Author response:
Reviewer #1 (Public Review):
Summary:
The authors want to elucidate which are the mechanisms that regulate the immune response in physiological conditions in cortical development. To achieve this goal, authors used a wide range of mutant mice to analyse the consequences of immune activation in the formation of cortical ectopia in mice.
Strengths:
The authors demonstrated that Abeta monomers are anti-inflammatory and inhibit microglial activation. This is a novel result that demonstrates the physiological role of APP in cortical development.
Weaknesses:
-On the other hand, cortical ectopia has been already described in mouse models in which the amyloid signalling has been disrupted (Herms et al., 2004; Guenette et al., 2006), making the current study less novel.
We agree these previous studies have implicated amyloid precursor protein in cortical ectopia. However, since these studies use whole-body knockouts, they have not implicated the functional roles of specific cell types. Nor have they identified the specific mechanisms underlying the formation of this unique class of cortical ectopia. In contrast, our studies show that the disruption of a novel Abeta-regulated signaling pathway in microglia is the primary cause of ectopia formation in this class of ectopia mutants. This is the first time that microglia have been specifically implicated in the development of cortical ectopia. We further show that elevated MMP activity and resulting cortical basement membrane degradation is the underlying mechanism leading to ectopia formation. This is also the first time that MMP activity and basement membrane degradation (instead of maintenance) have been implicated in cortical ectopia development. As such, our results have provided novel insights into the diverse mechanisms underlying cortical ectopia formation in developmental brain disorders.
One of the molecules analysed is Ric8a, a GTPase activator involved in neuronal development. Authors used the conditional mutant mice Emx1-Ric8a to delete Ric8a from early progenitors and glutamatergic neurons in the pallium. Emx1-Ric8a mutant mice present cortical ectopias and authors attributed this malformation to the increase in inflammatory response due to Ric8a deletion in microglia. Several discordances do not fit this interpretation:
-The role of Ric8a in cortical development and function has been already described in several papers, but none of them has been cited in the current manuscript (Kask et al., 2015, 2018; Ruisu et al., 2013; Tonissoo et al., 2006).
We will include reference to these publications in revision.
-Ectopia formation in the cortex has been already described in Nestin-Ric8a cKO mice (Kask et al., 2015). In the current manuscript, authors analyzed the same mutant mice (Nestin-Ric8a), but they did not detect any ectopia. Authors should discuss this discordance.
The expression pattern of nestin-cre is known to vary dependent on factors including transgene insertion site, genetic background, and sex. Early studies show, for example, that the nestin gene promoter drives cre expression in many non-neural tissues in another transgenic line in the FVB/N genetic background (Dubois et al Genesis. 2006 Aug;44(8):355-60. doi: 10.1002/dvg.20226). The specific nestin-cre line used in Kask et al 2015 has also been shown to be active in brain microglia and lead to increased microglia pro-inflammatory activity upon breeding to a conditional allele of a cholesterol transporter gene (Karasinska et al., Neurobiol Dis. 2013 Jun:54:445-55; Karasinska et al., J Neurosci. 2009 Mar 18; 29(11): 3579–3589). The ectopia reported in Kask et al 2015 are also significantly more subtle than what we have observed and apparently not observed in all mutant animals (we observe severe ectopia in every single emx1-cre mutant). We presume the ectopia reported in Kask et al 2015 may result from a combined deletion of ric8a gene from microglia and neural cells due to unique combinations of factors affecting nestin-cre expression in a subset of mutants.
-Authors claim that microglia express Emx1, and therefore, Ric8a is deleted in microglia cells. However, the arguments for this assumption are very weak and the evidence suggests that this is not the case. This is an important point considering that authors want to emphasise the role of Ric8a in microglia activation, and therefore, additional experiments should demonstrate that Ric8a is deleted in microglia in Emx1-Ric8a mutant mice.
We have observed altered mRNA expression of several genes in purified microglia cultured from the emx1-cre mutants (Supplemental Fig. 8), which indicates that ric8a is deleted from microglia and suggests a role of microglial ric8a deficiency in ectopia formation. This interpretation is further strengthened by the observation that deletion of ric8a from microglia using a microglia-specific cx3cr1-cre results in similar ectopia (Fig. 2). We also have other data supporting this interpretation, including data showing induction of the expression of a cre reporter in brain microglia by emx1-cre and loss of ric8a gene expression in microglia cells isolated from emx1-cre mutants. We will include these data in revision.
Reviewer #2 (Public Review):
Kwon et al. used several conditional KO mice for the deletion of ric8a or app in different cell types. Some of them exhibited pial basement membrane breaches leading to neuronal ectopia in the neocortex.
They first investigated ric8a, a Guanine Nucleotide Exchange Factor for Heterotrimeric G Proteins. They observed the above-mentioned phenotype when ric8a is deleted from microglia and neural cells (ric8a-emx1-cre or dual deletion with cre combination cx3cr1 (in microglia) and nestin (in neural cells)) but not in microglia alone or neural cells alone (whether it is in CR cells (ric8a-Wnt3a-cre), post-mitotic neurons (nex-cre or dlx5/6-cre), or in progenitors and their progeny (nestin-cre or foxg1-cre). They also show that ric8a KO mutant microglia cells stimulated in vitro by LPS exhibit an increased TNFa, IL6 and IL1b secretion compared to controls (Fig 2). They therefore injected LPS in vivo and observed the neuronal ectopia phenotype in the ric8a-cx3cr1-cre (microglial deletion) cortices at P0 (Fig 2). They suggest that ric8a KO in neuronal cells mimics immune stimulation (but we have no clue how ric8a KO in neural cells would induce immune stimulation).
We agree we do not currently know the precise mechanisms by which mutant microglia are activated in the mutant brain. However, this does not affect the conclusion that deficiency in the Abeta monomer-regulated APP/Ric8a pathway in microglia is the primary cause of cortical ectopia in these mutants, since we have shown that genetic disruption of this pathway in microglia alone by different means targeting different pathway components, using cell type specific cre, all results in similar cortical ectopia phenotypes. Regarding the source of the immunogens, there are several possibilities which we plan to investigate in future studies. For example, the clearance of apoptotic cells and associated cellular debris is an important physiological process and deficits in this process have been linked to inflammatory diseases throughout life (Doran et al., Nat Rev Immunol. 2020 Apr;20(4):254-267; Boada-Romero et al., Nat Rev Mol Cell Biol. 2020 Jul;21(7):398-414.). In the embryonic cortex, studies have shown that large numbers of cell death take place starting as early as E12 (Blaschke et al., Development. 1996 Apr;122(4):1165-74; Blaschke et al., J Comp Neurol. 1998 Jun 22;396(1):39-50). Studies have also shown that radial glia and neuronal progenitors play critical roles in the clearance of apoptotic cells and associated cellular debris in the brain (Lu et al., Nat Cell Biol. 2011 Jul 31;13(9):1076-83; Ginisty et al., Stem Cells. 2015 Feb;33(2):515-25; Amaya et al., J Comp Neurol. 2015 Feb 1;523(2):183-96). Moreover, Ric8a-dependent heterotrimeric G proteins have been found to specifically promote the phagocytic activity of both professional and non-professional phagocytic cells (Billings et al., Sci Signal. 2016 Feb 2;9(413):ra14; Preissler et al., Glia. 2015 Feb;63(2):206-15; Pan et al. Dev Cell. 2016 Feb 22;36(4):428-39; Flak et al. J Clin Invest. 2020 Jan 2;130(1):359-373; Zhang et al., Nat Commun. 2023 Sep 14;14(1):5706). Thus, it is likely that the failure to promptly clear up apoptotic cells and debris by radial glia may play a role in the triggering of microglial activation in ric8a mutants. We have not included discussion of these possibilities since the precise mechanisms remain to be determined. Moreover, they also do not impact the conclusion of the current study.
The authors then turned their attention on APP. They observed neuronal ectopia into the marginal zone when APP is deleted in microglia (app-cxcr3-cre) + intraperitoneal LPS injection (they did not show it, but we have to assume there would not be a phenotype without the injection of LPS) (Fig 3). (The phenotype is similar but not identical to ric8a-cx3cr1-cre + LPS. They suggest that the reason is because they had to inject 3 times less LPS due to enhanced immune sensitivity in this genetic background but it is only a hypothesis). After in vitro stimulation by LPS, app mutant microglia show a reduced secretion of TNFa and IL6 but not IL1b (this is the opposite to ric8a-cx3cr1-cre microglia cells) while peritoneal macrophages in culture show increased secretion of TNFa, IL1, IL6 and IL23 (fig 3 and Suppl. Fig 9).
We have data showing that that app-cxcr3-cre mutants without LPS injection do not show ectopia and will include them in revision. The reason we employ LPS injection is, in the first place, we do not see a phenotype without the injection. We agree, and have also stated in the text, that the phenotype of the app mutants is not as severe as that of the ric8a mutant. Besides the low LPS dosage used, we also suggest that other app family members may compensate since the ectopia in the app family gene mutants reported previously were only observed in app/aplp1/2 triple knockouts, not even in any of the double knockouts (Herms et al., 2004). These potential causes are also not mutually exclusive. Nonetheless, the microglia specific app mutants clearly show ectopia upon immune stimulation, implicating a role of microglial APP in cortical ectopia formation.
The distinct response of ric8a and app microglia to LPS results from in vitro culturing of microglia. Indeed, we have shown that, when acutely isolated macrophages are used, these mutants show changes in the same direction (both increased cytokine secretion). The microglia used for analysis in this study have all been cultured in vitro for two weeks before assay. They have thus been under chronic stimulation exposing to dead cells and debris in the culture dish through this period. Dependent on the degree of perturbation to inflammation-regulating pathways, such exposures are known to significantly change microglial cytokine expression, sometimes in an opposite direction from expected. For example, under chronic immune stimulation, while the trem2+/- microglia, which are heterozygous mutant for the anti-inflammatory Trem2, show elevated pro-inflammatory cytokine expression as expected, trem2-/- (null) microglia under the same conditions instead not only do not show increases but for some pro-inflammatory cytokines, actually show decreases in expression (Sayed et al.,, Proc Natl Acad Sci U S A. 2018 Oct 2;115(40):10172-10177). In several systems, Ric8a-dependent heterotrimeric G proteins have been shown to act downstream of APP and mediate one of the branches of the signaling activated by APP (Milosch et al., Cell Death Dis. 2014 Aug 28;5(8):e1391; Fogel et al,, Cell Rep. 2014 Jun 12;7(5):1560-1576; Ramaker et al., J Neurosci. 2013 Jun 12;33(24):10165-81; Nishimoto et al., Nature. 1993 Mar 4;362(6415):75-9). It is likely that in microglia Ric8a-dependent heterotrimeric G proteins may also mediate only a subset of the signaling downstream of APP. As such, app knockout in microglia may have more severe effects than ric8a knockout on microglial immune activation and lead to changes in the opposite direction compared to ric8a knockout, as has been observed for trem2 null mutation vs heterozygosity discussed above. This may explain the subdued TNF and IL6 secretion by cultured app mutant microglia.
Amyloid beta (Ab) being one of the molecules binding to APP, the authors showed that Ab40 monomers (they did not test Ab40 oligomers) partially inhibit cytokines (TNFa, IL6, IL1b, MCP-1, IL23a, IL10) secretion in vitro by microglia stimulated by LPS but does not affect secretion by microglia from app-cx3cr1-cre (tested for TNFa, IL6, IL1b, IL23a, IL10) (Fig 4, Suppl fig 10) (but still does it in aplp2-cx3cr1-cre) and does not affect secretion by ric8a-cx3cr1-cre microglia (tested for TNFa and IL6 but still suppress IL1b) (Therefore here is another difference between app and ric8a KO microglia).
We have tested the effects of Abeta40 oligomers, which induce instead of suppressing microglial cytokine secretion, and will include the data in revision. As mentioned above, in several systems, Ric8a-dependent heterotrimeric G proteins have been shown to act downstream of APP and mediate one of the branches of the signaling activated by APP (Milosch et al., Cell Death Dis. 2014 Aug 28;5(8):e1391; Fogel et al,, Cell Rep. 2014 Jun 12;7(5):1560-1576; Ramaker et al., J Neurosci. 2013 Jun 12;33(24):10165-81; Nishimoto et al., Nature. 1993 Mar 4;362(6415):75-9). We assume that this is likely also true in microglia and that Ric8a-dependent heterotrimeric G proteins may mediate only a subset of the signaling downstream of APP. This may explain the difference in the effects of APP and ric8a knockout mutation in abolishing the anti-inflammatory effects of Abeta monomers on IL-1b vs TNF/IL-6. It also suggests that TNF/IL-6 and IL-1b secretion must be regulated by different mechanisms. Indeed, it is well established in immunology that the secretion of IL1b, but not of TNF or IL6, is regulated by inflammasome-dependent mechanisms (see, for example, Proz & Dixit. Nat Rev Immunol. 2016 Jul;16(7):407-20. doi: 10.1038/nri.2016.58).
The authors injected inhibitors of Akt or Stat3 in the ric8a-emx1-cre cortex and found it suppressed neuronal ectopia (Fig 5, Suppl fig 11). It is not clear whether it suppresses immune stimulation from neuronal cells or immune reaction from microglia cells.
We agree at present the pharmacological approaches we have taken are not able to distinguish these possibilities. However, whichever of these possibilities turns out to be the case would still implicate a role of excessive microglial activation in the formation of cortical ectopia and support the conclusion of the study. Thus, while potentially worthwhile of further investigation, this question does not impact the conclusion of this study. Furthermore, as mentioned, we plan to determine the mechanisms of how ric8a mutation in neural cells induces immune activation in future studies. These results will likely enable us to adopt more specific approaches to address this question.
Finally, the authors examined the activities of MMP2 and MMP9 in the developing cortex using gelatin gel zymography. The activity and protein levels of MMP9 but not MMP2 in the ric8a-emx1-cre cortex were claimed significantly increased (Fig 5, Suppl fig 12). Unfortunately, they did not show it in the app-cx3cr1-cre +LPS mouse. They make a connection between ric8a deletion and MMP9 but unfortunately do not make the connection between app deletion and MMP9, which is at the center of the pathway claimed to be important here). Then they injected BB94, a broad-spectrum inhibitor of MMPs or an inhibitor specific for MMP9 and 13. They both significantly suppress the number and the size of the ectopia in ric8a mutants (Fig5).
For all the gelatin gel zymography analysis, we quantify protein concentrations in the cortical lysates using the Bio-Rad Bradford assay kit and load the same amounts of proteins per lane. The results across lanes are thus directly comparable. From the quantification, our results clearly show that MMP9, but not MMP2, levels are increased in the mutants (supplemental Figure 12). The data on MMP2 also provide an internal control further supporting the observation of a specific change in MMP9. For this analysis, we focus on the ric8a-emx1-cre mutants since the app-cx3cr1-cre +LPS animals show less severe, more localized ectopia and in most cases only in one of the hemispheres. Any changes in MMP9 are therefore likely to be masked and the experiments unlikely to yield meaningful results. On the other hand, we have clearly shown that the administration of different classes of MMP inhibitors significantly eliminate ectopia in ric8a-emx1-cre mutants. This has strongly implicated a functional contribution of MMPs.
After reading the manuscript, I still do not know how ric8a in neural cells is involved in the immune inhibition. Is it through the control of Ab monomers? In addition, the authors did not show in vivo data supporting that Ab monomers are the key players here. As the authors said, this is not the only APP interactor. Finally, I still do not know how ric8a is linked to APP in microglia in the model.
As detailed above, there are several possibilities including potential deficits in the clearance of apoptotic cells and associated debris that may trigger microglial activation in ri8ca-emx1-cre mutants. We will investigate these possibilities in future studies. We have not included discussion since their roles remain to be determined. As for the role of Abeta monomers, we have indicated that we currently do not have evidence that in the developing cortex Abeta monomers play a role in inhibiting microglia. We have also indicated in the manuscript that our conclusion is that an Abeta monomer-activated microglial pathway regulates normal brain development, not that Abeta monomers themselves regulate brain development. Regarding the link between Ric8a and APP, the reviewer has missed several major lines of supporting evidence. For example, we have shown that Abeta monomers activates a pathway in microglia that inhibits the secretion of several proinflammatory cytokines including TNF, IL-6, IL-10, and IL-23 (Figure 4 and Supplemental Figures 8-10). This inhibition is abolished when either app or ric8a gene is deleted from microglia. This indicates that app and ric8a act in the same pathway activated by Abeta monomers in microglia. We also show that this Abeta monomer-activated pathway also inhibits the transcription of several cytokines in microglia. This inhibition is also abolished when either app or ric8a gene is deleted from microglia. This reinforces the conclusion that app and ric8a act in the same pathway in microglia. Furthermore, cell type specific deletion of app or ric8a from microglia in vivo also results in similar phenotypes of cortical ectopia. Together, these results thus strongly support the conclusion that app and ric8a act in the same pathway activated by Abeta monomers in microglia. This conclusion is also consistent with published findings that Ric8a dependent heterotrimeric G proteins bind to APP and mediate subsets of APP signaling across different different species (Milosch et al., Cell Death Dis. 2014 Aug 28;5(8):e1391; Fogel et al,, Cell Rep. 2014 Jun 12;7(5):1560-1576; Ramaker et al., J Neurosci. 2013 Jun 12;33(24):10165-81; Nishimoto et al., Nature. 1993 Mar 4;362(6415):75-9).
While several of the findings presented in this manuscript are of potential interest, there are a number of shortcomings. Here are some suggestions that could improve the manuscript and help substantiate the conclusions:
(1) As the title suggests it, the focus is on Ab and APP functions in microglia. However, the analysis is more focused on ric8a. The connection between ric8a and APP in this study is not investigated, besides the fact that their deletion induces somewhat similar but not identical phenotypes. Showing a similar phenotype is not enough to conclude that they are working on the same pathway. The authors should find a way to make that connection between ric8a and app in the cells investigated here.
As discussed above, the reviewer misses several major lines of evidence showing that APP and Ric8a acts in the same pathway in microglia. For example, besides the similarity of the ectopia phenotypes, we have shown that Abeta monomers activates a pathway in microglia that inhibits the secretion of several proinflammatory cytokines including TNF, IL-6, IL-10, and IL-23 (Figure 4 and Supplemental Figures 8-10). These inhibitory effects are completely abolished when either app or ric8a gene is deleted from microglia. This indicates that app and ric8a act in the same pathway activated by Abeta monomers in microglia. We also show that this Abeta monomer-activated pathway inhibits the transcription of several cytokine genes in microglia. These effects are again completely abolished when either app or ric8a gene is deleted from microglia. This further reinforces the conclusion that app and ric8a act in the same pathway in microglia. Not only so we also show that the same results are true in macrophages. Together, these results therefore strongly support the conclusion that app and ric8a act in the same pathway in microglia. This conclusion is also consistent with published findings that Ric8a dependent heterotrimeric G proteins bind to APP and mediate APP signaling across different species (Milosch et al., Cell Death Dis. 2014 Aug 28;5(8):e1391; Fogel et al,, Cell Rep. 2014 Jun 12;7(5):1560-1576; Ramaker et al., J Neurosci. 2013 Jun 12;33(24):10165-81; Nishimoto et al., Nature. 1993 Mar 4;362(6415):75-9).
(2) This would help to show the appearance of breaches in the pial basement membrane leading to neuronal ectopia; to investigate laminin debris, cell identity, Wnt pathway for app-cxcr3-cre + LPS injection as you did for ric8a-emx1-cre.
We will provide further data on the breaches in the pial basement membrane. We have not observed any changes in cell identity or Wnt pathway activity in ric8a-emx1-cre mutants. The ectopia phenotype in the app-cxcr3-cre + LPS animals is also less severe. It is therefore likely of limited value to examine potential changes in these areas.
(3) As a control, this would help to show that app-cxcr3-cre without the LPS injection does not display the phenotype.
We have the data on app-cx3cr1-cre mutants without LPS injection, which show no ectopia, and will include the data in revision.
(4) This would help to show the activity and protein levels of MMP9 and MMP2 and perform the rescue experiments with the inhibitors in the app-cx3cr1-cre cortex +LPS.
As discussed above, we focus analysis on the ric8a-emx1-cre mutants since app-cx3cr1-cre +LPS animals show less severe, more localized ectopia and in most cases only in one of the hemispheres. Determining potential changes in MMP9 levels and effects of MMP inhibitors are therefore not likely to yield useful data. On the other hand, we have shown that MMP9 levels are increased and administration of different classes of MMP inhibitors eliminate cortical ectopia in ric8a-emx1-cre mutants. This has strongly implicated a functional contribution of MMPs.
(5) Is MMP9 secreted by microglia cells or neural cells?
Our in situ hybridization data show MMP9 is most highly expressed in macrophage-like cells in the embryonic cortex, suggesting that microglia may be a major source of MMP9. We will incorporate these data in revision.
(6) The in vitro evidence indicates that one of the multiple APP interactors, ie Ab40 monomers, is less effective in suppressing the expression of some cytokines by microglia cells mutants for ric8a (TNFa and IL6 but still suppress IL1b) or APP (TNFa, IL6, IL1b, IL23a, IL10) when compared to WT. But there are other interactors for APP. In order to support the claim, it seems crucial to have in vivo data to show that Ab40 monomers are the molecules involved in preventing the breach in the pial basement membrane.
As addressed in detail above, we have indicated that our conclusion is that an Abeta monomer-activated microglial pathway regulates normal brain development, not that Abeta monomers themselves regulate brain development. We currently do not have evidence that the Abeta monomers play a role in inhibiting microglia in the developing cortex. There are candidate ligands for the pathway in the developing cortex, the functional study of which, however, is a major undertaking and beyond the scope of the current study.
(7) In order to claim that this is specific to Ab40 monomers and not oligomers, it is necessary to show that the Ab40 oligomers do not have the same effect in vitro and in vivo. Also, an assay should be done to show that your Ab preparations are pure monomers or oligomers.
We have tested the effects of Abeta40 oligomers, which induce instead of suppressing microglial cytokine secretion, and will include the data in revision. The protocols we use in preparing the monomers and oligomers are standard protocols employed in the field of Alzheimer’s disease research and have been optimized and validated repeatedly over the past several decades.
(8) Most of the cytokine secretion assays used microglia cells in culture. Two results draw my attention. Ric8a deletion increases TNFa and IL6 secretion after LPS stimulation in vitro on microglia cells while app deletion decreases their secretion. Then later, papers show that the decrease in IL1b induced by Ab on microglia cells is prevented by APP deletion but not ric8a deletion. Those two pieces of data suggest that ric8a and APP might not be in the same pathway. In addition, the phenotype from app-cxcr3-cre + LPS injection and ric8a-cxcr3-cre + LPS injection are not exactly the same. It could be due to the level of LPS as the author suggests or it might not be. More experiments are needed to prove they are in the same pathway.
As discussed above, the reviewer misses several major lines of evidence, which strongly support the conclusion that APP and Ric8a act in the same pathway activated by Abeta monomers in microglia (see detailed discussion in point 1). The differential response of app and ric8a mutant microglia likely results from chronic immune stimulation during in vitro culturing, which is known to alter microglia cytokine expression (see detailed discussion in point 9 below on how chronic immune stimulation changes microglial cytokine expression). We have demonstrated this by showing that, without culturing, acutely isolated app and ric8a mutant macrophages both display elevated cytokine secretion (Figure 4). Regarding the distinct regulation of TNF/IL-6 and IL-1b by APP and Ric8a, as discussed above, in several systems, Ric8a-dependent heterotrimeric G proteins have been shown to act downstream of APP and mediate one of the branches of the signaling activated by APP (Milosch et al., Cell Death Dis. 2014 Aug 28;5(8):e1391; Fogel et al,, Cell Rep. 2014 Jun 12;7(5):1560-1576; Ramaker et al., J Neurosci. 2013 Jun 12;33(24):10165-81; Nishimoto et al., Nature. 1993 Mar 4;362(6415):75-9). It is likely this is also the case in microglia and Ric8a-dependent heterotrimeric G proteins may mediate only a subset of the anti-inflammatory signaling activated by APP. As such, this may explain why app, but ric8a, mutation abolishes the inhibitory effects of Abeta monomers on IL-1b. This also suggests that the secretion of TNF/IL-6 and IL-1b must be regulated by different mechanisms. Indeed, it is well established in immunology that the secretion of IL1b, but not that of TNF or IL6, is regulated by inflammasome-dependent mechanisms (see, for example, Proz & Dixit. Nat Rev Immunol. 2016 Jul;16(7):407-20. doi: 10.1038/nri.2016.58).
(9) How do the authors reconcile the reduced TNFa and IL6 secretion upon stimulation of app mutant microglia with the model where app is attenuating immune response in vivo? Line 213 says that microglia exhibit attenuated immune response following chronic stimulation but I don't know if 3 hours of LPS in vitro is a chronic stimulation.
The reviewer has misunderstood. The microglia used in this study have all been cultured in vitro for approximately two weeks before assay. They have thus been under chronic stimulation exposing to dead cells and debris in the culture dish throughout this period. Dependent on the degree of perturbation to inflammation-regulating pathways, such exposures are known to significantly change microglial cytokine expression, sometimes in an opposite direction than expected. For example, under chronic immune stimulation, while the trem2+/- microglia, which are heterozygous mutant for the anti-inflammatory Trem2, show elevated pro-inflammatory cytokine expression as expected, trem2-/- (null) microglia under the same conditions instead not only do not show increases but for some pro-inflammatory cytokines, actually show decreases in expression (Sayed et al.,, Proc Natl Acad Sci U S A. 2018 Oct 2;115(40):10172-10177). As mentioned, in several systems, Ric8a-dependent heterotrimeric G proteins have also been shown to bind to APP and mediate one of the branches of the signaling activated by APP (Milosch et al., Cell Death Dis. 2014 Aug 28;5(8):e1391; Fogel et al,, Cell Rep. 2014 Jun 12;7(5):1560-1576; Ramaker et al., J Neurosci. 2013 Jun 12;33(24):10165-81; Nishimoto et al., Nature. 1993 Mar 4;362(6415):75-9). It is likely that Ric8a-dependent heterotrimeric G proteins also mediate only a subset of the anti-inflammatory signaling activated by APP in microglia. As such, app knockout in microglia may have more severe effects than ric8a knockout on microglial immune activation, similar to the relationship between trem2 null mutation vs heterozygosity discussed above. This likely explains why TNF and IL6 secretion by cultured app mutant microglia is subdued. In contrast, we find that acutely isolated app mutant macrophages show increased cytokine secretion. This is likely more representative of the response of app mutant microglia in the absence of chronic stimulation.
(10) Line 119: In their model, the authors suggest that there is a breach in pial basement membrane but that the phenotype is different from the retraction of the radial fibers due to reduced adhesion. So, could the author discuss to what substrate the radial fibers are attached to, in their model where the pial surface is destroyed?
Radial glial endfeet normally bind to the basement membrane via cell surface receptors including the integrin and the dystroglycan protein complexes. We observe free radial glial endfeet at the breach sites, apparently without attachment to any basement membrane. However, we cannot exclude the possibility that there may be residual basement components not detected by the methodology employed.
(11) The authors should show that the increased cytokine secretion observed in vitro is also happening in vivo in ric8a-emx1-cre compared to WT mice and compared to ric8a-nestin-cre mice. Or when app is deleted in microglia (app-cxcr3-cre) + LPS injection compared to WT mice +LPS.
Unfortunately, this is not technically feasible since it is impossible to extract the extracellular (secreted) fractions of cytokines from an embryonic brain without causing cell lysis and the release of the intracellular pool. This, however, does not affect our conclusion that the Abeta monomer-regulated microglia pathway plays a key role in regulates normal brain development since its genetic disruption, by different approaches, clearly results in brain malformation.
(12) The authors injected inhibitors of Akt or Stat3 in the ric8a-emx1-cre cortex and found that it suppressed neuronal ectopia (Fig 5, Suppl fig 11). Does it suppress immune stimulation from neuronal cells or immune reaction from microglia cells?
As discussed above, we agree at present the pharmacological approaches we have taken are not able to distinguish these two possibilities. However, no matter which possibility is true, it does not affect our conclusion. Furthermore, we also plan to determine the mechanisms of how ric8a mutation in neural cells induce immune activation in future studies. These results will likely enable us to adopt specific approaches to address this question.
(13) Fig 5 and Supplementary fig 12: Please show a tubulin loading control in Fig 5i as you did in suppl fig 12 d (gel zymography). Please provide a gel zymography showing side by side Control, mutant and mutant +DM/S3I treatment. The same request for the MMP9 staining. Please provide statistics for control vs mutant for suppl fig 12c and d.
For all experiments of the gelatin gel zymography analysis, we quantify protein concentrations in the cortical lysates using the Bio-Rad Bradford assay kit and load the same amounts of proteins per lane. The results across lanes are thus all comparable. These experiments were also performed several years ago before the pandemic and we unfortunately no longer have the samples. We will, however, provide the protein quantification information in revision. The MMP9 staining images for the controls and mutants have also all been taken with the same parameters on the microscope and can be directly compared. The statistics will be provided as suggested.
(14) Please provide the name and the source of the MMP9/13 inhibitor used in this study.
This inhibitor is MMP-9/MMP-13 inhibitor I (CAS 204140-01-2), from Santa Cruz Biotechnology. This information will be included in revision.
(15) The results show that deletion of ric8a in microglia and neural cells induced pia membrane breaches but no phenotype is apparent in ric8a deletion in microglia or neural cells alone. Then, the results showed that intraperitoneal injection of LPS induced the phenotype in ric8a-cxcr3-cre mutants. It would be beneficial as a control supporting the model to show that the insult induced by LPS injection does not induce the phenotype in the ric8a-foxg1-cre mice.
We agree it may potentially be useful to show that LPS injection does not induce ectopia in ric8a-foxg1-cre mice. Unfortunately, since the ric8a-foxg1-cre mutation shows no phenotype, we are no longer in possession of this line.
Reviewer #1 (Recommendations For The Authors):
-The information in the abstract and the introduction is only related to app. So, it is very abrupt how authors start the manuscript studying the role of Ric8a, with no information at all about this protein and why the authors want to investigate this role in microglial activation. Later in the manuscript, the authors tried to link Ric8a with app to study the role of app in the inflammatory response and ectopia formation. This link is quite weak as well.
In the last paragraph of the Introduction, we explain the use of the ric8a mutant and how it leads to discovery of the Abeta monomer-regulated pathway. We will improve the writing in revision to make these points clearer. We will also improve the writing of the potential link of Ric8a to APP by highlighting, especially, the fact that ric8a and app pathway mutants are among a unique group of only three mouse mutants (ric8a, app/aplp1/2, and apbb1/2) that show cortical ectopia exclusively in the lateral cortex, while all other cortical ectopia mutants show the most severe ectopia are at the midline.
-In order to validate the mouse model, double immunofluorescence or immunofluorescence+in situ hybridization should be performed to show that microglia express ric8a and that is eliminated in the Emx1-Ric8a mutant mice.
As mentioned above, we have additional lines of evidence showing that ric8a is deleted from microglia in emx1-cre mutants. This includes data showing induction of the expression of a cre reporter in brain microglia by emx1-cre and loss of ric8a gene expression in microglia cells isolated from emx1-cre mutants. We will include these data in revision.
-In Supplemental Fig. 6, the authors claimed that cell proliferation is normal in Ric8a mutant mice without doing any quantification. They also quantified the angle of mitotic division of progenitors in the ventricular zone, but there are no images for the spindle orientation quantification, and no description of how they did it. In addition, this data is contrary to what has already been published in conditional Ric8a mutant mice (Kask et al., 2015). The Vimentin staining should be improved.
We will provide quantification of cell proliferation in revision. We will also provide details on the quantification on mitotic spindle orientation. We are not sure why the results are different from the other study. We were indeed anticipating deficits in mitotic spindle orientation and spent major efforts in the analysis. However, based on the data, we could not draw the conclusion.
-Analysis of the MMP9 expression should be done by western blot and not by immunofluorescence. In fact, the MMP9 expression shown in Figure 5g,h, does not correspond with RNA expression shown in gene expression atlas like genepaint or the allen atlas, doubting the specificity of the antibody. The expression of Mmp9 is quite low or absent in the cortex at E13.5-E14.5, making this protein very unlikely to be responsible for laminin degradation during development.
We perform gelatin gel zymography on MMP2/9, which shows increased MMP9 activity levels in the mutant cortex. This is similar to Western blot analysis (all lanes are loaded with the same amounts of cortical lysates). The immunofluorescence staining, a different type, of analysis, was designed as a complementary approach. Regarding RNA expression, please also note that MMP9 is a secreted protein and the protein expression pattern is expected to be different from that of RNA. We also have in situ data showing that, while MMP9 mRNA is indeed low, it is strongly expressed in macrophage-like cells most prominently in cortical blood vessels at E12-E13 (we will include these data in revision). We suspect that these cells are microglial lineage cells populating the embryonic cortex at this stage (see, for example, Squarzoni et al., Cell Rep. 2014 Sep 11;8(5):1271-9. doi: 10.1016/j.celrep.2014.07.042.) and may be a major source of cortical MMP9. As for functional contributions, we agree that we cannot rule roles played by other MMPs. However, based on the ectopia suppression data, our results clearly indicate a key functional contribution by MMP9/13.
For MMP9 activity, authors should show the whole membrane with a minimum of three control and three mutant individual samples and with the quantification.
-The graphs should be improved, including individual values and titles of the Y axes.
We will include these data in revision (the quantification of MMP9 activity is provided in Supplemental Figure 12d) and improve the graphs as suggested.
