Cancer-associated fibroblasts promote drug resistance in ALK -driven lung adenocarcinoma cells by upregulating lipid biosynthesis

  1. Ann-Kathrin Daum
  2. Lisa Schlicker
  3. Marc A. Schneider
  4. Thomas Muley
  5. Ursula Klingmüller
  6. Almut Schulze
  7. Michael Thomas
  8. Petros Christopoulos
  9. Holger Sültmann

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Targeted therapy interventions using tyrosine kinase inhibitors (TKIs) provide encouraging treatment responses in ALK -rearranged lung adenocarcinomas, yet resistances occur almost inevitably. Apart from tumor cell-intrinsic resistance mechanisms, accumulating evidence supports a role of cancer-associated fibroblasts (CAFs) in affecting the therapeutic vulnerability of lung cancer cells. Here, we aimed to investigate underlying molecular networks shaping the therapeutic susceptibility of ALK -driven lung adenocarcinoma cells via tumor microenvironmental cues using three-dimensional (3D) spheroid co-culture settings. We show that CAFs promote therapy resistance of lung tumor cells against ALK inhibition by reducing apoptotic cell death and increasing cell proliferation. Using single-cell RNA-sequencing analysis, we show that genes involved in lipogenesis constitute the major transcriptional difference between TKI-treated homo- and heterotypic lung tumor spheroids. CAF-conditioned medium and CAF-secreted factors HGF and NRG1 were both able to promote resistance of 3D-cultured ALK -rearranged lung tumor cells via AKT signaling, which was accompanied by enhanced de novo lipogenesis and supression of lipid peroxidation. Notably, simultaneous targeting of ALK and SREBP-1 was able to overcome the established CAF-driven lipid metabolic-supportive niche of TKI-resistant lung tumor spheroids. Our findings highlight a crucial role of CAFs in mediating ALK-TKI resistance via lipid metabolic reprogramming and suggest new ways to overcome resistance towards molecular directed drugs by targeting vulnerabilities downstream of oncogenic signaling.

Article activity feed

  1. Review Commons

    Note: This response was posted by the corresponding author to Review Commons. The content has not been altered except for formatting.

    Learn more at Review Commons

    Reply to the reviewers

    Manuscript number: RC-2023-02123

    Corresponding author(s): Holger Sültmann

    [The “revision plan” should delineate the revisions that authors intend to carry out in response to the points raised by the referees. It also provides the authors with the opportunity to explain their view of the paper and of the referee reports.

    The document is important for the editors of affiliate journals when they make a first decision on the transferred manuscript. It will also be useful to readers of the preprint and help them to obtain a balanced view of the paper.

    If you wish to submit a full revision, please use our "Full Revision" template. It is important to use the appropriate template to clearly inform the editors of your intentions.]

    1. General Statements [optional]

    This section is optional. Insert here any general statements you wish to make about the goal of the study or about the reviews.

    We would like to thank the editorial team of Review Commons for sending our manuscript for peer review and all Reviewers for carefully reading our manuscript. The reviewer’s detailed and constructive feedback and comments were instrumental to improve the quality and rigor of our manuscript. We highly appreciate the thoroughness of the review and have carefully considered all suggestions and concerns. Below, we have made point-by-point responses to the reviewer’s comments, and outlined revisions we plan to make, or have made. Textual changes in the revised manuscript are marked in Red.

    2. Description of the planned revisions

    Insert here a point-by-point reply that explains what revisions, additional experimentations and analyses are planned to address the points raised by the referees.

    Reviewer #1 (Evidence, reproducibility and clarity (Required)):

    Daum AK et al. indicated that CAFs promote TKI resistance via lipid biosynthesis in ALK-driven lung adenocarcinoma cells. This work provides novel CAF-induced drug resistant mechanisms in ALK-driven lung cancer cells, however, there are several issues to be resolved by the following.

    Major critiques:

    1. The authors claimed that CAF-produced HGF and NRG1 boosts AKT signaling and de novo lipogenesis to promote ALK-TKI resistance in lung cancer cells. They showed that CAF-secreted HGF and NRG1 inhibition attenuates tumor cell viability in the presence of FB2-CM, however, whether stromal HGF and NRG1 inhibition suppresses de novo lipogenesis in carcinoma cells has not yet been investigated. The authors should inhibit HGF and NRG1 expression in CAFs by shRNA prior to addition of CAF-CM onto cancer cells to evaluate AKT signaling and de novo lipogenesis.

    Author response:

    The necessity of demonstrating the direct impact of stromal HGF and NRG1 inhibition on de novo lipogenesis in cancer cells is a crucial aspect of our study, and appreciate the reviewer’s valuable suggestion to inhibit HGF and NRG1 expression in CAFs before assessing the effect on AKT signaling and de novo lipogenesis. To address this concern, we will conduct additional experiments to evaluate the impact of HGF and NRG1 expression knock-downs in CAFs by shRNA.

    This work lacks human correlation. Are HGF and NRG1 expressions in CAFs related with de novo lipogenesis, drug resistance and poor outcomes in ALK-driven lung cancer patients?

    __Author response: __

    We agree with the reviewer’s point that human correlation would underline the significance of the given findings. However, conducting such analyses presents certain challenges.

    1. The availability of ALK-mutated samples among TCGA samples is limited and the sample size is quite small (n = 5), making survival analyses less statistically meaningful due to low statistical power.
    2. Using bulk RNA-seq data for this analysis necessitates deconvolution methods to differentiate between tumor and stromal cell compartments. While deconvolution methods are valuable, they have limitations, including potential inaccuracies in estimating cell-specific gene expression due to the inherent heterogeneity of cell populations (1). This may lead to imprecise conclusions about the specific contributions of stromal factors, such as CAF-secreted HGF and NRG1, in the tumor microenvironment. Nonetheless, we are considering leveraging the available dataset of Maynard et al.(2), to address the raised concerns by the reviewer. Here, the authors performed single-cell RNA-seq on clinical biopsies, including a number of ALK+ samples from both treatment-naive and progressive-disease patients. The analysis of this dataset could allow us to investigate whether the effects observed in our study hold true in the in vivo human tissue environment, providing a more direct and clinically relevant assessment.

    The most experiments lack the appropriate control for CAFs. They would use primary isolated counterpart fibroblasts as the patient-specific control for lung cancer CAFs by extracting from non-cancerous regions of the same individual in their experiments.

    __Author response: __

    This point is well taken. Therefore, we will take this feedback into account and incorporate the suggested controls into our experimental design to enhance the robustness and validity of our results.

    Minor issues:

    In Fig 1B, coculture with TGF-b-treated MRC-5 attenuated cancer cell death with the lorlatinib treatment. However, HGF and NRG1 production is comparable between MRC-5 cells treated with or without TGF-b in Fig 5A. These data indicate that any fibroblasts but not CAFs could suppress cancer cell death with the lorlatinib treatment.

    __Author response: __

    We recognize the need for further clarification. To address this, we plan to include additional data to demonstrate the differences in tumor therapy response when cultured with CM derived from native fibroblasts versus TGF-β1-activated fibroblasts (CAFs). This will help elucidate the specific role of CAFs in suppressing cancer cell death with lorlatinib treatment and provide a more comprehensive understanding of the observed effects.

    The pAKT induction of H3122 treated with lorlatinib in the presence of CAM-CM or HGF or NRG in Fig 7A, B is barely observed and lacks the significance.

    Author response:

    We acknowledge the need for more robust data to demonstrate the significance of the observed pAKT induction in H3122 cells treated with lorlatinib in the presence of CAF-CM, HGF or NRG. We will work to provide additional data that strengthens the significance of this effect.

    Reviewer #1 (Significance (Required)):

    The concept of this work is interesting, however, molecular mechanisms underlying CAF-medicated ALK-TKI resistance remain poorly elucidated. Characterization of human primary fibroblasts (FB1, FB2) is not clearly described, and the most experiments lack proper control. Immunoblot in Fig 6 and 7 looks snap-shot and the reviewer has concerns about the reproducibility.

    Author response:

    In response to the comments, we will revise our manuscript to provide a more detailed characterization of the human primary fibroblasts to ensure transparency. To address the concern regarding controls, we will implement further controls in our experimental procedures. Regarding the concern about the reproducibility of the immunoblots, we appreciate the feedback, and we will provide additional data in the manuscript to ensure the reproducibility of our results.

    Reviewer #2 (Evidence, reproducibility and clarity (Required)):

    In this review, the authors describe a possible mechanism of resistance in lung tumor cells that is induced by CAFs in response to ALK-specific inhibitors. The manuscript is well written and conclusions are in general well supported by experimental data, however additional experiments are needed to fully support the conclusions stated by the authors.

    Brigatinib and lorlatinib are used to target ALK-driven lung tumor cells. In many of the experiments described heterotypic 3D co-cultures are used. Although both inhibitors act preferentially over the cancer cells, and no effect is expected in the CAFs, it would be desirable to confirm it.

    Author response:

    We thank the reviewer for this endorsement of our study, and we are pleased to learn that our conclusions are generally supported by the experimental data. Regarding the use of brigatinib and lorlatinib to target ALK-driven lung tumor cells in our experiments, we acknowledge the importance of confirming the specific action of these inhibitors. While these inhibitors act preferentially on the cancer cells, we agree that it is desirable to confirm their limited effect on CAFs. Therefore, we will address the reviewer’s suggestion by performing further cell viability analyses of TKI-treated fibroblast spheriods to specifically assess the impact of brigatinib and lorlatinib on CAFs.

    The authors demonstrated initial proliferation advantage and apoptosis protection of H2228 and H3122 cells in response to conditioned media (CM) of three independent CAF clones. However, once they identify lipid enzymes altered and specific ligand-receptor interaction, then they focus only in FB2. This imply that the mechanism described by authors is relevant for that particular clone, but it does not validate a general resistance mechanism induced by CAFs. In order to claim that this is a more general mechanism, other CAF clones should be tested.

    Author response:

    We appreciate the reviewer’s comment regarding the focus on a specific CAF clone (FB2) in our study. This point is well taken, and we understand the importance of demonstrating the generalizability of the resistance mechanism induced by CAFs.

    Since our results indicated consistent therapy response across all CAF clones tested (Figure 1), with the most pronounced effect observed with FB2-CAFs, we chose to focus our efforts on conducting sc-RNAseq experiments with FB2-CAFs and subsequently performed downstream validation experiments to corroborate our findings. This approach allowed us to prioritize the CAF clone with the most robust response while acknowledging the broader therapy response observed in all tested clones. Nevertheless, as requested by the reviewer, we will perform additional experiments using other independent CAF clones to assess whether the identified mechanism is broadly applicable.

    Authors show that the identified ligands secreted by CAFs (HGF, NRG1β1, etc) are found in conditioned media from CAFs. It would be good to determine if the amount of these ligands somehow is dependent on the presence of tumor cells and/or ALK-TKi. Additionally, both HGF, NRG1β1, are able to partially restore the expression of the lipogenic enzymes identified, or AKT activation pathway, but they are not able to completely restore it. Since CAF-derived CM would have both factors, maybe combination of both ligands may induce stronger rescue of the expression of these proteins.

    Author response:

    To provide a comprehensive understanding, we will investigate whether the levels of identified CAF-derived ligands are influenced by tumor cells and/or ALK-TKI treatment by performing additional assays on CAF-supernatants. Furthermore, we will explore the potential synergy between HGF and NRG1β1 in rescuing the expression of lipogenic enzymes and the AKT signal transduction pathway on protein level.

    Does PI3K/mTOR inhibitors revert the proliferation advantage of 3D heterotypic cultures? And expression of lipid biosynthesis genes?

    Author response:

    To address the impact of PI3K/mTOR inhibitors on the proliferation advantage of CAFs on ALK+ lung tumor spheroids and the expression of lipid metabolic genes, we will conduct treatment experiments using agents such as alpelisib (PI3Kα inhibitor), ipatasertib (panAKT inhibitor), or everolimus (mTORC1/2 inhibitor). Cell viability will be assessed using the 3D CellTiterGlo assay, and we will investigate changes in the expression of lipid biosynthesis genes to comprehensively evaluate the effects of these inhibitors on the resistance mechanism induced by CAFs.

    Reviewer #2 (Significance (Required)):

    This study underscores the multifaceted nature of resistance mechanisms in ALK-rearranged lung adenocarcinomas, highlighting the pivotal role of CAFs and lipid metabolic reprogramming. Lung adenocarcinoma remains a challenge in oncology, and while targeted therapy with tyrosine kinase inhibitors (TKIs) has shown promise in treating ALK-rearranged lung adenocarcinomas, the development of resistance to these therapies is nearly inevitable. This study delves into a critical aspect of this resistance by describing an important aspect of the intricate interplay between cancer-associated fibroblasts (CAFs) and tumor cells within the tumor microenvironment.

    One of the primary findings of this research is the impact of CAFs on the therapeutic response of ALK-driven lung adenocarcinoma cells. While intrinsic mechanisms within cancer cells are well-studied drivers of resistance, this study underscores the emerging importance of stromal components, particularly CAFs, in shaping therapeutic vulnerabilities. The observation that CAFs promote therapy resistance by hampering apoptotic cell death and fueling cell proliferation highlights the complexity of tumor-stroma interactions.

    The study utilizes three-dimensional (3D) spheroid co-culture models, providing a more physiologically relevant platform to investigate these interactions. This approach bridges the gap between conventional monolayer cultures and in vivo models, allowing for a deeper understanding of the role of the tumor microenvironment.

    Perhaps one of the most notable findings is the identification of lipogenesis-related genes as major players in TKI-treated lung tumor spheroids. This finding not only sheds light on a previously underexplored facet of cancer biology but also suggests that lipid metabolism may be a central determinant of therapeutic susceptibility in this context. Although data provided here suggests that it might not be the only mechanisms taking place in the development of resistance to ALK inhibitors, it clearly shows that it plays an important role in it.

    The study proposes a potential solution to overcome CAF-driven resistance by targeting vulnerabilities downstream of oncogenic signaling. The simultaneous targeting of ALK and SREBP-1, a key regulator of lipogenesis, emerges as a promising strategy to thwart the established lipid metabolic-supportive niche within TKI-resistant lung tumor spheroids.

    Author response:

    We thank the reviewer for this endorsement of our study and are gratified that the reviewer recognizes the critical implications of our research in the context of ALK-rearranged lung adenocarcinomas and their treatment resistance.

    One of the stronger limitations of this study is that it rely on a limited number of cell lines or patient-derived models, which may not fully capture the heterogeneity of ALK-rearranged lung adenocarcinomas. Furthermore, for most of the validatory assays performed, only a CAF cell line is used, higly limiting the significance of their conclusions to a more general resistance mechanism. Furthermore, the study provides valuable insights into the involvement of lipogenesis-related genes and AKT signaling but do not delve deeply into the precise molecular mechanisms underlying these processes. Further mechanistic studies are needed to understand the exact interactions and signaling pathways involved. In conclusion, while this study provides valuable insights into the role of CAFs and lipid metabolism in ALK-TKI resistance, its limitations underscore the need for further research, including more comprehensive in vivo models and clinical studies, to confirm and expand upon these findings.

    Author response:

    We appreciate the reviewer’s thoughtful assessment of our study and acknowledge the limitations highlighted in your comment. The limited number of cell lines and patient-derived models used in our study is indeed a limitation, and we agree that this may not fully capture the heterogeneity of ALK-rearranged lung adenocarcinomas. However, the number of ALK+ lung adenocarcinoma cell lines is limited (3), as is the availability of patient-derived tissue material. To address this, we are actively working on expanding our research to include a more comprehensive range of models (e.g. primary ALK+ lung cancer cells, patient-derived organoids (PDOs)) for future studies.

    We also acknowledge the importance of the limitations related to the use of a single CAF cell line in many of our validation assays. We are committed to broadening our experimental scope to involve multiple CAF cell lines to strengthen the significance of our conclusions.

    Regarding the need for deeper mechanistic studies to understand the precise molecular interactions and signaling pathways, we agree that this is a crucial point. To this end, we are planning additional mechanistic studies to uncover the exact molecular mechanisms underlying the described resistance processes in future studies.

    Reviewer #3 (Evidence, reproducibility and clarity (Required)):

    Summary: In this manuscript the authors use a spheroid co-culture model of human EML4-ALK non-small cell lung cancer (NSCLC) cell lines and fibroblasts to investigate the mechanism by which tumor-CAF crosstalk mediates non-genetic resistance to ALK inhibition. Spheroid culture of tumor cell lines with CAFs or CAF conditioned media was sufficient to reduce apoptosis and boost proliferation in the context of ALK inhibition. Using single-cell RNA-seq (scRNA-seq) of co-cultures treated with lorlatinib, the authors show that lipogenesis-associated genes are enriched in drug-treated tumor cells co-cultured with CAFs. Specifically, the authors propose that CAF-derived HGF and NRG1 derepress lorlatinib-induced changes in oncogenic Akt and mTOR signaling to boost expression of SREBP and FASN and restore lipid composition in NSCLC cancer cells.

    Major comments:

    The authors' conclusion that the effect of CAF CM on lorlatinib sensitivity is mediated by HGF and NRG is somewhat weak. Nrg1B1 was sufficient to rescue cell viability in the context of lorlatinib treatment (Figure 5B) only at a concentration significantly higher than that which was produced by fibroblast lines in culture (Figure 5A). Although the authors note that the real level of NRG1 could be higher than detected, this is speculative. Neither HGF nor NRG1 blocking antibodies appear to have rescued the elevated cell viability driven by CAF CM in the context of lorlatinib treatment (Figure 5C). These results, though statistically significant, do not appear biologically relevant. To strengthen their conclusions, the authors should consider ablating HGF or NRG1 in CAFs via shRNA or CRISPRi and then testing if CAF CM is no longer sufficient to rescue the viability of lorlatinib treated cancer spheroids.

    Author response:

    We refer the reviewer to our response to comment #1 of reviewer 1.

    In addition to the western blots used to demonstrate and effect of CAF CM or HGF/NRG1 on Akt and mTOR signaling, the authors could strengthen their conclusions by testing the effect of Akt and mTOR inhibitors on the rescue effect of CAF CM.

    Author response:

    We refer the reviewer to our response to comment #6 of reviewer 2.

    Minor comments:

    In addition to the cell death and proliferation assays shown in Figure 1B-E, it would be helpful to show and quantify images of spheroid co-cultures treated +/- lorlatinib (as in Figure 1A, Supplemental Figure S9). Although the effects on percent cell death and proliferation are significant, images of spheroid size and morphology would make these results more convincing.

    __Author response: __

    We can certainly incorporate representative images in the revised manuscript. However, we'd like to clarify that comparing spheroid mono- and direct co-cultures can be challenging due to differences in initial cell seeding numbers, variations in the growth rates of fibroblasts and tumor cells, and subsequently, differences in spheroid sizes at the initiation of treatment. These factors can confound direct comparisons between the two culture conditions upon treatment.

    In Figure 5B, the effect of CAF secreted factors on cell viability should be tested in comparison to CAF CM as a biological control. This would allow the reader to understand how the effect of each factor alone compares to the effect of CM.

    __Author response: __

    We will incorporate a comparison to CAF-CM as a biological control to provide a clearer understanding of the individual effects of CAF-secreted factors.

    **Referees cross-commenting**

    I am in agreement with all of the points made by Reviewer #1. I also suggested that the authors should use shRNA to inhibit HGF and NRG1 expression in CAFs, and am similarly concerned about both human and in vivo relevance of the authors' findings. The experiments suggested by Reviewer #1 to further characterize the fibroblast subtypes and to use non-CAF control cells are also reasonable.

    Author response:

    The reviewers alignment on the points raised is duly noted, and we understand the importance of addressing the concerns regarding the relevance of our findings. The use of shRNA to inhibit HGF and NRG1 expression in CAFs is a valuable suggestion, and we are actively considering this approach to enhance the specificity of our findings. Furthermore, we acknowledge the need for a deeper characterization of fibroblast subtypes and the inclusion of non-CAF control cells to strengthen the robustness of our research.

    I am also in agreement with the critiques presented by Reviewer #2 and find them reasonable; they would strengthen the manuscripts and better support the authors' findings. The work is indeed limited by the models used here, and mechanistic findings would be better supported by further metabolic analysis such as Seahorse or assessment of lipid synthesis.

    Author response:

    We greatly appreciate your alignment with Reviewer #2's critiques and your recognition of their reasonableness. Expanding our research to include additional models and conducting further metabolic analyses are valuable suggestions that we are actively considering to bolster the mechanistic underpinnings of our work.

    Reviewer #3 (Significance (Required)):

    As a reviewer, I have expertise in fibroblast biology and the contributions of the tumor microenvironment to pancreatic tumor development. Although my research has not focused on lung cancer specifically, I also have experience in lipid metabolism, therapy resistance, and tumor heterogeneity. In this manuscript the authors use a co-culture system to show that soluble CAF factors drive tyrosine kinase inhibitor (TKI) resistance in vitro in Alk-fusion driven NSCLC in line with prior work (Reviewed by Wong et al. 2021, Domen et al. 2021, Li et al. 2022). Mechanistically, the authors propose that CAF-secreted HGF and NRG1 restore Akt and mTor signaling pathways suppressed by lorlatinib, thus rescuing SREBP expression and the phospholipidome in TKI-treated NSCLC cells. Prior work has specifically demonstrated the ability of CAFs to rescue the effect of lorlatinib on NSCLC cell lines with ALK fusions via HGF/Met signaling (Hu et al. 2021), and the general effect of CAF-secreted HGF on therapy resistance through Akt/mTor signaling has been well characterized (CITE). The regulation of SREBP and lipid metabolism by Akt/mTOR signaling in cancer and TKI resistance have been similarly described (CITE). Thus, the authors largely connect these well-known pathways, demonstrating that CAF co-culture restores lipid-associated transcriptional programs and lipidomic profile in lorlatinib-treated cells via HGF/NRG1 activation of Akt, mTOR, and SREBP. A few points presented in the manuscript that could represent potential scientific advances include scRNA-seq analysis of CAF/NSCLC co-cultures and the implication of CAFs in TKI-resistance through the modulation of lipid metabolism. However, the scientific and clinical significance of these findings are limited by the biological systems used and by their incremental contribution in context of the current literature.

    The scRNA-seq analysis of NSCLC cells co-cultured with CAFs generated here could represent a potential advance and resource for future study; however, the application of this analysis to 2D cell lines in vitro may have limited utility as the heterogeneity of these long-culture lines is likely quite narrow (CITE OTHER scRNAseq in vivo or PDOs). The authors themselves did not leverage the scRNA-seq data for a deep analysis of cancer cell heterogeneity but rather uncovered lipid-associated transcriptional programs using bulk analysis across all tumor cells in their dataset. The significance of the authors' finding that CAF conditioned media (CM) mediates lorlatinib-sensitivity through the regulation of lipid metabolism is also somewhat limited by prior work directly implicating SREBP and phospholipid remodeling in TKI resistance (Xu et al. 2021, OTHERS). Although focusing on EGFR-mutant NSCLC, Xu et al. 2021 showed that SREBP upregulation and increased lipogenesis and decreased oxidative stress was associated with resistance to gefitinib and could be reversed by treatment with the SREBP inhibitor fatostatin in vivo. TKI-resistance is often driven by the activation of convergent signaling pathways (Akt, mTOR) in both in EGFR-mutant ALK-fusion NSCLC, so it is perhaps not particularly surprising that the authors find that lipid programs are similarly important in lorlatinib-resistance. The novelty in this manuscript is limited to the connection between CAFs and these critical lipid metabolism pathways, and the implication that SREBP inhibition similarly blocks non-genetic CAF-mediated TKI resistance. The significance of this finding might be greater if the authors explored whether fatostatin could improve therapy response to lorlatinib in vivo.

    Author response:

    We highly appreciate the reviewer’s detailed review and expertise in the fields of fibroblast biology, tumor microenvironment, lipid metabolism, and therapy resistance.

    The reviewer’s perspective on the utility of scRNA-seq analysis in our study is justified. We acknowledge that applying this analysis to 2D cell lines in vitro may have limitations due to the narrow heterogeneity of long-culture lines. We therefore attempted to enhance the relevance of our findings by applying 3D cell culture models, which are known to resemble the *in vivo *situation more closely than conventional monolayer cultures (4, 5). Nevertheless, we agree that incorporation of additional models (e.g. patient-derived organoids) would better capture the heterogeneity of the tumor and its surrounding microenvironment. We concur that a deeper analysis would enhance our understanding of the interactions between CAFs and tumor cell (sub)populations.

    The insights into the significance of our findings in the context of prior research on SREBP-dependent phospholipid remodeling in TKI resistance are well taken. We agree that the novelty of our study lies in the connection between CAFs and lipid metabolism pathways as a non-genetic CAF-mediated TKI resistance mechanism. However, it is also important to note that no prior studies have investigated stroma-driven lipid metabolic reprogramming in EML4-ALK-positive NSCLC. This unique aspect of our research adds to its originality and potential significance in advancing the understanding of ALK-positive NSCLC and therapy resistance.

    We agree with the reviewer’s point that an in vivo study would be important in exploring whether fatostatin could improve therapy response to lorlatinib. However, due to technical and timing limitations, the establishment of corresponding mouse models is beyond the scope of our present study.

    3. Description of the revisions that have already been incorporated in the transferred manuscript

    Please insert a point-by-point reply describing the revisions that were already carried out and included in the transferred manuscript. If no revisions have been carried out yet, please leave this section empty.

    Reviewer #1

    Minor issues:

    "----stronger in in case of-------" needs to be improved to "----stronger in case of-------", line 128, page 5.

    __Author response: __

    This was changed accordingly.

    Reviewer #2

    1. Based on sc-RNA-seq data authors then explore molecular processes facilitating survival of ALK-driven lung cancer cells under the influence of CAFs and mention that "among the enriched biological processes and pathways related to metabolic activities, the most striking terms were linked to lipid metabolism" (lines 182-184). Attending to the graph depicted in Fig. 3A, that is not true, finding other processes more significant. In fact, linked to metabolism, glycolysis is more significant than lipid metabolism. This sentence should be changed accordingly and a different rationale should be done to focus on lipid metabolism.

    __Author response: __

    The phrasing was changed accordingly.

    Reviewer #3

    Major comments:

    Although the authors note that the scRNA-seq data generated here may be an important resource in the field, it could also be explored in greater depth to further support the conclusions of this manuscript. As is, the scRNA-seq data is primarily analyzed at a bulk level to identify lipid-associated genes and gene sets as down-regulated by lorlatinib. In this case, it would be more useful and perhaps a better resource to conduct bulk RNA-seq in triplicate to generate a stronger dataset and generate a set of genes significantly regulated by lorlatinib and CAF co-culture or CAF CM. The scRNA-seq data could be leveraged to support the conclusions of the manuscript by plotting lipid-associated genes identified in Figure XB-C by U-map. This analysis would identify which clusters are enriched for lipid-associated genes and demonstrate whether these particular clusters are depleted by lorlatinib or rescued by CAF co-culture.

    __Author response: __

    We opted for scRNA-seq as it allowed us to simultaneously sequence co-cultivated tumor cells and fibroblasts, without the need for sorting experiments typically required for bulk RNA-sequencing experiments. With this we intended to avoid potential biases introduced by sorting procedures, which can be challenging, particularly in the case of identifying appropriate markers for fibroblasts.

    In response to the reviewer’s suggestions we have now refined our analysis to depict lipid-associated genes in a cluster-dependent manner (Supplementary Figure S9). This analysis, however, did not showcase a cluster-specific enrichment of lipid-associated genes and a demonstrated a TKI-induced depletion of these genes across all tumor cell cluster.

    The authors' conclusion that CAF co-culture restores the lipid profile of lorlatinib-treated tumor cells is somewhat weak due to the representation of lipidomic data. Although the Figure legends note that lipidomic analyses were conducted at n=3 replicates, the data as represented in Figure 8A-B do not allow the reader to assess variability across samples or the significance of the fold change differences in lipid species. Although it can be useful to view the data this way, the authors should also show variability across samples in some way via PCA plot or by including a heatmap of lipid abundance across all treatment groups and replicates. Especially as some differences appear subtle, it is also difficult to understand to what extent CAF CM rescues lorlatinib-induced effects on lipid species as values are shown as fold change relative to control for the independent groups. In this way, the reader cannot assess, for example, how lipid species abundance compares in lorlatinib-treated tumor cells +/- CAF CM. Again, a heatmap across treatment groups might be helpful in addition to an analysis for statistically significant differences in lipid abundance across treatment groups. The issues outlined here make it difficult to assess whether "addition of CAF-CM to H3122 lung tumor spheroids was able to partly abrogate this shift towards higher levels of poly-unsaturated lipid". As is, the statements describing the results in Figure 8A-B are vague and don't appear to totally align with the data. To my eyes, there is no apparent general trend in SFA or MUFA reduction in lorlatinib-treated cells as implied by the authors, though particular species may be down-regulated. The authors should also calculate saturation indices across lipid species to support their conclusion that lipid saturation is modulated by lorlatinib and rescued by CAF CM.

    __Author response: __

    Given the reviewer’s suggestions, we have made significant improvements to the presentation of our lipidomic data in the revised manuscript. We now provide a more comprehensive view of the data to allow for a better assessment of variability across samples and the significance of saturation index differences across lipid species. Specifically, we have included a PCA plot (Figure 8A) and a heatmap of lipid abundances across all treatment groups and replicates to address the issue of variability (Supplementary Figure S11). Furthermore, we have performed additional analyses to calculate saturation indices across lipid species (Supplementary Figure S12A), which support our conclusion that lipid saturation, i.e. de novo lipogenesis, is modulated by lorlatinib and rescued by CAF-CM. These additions provide a clearer visualization of the data and enhance the robustness of our findings.

    Minor comments:

    The ablation of specific cell clusters upon lorlatinib treatment in Figure 2 is compelling and visually striking. To make it easier for the reader to interpret this data, it might be useful to denote the general functional annotations of each cluster in the legend (for example, "cluster 3: proliferative"). This would allow the reader to visualize which populations are preferentially depleted by the inhibitor and rescued by CAF co-culture. Further, some quantification showing the number of cells in each cluster by treatment would group (or fold-reduction per cluster upon inhibitor treatment) would more clearly show how each cluster is impacted by the inhibitor and CAF co-culture.

    __Author response: __

    To facilitate a clearer understanding of which populations are preferentially affected by the ALK-inhibitor and rescued by CAF co-culture, we provided the cluster-specific annotations in the legend of Figure 2.

    Furthermore, we included quantifications showing the number of cells in each cluster by culture condition and treatment group (Supplementary Table S3), to provide a more comprehensive view of how each cluster is impacted by the inhibitor and CAF co-culture.

    In Supplemental Figure S3A please specify which gate is being used to quantify the percentage of dead cells shown in subsequent plots. It would also be useful to show the gating strategy used to separate labelled tumor cells and CAFs in heterotypic co-cultures by FACS so it is clear that CAF cells are not included in the cell death/proliferation analysis.

    __Author response: __

    Gates for quantification of dead cells are now specified, while the gating strategy used for analyzing cell death rates of separated tumor cells is given as requested in Supplementary Figure S3A. This gating strategy was likewise used to separate labelled tumor cells from CAFs to analyze cell cycle distributions.

    In Supplemental Figure S1B and C, please briefly clarify in the legend how fibroblast lines were cultured for collection of RNA and protein. It would be useful to know if the cells were assessed in spheroid culture and thus representative of their cell state when used for the following heterotypic co-culture experiments.

    __Author response: __

    Culture conditions were added in the figure legend as requested. Prior to generation of heterotypic tumor spheroids, fibroblasts were cultured as monolayers. Nevertheless, we also verified that the activation status is maintained following TGF-β1 removal and subsequent cultivation as homotypic fibroblast spheroid via WB analysis. We added the results as shown in Supplementary Figure S1D.

    In Figure 8C, the authors should plot all individual values in the bar graph as done in all other panels.

    __Author response: __

    This was changed accordingly

    4. Description of analyses that authors prefer not to carry out

    *Please include a point-by-point response explaining why some of the requested data or additional analyses might not be necessary or cannot be provided within the scope of a revision. *

    Reviewer #1

    Major critiques:

    The concept of this work is largely based on findings in vitro culture. The authors should perform animal experiments to convince their findings in vivo. CAFs and tumor cells would be implanted into recipient mice to determine whether AKT signaling, de novo lipogenesis and ALK-TKI resistance are increased in tumor cells by the presence of CAFs.

    __Author response: __

    We agree with the reviewer’s point that an *in vivo *study would be important in interrogating the full impact of CAFs on therapy response, AKT signaling, and *de novo *lipogenesis of ALK-driven lung adenocarcinoma cells. However, due to technical and timing limitations, establishing and performing co-injection and treatment experiments of corresponding mouse models is beyond the scope of our present study.

    The author described that human primary fibroblasts (FB1, FB2) were derived from NSCLC adenocarcinoma patients---. Have FB1 and FB2 been isolated from tumor tissues or no-tumor tissues? If these fibroblasts were isolated from tumor tissues as CAFs, why the authors added TGF-b onto the cells? The TGF-b treatment generates myofibroblastic CAFs, which is one of CAF subtypes, but fails to have inflammatory CAFs, which is another CAF subtypes.

    __Author response: __

    The fibroblasts FB1 and FB2 were indeed isolated from tumor tissue obtained from lung adenocarcinoma patients. In our study, the addition of TGF-β1 was employed as a strategy to maintain the CAF phenotype. It's important to note that CAFs exhibit considerable plasticity and can potentially lose their distinctive CAF characteristics during in vitro cultivation (6). The introduction of TGF-β1 was aimed at mimicking the tumor microenvironment and assisting in the preservation of the CAF phenotype, which was partially reflected in the increased expression of CAF markers such as αSMA and FAP (Supplementary Figure S1B and C).

    We acknowledge the existence of various CAF subtypes, including myofibroblastic and inflammatory CAFs, which can be induced by different stimuli. While TGF-β1 treatment tends to push fibroblasts more toward a myofibroblastic phenotype, other factors like IL-1 can induce an inflammatory phenotype (7). In our study, we chose to focus on the myofibroblastic CAF subtype. This decision was based on the prevalence of myofibroblastic CAFs in lung tumors and their established roles in tumor progression, poor prognosis across different cancer types, and resistance to immunotherapy in non-small cell lung cancer (NSCLC) (8, 9).

    Reviewer #2

    H2228 and H3122 cells are used indistinctively through the paper as ALK-driven lung tumor cells and, although in the discussion some reference is made regarding the worst outcomes observed for v3-driven ALK+ H2228 cells, results are considered similar for both cell lines, including sc-RNA-seq data. During analysis of sc-RNA-seq data numbers of specific genes identified at the different analysis are different, similar to the clusters identified (0-6). In order to determine the degree of overlap in the identified genes on the analysis and within clusters, it would be convenient to show tables with identified genes for each of the cell lines, together with the cluster classification of those genes.

    Author response:

    Regarding the comparison of v1- vs. v3-driven ALK+ tumors, we would like to clarify that the primary focus of our study is on the interactions CAFs and ALK-driven lung tumor cells, particularly in the context of therapy resistance. While the different ALK fusion variants are certainly of interest, our intention is not to delve into the comparative analysis of these variants in this paper. Instead, we aim to emphasize the broader impact of CAFs on ALK-driven lung tumors.

    The comparison of v1- vs. v3-driven tumors, as well as a detailed analysis of the differences between H2228 and H3122 cells, goes beyond the main focus of this paper. Incorporating a comparative analysis of specific genes for each cell line and their cluster classification would require a significantly expanded scope and could lead to a more complex and detailed study.

    In order to fully validate the effect of CAF/CAF CM in lipid biosynthesis in tumor cells, seahorse analysis would be highly beneficial, providing simultaneous measurement of multiple metabolic parameters, including glycolysis, oxidative phosphorylation, and fatty acid oxidation in homo (with and without CAF's CM or secreted ligands) and heterotypic conditions. Furthermore, it should be combined with specific substrates and inhibitors (i.e. glucose to measure acetyl-CoA production, labelled fatty acids, etc), to dissect various aspects of lipid biosynthesis and lipid metabolism and assess de novo lipogenesis, fatty acid uptake, or triglyceride.

    Author response:

    This point is well taken and we acknowledge the potential value of such comprehensive metabolic assessments. However, we would like to clarify that the Seahorse XF Analyzer can primarily measure oxygen consumption rate (OCR) and extracellular acidification rate (ECAR), in response to different substrates to interrogate key metabolic functions such as mitochondrial respiration and glycolysis. At least to our knowledge, the Seahorse analyzer does not specifically measure de novo lipogenesis and fatty acid uptake. Therefore, incorporating these assays in the revised manuscript may not directly address the central question of CAF-driven enhanced lipid biosynthesis.

    Nonetheless, we do agree with the reviewer that a more in-depth investigation of various metabolic alterations could be of interest in future studies. Given the GSEA data derived from our scRNA-seq analysis, which hints at alterations in glycolysis (Figure 3A), exploring these aspects of metabolic alterations in the context of CAF-mediated resistance effects could indeed provide valuable insights in the broader mechanisms underlying ALK-TKI resistance.

    This can be due to time or resource limitations or in case of disagreement about the necessity of such additional data given the scope of the study. Please leave empty if not applicable.

    References

    1. Nadel BB, Oliva M, Shou BL, Mitchell K, Ma F, Montoya DJ, et al. Systematic evaluation of transcriptomics-based deconvolution methods and references using thousands of clinical samples. Brief Bioinform. 2021;22(6).
    2. Maynard A, McCoach CE, Rotow JK, Harris L, Haderk F, Kerr DL, et al. Therapy-Induced Evolution of Human Lung Cancer Revealed by Single-Cell RNA Sequencing. Cell. 2020;182(5):1232-51 e22.
    3. Bairoch A. The Cellosaurus, a Cell-Line Knowledge Resource. J Biomol Tech. 2018;29(2):25-38.
    4. Friedrich J, Seidel C, Ebner R, Kunz-Schughart LA. Spheroid-based drug screen: considerations and practical approach. Nat Protoc. 2009;4(3):309-24.
    5. Pampaloni F, Reynaud EG, Stelzer EH. The third dimension bridges the gap between cell culture and live tissue. Nat Rev Mol Cell Biol. 2007;8(10):839-45.
    6. Sahai E, Astsaturov I, Cukierman E, DeNardo DG, Egeblad M, Evans RM, et al. A framework for advancing our understanding of cancer-associated fibroblasts. Nat Rev Cancer. 2020;20(3):174-86.
    7. Biffi G, Oni TE, Spielman B, Hao Y, Elyada E, Park Y, et al. IL1-Induced JAK/STAT Signaling Is Antagonized by TGFbeta to Shape CAF Heterogeneity in Pancreatic Ductal Adenocarcinoma. Cancer Discov. 2019;9(2):282-301.
    8. Mhaidly R, Mechta-Grigoriou F. Fibroblast heterogeneity in tumor micro-environment: Role in immunosuppression and new therapies. Semin Immunol. 2020;48:101417.
    9. Hanley CJ, Waise S, Ellis MJ, Lopez MA, Pun WY, Taylor J, et al. Single-cell analysis reveals prognostic fibroblast subpopulations linked to molecular and immunological subtypes of lung cancer. Nat Commun. 2023;14(1):387.
  2. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #3

    Evidence, reproducibility and clarity

    Summary:

    In this manuscript the authors use a spheroid co-culture model of human EML4-ALK non-small cell lung cancer (NSCLC) cell lines and fibroblasts to investigate the mechanism by which tumor-CAF crosstalk mediates non-genetic resistance to ALK inhibition. Spheroid culture of tumor cell lines with CAFs or CAF conditioned media was sufficient to reduce apoptosis and boost proliferation in the context of ALK inhibition. Using single-cell RNA-seq (scRNA-seq) of co-cultures treated with lorlatinib, the authors show that lipogenesis-associated genes are enriched in drug-treated tumor cells co-cultured with CAFs. Specifically, the authors propose that CAF-derived HGF and NRG1 derepress lorlatinib-induced changes in oncogenic Akt and mTOR signaling to boost expression of SREBP and FASN and restore lipid composition in NSCLC cancer cells.

    Major comments:

    1. Although the authors note that the scRNA-seq data generated here may be an important resource in the field, it could also be explored in greater depth to further support the conclusions of this manuscript. As is, the scRNA-seq data is primarily analyzed at a bulk level to identify lipid-associated genes and gene sets as down-regulated by lorlatinib. In this case, it would be more useful and perhaps a better resource to conduct bulk RNA-seq in triplicate to generate a stronger dataset and generate a set of genes significantly regulated by lorlatinib and CAF co-culture or CAF CM. The scRNA-seq data could be leveraged to support the conclusions of the manuscript by plotting lipid-associated genes identified in Figure XB-C by U-map. This analysis would identify which clusters are enriched for lipid-associated genes and demonstrate whether these particular clusters are depleted by lorlatinib or rescued by CAF co-culture.
    2. The authors' conclusion that the effect of CAF CM on lorlatinib sensitivity is mediated by HGF and NRG is somewhat weak. Nrg1B1 was sufficient to rescue cell viability in the context of lorlatinib treatment (Figure 5B) only at a concentration significantly higher than that which was produced by fibroblast lines in culture (Figure 5A). Although the authors note that the real level of NRG1 could be higher than detected, this is speculative. Neither HGF nor NRG1 blocking antibodies appear to have rescued the elevated cell viability driven by CAF CM in the context of lorlatinib treatment (Figure 5C). These results, though statistically significant, do not appear biologically relevant. To strengthen their conclusions, the authors should consider ablating HGF or NRG1 in CAFs via shRNA or CRISPRi and then testing if CAF CM is no longer sufficient to rescue the viability of lorlatinib treated cancer spheroids.
    3. The authors' conclusion that CAF co-culture restores the lipid profile of lorlatinib-treated tumor cells is somewhat weak due to the representation of lipidomic data. Although the Figure legends note that lipidomic analyses were conducted at n=3 replicates, the data as represented in Figure 8A-B do not allow the reader to assess variability across samples or the significance of the fold change differences in lipid species. Although it can be useful to view the data this way, the authors should also show variability across samples in some way via PCA plot or by including a heatmap of lipid abundance across all treatment groups and replicates. Especially as some differences appear subtle, it is also difficult to understand to what extent CAF CM rescues lorlatinib-induced effects on lipid species as values are shown as fold change relative to control for the independent groups. In this way, the reader cannot assess, for example, how lipid species abundance compares in lorlatinib-treated tumor cells +/- CAF CM. Again, a heatmap across treatment groups might be helpful in addition to an analysis for statistically significant differences in lipid abundance across treatment groups. The issues outlined here make it difficult to assess whether "addition of CAF-CM to H3122 lung tumor spheroids was able to partly abrogate this shift towards higher levels of poly-unsaturated lipid". As is, the statements describing the results in Figure 8A-B are vague and don't appear to totally align with the data. To my eyes, there is no apparent general trend in SFA or MUFA reduction in lorlatinib-treated cells as implied by the authors, though particular species may be down-regulated. The authors should also calculate saturation indices across lipid species to support their conclusion that lipid saturation is modulated by lorlatinib and rescued by CAF CM.
    4. In addition to the western blots used to demonstrate and effect of CAF CM or HGF/NRG1 on Akt and mTOR signaling, the authors could strengthen their conclusions by testing the effect of Akt and mTOR inhibitors on the rescue effect of CAF CM.

    Minor comments:

    1. In addition to the cell death and proliferation assays shown in Figure 1B-E, it would be helpful to show and quantify images of spheroid co-cultures treated +/- lorlatinib (as in Figure 1A, Supplemental Figure S9). Although the effects on percent cell death and proliferation are significant, images of spheroid size and morphology would make these results more convincing.
    2. The ablation of specific cell clusters upon lorlatinib treatment in Figure 2 is compelling and visually striking. To make it easier for the reader to interpret this data, it might be useful to denote the general functional annotations of each cluster in the legend (for example, "cluster 3: proliferative"). This would allow the reader to visualize which populations are preferentially depleted by the inhibitor and rescued by CAF co-culture. Further, some quantification showing the number of cells in each cluster by treatment would group (or fold-reduction per cluster upon inhibitor treatment) would more clearly show how each cluster is impacted by the inhibitor and CAF co-culture.
    3. In Supplemental Figure S3A please specify which gate is being used to quantify the percentage of dead cells shown in subsequent plots. It would also be useful to show the gating strategy used to separate labelled tumor cells and CAFs in heterotypic co-cultures by FACS so it is clear that CAF cells are not included in the cell death/proliferation analysis.
    4. In Supplemental Figure S1B and C, please briefly clarify in the legend how fibroblast lines were cultured for collection of RNA and protein. It would be useful to know if the cells were assessed in spheroid culture and thus representative of their cell state when used for the following heterotypic co-culture experiments.
    5. In Figure 5B, the effect of CAF secreted factors on cell viability should be tested in comparison to CAF CM as a biological control. This would allow the reader to understand how the effect of each factor alone compares to the effect of CM.
    6. In Figure 8C, the authors should plot all individual values in the bar graph as done in all other panels.

    Referees cross-commenting

    I am in agreement with all of the points made by Reviewer #1. I also suggested that the authors should use shRNA to inhibit HGF and NRG1 expression in CAFs, and am similarly concerned about both human and in vivo relevance of the authors' findings. The experiments suggested by Reviewer #1 to further characterize the fibroblast subtypes and to use non-CAF control cells are also reasonable.

    I am also in agreement with the critiques presented by Reviewer #2 and find them reasonable; they would strengthen the manuscripts and better support the authors' findings. The work is indeed limited by the models used here, and mechanistic findings would be better supported by further metabolic analysis such as Seahorse or assessment of lipid synthesis.

    Significance

    As a reviewer, I have expertise in fibroblast biology and the contributions of the tumor microenvironment to pancreatic tumor development. Although my research has not focused on lung cancer specifically, I also have experience in lipid metabolism, therapy resistance, and tumor heterogeneity. In this manuscript the authors use a co-culture system to show that soluble CAF factors drive tyrosine kinase inhibitor (TKI) resistance in vitro in Alk-fusion driven NSCLC in line with prior work (Reviewed by Wong et al. 2021, Domen et al. 2021, Li et al. 2022). Mechanistically, the authors propose that CAF-secreted HGF and NRG1 restore Akt and mTor signaling pathways suppressed by lorlatinib, thus rescuing SREBP expression and the phospholipidome in TKI-treated NSCLC cells. Prior work has specifically demonstrated the ability of CAFs to rescue the effect of lorlatinib on NSCLC cell lines with ALK fusions via HGF/Met signaling (Hu et al. 2021), and the general effect of CAF-secreted HGF on therapy resistance through Akt/mTor signaling has been well characterized (CITE). The regulation of SREBP and lipid metabolism by Akt/mTOR signaling in cancer and TKI resistance have been similarly described (CITE). Thus, the authors largely connect these well-known pathways, demonstrating that CAF co-culture restores lipid-associated transcriptional programs and lipidomic profile in lorlatinib-treated cells via HGF/NRG1 activation of Akt, mTOR, and SREBP. A few points presented in the manuscript that could represent potential scientific advances include scRNA-seq analysis of CAF/NSCLC co-cultures and the implication of CAFs in TKI-resistance through the modulation of lipid metabolism. However, the scientific and clinical significance of these findings are limited by the biological systems used and by their incremental contribution in context of the current literature.

    The scRNA-seq analysis of NSCLC cells co-cultured with CAFs generated here could represent a potential advance and resource for future study; however, the application of this analysis to 2D cell lines in vitro may have limited utility as the heterogeneity of these long-culture lines is likely quite narrow (CITE OTHER scRNAseq in vivo or PDOs). The authors themselves did not leverage the scRNA-seq data for a deep analysis of cancer cell heterogeneity but rather uncovered lipid-associated transcriptional programs using bulk analysis across all tumor cells in their dataset. The significance of the authors' finding that CAF conditioned media (CM) mediates lorlatinib-sensitivity through the regulation of lipid metabolism is also somewhat limited by prior work directly implicating SREBP and phospholipid remodeling in TKI resistance (Xu et al. 2021, OTHERS). Although focusing on EGFR-mutant NSCLC, Xu et al. 2021 showed that SREBP upregulation and increased lipogenesis and decreased oxidative stress was associated with resistance to gefitinib and could be reversed by treatment with the SREBP inhibitor fatostatin in vivo. TKI-resistance is often driven by the activation of convergent signaling pathways (Akt, mTOR) in both in EGFR-mutant ALK-fusion NSCLC, so it is perhaps not particularly surprising that the authors find that lipid programs are similarly important in lorlatinib-resistance. The novelty in this manuscript is limited to the connection between CAFs and these critical lipid metabolism pathways, and the implication that SREBP inhibition similarly blocks non-genetic CAF-mediated TKI resistance. The significance of this finding might be greater if the authors explored whether fatostatin could improve therapy response to lorlatinib in vivo.

  3. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #2

    Evidence, reproducibility and clarity

    In this review, the authors describe a possible mechanism of resistance in lung tumor cells that is induced by CAFs in response to ALK-specific inhibitors. The manuscript is well written and conclusions are in general well supported by experimental data, however additional experiments are needed to fully support the conclusions stated by the authors.

    • Brigatinib and lorlatinib are used to target ALK-driven lung tumor cells. In many of the experiments described heterotypic 3D co-cultures are used. Although both inhibitors act preferentially over the cancer cells, and no effect is expected in the CAFs, it would be desirable to confirm it.
    • H2228 and H3122 cells are used indistinctively through the paper as ALK-driven lung tumor cells and, although in the discussion some reference is made regarding the worst outcomes observed for v3-driven ALK+ H2228 cells, results are considered similar for both cell lines, including sc-RNA-seq data. During analysis of sc-RNA-seq data numbers of specific genes identified at the different analysis are different, similar to the clusters identified (0-6). In order to determine the degree of overlap in the identified genes on the analysis and within clusters, it would be convenient to show tables with identified genes for each of the cell lines, together with the cluster classification of those genes.
    • Based on sc-RNA-seq data authors then explore molecular processes facilitating survival of ALK-driven lung cancer cells under the influence of CAFs and mention that "among the enriched biological processes and pathways related to metabolic activities, the most striking terms were linked to lipid metabolism" (lines 182-184). Attending to the graph depicted in Fig. 3A, that is not true, finding other processes more significant. In fact, linked to metabolism, glycolysis is more significant than lipid metabolism. This sentence should be changed accordingly and a different rationale should be done to focus on lipid metabolism.
    • The authors demonstrated initial proliferation advantage and apoptosis protection of H2228 and H3122 cells in response to conditioned media (CM) of three independent CAF clones. However, once they identify lipid enzymes altered and specific ligand-receptor interaction, then they focus only in FB2. This imply that the mechanism described by authors is relevant for that particular clone, but it does not validate a general resistance mechanism induced by CAFs. In order to claim that this is a more general mechanism, other CAF clones should be tested.
    • Authors show that the identified ligands secreted by CAFs (HGF, NRG1β1, etc) are found in conditioned media from CAFs. It would be good to determine if the amount of these ligands somehow is dependent on the presence of tumor cells and/or ALK-TKi. Additionally, both HGF, NRG1β1, are able to partially restore the expression of the lipogenic enzymes identified, or AKT activation pathway, but they are not able to completely restore it. Since CAF-derived CM would have both factors, maybe combination of both ligands may induce stronger rescue of the expression of these proteins.
    • Does PI3K/mTOR inhibitors revert the proliferation advantage of 3D heterotypic cultures? And expression of lipid biosynthesis genes?
    • In order to fully validate the effect of CAF/CAF CM in lipid biosynthesis in tumor cells, seahorse analysis would be highly beneficial, providing simultaneous measurement of multiple metabolic parameters, including glycolysis, oxidative phosphorylation, and fatty acid oxidation in homo (with and without CAF's CM or secreted ligands) and heterotypic conditions. Furthermore, it should be combined with specific substrates and inhibitors (i.e. glucose to measure acetyl-CoA production, labelled fatty acids, etc), to dissect various aspects of lipid biosynthesis and lipid metabolism and assess de novo lipogenesis, fatty acid uptake, or triglyceride.

    Significance

    This study underscores the multifaceted nature of resistance mechanisms in ALK-rearranged lung adenocarcinomas, highlighting the pivotal role of CAFs and lipid metabolic reprogramming. Lung adenocarcinoma remains a challenge in oncology, and while targeted therapy with tyrosine kinase inhibitors (TKIs) has shown promise in treating ALK-rearranged lung adenocarcinomas, the development of resistance to these therapies is nearly inevitable. This study delves into a critical aspect of this resistance by describing an important aspect of the intricate interplay between cancer-associated fibroblasts (CAFs) and tumor cells within the tumor microenvironment.

    One of the primary findings of this research is the impact of CAFs on the therapeutic response of ALK-driven lung adenocarcinoma cells. While intrinsic mechanisms within cancer cells are well-studied drivers of resistance, this study underscores the emerging importance of stromal components, particularly CAFs, in shaping therapeutic vulnerabilities. The observation that CAFs promote therapy resistance by hampering apoptotic cell death and fueling cell proliferation highlights the complexity of tumor-stroma interactions. The study utilizes three-dimensional (3D) spheroid co-culture models, providing a more physiologically relevant platform to investigate these interactions. This approach bridges the gap between conventional monolayer cultures and in vivo models, allowing for a deeper understanding of the role of the tumor microenvironment.

    Perhaps one of the most notable findings is the identification of lipogenesis-related genes as major players in TKI-treated lung tumor spheroids. This finding not only sheds light on a previously underexplored facet of cancer biology but also suggests that lipid metabolism may be a central determinant of therapeutic susceptibility in this context. Although data provided here suggests that it might not be the only mechanisms taking place in the development of resistance to ALK inhibitors, it clearly shows that it plays an important role in it.

    The study proposes a potential solution to overcome CAF-driven resistance by targeting vulnerabilities downstream of oncogenic signaling. The simultaneous targeting of ALK and SREBP-1, a key regulator of lipogenesis, emerges as a promising strategy to thwart the established lipid metabolic-supportive niche within TKI-resistant lung tumor spheroids.

    One of the stronger limitations of this study is that it rely on a limited number of cell lines or patient-derived models, which may not fully capture the heterogeneity of ALK-rearranged lung adenocarcinomas. Furthermore, for most of the validatory assays performed, only a CAF cell line is used, higly limiting the significance of their conclusions to a more general resistance mechanism. Furthermore, the study provides valuable insights into the involvement of lipogenesis-related genes and AKT signaling but do not delve deeply into the precise molecular mechanisms underlying these processes. Further mechanistic studies are needed to understand the exact interactions and signaling pathways involved.

    In conclusion, while this study provides valuable insights into the role of CAFs and lipid metabolism in ALK-TKI resistance, its limitations underscore the need for further research, including more comprehensive in vivo models and clinical studies, to confirm and expand upon these findings.

  4. Review Commons

    Note: This preprint has been reviewed by subject experts for Review Commons. Content has not been altered except for formatting.

    Learn more at Review Commons

    Referee #1

    Evidence, reproducibility and clarity

    Daum AK et al. indicated that CAFs promote TKI resistance via lipid biosynthesis in ALK-driven lung adenocarcinoma cells. This work provides novel CAF-induced drug resistant mechanisms in ALK-driven lung cancer cells, however, there are several issues to be resolved by the following.

    Major critiques

    1. The authors claimed that CAF-produced HGF and NRG1 boosts AKT signaling and de novo lipogenesis to promote ALK-TKI resistance in lung cancer cells. They showed that CAF-secreted HGF and NRG1 inhibition attenuates tumor cell viability in the presence of FB2-CM, however, whether stromal HGF and NRG1 inhibition suppresses de novo lipogenesis in carcinoma cells has not yet been investigated. The authors should inhibit HGF and NRG1 expression in CAFs by shRNA prior to addition of CAF-CM onto cancer cells to evaluate AKT signaling and de novo lipogenesis.
    2. The concept of this work is largely based on findings in vitro culture. The authors should perform animal experiments to convince their findings in vivo. CAFs and tumor cells would be implanted into recipient mice to determine whether AKT signaling, de novo lipogenesis and ALK-TKI resistance are increased in tumor cells by the presence of CAFs.
    3. This work lacks human correlation. Are HGF and NRG1 expressions in CAFs related with de novo lipogenesis, drug resistance and poor outcomes in ALK-driven lung cancer patients?
    4. The author described that human primary fibroblasts (FB1, FB2) were derived from NSCLC adenocarcinoma patients---. Have FB1 and FB2 been isolated from tumor tissues or no-tumor tissues? If these fibroblasts were isolated from tumor tissues as CAFs, why the authors added TGF-b onto the cells? The TGF-b treatment generates myofibroblastic CAFs, which is one of CAF subtypes, but fails to have inflammatory CAFs, which is another CAF subtypes.
    5. The most experiments lack the appropriate control for CAFs. They would use primary isolated counterpart fibroblasts as the patient-specific control for lung cancer CAFs by extracting from non-cancerous regions of the same individual in their experiments.

    Minor issues

    1. In Fig 1B, coculture with TGF-b-treated MRC-5 attenuated cancer cell death with the lorlatinib treatment. However, HGF and NRG1 production is comparable between MRC-5 cells treated with or without TGF-b in Fig 5A. These data indicate that any fibroblasts but not CAFs could suppress cancer cell death with the lorlatinib treatment.
    2. The pAKT induction of H3122 treated with lorlatinib in the presence of CAM-CM or HGF or NRG in Fig 7A, B is barely observed and lacks the significance.
    3. "----stronger in in case of-------" needs to be improved to "----stronger in case of-------", line 128, page 5.

    Significance

    The concept of this work is interesting, however, molecular mechanisms underlying CAF-medicated ALK-TKI resistance remain poorly elucidated. Characterization of human primary fibroblasts (FB1, FB2) is not clearly described, and the most experiments lack proper control. Immunoblot in Fig 6 and 7 looks snap-shot and the reviewer has concerns about the reproducibility.

  5. Version published to 10.1101/2023.08.08.552439v1 on bioRxiv