Ras/MAPK signalling intensity defines subclonal fitness in a mouse model of hepatocellular carcinoma

  1. Anthony Lozano
  2. Francois-Régis Souche
  3. Carine Chavey
  4. Valérie Dardalhon
  5. Christel Ramirez
  6. Serena Vegna
  7. Guillaume Desandre
  8. Anaïs Riviere
  9. Amal Zine El Aabidine
  10. Philippe Fort
  11. Leila Akkari
  12. Urszula Hibner
  13. Damien Grégoire

Reviewed by

Read the full article

Listed in

Log in to save this article

Abstract

Quantitative differences in signal transduction are to date an understudied feature of tumour heterogeneity. The MAPK Erk pathway, which is activated in a large proportion of human tumours, is a prototypic example of distinct cell fates being driven by signal intensity. We have used primary hepatocyte precursors transformed with different dosages of an oncogenic form of Ras to model subclonal variations in MAPK signalling. Orthotopic allografts of Ras-transformed cells in immunocompromised mice gave rise to fast-growing aggressive tumours, both at the primary location and in the peritoneal cavity. Fluorescent labelling of cells expressing different oncogene levels, and consequently varying levels of MAPK Erk activation, highlighted the selection processes operating at the two sites of tumour growth. Indeed, significantly higher Ras expression was observed in primary as compared to secondary, metastatic sites, despite the apparent evolutionary trade-off of increased apoptotic death in the liver that correlated with high Ras dosage. Analysis of the immune tumour microenvironment at the two locations suggests that fast peritoneal tumour growth in the immunocompromised setting is abrogated in immunocompetent animals due to efficient antigen presentation by peritoneal dendritic cells. Furthermore, our data indicate that, in contrast to the metastatic-like outgrowth, strong MAPK signalling is required in the primary liver tumours to resist elimination by NK (natural killer) cells. Overall, this study describes a quantitative aspect of tumour heterogeneity and points to a potential vulnerability of a subtype of hepatocellular carcinoma as a function of MAPK Erk signalling intensity.

Article activity feed

  1. Version published to 10.7554/elife.76294 on eLife
  2. Review Commons

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

    Learn more at Review Commons

    Reply to the reviewers

    1. General Statements

    We thank the reviewers for their appreciation of our study and for their comments, which we believe helped us to improve our manuscript.

    We have carefully considered both the general comment on the significance of our work expressed by the Reviewers 1 and 3 and a few specific points requested by the Reviewer 2 and we believe we have answered all of the reviewers' concerns.

    2. Point-by-point description of the revisions

    Reviewers 1 and 3 agreed that our work was both original and well performed. Although they have not raised any specific issues, apart from minor editorial changes, they asked us to clarify the potential interest of our results in designing novel therapeutic interventions for hepatocellular carcinoma.

    Reviewer 1: While studies are very elegant and results convincing, it is unclear how they might be deployed to therapeutic ends.

    Reviewer 3: The authors should explain why this is an interesting finding. They mention in the abstract that this heterogeneity highlights potential vulnerabilities that could be therapeutically exploited. How do they envision this? Why is this not a trivial result and in what way can this observation help design new therapies?

    We believe that our results suggest exciting opportunities in the search for novel therapeutic options and we agree further discussion on this important issue should be included in the revised manuscript. We have now expanded the discussion on these points and commented on the clinical relevance of our findings to answer the reviewers' concern (Discussion section page 15 lines 7-15).

    Our data are consistent with the widely acknowledged role of NK cells in anti-tumour immunity (e.g. Pende et al, Frontiers Immunol. 2019, 10:1179, for review). NK activity is governed by engagement of a repertoire of activating and inhibitory receptors expressed on their surface (Shimasaki et al., Nature Rev Drug Discovery, 2020, 19, 200-218). Among the latter, homophilic interactions of CEACAM1 (CD66A) expressed on melanoma cells have been shown to protect the tumor from NK-mediated toxicity (Markel et al, J Immunol 2002; 168:2803-2810;), in a strict parallel to our interpretation.

    NK cells are relatively sparse in the peripheral blood and abundant in a healthy liver. In patients with HCC, the numbers of peripheral, liver resident and tumor-infiltrating NK all drop significantly, mainly due to the disappearance of CD56dimCD16pos cell subset, corresponding to the cytotoxic NK population. Moreover, despite the continuous expression of activating receptors, the functionality of both the cytotoxic (and the cytokine producing (CD56brightCD16neg) remaining NKs is severely impaired (Cai et al Clinical Immunology (2008) 129, 428–437). The molecular mechanisms underlying NK anergy in the context of HCC have yet to be fully elucidated. However, CEACAM1 expression has been shown to suppress NK function in hepatitis C patients (Suda et al Hepatology Communications 2018;2:1247-1258) and there is ample evidence of CEACAM1 playing a major role in hepatic disease and in particular in protection against inflammation and immune-induced hepatitis (reviewed in Horst et al Int. J. Mol. Sci. 2018, 19, 3110). Thus, CEACAM1 is a bona fide regulator of NK function that is relevant in cancer and in non-cancerous liver pathology.

    In this context, our data introduce an additional notion, namely the tumor-promoting effect of a strong ERK activation in HCC that leads to CEACAM1-mediated anergy of NK.

    How might these findings be translated into future therapeutic options for HCC? Several scenarios can be envisaged, a very attractive being a cell-mediated immunotherapy, notably either autologous or allogeneic NK transfer. These therapies, which were initially developed for hematopoietic malignancies, are currently gaining momentum for solid tumors. Infusion of modified NK cells, including CAR-NK, presents major advantages over T-cell based therapies, mainly due to a very much diminished risk of GVDH in allogenic setting and of cytokine release syndrome and neurotoxicity for autologous transfer. Moreover, because NK-mediated cytotoxicity is HLA-independent, it does not require careful haplotype matching, thus greatly increasing the speed and availability of cellular preparations (recently reviewed in Xie et al. EBioMedicine 59 (2020) 102975).

    Currently, there are 219 registered clinical trials, including 31 on HCC, for NK-mediated anti- solid tumor responses (clinicaltrials.gov). Although most of these are only in phase I or phase II, they bear a great promise for the future. Our data strongly suggest that a new combination therapy might have an improved efficiency in a subset of HCC characterized by a strong ERK activation. This would involve either activated NK or CAR-NK in combination with a FDA-approved inhibitor of the MAPK ERK, such as trametinib. Our data lead us to predict that even a partial decrease in the intensity of ERK signaling would be likely to significantly increase the efficacy of NK-mediated anti-tumor activity, at least in a subset of HCC. While we appreciate that this suggestion remains speculative at this point in time, we believe the strength and novelty of our data warrants an exploration of such novel therapeutic opportunity for this tumor type that dramatically lacks reliable treatment options.

    • Specific points Reviewer 1

    Minor edits: Editorial review for minor, infrequent word usage edits

    We apologize for any English language mistakes in the manuscript. While the formulation of the remark makes us believe that our word usage does not impair the understanding of the text, we shall of course be willing to correct it.

    Figure 1E: Not possible to read genes in left heatmap, middle heatmap very small. Figure 3D: Units at x and y axes not legible/small. 3E: scales not legible. Figure 4: typo in legend H-> G.

    We apologize for not being more careful in preparing these figures, this has now been corrected. We realize that due to the high number or genes, their names are still in a very small print in the left panel of Fig. 1E, however, the complete list the genes is given in the supplementary table 1, and we added larger image of the heat map with the table.

    Reviewer 2

    1. Is there any significant change of EMT like status in BMEL cells having H-RAS (high) vs H-RAS (low)? Several EMT markers (e.g. vimentin or loss of E-cadherin) are induced by H-Ras in BMEL cells, as we have previously reported (Akkari et al. J Hepatol, 2012). Moderate levels of Ras expression appear to be sufficient for this phenotype, since we did not detect significant differences in their expression profiles either between RASHIGH and RASLOW populations or between cells isolated from the hepatic versus the peritoneal tumors. We conclude that the phenotype of a selective advantage afforded by a high RAS expression level is not due to the EMT.

    There is no significant fold difference (MFI number) to put the sorting gates to enrich H-RAS high vs H-RAS low cells. The mRNA expression level was almost 3 fold difference. Is it correlated with protein expression level?

    This is a very valid point. We agree that the MFI difference is not strong, although it is in fact statistically significant in three independent cell sorting experiments. We were confident that the differences in the H-Ras mRNA level were reflected in the level of protein expression, since we have observed distinct transcriptional signatures as well as significant phenotypic differences in RasHIGH and RasLOW cells (Fig. 1 B and D). Nevertheless, we quite agree that the difference in protein expression level needed to be confirmed. This has now been done by immunoblot analysis of protein extracts with an antibody specific to RasG12V (Cell signaling #14412). These data have now been included in Figure 1A, and the text modified accordingly (page 4 line 9-19).

    Is there any translational relevance of these genes Al467606, Aim2, Dynap, Htra3, Itgb7, Tspan13 in HCC patients with poor survivability?

    The expression of these genes positively correlated with the level of the Ras oncogene in the ex vivo cell culture model, thus providing a nice demonstration that variation in HRAS oncogenic dosage translates into differential transcriptomic outputs. The analysis of publicly available data from the cancer genome atlas (TCGA) also showed their expression in the HCC cohort (372 patients samples). The clinical outcome of the level of their expression (shown below) is somewhat ambiguous: strong expression of ITGB7 and C16ORF54 (Human ortholog of Al467606) correlated with a better prognosis, while expression of AIM2 and DYNAP had no impact on patient overall survival. Finally, HTRA3 and TSPAN13 were associated with worse outcomes and thus constitute particularly interesting candidates for future investigations. These somewhat unexpected divergent correlations likely reflect the fact that RAS/MAPK signalling is unlikely to be the sole regulator of their expression.

    4.Is there any difference between survival curve upon grafting of H-RAS (high) vs H-RAS (low) cells in Fig.2A?

    This experiment has not been performed for ethical reasons. Indeed, the difference in tumor growth upon injection of RasHIGH vs RasLOW is statistically significant 21 days after injection (Fig. 2A, p-value= 0.008). The size of the RasHIGH tumours is rather large and we chose to sacrifice the animals before they developed any signs of suffering.

    Is there any difference of H-RAS expression between liver tumor and peritoneal tumors?

    We have quantified H-Ras expression levels by RTqPCR in the flow cytometry sorted tumoral cells derived from the liver and peritoneal tumours (Fig. 3C). In the revised version of the manuscript we provide evidence that the mRNA expression levels of the oncogene correlate with the protein expression. Therefore, while the measurement of H-Ras protein has not been performed on the tumours, we would argue that it will indeed be different in the two tumour locations.

    Please provide the data for pro-inflammatory cytokines in TME.

    These data have been shown in the Suppl. Fig. 4C.

    Please provide an explanation of the DC activation with antigen presentation though the tumor is non-necrotic or apoptotic.

    While it is true that peritoneal tumors are less necrotic and have a lower apoptotic index than the matched hepatic primary ones (Fig. 3E), significant cell death can be detected at both locations. We assume that the released antigens are sufficient for presentation by the DC, as supported by the data in Fig. 4C, D and G.

    Is the TAM showing M2 phenotypes at peritoneal tumors?

    The reviewer correctly points out that the distinction between liver and peritoneal TAM polarization is not perfectly clear-cut, since some immunosuppressive but also some inflammatory markers are present at both tumor locations (Fig. 4B and Suppl Fig4). This is not unexpected, as the spectrum of activation macrophages can undertake in vivo is neither static nor fully faithful to the M1/M2 polarization extremes inducible in vitro (see e.g. Ringelhan et al., Nat Immunol. 2018;19(3):222-232 ; Ruffell et al., Trends Immunol. 2012 33(3):119-26). We thus integrate these results with our observations of other modulated immune cell phenotypes in these tumors. Indeed, in addition to the macrophage polarization markers, we noted a more mature, activated phenotype in the peritoneal TAMs. Together with the cytokine expression profile in the two tumor locations (which is included as a supplementary table in the revised version of the manuscript) our data argue for a less inflammatory environment in the peritoneal tumors.

    Significance

    The data showed pretty promising and has a seminal impact on H-RAS high expressing HCC patients. TAM and DC showed some important immune regulation to promote HCC.

    We thank the reviewer for his appreciation of the significance of our study.

    Reviewer 3

    It is possible that RAS levels may not stay constant but dynamically go up and down. While this is a possibility that would complicate interpretations of the results, I am ok with the conclusions in the manuscript as it is, since there seems to be a significant difference between the different populations assayed.

    This is a valid point that we have addressed by comparing the H-RAS expression level in the parental BMEL population (labelled “cells before injection” in Fig 2D) to those either freshly isolated from the tumors after a rapid cell-sorting by flow cytometry (“tumors” in Fig. 2D) and then to those isolated from tumors and kept in culture for 14 days (“tumoral cell lines” in Fig. 2D). Our conclusion was that the level of RAS expression was stable upon ex vivo culture. This result does not exclude a possibility of epigenetic regulation that operated in vivo and was maintained in the subsequent cell culture. However, even if this was the case, it would not alter the conclusion of distinct selective advantage of the HRAS expression levels in the two tumoral locations.

    Significance

    The authors should explain why this is an interesting finding. They mention in the abstract that this heterogeneity highlights potential vulnerabilities that could be therapeutically exploited. How do they envision this? Why is this not a trivial result and in what way can this observation help design new therapies?

    This important point is very similar to the concern raised by the reviewer 1 and we have answered them together at the beginning of the rebuttal.

    We would like to thank again the reviewers for raising this issue, which prompted us to include the considerations of potential usefulness of our findings in the revised discussion.

  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 #3

    Evidence, reproducibility and clarity

    In this manuscript, Lozano et al. used primary hepatocyte precursor cells expressing high and low levels of oncogenic RAS to determine the dose dependent effects of RAS on tumor formation. The authors found that cells expressing different levels of RAS encounter different selection forces that modulate tumor growth. That is, high RAS expression was required in tumor cells growing in the liver, but not in the peritoneum. These findings suggest that differences in tumor microenvironment activate selection mechanisms that trigger tumor heterogeneity. The authors also show that different levels of RAS signaling cause resistance/sensitivity to NK cell attack.

    The experiments are done well, and the conclusions follow from the data, with the challenge that the RAS high and low cells are not clones but are purified from a population of cells expressing a continuum of different levels of RAS. It is possible that RAS levels may not stay constant but dynamically go up and down. While this is a possibility that would complicate interpretations of the results, I am ok with the conclusions in the manuscript as it is, since there seems to be a significant difference between the different populations assayed.

    Significance

    I find this finding conceptually only somewhat interesting, the most interesting aspect being that the liver is a more selective environment than the peritoneum. Do the authors have an explanation for this? It is certainly no surprise that tumor cells require different signaling activities to survive and proliferate in different environments. Here it is different levels of RAS, why not.

    The authors should explain why this is an interesting finding. They mention in the abstract that this heterogeneity highlights potential vulnerabilities that could be therapeutically exploited. How do they envision this? Why is this not a trivial result and in what way can this observation help design new therapies?

  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 #2

    Evidence, reproducibility and clarity

    [NOTE FROM THE EDITOR: THIS REVIEWER INDICATED S/HE DOES NOT WISH TO BE CONTACTED AGAIN]

    Anthony Lozano et al manuscript"Ras/MAPK signalling intensity defines subclonal fitness in a mouse model of primary and metastatic hepatocellular carcinoma" showed a mechanistic role of Ras/MAPK pathway in HCC with some immune mechanism. However the author needs to address the following concerns in the manuscript.

    Major Comments-

    1. Is there any significant change of EMT like status in BMEL cells having H-RAS (high) vs H-RAS (low)?
    2. There is no significant fold difference (MFI number) to put the sorting gates to enrich H-RAS high vs H-RAS low cells. The mRNA expression level was almost 3 fold difference. Is it correlated with protein expression level?
    3. Is there any translational relevance of these genes Al467606, Aim2, Dynap, Htra3, Itgb7, Tspan13 in HCC patients with poor survivability? 4.Is there any difference between survival curve upon grafting of H-RAS (high) vs H-RAS (low) cells in Fig.2A?
    4. Is there any difference of H-RAS expression between liver tumor and peritoneal tumors? 6.Please provide the data for pro-inflammatory cytokines in TME.
    5. Please provide an explanation of the DC activation with antigen presentation though the tumor is non-necrotic or apoptotic.
    6. Is the TAM showing M2 phenotypes at peritoneal tumors?

    Significance

    The data showed pretty promising and has a seminal impact on H-RAS high expressing HCC patients. TAM and DC showed some important immune regulation to promote HCC.

  5. 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

    This is a well-written, elegant manuscript where authors demonstrate that hepatocyte precursors (BMEL), when transformed with different copy numbers of oncologenic mutant H-Ras G12V, exhibit a dose-dependent fitness. Authors find that low dose H-ras mutant clones appear to be eliminated from primary tumors, but not from secondary tumours. Authors suggest that the different (primary v secondary) microenvironments, especially with respect to innate immunity, influence selection pressure differences. Specifically, investigators find that ceacam1-driven NK inhibition contributes to this clonal selection using murine models with and without adaptive immunity.

    Minor Weakness: While RAS pathway activation is present in 40% of HCC, given that few HCC are driven by Ras mutations, relevance could be called into question. Authors offer reasonable explanations for why these used the models they did.

    Minor edits: Editorial review for minor, infrequent word usage edits

    Figure 1E: Not possible to read genes in left heatmap, middle heatmap very small. Figure 3D: Units at x and y axes not legible/small. 3E: scales not legible. Figure 4: typo in legend H-> G.

    Significance

    HCC treatment options remain limited and outcomes are poor. Defining which pathways are coopted by HCC to improve fitness as primary or secondary tumors may improve treatment options. While studies are very elegant and results convincing, it is unclear how they might be deployed to therapeutic ends.

  6. Version published to 10.1101/2021.08.13.456223v1 on bioRxiv