TBC1D9 regulates TBK1 activation through Ca2+ signaling in selective autophagy

Invading microbial pathogens can be eliminated selectively by xenophagy. Ubiquitin-mediated autophagy receptors are phosphorylated by TANK-binding kinase 1 (TBK1) and recruited to ubiquitinated bacteria to facilitate autophagosome formation during xenophagy, but the molecular mechanism underlying TBK1 activation in response to microbial infection is not clear. Here, we show that bacterial infection increases Ca2+ levels to activate TBK1 for xenophagy via the Ca2+-binding protein TBC1 domain family member 9 (TBC1D9). Mechanistically, the ubiquitin-binding region (UBR) and Ca2+-binding motif of TBC1D9 mediate its binding with ubiquitin-positive bacteria, and TBC1D9 knockout suppresses TBK1 activation and subsequent recruitment of the ULK1 complex. Treatment with a Ca2+ chelator impairs TBC1D9–ubiquitin interactions and TBK1 activation during xenophagy. TBC1D9 is also recruited to damaged mitochondria through its UBR and Ca2+-binding motif, and is required for TBK1 activation during mitophagy. These results indicate that TBC1D9 controls TBK1 activation during xenophagy and mitophagy through Ca2+-dependent ubiquitin-recognition.

In this manuscript Nozawa et al., intended to explore the role of a Ca2+ binding protein TBC1D9 in TBK1 activation and consequent selective autophagy (xenophagy and mitophagy). Interestingly, TBK1 activation shown here is independent of STING, a previously reported activator of TBK1 during infection. Authors screened several TBC/RabGAPs during Group A strep infection and found that TBC1D9 affects autophagy activation. Furthermore, TBC1D9 interacts with TBK1 and KO of TBC1D19 reduced phosphorylation of TBK1 and recruitment of pTBK1 and TBK1 effectors p62, NDP52 and LC3 to bacteria. Authors also show that TBC1D9 regulates mitophagy and that Ca2+ signaling is required for recruitment of TBC1D9 to mitochondria and consequent mitophagy. Overall, this is very interesting study identifying the role of TBC1D9 in microbial autophagy and mitophagy, the finding that TBK1 can be activated by means other than its published interactors STING, MAVS, TRIF, STX17, etc, the data presented in this study are clear and experiments in most parts are well controlled. There are however several issues that need to be addressed before this manuscript is acceptable for publication.
1. What do authors suppose is happening after TBK1 activation? Please see points 9-12 below, for the missed mechanisms and references for recently identified autophagy processes that depend on TBK1 that are completely (hopefully unintentionally) ignored here. 2. Authors seem to contradict their statement in data shown in Fig 3d where they show that TBC1D9 recruitment is independent of ATG5. Does this mean that this process is independent of autophagy? 3. Alternatively, the xenophagy reported here is dependent on autophagy which is independent of ATG5? 4. Is LC3 recruitment also independent of ATG5? 5. Given that this study is mainly focused on autophagy (xenophagy and mitophagy), authors should firm their results using KO/KD of earliest effectors of autophagy such as ULK complex, which has recently been shown to regulate xenophagy and mitophagy (Mol Cell. 2019 Apr 18;74(2):320-329; Mol Cell. 2019 Apr 18;74(2): 347-362.), to test if this complex is required for recruitment of TBC1D9 to bacteria. 6. Figure 4m: authors need to replace the images or show all channels as the difference between control and IP3Rs samples is not apparent, the representative images should match the data shown in graphs. 7. Authors should also test if BAPTA-AM reduces colocalization of LC3 and TBC1D9 on bacteria. 8. Authors need to show quantifications related to figure 5d and 5e and should do some additional assay such as mitochondrial DNA quantifications or degradation of COXII (Nature 524, 309-314 (2015) to measure mitophagy. 9. Authors seem to be ignoring the published literature related to their study: Authors show that TBK1 phosphorylation/activation by GAS infection was independent of STING, which looks convincing but authors need to discuss the published literature related to STING and TBK1 in response to dsDNA (Proc Natl Acad Sci U S A. 2009 Dec 8;106(49): 20842-20846 (2):347-362.). 12. Authors also did not discuss and appear to ignore (again, the message here is that there is nothing gained by ignoring established work) some important papers showing role of Galectin in In this manuscript, Nozawa et al report about the importance of Ca2+ signalling in activation of TBK1 kinase during xenophagy and mitophagy. TBK1 has previously been shown to be recruited to Salmonella, leading to activation of xenophagy, thus restricting its proliferation. The authors have shown that TBK1 interacts with TBC1D9, which is recruited to Group A Streptoccocus (GAS) or damaged mitochondria via Ca2+-dependent ubiquitin binding. The research performed by Nozawa et al sheds light on the mechanism underlying the recruitment of autophagosomal machinery to GAS, with strong biochemical data to support it. For the publication, the authors should address the functional relevance of this mechanism, outlined in the comments below, which is the main part lacking in this manuscript.
1. The authors throughout the manuscript predominantly use the stable Hela EmGFP-TBC1D9 cell line. In Figures 3c and 5a, the authors used an antibody which appears to stain the endogenous protein very nicely. What is missing is a control, as the authors do not show endogenous TBC1D9 under normal conditions, non-infected ( Figure 3c) and untreated ( Figure 5a). The authors should also show the endogenous TBC1D9 staining with and without Parkin, as well as with and without BAPTA-AM in Figure 5. Another point is that all images in Figure 5 are upon AO treatment. It would be good to include the untreated images, at least in supplementary material, and see if there is any recruitment of TBC1D9 to mitochondria as part of basal mitophagy. 2. The authors concluded that TBC1D9 and Ca2+ signalling are important for the activation of xenophagy and mitophagy. While it is clear that the autophagy machinery recruitment to GAS and damaged mitochondria is inhibited upon depletion of TBC1D9, or Ca2+ (by using BAPTA-AM), the authors have not performed any functional experiments to show the biological significance of the mechanism they revealed in this manuscript. The authors should include some functional assays to measure mitophagy and xenophagy, possibly even bacterial proliferation. 3. Figure 3e, delta2-722 is missing EF-hand domain. 4. Figure 3f, the last image is labelled with delta919-1266, whereas the text says it is delta926-1266. This is confusing, and it should be corrected. 5. Figure 3j, the authors should, if possible, include the full-length GST-TBC1D9, or explain in case they had issues purifying it. 6. Not essential, but it would be good to label Dapi in grayscale, to increase the contrast between the DNA labelling and the background. 10. Discussion is full of spelling mistakes and typos. The whole manuscript should be proofread before publication. 11. Some blots have kDa labels and some don't. It would be best to include the molecular weight next to the bands shown.
Reviewer #3 (Remarks to the Author): The paper prepared by Nakagawa and team members reveals a novel calcium-dependent mechanism by which pathogens entering the cytosol of mammalian cells are removed by xenophagy. The authors show that TBC1D9 is recruited to invading Group A Streptococcus infection in a calcium and ubiquitin-dependent manner. This provokes "activation" of TBK1 through phosphorylation and phagophore formation around the bacteria. This is a completely novel mechanism. Furthermore, the authors have expanded the relevance of their findings to mitophagy, which provides a more general cell biological implications. Until now, not much was known about the protagonists (TBC1D9 and TBK1) revealed in this work and the link towards calcium signaling is completely novel.
Overall, the work is novel and exciting. I enjoyed reading this paper. The experiments and results are well described and underpinned by a plethora of approaches, mainly at the cell biological level.
Some aspects though, particularly related to calcium signaling, require further scrutiny and study to underpin (some of) the conclusions. Comments 1. Wild-type GAS provokes a cytosolic calcium rise that appears to be dependent on the presence of slo in the bacteria. The calcium rise is anticipated to originate from intracellular Ca2+ stores, but evidence is scant. Moreover, wild-type GAS appears to provoke Ca2+ rise that are even somewhat higher than A23187, which is due to release of ER but also from calcium fluxes across the plasma membrane. Thus, further work is needed: a) In addition to the snap shots taken at 1, 2 or 4 hours, it could be informative to monitor livecell calcium dynamics of wild-type GAS at much higher time resolution. (suggestion) b) It will instrumental to use ratiometric calcium indicators such Fura-2 in order to rule out changes in cell volume or other artefacts occurring upon wild-type GAS infection. c) The authors should demonstrate the effect of wild-type GAS on cytosolic Ca2+ levels in the absence of extracellular Ca2+ (thus using extracellular BAPTA) as well as in conditions of chelating intracellular Ca2+ (thus using BAPTA-AM). This is often omitted in studies, but an important control. Furthermore, it will be critical to assess the effect of IP3R knockdown on the cytosolic calcium levels in response to wild-type GAS. The same is true for 2-APB (see also below) and other inhibitors that should be tested. The use of PLC inhibitor U73122 (and its inactive enantiomer) could shed light on the necessity of PLC signaling. Moreover, the data with EGTA or EGTA-AM are puzzling on TBK-phosphorylation and also require further analysis. d) There is no statistical analysis presented of the Ca2+ data presented in Figure 4A. 2. a) The interpretation of the BAPTA-AM data to indicate the involvement of Ca2+ has to be taken with care, as BAPTA-AM treatments can also affect the Na/K ATPases. Besides validating that BAPTA-AM can reduce wild-type GAS-induced calcium rises, the authors should assess the effect of low-affinity Ca2+ chelators BAPTA-AM variants such dibromo-or difluoro-BAPTA-AM on wild-type GAS-induced TBK1 phosphorylation. Also, the use of EGTA-AM is needed (see also point 2b); EGTA-AM is a slow calcium chelator that is cell-permeable, but given the slow kinetics of the Ca2+ rise, this should work. b) The authors should clarify whether they have used EGTA (as described in the results) or EGTA-AM (as indicated in the panel). c) Unfortunately, 2-APB is not a selective IP3R inhibitor. In fact, 2-APB also inhibits SERCA and depletes ER calcium stores. This is not per se a problem, but will affect interpretation of the data. As such, wild-type GAS could fail to provoke a calcium rise, simply because ER Ca2+ stores were depleted. This should be further scrutinized using xestospongin B or C (but is less selective than XeB) and the use of thapsigargin, an irreversible SERCA inhibitor. The use of U73122 as PLC inhibitor could provide insight in the role of this enzyme. d) It would be advisable to underpin these data with the overexpression of Ca2+-buffering proteins such as parvalbumin or calbindin D-28k. 3. It should be validated that wild-type GAS-induced calcium rise is independent of downstream components such as STING or TBC1D9 or TBK1 by using the available knockout cell models. 4. It would be interesting to assess whether the xenophagy of delta slo GAS could be restored upon eliciting calcium signaling through artificial ways (such as A23187 exposure)? 5. It should be validated whether or not the Rab35 recruitment to GAS is dependent on cytosolic Ca2+ increases. 6. What is the consequence of failed xenophagy in case of GAS infection? What is the cell fate in experimental conditions with unresolved xenophagy upon GAS infection (such as TBC1D9 KO's; IP3R KO/KD cells). 7. The role of TBK1 phosphorylation in self-association requires further study. In other words, could TBK1-S172E (a phospho-mimic variant) self-assemble with TBK1 (wild-type or S172E mutant) in the absence of TBC1D9? 8. TBK1 phosphorylation is particularly affected by the absence of TBC1D9 at 4 hours, while the effect at 2 hours is much less evident. Moreover, there remains some TBPK1 phosphorylation in the absence of TBC1D9. How can this be explained? Is still inhibited upon cytosolic calcium buffering / IP3R knockdown. In other words, is the effect of TBC1D9 KO and calcium chelation/IP3R KD on TBK1 phosphorylation additive? 9. It should be clarified which IP3R isoform has been knocked down. Indeed, IP3Rs are not a protein as such, but are expressed as three different isoforms (IP3R1, IP3R2 and IP3R3).

Comment #1-1)
Manuscript entitled "TBC1D9 promotes 1 TBK1 activation through Ca2+ signaling in selective autophagy". This study appears to be of interest and is relatively convincing, hence the recommendation is a major revision. In  Response to comment #1-1) First, we wish to thank the reviewer for these positive comments on our study. We have performed additional experiments to clarify some concepts, as described below.

Comment #1-2)
What do authors suppose is happening after TBK1 activation? Please see points 9-12 below, for the missed mechanisms and references for recently identified autophagy processes that depend on TBK1 that are completely (hopefully unintentionally) ignored here.

Response to comment #1-2)
As suggested by the reviewer, recent advances have revealed that TBK1 activation leads to various downstream signaling molecules such as p62, NDP52, STX17, RAB7, and the ULK1 complex. In the original manuscript, as we focused on the activation mechanism of TBK1 during xenophagy, we did not mention the processes depending on TBK1. However, as suggested by reviewers, we agree that the importance of the autophagy process regulated by TBK1 that is activated through TBC1D9 and Ca 2+ signaling needs to be emphasized. Thus, we have cited the relevant references and performed additional experiments to reveal which autophagy process is influenced by TBK1 activation downstream of TBC1D9 recruitment. We previously showed that TBK1 activation leads to NDP52 and LC3 recruitment to GAS. Moreover, recent reports have revealed that TBK1 and NDP52 regulate ULK1 complex recruitment to bacteria in xenophagy. We then examined the localization of ULK1 complex during GAS infection. As shown in Fig. 2e, ULK1 was recruited to ubiquitin-coated GAS and this recruitment was suppressed by TBC1D9 knockout. We also observed that ULK1 and its complex components (ULK2, FIP200, and ATG13) were required for xenophagic degradation of GAS. Therefore, the results suggest that TBC1D9-mediated TBK1 activation leads to ULK1 complex recruitment and initiation of xenophagy.

Authors seem to contradict their statement in data shown in Fig 3d where they show that
TBC1D9 recruitment is independent of ATG5. Does this mean that this process is independent of autophagy?

Response to comment #1-3)
We believe that the recruitment of TBC1D9 to bacteria is an ATG5-independent mechanism whereas the TBC1D9-mediated xenophagy process is ATG5-dependent. In Fig. 3d of the original manuscript, we observed TBC1D9-positive GAS in both wild-type and ATG5-knockout cells. This suggested that TBC1D9 can be recruited to bacteria prior to LC3 recruitment. However, TBC1D9 recruitment efficiency in ATG5-knockout cells was significantly lower than that in wild-type cells, indicating that LC3 might promote the recruitment of TBC1D9. Taken together, TBC1D9 recruitment to GAS via ubiquitin and Ca 2+ -binding might occur prior to ATG5-mediated LC3 recruitment (autophagic membrane formation), and LC3 facilitates TBC1D9 accumulation around GAS via an unknown mechanism. During mitophagy, autophagy adaptors such as NDP52 and Optineurin are first recruited to ubiquitin-coated mitochondria, and following LC3 recruitment they provide positive feedback to recruit autophagy adaptors (Padman et al., Nat Commun. 2019); thus, the TBC1D9-TBK1-autophagy adaptors axis and LC3 might amplify autophagy signaling at the targeting sites during selective autophagy.

Comment #1-4)
Alternatively, the xenophagy reported here is dependent on autophagy which is independent of ATG5?

Response to comment #1-6)
As suggested by the reviewer, we examined the involvement of the ULK1 complex in TBC1D9 recruitment in the revised experiments. We knocked down the expression of FIP200 and ULK1 using siRNA, and observed the localization of TBC1D9 during GAS infection. Recruitment of TBC1D9 to ubiquitin-positive GAS was not affected by knockdown of FIP200 or ULK1 (below figure). Alternatively, we found that ULK1-positive GAS was significantly decreased in TBC1D9-knockout cells (Fig. 2 e, f).
Recruitment of ATG13 to depolarized mitochondria was also suppressed by TBC1D9 knockout (Fig. 6 d,e). These observations suggest that TBC1D9 is an upstream molecule of the ULK1 complex during the xenophagy and mitophagy initiation process.

Comment #1-7)
Figure 4m: authors need to replace the images or show all channels as the difference between control and IP3Rs samples is not apparent, the representative images should match the data shown in graphs.

Response to comment #1-7)
As suggested, we showed all channels and included insets of LC3 images in the control and IP3R-knockdown cells.

Comment #1-8)
Authors should also test if BAPTA-AM reduces colocalization of LC3 and TBC1D9 on bacteria.

Response to comment #1-8)
We have examined the effects of BAPTA-AM on the colocalization of LC3 and TBC1D9 in bacteria. As expected, colocalization was significantly decreased by BAPTA-AM (Fig.   4d).

Comment #1-9)
Authors need to show quantifications related to figure 5d and 5e and should do some additional assay such as mitochondrial DNA quantifications or degradation of COXII (Nature 524, 309-314 (2015) to measure mitophagy.

Response to comment #1-9)
As suggested by the reviewer, to evaluate mitophagy activity, we quantified COXII degradation in TBC1D9 KO cells. As shown in Fig. 6f and 6g

20842-20846).
Response to comment #1-10) As suggested, we agree that these references are important to discuss TBK1 activation in autophagy. We have thus cited these references in the revised manuscript.

Comment #1-11)
The authors should discuss latest preprint published

Response to comment #1-11)
As suggested, we have cited this reference and discussed it in the revised manuscript.

Response to comment #1-12)
As suggested, we have cited this and discussed it in the revised manuscript.

Authors also did not discuss and appear to ignore (again, the message here is that there is nothing gained by ignoring established work) some important papers showing role of
Galectin in endomembrane damage (Nature. 2012;482: 414-418) and a paper showing role of galectin3 in lysophagy (Dev Cell. 2016 Oct 10;39(1):13-27.) Response to comment #1-13) As suggested, we have added to the discussion in the revised manuscript. Specifically, since we showed that ULK1 complex recruitment is regulated by the TBC1D9-TBK1 axis in xenophagy, we discussed the difference in xenophagy and lysophagy.

Comment #2-1)
In this manuscript, Nozawa et al report about the importance of Ca2+ signalling in activation of TBK1 kinase during xenophagy and mitophagy. TBK1 has previously been shown to be recruited to Salmonella, leading to activation of xenophagy, thus restricting its proliferation. The authors have shown that TBK1 interacts with TBC1D9, which is recruited to Group A Streptoccocus (GAS) or damaged mitochondria via

Ca2+-dependent ubiquitin binding. The research performed by Nozawa et al sheds light
on the mechanism underlying the recruitment of autophagosomal machinery to GAS, with strong biochemical data to support it. For the publication, the authors should address the functional relevance of this mechanism, outlined in the comments below, which is the main part lacking in this manuscript.

Response to comment #2-1)
We first wish to thank the reviewer for the positive comments on our study. We have performed additional experiments to clarify some concepts, as described below.

Comment #2-2)
The authors throughout the manuscript predominantly use the stable Hela EmGFP-TBC1D9 cell line. In Figures 3c and 5a Response to comment #2-1) As suggested, we showed endogenous TBC1D9 localization in control cells (Fig. 3c and   5a). We also exhibited the subcellular localization of endogenous TBC1D9 with and without Parkin (Supplementary Fig. 9a). In addition, we showed TBC1D9 mutants localization under normal conditions (Supplementary Fig. 9b). As the reviewer commented, TBC1D9 partially colocalized with mitochondria even in the control condition. Thus, TBC1D9 might also be involved in basal mitophagy.

Comment #2-3)
The authors concluded that TBC1D9 and Ca2+ signalling are important for the activation of xenophagy and mitophagy. While it is clear that the autophagy machinery recruitment to GAS and damaged mitochondria is inhibited upon depletion of TBC1D9, or Ca2+ (by using BAPTA-AM), the authors have not performed any functional experiments to show the biological significance of the mechanism they revealed in this manuscript. The authors should include some functional assays to measure mitophagy and xenophagy, possibly even bacterial proliferation.

Response to comment #2-3)
As suggested, to investigate the functional significance of TBC1D9 in xenophagy and mitophagy, we assessed the degradation of GAS and mitochondria using TBC1D9-knockout cells. In Fig. 2g of the revised manuscript, survival GAS in TBC1D9 knockout HeLa cells at 6 h after infection was significantly higher than that in wild-type HeLa cells, suggesting that degradation of GAS is diminished by TBC1D9 knockout in HeLa cells. Moreover, AO-treatment-induced COXII degradation was also suppressed in TBC1D9 KO cells (Fig. 6f and 6g). Therefore, we concluded that TBC1D9 is functionally important for xenophagy and mitophagy.

Response to comment #2-4)
We appreciate your suggestion. We have modified Figure 3e in the revised manuscript.

Figure 3f, the last image is labelled with delta919-1266, whereas the text says it is
delta926-1266. This is confusing, and it should be corrected.
The paper prepared by Nakagawa and team members reveals a novel calcium-dependent mechanism by which pathogens entering the cytosol of mammalian cells are removed by xenophagy. The authors show that TBC1D9 is recruited to invading Group A Streptococcus infection in a calcium and ubiquitin-dependent manner. This provokes "activation" of TBK1 through phosphorylation and phagophore formation around the bacteria. This is a completely novel mechanism. Furthermore, the authors have expanded the relevance of their findings to mitophagy, which provides a more general cell biological implications. Until now, not much was known about the protagonists (TBC1D9 and TBK1) revealed in this work and the link towards calcium signaling is completely novel.
Overall, the work is novel and exciting. I enjoyed reading this paper. The experiments and results are well described and underpinned by a plethora of approaches, mainly at the cell biological level.
Some aspects though, particularly related to calcium signaling, require further scrutiny and study to underpin (some of) the conclusions.

Response to comment #3-1)
We appreciate your positive comments on our study and your advice. In response to all the comments, we have performed the suggested experiments and included data in the revised submission.

Comment #3-2a)
Wild-type GAS provokes a cytosolic calcium rise that appears to be dependent on the presence of slo in the bacteria. The calcium rise is anticipated to originate from intracellular Ca2+ stores, but evidence is scant. Moreover, wild-type GAS appears to provoke Ca2+ rise that are even somewhat higher than A23187, which is due to release of ER but also from calcium fluxes across the plasma membrane. Thus, further work is needed: In addition to the snap shots taken at 1, 2 or 4 hours, it could be informative to monitor live-cell calcium dynamics of wild-type GAS at much higher time resolution.

Response to comment #3-2a)
We agree that live-cell imaging could be informative to monitor the calcium dynamics during infection; however, unfortunately we could not perform this experiment in our research environment.

Comment #3-2b)
It will instrumental to use ratiometric calcium indicators such Fura-2 in order to rule out changes in cell volume or other artefacts occurring upon wild-type GAS infection.

Response to comment #3-2b)
As suggested, we used Fura 2-AM to monitor the Ca 2+ dynamics during GAS infection.
Time course of the Fura 2-AM ratio during GAS infection revealed that cytosolic Ca 2+ increased from 2 h after infection and that this elevation was SLO-dependent (Fig. 4a).
These results correspond with the data obtained using G-GaMP3 (Supplementary Fig. 5a and 5b).

Comment #3-2c)
The authors should demonstrate the effect of wild-type GAS on cytosolic Ca2+ levels in the absence of extracellular Ca2+ (thus using extracellular BAPTA) as well as in conditions of chelating intracellular Ca2+ (thus using BAPTA-AM). This is often omitted in studies, but an important control. Furthermore, it will be critical to assess the effect of IP3R knockdown on the cytosolic calcium levels in response to wild-type GAS. The same is true for 2-APB (see also below) and other inhibitors that should be tested. The use of PLC inhibitor U73122 (and its inactive enantiomer) could shed light on the necessity of PLC signaling. Moreover, the data with EGTA or EGTA-AM are puzzling on TBK-phosphorylation and also require further analysis.

Response to comment #3-2b)
As suggested, we first examined the Ca 2+ elevation during GAS infection in BAPTA-AM and BAPTA treated conditions using Fura 2-AM. We found that BAPTA partially suppressed Ca 2+ elevation, and that BAPTA-AM effectively inhibited Ca 2+ mobilization (Supplementary Fig. 5f and 5g). We also quantified TBK1 activation in cells with BAPTA-AM, BAPTA, EGTA-AM, and EGTA. As shown in Supplementary Fig. 5f and 5g, EGTA-AM and EGTA did not influence TBK1 activation, whereas BAPTA-AM clearly suppressed TBK1 activation in a dose-dependent manner. BAPTA affected TBK1 activation only slightly. Collectively, although extracellular Ca 2+ might also be involved in TBK1 activation, intracellular Ca 2+ would be critical for xenophagy during GAS infection.
In addition, we used the PLC inhibitor U731122 to test whether PLC signaling is involved in Ca 2+ -mediated TBK1 activation and found that U73122 significantly inhibited Ca 2+ elevation and TBK1 activation during GAS infection (Supplementary Fig.   5f and 5g). These results support that IP3 signaling is involved in Ca 2+ mobilization and TBK1 activation during GAS infection for xenophagy.

Comment #3-2c)
There is no statistical analysis presented of the Ca2+ data presented in Figure 4A.

Response to comment #3-2c)
We performed the statistical analysis and included the data in Supplementary Fig. 5b.

Comment #3-3a)
The interpretation of the BAPTA-AM data to indicate the involvement of Ca2+ has to be taken with care, as BAPTA-AM treatments can also affect the Na/K ATPases. Besides validating that BAPTA-AM can reduce wild-type GAS-induced calcium rises, the authors should assess the effect of low-affinity Ca2+ chelators BAPTA-AM variants such dibromo-or difluoro-BAPTA-AM on wild-type GAS-induced TBK1 phosphorylation.
Also, the use of EGTA-AM is needed (see also point 2b); EGTA-AM is a slow calcium chelator that is cell-permeable, but given the slow kinetics of the Ca2+ rise, this should work.

Response to comment #3-3a)
We appreciate your informative advice about calcium inhibitors. We used dibromo-BAPTA-AM and quantified TBK1 activation in response to GAS infection.
Treatment with dibromo-BAPTA-AM suppressed TBK1 activation, but was only mildly-effective compared to BAPTA-AM (Supplementary Fig. 5f and 5g). Also, we have added the data for EGTA-AM in the revised manuscript. We did not observe any effect of EGTA-AM on TBK1 activation during GAS infection. We have mentioned this result in the revised manuscript.

Comment #3-3b)
The authors should clarify whether they have used EGTA (as described in the results) or

Response to comment #3-3b)
We made an error in the label in Fig. 4h of the original manuscript. In the original manuscript, we showed the data using EGTA (not EGTA-AM). As described above, we have added the results of both EGTA-AM and EGTA.

Comment #3-3c)
Unfortunately, 2-APB is not a selective IP3R inhibitor. In fact, 2-APB also inhibits SERCA and depletes ER calcium stores. This is not per se a problem, but will affect interpretation of the data. As such, wild-type GAS could fail to provoke a calcium rise, simply because ER Ca2+ stores were depleted. This should be further scrutinized using xestospongin B or C (but is less selective than XeB) and the use of thapsigargin, an irreversible SERCA inhibitor. The use of U73122 as PLC inhibitor could provide insight in the role of this enzyme.

Response to comment #3-3c)
We investigated the effects of these inhibitors on TBK1 activation during wild-type GAS infection, and found that in our experiment condition, XeC and U73122 significantly inhibited TBK1 activation whereas thapsigargin did not. These results suggest that IP3R-mediated calcium release is involved in TBK1 activation during GAS infection.

Comment #3-3d)
It would be advisable to underpin these data with the overexpression of Ca2+-buffering proteins such as parvalbumin or calbindin D-28k.
Moreover, recruitment of TBC1D9 to ubiquitin-positive GAS was also suppressed by overexpression of PVALB or CALB (Supplementary Fig. 6a). There results support that cytosolic Ca 2+ is critical for TBC1D9 recruitment and TBK1 activation against GAS infection.

Comment #3-4)
It should be validated that wild-type GAS-induced calcium rise is independent of downstream components such as STING or TBC1D9 or TBK1 by using the available knockout cell models.

Response to comment #3-4)
As suggested by the reviewer, we examined the Ca 2+ level in STING, TBC1D9, and TBK1 knockout cells during wild-type GAS infection, and observed that Ca 2+ elevation was observed even in these knockout cells (Supplementary Fig. 5i).

Comment #3-5)
It would be interesting to assess whether the xenophagy of delta slo GAS could be restored upon eliciting calcium signaling through artificial ways (such as A23187 exposure)?

Response to comment #3-5)
We infected HeLa cells with GAS Δslo and treated them with A23187 to elicit calcium signaling, and then examined xenophagy. None of TBC1D9, p-TBK1, and LC3 was recruited to GAS Δslo even when with A23187 treatment (below figures). Although Ca 2+ elevation in the cytosol is critical for TBC1D9-mediated TBK1 activation, ubiquitin accumulation on bacteria is also essential for the TBC1D9-TBK1 axis and this ubiquitin recruitment is independent on Ca 2+ signaling (Supplementary Fig. 5c and 5d). Therefore, we think that Ca 2+ signaling and the ubiquitination process are activated in parallel in response to wild-type GAS infection, and that both events are required for TBC1D9-mediated xenophagy by promoting TBK1 activation.

Comment #3-6)
It should be validated whether or not the Rab35 recruitment to GAS is dependent on cytosolic Ca2+ increases.
Response to comment #3-6) As suggested, we observed Rab35 localization during infection with BAPTA-AM.
BAPTA-AM did not inhibit the recruitment of Rab35 to bacteria (supplementary Fig. 5e).
This result is consistent with our previous results that Rab35 is recruited to GAS before cytosolic invasion of GAS.

Comment #3-7)
What is the consequence of failed xenophagy in case of GAS infection? What is the cell fate in experimental conditions with unresolved xenophagy upon GAS infection (such as

Response to comment #3-7)
In the revised manuscript, we examined bacterial survival in several knockout or knockdown cells. Bacterial survival at 6 h after infection was significantly increased in TBC1D9 knockout cells as well as in ATG5 knockout cells ( Fig. 2g; Minowa Nozawa et al. EMBO J), indicating that TBC1D9 is required for xenophagic degradation of GAS.
We also found that knockdown of IP3Rs increased bacterial survival (Fig. 4k).

Comment #3-8)
The role of TBK1 phosphorylation in self-association requires further study. In other words, could TBK1-S172E (a phospho-mimic variant) self-assemble with TBK1 (wild-type or S172E mutant) in the absence of TBC1D9?
Response to comment #3-8) As suggested, we examined the self-association of TBK1-S172E in TBC1D9 knockout cells. Similar to wild-type TBK1, interaction between FLAG-TBK1-S172E and EmGFP-TBK1-S172E was still attenuated in TBC1D9 knockout cells. We therefore concluded that TBC1D9 is involved in the self-association of TBK1. We have added these results to Fig. 1i in the revised manuscript.

Comment #3-9)
TBK1 phosphorylation is particularly affected by the absence of TBC1D9 at 4 hours, while the effect at 2 hours is much less evident. Moreover, there remains some TBK1 phosphorylation in the absence of TBC1D9. How can this be explained? Is still inhibited upon cytosolic calcium buffering / IP3R knockdown. In other words, is the effect of TBC1D9 KO and calcium chelation/IP3R KD on TBK1 phosphorylation additive?
Response to comment #3-9) In TBC1D9 knockout cells, the phosphorylated TBK1 (p-TBK1) level was significantly increased in normal conditions (the reason of this is not clear yet). Then, to evaluate the increase in p-TBK1 during infection, we quantified the p-TBK1 amount relative to 0 h.p.i in WT and knockout cells. As shown in Fig. 1b, increase of p-TBK1 relative to that before infection significantly decreased in TBC1D9 knockout cells at both 2 and 4 h. To validate this, we transiently knocked down the TBC1D9 expression using siRNA and found that the siRNA for TBC1D9 inhibited TBK1 phosphorylation even at 2 h after infection ( Supplementary Fig. 7b, c). In addition, we found that the p-TBK1 level was comparable between TBC1D9-knockdown and TBC1D9/IP3R1 double-knockdown cells, suggesting that the effect of TBC1D9 knockdown and IP3Rs knockdown on TBK1 activation is not additive.

Comment #3-10)
It should be clarified which IP3R isoform has been knocked down. Indeed, IP3Rs are not a protein as such, but are expressed as three different isoforms (IP3R1, IP3R2 and IP3R3).
To identify which IP3R isoforms is involved in xenophagy induction, we selectively knocked down the expression of each IP3R isoform. We found that IP3R1 knockdown most profoundly diminished TBK1 activation and bacterial degradation ( Fig. 4f and 4k).