Control of cytokine responses through the MITF-IRF4 transcription factor network in melanoma
Hematology, Oncology
Final Report Abstract
3. Summary of the results 3.1 Role of MITF in the control of inflammation-induced dedifferentiation Results published in: Riesenberg S, Groetchen A, Siddaway R, Bald T, Reinhardt J, Smorra D, Kohlmeyer J, Renn M, Phung B, Aymans P, Schmidt T, Hornung V, Davidson I, Goding CR, Jönsson G, Landsberg J, Tüting T, Hölzel M. MITF and c-Jun antagonism interconnects melanoma dedifferentiation with pro-inflammatory cytokine responsiveness and myeloid cell recruitment. Nature Communications 2015;6:87552. (pdf attached to the final report) Soon after HO4281/2-1 was accepted and funding had started, a prominent publication in Cell by other researchers (Praetorious et al.) showed that MITF/TFAP2 indeed regulates IRF4 expression and the authors also confirmed the importance of the pigmentation-related SNP as it was exactly proposed by us in the grant proposal 10. In addition, and to our surprise, siRNA loss-of-function experiments showed that IRF4 depletion had marginals or no effects on cytokine responses in melanoma cells, despite our initial overexpression experiments and the importance of IRF4 in immune cells controlling immune signaling cascades. Instead, MITF itself emerged as a very potent negative regulator of TNF-responses and cytokine expression. Together with the work by Praetorius et al. we concluded that IRF4 downstream of MITF is largely involved in the control of the melanocytic phenotype, because IRF4 cooperates with MITF in the induction of the tyrosinase gene, the key enzyme in melanin production 10. Based on this, we decided to adjust our research plans and scrutinize our interesting observation that MITF depletion strongly augmented transcriptional responses to inflammatory signaling and cytokine expression by TNF-. 3 Final scientific report HO4281/2-1 As pointed out above, inflammation promotes phenotypic plasticity in melanoma cells, a source of non-genetic heterogeneity as shown and reviewed by us (Landsberg et al., Nature 2012, Hölzel and Tüting, Nat Rev Cancer 2013)1,11, but the molecular framework was poorly understood at that time. In order to address the underlying molecular mechanisms and networks, we used various functional genomic approaches (siRNA depletion – loss-of-function experiments; doxycycline-induced conditional expression – gain-of-function experiments) and we identified a reciprocal antagonism between MITF and the stress responsive transcription factor JUN (also known as c-Jun), which regulated inflammation-induced melanoma cell dedifferentiation and cytokine responsiveness as well as cytokine production. In detail, we found that pro-inflammatory cytokines such as TNF-α instigated a gradual suppression of MITF expression through JUN. Integrating ChIP-Seq data, we showed that MITF itself binds to the JUN regulatory genomic region mediating repression of JUN expression, as confirmed by CRISPR-Cas9 based engineering of the genomic MITF binding site in melanoma cells. Consequently, the gradual reduction of MITF level in response to signals like TNF further increased JUN expression that in turn amplified TNF-stimulated cytokine expression and again led to a reduction of MITF, depicting a clear feed-forward loop of dedifferentiation. JUN is a member of the AP-1 heteromeric transcription factor complex coordinating diverse stress responses and cooperating with the NF-kB transcription factor complex in cytokine expression, which explained why the MITFlow/ JUN high (short MITFlow/JUNhigh) melanoma cell state strongly promoted cytokine expression. Furthermore, the feed-forward mechanism of gradual MITF loss and reciprocal JUN induction turned poor peak-like transcriptional responses to TNF-α into progressive and persistent cytokine and chemokine induction, which also separated human melanoma cells in “strong switchers” and “poor switchers”. Based on these molecular insights, we also hypothesized that the MITFlow/JUNhigh melanoma cell state should be associated with a distinct TME immune cell composition in human melanomas. Indeed, TCGA transcriptome data confirmed a preferential myeloid immune cell infiltrate in MITFlow/JUNhigh melanomas. We validated this finding by immunohistochemistry by analyzing an independent melanoma patient cohort. To corroborate that the MITFlow/JUNhigh cell state actively instructs a myeloid immune cell-rich microenvironment, we used a serious of syngeneic melanoma cell lines derived from the Hgf x Cdk4R24C mouse model with distinct phenotypes (MITFlow/JUNhigh, metastable MITFlow/JUNhigh; stable MITFlow/JUNhigh). Consistently, MITFlow/JUNhigh syngeneic mouse melanomas strongly recruited myeloid immune cells into the TME and therefore recapitulated the findings obtained in human MITFlow/JUNhigh melanomas. In summary, our results of the first part of HO4281/2-1 provided strong evidence that myeloid cell-directed therapies may be particularly useful for MITFlow/JUNhigh melanomas to counteract their growth-promoting and immunosuppressive functions, though this subject of ongoing and future studies. Apart from the original research publication, we also reviewed the findings of the first part of HO4281/2-1 in: Hölzel M, Tüting T. Inflammation-Induced Plasticity in Melanoma Therapy and Metastasis. Trends Immunol. 2016 Jun;37(6):364-3749. Importantly, the MITFlow/JUNhigh cell state has been also confirmed by several other studies 12,13 and discussed in several prominent reviews by other researchers14,15. Figure 1 shows a graphical summary from a recent review by Arozarena and Wellbrock, Nat Rev Cancer 2019 15. 4 Final scientific report HO4281/2-1 Figure 1. Graphical summary (figure 2) from Arozarena and Wellbrock, Nat Rev Cancer 2019. Panel b) refers to our findings from Riesenberg et al.2 that the MITFlow/JUNhigh melanoma cell state promotes a myeloid-cell rich TME. MITFlow/JUNhigh melanoma cells also express high levels of AXL, for which reason MITFlow/JUNhigh MITFlow/AXLhigh likely define largely overlapping melanoma cell states, though this has not been investigated in detail. In conclusion, with regard to aim 1, HO4281/2-1 has achieved its goals and has made significant contribution to the field of melanoma biology and adaptive cell resistance through phenotypic plasticity. In addition, results from Riesenberg et al. were a central part of a PhD thesis (Dr. rer. nat) which was awarded with summa cum laude to Stefanie Riesenberg (25.11.2016). „The melanocytic lineage factor MITF suppresses inflammatory cytokine responsiveness and myeloid cell recruitment in melanoma by antagonizing AP-1”. 3.2 Mitogenic HGF/MET signaling and inflammation-induced dedifferentiation of melanoma cells The second part of HO4281/2-1 aimed at investigating the impact of melanoma dedifferentiation on HGF/MET signaling and vice versa. Firstly, HGF is a major growth factor in the TME known to promote cancer growth. Physiologically, HGF is strongly implicated in processes related to tissue regeneration, thus wound healing mechanisms in a broader sense. Given that i) the HGF receptor MET, a receptor tyrosine kinase, is a target of MITF and ii) HGF a major TME growth factor and also transgenic driver in the Hgf x Cdk4 R24C mouse model, we explored the relationships between HGF/MET signaling, inflammatory pathways like TNF and melanoma dedifferentiation or cell states in general. Our central publication of this work is: Reinhardt J, Landsberg J, Schmid-Burgk JL, Ramis BB, Bald T, Glodde N, Lopez-Ramos D, Young A, Ngiow SF, Nettersheim D, Schorle H, Quast T, Kolanus W, Schadendorf D, Long GV, Madore J, Scolyer RA, Ribas A, Smyth MJ, Tumeh PC, Tüting T, Hölzel M. MAPK Signaling and Inflammation Link Melanoma Phenotype Switching to Induction of CD73 during Immunotherapy. Cancer Research 2017;77(17):4697–4709. (pdf attached to the final report).3 We started the project with bioinformatic analyses of large panels of melanoma cell lines generating a trajectory from a differentiated to a dedifferentiated cell state. Our intention was to identify and characterize ‘transitional cell states’ and drivers of incipient dedifferentiation. Unbiased analyses identified HGF/MET signaling and we corroborated this by complementary experimental approaches that HGF promoted melanoma phenotype switching. Firstly, ERK1/2 are known to phosphorylate MITF on S73, which on the one hand is believed to increase 5 Final scientific report HO4281/2-1 transcriptional activity, but on the other hand also primes MITF for degradation. Most notably, when combined with inflammatory signals like TNF, HGF profoundly accelerated and promoted melanoma dedifferentiation. HGF is a mitogenic factor activating ERK1/2 activity, thus triggering MITF degradation, and we also observed strong accumulation of JUN, further enforcing dedifferentiation. Proinflammatory signals (TNF) and mitogenic signals (HGF), which together represent a regenerative wound healing environment, apparently drive melanoma phenotype adaptation towards a neural-crest progenitor and/or mesenchymal-like phenotype. Apart from our previous study in the field of T cell therapy 1, others have implicated this phenotype switch, in particular in the context of resistance to BRAF inhibitors16,17. Therefore, we were also curious about our finding that HGF strongly promoted the induction of the cell surface ectonucleotidase CD73 (encoded by NT5E). CD73 converts AMP into immunosuppressive adenosine in the TME and is an emerging and promising target for cancer immunotherapy currently under intensive clinical evaluation (small molecule inhibitors, monoclonal antibodies)18. CD73 is also known as a mesenchymal stem cell marker and therefore it was noteworthy that CD73 expression labelled not only dedifferentiated melanoma cells but also those melanoma cells with an incipient dedifferentiation phenotype expressing markers of both phenotypes, differentiated and dedifferentiated, respectively. Correlative bioinformatic analyses further suggested a direct role for JUN in the regulation of NT5E (CD73) expression, which we confirmed by a serious of experiments in human melanoma cell lines. In essence, we mapped and validated the relevant JUN binding sites in the regulatory region of NT5E (CD73). Given our previous that JUN is a central orchestrator of inflammatory and mitogenic signals in melanoma cells, these results then prompted us to explore the regulation of CD73 in mouse and human melanomas under T cell immunotherapy and how this related to therapeutic efficacy/resistance. In our melanoma mouse models, we found that dedifferentiated melanomas that had escaped from adoptive T-cell therapy directed against the melanocyte differentiation antigen Pmel (also known as gp100) showed strong upregulation of CD73. Of note, we also observed a similar induction of CD73 expression by immunohistochemistry (IHC) in a human melanoma lesion that showed progressive growth under adoptive T cell therapy directed against the melanocyte differentiation antigen MART-1 (also known as MLANA). We had access to this unique patient material via our international collaboration partner Prof. Antoni Ribas from UCLA, who has conducted several clinical trials of adoptive T cell therapy in melanoma patients19,20. Actually, this patient case also nicely confirmed melanoma dedifferentiation as an adaptive resistance mechanism in human melanomas as first described in our mouse models 20. In addition, we analyzed CD73 expression by IHC in melanoma samples (pre- and under treatment with anti PD-1 immune checkpoint inhibitors) from more than 30 cases from two clinical centers (UCLA, Sydney). Our study revealed that CD73 expression is highly dynamic as anticipated from our mechanistic cell culture experiments. Of note, none of the cases in which we observed a reduction in CD73 expression was classified as progressive disease. Despite the dynamics and heterogeneity of CD73 expression, this indicated that CD73 might contribute to resistance to immunotherapy in melanoma patients, but dynamic markers and monitoring will be needed to stratify patients in order to optimize the benefit of an anti-CD73 based therapies in combination with anti-PD-1. Taken together, the second part of HO4281/2-1 also achieved its goals and elucidated the interconnection between HGF/MET signaling and inflammation-induced melanoma cell dedifferentiation. The central findings of Reinhardt et al.3 are i) proinflammatory and mitogenic 6 Final scientific report HO4281/2-1 signals from the TME strongly cooperate to induce melanoma dedifferentiation, ii) the immunotherapeutic target CD73 is a marker of established and incipient melanoma dedifferentiation and iii) the direct involvement of the stress-responsive transcription factor JUN provides an explanation for the dynamic expression of CD73 under immunotherapy in human and mouse melanomas. The findings by our work have been discussed in many recent prominent reviews on CD73/adenosine in cancer immunotherapy, melanoma resistance and melanoma cell plasticity14,15,18,21,22. In addition, results from Reinhardt et al. were a central part of a PhD thesis (Dr. rer. nat) which was awarded with manga cum laude to Julia Reinhardt (17.09.2019). „Molecular and functional characterization of inflammation-induced cell state transitions in malignant-transformed melanocytes”. 3.3 Additional contributions of HO4281/2-1 to studies on melanoma cell plasticity Within HO4281/2-1 we established substantial expertise and central datasets for human and mouse melanoma cell plasticity that critically contributed to several other publications during the funding period. In the following sections, these contributions are discussed briefly. Bald T, Quast T, …, Hölzel M*, Gaffal E*, Tüting T*. Ultraviolet-radiation-induced inflammation promotes angiotropism and metastasis in melanoma. Nature. 2014 Mar 6;507(7490):109-13. *co-corresponding authors This work identified neutrophilic inflammation in response to UV irradiation of the skin as a driver of angiotropic metastasis in mouse models of melanoma. Inflammation-induced dedifferentiation of melanoma cells promoted the reactivation of migratory cues of neuralcrest progenitor cells, which are known to be a highly migratory cell population during embryonic development. Expertise, generation and analysis of data sets of TNF-treated human melanoma cells generated within HO4281/2-1 contributed to this work. Bald, T., Landsberg, J., …, Jönsson, G., Hölzel, M. & Tüting, T. Immune cell-poor melanomas benefit from PD-1 blockade after targeted type I IFN activation. Cancer Discov. 4, 674–687 (2014). This work identified a strategy how to overcome the immune cell poor phenotype of MITF high melanomas by targeted activation of the type I interferon system using innate immune ligands in mouse models. Expertise and analysis of melanoma phenotypes established within HO4281/2-1 contributed to this work. Hölzel M, Landsberg J, …, Jönsson G, Tüting T. A Preclinical Model of Malignant Peripheral Nerve Sheath Tumor-like Melanoma Is Characterized by Infiltrating Mast Cells. Cancer Research 2016;76(2):251–263. This work characterized a novel mouse melanoma model based on the two oncogenic driver lesion BrafV600E x Cdk4R24C. Expertise in integrative bioinformatic analyses of melanoma phenotypes established within HO4281/2-1 critically contributed to this work. In this particular model, it was found that Braf V600E x Cdk4R24C mouse melanoma tend to dedifferentiate spontaneously exhibiting a broad spectrum of morphological phenotypes resembling malignant peripheral nerve sheath tumors (MPNST-like) also with prominent angioblastic components. Of note, such phenotypes could be recapitulated in the TCGA 7 Final scientific report HO4281/2-1 dataset of human melanomas by correlating transcriptome signature with respective morphological features of the MPNST-like human melanomas. A B C Figure 2. (A) Morphologic features of MPNST in amelanotic BrafV600E-Cdk4R24C melanomas. Representative H&E- stained tissue sections demonstrating the spectrum of histomorphologic appearances of an amelanotic BrafV600E- Cdk4R24C tumor. The larger panel in the middle shows an overview of the tumor at lower magnification (x2.5). The smaller panels exemplify the heterogeneity of the histologic appearance found in a single tumor (magnification, x40). Top left, nest of melanocytic tumor cells. Top right, hemangiopericytoma-like vascular growth pattern. Bottom left, cell- poor myxoid growth pattern. Bottom right, tumor cell–rich fascicular growth pattern. Size bars, magnifications. (From figure 2 in Hölzel et al., Cancer Res 2016)6. (B) Comparative transcriptomic characterization of amelanotic MPNST-like and pigmented BrafV600E-Cdk4R24C melanomas. Outline of experimental approach for transcriptomic characterization of MPNST-like (n=5) versus pigmented (n=5) BrafV600E-Cdk4R24C melanomas using microarrays and GSEA. (C) Left panel: volcano plot visualizing the strategy to identify top differentially expressed genes between amelanotic MPNST-like (n = 5) and pigmented (n = 5) BrafV600E-Cdk4R24C tumors using moderated empirical Bayes t-test statistics for building the MPNST-like gene (= BCMPNST-like) signature. FDR-corrected P values are represented as log10 values on the y-axis. The x-axis shows differences (log2) in gene expression between the groups. Magenta dots represent genes (n = 265) that were used to build the BCMPNST-like signature. Right panel: Expression of the BCMPNST-like signature (human homologs converted) in different human tumor entities. Significance was determined by an unpaired two-sided pairwise t-test with FDR adjustment for multiple comparisons. **, p < 0.01; ***, p < 0.001. (From figure 3 in Hölzel et al., Cancer Res 2016)6. Falletta P, …, Hölzel M, Goding CR. Translation reprogramming is an evolutionarily conserved driver of phenotypic plasticity and therapeutic resistance in melanoma. Genes & Develompent 2017;31(1):18–33. This work found translational reprogramming as an important level of regulation of melanoma cell plasticity in response to nutrient starvation and inflammation. Datasets, expertise and analysis of melanoma phenotypes established within HO4281/2-1 contributed to this work. Glodde N, Bald T, …, Scolyer RA, Long GV, Janzen V, Teng MWL, Kastenmüller W, Mazzone M, Smyth MJ, Tüting T*, Hölzel M*. Reactive Neutrophil Responses Dependent on the Receptor Tyrosine Kinase c-MET Limit Cancer Immunotherapy. Immunity 2017;47(4):789- 802.e9. *co-senior authors. This work identified reactive recruitment of neutrophils into melanoma tissues and secondary lymphoid organs during adoptive T cell therapy as immune suppressive cell 8 Final scientific report HO4281/2-1 population. Expertise and analysis melanoma phenotypes driving enhanced myeloid cell recruitment established within HO4281/2-1 contributed to this work.
Publications
- Immune cell-poor melanomas benefit from PD-1 blockade after targeted type I IFN activation. Cancer Discov. 4, 674–687 (2014).5
Bald, T., Landsberg, J., Lopez-Ramos, D., Renn, M., Glodde, N., Jansen, P., Gaffal, E., Steitz, J., Tolba, R., Kalinke, U., Limmer, A., Jönsson, G., Hölzel, M. & Tüting, T.
(See online at https://doi.org/10.1158/2159-8290.cd-13-0458) - Ultraviolet-radiation-induced inflammation promotes angiotropism and metastasis in melanoma. Nature. 2014 Mar 6;507(7490):109-13
Bald T, Quast T, Landsberg J, Rogava M, Glodde N, Lopez-Ramos D, Kohlmeyer J, Riesenberg S, van den Boorn-Konijnenberg D, Hömig-Hölzel C, Reuten R, Schadow B, Weighardt H, Wenzel D, Helfrich I, Schadendorf D, Bloch W, Bianchi ME, Lugassy C, Barnhill RL, Koch M, Fleischmann BK, Förster I, Kastenmüller W, Kolanus W, Hölzel M, Gaffal E, Tüting T
(See online at https://doi.org/10.1038/nature13111) - MITF and c-Jun antagonism interconnects melanoma dedifferentiation with pro-inflammatory cytokine responsiveness and myeloid cell recruitment. Nature Communications 2015;6:8755.2
Riesenberg S, Groetchen A, Siddaway R, Bald T, Reinhardt J, Smorra D, Kohlmeyer J, Renn M, Phung B, Aymans P, Schmidt T, Hornung V, Davidson I, Goding CR, Jönsson G, Landsberg J, Tüting T, Hölzel M
(See online at https://doi.org/10.1038/ncomms9755) - A Preclinical Model of Malignant Peripheral Nerve Sheath Tumor-like Melanoma Is Characterized by Infiltrating Mast Cells. Cancer Research 2016;76(2):251–263.6
Hölzel M, Landsberg J, Glodde N, Bald T, Rogava M, Riesenberg S, Becker A, Jönsson G, Tüting T
(See online at https://doi.org/10.1158/0008-5472.can-15-1090) - Inflammation-Induced Plasticity in Melanoma Therapy and Metastasis. Trends Immunol. 2016 Jun;37(6):364-374.9
Hölzel M, Tüting T
(See online at https://doi.org/10.1016/j.it.2016.03.009) - MAPK Signaling and Inflammation Link Melanoma Phenotype Switching to Induction of CD73 during Immunotherapy. Cancer Research 2017;77(17):4697–4709.3
Reinhardt J, Landsberg J, Schmid-Burgk JL, Ramis BB, Bald T, Glodde N, Lopez-Ramos D, Young A, Ngiow SF, Nettersheim D, Schorle H, Quast T, Kolanus W, Schadendorf D, Long GV, Madore J, Scolyer RA, Ribas A, Smyth MJ, Tumeh PC, Tüting T, Hölzel M
(See online at https://doi.org/10.1158/0008-5472.can-17-0395) - Reactive Neutrophil Responses Dependent on the Receptor Tyrosine Kinase c-MET Limit Cancer Immunotherapy. Immunity 2017;47(4):789-802.e9
Glodde N, Bald T, van den Boorn-Konijnenberg D, Nakamura K, O’Donnell JS, Szczepanski S, Brandes M, Eickhoff S, Das I, Shridhar N, Hinze D, Rogava M, van der Sluis TC, Ruotsalainen JJ, Gaffal E, Landsberg J, Ludwig KU, Wilhelm C, Riek-Burchardt M, Müller AJ, Gebhardt C, Scolyer RA, Long GV, Janzen V, Teng MWL, Kastenmüller W, Mazzone M, Smyth MJ, Tüting T, Hölzel M
(See online at https://doi.org/10.1016/j.immuni.2017.09.012) - Translation reprogramming is an evolutionarily conserved driver of phenotypic plasticity and therapeutic resistance in melanoma. Genes & Develompent 2017;31(1):18–33.7
Falletta P, Sanchez-Del-Campo L, Chauhan J, Effern M, Kenyon A, Kershaw CJ, Siddaway R, Lisle R, Freter R, Daniels MJ, Lu X, Tüting T, Middleton M, Buffa FM, Willis AE, Pavitt G, Ronai ZA, Sauka-Spengler T, Hölzel M, Goding CR
(See online at https://doi.org/10.1101/gad.290940.116)
