Cilia-localized LKB1 regulates chemokine signalling, macrophage recruitment and tissue homeostasis in the kidney

Transcription

1 Cilia-localized LKB1 regulates chemokine signalling, macrophage recruitment and tissue homeostasis in the kidney Amandine Viau, Frank Bienaimé, Kamile Lukas, Abhijeet P. Todkar, Manuel Knoll, Toma A. Yakulov, Alexis Hofherr, Oliver Kretz, Martin Helmstädter, Wilfried Reichardt, Simone Braeg, Tom Aschman, Annette Merkle, Dietmar Pfeifer, Verónica I. Dumit, Marie-Claire Gubler, Roland Nitschke, Tobias B. Huber, Fabiola Terzi, Jörn Dengjel, Florian Grahammer, Michael Köttgen, Hauke Busch, Melanie Boerries, Gerd Walz, Antigoni Triantafyllopoulou, E. Wolfgang Kuehn Review timeline: Submission date: 9 November 2017 Editorial Decision: 10 January 2018 Revision received: 28 March 2018 Editorial Decision: 27 April 2018 Revision received: 13 May 2018 Accepted: 15 May 2018 Editors: Ieva Gailite and Deniz Senyilmaz Tiebe Transaction Report: (Note: With the exception of the correction of typographical or spelling errors that could be a source of ambiguity, letters and reports are not edited. The original formatting of letters and referee reports may not be reflected in this compilation.) 1st Editorial Decision 10 January 2018 Thank you for submitting your manuscript for consideration by the EMBO Journal. We have now received a full set of referee reports on your manuscript, which are included below for your information. As you can see from the comments, reviewers #1 and #3 express interest in the presented role of ciliary LKB1 in regulation of inflammation in polycystic kidneys. However, all three referees also raise substantive concerns with the analysis that would need to be addressed before they can support publication here. Based on the overall interest expressed in the reports I would like to invite you to submit a revised version of your manuscript in which you address the comments of all three referees. Please note that it is The EMBO Journal policy to allow only a single major round of revision and that it is therefore important to resolve the main concerns at this stage. We generally allow three months as standard revision time, but the revision period can be extended to six months in the case of extensive revisions. Please contact us in advance if you would need an additional extension. As a matter of policy, competing manuscripts published during this period will not negatively impact on our assessment of the conceptual advance presented by your study. However, please contact me as soon as possible upon publication of any related work to discuss how to proceed. Please feel free to contact me if have any further questions regarding the revision. Thank you for the opportunity to consider your work for publication. I look forward to your revision European Molecular Biology Organization 1

2 Referee #1: In the present study, Viau and co-authors investigated the interaction between the primary cilium of kidney cells and inflammation processes in cilia-related kidney disorders. Here, they uncovered an essential role of LKB1 for normal kidney functioning, and showed that loss of LKB1 leads to a CCL2 chemokine release with subsequent recruitment of immune cells, and also that this process is absolutely dependent on the primary cilium. This is a very interesting area, which to date remains poorly investigated. Overall, this study is novel and has a significant impact in understanding the mechanisms underlying the cilia signaling and cilia-related disorders. The data shown in the study are generally of high quality, however there are some points to address: 1. Loss of Lkb1 in Supp Fig 1J is associated with reduced size of the kidney. Most ciliopathies are associated with increased size of the kidney. What is the critical difference here? 2. The authors report that deletion of Lkb1 does not result in induction of AMPK or mtor by microarray, and hence some other event must be important for renal pathology (Fig 4). The explanation for this lack of transcriptional effect is likely the observation made in the past by some of the authors, which has shown that Lkb1 at the cilia indeed regulates AMPK and mtor in a physiologically important way, through phosphorylation occurring at the basal body (PMID: ), in a manner that controls renal cell growth. Many other important biological events regulating cystic phenotypes also occur at the cilia and basal body, but do not necessarily affect transcriptional readouts. The authors should emphasize that pathology does not necessarily require changes in the transcription of regulated proteins, to integrate their current and past work. 3. The mechanism of CCL2 production control by LKB1 has been already reported for cancer cells, as the authors note (PMID: ). However, the authors for the first time demonstrated specific cooperation between LKB1 and PKD1 controls CCL2 signaling, suggesting that disruption of this mechanism has an effect on ADPKD progression, a ciliopathy which is characterized by chronic inflammation and fibrosis. A puzzling part of this observation is the fact that many cancer cells lose ciliation, making it hard to understand how the same signaling mechanism can be observed in ciliated renal cells and likely unciliated endometrial cancer cells, particularly when loss of cilia in the current study results in reduced CCL2 signaling. The authors should reconcile these data. 4. Many other studies have described the upregulation of CCL2/MCP-1 in renal tubules as an early step in inflammation as part of the cystogenesis process: for example, with the Han:SPRD rat model (Cowley et al 2001, PMID: ), in human patients (Zheng et al 2003, PMID: ) and as a result of polycystin mutation (Chen et al 2015, PMID: ), and with the Six2cremediated deletion of Frs2α (Puri et al, 2016, PMID: ) and other stimuli. The results should be discussed in the context of the broader findings, that CCL2/MCP-1 induction is broadly linked to cystogenesis arising from multiple stimuli. Is the major new idea that CCL2 release and macrophage infiltration distinguish the more aggressive pathology of cysts associated with defective ciliary signaling, versus the less aggressive phenotype seen with loss of cilia? This should be better developed, and supported with some experiments. 5. It is also of interest that Ta and colleagues using a model of NPHP9-dependent cystic kidneys have shown that sirolimus inhibits the LKB1 regulated protein mtorc1, reducing disease pathology but not CCL2 expression (PMID: ). This also may inform mechanism, and could add to the discussion. Minor point. 1. Figure 5B. Could the authors enlarge the images demonstrating the localization of LKB1 in the primary cilium? Referee #2: General Summary In this manuscript, the authors conclude that Pkd1 and an Lkb1-based module comprising European Molecular Biology Organization 2

3 Lkb1/NPHP1/ANKS3/NEK7 form a functional unit that inhibits a ciliary Ccl2 inducing signal, which otherwise leads to inflammation and PKD progression in Lkb1 or Pkd1 deficient mice. They based this conclusion on the following observations. 1) A proteomic screen of Lkb1 precipitates identified Stradα and 30 other proteins that included the NPHP1-binding ANKS3 as well as NEK7 (known to interact with ANKS3); 2) NPHP1 and Lkb1 functionally interact in cystogenesis in zebrafish; 3) Ccl2 (Mcp1) and ANKRD1 are upregulated based on comparisons of microarray analysis in Lkb1 mutant kidneys at 5 weeks with RNA seq. in Lkb1-silenced MDCK; 4) Ccl2 is increased in Lkb1ΔTub kidneys at 5 weeks and in Lkb1-silenced MDCK, and neutrophils and CCR2+ Mø (but not T, B or total F4/80hiCD11b+ Mø) are increased in Lkb1ΔTub kidneys at 10 weeks; 5) Conditional silencing of Stradβ (but not Stradα) in MDCK increased Ccl2, and coablation of cilia together with Lkb1 prevented the increase in Ccl2 in Lkb1-silenced MDCK. Finally the authors showed that Pkd1 co-ip with Lkb1 and both Strads (α>>>β) in HEK293, and that ablated cilia in Pkd1 null MDCK cells or in 12 week ipkd1δtub ikif3aδtub mice decreased Ccl2 expression and kidney CCR2+ Mø. Specific concerns (can also be transmitted to the authors) Despite the extensive data presented, the paper do not convincingly support the authors major conclusion that a ciliary Pkd1 Lkb1-based complex regulates Ccl2 and that this Ccl2 is the elusive cilia-dependent cyst activating signal. First, Lkb1 expression in the kidney is massive (Fig. 1A) and its role cannot be accounted for by expression in cilia, unless its knockdown is limited to cilia, which has not been done. Second, it is unclear that the Lkb1-based module comprising Lkb1/NPHP1/ANKS3/NEK7 or Stradβ is cilia-limited as none of these proteins are expressed ONLY in cilia. Third, the relevance of the zebrafish data is suspicious as zebrafish pronephros does not have nonmotile primary cilia that are found in mammalian kidneys. Forth, identification of Ccl2 was based on the unusual comparison of microarray analysis in Lkb1 mutant kidneys at 5 weeks with RNA seq in the Lkb1-silenced canine cell line MDCK; no protein expression has been determined in any experiment presented, and cyst fluid in ADPKD patients contains several other cytokines. Fifth, the authors show that infiltration of CCR2+ macrophages (Fig 4D) but not total macrophages (Suppl. Fig. 6 J) is increased in Lkb1Δ Tub kidneys at 10 weeks. How significant is this subpopulation? What percent it represents of the total infiltrating Mø? Sixth, Stradβ does not bind Lkb1 in the absence of MO25, and PKd1 associates much better with Stradα (reportedly not involved in Ccl2 expression). Finally, how do these data explain the differences in clinical presentation of NPHP where inflammation and subsequent fibrosis appear early and are much more prominent than cysts vs. ADPKD if Ccl2 is the common factor in cystogenesis? Referee #3: This is an interesting manuscript that addressed the role of ciliary LKB1 in regulating chemokine signalling, macrophages and tissue homeostasis in polycystic kidney disease (PKD). Viau et al found that deletion of LKB1 in renal tubular cells not only disrupted renal morphology and function but also induced the expression of chemokine CCL2, resulting in the recruitment and activation of macrophages in kidneys. This study provided integrated evidences to link ciliary LKB1 to renal inflammation, in which LKB1 in the primary cilia of renal tubular epithelial cells was involved in macrophage infiltration and played an important role in the progress of ADPKD or NPHP. However, several concerns need to be addressed. Major concerns: 1. In Fig.1 the authors provided evidences that deletion of LKB1 affected kidney morphology and function. Some panels in this figure are confused and need to be reorganized by adding the missing controls. In Fig. 1C, the kidneys sizes were not diminished as mentioned in MRI analysis. In Fig. 1F, is it to be implied that the control kidney sections (for 14 and 23 weeks) do not change from the 5 week time point? 2. In Fig. 2, the authors found that LKB1 interacted with ANKS3, NPHP1 and NEK7 in kidney medulla and HEK 293 cells. In Fig. 5, depletion of NPHP1, ANKS3 or NEK7 resulted in European Molecular Biology Organization 3

4 upregulation of Ccl2 in MDCK cells. Does knockdown of NPHP1, ANKS3, and NEK7 affect the protein level and activity of LKB1? 3. In Fig. 6, the authors found that LKB1 also interacted with polycystin-1 (PC1) to form a functional unit to contribute to the progression of disease. Does deletion of PC1 affect the expression and activity of LKB1, ANKS3, NPHP3 as well as NEK7 in the PKD1 mutant cells and kidneys compared to the control? Does PKD1 mutation induce the decrease of LKB1 and the module proteins, leading to increase of Ccl2 expression in in PKD1 mutant cells and the subsequent recruitment of macrophages in PKD1 mutant kidneys. In addition, the effect of double knockout of Pkd1 and LKB1 on cyst growth, the expression of Ccl2 and the recruitment of macrophages in kidneys need to be added in Fig. 6 to strength the role of LKB1 in PKD. 4. The authors provided evidence that LKB1/NPHP1/ANKS3/NEK7 complex controls Ccl2 expression. Is Ccl2 expression in PKD1 mutant also LKB1 dependent? 5. What's the mechanism of LKB1 in regulating the expression of Ccl2. It is unclear how LKB1 inhibits Ccl2 expression at homeostasis or upregulates Ccl2 expression in Pkd1 mutant kidneys. Please justify. Minor concerns: In Fig. 6, it should be "Polycystin 1 (PC1)" interacts with LKB1 but not "PKD1", which is the gene name but not the protein name. The size of PC1 in Fig. 6a needs to be double checked. 1st Revision - authors' response 28 March 2018 We thank the reviewers for their careful assessment of our submission and for thoughtful comments which we address here in full. The process has helped to further strengthen of our findings and results in an improved manuscript. Referee #1: In the present study, Viau and co-authors investigated the interaction between the primary cilium of kidney cells and inflammation processes in cilia-related kidney disorders. Here, they uncovered an essential role of LKB1 for normal kidney functioning, and showed that loss of LKB1 leads to a CCL2 chemokine release with subsequent recruitment of immune cells, and also that this process is absolutely dependent on the primary cilium. This is a very interesting area, which to date remains poorly investigated. Overall, this study is novel and has a significant impact in understanding the mechanisms underlying the cilia signaling and cilia-related disorders. The data shown in the study are generally of high quality, however there are some points to address: 1. Loss of Lkb1 in Supp Fig 1J is associated with reduced size of the kidney. Most ciliopathies are associated with increased size of the kidney. What is the critical difference here? Renal mass in ciliopathies correlates with the number and size of cysts. This is true for human ciliopathies, as well as mouse models (PMIDs: , ). ADPKD is the prime example of massive cyst formation. The same is true for autosomal recessive polycystic kidney disease (ARPKD). Nephronophthisis (NPHP) on the other hand results in small kidneys in the majority of patients, for instance in patients with NPHP1 mutations. A recent survey of children with NPHP showed that 78% had small kidneys on ultrasound (PMID: ). Our findings suggest that in NPHP the inflammatory phenotype could be driven by CCL2 release, inflammation and fibrosis. In ADPKD, macrophages activate cyst growth, as has been described by others, explaining that the kidneys are larger. Our results do not explain what initiates cyst formation in the first place. They confirm that macrophages fuel cyst growth and establish for the first time that this is a direct effect of the loss of polycystin 1 function in the cilium and the subsequent CCL2 release. We have now added new in-vivo data to confirm that tubular CCL2 drives cyst growth in PKD1 mutant kidneys: tubule specific knock-out of PKD1 + CCL2 strongly ameliorates cyst growth compared to PKD1 targeted animals alone (Fig7 and Fig EV5). We have now altered the text in the results section to make this point clearer. European Molecular Biology Organization 4

5 2. The authors report that deletion of Lkb1 does not result in induction of AMPK or mtor by microarray, and hence some other event must be important for renal pathology (Fig 4). The explanation for this lack of transcriptional effect is likely the observation made in the past by some of the authors, which has shown that Lkb1 at the cilia indeed regulates AMPK and mtor in a physiologically important way, through phosphorylation occurring at the basal body (PMID: ), in a manner that controls renal cell growth. Many other important biological events regulating cystic phenotypes also occur at the cilia and basal body, but do not necessarily affect transcriptional readouts. The authors should emphasize that pathology does not necessarily require changes in the transcription of regulated proteins, to integrate their current and past work. We agree with the reviewer, that transcriptional read-outs are not an adequate way to assess mtor activity. This is reflected in Suppl. Figure 4 (now Appendix Fig S2 - please note that the figures had to be rearranged according to journal standards) which she/he refers to here: Similar to our previous work in MDCK cells under flow, here we do not observe an overall decrease of AMPK phosphorylation in LKB1 targeted kidneys (Appendix Fig S2A D). However, we do find a trend towards increased mtor activity in the LKB1 targeted kidneys (ps6rp by Western blot - Appendix Fig S2B right) without reaching statistical significance. For this reason we stated that 'no consistent effect' was found. We would like to point out that in the cited paper, due to optimized experimental conditions, we were able to detect differences of p-ampk at the basal body under flow vs. no flow, but this may not be measurable in the in vivo model and therefore doesn't negate a role of ciliary LKB1 in mtor signalling. Interestingly, in the referenced paper loss of LKB1 only partially affected mtor activity when compared to cilia loss (the size difference was much less) suggesting that the effect is contributory rather than exclusive. We have now added new data to the manuscript showing that treating LKB1 deficient cells with AICAR, a stabilizer of AMPK phosphorylation, decreased CCL2 expression (Fig EV3E). This suggests that phosphorylation of AMPK by LKB1 at the basal body may affect CCL2 in addition to mtor. We agree that the question of the precise role of mtor in PKD is important, and realize that many investigations are conducted to this end. Since our manuscript has a different focus and due to the large number of data that the manuscript contains, we decided against discussing this point further in the text. Since the review process will be published alongside the paper, in case it is accepted, the readers interested in this point will be able to follow the discussion here. 3. The mechanism of CCL2 production control by LKB1 has been already reported for cancer cells, as the authors note (PMID: ). However, the authors for the first time demonstrated specific cooperation between LKB1 and PKD1 controls CCL2 signaling, suggesting that disruption of this mechanism has an effect on ADPKD progression, a ciliopathy which is characterized by chronic inflammation and fibrosis. A puzzling part of this observation is the fact that many cancer cells lose ciliation, making it hard to understand how the same signaling mechanism can be observed in ciliated renal cells and likely unciliated endometrial cancer cells, particularly when loss of cilia in the current study results in reduced CCL2 signaling. The authors should reconcile these data. This is a very interesting point. Ciliation is heterogeneous across different cancers and not uniform for any type of malignancy. Indeed HeLa cells, which are derived from a highly aggressive cervical cancer contain cilia (PMID: ). Endometrial cancer is mentioned and has been shown to contain cilia (PMID: ). While the number of ciliated cells in that paper was lower compared to normal tissue, cilia were still present. One reason for the decreased number of cilia may be increased proliferation since mitotic cells are unciliated. In pancreatic carcinoma the presence of cilia correlates with more metastases worse prognosis (PMID: ). We have now added this reference to the text. Taking into account the genetic heterogeneity of cancers, even cancers of the same type, it is clear that the prognosis of cancers can be influenced by many defects. Ciliation may only be one of them. Future studies in different cancers are needed to achieve a better understanding of their role. 4. Many other studies have described the upregulation of CCL2/MCP-1 in renal tubules as an early step in inflammation as part of the cystogenesis process: for example, with the Han:SPRD rat model (Cowley et al 2001, PMID: ), in human patients (Zheng et al 2003, PMID: ) and as a result of polycystin mutation (Chen et al 2015, PMID: ), and with the Six2cremediated deletion of Frs2α (Puri et al, 2016, PMID: ) and other stimuli. The results should European Molecular Biology Organization 5

6 be discussed in the context of the broader findings, that CCL2/MCP-1 induction is broadly linked to cystogenesis arising from multiple stimuli. Is the major new idea that CCL2 release and macrophage infiltration distinguish the more aggressive pathology of cysts associated with defective ciliary signaling, versus the less aggressive phenotype seen with loss of cilia? This should be better developed, and supported with some experiments. We thank the reviewer for pointing out these studies supporting the role of inflammation in different models of cystic kidney disease. We have included the indicated references in the discussion section. Since we have not studied cyst formation in mice lacking cilia, such as Kif3a targeted mice, we cannot comment on how cysts form in these models. We now show new data from Pkd1/Ccl2 targeted mice, which recapitulate the phenotype of Pkd1/Kif3a targeted mice and demonstrate that interference with tubular CCL2 expression decreases macrophage infiltration and cyst growth (Fig 7 and Fig EV5). These findings strengthen the concept that lack of PKD1 interferes with the control of a ciliary CCL2 inducing signal, leading to inflammation and more rapid cyst growth. 5. It is also of interest that Ta and colleagues using a model of NPHP9-dependent cystic kidneys have shown that sirolimus inhibits the LKB1 regulated protein mtorc1, reducing disease pathology but not CCL2 expression (PMID: ). This also may inform mechanism, and could add to the discussion. We thank the reviewer for mentioning this reference. It is interesting that in their model CCL2 regulation was independent of rapamycin, suggesting that it does not occur downstream of mtor. Since the model in the cited paper is based on an NPHP9 mutation we decided against discussing it in our manuscript. We have not tested NPHP9 and find that not all of the NPHPs we examined bind to LKB1. We suggest leaving the role of NPHP9 to be defined in future studies. Minor point. 1. Figure 5B. Could the authors enlarge the images demonstrating the localization of LKB1 in the primary cilium? We have now added enlargements of cilia in this subfigure (Fig 5D). Referee #2: General Summary In this manuscript, the authors conclude that Pkd1 and an Lkb1-based module comprising Lkb1/NPHP1/ANKS3/NEK7 form a functional unit that inhibits a ciliary Ccl2 inducing signal, which otherwise leads to inflammation and PKD progression in Lkb1 or Pkd1 deficient mice. They based this conclusion on the following observations. 1) A proteomic screen of Lkb1 precipitates identified Stradα and 30 other proteins that included the NPHP1-binding ANKS3 as well as NEK7 (known to interact with ANKS3); 2) NPHP1 and Lkb1 functionally interact in cystogenesis in zebrafish; 3) Ccl2 (Mcp1) and ANKRD1 are upregulated based on comparisons of microarray analysis in Lkb1 mutant kidneys at 5 weeks with RNA seq. in Lkb1-silenced MDCK; 4) Ccl2 is increased in Lkb1ΔTub kidneys at 5 weeks and in Lkb1-silenced MDCK, and neutrophils and CCR2+ Mø (but not T, B or total F4/80hiCD11b+ Mø) are increased in Lkb1ΔTub kidneys at 10 weeks; 5) Conditional silencing of Stradβ (but not Stradα) in MDCK increased Ccl2, and coablation of cilia together with Lkb1 prevented the increase in Ccl2 in Lkb1-silenced MDCK. Finally the authors showed that Pkd1 co-ip with Lkb1 and both Strads (α>>>β) in HEK293, and that ablated cilia in Pkd1 null MDCK cells or in 12 week ipkd1δtub ikif3aδtub mice decreased Ccl2 expression and kidney CCR2+ Mø. Specific concerns (can also be transmitted to the authors) Despite the extensive data presented, the paper do not convincingly support the authors major conclusion that a ciliary Pkd1 Lkb1-based complex regulates Ccl2 and that this Ccl2 is the elusive cilia-dependent cyst activating signal. First, Lkb1 expression in the kidney is massive (Fig. 1A) and its role cannot be accounted for by expression in cilia, unless its knockdown is limited to cilia, which has not been done. European Molecular Biology Organization 6

7 Cilia proteins are synthesized in the protein machinery of the cell and then transported to cilia by different routes, whereby protein access to cilia is restricted in a number of different molecular mechanisms that are only partly understood. A knock-down limited to cilia is therefore not technically possible. As the reviewer mentions above under point 5, we find that depletion of STRADb, but not STRADa affects CCL2 release. STRAD is a necessary interactor of LKB1 required for its function. The fact that STRADa is the cytosolic binding partner of LKB1, yet is dispensable for CCL2 regulation, clearly demonstrates, that cytosolic LKB1 is not involved. Conversely, the fact that targeting of STRADb, the ciliary binding partner required for its localization (PMID and our data Fig 5E), leads to increased CCL2 expression, is solid evidence supporting our claim that LKB1 acts in the cilium to regulate CCL2. Second, it is unclear that the Lkb1-based module comprising Lkb1/NPHP1/ANKS3/NEK7 or Stradβ is cilia-limited as none of these proteins are expressed ONLY in cilia. We do not claim that this complex is limited to cilia. Yet, our findings from cell based as well as animal models support that this complex acts in cilia, see for instance the points raised above. Third, the relevance of the zebrafish data is suspicious as zebrafish pronephros does not have nonmotile primary cilia that are found in mammalian kidneys. Pronephric cyst formation in zebrafish is a widely established model in the ciliopathy field. It is used for genetic epistasis experiments, but not as a model for cell biology. This means that the experiment demonstrates a genetic interaction between LKB1 and NPHP1, but not a mechanism. The fact that cilia in the pronephros are motile does not exclude sensory functions. Forth, identification of Ccl2 was based on the unusual comparison of microarray analysis in Lkb1 mutant kidneys at 5 weeks with RNA seq in the Lkb1-silenced canine cell line MDCK; The strength of our chosen comparison between LKB1 mutant kidneys and LKB1 silenced cells lies in the advantage that we were able to discard regulated transcripts from the kidney that came from outside of renal epithelia. A case in point is CCL2, which might have been discarded based on the assumption that it is derived from infiltrating immune cells. Including transcriptional data from LKB1 silenced cells in the analysis indicated that CCL2 is expressed in the tubular epithelial cell as a consequence of LKB1 loss. no protein expression has been determined in any experiment presented, We agree that this is an important point. We have now analyzed CCL2 from supernatants of LKB1 depleted and PKD1 null cells by ELISA and find that CCL2 protein is secreted in higher amounts compared to control cells (Fig 3E and 6C). This result is in tune with our in-vivo findings that CCL2 expression in LKB1- and PKD1-knock-out kidneys leads to macrophage recruitment and that deletion of CCL2 strongly ameliorates inflammation (new data: Fig 7 and Fig EV5), suggesting that CCL2 expression as measured by quantitative RT-PCR translates into secretion of this chemokine to yield its function in tissue homeostasis. and cyst fluid in ADPKD patients contains several other cytokines. We do not argue that other cytokines have no role in the pathogenesis of ADPKD. We now present new data showing that co-targeting CCL2 with PKD1 in the kidney ameliorates the phenotype (Fig 7 and Fig EV5). This supports our model that tubular secretion of CCL2 plays a patho-physiologically meaningful role in ADPKD. Fifth, the authors show that infiltration of CCR2+ macrophages (Fig 4D) but not total macrophages (Suppl. Fig. 6 J) is increased in Lkb1Δ Tub kidneys at 10 weeks. How significant is this subpopulation? What percent it represents of the total infiltrating Mø? This is indeed an important question and we thank the reviewer for raising this. Our data show that about 20% of total kidney macrophages (defined as F4/80hiCD11b+Ly6C-Ly6G-) at homeostasis express CCR2. In Lkb1 deficient mice this population increases to about 30%. We have added this European Molecular Biology Organization 7

8 information in Appendix Fig S5L. This represents a biologically relevant fraction of macrophages and supports the observations on the Pkd1/Ccl2 targeted mice. Sixth, Stradβ does not bind Lkb1 in the absence of MO25, We do not show that STRADb does not bind LKB1 in the absence of MO25. and PKd1 associates much better with Stradα (reportedly not involved in Ccl2 expression). The reviewer presumably refers to Fig. 6A. This experiment tests the interaction of PKD1 with STRAD in an overexpression system (tagged constructs in HEK 293T cells). It demonstrates that PKD1 can interact with STRADa as well as STRADb. Since the interaction is not between endogenous proteins, the strength of interaction does not reflect physiological amounts. Please note that the input of Flag-PC1 is stronger in the STRADa lanes compared to STRADb lanes, which may explain why more STRADa is detected in the IP compared to STRADb. Our findings do not rule out that PKD1 interacts with STRADa outside the cilium in a functional way. Finally, how do these data explain the differences in clinical presentation of NPHP where inflammation and subsequent fibrosis appear early and are much more prominent than cysts vs. ADPKD if Ccl2 is the common factor in cystogenesis? This is an important point and has been raised by other reviewers as well. Our data suggest that increased CCL2 drives macrophage numbers. This can have different effects depending on context. In ADPKD, macrophages promote cyst growth in epithelia that are prone to cystogenesis due to other effects of PKD1 loss which our manuscript does not address. In other words, the difference between the models is that the PKD1 mutation confers a pro-cystogenic phenotype on epithelial cells that does not occur in LKB1 mutated kidneys (and presumably not in NPHP1 mutation either). The hypothesis that cyst formation occurs through different mechanisms in ADPKD and NPHP has been raised by others (PMID: ). However, in both cases macrophages drive fibrosis and nephron loss. In addition cyst growth (not cyst initiation) is fueled in ADPKD. While macrophages and fibrosis occur early in ADPKD, their presence is hidden by the much more prominent cysts. We have now tested this hypothesis by targeting PKD1 together with tubular CCL2 in the kidney. This strongly ameliorated cyst growth (Fig 7 and Fig EV5) but did not prevent cyst formation altogether and thus supports our hypothesis. With respect to ADPKD, our findings are in line with previous data by others: (1) macrophages drive cyst growth in genetic models of PKD (PMID: ). (2) ablation of cilia in ADPKD models ameliorates the phenotype (PMID: ). Parts of the results section and the discussion have been rewritten to accentuate these issues. Referee #3: This is an interesting manuscript that addressed the role of ciliary LKB1 in regulating chemokine signalling, macrophages and tissue homeostasis in polycystic kidney disease (PKD). Viau et al found that deletion of LKB1 in renal tubular cells not only disrupted renal morphology and function but also induced the expression of chemokine CCL2, resulting in the recruitment and activation of macrophages in kidneys. This study provided integrated evidences to link ciliary LKB1 to renal inflammation, in which LKB1 in the primary cilia of renal tubular epithelial cells was involved in macrophage infiltration and played an important role in the progress of ADPKD or NPHP. However, several concerns need to be addressed. Major concerns: 1. In Fig.1 the authors provided evidences that deletion of LKB1 affected kidney morphology and function. Some panels in this figure are confused and need to be reorganized by adding the missing controls. In Fig. 1C, the kidneys sizes were not diminished as mentioned in MRI analysis. In Fig. 1F, is it to be implied that the control kidney sections (for 14 and 23 weeks) do not change from the 5 week time point? We thank the reviewer for pointing out that in Fig 1C the MRI images are not representative of smaller kidneys in Lkb1 mutant animals. This is due to the fact that the section shown does not always represent the maximum diameter. We have now calculated the kidney volumes obtained in European Molecular Biology Organization 8

9 the 16 week MRI analyses and normalized them for body weight (Appendix Fig S1A). This shows a trend towards smaller kidneys in the mutants which corresponds to the actually measured kidney/body weights (Fig EV1K please note that the figure terminology was changed according to journal standards). The effect is less pronounced in the MRIs, presumably due to hydronephrosis in the mutant kidneys which add to the kidney volumes of Lkb1 mutants. We have also changed the displayed section in Fig 1C to better represent these findings. The reviewer is correct that the 5 week wild type kidney in Fig. 1F is smaller than they would be at 14 and 23 weeks (5 weeks: 92mg, 14 and 23 weeks: 140mg). In this figure the wild type kidney is displayed for reasons of morphology, not size. As the kidney texture does not change at later time points and for reasons of space we display only the 5 week wild type control. 2. In Fig. 2, the authors found that LKB1 interacted with ANKS3, NPHP1 and NEK7 in kidney medulla and HEK 293 cells. In Fig. 5, depletion of NPHP1, ANKS3 or NEK7 resulted in upregulation of Ccl2 in MDCK cells. Does knockdown of NPHP1, ANKS3, and NEK7 affect the protein level and activity of LKB1? To address this question we performed western blots for LKB1, AMPK, and p-ampk in cells after knock-down of NPHP1, ANKS3 and NEK7. We did not find altered expression of LKB1 or different amounts of p-ampk (Appendix Fig. S4 A C and I - K). 3. In Fig. 6, the authors found that LKB1 also interacted with polycystin-1 (PC1) to form a functional unit to contribute to the progression of disease. Does deletion of PC1 affect the expression and activity of LKB1, ANKS3, NPHP3 as well as NEK7 in the PKD1 mutant cells and kidneys compared to the control? Does PKD1 mutation induce the decrease of LKB1 and the module proteins, leading to increase of Ccl2 expression in in PKD1 mutant cells and the subsequent recruitment of macrophages in PKD1 mutant kidneys. In addition, the effect of double knockout of Pkd1 and LKB1 on cyst growth, the expression of Ccl2 and the recruitment of macrophages in kidneys need to be added in Fig. 6 to strength the role of LKB1 in PKD. We examined but did not find consistent differences in the expression of ANKS3, NEK7 or NPHP1 in PKD1 knock-out cells (new data: Appendix Fig S4D, F - H). LKB1 expression was not decreased and no difference was observed in p-ampk (Appendix Fig S4D, E, L). In PKD1 targeted kidneys no difference was observed with regards to the expression of LKB1, ANKS3, NEK7 or NPHP1 (Appendix Fig S4M - P). This leads us to conclude that altered expression of members of the CCL2 regulating complex do not account for the observed increase in CCL2 expression in PKD1 mutated cells or kidneys. We thank the reviewer for asking about the simultaneous deletion of PKD1 and LKB1. This is an instructive experiment, as it may delineate if PKD1 is required for the activity of the LKB1 regulatory complex or not. We engineered PKD1 knock-out cells to express tetracycline inducible shrna against LKB1 (Appendix Fig S3S). We found CCL2 expression to be increased in these cells at a similar level compared to PKD1 knock-out cells with intact LKB1 (Fig EV3H). The lack of a further increase of CCL2 upon depletion of LKB1 in PKD1 knock-out cells suggests that loss of PKD1 fully deregulates CCL2 and indicates that LKB1 requires PKD1 for its CCL2 regulating function. We did not perform dual targeting of LKB1 and PKD1 in mice. Setting up this experiment would take about 1 year and would far exceed the time allotted for revisions by EMBO Journal. To strengthen the hypothesis that PKD1 interacts with the CCL2 regulatory complex in a pathologically relevant way, we completed an experiment where PKD1 and CCL2 were inactivated in the mouse kidney using the same tubule specific Cre-driver. Indeed, compared to PKD1 inactivation alone, this lessens macrophage recruitment and ameliorates the phenotype, similar to the effect of PKD1/KIF3a targeted mice (Fig 7 and Fig EV5) and thus links PKD1 and CCL2 to disease progression in-vivo. 4. The authors provided evidence that LKB1/NPHP1/ANKS3/NEK7 complex controls Ccl2 expression. Is Ccl2 expression in PKD1 mutant also LKB1 dependent? This is not the case as is indicated in point #3. 5. What's the mechanism of LKB1 in regulating the expression of Ccl2. It is unclear how LKB1 European Molecular Biology Organization 9

10 inhibits Ccl2 expression at homeostasis or upregulates Ccl2 expression in Pkd1 mutant kidneys. Please justify. LKB1 phosphorylates a number of downstream targets, AMPK being the most prominent. Ciliary LKB1 has been shown to phosphorylate AMPK at the basal body where the cilium is anchored. We now treated LKB1 depleted cells with AICAR, a stabilizer of AMPK phosphorylation. This led to a strong decrease in CCL2 expression and indicated that the LKB1 effect on CCL2 is mediated via AMPK (Fig EV3E). We find that removal of the cilium decreases CCL2, irrespective of LKB1, suggesting that the cilium receives a positive signal inducing CCL2. The question of the nature of this signal, it's physiological and pathophysiological context, is very interesting and will need to be addressed in further studies. Minor concerns: In Fig. 6, it should be "Polycystin 1 (PC1)" interacts with LKB1 but not "PKD1", which is the gene name but not the protein name. The size of PC1 in Fig. 6a needs to be double checked. Thank you for bringing this to our attention. It was changed. The predicted molecular weight of PC1 is 460 kda and is usually visible above the 250 kda marker. The distance can seem less on a gradient gel, as is the case here. 2nd Editorial Decision 27 April 2018 Thank you for submitting a revised version of your manuscript. It has now been seen by two of the original referees (#1 and 3) whose comments are shown below. As you will see they both find that all criticisms have been sufficiently addressed and recommend the manuscript for publication. However, before we can officially accept the manuscript there are a few editorial issues concerning text and figures that I need you to address. Please include your response to the following point raised by the referee #2 in the manuscript text. Below is the point of the referee #2: Referee point: Forth, identification of Ccl2 was based on the unusual comparison of microarray analysis in Lkb1 mutant kidneys at 5 weeks with RNA seq in the Lkb1-silenced canine cell line MDCK; Your response: The strength of our chosen comparison between LKB1 mutant kidneys and LKB1 silenced cells lies in the advantage that we were able to discard regulated transcripts from the kidney that came from outside of renal epithelia. A case in point is CCL2, which might have been discarded based on the assumption that it is derived from infiltrating immune cells. Including transcriptional data from LKB1 silenced cells in the analysis indicated that CCL2 is expressed in the tubular epithelial cell as a consequence of LKB1 loss. Please make sure to address the remaining concerns raised by the referee #3 in both point-by-point response file and corresponding changes in the text Referee #1: Based on the revisions, the manuscript is significantly improved, and has thoroughly and substantively addressed points I originally raised. Referee #3: This study tried to link ciliopathy to inflammation through a LKB1. The authors found that deletion of LKB1 in renal tubular cells induced the expression of chemokine CCL2, which resulted in the recruitment and activation of macrophages in kidneys, contributing to ciliopathy. The authors addressed the reviewers' concerns point-by-point in the rebuttal letter but some concerns were not clearly answered. In addition, Flowers et al., has published a paper entitled "Lkb1 deficiency confers European Molecular Biology Organization 10

11 glutamine dependency in polycystic kidney disease" (Nature Comm. Volume 9, 814, 2018). The major concerns still need to be addressed: 1. Regarding the downstream signaling of LKB1, the authors originally stated that "deletion of Lkb1 does not result in induction of AMPK or mtor by microarray, and hence some other event must be important for renal pathology", and now they thought "that phosphorylation of AMPK by LKB1 at the basal body may affect CCL2 in addition to mtor." Please clarify. 2. It seems that deletion of Pkd1 alone is enough to affect the expression of Ccl2. If this is the case, a. how to explain the upregulation of Ccl2 in ADPKD if the expression and activity of LKB1 didn't change. b. did it mean that the upregulation of Ccl2 is not LKB1 dependent in ADPKD? c. what is the functional role of the interaction between PKD1 and LKB1? 3. As the authors mentioned there are two independent pathways to regulate the Ccl2 expression in PKD. One is Pkd1 dependent, and another is LKB1/NPHP1/ANKS3/ NEK7 unit dependent, which are not associated and function in parallel. The authors stated that "The lack of a further increase of CCL2 upon depletion of LKB1 in PKD1 knock-out cells suggests that loss of PKD1 fully deregulates CCL2 and indicates that LKB1 requires PKD1 for its CCL2 regulating function." It is not clear how these two pathways works in ADPKD and in NPHP. Whether LKB1 works in NPHP as the authors proposed in ADPKD. 4. It was pointed out that the mechanism of CCL2 production control by LKB1 has been already reported for cancer cells (PMID: ), which suggested that LKB1 could regulate CCL2 expression with mechanisms independent of PC1. Since LKB1 should be presented in both ciliated and non-ciliated cells, whether LKB1 in non-ciliated cells, such as IFT88 and Kif3a knockout cells, also regulated the expression of Ccl2. 5. The authors found that removal of the cilium decreases CCL2, irrespective of LKB1, suggesting that the cilium receives a positive signal inducing CCL2. In this case how to explain the regulation of Ccl2 by LKB1 in the mouse models published by Somlo's lab (Nat Genet. 45(9): ). 6. The zebrafish data did not provide any support to the study in mammalian kidneys and in this study. 2nd Revision - authors' response 13 May 2018 This study tried to link ciliopathy to inflammation through a LKB1. The authors found that deletion of LKB1 in renal tubular cells induced the expression of chemokine CCL2, which resulted in the recruitment and activation of macrophages in kidneys, contributing to ciliopathy. The authors addressed the reviewers' concerns point-by-point in the rebuttal letter but some concerns were not clearly answered. In addition, Flowers et al., has published a paper entitled "Lkb1 deficiency confers glutamine dependency in polycystic kidney disease" (Nature Comm. Volume 9, 814, 2018). The major concerns still need to be addressed: 1. Regarding the downstream signaling of LKB1, the authors originally stated that "deletion of Lkb1 does not result in induction of AMPK or mtor by microarray, and hence some other event must be important for renal pathology", and now they thought "that phosphorylation of AMPK by LKB1 at the basal body may affect CCL2 in addition to mtor." Please clarify. The reviewer refers to a statement in the response to reviewer 1 in the first revision. Phosphorylated AMPK inhibits mtor through the TSC complex, but also has mtor independent effects through an enlarging group of AMPK targets (please see PMID: for a current review). We interpret our data in a way that LKB1 may phosphorylate AMPK, but that this leads to mtor independent events regulating CCL2. 2. It seems that deletion of Pkd1 alone is enough to affect the expression of Ccl2. If this is the case, European Molecular Biology Organization 11

12 a. how to explain the upregulation of Ccl2 in ADPKD if the expression and activity of LKB1 didn't change. b. did it mean that the upregulation of Ccl2 is not LKB1 dependent in ADPKD? c. what is the functional role of the interaction between PKD1 and LKB1? a) In our western blots we measure predominantly cytosolic LKB1 activity. The preserved activity as measured by levels of phosphorylated AMPK is consistent with the fact that the level of LKB1 doesn't change. This observation does not preclude, that LKB1 activity in the cilium is disturbed in the absence of PKD1. We now clarified this point in the results section by changing "LKB1 was not affected" into "total LKB1 expression was not affected" (section: 'The LKB1/NPHP1/ANKS3/NEK7 complex controls CCL2 expression'). b) This means that in our model PKD1 is required for CCL2 down-regulation in the cilium. Loss of PKD1 or LKB1 prevent the down-regulation of CCL2, leading to increased expression. This is illustrated in Fig. 8C and D. c) The functional interaction is that of ciliary CCL2 control. See also the answer to the next question. 3. As the authors mentioned there are two independent pathways to regulate the Ccl2 expression in PKD. One is Pkd1 dependent, and another is LKB1/NPHP1/ANKS3/ NEK7 unit dependent, which are not associated and function in parallel. The authors stated that "The lack of a further increase of CCL2 upon depletion of LKB1 in PKD1 knock-out cells suggests that loss of PKD1 fully deregulates CCL2 and indicates that LKB1 requires PKD1 for its CCL2 regulating function." It is not clear how these two pathways works in ADPKD and in NPHP. Whether LKB1 works in NPHP as the authors proposed in ADPKD. In fact, we do not claim that there are two different pathways regulating CCL2 in PKD. Instead, we argue that there is a common mechanism: PKD1 is part of a ciliary multi-protein complex consisting of PKD1, NPHP1, LKB1, STRADb, ANKS3 and NEK7. Under physiological conditions, this complex negatively regulates a CCL2 inducing signal that is sensed by cilia, and the nature of which is unknown. In the pathological setting CCL2 is deregulated. We demonstrate this fact invitro in loss of function models for each of the components of the complex. In-vivo, this is shown in LKB1 targeted kidneys - a situation that has no correlate in man and PKD1 targeted kidneys. These two animal models confirm our in-vitro findings in vivo for two members of the regulatory complex. Our model predicts that the same is true in NPH patients with mutations in NPHP1. Future studies will need to determine if this model is true in patients with NPHP1 mutations. We have stated this model of a single ciliary complex regulating CCL2 in several parts of the manuscript: abstract: "we identify a complex of LKB1, a ciliary kinase, and several ciliopathy proteins including NPHP1 and PKD1. Deletion of LKB1 or PKD1 in renal tubules elevates CCL2 expression." results: paragraph 'PKD1 interacts with LKB1 and STRAD': "these findings.. demonstrate that a physical and functional interaction exists between PKD1 and the LKB1 module. Our findings suggest that PKD1 and the LKB1 module form a functional unit to inhibit a ciliary CCL2 inducing signal and that PKD1 is required for its function." discussion, first paragraph: "The signal inducing CCL2 is regulated by a ciliary module involving LKB1, the ciliopathy proteins NPHP1, PKD1, and ANKS3, as well as NEK7." Figure 8 gives a detailed depiction of our model and a summary of our findings. 4. It was pointed out that the mechanism of CCL2 production control by LKB1 has been already reported for cancer cells (PMID: ), which suggested that LKB1 could regulate CCL2 expression with mechanisms independent of PC1. Since LKB1 should be presented in both ciliated and non-ciliated cells, whether LKB1 in non-ciliated cells, such as IFT88 and Kif3a knockout cells, also regulated the expression of Ccl2. The question here isn't clear. We show in the manuscript that removal of cilia by targeting IFT88 or KIF3a in LKB1 knock-down cells prevents a CCL2 increase (Fig. 5H). We interpret this finding such that LKB1 does not regulate CCL2 in this case, because the CCL2 inducing signal cannot be sensed in the absence of cilia. We do not have any data on cancer cells. European Molecular Biology Organization 12

13 5. The authors found that removal of the cilium decreases CCL2, irrespective of LKB1, suggesting that the cilium receives a positive signal inducing CCL2. In this case how to explain the regulation of Ccl2 by LKB1 in the mouse models published by Somlo's lab (Nat Genet. 45(9): ). Again, the question here isn't clear. The mentioned paper does not show data regarding, or discuss, CCL2 or LKB1. As far as the inducible PKD1- and PKD1/KIF3a knock-out models are concerned, we find that in the absence of PKD1 CCL2 is deregulated in the following way: The inhibitory complex in the cilium does not function in the absence of PKD1. This leads to deregulation of the inhibitory signal as shown in Fig.8 D, irrespective of LKB1. In the absence of cilia, CCL2 is not elevated due to the inability of the cell to sense the inducing signal. This is illustrated in Fig. 8E. 6. The zebrafish data did not provide any support to the study in mammalian kidneys and in this study. We disagree. The zebrafish data contribute to our model in that they show a functional interaction between LKB1 and NPHP1 in-vivo. While the findings in the Lkb1 targeted kidney are strongly reminiscent of the clinical phenotype of human patients with NPHP1 mutations, we have no basis to to link the mouse phenotype to NPHP1, other than that these two proteins interact in vitro. Zebrafish are an established model in ciliopathy research. The fact that NPHP1 and LKB1 interact in this system supports our model that these two proteins interact in-vivo in a physiologically important way. European Molecular Biology Organization 13