“Using genome-scale CRISPR screening to create a desmoid tumor dependency map.” Year 1 of 3

Lay Abstract

While some aspects of individual desmoid tumors are shared between patients, each tumor has unique molecular and clinical characteristics. This means that selecting the right therapies for future patients with desmoid tumors requires a special type of map. This map, which we call a “dependency map,” would enable a clinician in the future to know therapy(ies) to prescribe based on the genetic characterization of that individual patient’s tumor. In our project, we aim to produce freely available foundational resources and develop a pilot version of a desmoid tumor dependency map that the entire scientific community can use. We will first test a large number (>6,000) of existing drugs that might be repurposed for desmoid tumors. But, we hypothesize that the most promising future drugs for desmoid tumors likely don’t yet exist. So, in our project we will focus on using new genetic tools, including a special type of molecular “scissors” called CRISPR/Cas9 to disable every gene in the genome in a systematic fashion and look for evidence of desmoid tumor cell killing in the laboratory. These experiments will prioritize the best targets to ensure drug companies know which targets to work on for desmoid tumors and ensure they include desmoid tumors in future drug discovery programs that might be focused otherwise on other tumor types. At the end of our project, we will complete (1) an inventory of nearly 20 desmoid tumor laboratory models that will be fully genomically characterized and available at a third-party distributor for all scientists to access, and (2) an initial prioritization of existing drugs and drug targets for desmoid tumors. This effort will help us understand what size and scale of a full Desmoid Tumor Dependency Map will ultimately be needed until we can make accurate dependency predictions for each patient.

Scientific Abstract

The success of cancer precision medicine is contingent on the ability to determine an optimal therapeutic strategy given the molecular classification of a patient’s tumor. While desmoid tumors are relatively homogenous at the genomic level, we do not yet know to what extent this homogeneity induces dependency on a limited or an expansive number of cellular targets for survival and whether such targets are shared with other common cancers. Here, we hypothesize that desmoid tumors have a finite number of cellular dependencies and that a fraction of these are shared with other common cancers with active b-catenin signaling. We propose creating a pilot version of an experimental map of such dependencies in a manner that can be fully integrated with data emerging from common cancers to provide sufficient statistical power to prioritize dependencies driven by b-catenin mutations in desmoid tumors. To do this, we will perform genome-wide CRISPR/Cas9 viability screens and test over 6,100 existing therapeutics that have been developed for any human disease for evidence of desmoid tumor cell killing, with all data going into the public domain pre-publication to serve the entire desmoid tumor research community. This resource project will yield a reference map to nominate high priority targets for drug discovery, and may support the inclusion of desmoid tumors as inclusion criteria for ongoing clinical studies.

This project aims to provide a fast, semi-high throughput and cheap animal model for identifying and/or characterizing promising drug targets for treating desmoid tumors. We have obtained proof of principle that, in an approach of mosaic multiplexed gene knockout using CRISPR/Cas9, we can identify genes that are critical for desmoid tumor formation. Using a streamlined experimental pipeline, we will be able to further assess whether additional  genes that are highly expressed in human desmoid tumors play an essential role in tumor proliferation and could serve as anchor points for novel therapeutic drug development. In addition we will further evaluate and optimize the EZH2 inhibitors GSK126 and Tazemetostat (EPZ-6438) on desmoid tumor formation in our experimental Xenopus model.  We will also test inhibitors for membrane cleavage – hence activation – of the transcription factor CREB3L1. Such inhibitors – AEBSF and PF429242 – have recently been described to block CREB3L1 activation and AEBSF has been used in an in vivo context (Feng et al. , Nat. Comm. 2017). We will copy the treatment protocol (intraperitoneal  injection) to our Xenopus model. Evidently more specific inhibitors will have to be developed since AEBSF also interferes with the proteolytic activation of SREBP, a master transcription factor for lipogenesis. We believe that our model offers a unique experimental platform that can foster the research lines of several groups active in the field.

LAY VERSION OF ABSTRACT-  “Identifying targets for therapy in a novel genetic Xenopus model for desmoid tumor formation.”

The tumor stroma includes many nonneoplastic cells and the extracellular matrix, making up the microenvironment in which neoplastic cells grow. The non-neoplastic cells are predominantly stromal fibroblasts, sometimes called “tumor associated fibroblasts”. These cells are important for the maintenance and remodeling of the microenvironment, providing the appropriate conditions for neoplastic cell growth and invasion. The role of stromal fibroblasts in promoting tumor invasion has recently been highlighted in a number of cancers such as breast cancer, gastric carcinoma, non-small cell lung carcinoma, and colorectal cancer. This interaction between the stromal fibroblasts and the neoplastic cells can occur indirectly through secreted paracrine factors, or directly by physical cell-cell contact. Desmoid tumors (DTs) are characterized by proliferating and invading fibroblastic cells embedded in depositions of extracellular matrix driven by mutations that activate β-catenin signaling. We hypothesized that desmoid tumor cells interact with the stromal fibroblasts and that this interaction is responsible for maintaining the neoplastic phenotype of tumor cells. To test our hypothesis, we proposed to answer the following questions:

Aim 1: How do the neoplastic and stromal populations of desmoid tumors differ?

Our main goal is to study the composition and potential cross-talk of sub-populations within DTs. However, it is difficult to distinguish the neoplastic, mutant DT cells from the non-neoplastic stromal, “wild-type” fibroblasts due to the invasive nature of the tumor and due to both cellular populations exhibiting a mesenchymal fibroblastic phenotype. In the past year, we utilized a single cell-based method we developed to isolate individual cells from DT patient samples to establish several individual clones in fibroblastic colony forming culture media. We characterized individual clones to study whether it is a mutant neoplastic cell, or a non-mutant stromal cell. We also characterized differences in the expression of cell surface markers. This helped us identify a unique cell surface marker that can accelerate our isolation efforts and can more sensitively quantify tumor composition of heterogeneous samples. In the coming year, we aim to take advantage of our identified markers to establish more mutant and non-mutant pairs from more samples. We also aim to perform gene expression analyses which will be used to further characterize similarities and differences between the different populations.

Aim 2: Do desmoid tumor cells interact with the surrounding stromal cells by paracrine signals?

In the past year, we analyzed the differential expression of secreted factors between mutant and nonmutant populations and found several factors that are uniquely expressed by either subpopulation. Modulation of the activity of selected factors altered the growth of DT cultures. We also co-cultured DT cells with normal cells separately using a permeable co-culture system that allows the sharing of released factors but without cell contact. We measured the effect of co-culturing on cells by viability, proliferation and migration assays to study how the two subpopulations can influence cell behavior. This work allowed us to optimize our assays and decide on readouts for subsequent studies. In the coming year, we aim to study gene expression differences to study the underlying changes in molecular biology due to secreted factors. We also aim to further study the role of candidate secreted factors on DT culture growth. The selected pathways will be disrupted by selective drugs or neutralizing antibodies and the effect on the co-culturing experiment will be measured by the same outlined assays that we have established in the last year. Our work so far has identified potentially novel pathways for further studies, and we expect that we will identify more pathways that modulate tumor-stromal interaction which can be targeted for therapy.

Aim 3: Do desmoid tumor cells interact with the surrounding stromal cells by direct cell-cell contact?

In addition to paracrine signals, invading tumor cells can also interact with the microenvironment via direct cell-cell adhesion. Molecules involved in this process include integrins, cadherins, and Notch signaling pathway components. Since β-catenin is known to interact with these molecules, directly or indirectly, we anticipate that direct cell contact will have an effect on the mixed populations. In the past year, we conducted a high throughput cell surface antigen screen to study the expression of proteins that can play a role in cell-cell interaction and how they differ between mutant and non-mutant populations. Our preliminary analysis has identified a number of integrin molecules that exhibit aberrant expression in the mutant populations. We anticipate that these molecules play a role in direct cell-cell contact. In the coming year, we will label DT cells by generating cell lines that stably express a marker, such as GFP (green fluorescence protein). We will then mix DT cells with normal cells and allow the two cell populations to grow together in direct cell contact. We will measure the effect of direct co-culturing on each cell population, taking advantage of our labeling method, by conducting viability and proliferation assays in addition to gene expression analyses. We will take a candidate approach at inhibiting these molecules that are likely to play a role in this process based on our surfaceomics data to date, in addition to any selected pathways from Aim 1, and we expect that we will identify a number of potential therapeutic targets.

By elucidating the composition of desmoid tumors, and how the tumor cells interact with the normal stromal cells, we can greatly enhance our knowledge of how they grow. DTs are invasive tumors characterized by abundant extracellular matrix deposition. Therefore, it is very likely that tumorstromal interactions are important in the way they grow and infiltrate the surrounding tissue. By isolating and characterizing individual cell populations from the tumor mass, we can gain a better understanding of what distinguishes the mutant cells from the normal cells. Our method of separating individual cells will also allow us to study variability between patient samples at the cellular level, without the confounding factor of having a different proportion of normal cells in each tumor sample. Since DTs are known to be heterogeneous, this method is key at understanding the source of this heterogeneity, which can influence treatment decisions. Elucidating what signaling pathways are involved in the process of tumor-stroma interaction will also allow us to potentially identify novel therapeutic targets that we can use to inhibit tumor growth and infiltration. We anticipate that the established single cell-derived clones, and the knowledge gained from this project of desmoid tumor biology, will be the basis for future experiments that build on our knowledge of desmoid tumor heterogeneity and tumor-stromal interactions.

LAY VERSION OF ABSTRACT- “Single cell-derived clonal analysis of desmoid tumors to investigate tumor-stroma interactions.”

There are very few randomized studies providing thorough data and evidence on treatment decisions in desmoids. In this context we need to evaluate existing evidence from publications in order to draw data-based conclusions, particularly if such an approach is intended to find transatlantic acceptance. We will focus on the following aspects of desmoid treatment by answering pico questions for:

1) Pathology and molecular genetics of desmoids and its role in prognosis and prediction of therapeutic results 2) Indications to an active treatment in different populations of desmoid tumor patients (sporadic, FAP-associated, beta-catenin mutated vs wildtype, type of mutation) a) Is watchful waiting the primary approach indicated for all? b) When to switch to active therapy in the different populations? 3) Definition of a hierarchy of available medical therapies in different indications (tumor location, extent of progression, type of desmoid) – Which assessment of treatment effect was used in the studies (no pico question) 4) Role of pain control and physical therapy?

Data will be discussed in a consensus meeting in Milano, Italy at the Istituto Nazionale dei Tumori, on June 17-18, 2018.

LAY VERSION OF ABSTRACT-  “An evidence-based consensus on the treatment of desmoids/aggressive fibromatosis.”

Desmoid tumors are proliferations of fibroblasts, myofibroblasts, and dense collagen. They are histologically similar to superficial fibromatoses such as palmar (Dupuytren’s disease) and plantar fibromatoses, as well as keloids. Unlike the superficial fibromatoses, desmoid tumors are located in the deep tissue and stratified based on an abdominal or extra-abdominal location. Historically, wide surgical resection was the treatment of choice. This often resulted in a disfiguring appearance and recurrence was common. Additionally, radiation and systemic therapies are performed with an approximate 26% rate of objective response based on RECIST. These therapies are however not without side effects.

Corticosteroids, such as triamcinolone acetonide, have long been used in the treatment of hypertrophic scars and keloids in order to decrease the size of the lesion. Proposed mechanisms of action include a decrease in the production of collagen, dissolution of insoluble collagen, a decrease in the local inflammatory process, and an increased rate of apoptosis of fibroblast and inflammatory cells. Recent reports have also evaluated the use of corticosteroid in palmar fibromatosis with promising results. We therefore aim to perform a pilot study to evaluate the response rate and tolerability of intralesional injections of triamcinolone acetonide into desmoid tumors, using RECIST to evaluate for response.

In addition to evaluating for response of the triamcinolone acetonide we will perform genetic testing for CTNNB1 mutations. New research has shown specific mutations, such as the 45F have been correlated with worse recurrence-free survival following treatment. We will observe the frequency of CTNNB1 mutations in our patient population.

LAY VERSION OF ABSTRACT-  “A Pilot Study of Intralesional Injection of Triamcinolone Acetonide for desmoid tumors.”

Desmoid-type fibromatosis (DTF) are locally aggressive benign clonal fibroblastic/myofibroblastic proliferations that are defined by infiltrative growth in deep soft tissue have an increased propensity for local recurrence but no metastatic potential. They are known to be related to perturbation in the APC/beta-catenin/Wnt-signaling pathway and are characterized by demonstration of frequent beta-catenin nuclear localization on immunohistochemistry. The majority of cases occur sporadically while rare cases arise infrequently as part of a familial syndrome (called familial adenomatous polyposis) due to germline mutation in the adenomatous polyposis gene (APC). There are 3 types of DTF based on anatomic distribution: 1. Intra-abdominal (mesenteric) fibromatosis; 2. Abdominal fibromatosis; 3. Extra-abdominal fibromatosis. These tumors are morphologically identical and essentially indistinguishable from each other. More so, the biology of these tumors is unpredictable and consistent reproducible predictive biomarkers in terms of response to therapy are lacking. Emerging data from recent large clinical studies and molecular profiling seems to indicate significant heterogeneity in tumor biology among the different types of DTF. Specifically, desmoid tumors with CTNNB1 41A gene mutation have been found to be associated with a better statistically significant 5-year recurrence free survival compared to those with S45F mutation and desmoids of abdominal site behave less aggressively than those in extra-abdominal sites. Indeed, the largest prospective outcome data on DTF from the French Sarcoma Group, suggests that anatomic location is the single most important determinant of event free survival.

In spite of the recent strides in the biology of DF, the optimal treatment remains a challenge with an unacceptably high recurrence rate with various multi-modality approaches including surgery, chemoradiation, anti-estrogenic and different molecular targeted therapies. There is thus an urgent need for thorough understanding of the molecular biology of DTF with the ultimate goal of developing better treatment strategies. There is paucity of data in the proteomic analysis of desmoid tumors. Using a novel quantitative mass spectrometry based proteomics platform, we propose to elucidate any differences in protein signatures between the different types of desmoid type fibromatosis.

LAY VERSION OF ABSTRACT-  “Quantitative Proteomics Analysis of Desmoid-Type Fibromatosis.”