Astrocytes detect and upregulate transmission at inhibitory synapses of somatostatin interneurons onto pyramidal cells

Astrocytes are important regulators of excitatory synaptic networks. However, astrocytes regulation of inhibitory synaptic systems remains ill defined. This is particularly relevant since GABAergic interneurons regulate the activity of excitatory cells and shape network function. To address this issue, we combined optogenetics and pharmacological approaches, two-photon confocal imaging and whole-cell recordings to specifically activate hippocampal somatostatin or paravalbumin-expressing interneurons (SOM-INs or PV-INs), while monitoring inhibitory synaptic currents in pyramidal cells and Ca2+ responses in astrocytes. We found that astrocytes detect SOM-IN synaptic activity via GABABR and GAT-3-dependent Ca2+ signaling mechanisms, the latter triggering the release of ATP. In turn, ATP is converted into adenosine, activating A1Rs and upregulating SOM-IN synaptic inhibition of pyramidal cells, but not PV-IN inhibition. Our findings uncover functional interactions between a specific subpopulation of interneurons, astrocytes and pyramidal cells, involved in positive feedback autoregulation of dendritic inhibition of pyramidal cells.

extracellular space via hemichannels. The authors should make sure that BAPTA does not leak to the extracellular space. If it did, the high concentration of BAPTA (20 mM) would have a serious impact on extracellular Ca2+ and hence synaptic transmission. (The increased IPSC amplitude in Figs 6j-l may argue against BAPTA leak. However, most of the IPSCs are of perisomatic origin which may be out of reach of BAPTA diffusion and might reflect some homeostatic IPSC changes.) Suppl. Fig. 3: The correctness of GAT-3, GABAB-R, and A1R IHC needs to be demonstrated (or cite the literature that uses the respective antibodies; GAT-3 and GABAB-R should not be abundant in the liver and lung. ).
Minor points Scales in micrographs µm:: µM Lines 268-270 & 297-299: These statements must be supported in a quantitative manner. It seems that any axons in the hippocampus is close to GAT-3 given the astrocytic coverage of the brain and the frequent punctate distribution of GAT-3. More detailed analyses are needed for Ca2+ imaging of astrocytes. Although "optogenetic stimulation of SOM-IN-induced Ca2+ transients in all astrocytic processes analysed (n=24)", the current manuscript does not clearly described how ROIs were selected. How similar/different are opto-SOM-IN-induced Ca2+ transients to/from spontaneously-observed ones? How does the Ca2+ length change with stimulus strength? How does the optogenetic excitation of SOM-INs compare with known firing rates of SOM INs in literature? A spatial analysis of astrocytic Ca2+ elevation in relation to SOM-INs axon terminals should also be performed. Shigetomi et al.'s (2012;doi:10.1038/nn.3000) finding that GAT-3 surface expression can be rapidly modulated by astrocytic BAPTA dialysis could be mentioned.

Reviewer #2 (Remarks to the Author):
In this study the authors focus on somatostatin expressing interneurons (SOM-IN) and the ability of an astrocytic source of adenosine to modulate this pathway. The experiments that are performed utilize the selective expression of channelrhodopsin in SOM IN to permit selective activation of this specific sub-class of interneuron. The authors report that astrocytes detect SOM-IN activity through a combination of GABAbRs and GAT-3 dependent Ca2+ signaling mechanisms and that this leads to triggering the release of ATP which is hydrolysed to adenosine which in turn causes an A1R upregulation of SOM-IN IPSCs. This study is extremely interesting and provides novel insights into interactions between a specific sub class of interneuron and astrocytes. However, the studies are too preliminary in terms of understanding the GAT-3 mechanism as well as superficial in terms of target identification. Additional studies, including the use of A1R, CD73 and GAT-3 KO mice, are required to fully examine the proposed pathway.
P7: Are the effects of AMP-CP prevented by the A1R antagonist DPCPX as would be predicted? Are the effects of adenosine on the SOM IPSCs mediated pre or postsynaptically?
P8: Important experiments are performed using BAPTA to prevent changes in astrocytic Ca2+. The effects of 01 versus 20mM BAPTA are compared. Were these solutions isosmotic given the great sensitivity of astrocytes to small osmotic changes? P9: GAT-3 immunoreactivity is presented which is appropriate. It is also important to examine existing databases of cell specific transcriptome studies to confirm that those studies demonstrate expression of the mRNA, and to cite accordingly.
On page 9 the authors introduce the idea of GABA transporters by pharmacologically interfering with GAT-3. This is important work. Presumably GAT-3 was focused on because of its enrichment in astrocytes. However, the work would be augmented by i) using antagonists to other GABA transporters and determining whether effects are specific to GAT-3, and ii) by using GAT-3 KO mice to confirm on target activity of the GAT-3 antagonist. These additional studies are appropriate both in the Ca2+ and synaptic modulation components of the ms. P10 last line: A lack of statistical difference is concluded but only an N=4 is used. This is often an insufficient N number to observe statistically significant differences which emerge as larger sample sizes are examined. Please perform additional studies to determine whether there is indeed no difference. With further n values I anticipate statistical significance will be reached which would significantly change some of the conclusions. P11: Studies are consistent with the GAT-3 being important in controlling the A1R modulation of SOM-IN inhibition of pyramidal neurons. However, all A1R evidence in the ms is weak, relying on a single does of the antagonist DPCPX. More effort should be taken into clearly establishing the role of A1R. This evidence should include the use of A1R agonists, as well as A1R knockout mice.
P12: Immuno co-localization is presented to show a correlation between the localization of GAT-3 and A1Rs. However, for this evidence to be more compeling quantitative approaches are necessary.
P13: AMP-CP is used to examine the role of CD73 in mediating the hydrolysis of ATP to adenosine. This should be extended by using CD73 knockout mice.
The mechanism of GAT3 inhibition leading to alterations in Ca2+ signals is speculative at best and further studies are required to understand mechanism. Figure 2d -Does an A1R agonist rescue the BAPTA induced reduction of the IPSC? Similarly, one would predict rescue by ATP and that these effects would be prevented by A1R antagonists and in A1R KO mice.
Figure 4g -Does A1R agonist rescue the IPSC magnitude in the presence of SNAP and BAPTA? Is the effect of A1R agonist prevented in A1R KO mice and by DPCPX?

Reviewer #3 (Remarks to the Author):
The manuscript by Matos et al. describes a form of feedback modulation of the inhibitory events evoked by CA1 PC dendrite-innervating SOM+ IN through activation of astrocytes. The feedback is completed by the activation of A1Rs by adenosine following the conversion of ATP released by the astrocytes. The authors also claim that this effect is specific to SOM+ IN. There are two major flaws with the claims: 1) The comparison of the drug effects onto sIPSCs and those evoked by optogenetic stimulation of SOM+ IN is equivocal. Most of the sIPSCs recorded in the somata of CA1 PCs originate not from the activity of IN innervationg distant dendrites, but from the somatargeting IN (e.g., PV+ and CCK+ basket cells). A fair comparison would have been to express ChR2 in other dendrite-targeting IN (e.g., PV+ bistratified cells, or other IN types), and to carry out a side-by-side comparison of the pharmacological effects on the two (or more) types of optogenetically evoked IPSCs. Such experiments would substantiate the claim that the modulation is unique to the SOM+ IN, which would constitute a significant finding. Otherwise the modulation of IPSCs by gliotransmitters or glial intermediaries is by itself not novel.
2) The experiments appear to be seriously underpowered. The n's range from 4-8 with a very few exceptions. No attempt has been made to show that the data are normally distributed (I doubt that this can be shown on datasets with an n of 4). Moreover, the ttests are repeateadly used on normalized data, which by definition reduces the variance. At least, repeated measures ANOVA statistics should be performed when there are repeated measurements in the data sets, such as control->drug->wash.
Reviewer #1: In this manuscript, Matos and colleagues describe that the amplitude of SOM-IN to PYR inhibitory post-synaptic currents is enhanced by A1Rs in the hippocampal CA1 stratum radiatum. Using mouse hippocampal slices, they showed that GABA released by SOM-INs elicits Ca2+ elevations in astrocytes via the GAT-3 GABA transporter. The elevated Ca2+ triggers the release of ATP from astrocytic processes. The ATP in the extracellular space is converted to adenosine which in turn activates A1Rs. While the finding is potentially interesting and novel in that synaptic transmission of a specific class of interneuron is modulated by astrocytes, the dynamical it remains unclear how this circuit dynamically operates and contributes to the function of the hippocampus.
1-For instance, is the SOM-IN to Pyr IPSC enhancement tonic or phasic? If latter, what is the time course of enhancement/decay since astrocytic Ca2+ elevation? How does it contribute/influenced to/by synaptic plasticity or hippocampal LFP events (ripples, sharp waves, theta, gamma, etc.)?
We share the reviewer's interest for these unresolved questions. In the present study, we aimed first at establishing and deciphering how astrocytes could interact with interneurons to regulate inhibition of excitatory cells. and shape the network function in the hippocampus. Investigating how, as a consequence, these mechanisms shape the network function in the hippocampus is important, but it represents a substantial amount of work that is beyond the scope of the present paper. Nevertheless we added a comment in the discussion section, line 492-496 to address these points.
"At another level of regulation, astrocytes are able to influence rhythmic firing of neurons (Morquette & Tan), therefore it would also be relevant to investigate the relationship between astrocyte and inhibitory synapses in the modulation of rhythmic brain activities that are important for hippocampal functions." 2-In addition, there are a few points that need to be resolved to assure that the authors' proposal.
It remains unclear whether SOM-INs-driven Ca2+ elevations in astrocytes enhance SOM-INs-evoked IPSCs in pyramidal neurons, because BAPTA dialysis or GAT-3 inhibition reduces Ca2+ activities in wide-spread astrocytes. As it stands, endogenous activities of other GABAergic cell activity could also activate astrocytes through GAT-3 and/or GABA-BRs.
To address this point, we performed additional experiments with PV-ChR2/EYFP transgenic mice and found that bath application of the GAT-3 specific antagonist (S)-SNAP-5114 (10 µM) did not affect the amplitude of PV-IPSCs in pyramidal cells (95.3 ± 9.1% of control, n = 6, p > 0.05, Friedman test; Fig. 6e, f). These results show that inhibition by at least another major GABAergic cell type (PV-IN) in the hippocampus is not modulated through GAT-3 activation of astrocytes. We have also modified discussion section accordingly. In our conditions (i.e. with AP5 and NBQX), the spontaneous firing rate at resting membrane potential of SOM-IN is around 3-7 Hz as described in Chittajallu R et al., 2013 andAmilhon B et al., 2015. As pointed out by the reviewer, at present it is not known if SOM-IN activity sets the extracellular adenosine tone or if it is influenced solely by spontaneous astrocytic Ca 2+ activity, or a combination of both.
The spontaneous astrocytic Ca 2+ activities in the stratum radiatum of the hippocampus without AP5 and NBQX has been shown to occur approximately in 49 % of the astrocytic network with a mean frequency of 0.49 event/min and in 57.9 % of the microdomains with a mean frequency of 0.59 event/min as described by Nakayama R et al., 2016, Rungta RL et al., 2016and Bosson A et al., 2017. Spontaneous astrocytic Ca 2+ activities in the stratum radiatum are mainly mediated by influx of extracellular Ca 2+ and internal store opening rather than AMPA-R and NMDA-R. As a consequence, in presence of AP5 and NBQX, these parameters should reflect astrocytes Ca 2+ activity.
In addition, each astrocyte could contact hundreds to thousands of synapses, both excitatory and inhibitory. This represents as much different inputs that could simultaneously modulate astrocyte Ca 2+ activities at the network and microdomains levels. At present we feel that determining how these particular spontaneous Ca 2+ activities contribute to the SOM-IN synapse regulation would require lots of experimentation and is beyond the scope of the present paper.
Thus given the uncertainty about the mechanisms responsible for extracellular adenosine tone, we preferred to refrain from discussing speculations.
5-The idea of distinct regulation of dendritic SOM-INs-evoked IPSCs relies on the comparison between SOM-INs-evoked IPSCs and spontaneous IPSCs which reflect perisomatic inhibition. Recording of IPSCs evoked by optogenitic activation of perisomatically targeting interneurons enables a fair comparison, which makes this interesting idea more acceptable.
We agree with the reviewer and to address this question we generated PV-ChR2 mice to selectively activate paravalbumin expressing interneurons (PV-INs), known to target the perisomatic area of pyramidal cells (Kenneth A et al., 2017). We tested if PV-IN inhibition of pyramidal cells was sensitive to endogenous A1R activation or GAT-3 activity and found it was not. We have modified the manuscript in the light of these new experiments: Results section, line 302-324

"Endogenous activation of A1R and astrocytic GAT-3 do not regulate inhibition of pyramidal cells by PV-INs.
We Fig. 1d) that were GABAAR-mediated (Extended Data Fig. 1f).

next examined if A1R-and GAT-3-mediated astrocytic modulation of synaptic inhibition of pyramidal cells is specific to inhibition by SOM-INs or also regulate inhibition by other interneurons types. We targeted ChR2 expression to another major type of interneurons, the parvalbumin-expressing interneurons (PV-INs) and recorded IPSCs evoked in CA1 pyramidal cells of PV-ChR2/EYFP transgenic mice by optogenetic stimulation (Fig. 6a-b). Optogenetic stimulation of PV-INs with light pulses of different duration (0.4 to 1 ms; 0.1 Hz) evoked IPSCs (PV-IPSCs) of increasing amplitude in pyramidal cells (Extended Data
Next we used a similar pharmacological approach as previously to determine if endogenous activation of A1Rs regulates inhibition of pyramidal cells by PV-INs. Bath application of the A1R antagonist DPCPX (100 nM) failed to affect the amplitude of PV-IPSCs (99.9 ± 6.0% of control, n = 7, p > 0.05, Fig. 6d, f). These results indicate that PV-IPSCs in pyramidal cells are not subject to regulation by endogenous adenosine activation of A1Rs. We next assessed if astrocytic GAT-3 activation regulates inhibition of pyramidal cells by PV-INs using the GAT-3 specific antagonist (S)-SNAP-5114 as previously. Bath application (S)-SNAP-5114 (10 µM) did not change the amplitude of PV-IPSCs (95.3 ± 9.1% of control, n = 6, p > 0.05, Fig. 6e, f)

. These results indicate that inhibition of pyramidal cells by PV-INs is unaffected by the blockade of either GAT-3 or A1R suggesting that A1R-and GAT-3-mediated astrocytic regulation of synaptic inhibition of pyramidal cells may be specific to inhibition by SOM-INs."
We have also added a new figure (Fig 6) to illustrate these data. 6-The mechanism for the A1R-dependent IPSC enhancement is not clear. How does the author conclude the role of post-synaptic A1R while (s)-SNAP-5114 would block astrocytic, pre-, and post-synaptic A1Rs. Did the authors measure the paired pulse ratio for SOM-IN -PYR synaptic transmission?
We agree with the reviewer that we did not address experimentally this point in the present manuscript. We did not assess changes in paired pulse ratio of IPSCs because we found an absence of paired pulse modulation with paired pulse optogenetic stimulation (IPSC2 = IPSC1). We acknowledged in the discussion (line 510-515 & 520-523) that we could not distinguish between pre-and post-synaptic mechanisms, but we acknowledge previous work of others (references 58 59 60 62) that have shown a postsynaptic mechanism involving inhibition of postsynaptic adenylate cyclase, reduced PKA activity and increased postsynaptic GABA A R function. 7-One of the growing concerns in the community is the BAPTA-leakage problem: If BAPTA can spread to gap junction-connected neighbouring astrocytes, it could as well leak to the extracellular space via hemichannels. The authors should make sure that BAPTA does not leak to the extracellular space. If it did, the high concentration of BAPTA (20 mM) would have a serious impact on extracellular Ca2+ and hence synaptic transmission. (The increased IPSC amplitude in Figs 6j-l may argue against BAPTA leak. However, most of the IPSCs are of perisomatic origin which may be out of reach of BAPTA diffusion and might reflect some homeostatic IPSC changes.) As suggested by the reviewer, the BAPTA-leakage is important to take into account. As a consequence we and others previously carried out experiments with a pipette containing BAPTA in the extracellular space to rule out potential effects of BAPTA leakage (Serrano A et al., 2006, Panatier A et al., 2011, Martin-Fernandez M, et al., 2017. Thus as previously shown, in our experimental conditions the effects of BAPTA dialysis in astrocytes are unlikely to be due to BAPTA-leakage. 8-Suppl. Fig. 3: The correctness of GAT-3, GABAB-R, and A1R IHC needs to be demonstrated (or cite the literature that uses the respective antibodies; GAT-3 and GABAB-R should not be abundant in the liver and lung).
Due to Nature Communications guidelines and references limitations, we did not cite the original controls. Nevertheless, in the Methods section we provide catalog numbers and suppliers, to which readers may refer for controls for all the antibodies used in the manuscript.
These statements refer to the immunohistochemical data that we obtained and that provides qualitative evidence of co-localization at the light microscope level to support "putative mechanism involving GAT-3-dependent activation of a Ca 2+ -dependent astrocytic purinergic regulation of SOM-IN inhibitory synapses on pyramidal cell dendrites". A detailed quantification of GAT-3, GFAP, S100B, SOM-IN-YFP and A 1 R co-localization is not necessary to make our points. In addition such quantification would require extensive analysis (deconvolution methods, super or pseudo super-resolution methods such as airy-scan or even electron microscopy studies) and are beyond the scope of the present paper.

11-Fig 3d: scale bar is missing. (assumed to be the same as 3j)
The scale bar is the same for 3d & e, but only present on 3e to avoid redundancy, 12- Fig. 3k, the authors should try 2-way ANOVA to assess the interactive effects of SNAP and CGP. [Optional: likewise, would combined application of SNAP and CGP further attenuate SOM-IN -PYR IPSC?] As the drugs SNAP and CGP are considered to be the same factor, one-way ANOVA followed by Tukey's multiple comparison post hoc test is relevant. To this end, the interactive effect has been assessed by the combination of SNAP + CGP, which is significantly different (higher reduction) from SNAP and CGP alone. These questions about Ca 2+ transients were answered in point 4 above. In addition a description of how ROIs were selected was added in Methods (line 644).

"Fluorescence intensity was determined in individual astrocytes by measuring the average pixel values in defined in 2-3 circular regions of interest (ROIs -2 µm diameter) placed over random proximal astrocytic processes (1-2 processes per astrocyte) and subtracted to a control extracellular background ROI."
Finally, related to point 10 above, a quantitative analysis of the spatial relationship between astrocyte Ca 2+ elevation and SOM-IN axon terminals would require exhaustive anatomical experimentation which is beyond the scope of the present paper.
This was already acknowledged in the original manuscript, but to be more explicit the sentence was modified in discussion section, line 421 "This may be an auto-regulated mechanism since astrocytic Ca2+ signals can conversely modulate GAT-3 activity and expression 24 " Reviewer #2: In this study the authors focus on somatostatin expressing interneurons (SOM-IN) and the ability of an astrocytic source of adenosine to modulate this pathway. The experiments that are performed utilize the selective expression of channelrhodopsin in SOM IN to permit selective activation of this specific sub-class of interneuron. The authors report that astrocytes detect SOM-IN activity through a combination of GABAbRs and GAT-3 dependent Ca2+ signaling mechanisms and that this leads to triggering the release of ATP which is hydrolysed to adenosine which in turn causes an A1R upregulation of SOM-IN IPSCs.
This study is extremely interesting and provides novel insights into interactions between a specific sub class of interneuron and astrocytes. However, the studies are too preliminary in terms of understanding the GAT-3 mechanism as well as superficial in terms of target identification. Additional studies, including the use of A1R, CD73 and GAT-3 KO mice, are required to fully examine the proposed pathway.
1-P7: Are the effects of AMP-CP prevented by the A1R antagonist DPCPX as would be predicted? Are the effects of adenosine on the SOM IPSCs mediated pre or postsynaptically?
As suggested we performed new experiment to test if DPCPX actions occlude the effect of AMP-CP. As predicted, AMP-CP failed to produce an effect in the presence of DPCPX (66.5 ± 5.9 % of control, n = 6, p< 0.01 relative to control, p> 0.05 relative to DPCPX, ANOVA; Fig. 1f, i). Figure 1 was modified to illustrate these results (panel f & i).
About pre-or postsynaptic adenosine mechanisms, this is the same point as reviewer 1 point 6. We agree with the reviewer that we did not address this point experimentally in the present manuscript. We acknowledged in the discussion (line 510-511 & 521-524) that we could not distinguish between pre-and post-synaptic mechanisms, but previous work of others (ref 60) have shown a postsynaptic mechanism involving inhibition of postsynaptic adenylate cyclase, reduced PKA activity and increased postsynaptic GABA A R function 2-P8: Important experiments are performed using BAPTA to prevent changes in astrocytic Ca2+. The effects of 01 versus 20mM BAPTA are compared. Were these solutions isosmotic given the great sensitivity of astrocytes to small osmotic changes?

mM, Sigma/Aldrich) the concentration of KMeSO 4 was adjusted to maintain the concentration of potassium ions 11 ."
In addition, we know from our previous study (reference 11) that these different concentrations do not alter the astrocytic properties. "Finally, all criteria used to monitor astrocyte properties revealed that the cells were healthy and their basic properties were unaltered by BAPTA. These include membrane resistance, the stability of the whole-cell recordings and fine morphological structures such as the shape and size of the compartments that were monitored throughout the entire length of the experiments. " from Panatier et al., 3-P9: GAT-3 immunoreactivity is presented which is appropriate. It is also important to examine existing databases of cell specific transcriptome studies to confirm that those studies demonstrate expression of the mRNA, and to cite accordingly.
In agreement with the reviewer comment, we added the citation of the recently published paper by Ghirardini E, et al., 2018. In this paper, the authors showed functional expression of GAT-3 in hippocampal astrocytes by using whole-cell patch-clamp and single-cell reverse transcritption-PCR. The reference has been added in Results section line 217, Discussion section line 406. 4-On page 9 the authors introduce the idea of GABA transporters by pharmacologically interfering with GAT-3. This is important work. Presumably GAT-3 was focused on because of its enrichment in astrocytes. However, the work would be augmented by i) using antagonists to other GABA transporters and determining whether effects are specific to GAT-3, We chose to focus on GAT-3 because it's the only GABA transporter known to be specific to astrocytes (whereas GAT-1 or GlyTs are also expressed in neurons). Due to this specificity of expression we also take advantage of using (S)-SNAP-5114 which is a competitive GAT-3 specific antagonist, allowing us to target specifically and reversibly astrocytes, That in itself is quite unique in the glial field in terms of specificity. Even if GAT-1 and GAT-3 could be involved in the regulation of extracellular level of GABA, only GAT-3 has been implicated by other groups in Ca 2+ signaling in astrocytes as stated in Discussion section, line 409 to 414.
Thus we feel that using antagonists of other transporters would not alter our conclusion that the astrocyte specific GABA transporter (GAT-3) is involved in the detection and upregulation of inhibitory transmission by somatostatin interneurons onto pyramidal cells and that these additional experiments are not necessary. 5-and ii) by using GAT-3 KO mice to confirm on target activity of the GAT-3 antagonist. These additional studies are appropriate both in the Ca2+ and synaptic modulation components of the ms.
We are looking in this study at a hippocampal specific acute effect in mature brain, and so take advantage of the pharmacological selectivity of (S)-SNAP-5114 to GAT-3. A Cre-lox strategy for GAT-3 KO is impossible since we already use SOM Cre mice to target ChR2. Also, the use of a general KO targeting GAT-3 would impact the whole brain, both during and after development. Moreover the generation and characterization of these mice would take a substantial amount of time beyond the scope of the present paper.
However : 1) we carried out additional experiments with another mouse line to express ChR2 in a parvalbumin expressing interneurons to show that the GAT-3-mediated modulation is not present at parvalbumin interneuron synapses; 2) we performed new pharmacological experiments showing that application of an A1R agonist reverses the modulation of IPSCs produced by GAT3 inhibition; and 3) we carried out other additional experiments to increase N numbers in instances requested by the reviewers (eg. point 6 below).
6-P10 last line: A lack of statistical difference is concluded but only an N=4 is used. This is often an insufficient N number to observe statistically significant differences which emerge as larger sample sizes are examined. Please perform additional studies to determine whether there is indeed no difference. With further n values I anticipate statistical significance will be reached which would significantly change some of the conclusions.
As requested, we performed additional experiments and our conclusions remained unchanged. With n = 7 for the vehicle and CGP55845 experiments, the results are as previously, inhibition of GAT-3 but not GABA B R regulates SOM-INs inhibition of pyramidal cells via astrocyte Ca 2+ signaling. The legend of Fig. 4 and the corresponding text in the Results section have been modified.

7-To avoid redundancy in our answers, we grouped the following points made by reviewer #2
P11: Studies are consistent with the GAT-3 being important in controlling the A1R modulation of SOM-IN inhibition of pyramidal neurons. However, all A1R evidence in the ms is weak, relying on a single does of the antagonist DPCPX. More effort should be taken into clearly establishing the role of A1R. This evidence should include the use of A1R agonists, as well as A1R knockout mice.  We share the reviewer concerns about the need for direct evidence of A 1 R agonist. Regarding the use of A 1 R KO mice, we feel the effects will be too extensive and deleterious since those KO are known to deeply impact synaptic transmission and induce defects in excitatory transmission, lower seizure thresholds and cognitive/anxiogenic dysfunctions (Johansson et al., 2001;Fredholm et al., 2005 ;Fedele et al., 2006Kochanek et al., 2006. Strength of the present study is the ability to transiently modulate the network with the use of specific pharmacology in paired experiments where each cell is its own control.
But as requested by the reviewer we performed additional rescue experiments with a specific A 1 R agonist. We now report that N 6 -CPA rescues SNAP-5114 reduction of IPSCs and this effect is blocked by the A 1 R antagonist.
As a consequence the following paragraph and figure has been added in the Results sections line 280 to 292: (note that the effect of N 6 -CPA alone is not shown in the manuscript as described below but shown to the reviewer after figure 5) "To further confirm this mechanism we tested if bath application of the A1R agonist N6cyclopentyladenosine (N6-CPA, 1 µM) during the blockade of GAT-3 could up-regulate inhibition of pyramidal cells by SOM-INs. Under normal basal conditions (in absence of inhibitors), bath application of N6-CPA did not affect the amplitude of SOM-IPSCs evoked by optogenetic stimulation (99.0 ± 10.1% of control, n = 5, p= 0.33 Wilcoxon signed-rank test; data not shown). However in the presence of the GAT-3 inhibitor (S)-SNAP-5114 that decreased amplitude of SOM-IPSCs (59.3 ± 5.9 % of control, n = 4, p < 0.001 relative to control), application of the A1R agonist N6-CPA increased the amplitude of SOM-IPSCs (72.8 ± 5.9% of control, p < 0.001 relative to (S)-SNAP-5114 Fig 5b, d). Furthermore, this effect of N6-CPA was blocked by bath application of the A1R antagonist DPCPX (Fig 5b, d).
Overall, these results are consistent with a GAT-3 activation of astrocytes leading to ATP release, activation of A1Rs and upregulation of pyramidal cell inhibition by SOM-INs." The following figure 5 has been modified to represent these new data (Fig.5 b, d) Bath application of N 6 -CPA at 1 µM alone had no effect on SOM-IPSCs amplitude as illustrated on the figure below (due to space limitation we chose not to include this figure on the manuscript).
This absence of further inhibition is probably due to endogenous tone of adenosine in the extracellular space. 8-P12: Immuno co-localization is presented to show a correlation between the localization of GAT-3 and A1Rs. However, for this evidence to be more compeling quantitative approaches are necessary.
(Similar to reviewer 1 point 10) In the manuscript we are careful to point out in a qualitative manner the evident co-localization revealed by immunohistochemistry to support the "putative mechanism involving GAT-3-dependent activation of a Ca 2+ -dependent astrocytic purinergic regulation of inhibitory synapses on pyramidal cell dendrites". We feel that a quantification of GAT-3 and A 1 R co-localization is not necessary to make our point. In addition such quantification would require extensive analysis (deconvolution methods, super or pseudo super-resolution methods such as airy-scan or even electron microscopy studies) and we feel these exhaustive experiments are beyond the scope of the present paper. 9-P13: AMP-CP is used to examine the role of CD73 in mediating the hydrolysis of ATP to adenosine. This should be extended by using CD73 knockout mice.
In the light of the additional experiments, suggested by the reviewers, that we carried for the revisions of the manuscript (see point 5 above) and for similar reasons pointed out above (about the effect of global deletions; time consuming nature of KO mice generation and characterization, in addition to the two different Cre mice lines used (SOM-ChR2/EYFP and PV-ChR2/EYFP) in the present study, we think that carrying out these additional experiments is beyond the scope of the present paper.
10-The mechanism of GAT3 inhibition leading to alterations in Ca2+ signals is speculative at best and further studies are required to understand mechanism.
We addressed this point of the reviewer by acknowledging previous published work about GAT-3 and astrocyte Ca 2+ mechanisms in a paragraph in the discussion (line 417-425).
"As previously suggested 17, 18 , GAT-3 mediated Ca2+ events could involve GAT-3 activation leading to inhibition of Na+/Ca2+ exchanger and subsequent Ca2+-induced Ca2+ release from internal stores. This may be an auto-regulated mechanism since astrocytic Ca2+ signals can conversely modulate GAT-3 activity and expression 24 . The possibility of a coordinated transporter/receptor mechanism involving GAT-3 and GABABRs in Ca2+ transients in astrocytes is also possible, owing to their intimate co-localization in astrocytes (Fig. 3 and Extended Data Fig. 4) and a recent report showing that GABABRs controls GAT-3 levels in astrocytes in vivo during synaptogenesis 57 ." Reviewer #3: The manuscript by Matos et al. describes a form of feedback modulation of the inhibitory events evoked by CA1 PC dendrite-innervating SOM+ IN through activation of astrocytes. The feedback is completed by the activation of A1Rs by adenosine following the conversion of ATP released by the astrocytes. The authors also claim that this effect is specific to SOM+ IN. There are two major flaws with the claims: 1) The comparison of the drug effects onto sIPSCs and those evoked by optogenetic stimulation of SOM+ IN is equivocal. Most of the sIPSCs recorded in the somata of CA1 PCs originate not from the activity of IN innervationg distant dendrites, but from the somatargeting IN (e.g., PV+ and CCK+ basket cells). A fair comparison would have been to express ChR2 in other dendrite-targeting IN (e.g., PV+ bistratified cells, or other IN types), and to carry out a side-by-side comparison of the pharmacological effects on the two (or more) types of optogenetically evoked IPSCs. Such experiments would substantiate the claim that the modulation is unique to the SOM+ IN, which would constitute a significant finding. Otherwise the modulation of IPSCs by gliotransmitters or glial intermediaries is by itself not novel.