STK11 (LKB1) missense somatic mutant isoforms promote tumor growth, motility and inflammation

Elucidating the contribution of somatic mutations to cancer is essential for personalized medicine. STK11 (LKB1) appears to be inactivated in human cancer. However, somatic missense mutations also occur, and the role/s of these alterations to this disease remain unknown. Here, we investigated the contribution of four missense LKB1 somatic mutations in tumor biology. Three out of the four mutants lost their tumor suppressor capabilities and showed deficient kinase activity. The remaining mutant retained the enzymatic activity of wild type LKB1, but induced increased cell motility. Mechanistically, LKB1 mutants resulted in differential gene expression of genes encoding vesicle trafficking regulating molecules, adhesion molecules and cytokines. The differentially regulated genes correlated with protein networks identified through comparative secretome analysis. Notably, three mutant isoforms promoted tumor growth, and one induced inflammation-like features together with dysregulated levels of cytokines. These findings uncover oncogenic roles of LKB1 somatic mutations, and will aid in further understanding their contributions to cancer development and progression.


5.
The clinical significance of these particular mutations is unclear. In searching through the Lung adenocarcinoma TCGA, I did not see any examples of R87K, Y49D, or G135R and just a few cases of D194Y. In looking at common inherited mutations in Peutz Jager syndrome I did not see these mutations occur commonly (e.g. Amos et al, J Med Genetics 2004). The authors should clarify: What is the rationale for these particular mutations? What percentage of lung cancer or other cancer patients are affected by these particular mutations? This data should be included in the manuscript of supplement. 6. Comparison of gene expression changes (e.g. Fig 4, Supp Fig S3) seems to be done comparing triplicate copies of a single cell line expressing the particular mutant. This makes it difficult to know to what extent the changes in gene expression are truly related to the particular mutation or an artefact specific to the cell line expressing a particular mutant at a particular expression level (also note that A549 have other mutations such as KEAP1 and may behave differently at times than other LKB1deficient cells). Given that there are publicly available datasets and robust gene expression signatures for LKB1-deficient tumors (e.g. Kaufman et al JTO 2014), it would be more appropriate to A) focus on genes known to be regulated by LKB1 in clinical datasets, and then B) Assess whether those particular LKB1-rgulated genes are differentially regulated by different mutants, ideally using clinical data or at least more than one model expressing a particular mutation. There are similar concerns regarding the comparisons of the secreted proteins. 7. The significance of the growth rates in the in vivo models is unclear because we see that the growth rates of the control group (no doxycycline and hence no mutant expression) should be similar across the pairs (since it should be an identical LKB1 null background); in fact there is wide variability with about 6 fold growth in the R87K experiment ( Fig 6C) and more than 20 fold growth in the Y49k experiment. This suggests that there could be some leakiness in the dox promoter or that there are simply differences in growth rates across the different stable cell lines. Therefore basing comparisons in the growth with or without dox when the baseline is different for each does not seem reasonable. It would be more appropriate to see if similar mutation-specific differences are observed across different models expressing the particular mutant. 8. LKB1 is a master kinase that activates several downstream pathways by phosphorylating different substrates. In term of characterization of loss of function and kinase activity in the different mutant isoforms, data provided in the manuscript should be further supported by assessing LKB1 downstream targets such us AMPK/mTOR (cell proliferation and metabolism control) and/or MARK/SAD/SIK (cell polarization) should be performed. Recent papers (Hollstein et al, Cancer Discovery 2019 and Murray et al Cancer Discovery 2019) are highly relevant in this regard and should be cited. 9. In vitro assays in figure 1, as well as, in vivo experiments shown in figure 6 revealed a potentially interesting result: expression of the G135R and D194Y mutations in an LKB1 deficient cell line (A549), accelerated tumor growth compared with parental cells, which already has a high proliferative phenotype due to lack of LKB1 expression. This would suggest that the mutations could activate pathways beyond simply LKB1 loss. This should be evaluated in a different model and the authors should discuss possible explanations for this phenomenon.
The finding that some mutants may accelerate tumor growth (e.g. Figure 6D) compared with the null background does not seem to fit with the earlier in vitro experiments showing that the mutants did not suppress tumor growth but did not accelerate it either (figure 1), which would be more consistent with the role of a tumor suppressor. The statement on page 16 "Thus, the in vitro tumor suppressor capabilities of the investigated LKB1 mutants were reflected in vivo." is not accurate; there is no significant suppression for the R87K and there is significant acceleration in Fig 6B, D, and E. The authors should discuss the differences in these findings.
10. In figure 1, data revealed that Q135R and D194Y mutations increased cell proliferation when expressed in a LKB1 deficient cell line (A549), while R87K and wt isoforms reduced cell growth. By contrast, Y49D mutant isoform seems not to have effects in term of cells proliferation. In figure 2, by contrast, Q135R and D194Y only partially lost kinase activity, while Y49D also significantly reduced its kinase activity and this isoform seems to lose ability to interact with STRADα. On the other hand, R87K (wt-like isoform), Y49D and D194K, but not G135R display a shorter protein half-life. These data are inconsistent with the proliferative phenotype shown in figure 1. The authors should discuss possible explanations for these findings.
Minor issues: 1. The statement (page 20) that "LKB1 Y49D showed signs of inflammation and disorganized angiogenesis (hemorrhagic) confirming the role of LKB1 in regulating cytokine production and inflammation" is an overstatement because the models were tested in immunocompromised mice so inflammation could not be accurately assessed; furthermore, hemorrhage is not an established marker for inflammation. 2. In Fig 1E: panel with D194Y at 48h + Dox has dark blue bar (G1 arrest) in the middle of the other two blue bars, whereas elsewhere the dark blue bar is at the bottom. Should the bar be dark blue (meaning G1 arrest), in which case it should be at the bottom, or medium blue? Also, elsewhere in the figure, there are 4 different shades of blue but only 3 in the legend-one color is nearly black. Please clarify the colors and labels. 3. The concluding statement in the Discussion is not adequately supported by the data, particularly the effects on immune modulation given the immunocompromised mouse models: "In summary, we show that beyond the role of the non-mutated protein as a tumor suppressor, missense LKB1 somatic mutations could contribute to tumor development and/or progression by modifying not only intrinsic cell capabilities such as proliferation, motility or adhesion but also the tumor microenvironment, affecting inflammatory responses and likely the immune system. "These experiments could be repeated in syngeneic murine models to better get at the impact on the microenvironment. In addition, public data could be mined (e.g. TCGA using CIBERSORT) to see if it supports the association of different patterns associated with different mutations. 4. Analysis of cells cycle phase distribution is shown in Figure 1F to support higher or lower rate of proliferation across the different mutants. Some issue with these data: -First, quantification of all cell cycle phases together should be close to 100%. Are the authors missing any population? For example, is SubG1 peak (Dead cells which display lower probe staining) quantified? Please, reanalyzed carefully these data.
-Cells cycle analysis are typically performed when cells are growing exponentially to avoid include additional effects that could interfere in the cell cycle progression. Could the authors explain why G1 phase percentage is higher in -Dox treatment at 48h compared with 0h for WT and Y49D isoform? If cell cycle analysis was performed when cells are growing exponentially, G1 phase percentage at 48h should be similar to 0h. This data seems to indicate a G1 arrest at 48h due to low nutrient availability or confluent state rather that expression of LKB1 wt. Initial number of plated cells should be recalculated to allow exponential grow still at 48h. -Finally, averages for 2 or 3 independent experiments should be included. 5. In figure 1D and Figure S1 C authors show the quantification of clonogenic assay performed in the different mutated isoform. Are these differences statistically difference? If it does, please include this data. 6. Figure 6C is cited in the text before than Figure 6B. 7. In Discussion section, second paragraph the sentence "Thus, selected missense LKB1 selected mutations…" should be corrected by deleting the second "selected".
This manuscript by Paula Granado-Martinez et al. focuses on the functional study of STK11 (LKB1) missense somatic mutant isoforms such as LKB1Y49D, LKB1R87K, LKB1G135R and LKB1D194Y in cancer. They performed several experiments including omics analysis such as secretome to reveal the functions of these mutant isoforms. This is interesting work to reveal the importance of the STK11 (LKB1) missense somatic mutant isoforms. Before it can be published in Communications Biology, I have several suggestions as described below.
1. The statistics analysis should be performed in Figure 1D, 3A and 3E. 2. The supplementary tables for proteome should be described the protein full names and the detail information about mass identification. Since many errors happened in protein identification even using software such Proteome Discoverer, to check the mass peaks carefully and list the results are important in proteome field. 3. I strongly suggest that the authors should deposit the proteome data to public database such as ProteomeXchange (http://www.proteomexchange.org/).

Reviewer #1 (Remarks to the Author):
This is a timely and well-written report on the functional characterization of four STK11 mutations in tumor cells. The manuscript is interesting because STK11(LKB1)is often somatically mutated (or deleted) in NSCLC and its alterations have been associated with poor response to immunotherapy with ICI (as well as to metabolic alterations). Since LKB1 mutations are not concentrated in hotspots, there are many cancer-associated mutations which remain poorly investigated in terms of their impact on tumor cell functions. A few points which deserve to be improved:  (Esteve-Puig et al., 2009;Gonzalez-Sanchez et al 2013;Esteve-Puig et al., 2014). Due to the multitask functions of LKB1, we chose these four missense LKB1 somatic mutations because we are interested in studying different functional parts of the protein contextualized in human cancer. The great majority of missense mutations in STK11, except for D194 residue, do not represent hotspots.
D194N mutant was described to be kinase dead. However, the functional consequences in cancer of this mutant were mostly unknown. This residue is located in the ATP binding cleft. G135 residue is also located in the same functional region. Thus, G135R mutant was selected to be compared with D194Y since it was located in a similar 3D functional location (ATP binding cleft). Y49 residue was embedded within a β-sheet in the LKB1 N-lobe close to the 3D interaction region with STRAD. Since LKB1 subcellular localization and kinase activity is STRAD-binding dependent we were interested in study the possible consequences of this mutation. Finally, R87 is exposed at the surface of the molecule being susceptible to be modified or interact with other molecules. Moreover, the change for the lysine is particularly interesting because is not very disruptive respect the charge, however Lys and Arg have different H-bonding capabilities and hydration free energies, and they can be post-translationally modified differently. Importantly, all selected mutations were initially detected in tumor samples and were predicted to be oncogenic or likely oncogenic (polyphen or OncoKB team TCGA). We have added a couple of sentences in the introduction explaining our motivation selecting the mutants. We hope this explanation will satisfy the reviewer concerns.
2. With regard to the study design and the methodology used here, one concern is that by their lentiviral vector-mediated approach the Authors could actually over-express LKB1 mutants, compared with levels found in tumor cells bearing these mutations. Some biological effects measured here could be "dosage dependent". Can the Authors provide some evidence of the relative expression levels of LKB1 in their tumor cells compared with "physiological" levels detected in LKB1-mutant tumor cells?
This is a very interesting observation and at the same time is a complex and philosophical issue. We completely agree with the reviewer when state that in order to be able to interpret the results, we have to work within the "physiological range of the protein". To sustain the expression of the protein at the "physiological range we titrated both, the concentration of the doxycycline and the time of induction in all cell lines (Supplementary figure 1). However, the physiological levels for normal cells varies according to the type of cell (melanocyte, hepatocyte…etc 3. Page 13, chapter "Vesicle trafficking regulatory molecules." It is not clear which tumor cells are used in this set of experiments. Please correct the text accordingly.
We are sorry for not include this information. This have been corrected accordingly.
4. With regard to the translational relevance of these findings, is any of these 4 mutations being reported in patients treated with ICI? Do they associate with resistance to ICI?
As above stated missense mutations represent about 30% of the genetic mutation of STK11. Among those 50% of the punctual mutation described in lung cancer, melanoma and cervix listed in TCGA and COSMIC databases have been described in patients subjected to immune-checkpoint inhibitors. We believe that the message in the manuscript will contribute to clarify the potential role of these mutation in patients.
5. LKB1 is best known for its effects on several metabolic pathways. It is surprising to see that the Authors did not perform any metabolic assay with these LKB1 variants. At a minim, Seahorse analysis should be performed and glycolysis/OXPHOS activity in the cell lines expressing the mutants compared with LKB1 WT or null cells should be shown. Also, since LKB1 mutations have been associated with response to Metformin in vitro, could the Authors evaluate this in their experimental system? In my opinion, these results would complete their comprehensive functional characterization of these mutants.

This is a very good question. We performed an OCR and ECAR seahorse analysis.
Overall, we did not observe significant differences in OCR among the wild type isoform All this information has been added to the manuscript accordingly and we hope this will satisfy the reviewer concerns.
Reviewer #2 (Remarks to the Author): In the manuscript entitled "STK11 (LKB1) missense somatic mutant isoforms promote tumor growth, motility and inflammation" authors explore the biological implication of four somatic mutations for the tumor suppressor gene STK11. Authors found that mutation in Y49D, G135R and D194Y increases proliferation, tumor growth and reduce kinase activity, while R87K mutation displays a similar phenotype to wt isoform. This greater tumorigenic phenotype is explained, only in part, by higher motility and modulation of vesicle trafficking, adhesion regulation or cytokines production. STK11, is the third most commonly altered gene in lung adenocarcinoma, and it is also a risk factor in other type of tumors such as pancreas, gastrointestinal, breast, cervical, uterine, and testicular cancer. The STK11 gene encoding LKB1 protein is emerging as an important tumor suppressor that may impact the metastatic propensity of tumors as well as therapeutic response. Most mutations are truncating or loss of function so understanding the functional impact of different mutations is a relevant issue. The study has some interesting data suggesting that there may be different mechanisms by which LKB1 function is lost including increased degradation vs reduced function of the kinase.
The manuscript has some significant shortcomings that limit the overall impact of the study, however. Significant issues include the following: 5. The clinical significance of these particular mutations is unclear. In searching through the Lung adenocarcinoma TCGA, I did not see any examples of R87K, Y49D, or G135R and just a few cases of D194Y. In looking at common inherited mutations in Peutz Jager syndrome I did not see these mutations occur commonly ( Basically, the comparison confirmed the relevance of the genes selected in the original version of the manuscript. We hope this will satisfy the reviewer concerns. This information has been included in the manuscript accordingly. 7. The significance of the growth rates in the in vivo models is unclear because we see that the growth rates of the control group (no doxycycline and hence no mutant expression) should be similar across the pairs (since it should be an identical LKB1 null background); in fact there is wide variability with about 6 fold growth in the R87K experiment ( Fig 6C) and more than 20 fold growth in the Y49k experiment. This suggests that there could be some leakiness in the dox promoter or that there are simply differences in growth rates across the different stable cell lines. Therefore basing comparisons in the growth with or without dox when the baseline is different for each does not seem reasonable. It would be more appropriate to see if similar mutation-specific differences are observed across different models expressing the particular mutant.  TESC  KLK10  GNAL  TNS4  TNS4  AREG  EMR2  CTSS  MMP2  FST  SOX4  RAB3B  FSTL4  CYP4F11  NEO1  DHTKD1  HBXIP  FAM105A  DIO2  PTS  CXCL2  FSTL4  ALDH3A1  MACF1  RBM47  SNRPG  SECTM1  NCR2  ALDH3B1  NOL7  TBC1D2  RND1  ATP8B1  PAPSS2  FGA  TESC  GLRX  EEA1  BCL3  GEM  PDLIM7  SLC37A1  WBSCR16  KIAA0319  TESC  ID1  ODC1  RPP40  ALDH3B1  PDK4  RGS2  CHAC1  TAGLN  NEO1  LGALS9  ASS1  SNX10  HOXB5  NRG1  SOD2  FSTL4  ID1  PDGFRL  TSKU  TESC  TNS4  DIO2  IRF1  B3GNT1  BCL3  INCENP  TDO2  GABRA5  HBXIP  RND1  FAM105A  ALDH3A1  STC1  RAP1GAP  CREB5  OAS1  UNC50  IFNGR1  ANGPTL4  IKBKE  TP53I11  PWP1  ID1  ALDH3B1  LARGE  GPRC5C  EMR2  ETS2  FST  MAML3  MACF1  MYLIP  CDCP1  RRP15  SFN  ERP29  RNF103  ETS2  LOC440792  CXCL5  SNTB1  MMD  MKNK2  DHRS3  TAGLN  TIMM23  GEM  PDK4  EFNA1  TRIB3  ALDH3B1  OSTF1  TIMP1  NPR3  SRM  FAM45B  GPRC5C  PTGES  TUFT1  C5  DEPDC6  MTERFD1  CSF2RA  RELB  CKS1B  CIT  FSTL3  KIAA0319  AKAP12  ARRB1  NF2  TNIP1  MACF1  RASA4  FAM105A  TESC  MAP3K8  ANPEP  TESC  MMP2  DHRS3  CDC42  CYP1B1  HBXIP  HBXIP  ANXA13  DUSP1  LARGE  MMP2  ASPM  RAP1GAP  CSRP1  TSPAN13  ID1  CTGF  AREG  TNC  EFNA1  FSTL4  GEM  NUPR1  DUSP1  ADAM12  DUSP1 LGR4 DIO2  SOD2  SLC46A3  SGK1  PELI2  CKS1B  DUSP1  EEF1A2  HSPB8 LGR4

This is a good observation that we pursuing separately. It is very interesting how a tumor suppressor conceptually based on the presence or absence of the protein, could
acquire oncogene features upon punctual mutations. This is supported by the case of D194 position which is the most frequent selected missense mutation across different type of cancers. In supplementary Figure 1C it is shown that this effect is also observed in the colony formation assay performed in two additional models, G361

melanoma cells and HeLa cells. In addition to this, we have generated proliferation data in G361 melanoma cells showing the same effect (showed below). We are currently investigating this subject.
The finding that some mutants may accelerate tumor growth (e.g. Figure 6D) compared with the null background does not seem to fit with the earlier in vitro experiments showing that the mutants did not suppress tumor growth but did not accelerate it either (figure 1), which would be more consistent with the role of a tumor suppressor. The statement on page 16 "Thus, the in vitro tumor suppressor capabilities of the investigated LKB1 mutants were reflected in vivo." is not accurate; there is no significant suppression for the R87K and there is significant acceleration in Fig 6B, D, and E. The authors should discuss the differences in these findings.
We thank the reviewer comment. We wanted to clarify that the experiments that the reviewer is referring to, are not equivalent and the results although in the same direction might differ in intensity due to multiple factors. First, and probably more important is the -in vitro, in vivo-feature of the experiments. Cells in tissue culture media are not exposed to the same factors conditions than in vivo that might determine the final result. In addition to this, in "in vivo" settings, there is always a selection of what will grow better within the injected cell lines (cell lines are heterogeneous).
Moreover, the timing for exponential multiplication is different when we talk about in vitro or in vivo experiments. We believe that the effect that we described in vitro is reflected in vivo. In the case R87K where the reviewer states that the suppression is not significant. We think that there is a suppression, might be not significant for  10. In figure 1, data revealed that Q135R and D194Y mutations increased cell proliferation when expressed in a LKB1 deficient cell line (A549), while R87K and wt isoforms reduced cell growth. By contrast, Y49D mutant isoform seems not to have effects in term of cells proliferation. In figure 2, by contrast, Q135R and D194Y only partially lost kinase activity, while Y49D also significantly reduced its kinase activity and this isoform seems to lose ability to interact with STRADα. On the other hand, R87K (wt-like isoform), Y49D and D194K, but not G135R display a shorter protein half-life. These data are inconsistent with the proliferative phenotype shown in figure 1. The authors should discuss possible explanations for these findings.
We appreciate the reviewer observation, but we do not understand which is the inconsistent link between the half-life of the proteins, the relative kinase activity of every mutant and proliferation. Minor issues: 1. The statement (page 20) that "LKB1 Y49D showed signs of inflammation and disorganized angiogenesis (hemorrhagic) confirming the role of LKB1 in regulating cytokine production and inflammation" is an overstatement because the models were tested in immunocompromised mice so inflammation could not be accurately assessed; furthermore, hemorrhage is not an established marker for inflammation.
The reviewer is right, this statement was not very accurate, we have changed the term inflammation for inflammation-like, and corrected the whole sentence accordingly. In the case of hemorrhage, we did not mean to say that was a marker of inflammation, we just wanted to link two different events that have been linked in the literature in both directions Hemorrhage promotes inflammation (Ahn SH et al., 2019) and inflammation induces hemorrhage (Valance Washington A. et al 2009;Goerge et al., 2005). Fig 1E: panel with D194Y at 48h + Dox has dark blue bar (G1 arrest) in the middle of the other two blue bars, whereas elsewhere the dark blue bar is at the bottom. Should the bar be dark blue (meaning G1 arrest), in which case it should be at the bottom, or medium blue? Also, elsewhere in the figure, there are 4 different shades of blue but only 3 in the legend-one color is nearly black. Please clarify the colors and labels.

In
We believe that the reviewer refers to Figure 1F 3. The concluding statement in the Discussion is not adequately supported by the data, particularly the effects on immune modulation given the immunocompromised mouse models: "In summary, we show that beyond the role of the non-mutated protein as a tumor suppressor, missense LKB1 somatic mutations could contribute to tumor development and/or progression by modifying not only intrinsic cell capabilities such as proliferation, motility or adhesion but also the tumor microenvironment, affecting inflammatory responses and likely the immune system. "These experiments could be repeated in syngeneic murine models to better get at the impact on the microenvironment. In addition, public data could be mined (e.g. TCGA using CIBERSORT) to see if it supports the association of different patterns associated with different mutations.
We appreciate the reviewer comment and we agree in that our model does not reflect the complex and orchestrated anti-tumoral immune response that would happen in an immunocompetent mouse model. However, one of the main purposes of this manuscript is to emphasize the possible roles o mutated LKB1 molecules in cancer.  Figure 1F to support higher or lower rate of proliferation across the different mutants. Some issue with these data: -First, quantification of all cell cycle phases together should be close to 100%. Are the authors missing any population? For example, is SubG1 peak (Dead cells which display lower probe staining) quantified? Please, reanalyzed carefully these data.

Analysis of cells cycle phase distribution is shown in
The reviewer is right. The cause for this was that in the original figure we only considered the 2n population. These cells have a polyploid population that oscillates and we did not include in the data (See figure below). We have corrected this in the manuscript and this time the data is referred to the 100% of the 2n population.
-Cells cycle analysis are typically performed when cells are growing exponentially to avoid include additional effects that could interfere in the cell cycle progression. Could the authors explain why G1 phase percentage is higher in -Dox treatment at 48h compared with 0h for WT and Y49D isoform? If cell cycle analysis was performed when cells are growing exponentially, G1 phase percentage at 48h should be similar to 0h. This data seems to indicate a G1 arrest at 48h due to low nutrient availability or confluent state rather that expression of LKB1 wt. Initial number of plated cells should be recalculated to allow exponential grow still at 48h.
As above explained this figure has been corrected accordingly.
-Finally, averages for 2 or 3 independent experiments should be included.

As explained above and in the figure legend these experiments were done in triplicates
where dark blue bars represent significative differences 5. In figure 1D and Figure S1 C authors show the quantification of clonogenic assay performed in the different mutated isoform. Are these differences statistically difference? If it does, please include this data.
We apologize for not include this in the original version of the manuscript. This issue has been corrected accordingly.
6. Figure  6C is cited in the text before than Figure  6B.
We apologize for this mistake. This issue has been fixed.
7. In Discussion section, second paragraph the sentence "Thus, selected missense LKB1 selected mutations…" should be corrected by deleting the second "selected".
We have corrected this mistake.
Reviewer #3 (Remarks to the Author): This manuscript by Paula Granado-Martinez et al. focuses on the functional study of STK11 (LKB1) missense somatic mutant isoforms such as LKB1Y49D, LKB1R87K, LKB1G135R and LKB1D194Y in cancer. They performed several experiments including omics analysis such as secretome to reveal the functions of these mutant isoforms. This is interesting work to reveal the importance of the STK11 (LKB1) missense somatic mutant isoforms. Before it can be published in Communications Biology, I have several suggestions as described below.
1. The statistics analysis should be performed in Figure 1D, 3A and 3E.
We thank the reviewer comment. We have added the statistics analysis to the suggested figures.
2. The supplementary tables for proteome should be described the protein full names and the detail information about mass identification. Since many errors happened in protein identification even using software such Proteome Discoverer, to check the mass peaks carefully and list the results are important in proteome field. Table 2. 3. I strongly suggest that the authors should deposit the proteome data to public database such as ProteomeXchange (http://www.proteomexchange.org/).

We apologize form not include this information in the previous version. This information has been added to the new version of the manuscript as Supplementary
We appreciate and agree with the reviewer suggestion. We have deposited the data in proteomeXchange database PRIDE with accession number: PXD018041.
List of new data added to manuscript 1-We have added in the introduction an explanatory phrase justifying the selection of the mutants (page 5 lines 7-9).
2-We have added to supplementary Figure S1 a western blot showing the expression amounts of endogenous LKB1 in 14 lung tumor cell lines and 8 melanoma cell lines. Figure 1E showing the metabolic profiles (Seahorse technology: OCR, ECAR, and some mitochondrial parameters for mitochondrial use of the different cell lines expressing the different isoforms of LKB1. Figure figure 2D showing a physiological functional assay to confirm the described kinase activity. We have measured the response of cells to metabolic stress measuring the amounts of p-AMPK as a surrogate marker of the LKB1 kinase activity.

4-In supplemental
8-We have fixed the cell cycle experiment following the reviewer suggestions.
9-We also have added all the statistical analysis to the graphs lacking the p-values. Table 2 showing the proteins detected by mass spectrometry in secretomes before and after the expression of the different LKB1 isoform. Old Supplementary Table 2 become Supplementary Table 3 and  Supplementary Table 3 has become Supplementary Table 4. 11-We have uploaded the gene expression data and the proteomics data to Arrayexpress and ProteomeXchange databases.

10-We have added new Supplementary
pathways with that many genes (e.g. the pathway maps in Figure 5) but the significance of any of these pathways and their association with a specific mutation remains unclear. If the authors would like to demonstrate that specific mutations lead to differences in pathway activation, it would be more convincing to show that changes observed in vitro with a specific mutation (e.g. D194) can be validated using tumors from patients with that specific mutation (e.g. using TCGA data or Kaufman dataset). Alternatively, it would be useful to validate that expression of specific mutations lead to the same differences in an independent in vitro system (e.g. different cell line).
We agree that validating our results in other models will be ideal, and probably some of our results will be more universal that others. Cancer is a very heterogenous disease (even intratumoral) and therapeutic responses (i.e melanoma and thyroid or colon cancer). We believe that our results are solid and strong. We obtained the data using different technical approaches, from a global view (OMICs), to the detail (biochemical and molecular techniques), in vitro and in vivo. It is possible that validation of our results could make the described mechanisms broader, but not with better quality nor relevant. We could validate our results in different systems and still have the same results, but we can also obtain different results and that would not invalidate the first ones. Obviously, there are some results within the manuscript that are hypothesis-generating, and of course, all of the result will require independent validation in future work. In this particular case, using human samples to validate our results is almost impossible due to the low number of samples with these particular mutations and the limited information available about them (i.e. RNA seq, Exome…). The manuscript that the reviewer refers to (Kaufman et al 2014) is dedicated to study the "loss of function of LKB1" vs.  (Fig.4). However, the interpretation of these results with just 3 samples , no control of the protein amounts of LKB1 within samples, unknowing the mutation allele frequency of STK11 mutant in tumor cells, heterogeneity of tumor expressing this allele, tumor cell percentage of the sample… etc., it turns difficult. The real validation of this type of data will come from the rest of the scientific community over time, and the Kaufman study is a good example, where 15 independent studies investigated the same thing over time, in different groups of samples, tumor subtypes, and species.
2. The authors state (page 15) "To validate the relevance of our dataset, we compared our 817 regulated genes with the 2080 unique regulated genes obtained from 15 different datasets (top 200 regulated ge.nes in each data set) published in Kaufman et al., 201437, comparing the gene expression profiles of human and murine tumors with or without deleted STK11". It is unclear how this is validating the relevance of the dataset. While some overlap in the genes differentially regulated in STK11 mutant vs wt was observed with the current dataset, it does not seem that the differences induced by specific mutations was validated-e.g. the analyses of genes differentially regulated by D194 mutants vs other STK11 mutants in the Kaufman dataset compared with the genes or proteins specifically regulated by D194 in the current study.
In the first revision round the reviewer suggested the comparison of our regulated genes with public datasets (Kaufman et al., 2014) to "A) focus on genes known to be regulated by LKB1 in clinical datasets, and then B) Assess whether those particular LKB1-rgulated genes are differentially regulated by different mutants. We found this suggestion a very good contribution to assure that the selected genes were specific for LKB1 so we could compare the regulation of these genes by the LKB1WT with the mutant isoforms. It turned out that all the genes we had selected were also relevant in the manuscript (Kaufman et al., 2014, this information was included in the revised version of the manuscript). We believe that this suggestion strengthened our results and validated the selection of the genes regulated by the expression of LKB1 WT that was analyzed with the LKB1 mutant isoforms. As above explained due to the restricted number of samples harboring the appropriate mutations and data availability of these samples the type of analysis suggested are not significant (Figure enclosed).