Essentiality of fatty acid synthase in the 2D to anchorage-independent growth transition in transforming cells

Upregulation of fatty acid synthase (FASN) is a common event in cancer, although its mechanistic and potential therapeutic roles are not completely understood. In this study, we establish a key role of FASN during transformation. FASN is required for eliciting the anaplerotic shift of the Krebs cycle observed in cancer cells. However, its main role is to consume acetyl-CoA, which unlocks isocitrate dehydrogenase (IDH)-dependent reductive carboxylation, producing the reductive power necessary to quench reactive oxygen species (ROS) originated during the switch from two-dimensional (2D) to three-dimensional (3D) growth (a necessary hallmark of cancer). Upregulation of FASN elicits the 2D-to-3D switch; however, FASN's synthetic product palmitate is dispensable for this process since cells satisfy their fatty acid requirements from the media. In vivo, genetic deletion or pharmacologic inhibition of FASN before oncogenic activation prevents tumor development and invasive growth. These results render FASN as a potential target for cancer prevention studies.

: Levels of FASN are increased in MEFs following PyMT infection. This increase is not observed following KRAS or HER. Could this increase of FASN participate to the ability of MEFs to transform upon PyMT? In other words do levels of FASN impact on transformation by PyMT? Levels of RAS are higher in FASNlox/lox MEFs compared to FASN∆/∆ MEFs. As FASNlox/lox MEFs proliferate more (Fig1C) they probably are easier to infect and integrate more retroviral particules which could explain the increased RAS level. So one could conclude that increased RAS levels in FASNlox/lox MEFs participate to the increased clonogenic ability of these cells ( Fig S1C) and could be responsible to the rapid tumor growth at T3-T4 observed in FigS6B compared to the slower tumor growth at T8 for FASN∆/∆KRAS MEFs. Quantification of western blots (Fig S1) would help.
Figure S1E: HC11FASNKO cell lines are generated by the Crispr-C9 technology. sgRNA targeting human FASN were used. It appears that HC11FASNwt cells were not infected by recombinant lentivirus containing control sg. Such control is missing. Authors should compare both lentivirus infected HC11 cells with or without FASN. Figure S2A: Authors mention that the decrease in PDH activity in FASN∆/∆PyMT MEFs compared to FASNlox/loxPyMT MEFs is phenocopied by adding acetylCoA to FASNlox/loxPyMT MEFs which is not clearly illustrated in the panel S2A where no statistic are illustrated. Figure 2B. NADPH production is increased in FASN∆/∆PyMT MEFs compared with FASNlox/loxPyMT MEFs in 2D condition. Is this increase due to low ROS levels in FASN∆/∆PyMT MEFs or vice-versa? Authors mention that, in 3D, there would be insufficient intra-mitochondrial reduced equivalents which are usually consumed by excess of unquenched ROS produced during the 2D to 3D transition. NADPH measurements in 3D should be performed and added in Figure 4 or 5. Moreover, authors should better explain why accumulation of unquenched ROS during 2D to 3D transition would be followed by impaired mitochondrial respiration, as authors write that lack of ROS production in 2D in FASN∆/∆PyMT MEFs explains the low OCR observed (lines 165 to 168). These 2 different conclusions in 2D and 3D conditions are confusing. Collaboration with expert in ROS and OXPHOS mechanisms such as Dr N. Chandel, co-authors of this study, should help to make these points cleared up.    Figure 6B: Authors should illustrate higher magnifications for both illustrations. As authors write in lines 332, 333 that the number of foci is smaller in PyMT/FASN negative tumors compared to PyMT/FASN positive tumors, they have to quantify and illustrate such quantifications. Figure S6D,E: Authors claim that similar differences are observed in vivo using MEFs models or HC cells (line 338) which is not correct. A decrease in tumor size is indeed observed following FASN deletion using MEFs cell lines with KRAS and HER2 mutations (FigS6B, C) however using HC11 breast cancer cell lines (Fig6 D,E) tumor growth is only delayed and not reduced as authors suggest by mentioning "similar differences were observed". This points out the fact that FASN deletion is efficient to impede growth of tumors derived from MEFs (following KRAS or HER2 mutations) but not of tumors derived from HC11 cells for which only a delay of growth is observed. To increase the relevance of targeting FASN in vivo, I would suggest that authors combine the model illustrated in Fig6 with G28UCM treatment to eradicate remaining FASN positive foci and evaluate the impact of a total deletion of FASN on spontaneous breast cancer tumors.  Line 254 : "extra-mitochondrial" should be corrected by "intra-mitochondrial" Line 135 : Authors should add "and apoptotic rates were unaffected by FASN depletion in…" Reviewer #2: FASN (Remarks to the Author): Bueno et al describe a new role for FASN, quenching excess ROS produced during anchorage independent transformation through consumption of acetyl-CoA and IDH-dependent reductive carboxylation. The authors show how FASN activity is essential for the initial steps of transformation using different oncogene driven models, and characterize the levels and fate of metabolic intermediates in cells but that this activity is necessary to restore the NAD+/NADH ratio and is independent of palmitate production. Overall, the study is original, interesting, and potentially relevant, but some of the claims are not supported by the data provided and additional experiments are required to substantiate the conclusions.
1. The most relevant and novel finding in this paper is the fact that FASN is essential to sustain IDH-1 dependent reductive carboxylation of glutamine. The rescue of spheroids and soft agar growth after restoring NAD+/NADH ratio is very interesting and important in support of the hypothesis. To demonstrate that this is indeed a universal phenomenon, the authors should include data in figure 4 (or figure S5) presenting both NAC/GSH rescue in all oncogene models studied (PyMT, Kras, HER2).
2. The authors claim that the effects of FASN on cell transformation is independent of its enzymatic product palmitate, stating that the preferential source of fatty acids for the cells is exogenous. This is a pretty strong claim that is counter to several studies demonstrating that the lack of de novo palmitate synthesis is the main reason for the induction of apoptosis, cell growth reduction, etc, typically observed following FASN inhibition. However, this study is carried out in fibroblasts transformed by oncogenes. Normal cells do not require FASN activity as they rely on exogenous lipids and the MEFs being utilized perhaps have yet to achieve the metabolic re-wiring required during full transformation. Requirements for endogenously produced lipids (palmitate) is different in neoplastic cells. In addition, the palmitate rescue experiment in MEFs (figure S3C) was not performed with albumin-complexed palmitate, which might have negatively affected the result. Another issue is in figure S1D: in the absence of exogenous palmitate no colony formation was observed in MEFs with any of the oncogene models evaluated. Palmitate represents 26% of fatty acid content in regular FBS, as stated by the authors. Therefore, fatty-acid free FBS is also depriving the cells of many other lipids, confounding results. Two approaches could be done to support that de novo synthesis of palmitate is not playing a role in colony formation: add exogenous albumin-complexed palmitate to fatty acid free FBS to rescue the colony formation capability, or add sulfo-N-succinimidyl oleate (an inhibitor of fatty acid uptake) to medium containing normal FBS, ensuring that the uptake of fatty acids is, indeed, necessary for cell transformation.
3. In figure S3C, soraphen-A was used to decrease malonyl-CoA availability, preventing fatty acid oxidation (FAO) blockade and, consequently, impairment of colony formation. While no growth rescue was observed, the authors did not verify FAO activity, or whether soraphen-A could rescue the potential FAO blockade observed with FASN ablation. 4. In figure 3B the unlabeled lipid pools were similar in FASNlox/lox-PyMT and FASNΔ/Δ-PyMT MEFs. The authors believe this result confirms that intracellular lipids levels are not dependent on de novo lipogenesis and that cell transformation is not dependent on palmitate itself, which again conflicts with other studies that show the critical role of FASN in critically contributing to the intracellular pool of lipids. Neutral lipid accumulation assessed by Oil Red O staining could provide further confirmation of this result. 5. The authors claim that IDH1-driven reductive carboxylation supports anchorage independence, which is blocked in the absence of FASN activity due to acetyl-CoA accumulation that leads to an increased citrate/isocitrate ratio. The carbon tracing experiments and the addition of sodium citrate or ATP citrate lyase inhibitor lend support for the mechanism proposed. However, is not a decreased replication rate, but the inability to switch from 2D-growth to 3D-growth. We regret not having included those controls. When we were executing the HC11 experiments we had everything else already finished, and we were just validating the results in one more system with further oncogenes. At that moment the phenotype was so consistent that we sincerely doubted that the Sg integration could play any role on itself, at least on this phenotype. Regardless, and because academically speaking it is the correct control, although we performed all the experiments with the HC11 "wild-type" clones, we In addition, we would like to mention that the reference to human FASN in the methods section was a mistake. It is murine FASN. It has been corrected. Figure

Collaboration with expert in ROS and OXPHOS mechanisms such as Dr N.
Chandel, co-authors of this study, should help to make these points cleared up.
We thank the Reviewer from bringing up this point, since it is the "core" of the manuscript and was not well explained in the previous version.
In 2D, FASN ∆/∆ -PyMT displays a low incorporation of pyruvate into the Krebs cycle and decreased Krebs cycle activity ( Figure 3), which in turn implies low mitochondrial activity and respiration. This is associated to low ROS. The increased NADPH observed in 2D in FASN ∆/∆ -PyMT can be explained by the low ROS, but, above all, by the simple fact that FASN synthetizes fatty acids by condensing acetyl CoA, malonyl CoA and NADPH. If FASN is not working, there is not NADPH consumption by FASN, and FASN is one of the main consumers of NADPH (Figures 2b and f). In contrast, in 3D (where cells generate more ROS according to already published observations 1 ; we observed the same effect), the main contributor to low mitochondrial activity is the excess of ROS generated by the citrate lyase blockade. NADPH becomes key for tumor progression in 3D, since, wild type clones quench the excess of ROS by NADPH originated through reductive carboxylation; the knockouts can not produce these extra NADPH, although the amount of NADPH would never be "0" because FASN is inactive and it is not consuming NADPH at all. Thus, it would be expected to observe increased NADPH in the knockouts in 2D (already in previous version) but decreased NADPH in 3D.
How can we prove that the chain of events is as we say, and that it is the excess of ROS what negatively impacts mitochondrial respiration in 3D in the knockouts? By two experiments that are now included in the manuscript: -First, we have already showed low Krebs cycle activity in 2D, low ROS and high NADPH.
In 3D, we have shown high ROS, but not NADPH. We include now total and intra- Thus, the model is the following: -2D: background ROS production conditions. FASN lox/lox -PyMT clones are showing standard respiration and ROS production. FASN ∆/∆ -PyMT clones, however, have a decrease input in the Krebs cycle and thus produce less ROS. Both clones have intact mitochondrial electron transport chain complexes. NADPH is not required to quench 3Dderived ROS, and thus it is not unexpected to observe increased NADPH in FASN ∆/∆ -PyMT clones, since the mitochondrial ROS production is low, and NADPH is not used for fatty acid synthesis.
-3D: increased ROS production conditions. FASN lox/lox -PyMT clones can quench them by the excess NADPH produced by reductive carboxylation, and thus preserve complexes integrity and replicate/grow normally. FASN ∆/∆ -PyMT clones, however, have a decreased ability to perform reductive carboxylation, evidenced by the data in the previous version, and the lower NADPH levels, which results in increased unquenched ROS that damage the mitochondria. In this situation, they can not sustain 3D growth and thus tumor formation.
The title of the results section has been changed to the following: "Cells lacking FASN cannot grow in an anchorage-independent manner in 3D due to insufficient ROS quenching, leading to mitochondrial complex I disassembly from supercomplexes and explaining the inability to complete oncogenic transformation" The text of the final part of the section is now as follows: To test this hypothesis, first we measured intra-mitochondrial and total NADPH. Total NADPH levels were decreased in FASN ∆/∆ -PyMT compared to FASN lox/lox -PyMT MEFs ( Figure 5g). However, the difference was more pronounced in the intra-mitochondrial compartment (Figure 5g), which is congruent with the fact that cytoplasmic levels may still be high because of the lack of activity of FASN, which consumes NADPH. It has been shown that an increased ROS production could disrupt mitochondrial complexes integrity 2, 3 , stalling respiration. While in 2D the decreased respiration levels of FASN ∆/∆ -PyMT can be attributed to a decreased entrance of pyruvate in the mitochondria, in 3D the increased ROS might be a more important contributor. In fact, we observed that in 2D total NADPH was increased in FASN ∆/∆ -PyMT clones ( Figure 2b); however, in 3D in the context of decreased citrate lyase activity, we observed decreased intramitochondrial NADPH levels,  Figures 6 a, b).
Taken together, our data suggest the acetyl-CoA build-up secondary to FASN deletion inhibits ATP citrate lyase and induces an accumulation of citrate/isocitrate. Subsequently, this impairs IDH1-dependent reductive carboxylation, a tumor cell requirement limited to 3D growth, inducing a decrease in intramitochondrial NADPH, an increase in ROS, and finally mitochondrial supercomplexes assembly disruption, which ultimately impairs cell transformation and tumorigenesis.
Finally, the second-last paragraph of the discussion has been modified accordingly: "… impaired, what leads to the observed decreased in intra-mitochondrial NADPH ( Figure   5g) and decreased levels of mitochondrial complex I assembly into supercomplexes (Figure   5h), which accounts for the inability to transform (Figures 4 and 5). This was proven by This is a very good, albeit previously missing, point. We have studied the effects of G28UCM administered to FASN-lox/lox -PyMT MEFs in terms of spheroid formation, ROS production in 3D, and OCR. G28UCM disrupts spheroid formation, increases ROS production in 3D, and decreases OCR rates, although in a less intense manner than the genetic experiments. The reviewer is absolutely right, we apologize for having not been accurate in our wording.
We have change the wording -the new text is as follows: "We tested as well the in vivo effects on tumor growth of FASN ∆/∆ -MEFs infected with KRAS or HER2, grafting them into wild-type animals compared with FASN lox/lox counterparts (Supplementary Figures 7b, c). In line with the in vitro data (Supplementary  (Supplementary Figures 7g, h). A strong reduction in tumor growth was observed (Supplementary Figures 7b, c). Finally, when HC11 cells with CRISPR-deleted FASN were infected with PyMT or KRAS and grafted into wild-type animals, a delay on tumor onset was observed compared with wild-type counterparts (Supplementary Figures   7d, e)." The combination of a pharmacologic and genetic approach is an excellent idea, and we have performed the experiment. We have observed a further decrease in tumor burden in the FASN-negative PyMT animals when they received G28UCM treatment compared with vehicle. However, the treatment was toxic in these animals (likely because of the complete  We thank again the Reviewer for a great suggestion to make the mechanistic findings more robust. We have performed the experiment, and found that G28UCM induces high ROS levels in We have backcrossed the PyMT animal to a pure FVB background and to a pure C57/B6 background. We use the latter for studying the function of novel oncogenes or tumor promoters/accelerators, while we use the former for studies with genes involved in a potential delay or abrogation of tumor onset or growth (the model is also very good for oncologic drugs, since due to the fast growth it allows studying antitumor effects of novel compounds relatively fast).
Since the pure FVB-PyMT animals usually have to be sacrificed at 5-7 weeks after tumor onset, the metastatic ratio is very low -approximately 10% in our hands. Thus, it is an impractical model for answering this question. We have examined the lungs from 54 FVB-PyMt animals sacrificed at the humane endpoint, and we have found metastases in 5 animals (9.5% of the cases).
In addition, the following picture allows appreciating that the metastatic disease burden is quite low.
In summary, it is a great question that we also want to address. We believe that because FASN-/-cancer cells are unable to survive in 3D, we believe that they will be unable as well to form metastases because of the intermediate steps required for doing so (surviving in a cell-adhesion independent manner in blood, forming CTC-clumps, give rise to a 3D tumor in the lung, etc). The study of the process will require heavy experimentation and mechanistic studies, and also move to the C57/B6 background. Thus, it will be reported in a future manuscript in about 2-3 years from now. The number of replicates (and whether they are biological or technical), as well as information about the statistical test that was performed, has been incorporated in all principal and supplementary figures wherever it applies.

13) Discussion: In practice, how authors consider the feasibility to clinically target FASN before transformation? In high risk patients? This concept needs to be clarified.
We thank the reviewer for giving us the opportunity to clarify this point. Indeed, akin any other intervention in the prevention sphere involving drugs and not just lifestyle changes, it would be justified only in high-risk patients (at least until some benefit is demonstrated; subsequently, further trials in lower-risk or standard-risk populations can be planned).
Several trials have assessed various approaches for decreasing the rate of invasive cancer 4, 5 , and usually they rely on the same criteria: Gail score above 1.7 (in 5 years), previous history of lobular carcinoma in situ and/or previous history of atypical hyperplasia.
We have modified the last paragraph of the manuscript accordingly: "Taken together, these features indicate FASN is a potential target for cancer prevention.
Akin previous chemoprevention trials 4,5 , future interventions should be planned in healthy patients at high risk of developing invasive breast cancer, such as those with previous diagnosis of atypical hyperplasia, lobular carcinoma in situ, or high Gail 5-year risk score." Minor comments: 14) Figure 3D: There is no compensatory induction of Glut1 as claimed by authors as it appears ns. Authors should correct this in the manuscript.

Glut4 mRNA levels in FASNlox/loxPyMT is below 1, what is the sample of reference as =1?
We thank the Reviewer for noticing this. Regarding Glut 1, we have corrected the main text.
In the case of Glut4, it was incorrectly plotted: the reference (FASN lox/lox -PyMT in both charts) is 1 (Figure 3d). We apologize for the confusion -rather than a mislabeling it was the ECAR readout of the OCR experiment: i.e., in the previous version of the figure, since the Seahorse machine can measure simultaneously the oxygen consumption and the extracellular acidification rate, we provided the acidification rate in the experiment designed to measure OCR (that is, during the timecourse of adding oligomycin, FCCP and antimycin/rotenone). It is common to observe such type of simultaneous report in publications, however, it does not provide an accurate assessment of glycolysis. Thus, we have chosen to repeat the experiment, this time measuring ECAR with the ECAR-measuring protocol (i.e., sequential addition of glucose,

16) Line 254: "extra-mitochondrial" should be corrected by "intramitochondrial"
We thank the Reviewer for noticing this typo. It has been corrected.

17) Figure 1c: axis legends of needs scale
We are not 100% certain about the meaning of this question -in any case, we have tried our best to clarify the chart. The chart has two axes, X and Y, and plots the relative increase in the number of cells (Y axis) along time (X axis). Time is expressed in days (0 to 3). The Y axis is now re-labeled ("Fold-increase in cell number"), and expresses the relative number of cells of each clone at any given time compared to time 0, when the experiment started. We have clarified as well the figure legend: "Lack of significand differences in cell replication rates along time".

18) Title of Figure 1 should be focused on clonogenic ability of MEFs as proliferation is not altered by FASN deletion.
We have changed the title of Figure 1 accordingly.

19) Line 135 : Authors should add "and apoptotic rates were unaffected by FASN depletion in…"
We thank the Reviewer for the suggestion, which has been incorporated. We apologize for not having included all the possible variants. We include now proof or spheroid formation rescue on the missing cell lines/genotypes. As it can be appreciated (now Supplementary Figure 6), the results are consistent across different genotypes.

The authors claim that the effects of FASN on cell transformation is
independent of its enzymatic product palmitate, stating that the preferential source of fatty acids for the cells is exogenous. This is a pretty strong claim that is counter to several studies demonstrating that the lack of de novo palmitate synthesis is the main reason for the induction of apoptosis, cell growth reduction, etc, typically observed following FASN inhibition. However, this study is carried out in fibroblasts transformed by oncogenes. Normal We sincerely thank these suggestions, since we are aware that the commonly accepted notion is that the synthetic product itself is the essential feature linked to FASN upregulation in cancer. We were surprised as well to observe our results, and the suggestions made by the Reviewer strengthen even more the conclusion, which we believe that it is a major finding. We have performed both experiments, and, to our surprise, neither  We thank the Reviewer for pointing this out, since it has been proposed that one main mechanism by which FASN inhibition impairs tumor growth is because of the Malonyl-Coa build-up which in turn would inhibit fatty acid oxidation. We have found that the other Once again, we sincerely thank the Reviewer for a suggestion that allows us to strengthen the conclusions.
We have silenced IDH1 with siRNA (we chose this approach to observe the effects of short-term IDH1 inhibition akin with a compound -SB204990 -or with citrate, since CRISPR editing might cause some sort of metabolic rewiring that would mask elucidating the mechanism) and we observed that it phenocopies the three main features of FASN deletion in 3D: increased ROS production, abrogation of formation of spheroids, and decreased OCR. The results are shown in Supplementary Figure 10.
We have added two lines of text, and a new supplementary figure, reporting the experiment in the main text: "….. Figures 4i, j, Supplementary Figures 5d, e). When IDH1 was silenced with siRNA in 3D, it led to a disruption in the formation of tumor spheroids and suppression or OCR together with increased ROS, compared with no effect in these traits in 2D (Supplementary Determining on-target and off-target effects are important points on studies reporting on new drugs. Although our study is not interested in G28UCM (a compound that has been previously characterized 8 ) and our only interest on it was as a tool compound for completing some of the experiments, we think that it is a good suggestion to wrap-up our findings in a purist academic manner.
The question has two parts -showing the on-target effect and controlling off-target effects.
The first one is easy to demonstrate, while the other poses problems (see below).
Regarding the off-target effects: it is common to have this request in manuscripts dealing with "targeted therapies", in order to prove that a drug does have or does not have effects in absence of the target, after knocking it out. In the first case, the drug would have substantial off-target effects, and in the second, it would be relatively clean. As reviewers wisely usually ask for it, it is common now to assume the fact that most "targeted drugs" are not as "targeted" as they initially look, and thus is a legit question in order to characterize compounds.
In order to address such problem, two features must be taken into account: first, there has to be a phenotype (most commonly is tumor cell death), and second, there must be a pharmacodynamic surrogate. In a hypothetical example we could talk, for example, about a MEK inhibitor that has anticancer effect. We would observe certain cell killing (e.g., 70%), and we would observe pharmacodynamic correlate (i.e., in this case, decrease in pERK). In order to assess whether our MEK inhibitor has or not off target effect, we would KO MEK, and treat the Kos with the drug. Cells would be alive in absence of the drug (since our target would not be essential), and if our compound has off target effect, still a percentage of MEK KO cells would die with the compound -i.e., 20%, -but no further changes in pERK should be seen. Thus, the efficacy of our compound would be partly because of its effects in MEK (rough numbers 50%), and because of other off-target effects (roughly 20%). Now, if we translate this to our scenario, there is a technical hurdle difficult to overcome: our phenotype is "abrogation of 2D to 3D transition in absence of FASN". When we KO FASN, the penetrance of the phenotype is 100%. We have never been able to recover colonies in soft agar or espheroids in low adherence when FASN is knocked out before PYMT transformation (with the exception of escapers, but that´s a different story, since they still are FASN+). In other words, there´s no "residual" phenotype where we can check with G28UCM if there´s further decrease in "abrogation of 2D to 3D transition in absence of FASN, mediated through an effect through a different target than FASN" (thus proving that there´s some other pathways that abrogate this transition, outside FASN-IDH1reductive carboxylation) adding G28UCM to the knockouts and compare it to the knockouts alone, because the abrogation is already complete. In fact, when we add the compound in FASN wild type MEFs, what we observe is that it only partially abrogates 2D to 3D, because it only inhibits FASN partially. Unfortunately, there is no way to address whether in absence of FASN other G28UCM-modulated pathways come into play, because there´s nothing left to block with G28UCM.
Binding studies for sure would find that G28UCM binds other targets, since the amount of compound required to inhibit FASN is relatively high; however, as we said, the manuscript is not reporting on G28UCM characterization.

The authors suggest to target FASN before transformation, as a prevention
strategy. Normal cells, however, do not require FASN activity as they rely on exogenous lipids. This is different in neoplastic cells. Thus, this suggestion lacks the biological rationale.
We are not 100% sure we understand the question. What we meant is exactly what the Reviewer explains, which to the best of our understanding is the basis of prevention: using a compound that would be relatively innocuous to normal cells (Rev: "Normal cells, however, do not require FASN activity") that would, meanwhile, impair the fitness of cells that are undergoing transformation (Rev: "This is different in neoplastic cells"). Such compound could be administered indefinitely, and, whenever a malignant transformation event occurs, the compound would at least in some individuals abrogate or delay its effects. Since today it is -yet -impossible to predict or detect when/where a malignant transformation events takes/will take place, that is the basis of chemoprevention: in groups of patients at high-risk (in the case of breast cancer, BRCA mutants or patients with high Gail score), they would be taking the compound while there is no evidence of disease, so that in case in one patient an early tumor has started to grow, or will start to do so, is blocked by the compound with selective activity in blocking the initial steps of tumorigenesis, which is what a clinical-grade compound would do. We believe that targeting specific processes of malignant cells or cells undergoing transformation while sparing normal cells is a quite reasonable biologic rationale for oncology therapeutics, regardless whether it is aimed at the prevention or therapeutic phase, and this rationale is behind the immense majority of existing anticancer drugs.
We have changed the wording of the prevention paragraph in any case to facilitate the reading. " A clinical-grade compound that would selectively target FASN could be administered long-term to a healthy individual, sparing toxicity to normal tissues, while could target cells in their initial steps of transformation. Taken together, these features indicate FASN is a potential target for cancer prevention. Akin previous chemoprevention trials 4, 5 , future interventions should be planned in healthy patients at high risk of developing invasive breast cancer, such as those with previous diagnosis of atypical hyperplasia, lobular carcinoma in situ, or high Gail 5-year risk score." 8. There are certainly other mechanisms that require FASN activity when cells acquire invasive properties and these are not mentioned or tested (PMID: 22238651).
We have added the following sentence introducing the last paragraph of the discussion so that the reader can put our findings in context, citing the manuscript that the Reviewer suggests.
"FASN has been previously related to relevant diverse features of tumor progression such as increased cell replication 9 , HER2-signaling 10 or regulation of invadopodia 11 ."