astrocyte population with an outlined cell. Scale bar = 50 µm. (B) Selected astrocyte placed in polar coordinates with rings overlaid to assess structural fluorescence. Scale bar = 10 µm. (C) Average fluorescence of rings centered on astrocyte soma as radius is extended outward. Note, overlay of values from rings in B. (D) Fluorescence of rings in panel B as a function of angle. (E) Regions of interest (ROIs) from algorithm. (F) Calcium pseudocolor image during basal and stimulation. (G) Calcium traces from F during stimulation. Scale = F/Fo, 10 s. Semi-automatic method for the segmentation of astrocyte morphology. (A) Sulforhodamine 101 (SR101)-stained astrocyte population with an outlined cell. Scale bar = 50 µm. (B) Selected astrocyte placed in polar coordinates with rings overlaid to assess structural fluorescence. Scale bar = 10 µm. (C) Average fluorescence of rings centered on astrocyte soma as radius is extended outward. Note, overlay of values from rings in B. (D) Fluorescence of rings in panel B as a function of angle. (E) Regions of interest (ROIs) from algorithm. (F) Calcium pseudocolor image during basal and stimulation. (G) Calcium traces from F during stimulation. Scale = F/Fo, 10 s. The analysis of the sensory-evoked calcium activity from astrocyte arborization and soma uncovered that (1) the majority of responses occurred in both the soma and arborization (57.7 ± 4.5%; n=30 populations, 3 animals); (2) some responses occurred only in the arborization (15.1 ± 1.6%; n=30 populations, 3 animals; Figure 2A–D); (3) a small minority of responses included activity in the soma but not the arborization (3 ± 0.5%; n=30 populations, 3 animals; Figure 2D); and (4) some astrocytes did not respond (24.1 ± 3.6%; n=30 populations, 3 animals). Since the majority of cells showed responses in both the soma and arborization, we hypothesized that the proportion of domain activity within the arborization and the somatic calcium activity were correlated across a population of astrocytes. To test this hypothesis, we first examined the average percentage of responding arborizations vs the percentage of soma activation within a population and found a significant linear correlation between these subcellular measures of activity (linear correlation: p<0.001, R2=0.90; n=30 populations, 3 animals; Figure 2E). Additionally, averaging the percentage of active domains per cell over a population vs the percentage of somas active showed a significant linear correlation (linear correlation: p<0.001, R2=0.87; n=30 populations, 3 animals; Figure 2F). These results indicate that, on average, subcellular calcium events located in astrocyte arborizations are related to soma activation. Population arborization calcium is correlated to population soma activity. (A) Sulforhodamine 101 (SR101) staining. Scale bar = 50 µm. (B) Pseudocolor Ca2+ images at basal and stimulation. (C) Regions of interest (ROIs) of soma and arborizations/domains along with activity during stimulation. (D) Proportion of subcellular responses to stimulation. (E) Percentage of active arborizations vs percent of somas active. (F) Percentage of domains active vs percent of somas active. Mean ± SEM. Pearson correlation. Subcellular astrocyte calcium originates in the arborization We then analyzed the spatial and temporal properties of the intracellular calcium dynamics in astrocytic somas and arborizations (Figure 3A–D). First, we determined the kinetics of the sensory-evoked astrocyte calcium signal. Sensory-evoked calcium rises in arborizations occurred with a delay of 11.1±0.3 s from the onset of the peripheral stimulation and significantly preceded those occurring in the soma with a 13.2±0.2 s delay from stimulus onset (p<0.001; n=30 populations, 3 animals; Figure 3D–F). Moreover, rise time to peak and decay time back to baseline of the calcium traces were faster in somas than arborizations (10–90% rise time: 5.7±0.2 s in arborizations vs 3.5±0.2 s in somas; p<0.001; 90–10% decay time: 4.8±0.2 s in arborizations vs 4.3±0.2 s in somas; p<0.01; n=30 populations, 3 animals; Figure 3E and F). These results indicate that astrocyte responses occurred initially in the arborizations, which is consistent with the idea that synapses are likely to be accessed at the astrocyte arborization (Arizono et al., 2020; Papouin et al., 2017). Astrocyte calcium responses originate in the arborization before the soma. (A) Astrocyte with regions of interest (ROIs). Scale bar = 10 µm. (B) Pseudocolor Ca2+ image. (C) Ca2+ traces in B from domains (pink), arborization (red), and the soma (blue). Scale = F/Fo, 5 s. (D) Raster plot of astrocyte somas (blue) and arbors (red) in response to stimulation (gray). (E) Average calcium traces from somas (blue) and arborizations (red) aligned to their respective soma onset. (F) Soma and arbor latency to response (left), event rise time (center), and event decay time (right). n = 995 astrocyte reponses. Mean ± SEM. ‘**’ ≡ p<0.01 and ‘***’ ≡ p<0.001 using paired Student’s t-test. Figure 3—figure supplement 1.Distinct dynamics of domains before and after soma onset.(A) Average calcium traces from somas (blue) and domains activating before the soma (pre-soma; green) and after the soma (post-soma; pink) aligned to their respective soma onset. (B) Pre-soma and post-soma latency to response relative to their respective soma onset. (C) Event rise time. (D) Event decay time. n = 995 astrocyte reponses. Mean ± SEM. ‘***’ ≡ p<0.001 using paired Student’s t-test. Distinct dynamics of domains before and after soma onset. (A) Average calcium traces from somas (blue) and domains activating before the soma (pre-soma; green) and after the soma (post-soma; pink) aligned to their respective soma onset. (B) Pre-soma and post-soma latency to response relative to their respective soma onset. (C) Event rise time. (D) Event decay time. n = 995 astrocyte reponses. Mean ± SEM. ‘***’ ≡ p<0.001 using paired Student’s t-test. A spatial threshold to activate the soma and calcium surge Next, we determined the relative spatial relationship of calcium activity of domains within the arborizations and somas of individual astrocytes (Figure 4A). We quantified the proportion of subcellular domains in individual astrocytic arborizations that responded to electrical stimuli with varied parameters and assessed whether the corresponding soma responded (Figure 4A–C). When changing the stimulus parameters (duration, frequency, and intensity), the number of responding domains increased as the stimulus duration, frequency, and intensity increased (ANOVA: duration: p<0.001, frequency: p<0.001, intensity: p<0.001; n=11 populations, 4 animals; Figure 4D). As described above, the probability of soma activation vs the percentage of active domains could be accurately fit to a linear regression (see Figure 2F) indicating a correlation between these variables. To further characterize this relationship, we plotted paired values of on/off active soma (i.e. activated or not) vs the proportion of active domains from individual astrocytes. We found that the activation of a relatively low proportion of domains occurred without activation of the soma (Figure 4E). Conversely, large proportions of activated domains were accompanied by a calcium elevation in the soma (Figure 4E). This relationship suggested the existence of a threshold. Fitting these values to the Heaviside step function (in Methods, Equation 4; Davies, 2002) indicated that somas were active when at least 22.6% of their respective domains were active (R2=0.42; n=995 astrocytes from 30 populations and 3 animals; Figure 4E). This spatial threshold value was independent of the sensory input because similar values were found across various stimulus parameters (one-way ANOVA: duration: p=0.50, frequency: p=0.29, intensity: p=0.38; n=11 populations, 4 animals; Figure 4F), suggesting that it is determined by intrinsic astrocyte properties. Consolidating spatial threshold measurements from various stimulation parameters, we quantified the spatial threshold to be within 95% confidence intervals of [21.2%, 24.0%]. Moreover, plotting the percent of active domains for an individual astrocyte vs the amplitude of the somatic calcium response was fit to a sigmoid curve (R2=0.58; n=995 astrocytes from 30 populations and 3 animals; Figure 4G). These fits to cellular data as well as the large cluster of unchanged somatic amplitude with subthreshold domain activity further confirms that nonresponsive somas were not just below event detection, but indeed the soma does not become active. Taken together, these results indicate the existence of a spatial threshold for soma activation determined by astrocyte intrinsic properties and the proportion of active domains. A spatial threshold for astrocyte calcium in the soma to reach astrocyte calcium surge. (A) Astrocyte and regions of interest (ROIs). Scale bar = 10 µm. (B) Pseudocolor Ca2+ images during basal and different frequency of stimulations. (C) Scheme of domains (red) and soma (blue) Ca2+ activity from B. (D) Percentage of active domains vs stimulus duration, intensity, and frequency. n = 11 populations. (E) Active state of soma for individual astrocytes vs percentage of active domains (red). Data were fit to a Heaviside step function (blue dotted line). (F) Percentage of active domains necessary to elicit soma activation vs stimulus duration, intensity, and frequency. Blue dotted lines denote 22.6% spatial threshold. n = 11 populations. (G) Soma fluorescence vs percentage of active domains (red). Data were fit to a sigmoidal function (blue) and a blue dotted line denotes 22.6% spatial threshold. (H) Percentage of active domains in the absence of soma activation vs active domains before and after soma activation. Blue line denotes 22.6% spatial threshold. (I) Schematic showing subthreshold and suprathreshold astrocyte calcium activity. Mean ± SEM. ‘***’ ≡ p<0.001 and ‘ns’ ≡ p>0.05 using one-way ANOVA or Student’s t-test. The presence of a spatial threshold for somatic responses suggests that cells that respond with soma activity would have a higher response of domains prior to the soma response (pre-soma) when compared to astrocytes without a soma response (no-soma). Indeed, when comparing these populations we found that astrocytes with responding somas had a significantly larger proportion of surrounding domains active prior to soma activation (pre-soma) when compared to cells without a somatic response (no-soma), confirming our hypothesis of a spatial cellular threshold (18.5 ± 1.7% of active domains in no-soma vs 23.9 ± 0.8% of domains in pre-soma cells, p<0.05; n=607 active vs n=388 not active, 30 populations and 3 animals; Figure 4H). Further, we found that astrocytes with an excess of 22.6% of domain activity, that induced a somatic response, led to increased domain activation throughout the remaining arborization (post-soma), i.e., domain activity before soma activation (pre-soma) vs domain activity after soma activation (post-soma) (23.9 ± 0.8% in pre-soma firing compared to 45.0 ± 1.8% in post-soma activation, p<0.001; n=607 astrocytes, 30 populations in 3 animals; Figure 4H, Figure 3—figure supplement 1). Together, these results confirm that astrocytes that respond with domain activity in excess of the spatial threshold precipitates somatic activity and a calcium surge of expanded responses throughout the astrocytic arborization (Figure 4I). The spatiotemporal characteristics of active domains in astrocyte calcium responses Our results demonstrate the relationship between the percentage of active domains and soma activation and subsequent calcium surge. Next, we were interested in the spatiotemporal properties of domain activity leading up to and during calcium surge. Because we imaged groups of astrocytes, we were able to constrain our analyses to early responders (onset As discussed above, the evidence does not seem to be strong enough to assert that IP3R2 is necessary to trigger somatic events. The results indicate that the IP3R2 pathway seems to facilitate the emergence of somatic events. As astrocytes differ strongly in terms of morphology and expression profiles depending on physiological conditions, the conclusions of this study might only apply to the specific experimental conditions used: region studied, age of the animal, type of sensory stimuli performed, and so on. - "These results indicate that spatial threshold of the astrocyte calcium surge has a functional impact on gliotransmission, which have important consequences on the spatial extension of the astrocyte-neuron communication and synaptic regulation", l.41-48 page 11. Figure 6 seems to indicate a correlation between the proportion of astrocyte domains activated and the frequency of SICs. The data seems insufficient to conclude that there is a causal relationship between calcium surge in the astrocyte and gliotransmission or SIC frequency. -" These results indicate that, on average, subcellular calcium events located in astrocyte arborizations are related to soma activation.", page 6 l 15-16. It may be more informative to specify the correlation measured: i.e the larger the arborization activity, the larger the percentage of active somas. Figures Figure 2: Adding more details in the figure legend explaining how the different parameters are calculated might be useful to the reader. Notably, what does soma active (%) refer to? Figure 3: Could it be possible to add individual traces of calcium activity in the soma and arborization of individual cells to provide a glimpse of the variability of the signals measured? Fig4. B-C: Could it be possible to add in the legend information on the timeline between stimulation and calcium signal recording? (and the duration of the latter). Fig4 D-E: Why is the maximum number of active domains in panel D ~50-60% but goes up to ~100% in panel E? Could it be that plotting SEM rather than STD might misrepresent the variability in the percentage of active domains for each stimulus property? Fig4F: It seems that the threshold changes with the frequency of the stimulus: e.g. at 10 Hz, the threshold seems larger than 22.6%. What would that mean? Fig4G: - Why do some data points display a soma amplitude < 0 DF/F0 ? - Why choose a sigmoid fit? What are the statistics associated to the fit? Is it in accordance with the threshold of 23%? Would a linear fit provide a good fit? Fig5F: - It seems that a few IP3R2-/- astrocytes displayed somatic events? If so, it might be interesting to mention this in the discussion section and to speculate on why that might be. - It seems that panel 5F displays the average percentage of somas that got activated rather than the probability of somatic events. - Is it possible that the effect seen in domains vs arborization is due to statistical effects (as n=2450 vs 112)? Fig S1: Panel D legend: double labeling of the radius used for each plot might be useful, notably for colorblind readers as the colors might be hard to see. Discussion - The discussion section might benefit from a discussion on the similitude between the data presented here and previous reports that reported similar results, i.e that most calcium signals in astrocytes were located in the distal processes, forming microdomains that rarely propagated to the soma. These include Bindocci et al 2017 Science (DOI:10.1126/science.aai8185) and Georgiou et al, Science Advances, 2022 (DOI: 10.1126/sciadv.abe5371). Thank you for the suggestions. We have now changed portions of the Methods, Results and Discussion sections. Reviewer #3 (Recommendations For The Authors): The text could potentially be improved somewhat. Thank you.