1. Neuroscience
Download icon

Astrocytes contribute to synapse elimination via type 2 inositol 1,4,5-trisphosphate receptor-dependent release of ATP

  1. Junhua Yang
  2. Hongbin Yang
  3. Yali Liu
  4. Xia Li
  5. Liming Qin
  6. Huifang Lou
  7. Shumin Duan  Is a corresponding author
  8. Hao Wang  Is a corresponding author
  1. Zhejiang University School of Medicine, China
Research Article
  • Cited 36
  • Views 2,719
  • Annotations
Cite this article as: eLife 2016;5:e15043 doi: 10.7554/eLife.15043

Abstract

Selective elimination of unwanted synapses is vital for the precise formation of neuronal circuits during development, but the underlying mechanisms remain unclear. Using inositol 1,4,5-trisphosphate receptor type 2 knockout (Itpr2−/−) mice to specifically disturb somatic Ca2+ signaling in astrocytes, we showed that developmental elimination of the ventral posteromedial nucleus relay synapse was impaired. Interestingly, intracerebroventricular injection of ATP, but not adenosine, rescued the deficit in synapse elimination in Itpr2−/− mice. Further studies showed that developmental synapse elimination was also impaired in P2ry1−/− mice and was not rescued by ATP, indicating a possible role of purinergic signaling. This hypothesis was confirmed by MRS-2365, a selective P2Y1 agonist, could also rescue the deficient of synapse elimination in Itpr2−/− mice. Our results uncovered a novel mechanism suggesting that astrocytes release ATP in an IP3R2-dependent manner to regulate synapse elimination.

https://doi.org/10.7554/eLife.15043.001

eLife digest

Neighbouring neurons connect to each other and share information through structures known as synapses. As the brain develops, many synapses turn out to be redundant. Just like trees in a garden that need to be trimmed, these redundant synapses must be pruned in order to form the right pattern of connections between different neurons. Brain cells called astrocytes play a key role in synaptic pruning, but it is unclear exactly how astrocytes coordinate this process.

One important way in which astrocytes communicate with neurons is through a process called calcium signaling, in which the movement of calcium ions into or out of the cell sets off a cascade of activity inside the astrocytes. Yang et al. have now studied developing mice that lacked a gene that is essential for calcium signaling in astrocytes. Two weeks after they were born, these mice still had redundant synapses that are normally lost after birth. However, injecting the developing brain with a substance called ATP prevented this defect and allowed synapses to be correctly pruned. This is likely to be because astrocytes also use ATP to communicate with neurons, and ATP compensated for the missing calcium signaling.

The experiments also uncovered the specific structure – called the P2Y1 receptor – on the outer surface of a neuron that ATP latches on to in order to help remove synapses. Further work is now needed to reveal how activating the P2Y1 receptor coordinates synaptic removal.

https://doi.org/10.7554/eLife.15043.002

Introduction

Synapse elimination, a process of pruning inappropriate synapses during development is essential for the formation of neuronal circuits and proper brain function (Katz and Shatz, 1996; Paolicelli et al., 2011; Sanes and Lichtman, 1999). Disruption of this process is likely involved in many neurological diseases such as schizophrenia and autism (Hoffman and McGlashan, 1997; Tang et al., 2014; Tsai et al., 2012). Although activity-dependent competition among different inputs has been suggested as the key mechanisms for synapse elimination, the underlying molecular and cellular mechanisms remain unclear. In addition to the retino-geniculate and climbing fiber-Purkinje cell pathways (Chen and Regehr, 2000; Kano and Hashimoto, 2009), the principal trigeminal nucleus-ventral posteriormedial thalamic nucleus (Pr5-VPm) connection in mice has proven to be an excellent model in which to investigate developmental synapse elimination in the central nervous system (Arsenault and Zhang, 2006). VPm neurons receive a contra-lateral Pr5 projection and relay the information to the somatosensory cortex. At an early age like postnatal day 7 (P7), each VPm neuron receives an average of 7–8 Pr5 inputs. Most of these inputs are eliminated in an activity-dependent manner during development and the majority of VPm neurons only receive a single Pr5 input by P16.

Over the past decade, emerging evidence has suggested that astrocytes not only provide structural and metabolic support to neighboring neurons, but also participate in synapse formation, synaptic transmission, and synaptic plasticity (Allen, 2014; Henneberger et al., 2010; Zhang et al., 2003). Structurally, astrocytes contact 50–90% of the synapses in a given brain region (Genoud et al., 2006; Oliet et al., 2001) and form one component of the “tripartite synapse” (Araque et al., 1999). Functionally, intracellular Ca2+ ([Ca2+]i) elevation in neighboring astrocytes can be induced by activation of the synapse, triggering the release of gliotransmitters such as ATP and D-serine that modulate synaptic efficacy and plasticity (Chen et al., 2013; Fields and Burnstock, 2006; Henneberger et al., 2010; Yang et al., 2003). It is generally accepted that astrocytes release gliotransmitters in a G-protein-coupled receptor/ inositol 1,4,5-trisphosphate receptor (GPCR/IP3R)-mediated Ca2+-dependent manner. Studies in both acute slice preparations and in vivo have shown that, in response to neuronal activity, GPCR-mediated Ca2+ signaling in astrocytes depends on the activation of IP3R2, a subunit that is solely expressed in astrocytes (Fiacco and McCarthy, 2004; Navarrete et al., 2012). Controversially, studies also indicated that IP3R2-mediated Ca2+ signaling in astrocytes may not be involved in the acute modulation of neuronal activity and synaptic transmission (Agulhon et al., 2010; Petravicz et al., 2008). Indeed, there are other Ca2+ signaling pathways in astrocytes which do not involve IP3R2. A recent study showed that although the somatic Ca2+ signaling was eliminated, Ca2+ signaling in processes was intact in the astrocyes of Itpr2−/− mice (Srinivasan et al., 2015). Therefore, IP3-induced Ca2+ increase is only the major source for somatic Ca2+ signaling in astrocytes. Recent studies have shown that astrocytes could mediate synapse elimination in a neuronal activity-dependent manner through two distinct pathways: either directly by phagocytosis through the MEGF10 and MERTK pathways, or by activating the classical complement cascade via astrocyte-derived transforming growth factor-β (TGF-β) (Bialas and Stevens, 2013; Chung et al., 2013). Here, we hypothesized that IP3R2-dependent Ca2+ signaling in astrocytes might be involved in regulating developmental synapse elimination. The results from studies in the cerebellum suggested that Ca2+-permeableα-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs) in Bergmann glia are required for the developmental elimination of climbing fiber-Purkinje cell inputs, which provides a clue that astrocytic Ca2+ signaling might be involved in synapse elimination (Iino et al., 2001). Unfortunately, it is difficult to draw an exclusive conclusion since Bergmann glia are morphologically distinct from astrocytes and detect synaptic activity at cerebellar synapses primarily though activation of Ca2+-permeable AMPARs as well as GPCRs.

By combining electrophysiological, pharmacological, and immunohistochemical methods, we discovered that selectively disturbing [Ca2+]i signaling in astrocytes using Itpr2−/− mice impaired the developmental elimination of VPm relay synapses. Intracerebroventricular injection of ATP, but not adenosine, rescued the impaired synapse elimination in these mice. We further found that developmental synapse elimination was also impaired in P2ry1−/− mice and could not be rescued by ATP. Last, the deficit of synapse elimination in Itpr2−/− mice was rescued by intracerebroventricular injection a selective P2Y1 receptors agonist MRS-2365. Overall, we provide direct evidence to show that astrocytes contribute to synapse elimination in an IP3R2-dependent manner through activation of purinergic signaling.

Results

Synapse elimination is disrupted at the VPm relay synapse in Itpr2−/− mice at P16-17

To determine the role of astrocytic Ca2+ signaling in synapse elimination, we need an effective way to selectively disrupt the function of astrocytes. In mammals, astrocytes rely on IP3R2-mediated intracellular Ca2+ signaling to perform their functions. A large literature suggested that IP3R2 is the only subtype astrocytes expressed and is not expressed in microglia and neurons in the cerebrum. Thus Itpr2−/− mice could be used to study the specific roles of Ca2+ signaling in astrocytes for synapse function (Hertle and Yeckel, 2007; Li et al., 2015; Sharp et al., 1999). In line with previous studies, we also found that IP3R2 was co-expressed with GFAP, but not bio-markers for microglia or neurons in the brain areas we tested including hippocampus (Figure 1—figure supplement 1). Using Ca2+ imaging in acute brain slices, we next found that the ATP -induced somatic [Ca2+]i elevation in the astrocytes but not in the neurons of Itpr2−/− mice (Figure 1—figure supplement 2b,c) was abolished in both of the VPm and hippocampus, confirming that Ca2+ signaling was selectively impaired in astrocytes in Itpr2−/− mice. Next, we examined developmental synapse elimination by whole-cell patch recording in acute brain slices. Interestingly, we found a marked difference in the mean number of inputs received by each VPm neuron between WT and Itpr2−/− mice at P16-18 (WT = 1.2 ± 0.02, n = 26 cells from 4 mice; Itpr2−/− = 2.1 ± 0.10, n = 40 cells from 6 mice; p<0.01, Figure 1d). In WT mice, only 27% (7 of 26) of VPm relay neurons received multiple Pr5 inputs at this age (Figure 1a,c), whereas most of these neurons (72%, 32 of 42) in Itpr2−/− mice received multiple Pr5 inputs (Figure 1b,c). VPm relay neurons receive two major excitatory inputs: from layer VI cortex and the other from the Pr5 that express vesicular glutamate transporter 1 (VGluT1) and VGluT2, respectively (Graziano et al., 2008). Each Pr5 input forms multiple synaptic contacts with VPm neurons and thus, the number of inputs indicates how many axonal projections while the VGluT2 staining represents number of synaptic terminals. The pruning of somatic innervations by Pr5 inputs in the VPm is always related to the elimination of VPm relay synapses, as showed by previous studies (Takeuchi et al., 2014; Zhang et al., 2012). To further verify that there were more synapses in KO mice, we immunostained for VGluT2. Consistent with the electrophysiological results, we observed more VGluT2 puncta around the soma as well as the total numbers of puncta in Itpr2−/− mice (Figure 1e,f), indicating that there were more synapses in KO mice. In addition, we found that neuron number did not significantly change in the VPm between WT and Itpr2−/− mice at this age (Figure 1—figure supplement 3). These results strongly suggested that astrocytic IP3R2-dependent Ca2+ signaling is required for synapse elimination.

Figure 1 with 4 supplements see all
Developmental synapse elimination was impaired in Itpr2−/− mice at P16-17.

(a and b) Left panels, sample traces of membrane current in response to stimuli over a range of intensities in VPm neurons at P16-17 in WT (a) and Itpr2−/− mice (b). Currents recorded at +40 mV are mediated by NMDA receptors, and those at -70 mV are mediated by AMPA receptors. Right panels, peak current versus stimulus intensity for WT (a) and Itpr2−/− mice (b). (c) Distributions of the number of Pr5 axons innervating each VPm neuron during P16-17 in WT (n = 26 cells) and Itpr2−/− (n = 42 cells) mice. ***p<0.001, χ2 test. (d) Histogram of average number of inputs received by each VPm neuron at P16-17 in WT and Itpr2−/− mice. ***p<0.001, unpaired Student’s t test. Error bars indicate SEM. (e) Sample confocal images of immunostained neurons and Pr5 axonal terminals in the VPm at P16. Neurons were labeled by the NeuN antibody (green), and Pr5 axon terminals were labeled by the VGluT2 antibody (red). Inset is higher-magnification of the boxed area. Scale bar, 5 µm. (f) Quantification of VGluT2 puncta/soma (left, n = 40 cells/group) and VGluT2 puncta/neuron (right, n = 12 sections from 4 mice/group) for WT and Itpr2−/− mice. ***p<0.001, unpaired Student’s t test. Error bars indicate SEM.

https://doi.org/10.7554/eLife.15043.003

During development, the strengthening of immature synapses often concurrent with the removal of redundant inputs. In this study, we did not find any change in maximal AMPAR-EPSCs, N-methyl-D-aspartate receptor (NMDA-) EPSCs and APMAR-EPSC/NMDAR-EPSC ratios recorded at VPm relay synapses when WT mice were compared with Itpr2−/− at P16 (Figure 1—figure supplement 4). Given that the Itpr2−/− mice had more inputs and synapses, the strength of each input or individual synapse was weaker than that in WT mice (Figure 1—figure supplement 4b). These data revealed that the deficit of synapse elimination was sufficient to affect synaptic strengthening during development.

Connectivity of Pr5-VPm pathway is not affected in Itpr2−/− mice at P7

To distinguish whether the deficit in synapse elimination was due to the failure of developmental elimination or abnormal synaptogenesis at an early age, we tested the connectivity of the Pr5-VPm projection at P7. We found that the number of inputs received by VPm neurons was comparable in the WT and Itpr2−/− mice (Figure 2a–d). Neither the numbers of VGluT2 puncta per soma nor per neuron differed between WT and Itpr2−/− mice at this age, indicating similar numbers of synapses in these mice (Figure 2e,f). These results suggested that synapse formation was normal at early postnatal stage in Itpr2−/− mice. To address the possibility that the deficit in synapse elimination we found at P16 was due to a delay in development, we assessed the synapse numbers at P30 by immunostaining for VGluT2. We found that even at P30, there were more Pr5-VPm synapses in Itpr2−/− mice than in WT mice (Figure 2—figure supplement 1a,b). Therefore, knockout of IP3R2 disrupted developmental synapse elimination at the VPm relay synapses.

Figure 2 with 1 supplement see all
Connectivity of Pr5-VPm pathway was comparable in WT and Itpr2−/− mice at P7.

(a and b) Left panels, sample traces showing membrane current in response to stimuli at a range of intensities in VPm neurons at P7 in WT (a) and Itpr2−/− (b) mice. Right panels, peak current versus stimulus intensity for WT (a) and Itpr2−/− mice (b). (c) Distributions of the number of Pr5 axons innervating each VPm neuron at P7 did not differ between WT (n = 33 cells) and Itpr2−/− (n = 35 cells) mice. p=0.78, χ2 test. (d) Histogram of average number of inputs received by each VPm neuron at P7 in WT and Itpr2−/− mice (WT = 4.8 ± 0.3, n = 33; Itpr2−/− = 4.9 ± 0.3, n = 33). p=0.74, unpaired Student’s t test. (e) Sample confocal images of immunostained neurons and Pr5 axonal terminals in the VPm at P7 in WT and Itpr2−/− mice. Neurons were visualized with the NeuN antibody (green), and Pr5 axonal terminals were labeled by the VGluT2 antibody (red). Inset is higher-magnification of the boxed area. Scale bar, 5 µm. (f) Quantification of VGluT2 puncta/soma (left, n = 24 cells/group,) and VGluT2 puncta/neuron (right, n = 9 sections from 3 mice/group) for WT and Itpr2−/− mice. p>0.05, unpaired Student’s t test. Error bars indicate SEM.

https://doi.org/10.7554/eLife.15043.008

Intracerebroventricular injection of ATP from P11 to P15 rescued the deficit in synapse elimination in Itpr2−/− mice

Increasing evidence suggests that astrocytes release a number of gliotransmitters, such as glutamate, ATP, and D-serine to regulate synaptic transmission and synaptic plasticity (Chen et al., 2013; Henneberger et al., 2010; Jourdain et al., 2007). Among these gliotransmitters, we were particularly interested in ATP and assumed that it might play a role in synapse elimination for the following reasons: 1) the basal ATP levels are reduced in Itpr2−/− mice (Cao et al., 2013); 2) astrocytes release ATP in a Ca2+-dependent manner (Lalo et al., 2014; Zhang et al., 2007); and 3) ATP regulates synaptic plasticity (Chen et al., 2013). We then tested whether the basal ATP level in the VPm changes during development at the time when exploratory activity increases in WT mice. A significant increase in the basal ATP level occurred at P18 compared with that at P7 in WT mice (Figure 3a). This developmental up-regulation of the basal ATP level was absent in Itpr2−/− mice (Figure 3a). In addition, we found that the basal ATP level was comparable between WT and Itpr2−/− mice at P7 but was significantly reduced in KO mice at P18 (Figure 3a). These results suggested that, during development, the basal ATP level increases in WT but not in Itpr2−/− mice. Next, we tested whether the deficit of synapse elimination in Itpr2−/− mice could be rescued by compensatory ATP. To achieve this goal, we implanted a cannula in the left lateral ventricular of the brain (Figure 3b) and found that intracerebroventricular injection of ATP from P11 to P15 rescued the synapse elimination deficit (Figure 3e,i,l,m). At P16-17, unlike aCSF-treated mice, the majority of VPm neurons in Itpr2−/− mice that had received ATP (50 µM) treatment was innervated by a single Pr5 input (Figure 3l). Immunostaining for VGluT2 also revealed that the number of synapses significantly decreased in Itpr2−/− mice with ATP treatment (Figure 3p,q). These results indicate that ATP treatment is sufficient to rescue synapse elimination deficit in Itpr2−/− mice and the removal of redundant synapses is independent of IP3R2-dependent Ca2+ signaling. Cannula implantation caused injury may induce inflammatory responses of astrocytes and microglia differentially between aCSF and ATP treated-Itpr2−/− mice, thereby affected the rescue effects. We found that reactive astrocytes and activated microglia around the site of cannula placement in aCSF and ATP treated-Itpr2−/− mice were identical (Figure 3—figure supplement 1), and ruled out this possibility. We next applied a low dose ATP (5 µM) and found that the impaired synapse elimination cannot be rescued in Itpr2−/− mice, suggesting the rescue effect of ATP was in a dose-dependent manner (Figure 3n,o).

Figure 3 with 1 supplement see all
Intracerebroventricular injection of ATP from P11 to P15 rescued the synapse elimination deficit in Itpr2−/− mice.

(a) Basal ATP levels at P8 and P17-19 in WT and Itpr2−/− mice. *p<0.05, two-way ANOVA followed by Bonferroni post hoc test, n = 5 mice per age. n.s, not significant. Error bars indicate SEM. (b) Schematic illustrating the cannula was implanted to the left lateral ventricle. (c) Experimental design of aCSF, ATP (50 µM), ATPγS (50 µM), and adenosine (50 µM) administration and analysis. Injections were given twice per day from P11 to P15. (d-g) Sample traces showing membrane current in response to stimuli with a range of intensities in VPm neurons from Itpr2−/− mice at P16-17 with a short period of aCSF (d), ATP (e), ATPγS (f) and adenosine (g) treatment. (h-k) Peak current versus stimulus intensity for aCSF (h), ATP (i), ATPγS (j) and adenosine (k) treatment in Itpr2−/− mice. (l) Distributions of the number of Pr5 axons innervating each VPm neuron at P16-17 with aCSF (n = 29 cells), ATP (n = 32 cells), adenosine (n = 27 cells), and ATPγS (n = 32 cells) treatment in Itpr2−/− mice. **p<0.01, ***p<0.001, χ2 test. (m) Histogram of average number of inputs received by each VPm neuron at P16-17 in WT and Itpr2−/− mice followed by different injections. ***p<0.001, one-way ANOVA followed by Bonferroni post hoc test. Error bars indicate SEM. (n) Distributions of the number of Pr5 axons innervating each VPm neuron at P16-17 with different concentrations of ATP treatment in Itpr2−/− mice. ***p<0.001, χ2 test. (o) Histogram of average number of inputs received by each VPm neuron at P16-17 in Itpr2−/− mice followed by different concentrations of ATP injection. ***p<0.001, unpaired Student’s t test. (p) Sample confocal images of immunostained neurons and Pr5 axon terminals in the VPm at P16. Inset is higher-magnification of the white boxed area. Scale bar, 5 µm. (q) Quantification of VGluT2 puncta/soma (left, n = 40 cells/group) and VGluT2 puncta/neuron (right, n = 9 sections from 3 mice/group) for WT, Itpr2−/−, ATP and adenosine-treated Itpr2−/− mice. ***p<0.001, one-way ANOVA. Error bars indicate SEM.

https://doi.org/10.7554/eLife.15043.010

Given that ATP is readily hydrolyzed, and to exclude the possibility that the degradation products of ATP play a role in synapse elimination, we injected the non-hydrolyzable ATP analog ATPγS and found that, ATPγS (50 µM) also rescued the synapse elimination impairment in Itpr2−/− mice (Figure 3f,j,l,m). These results suggested that astrocytes mediate synapse elimination via the Ca2+-dependent release of ATP. Since adenosine is a degradation product of ATP and could modulate synaptic plasticity by activating adenosine receptors (Pascual et al., 2005), we set out to exclude the possibility that adenosine rescued the synapse elimination deficit in Itpr2−/− mice. We found intracerebroventricular injection of adenosine (50 µM) throughout P11-P15 had no effect on Pr5-VPm connectivity at P16 in these mice (Figure 3g, k, l, m). Thus, the synapse elimination deficit in Itpr2−/− mice was rescued by ATP, but not adenosine.

ATP modulates synapse elimination through activation of P2Y1 receptors

Having shown that ATP rescued the synapse elimination deficit at the VPm relay synapses in Itpr2−/− mice, the next question is how this process occurs. In the CNS, ATP could activate purinegic signaling through P2X or P2Y receptors to modulate synapse function. Particularly, P2Y1 receptors have been shown participated in ATP-induced heterosynaptic long-term depression (LTD) in hippocampus (Chen et al., 2013), and LTD has been suggested correlated to synapse elimination (Bastrikova et al., 2008; Wiegert and Oertner, 2013). To test the possibility of involvement of P2Y1 receptors in synapse elimination at the VPm, we first examined the expression pattern of P2Y1 receptors and found that unlike in the hippocampus (Zhu and Kimelberg, 2004), they were only expressed in neurons in the VPm (Figure 4—figure supplement 1). We next found that ATP-induced somatic Ca2+ elevation in astrocytes was impaired in hippocampus, but not in the VPm in P2ry1−/− mice (Figure 4—figure supplement 2), further indicated that the expression pattern of P2Y1 receptors might be different in hippocampus when compared with that in the VPm. In line with immunostaining results, we also found that ATP-induced Ca2+ elevation in VPm neurons of P2ry1−/− mice was impaired (Figure 4—figure supplement 2). These results suggested that even in the VPm, the expression pattern of P2Y1 may have specificity among different cell types. Next, we determined whether synapse elimination was disrupted in P2ry1−/− mice. Compared to WT mice, the majority of VPm neurons in P2ry1−/− mice received multiple Pr5 inputs at P16 (WT = 30%, 13 of 33 cells; P2ry1−/− = 72%, 34 of 47 cells, Figure 4g). Consistent with the electrophysiological results, we confirmed the synapse elimination deficit in P2ry1−/− mice with immunostaining for VGluT2 (Figure 4i). There was more VGluT2 staining around the somata of VPm neurons in P2ry1−/− mice than in WT mice (Figures 4i,j). Therefore, synapse elimination was also impaired in P2ry1−/− mice. To address the abnormal of synaptic connectivity in P2ry1−/− mice at P16 was actually due to developmental failure, we tested Pr5-VPm synapses at P7. We found that the connectivity of Pr5-VPm pathway was comparable between WT and P2ry1−/− mice at P7 (Figure 4—figure supplement 3). In addition, the synapse elimination deficit in these mice was not rescued by intracerebroventricular injection of ATP from P11 to P15 (Figure 4c, f, g, h). These results suggested that synapse elimination promoted by ATP may depend on P2Y1 receptors. If this is the case, one direct evidence would be the rescue of the synapse elimination defect with application of the P2Y1 agonist in Itpr2−/− mice. We next found that the deficit of synapse elimination was rescued by intracerebroventricular injection of a selective P2Y1 receptors agonist, MRS-2365 (20 µM) from P11 to P15 (Figure 5). Unlike aCSF treatment, the majority of VPm neurons in Itpr2−/− mice with treatment of MRS-2365 innervated by a single Pr5 input (Figure 5c). The mean number of inputs received by each VPm neuron was significantly decreased in Itpr2−/− mice with intracerebroventricular injection of MRS-2365 (2.0 ± 0.76 MRS-2365 versus 1.33 ± 0.48 aCSF, Figure 5d). Taken together, these data confirmed that ATP promotes synapse elimination via P2Y1 receptors.

Figure 4 with 3 supplements see all
Synapse elimination was also impaired in P2ry1−/− mice at P16-17 and cannot be rescued by ATP.

(a-c) Sample traces of membrane current in response to stimuli at a range of intensities in VPm neurons at P16-17 in WT (a), P2ry1−/− (b), and P2ry1−/− with ATP injection (c) mice. (d-f) Peak current versus stimulus intensity for WT (d), P2ry1−/− (e), and P2ry1−/− with ATP injection (f) mice. (g) Distributions of the number of Pr5 axons innervating each VPm neuron at P16-17 in WT (n = 33 cells), P2ry1−/− (n = 47 cells) and P2ry1−/− with ATP injection (n = 39 cells) mice. **p<0.01, χ2 test. (h) Average number of Pr5 axonal inputs received by individual VPm relay neurons significantly increased in P2ry1−/− mice compared to WT controls. ***p<0.001 by one-way ANOVA. (i) Sample confocal images of immunostained neurons and Pr5 axon terminals in the VPm at P16. Neurons were visualized with the NeuN antibody (green), and Pr5 axon terminals were labeled by the VGluT2 antibody (red). Inset is higher-magnification of the boxed area. Scale bar, 5 µm. (j) Quantification of VGluT2 puncta/soma (left, n = 42 cells/group) and VGluT2 puncta/neuron (right, n = 9 sections from 3 mice/group) in WT and P2ry1−/− mice. ***p<0.001, unpaired Student’s t test. Error bars indicate SEM.

https://doi.org/10.7554/eLife.15043.012
Synapse elimination was rescued by P2Y1 receptor agonist MRS-2365 in Itpr2−/− mice.

(a and b) Left panels, sample traces of membrane current in response to stimuli over a range of intensities in VPm neurons at P16-17 in aCSF (a) and MRS-2365 (b) treated-Itpr2−/− mice. Right panels, peak current versus stimulus intensity for aCSF (a) and MRS-2365 (b) treated-Itpr2−/− mice. (c) Distributions of the number of Pr5 axons innervating each VPm neuron during P16-17 in aCSF (n = 32 cells) and MRS-2365 treated-Itpr2−/− mice(n = 30 cells). ***p<0.001, χ2 test. (d) Histogram of average number of inputs received by each VPm neuron at P16-17 in aCSF (n = 32 cells) and MRS-2365 (n = 30 cells) treated-Itpr2−/− mice. ***p<0.001, unpaired Student’s t test. Error bars indicate SEM.

https://doi.org/10.7554/eLife.15043.016

Discussion

Here, by using electrophysiological recording in transgenic mice, we found that astrocytes contribute to synapse elimination through Ca2+-dependent release of ATP and activation of P2Y1 receptors. Specifically, we demonstrated that (1) developmental synapse elimination in the VPm was impaired in the Itpr2−/− mice; (2) intracerebroventricular injection of ATP and ATPγS, but not adenosine, from P11 to P15, rescued the synapse elimination deficit in Itpr2−/− mice; (3) P2ry1−/− mice also showed impaired synapse elimination and this could not be rescued by ATP; and (4) deficient of synapse elimination in Itpr2−/− mice was rescued by intracerebroventricular injection of MRS-2365, a selective P2Y1 receptors agonist, from P11 to P15.

In the CNS, neurons make redundant synaptic connections early in development, and most of inappropriate synapses are eliminated later to form precise neural circuitry (Lichtman and Colman, 2000; Luo and O'Leary, 2005). This process is believed to largely rely on neuronal activity, while the underlying mechanisms are still unclear. Recent studies have shown that astrocytes show a transient [Ca2+]i elevation, which depends on the GPCR/IP3R pathway, in response to neuronal activity and subsequently release gliotransmitters to modulate synaptic transmission and plasticity (Araque et al., 2014). Therefore, astrocytic [Ca2+]i elevation is a consequence of neuronal activity. Based on these reasons, we proposed that activity-dependent competition for synapse elimination involves both postsynaptic neuronal activation and astrocytic [Ca2+]I elevation. A large literature has shown that GPCR/IP3R-dependent Ca2+ signaling is critical for astrocytes to release gliotransmitters and modulate synaptic transmission. Conversely, recent studies from McCarthy’s laboratory have suggested that IP3R2-mediated Ca2+ signaling in astrocytes may not be involved in the acute modulation of neuronal activity and synaptic transmission (Fiacco et al., 2007; Petravicz et al., 2008). Two studies have shown that there were still some Ca2+ fluctuations occurring in astrocytes process, although the somatic Ca2+ signaling was eliminated in Itpr2−/− mice (Kanemaru et al., 2014; Srinivasan et al., 2015). These results suggested that Ca2+ signaling was not totally removed in these mutants. In line with previous studies, we found that the lack of ATP-induced Ca2+ increase in astrocytes soma in these mutants. Our studies strongly suggested that astrocytic IP3R2-dependent somatic Ca2+ signaling is required for developmental synapse elimination. Considering the dynamic and complexity, the astrocytic Ca2+ signaling in the process and the soma may play different roles in gliotransmitter release and regulating neuronal functions (Volterra et al., 2014).

It is important to note that synapse elimination was only partly impaired in the Itpr2−/− and P2ry1−/− mice. Although we found both anatomical and electrophysiological evidence that these mice had significant deficits in developmental synapse elimination, some degree of synaptic pruning still occurred during P7 to P16, suggesting that there are multiple mechanisms for controlling synapse elimination. Indeed, synaptic pruning in the VPm relay synapses occurs at several developmental stages (Wang and Zhang, 2008). Before P10, pruning is insensitive to whisker deprivation, but then later, the elimination of redundant inputs is largely dependent on somatosensory experience. In the mutants, such as Itpr2−/− mice, the nature of sensory information transmitted from whisker to the VPm might be altered, thus causes the defects of synapse elimination. We found surprisingly, the connectivity between Pr5 and VPm in Itpr2−/− mice undergone ATP and ATPγS treatments restored to WT mice level. These results suggested that no matter sensory information had been changed or not, ATP was sufficient to rescue the impairment of synapse elimination in Itpr2−/− mice. However, the rescue effect of ATP may occur at a site between the whisker and VPm, and this possibility could not be ruled out by our study.

It is about P13 that mice open their eyes and start to use their vibrissae to explore the environment. Accompanying this behavioral change, we found a significant enhancement of the basal ATP level in the VPm of WT mice between P7 and P18 (Figure 3a). However, the basal ATP level failed to be up-regulated in Itpr2−/− mice during development. In line with previous studies (Cao et al., 2013), we also found a decreased basal ATP level in the brain of Itpr2−/− mice. ATP can be released by both neurons and glia. The decreased ATP level in Itpr2−/− mice was most likely due to astrocytes since IP3R2 was not expressed in neurons and [Ca2+]i elevation did not change in neurons. Although ATP itself is a signaling molecule by activating P2X or P2Y receptors, the degraded form adenosine is also a ligand for adenosine receptors and causes synaptic depression (Wu and Saggau, 1994). The fact that adenosine injection through P11-15 did not rescue the synapse elimination deficit in Itpr2−/− mice excludes the possibility of a role for adenosine in synapse elimination. In addition to being a signal molecule, ATP is also energy currency in the brain and the phenotype of Itpr2−/− mice could be a result of global metabolic stress. This possibility is very unlikely because either ATPγS, a non-hydrolyzable ATP analog or one selective P2Y1 receptors agonist MRS-2365 could rescue the deficit of synapse elimination in Itpr2−/− mice. In our study, we concluded that ATP was not acting as an energy supply to play a role in synapse elimination.

Although both of Itpr2−/− and P2ry1−/− mice showed deficits of developmental elimination of relay synapses at the VPm, it is not necessary that they are mechanistically related. That is, the phenotypes in these two mutants are two independent events. The results of P2Y1 receptors agonist injection rescued the deficient of synapse elimination in Itpr2−/− mice provided a direct linkage between IP3R2-dependent releasing of ATP and its downstream purinergic pathway. Unlike IP3R2, studies have shown that P2Y1 receptors are widely expressed in astrocytes, microglia and neurons. Thus, loss function of P2Y1 receptors in all three types of cell may contribute to the phenomenon of impaired synapse elimination in P2ry1−/− mice. First, P2Y1 receptors have been shown to be expressed by astrocytes (Pascual et al., 2012), therefore the finding that P2ry1−/− mice has impaired synapse elimination could be due to a failure to increase calcium levels in astrocytes in response to ATP. This possibility could be excluded by the results of astrocytes in the VPm of P2ry1-/- mice responded to ATP with a similar increase in intracellular Ca2+ when compared to cells in the WT mice (Figure 4—figure supplement 2). Additionally, we found that P2Y1 receptors were not expressed by astrocytes in the VPm (Figure 4—figure supplement 1a), but abundantly expressed in the hippocampus astrocytes (data not shown). Second, P2Y1 receptors have been reported to be expressed in microglia and ATP activates microglia. The phenotype of synapse elimination defects in P2ry1-/- mice could be explained by injection of ATP activates microglia-mediated synapse elimination. At present, there were some mRNA and functional clues showing that microglia expressed P2Y1 receptors, whereas the protein evidence was very few (Fields and Burnstock, 2006). In our study, we did not find any P2Y1 protein expression on microglia in both of hippocampus (data not shown) and the VPm (Figure 4—figure supplement 1b). Therefore, the protein level of P2Y1 receptors expressed on microglia in the VPm, if any, was very low. However, we still could not totally exclude the possibility that microglia may contribute to synapse elimination through P2Y1 receptors. Further studies by using cell-type specific knockout mice are needed to answer this question. Last, ATP-mediates synapse elimination may occur through activation of neuronal P2Y1 receptors. This speculation is most likely due to P2Y1 receptors are abundantly expressed in VPm neurons (Figure 4—figure supplement 1c). In addition, our Ca2+ imaging data showing that ATP-evoked Ca2+ response in VPm neurons was significantly decreased in P2ry1-/- mice. Our findings strongly suggest that ATP may activate neuronal P2Y1 receptors to promote synapse elimination. This hypothesis is supported by a previous study showing that activation of neuronal P2Y1 receptors is required for astrocytes derived ATP mediated LTD in hippocampus (Chen et al., 2013). Due to a close linkage proposed between LTD and synapse elimination (Wiegert and Oertner, 2013), one possible model is that ATP and downstream purinergic signaling could recognize unwanted synapses. Certainly, further studies are needed to elucidate the whole picture of pathways that how activation of purinergic signal mediating synapse elimination.

Materials and methods

Animals

IP3R2-knockout (Itpr2−/−) mice were a kind gift from Dr. Ju Chen (University of California, San Diego) and maintained as a heterozygous line. Heterozygous (Itpr2+/−) mice were interbred to generate homozygous full-mutant mice (Itpr2−/−) and WT littermate controls (Itpr2+/+), which were used in experiments. Cx3cr1GFP/+ mice,hGFAP-GFP mice, and heterozygous P2ry1−/− mice were obtained from Jackson Laboratories.

Slice preparation

Request a detailed protocol

Sagittal slices were obtained as previously described (Wang and Zhang, 2008). Mice and littermate controls were anesthetized with sodium pentobarbital, and then perfused with ice-cold oxygenated slicing solution. The brain was removed rapidly and immersed in ice-cold slicing solution containing (in mM): 110 choline chloride, 7 MgCl2·6H2O, 2.5 KCl, 0.5 CaCl2·H2O, 1.3 NaH2PO4, 25 NaHCO3, 20 glucose, saturated with 95% O2 and 5% CO2. The brain was cut into 300 µm slices on a vibratome (MicromHM650V). Slices were allowed to recover for 40 min at 32°C and then at room temperature for recording in artificial cerebrospinal fluid (aCSF) containing (in mM): 125 NaCl, 2.5 KCl, 2 CaCl2·H2O, 1.3 MgCl2·6H2O, 1.3 NaH2PO4, 25 NaHCO3, 10 glucose. Oxygen was continuously supplied during recovery and recording. To block inhibitory synaptic transmission, picrotoxin (100 µM) was added to the bath.

Electrophysiology

Request a detailed protocol

All recordings were made at room temperature in a submerged recording chamber with constant aCSF perfusion. Patch electrodes had a resistance of 2–4 MΩ when filled with an internal solution containing (in mM): 110 Cs methylsulfate, 20 TEA-Cl, 15 CsCl, 4 ATP-Mg, 0.3 GTP, 0.5 EGTA, 10 HEPES, 4.0 QX-314, and 1.0 spermine (pH adjusted to 7.2 with CsOH, 290–300 mOsm with sucrose). Whole-cell voltage-clamp recordings were made from the soma of VPm neurons with an Axopatch 200B amplifier and Digidata 1322A with pCLAMP 8.1 software (Molecular Devices). Cells were visualized under an upright microscope (BX51WI, Olympus) with infrared optics. Signals were filtered at 2 kHz and digitized at 10 kHz. The series resistance (Rs) was <20 MΩ with no compensation and data were discarded when this varied by ≥ 20%.

To determine the numbers of inputs to a VPm neuron, we recorded evoked EPSCs from the same cells at holding potentials of -70 and +40 mV cell over a wide range of stimulus intensity. A concentric electrode (World Precision Instruments) was placed on the medial lemniscal fiber bundle and stimuli (typically from 20 to 600 µA, 100 µs) were delivered at 0.1 Hz via a Grass S88 stimulator. To distinguish lemniscal synaptic responses from corticothalamic responses, paired-pulse stimulation with an interval of 100 ms was used. First, we used increments of 50–100 µA to search for step numbers. Then we used small increments of 1–10 µA near each transition point to ensure that it was actually a single step. After that, we applied a stronger stimulus, the intensity of which was at least twice than the last step.

Extracellular ATP measurement

Request a detailed protocol

The concentration of extracellular ATP was determined with a bioluminescent ATP assay kit (Sigma), as previously described (Zhang et al., 2007). The ectonucleotidase inhibitor dipyridamole (10 µM) was added to the extracellular solution throughout the experiment to decrease ATP hydrolysis. Luminescence was measured with a luminometer (Varioskan Flash, Thermo) according to the manufacturer’s instructions. A calibration curve was constructed from standard ATP samples and the luminescence of the incubated medium was measured as the background ATP level. To measure the ATP content in acute VPm slices, they were incubated in oxygenated ACSF for 10 min at room temperature. The ACSF was then collected for ATP assay. For quantification, the amount of protein was used for normalization and was determined using the Enhanced BCA protein assay kit (Beyotime, China).

Immunostaining

View detailed protocol

Animals were anesthetized with sodium pentobarbital, and perfused with 0.9% saline followed by ice-cold 4% paraformaldehyde. Brains were removed and post-fixed overnight in 4% paraformaldehyde at 4°C and then transferred to 30% sucrose in PBS at 4°C for 2 days. Sagittal sections 30 or 40 µm thick were cut on a microtome (Leica CM 1850). Sections were dried, washed three times in 0.01M PBS and with 0.3% Triton-100X in 0.1M PBS (40 min) or frozen methanol (10 min at -20°C), then blocked with 10% BSA for 1 hr at room temperature. Sections were incubated with primary antibodies as follows: VGluT2 (guinea-pig polyclonal, 1:2000, Millipore), NeuN (monoclonal mouse, 1:500; Millipore), GFAP (rabbit, 1:500, Chemicon), GFAP (monoclonal mouse, 1:500; Synaptic System), Iba1 (polyclonal rabbit, 1:200; Chemicon), IP3R2 (polyclonal rabbit, 1:200; Santa Cruz), P2Y1 (polyclonal rabbit, 1:200; Abcom) at 4°C for 12–24 hr. Secondary Alexa-conjugated antibodies were added at 1:1000 in 0.1 M PBS for 2 hr at room temperature. Images were captured using an Olympus FV-1200 inverted confocal microscope.

Brain slice Ca2+ imaging

Request a detailed protocol

In brief, for Ca2+ imaging in hippocampus, WT, Itpr2−/− mice, and P2ry1-/- (P13-P16) were sacrificed and coronal slices of hippocampus were cut and incubated in ACSF at room temperature, saturated with 95% O2 and 5% CO2. For Ca2+ imaging in VPm, sagittal slices containing VPm were cut. Slices were incubated with 10 µM Fluo-4/AM (Invitrogen) or Cal-250 (AATBioquest) for 45 min at room temperature in oxygenated ACSF. In these conditions, most of the loaded cells in the stratum radiatum and VPm were astrocytes as confirmed by SR101 staining. CA1, dentate gyrus, and VPm neurons were also loaded in these conditions. Before imaging, slices were transferred to stain-free ACSF for at least 10 min to wash out the stain. Images were acquired every 3 s using an Olympus FV-1200 confocal microscope. Average fluorescence intensity was measured from analysis boxes over the cell-bodies of astrocytes and neurons. When the baseline was stable, ATP (100 µM) was added to evoke intracellular Ca2+ release from astrocytes and neurons. Increases in the fluorescence intensity over baseline were calculated for each trace and are reported as ΔF/F0.

Quantification of VGluT2

Request a detailed protocol

Confocal images were collected with a 60× objective (NA 1.3, water) on an Olympus FV-1200 inverted confocal microscope. Z-stacks were obtained by 10 consecutive steps with 0.5 µm interval thickness. To determine the VGluT2 puncta on the soma of a thalamic VPm neuron, data were opened with Fluoview software (Olympus) and then exported as 8-bit tiff files. Neurons were randomly selected and VGluT2 puncta were counted manually using the Cell Counter plugin of ImageJ. In most cases, the VGlut2 puncta are separated from each other. Sometimes there are multiple puncta that stay very close, but we can still discriminate them by the little gaps among them. Occasionally, there is a big puncta that has no gap, which is then referred as one puncta. We count the puncta that contact the NeuN as puncta/soma. To quantify the average VGluT2 puncta per neuron, the density of VGluT2 puncta was analyzed using the Analyze Particles command in ImageJ (version 1.43). Thalamic VPm neuron number was counted manually using the Cell Counter plugin of ImageJ. The puncta/neuron is then determined by all the puncta number divided by all neuron number in a given image.

Neuron density quantification in the VPm

Request a detailed protocol

For quantification of neuron density, 9 sections from 3 different animals (3 sections per animal) were used. The VPm was acquired by using a 10×objective (NA 0.4). Thalamic VPm neuron numbers were counted manually by using the Cell Counter plugin of Image J. To calculate the density of neurons, only the area of the VPm was measured.

Cannula implantation and pharmacological experiments

Request a detailed protocol

Mice at P10 were deeply anaesthetized with isoflurane and placed into a stereotactic apparatus. A 30-gauge stainless steel infusion cannula (RWD life science) was implanted unilaterally in the left lateral ventricle (-0.4 mm from bregma, 0.9 mm lateral from midline, and 2.5 mm vertical from the cortical surface) and was fixed to the skull with dental cement. After one day recovery, drugs including aCSF, ATP (5 or 50 µM, 2 µl), adenosine (50 µM, 2 µl), ATPγS (50 µM, 2 µl), MRS-2365 (20 µM, 2 µl) were delivered to lateral ventricle slowly by using a microsyringe pump which was connected to the infusion cannula by a PE20 tube. Injections were given twice per day at the interval of 12 hr from P11 to P15. After experiments, the location of the cannula was verified histologically and discarded the animal which did not target to lateral ventricle.

Inflammatory response quantification

Request a detailed protocol

To analysis the inflammatory responses of astrocytes and microglia, fixed sections were immunolabeled for GFAP and Iba1. For quantification the maximum distance of reactive cells to the injury site, three images were collected from each animal on a confocal microscopy by using 1 µm z-step. Astrocytes in the cortex express a low level of GFAP in normal situation. They were very sensitive to injury as indicated by an increased GFAP expression. Therefore, the distance of GFAP positive cells to injury site was measured as the distance of inflammatory responses. For quantification of reactive astrocyte, the number of GFAP positive cells within 150 µm from the injury site was counted. For quantification of microglia activation, the number of Iba1+ microglia within 150 µm from the injury site was counted. Then the percentage of amedoid-like activated microglia was calculated.

Statistical analysis

Request a detailed protocol

All results are presented as mean ± SEM. GraphPad Prism 5 (La Jolla, CA) software was used for all statistical analyses. Differences between two groups were evaluated by un-paired or paired Student’s t-test. For multi-group comparisons, one-way ANOVA with Tukey’s multiple-comparisons test was used. Two-way ANOVA followed by the Bonferroni post hoc test was used for Figure 3A. Frequency distributions were analyzed with χ2 test. The significance level was set at p<0.05.

References

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8
  9. 9
  10. 10
  11. 11
  12. 12
  13. 13
  14. 14
  15. 15
  16. 16
  17. 17
  18. 18
  19. 19
  20. 20
  21. 21
  22. 22
  23. 23
  24. 24
  25. 25
  26. 26
  27. 27
  28. 28
  29. 29
  30. 30
  31. 31
  32. 32
  33. 33
  34. 34
  35. 35
  36. 36
  37. 37
  38. 38
  39. 39
  40. 40
  41. 41
  42. 42
  43. 43
  44. 44
  45. 45
  46. 46
  47. 47
  48. 48
  49. 49

Decision letter

  1. Gary L Westbrook
    Reviewing Editor; Vollum Institute, United States

In the interests of transparency, eLife includes the editorial decision letter and accompanying author responses. A lightly edited version of the letter sent to the authors after peer review is shown, indicating the most substantive concerns; minor comments are not usually included.

Thank you for resubmitting your work entitled "Astrocytes contribute to synapse elimination via type 2 inositol 1,4,5-trisphosphate receptor-dependent release of ATP" for further consideration at eLife. Your article has been favorably evaluated by Gary Westbrook (Senior editor) and two reviewers. This study is interesting and informative as it identifies molecular mechanisms important to synapse refinement. In the revision, the authors have removed the problematic ATP induced LTD model, improved the quality of their images and rewritten their Discussion section. The manuscript is much improved with these changes. There are just a few remaining suggestions to be addressed:

1) It should be made very clear in the figure legend of Figure 1—figure supplement 1 that the co-expression data of IP3R2 was done in the hippocampus, and that the authors could not obtain similar data in VPm because the same antibodies that worked for hippocampus did not work for thalamus.

2) In the third paragraph of the Discussion: The authors argue against the possibility that their synapse elimination findings may arise from altered presynaptic activity because ATP and ATPγS rescue the phenotype. However, they cannot rule out that this rescue occurs at a site between the whisker and VPm.

3) The text needs copyediting. A number of sentences could be reworded to clarify the authors' meaning, including:

A) Results: the phrase "Astrocytes solely express IP3R2" could be interpreted to mean that IP3R2 is all that astrocytes express. Instead the authors mean that astrocytes are the only cell (as opposed to microglia and neurons) that express IP3R2.

B) Results: "If this is the case, one direct evidence": wording here could be clarified, such as "… one direct evidence would be the rescue of the refinement defect with application of the P2Y1 agonist".

C) The following sentence in the Discussion should be clarified as: "Due to a close linkage proposed between LTD and synapse elimination, one possible model is that ATP and downstream purinergic signaling could recognize unwanted synapses."

[Editors’ note: a previous version of this study was rejected after peer review, but the authors submitted for reconsideration. The previous decision letter after peer review is shown below.]

Thank you for submitting your work entitled "Astrocytes contribute to synapse elimination via type 2 inositol 1,4,5-trisphosphate receptor-dependent release of ATP" for consideration by eLife. Your article has been favorably evaluated by Gary Westbrook (Senior Editor) and two reviewers.

Our decision has been reached after consultation between the reviewers. Based on these discussions and the individual reviews below, we regret to inform you that your work will not be considered further for publication in eLife. However, if after considering the reviewer comments, you decide to address the comments with additional experiments or extensive rewriting, we would be willing to consider another submission.

The manuscript was extensively discussed by the reviewers and the editor. As you will see from reading the reviews, there was some difference of opinion at the time the reviews were submitted. However, after discussion both reviewers agreed that the findings that mutations in IP3R2 and P2Y1 lead to defects in developmental connectivity are interesting observations that could be worthy of publication in eLife. The work implicating the purinergic system in synapse elimination was also considered interesting, but the data suggesting that ATP acts by inducing presynaptic LTD was less compelling. We agreed that either additional experiments were required or that the manuscript required extensive rewriting to focus on the conclusions that are best supported by the data. Because it is not clear that this could be accomplished within 2 months, the timeframe eLife considers the maximum for revision, we are returning the manuscript to you.

Reviewer #1:

This manuscript by Yang et al. examines the contribution of IP(3)R2 dependent calcium signaling in astrocytes to developmental synapse elimination at the Pr5-VPm synapse. The authors use a IP(3)R2 knockout mouse (Itpr2–/–) that exhibits disrupted calcium signaling in astrocytes, combined with manipulations of CSF ATP, P2Y1 receptor agonists and P2Y1 KO mouse, to argue for a model where ATP is release from astrocytes through an activity- and IP3R2-dependent pathway. ATP then activates P2Y1 receptors on presumably presynaptic terminals, inducing a reduction of release probability, leading to elimination of presynaptic inputs.

Although the authors do show that both Itpr2–/– and P2Y1–/– mice exhibit abnormal convergence of afferent inputs onto relay neurons, their conclusions with regard to the model are overstated as they fail to consider and rule out alternative explanations including,

1) IP(3) R2 has been shown to be present in neuronal processes in the CNS (Holtzclaw et al., Glia 2002). The authors do not convincingly demonstrate that their manipulation targets only astrocytes and not neurons in the circuit, including relay neurons and afferent inputs onto relay neurons.

2) A previous study by one of the authors showed that sensory deprivation leads to defects in refinement at this synapse. In this study, the authors do not rule out the possibility that disruption of IP(3)R2 or P2Y1 (in astrocytes, glia or neurons) might alter the nature of sensory information transmitted from the whiskers to thalamus. To clearly support the authors' hypothesis, cell-type and region specific KOs are needed.

3) The phenotype of Itpr2–/– mice could be a nonspecific sequela of reduced ATP and global metabolic stress leading to halted or disrupted development.

4) Rescue by ATP in Itpr2–/– mice may be through an entirely different pathway from astrocyte signaling. For example, injection of ATP could activate microglia-mediated synapse elimination. P2RY1 has been reported to be expressed in microglia (Ballerini et al., 2005 Int J Immunopathol Pharmacol.18(2):255-68) and ATP activates microglia (Davalos et al., Nat Neurosci. 2005 Jun;8(6):752-8).

Specific Comments:

1) Figure 1F: More detail is needed on how the authors are counting VGlut2 puncta. For example, in Figures 1E or 2E, how do the authors distinguish between one large puncta from multiple smaller puncta that are so close together that they look like one large object? Are they counting just the puncta that are contacting the NeuN stain? How is puncta/neuron calculated?

2) Figure 1—figure supplement 1: Not clear what region of the brain this is. In addition to somas, there appears to be a lot of processes that are labeled for IP3R2. The low power magnification makes it difficult to assess whether these processes come from astrocytes, glia or neurons.

3) Figure 1—figure supplement 2B: How are neurons distinguished from microglia here?

4) According to the authors’ model, PPRatio of the Pr5-VPm synapse in Itpr2–/– mice should increase, but this was not examined.

5) Figure 3I: distribution of Itpr2–/– + adenosine does not look correct: the sum of the percentages appears to add up to ~55%.

6) Figure 3—figure supplement 1B: difficult to visualize the different cell types. Need higher magnification.

7) Figure 5C: synaptic recordings show slow and distorted currents, suggesting poor voltage clamp.

8) The authors try to correlate increasing ATP levels in VPm with synapse elimination; however, timepoints during the period of active pruning at this synapse (between P7 and P12, Arsenault and Zhang, 2006) are necessary to support their argument.

9) The authors’ model lacks a mechanism of synapse specificity.

10) Figure 5—figure supplement 1: Images are too low in magnification to rule out the possibility that P2Ry receptors are in the distal processes of microglia.

11) Does intraventricular injection of ATP or adenosine affect synaptic refinement in WT mice?

Reviewer #1 (Additional data files and statistical comments):

Higher resolution images that assess the presence of Ip(3)R2 and P2RY1 in cell type specific processes are important supporting data for the authors' model.

Reviewer #2:

The overall finding of this paper, that knocking out somatic calcium signaling in astrocytes leads to defective neuronal synapse elimination in the VPm, is novel and interesting. The authors propose that the synapse elimination defect is due to a decrease in astrocytic ATP release in the absence of astrocyte calcium signaling, and that ATP acts via presynaptic P2Y1 receptors to induce LTD at weak synapses, which leads to their elimination. The series of experiments to determine if calcium-dependent astrocyte release of ATP is mediating synapse elimination by inducing LTD via P2Y1 receptors could each have an alternative explanation by themselves, but taken together make a strong case in support of this hypothesis. For example, the P2Y1 KO phenocopies the IP3R2 KO, in terms of synapse elimination. Injection of a P2Y1 agonist to the IP3R2 KO mouse rescues the synapse elimination phenotype. The authors provide an extensive discussion of alternative explanations for their data, along with control experiments that they did to test these alternatives.

The authors have addressed all of the points that I raised when reviewing a previous version of this manuscript, and added additional data figures (Figures 3 and 6).

https://doi.org/10.7554/eLife.15043.022

Author response

The manuscript is much improved with these changes. There are just a few remaining suggestions to be addressed: 1) It should be made very clear in the figure legend of Figure 1—figure supplement 1 that the co-expression data of IP3R2 was done in the hippocampus, and that the authors could not obtain similar data in VPm because the same antibodies that worked for hippocampus did not work for thalamus. We thank the reviewers for this suggestion. To clarify, we added a sentence in the Figure 1—figure supplement 1 legend to state that the images were obtained from the hippocampus but not the VPm because the same antibodies (Santa cruz, sc-7278; Millipore, AB3000) that worked for hippocampus did not work for VPm. These changes could be found in the legend for Figure 1—figure supplement 1.

2) In the third paragraph of the Discussion: The authors argue against the possibility that their synapse elimination findings may arise from altered presynaptic activity because ATP and ATPγS rescue the phenotype. However, they cannot rule out that this rescue occurs at a site between the whisker and VPm. We agree with the reviewers that the rescue effects by ATP and ATPγS may occur at a site between the whisker and VPm. We re-discuss this issue in the third paragraph of the Discussion.

3) The text needs copyediting. A number of sentences could be reworded to clarify the authors' meaning, including: A) Results: the phrase "Astrocytes solely express IP3R2" could be interpreted to mean that IP3R2 is all that astrocytes express. Instead the authors mean that astrocytes are the only cell (as opposed to microglia and neurons) that express IP3R2.

We thank the reviewers for pointing out this confusing statement. Among three subtypes of IP3 receptors, IP3R2 is the only one that expressed in astrocytes. In addition, IP3R2 is not expressed in microglia and neurons in the cerebrum. To clarify, we modified the sentence in the first paragraph of the Results.

B) Results: "If this is the case, one direct evidence": wording here could be clarified, such as "… one direct evidence would be the rescue of the refinement defect with application of the P2Y1 agonist".

According to the reviewers’ suggestion, we modified the sentence: “If this is the case, one direct evidence would be the rescue of the synapse elimination defect with application of the P2Y1 agonist in Itpr2−/− mice”.

C) The following sentence in the Discussion should be clarified as: "Due to a close linkage proposed between LTD and synapse elimination, one possible model is that ATP and downstream purinergic signaling could recognize unwanted synapses."

We thank the reviewers for this suggestion and have changed the sentence in the last paragraph of the Results.

[Editors’ note: the author responses to the previous round of peer review follow.]

The manuscript was extensively discussed by the reviewers and the editor. As you will see from reading the reviews, there was some difference of opinion at the time the reviews were submitted. However, after discussion both reviewers agreed that the findings that mutations in IP3R2 and P2Y1 lead to defects in developmental connectivity are interesting observations that could be worthy of publication in eLife. The work implicating the purinergic system in synapse elimination was also considered interesting, but the data suggesting that ATP acts by inducing presynaptic LTD was less compelling. We agreed that either additional experiments were required or that the manuscript required extensive rewriting to focus on the conclusions that are best supported by the data. Because it is not clear that this could be accomplished within 2 months, the timeframe eLife considers the maximum for revision, we are returning the manuscript to you. Reviewer #1: This manuscript by Yang et al.

examines the contribution of IP(3)R2 dependent calcium signaling in astrocytes to developmental synapse elimination at the Pr5-VPm synapse. The authors use a IP(3)R2 knockout mouse (Itpr2–/–) that exhibits disrupted calcium signaling in astrocytes, combined with manipulations of CSF ATP, P2Y1 receptor agonists and P2Y1 KO mouse, to argue for a model where ATP is release from astrocytes through an activity- and IP3R2-dependent pathway. ATP then activates P2Y1 receptors on presumably presynaptic terminals, inducing a reduction of release probability, leading to elimination of presynaptic inputs. Although the authors do show that both Itpr2–/– and P2Y1–/– mice exhibit abnormal convergence of afferent inputs onto relay neurons, their conclusions with regard to the model are overstated as they fail to consider and rule out alternative explanations including, 1) IP(3) R2 has been shown to be present in neuronal processes in the CNS (Holtzclaw et al., Glia 2002). The authors do not convincingly demonstrate that their manipulation targets only astrocytes and not neurons in the circuit, including relay neurons and afferent inputs onto relay neurons.

We thank the reviewer for raising this important concern. As the reviewer mentioned, Holtzclaw et al. showed that IP3R2 was expressed not only in astrocytes, but also in neuronal processes in the cerebellum. Also, they found that IP3R2 was expressed in somata of few cells with neuronal profile in hippocampus and cortex, which is controversial from other studies suggested that IP3R2 was not expressed in neurons of these brain regions (Hertle and Yeckel, 2007; Li et al., 2015; Petravicz et al., 2014; Sharp et al., 1999). Our immunostaining results in hippocampus are in line with the opinion that IP3R2 was not expressed in the neurons, as showed in Author response image 1 and Figure 1—figure supplement 1. Unfortunately, we tried different types of IP3R2 antibodies (Santa cruz, sc-7278; Millipore, RB3000) and found no cells could be labeled in the VPm. In this study, we use IP3R2 deficient mice to selectively eliminate Ca2+ signaling in astrocytes but not in neurons, which is further confirmed by calcium imaging experiments as showed in Figure 1—figure supplement 2. Moreover, IP3R2 knockout mice have been extensively used to study the contribution of Ca2+ signaling in astrocytes for synaptic transmission (Martin et al., 2015; Petravicz et al., 2008), synaptic plasticity (Agulhon et al., 2010; Di Castro et al., 2011; Navarrete et al., 2012), cerebral blood flow (Bonder and McCarthy, 2014; Nizar et al., 2013), neuroprotection after brain injury (Hertle and Yeckel, 2007) and depression (Cao et al., 2013). These literatures together indicate that IP3R2-KO mice are well accepted in the field as a mouse model for studying the specific role of Ca2+ signaling in astrocytes.

2) A previous study by one of the authors showed that sensory deprivation leads to defects in refinement at this synapse. In this study, the authors do not rule out the possibility that disruption of IP(3)R2 or P2Y1 (in astrocytes, glia or neurons) might alter the nature of sensory information transmitted from the whiskers to thalamus. To clearly support the authors' hypothesis, cell-type and region specific KOs are needed. We agree with the reviewer that cell-type and region specific KOs’ experiments are very important, and needed to be carried out in the future. In the current study, we found that both of RTP and one selective P2Y1 receptors agonist MRS-2365 could rescue the deficient of synapse elimination in IP3R2-KO mice. These results indicate no matter the nature of sensory information has been changed or not in the IP3R2-KO mice, RTP and its downstream purinergic signaling are sufficient to rescue the defects of synapse elimination in the mutants. We discuss this issue in the third paragraph of the Discussion.

3) The phenotype of Itpr2–/– mice could be a nonspecific sequela of reduced ATP and global metabolic stress leading to halted or disrupted development. We agree with the reviewer that global metabolic stress caused by reduced RTP may be one of the reasons leading to disruption of synapse elimination in Itpr2–/–mice, but this is very unlikely. First, these mutant mice have normal brain cytoarchitecture, neuron and astrocyte numbers, and comparable body weight as well as lifespan with WT mice. Second, we found that impaired synapse elimination in Itpr2–/–mice could be rescued by either RTPçS, a non-hydrolyzable RTP analog, or MRS-2365, a selective P2ry1 agonist. Both molecules could barely provide energy. We discuss this issue in the fourth paragraph of the Discussion.

4) Rescue by ATP in Itpr2–/– mice may be through an entirely different pathway from astrocyte signaling. For example, injection of ATP could activate microglia-mediated synapse elimination. P2RY1 has been reported to be expressed in microglia (Ballerini et al., 2005 Int J Immunopathol Pharmacol.18(2):255-68) and ATP activates microglia (Davalos et al., Nat Neurosci. 2005 Jun;8(6):752-8). We agree with the reviewer that rescue effects by RTP may be through activation of microglia. At present, there are mRNR expressions and functional evidences (pharmacological data) showing that microglia express P2Y1 receptors, but the protein evidence is very few. In our study, we did not observe any P2Y1 receptors expression in VPm microglia (Figure 4—figure supplement 1B). So, the expression level of P2Y1 receptors in microglia of the VPm, if any, may be too low to be detected by antibody. However, we cannot totally exclude the possibility that RTP activates microglia P2RY1 to rescue synapse elimination defects in IP3R2-KO mice. We re-discuss this in the last paragraph of the Discussion.

Specific Comments:

1) Figure 1F: More detail is needed on how the authors are counting VGlut2 puncta. For example, in Figures 1E or 2E, how do the authors distinguish between one large puncta from multiple smaller puncta that are so close together that they look like one large object? Are they counting just the puncta that are contacting the NeuN stain? How is puncta/neuron calculated?

We count the number of VGlut2 puncta manually by using the Cell Counter plugin of ImageJ software. In most cases, the VGlut2 puncta are separated from each other. Sometimes there are multiple puncta that stay very close, but we can still discriminate them by the little gaps among them (indicated by the arrow between the No. 2 and 3, No. 4, 5, 6 in Author response image 2). Occasionally, there is a big puncta that has no gap, which is then referred as one puncta. We count the puncta that contact the NeuN as puncta/soma. The puncta/neuron is determined by all the puncta number divided by all neuron number in a given image. We also described this method in detail in the Materials and methods, subsection “Quantification of VGluT2”.

2) Figure 1—figure supplement 1: Not clear what region of the brain this is. In addition to somas, there appears to be a lot of processes that are labeled for IP3R2. The low power magnification makes it difficult to assess whether these processes come from astrocytes, glia or neurons. Figure 1—figure supplement 1 shows IP3R2 immunostaining from hippocampus CR1 region. We repeated this experiment again and found IP3R2 was almost expressed in astrocytes. We also found that microglia and neuronal processes did not express IP3R2, as showed in the new high power magnification Figure 1—figure supplement 1. Unfortunately, although we tried different types of IP3R2 antibodies including (Santa cruz, sc-f2f8; Millipore, RB3000), there were no cells in the VPm could be labeled.

3) Figure 1—figure supplement 2B: How are neurons distinguished from microglia here? We distinguish neurons from microglia by the big size of cell body.

Neurons normally have bigger cell body than microglia and astrocyte do. In addition, by using CZ3CR1-GFP mice we found that microglia cannot be labeled by calcium dye (Fluo-4 RM and Cal-520).

4) According to the authors’ model, PPRatio of the Pr5-VPm synapse in Itpr2–/– mice should increase, but this was not examined. According to the reviewer’s suggestion, we calculated the PPRatio (100ms interval) and found that there was no significant difference between WT and Itpr2–/– mice. This could be explained by a compensatory mechanism occurred during the embryonic knocking out of IP3R2. In addition, we agree with the editors’ opinion and remove our model of RTP induced LTD causing synapse elimination. We also extensively rewrote the Discussion.

5) Figure 3I: distribution of Itpr2–/– + adenosine does not look correct: the sum of the percentages appears to add up to ~55%. We thank the reviewer for pointing out this and have corrected the mistake in our revised manuscript.

6) Figure 3—figure supplement 1B: difficult to visualize the different cell types. Need higher magnification. We thank the reviewer for this suggestion. A higher magnification image is provided in the revised manuscript.

7) Figure 5C: synaptic recordings show slow and distorted currents, suggesting poor voltage clamp. We replaced the original sample traces with new one in our revised manuscript.

8) The authors try to correlate increasing ATP levels in VPm with synapse elimination; however, timepoints during the period of active pruning at this synapse (between P7 and P12, Arsenault and Zhang, 2006) are necessary to support their argument. We measured the RTP levels at P12-P13 in IP3R2-KO mice and found no significant difference when compared to P7-P8. This is not surprising because pruning of VPm relay synapses has different stages. An early stage from P7 to P12 which is insensitive whisker deprivation and a late stage from P12 to P16 which is largely rely on sensory experience. Thus, RTP dependent synapse elimination may only contribute to the late stage of synapse elimination.

9) The authors’ model lacks a mechanism of synapse specificity. We agree with the reviewer that our previous model lacks a

mechanism of synapse specificity. According to the editor’s suggestion, we removed our model of RTP induced LTD causing synapse elimination. We also extensively rewrote the Discussion and gave a speculation of how RTP and purinergic signaling recognized unwanted synapses (Discussion, last paragraph).

10) Figure 5—figure supplement 1: Images are too low in magnification to rule out the possibility that P2Ry receptors are in the distal processes of microglia. We thank the reviewer for this suggestion. In the revised manuscript, we replace the images with higher magnification ones, as showed in Figure 4—figure supplement 1. We did not observe P2Y1 receptors expressed in the distal processes of microglia.

11) Does intraventricular injection of ATP or adenosine affect synaptic refinement in WT mice?

We thank the reviewer for this suggestion. We preformed new experiments with injection of RTP or adenosine in WT mice from P11 to P15 and found no effects on synapse elimination, as showed in Author response image 5.

https://doi.org/10.7554/eLife.15043.023

Article and author information

Author details

  1. Junhua Yang

    Department of Neurobiology, Key Laboratory of Medical Neurobiology of Ministry of Health of China, Key Laboratory of Neurobiology, Zhejiang University School of Medicine, Hangzhou, China
    Contribution
    JY, Drafting the article, Final approval of the version to be published, Conception and design, Acquisition of data, Analysis and interpretation of data
    Contributed equally with
    Hongbin Yang
    Competing interests
    The authors declare that no competing interests exist.
  2. Hongbin Yang

    Department of Neurobiology, Key Laboratory of Medical Neurobiology of Ministry of Health of China, Key Laboratory of Neurobiology, Zhejiang University School of Medicine, Hangzhou, China
    Contribution
    HY, Drafting the article, Final approval of the version to be published, Conception and design, Acquisition of data, Analysis and interpretation of data
    Contributed equally with
    Junhua Yang
    Competing interests
    The authors declare that no competing interests exist.
  3. Yali Liu

    Department of Neurobiology, Key Laboratory of Medical Neurobiology of Ministry of Health of China, Key Laboratory of Neurobiology, Zhejiang University School of Medicine, Hangzhou, China
    Contribution
    YL, Discussing the revision of the article, Final approval of the version to be published, Acquisition of data, Analysis and interpretation of data
    Competing interests
    The authors declare that no competing interests exist.
  4. Xia Li

    Department of Neurobiology, Key Laboratory of Medical Neurobiology of Ministry of Health of China, Key Laboratory of Neurobiology, Zhejiang University School of Medicine, Hangzhou, China
    Contribution
    XL, Discussing the revision of the article, Final approval of the version to be published, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.
  5. Liming Qin

    Department of Neurobiology, Key Laboratory of Medical Neurobiology of Ministry of Health of China, Key Laboratory of Neurobiology, Zhejiang University School of Medicine, Hangzhou, China
    Contribution
    LQ, Discussing the revision of the article, Final approval of the version to be published, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.
  6. Huifang Lou

    Department of Neurobiology, Key Laboratory of Medical Neurobiology of Ministry of Health of China, Key Laboratory of Neurobiology, Zhejiang University School of Medicine, Hangzhou, China
    Contribution
    HL, Discussing the revision of the article, Final approval of the version to be published, Contributed unpublished essential data or reagents
    Competing interests
    The authors declare that no competing interests exist.
  7. Shumin Duan

    Department of Neurobiology, Key Laboratory of Medical Neurobiology of Ministry of Health of China, Key Laboratory of Neurobiology, Zhejiang University School of Medicine, Hangzhou, China
    Contribution
    SD, Discussing the revision of the article, Final approval of the version to be published, Conception and design
    For correspondence
    duanshumin@zju.edu.cn
    Competing interests
    The authors declare that no competing interests exist.
  8. Hao Wang

    Department of Neurobiology, Key Laboratory of Medical Neurobiology of Ministry of Health of China, Key Laboratory of Neurobiology, Zhejiang University School of Medicine, Hangzhou, China
    Contribution
    HW, Conception and design, Analysis and interpretation of data, Drafting or revising the article
    For correspondence
    haowang@zju.edu.cn
    Competing interests
    The authors declare that no competing interests exist.
    ORCID icon "This ORCID iD identifies the author of this article:" 0000-0003-1862-8229

Funding

National Natural Science Foundation of China (31471022)

  • Shumin Duan
  • Hao Wang

Major State Basic Research Program of China (2011CB504400)

  • Shumin Duan
  • Hao Wang

Major State Basic Research Program of China (2013CB945600)

  • Shumin Duan
  • Hao Wang

Major State Basic Research Program of China (2015CB755600)

  • Shumin Duan
  • Hao Wang

Major State Basic Research Program of China (2015AA020515)

  • Shumin Duan
  • Hao Wang

The Program for Introducing Talents in Discipline to Universities

  • Shumin Duan

The Fundamental Research Funds for the Central Universities (2014FZA7007)

  • Shumin Duan

Zhejiang Provincial Natural Science Foundation of China (Y2110057)

  • Shumin Duan
  • Hao Wang

National Natural Science Foundation of China (81221003)

  • Shumin Duan
  • Hao Wang

National Natural Science Foundation of China (91232000)

  • Shumin Duan
  • Hao Wang

National Natural Science Foundation of China (31490590)

  • Shumin Duan
  • Hao Wang

The funders had no role in study design, data collection and interpretation, or the decision to submit the work for publication.

Acknowledgements

We thank Dr. Zhong-wei Zhang, Dr. Da-ting Lin, Dr. Xiang-yao Li, Dr. Jia-dong Chen, and Dr. IC Bruce for critical comments on this manuscript. This work was supported by grants from the Major State Basic Research Program of China (2011CB504400, 2013CB945600, 2015CB755600, 2015AA020515), the National Natural Science Foundation of China (31471022, 81221003, 91232000, 31490590), the Program for Introducing Talents in Discipline to Universities, the Zhejiang Provincial Natural Science Foundation of China (Y2110057), and the Fundamental Research Funds for the Central Universities (2014FZA7007).

Ethics

Animal experimentation: Animal experiments were conducted in accordance with the Guidelines for the Care and Use of Laboratory Animals of Zhejiang University, and were approved by Committee of Laboratory Animal Center of Zhejiang University (ZJU201501402). All surgery was performed under sodium pentobarbital anesthesia, and every effort was made to minimize suffering.

Reviewing Editor

  1. Gary L Westbrook, Vollum Institute, United States

Publication history

  1. Received: February 7, 2016
  2. Accepted: March 15, 2016
  3. Version of Record published: April 12, 2016 (version 1)

Copyright

© 2016, Yang et al.

This article is distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use and redistribution provided that the original author and source are credited.

Metrics

  • 2,719
    Page views
  • 682
    Downloads
  • 36
    Citations

Article citation count generated by polling the highest count across the following sources: Scopus, Crossref, PubMed Central.

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)

Further reading

    1. Ecology
    2. Neuroscience
    Felix JH Hol et al.
    Tools and Resources

    Female mosquitoes need a blood meal to reproduce, and in obtaining this essential nutrient they transmit deadly pathogens. Although crucial for the spread of mosquito-borne diseases, blood feeding remains poorly understood due to technological limitations. Indeed, studies often expose human subjects to assess biting behavior. Here, we present the biteOscope, a device that attracts mosquitoes to a host mimic which they bite to obtain an artificial blood meal. The host mimic is transparent, allowing high-resolution imaging of the feeding mosquito. Using machine learning we extract detailed behavioral statistics describing the locomotion, pose, biting, and feeding dynamics of Aedes aegypti, Aedes albopictus, Anopheles stephensi, and Anopheles coluzzii. In addition to characterizing behavioral patterns, we discover that the common insect repellent DEET repels Anopheles coluzzii upon contact with their legs. The biteOscope provides a new perspective on mosquito blood feeding, enabling the high-throughput quantitative characterization of this lethal behavior.

    1. Neuroscience
    Claudia Scheckel et al.
    Research Article

    Prion diseases are caused by PrPSc, a self-replicating pathologically misfolded protein that exerts toxicity predominantly in the brain. The administration of PrPSc causes a robust, reproducible and specific disease manifestation. Here we have applied a combination of translating ribosome affinity purification and ribosome profiling to identify biologically relevant prion-induced changes during disease progression in a cell-type specific and genome-wide manner. Terminally diseased mice with severe neurological symptoms showed extensive alterations in astrocytes and microglia. Surprisingly, we detected only minor changes in the translational profiles of neurons. Prion-induced alterations in glia overlapped with those identified in other neurodegenerative diseases, suggesting that similar events occur in a broad spectrum of pathologies. Our results suggest that aberrant translation within glia may suffice to cause severe neurological symptoms and may even be the primary driver of prion disease.