Characterization of a Novel Mouse Model for Angiosarcoma in Which Combined Inhibition of mTOR and MEK Results in Tumor Suppression
A dissertation submitted to the Graduate School of the University of Cincinnati in partial fulfillment of the requirements to the degree of
Doctor of Philosophy (Ph.D.)
in the Department of Cancer and Cell Biology of the College of Medicine March 16, 2017
By
Michelle Lynn Chadwick
B.S. Purdue University, 2011
Dissertation Committee: Lionel ML. Chow, M.D., Ph.D. (Chair) Elisa Boscolo, Ph.D. Biplab Dasgupta, Ph.D. James Driscoll, M.D., Ph.D. Nicolas Nassar, Ph.D.
Abstract
Angiosarcoma is a rare malignancy of the endothelium, making up about 2% of sarcoma diagnoses. The causes of this disease are relatively unknown, especially in the pediatric population. In adults, angiosarcoma can be associated with exposure to certain chemicals such as vinyl chloride, prior radiation for other types of cancer, and chronic lymphedema. Angiosarcomas frequently occur in soft tissue, the liver, and bone. Currently, the best treatment for angiosarcoma is complete surgical resection. If complete resection is unachievable due to anatomic location, chemotherapeutics are used, although these are not very effective. Five year median survival remains poor around 31-43%. This is due to the metastatic nature of the disease as well as the ineffectiveness of current therapeutic regimens. As angiosarcoma is so rare, there has been little research done regarding the pathogenesis of the disease. There is a lack of model systems in which to study angiosarcoma in order to develop novel therapeutic strategies.
Some research has been conducted to determine the mutations and signaling aberrations that lead to angiosarcoma. High levels of activation of the mTOR
(mammalian target of rapamycin) and mitogen activated kinase (MAPK, or ERK, extracellular signal-regulated kinase, referred to ask MAPK throughout this thesis) have been discovered in patient samples. Mutations in members of the RAS (a
GTPase) gene family have been found. RAS can activate either the phosphoinositide
3 kinase (PI3K) pathway or the MAPK pathway, both of which have been implicated in angiosarcoma. Sequencing of angiosarcomas discovered frequent mutations in receptor-type protein tyrosine phosphatase beta (PTPRB) as well. This gene encodes
ii a protein tyrosine phosphatase that dephosphorylates receptors specifically in the endothelium, leading to negative regulation of the MAPK pathway.
Protein tyrosine phosphatase nonreceptor 12 (PTPN12) is another protein tyrosine kinase that is known to dephosphorylate receptor tyrosine kinases (RTKs) as well as tyrosine kinases involved in cell migration and adhesion.
Developmentally, Ptpn12 knockout (KO) in mouse models results in embryonic lethality around E9.5, largely due to defects in vasculogenesis and hepatogenesis.
Loss of this protein can also be found in a variety of cancers, including breast, prostate, colon, and esophageal. This loss is often associated with decreased survival. In breast cancer, loss of PTPN12 is typically found in the highly aggressive triple negative breast cancer (human epidermal growth factor receptor 2 (HER2), estrogen receptor (ER), and progesterone receptor (PR) negative). Loss of this protein is mutually exclusive with amplification of the RTKs epidermal growth factor receptor (EGFR) and HER2, leading to the hypothesis that this loss substitutes for amplification of the RTKs by activating the MAPK pathway. Loss of PTPN12 also results in an increase in migration and metastasis.
The immediate goal of these studies is to develop a novel mouse model of angiosarcoma to study the role of PTPN12, disease pathogenesis, and to create a model system in which to test new therapeutic strategies, which are greatly needed to improve the outcome for these patients. Here, we aim to 1) develop and characterize a mouse model that recapitulates human disease, 2) to find signaling pathways elevated in response to our genetic modifications and, more specifically, investigate the role of PTPN12, and 3) to determine whether inhibition of these
iii pathways results in disease regression. Our long-term goal is to translate the information we gain from our mouse model to find a novel therapeutic strategy for use in patients in order to improve the currently poor outcomes for these patients.
Here we describe a novel genetically engineered mouse model for angiosarcoma that utilizes the deletion of transformation related protein 53 (Trp53, the mouse equivalent of the TP53 gene in humans); phosphatase and tensin homolog (Pten); and Ptpn12 under a glial fibrillary acidic protein (GFAP)-CreER promoter. TP53 and PTEN abnormalities have been described in vascular anomalies;
PTPN12 is thought to play a role similar to PTPRB, which has recently been implicated in angiosarcoma. Mice of all genetic combinations were generated; only combinations that incorporate Ptpn12 deletion develop angiosarcoma, leading to the hypothesis that loss of this gene is important for pathogenesis of the disease.
Novel models for angiosarcoma are critical in order to develop better therapeutic strategies to prolong the life expectancy for these patients. Using this model, we find activation of the PI3K/mTOR (mammalian target of rapamycin) and MAPK pathways, pathways that have also been implicated in human disease. We hypothesized that these pathways were important as they are known to be regulated by PTEN and PTPN12, two of the genes we have deleted. These pathways are often highly elevated in human angiosarcoma as well, demonstrating their importance in both murine and human disease.
We further try to define the role of PTPN12 in these tumors utilizing cell culture. We tried numerous techniques to culture primary tumor cells.
Unfortunately, our attempts thus far have been unsuccessful. We would like to
iv pursue tissue culture to determine the role of PTPN12 as we believe loss of this tumor suppressor is important for angiosarcoma formation. Knowledge of the substrates of PTPN12 in this context could provide novel therapeutic targets for this disease.
We have demonstrated that the PI3K/mTOR and MAPK pathways are activated in these tumors, alterations that we expected due to the loss of PTEN and
PTPN12 which are responsible for negative regulation of these pathways. To determine the involvement of these pathways in the growth of tumors we treated the mice with an mTOR inhibitor (rapamycin), a MEK (MAPK/ERK kinase) inhibitor
(trametinib), or a combination of the two. We find that inhibition of both pathways in combination results in sustainable tumor suppression. This demonstrates the usefulness of our mouse model in testing therapies and provides a novel therapeutic option for people with angiosarcoma that could result in improved outcomes.
v vi Preface
Excerpts from the following work is in preparation for the following peer-reviewed journals:
1. Chadwick, ML; Lane, A; Thomas, D; Smith, AR; White, AR; Feng, Y; Boscolo, E;
Zheng, Y; Adams, D; Gupta, A; Veillette, A; Chow, LM. Combined mTOR and MEK inhibition is an effective therapy in a novel mouse model for angiosarcoma.
Manuscript in preparation for JCI.
vii Acknowledgements
During the course of my thesis work I have had the privilege to interact with many talented scientists. First, I have to thank my mentor, Dr. Lionel Chow, who spent countless hours training me and guiding me in data interpretation. I truly wouldn’t be the scientist I am without his guidance and support. As a third year student he encouraged me to present my research at an international conference.
Here I had the pleasure of meeting Dr. Elisa Boscolo, who would become a member of my thesis committee. I would like to thank her for allowing me to take advantage of her endothelial cell expertise and always making time to speak with me, as well as her generous sharing of reagents. Dr. Nicolas Nassar has been incredibly helpful and always willing to listen and offer advice, scientific and otherwise. Dr. James Driscoll and Dr. Biplab Dasgupta were part of my qualifying committee as well as my thesis committee; I would like to thank them for their guidance and help throughout my entire graduate career. I would also like to thank all the members of my committee for all of their helpful feedback during committee meetings; their insightful questions often made me think of my project in a new light. They have all been instrumental in my development as a scientist.
I would also like to thank all the members of the Chow, both past and present, for their support. I enjoyed our conversations; Mandi Smith and Angela
White truly made it a joy to come to lab every day. I’m glad I had the opportunity to meet them and work with them. This work wouldn’t have been possible without them.
viii My family has been integral for my success as well. I thank them for their constant encouragement. My mom has always been there to listen to me complain when my experiments don’t go as planned (which happens often in science) or meet me for a weekend trip to my alma mater. My dad and stepmom have been incredibly generous in their support. I appreciate my dad’s offers to ship me horseshoe crabs to determine whether they can be useful in treating cancer after he read that a group is studying the effects of their blood on cancer cells.
Finally, I need to thank my husband, Adrien Chadwick, without whom none of this would have been possible. I can’t express how much I appreciate your unconditional love and support that I have been so lucky to receive. You are the best partner I could ask for and I’m so excited to see what the future holds for us!
ix Table of Contents Abstract…………………………………………………………………………………………….…ii Preface………………………………………………………………………………………………..vi Acknowledgements……………………………………………………………………………vii List of Figures and Tables……………………………………………………………………xi Chapter 1: Introduction….…………………………………………………………………………….….1 Angiosarcoma………………………………………………………………………………….…...2 Phosphatases and PTPN12…..………………………………………………………….…..12 mTOR and MAPK Pathways…………………………………………………………….…...26 Chapter 2: Development and Characterization of a Novel Genetically Engineered Mouse Model for Angiosarcoma ………………………………………..……….36 Introduction……………………………………………………………………………………….37 Results…………………………..…………………………………………………………………...41 Discussion….……………………………………………………………………………………….46 Methods……………………………………..……………………………………………………….53 Figures……………………..…………………………………………………………………………57 Chapter 3: Development of an Angiosarcoma Cell Line from Murine Tumors ………………………………………………………………………………………………………..69 Introduction……………………………………………………………………………………….70 Results………………………………………………………………………………………………..73 Discussion…………………………………………………………………………………………..78 Methods…………………………………………..………………………………………………….83 Figures………………………………………………………………………………………………..87 Chapter 4: Combined Inhibition of MEK and mTOR Results in Sustained Tumor Suppression ………………………………………………………………………………………………..100 Introduction……………………………………………………………………………………..101 Results……………………………………………………………………………………………...103 Discussion………………………………………………………………………………………...109 Methods……………………………………………………………………………….…………...114 Figures…………………………………………………………………………………………...…118 Chapter 5: Discussion and Future Directions………………………………………………131 Summary…………………………………………………………………………………………..145 Acknowledgements…………………………………………………………………………..145 References…………………………………………………………………………………………………..146
x List of Figures and Tables Chapter 1: Introduction Figure 1.1 Angiosarcoma following treatment for breast cancer 3 Figure 1.2 Histology for angiosarcoma of the liver 4 Figure 1.3 Schematic of PTPN12 signaling 19 Figure 1.4 Schematic of mTOR and MAPK pathways 28
Ch. 2: Development and Characterization of a Novel Genetically Engineered Mouse Model for Angiosarcoma Figure 2.1 Ptpn12 deletion leads to angiosarcoma in mice 57 Figure 2.2 mTOR and MAPK pathways are activated in murine angiosarcoma 59 Figure 2.3 Murine angiosarcomas closely mimic human disease 61 Supp. Figure 2.1 Mouse model deletes genes in a subset of endothelial cells leading to angiosarcoma 63 Supp. Figure 2.2 PDGFR-β is phosphorylated in angiosarcoma while the VEGF receptors are not 65 Supp. Figure 2.3 Proteins around 40 and 70kd are phosphorylated in human and murine angiosarcomas 67 Supp. Table 2.1 Positions of RTK array spots 68
Chapter 3: Development of an Angiosarcoma Cell Line from Murine Tumors Figure 3.1 Cells grown in monolayer conditions grow slowly 87 Figure 3.2 Cells grow best with Collagenase I, but do not grow in a homogeneous sheet 88 Figure 3.3 Monolayer cells do not have Trp53, Pten, or Ptpn12 recombined 89 Figure 3.4 Stimulation experiment resulted in better growth but the cells do not express characteristic markers 90 Figure 3.5 Sphere cultures express expected markers but not recombination by PCR 92 Figure 3.6 Digestion results in a change in cell signaling 94 Table 3.1 Summary of monolayer cell culture trials 95 Table 3.2 Digestion with 0.1% Collagenase I results in most live cells 97 Supp. Figure 3.1 CD31 cell sorting results in a 70-80% pure population of endothelial cells 98
Chapter 4: Combined Inhibition of MEK and mTOR Results in Sustained Tumor Suppression Figure 4.1 Rapamycin is effective down to 1mg/kg 118 Figure 4.2 Trametinib is effective down to 1mg/kg 120 Figure 4.3 Five-day treatment with trametinib, rapamycin, or a combination of both results in disease regression 122
xi Figure 4.4 Long-term treatment using both rapamycin and trametinib results in sustainable disease regression 124 Table 4.1 Statistical analysis of waterfall data 126 Supp. Figure 4.1 Combination therapy results in best inhibition of mTOR and MAPK signaling pathways 127 Supp. Figure 4.2 Dual treatment results in toxicities, even at a dose of 0.5mg/kg of each drug 129 Supp. Figure 4.3 Therapeutic resistance is not due to re-activation of inhibited pathways 131
xii Chapter 1: Introduction
1
Angiosarcoma
Etiology of Angiosarcoma. Angiosarcoma is a malignancy thought to arise from endothelial cells that is quite rare, comprising only about 2% of all sarcomas. The five year survival remains low at around 31-43% (1-7). As the disease is so rare, there is little known about its etiology, although there are some known causes in adults. There is a known association between angiosarcoma formation and exposure to certain chemicals such as vinyl chloride (2). Prior radiation treatment for cancer and lymphedema, especially in breast cancer patients, is known to result in formation of secondary angiosarcomas (2). In rare cases, benign vascular lesions
(hemangiomas) can transform into angiosarcoma in children (8). However, the cause of these tumors is often unknown, especially in the pediatric population. Due to the low incidence of this disease, it is difficult to conduct meaningful studies concerning the pathogenesis of angiosarcoma. Therefore, generation of model systems in which to study this disease is critical for gaining knowledge of the pathology of angiosarcoma as well as for finding novel treatments.
Clinical Presentation. When these tumors present subcutaneously, they often appear like a bruise or lesion that does not go away (Figure 1.1). Other angiosarcomas may cause lymphedema, pain, or hemorrhage. Due to these subtle symptoms, tumors often are not detected until they are quite large and have already metastasized. The most common location for metastasis is the lung, although it has also been shown to metastasize to the liver, bone, and lymph nodes (2).
2
Figure 1.1: Angiosarcoma following treatment for breast cancer (2).
Angiosarcomas have been known to arise in the liver, breast (usually secondary following treatment for breast cancer), or subcutaneously (especially in the head and neck) (3). There is evidence that angiosarcomas occurring in different locations may express different mutations and gene expression levels. For example, one study showed that angiosarcomas of the liver, heart, bone, or spleen correlated with worse survival and that the location of the tumors was among the most influential factors in progression free survival- the amount of time that elapses before disease progression- (PFS) and overall survival (OS) in a cohort of 161 patients (3).
Histology. Histologically, these tumors are often infiltrating with no clear borders, making complete surgical resection difficult. They contain anastomosing channels, which are irregular blood vessels that form immediately adjacent to one another.
The endothelial cells forming the tumor have large atypical nuclei and a high proliferative index. The more aggressive high-grade tumors demonstrate decreased differentiation of the cells and necrosis. In order to diagnose angiosarcoma, typically immunohistochemical (IHC) stains for an endothelial marker, CD31 (or PECAM,
3 platelet endothelial cell adhesion molecule) or CD34 and a lymphatic marker, prospero homeobox protein 1 (PROX1) or podoplanin are performed.
Figure 1.2: Histology for angiosarcoma of the liver. Podoplanin (lymphatic marker) stains negatively while the endothelial marker CD34 stains positively. Adapted from (9).
CD31 positivity implies that the tumors are arising from endothelial cells while
PROX1 negativity demonstrates that the cells are not arising from the lymphatic system, leading to the conclusion that the cells are of vascular endothelial origin and therefore angiosarcoma rather than lymphangiosarcoma (Figure 1.2) (2, 10, 11).
Current treatment for angiosarcoma. The most effective treatment for these tumors is surgical resection; survival corresponds with extent of surgical resection (4).
However, due to the location of many of these tumors and extent of metastases, resection is often not possible. Some studies indicate improved survival when both surgery and radiotherapy are used compared with either treatment alone (12).
Chemotherapeutics, most often taxanes and anthracyclines, are used as well, either following surgery or on their own (4, 13, 14). Despite clinical responses of paclitaxel ranging from 58% to 89% (depending on location); median time until disease progression remain dismal at 5-7.6 months (15, 16). Further, ultimately the majority
4 of angiosarcoma patients relapse. Even surgical resection often results in relapse; in a retrospective study 24/59 of patients who underwent surgery experienced relapse (6). Unfortunately, it has been difficult to develop new therapies for angiosarcoma due to its low incidence. Clearly there is a need for model systems in which to study angiosarcoma to develop more effective treatments.
Prognosis remains poor for angiosarcoma patients. Despite use of surgical resection, chemotherapy, and radiotherapy, prognosis for angiosarcoma remains poor with a median five-year survival of 31-43%. A study by Fayette et al identified several factors that can influence survival including location of the tumor as mentioned previously, metastases, necrosis, size of the tumor, and extent of surgical resection.
The presence of metastatic disease shortened median overall survival to one year.
Necrosis is also a strong negative prognostic factor; no patients with necrosis were alive after five years. Small tumor size (<5 cm) is associated with better outcomes with a five-year survival of 32% compared with larger tumors at 13% (3). Another study found the same trend (17). The extent of surgical resection was the best- correlated factor with overall survival (3). A recent study by Buehler confirmed these prognostic factors. Further, this study quantified the advantage that surgical resection gives to survival: median overall survival for those who underwent surgery was >60 months compared with 8 months for those who did not undergo surgery (4). Better treatment options are critical, especially for the cohort of patients who are unable to achieve total surgical resection.
5 VEGF receptors in angiosarcoma. The role of the vascular endothelial growth factor receptors (VEGFRs) and their ligands in these tumors is a matter of interest as they are highly involved in angiogenesis and endothelial cell signaling. One study characterized the expression of these proteins in angiosarcoma, finding expression of VEGFA in 32/34, VEGFC in 4/34, VEGFR1 in 22/34, VEGFR2 in 22/34, and
VEGFR3 in 27/34 tumors. Interestingly, this study found that prognosis was significantly better for those with tumors expressing VEGFR2; median survival of low-expressing tumors was 14 months compared with 34 months for those with high expression. This suggests that the role of VEGFR2 may not be in signaling but may rather be a marker of a more-differentiated, less aggressive tumor (18).
This data is corroborated by the disappointing results of VEGFR inhibitors.
VEGF inhibitors are used to inhibit angiogenesis; therefore, they have been of interest for use in angiosarcoma. Bevacizumab (Avastin) is a monoclonal humanized antibody directed against VEGFA that has been tested to stop angiogenesis in a number of other tumor types. One study looking at the effects of Bevacizumab in patients with angiosarcoma found that of the 23 angiosarcoma patients, 2 had a partial response, 11 stable disease, and 10 progressive disease at first evaluation.
Median progression time for these patients was 12 weeks while overall survival was
52.7 weeks (19). Work examining the efficacy of sorafenib, an RTK inhibitor that inhibits the BRAF and VEGF receptors, in angiosarcoma has been equally disappointing. One study found that among 33 angiosarcoma patients and 4 epitheliod hemangioendothelioma patients, sorafenib treatment resulted in 5/37 responses as well as a 6-month progression free survival (PFS) rate around 30%
6 (20). Another study focusing solely on angiosarcoma patients found that only 2/31 patients experienced stable disease at 9 months of treatment, resulting in termination of the study. Median PFS was also poor between 1.8 and 3.8 months
(21). Comparatively, PFS with paclitaxel alone was found to be around 4 months
(22, 23).
Interestingly, while high levels of VEGFR2 correlates with better prognosis in angiosarcoma, high levels of the ligand, VEGF, correlates with higher tumor grade, increased risk of tumor recurrence and metastasis, and decreased survival in soft- tissue sarcomas, indicating that this ligand is playing a crucial role in sarcomas (24,
25). More research is necessary to explore the role of VEGF in angiosarcoma as well as to test other anti-angiogenic therapies.
Mutations in TP53 are common in angiosarcomas. As angiosarcomas are so rare, it has been difficult to gather enough samples to obtain meaningful results. Therefore, the mutations and pathways leading to this disease have not been extensively studied.
p53 (encoded by TP53, tumor protein 53) is a protein important in cellular responses to DNA damage, leading to cell cycle arrest, apoptosis, regulation of angiogenesis, and senescence; mutations in this pathway are quite common in cancer (26-28). Alterations of this pathway are involved in angiosarcomas as well.
One study found two cancer-related mutations by sequencing in the p53 pathway out of 52 tumors; they also uncovered a lesser-studied polymorphism in 58% of these tumors. They found overexpression of p53 protein by IHC correlating with
7 worsened prognosis in 23/52 tumors (including the 2 with confirmed mutation by sequencing) (13). Elevated levels of p53 protein are often seen in tumors that have mutated p53 as they result in an increased half-life of the protein, leading to an accumulation of the non-functional protein within cells (29-31). A study by Murali et al found that TP53 mutations were the most common mutations, occurring in 35% of tumors. (32). INK4a-ARF inactivation is seen as well in angiosarcoma, a tumor- suppressor that activates p53 (33).
The mTOR and MAPK pathways are often simultaneously activated in angiosarcoma.
The mTOR and MAPK pathways are the best-studied pathways in angiosarcoma.
These pathways are simultaneously activated in many forms of cancer and have a considerable amount of cross-talk. They are responsible for controlling cell survival, differentiation, proliferation, and migration (34). One study found increased phosphorylated S6 (a downstream substrate of mTOR, Figure 1.4) by immunohistochemistry (IHC) in 100% of 59 angiosarcoma patient samples (35).
Causes of elevation of the mTOR pathway include mutations in PIK3CA (a gene that encodes PI3K) and decreased expression of PTEN (36, 37). Sequencing of 341 genes commonly mutated in cancer was performed in 34 angiosarcoma samples. Over half
(53%) of the samples in this study harbored mutations in the MAPK pathway with gene mutations occurring in KRAS, HRAS, NRAS, BRAF (rapidly accelerating fibrosarcoma protein), MAPK1, NF1 (neurofibromatosis related protein 1), and CRAF
(32). Another study found that by IHC, 42% of angiosarcomas were found to have high levels of phosphorylated S6 and 4EBP1 (EIF4E-binding protein 1) protein, two
8 downstream targets of mTOR, and 31% had phosphorylated MAPK. Interestingly, in this study eleven of the seventeen tumors (65%) positive for phospho-S6 and/or phospho-4EBP1 also had activation of phospho-MAPK (13). Therefore, a large percentage of human tumors have activation of both the mTOR and MAPK pathways, indicating that these may be good targets for inhibition for the treatment of this disease.
Some studies of targeted inhibitors have indicated the importance of these pathways in angiosarcoma. In canine angiosarcoma cell lines, use of mTOR inhibitors in combination with MEK inhibitors results in decreased cell proliferation
(38). Treatment of a xenograft model using these canine angiosarcoma cells with
MEK inhibitors results in decreased tumor growth (39). Human angiosarcoma cells treated with PI3K/mTOR inhibitors show decreased cell proliferation, an effect that was replicated in patients in a phase II clinical trial for everolimus (40-42). One angiosarcoma patient received a combination mTOR and MEK inhibitor with little effect; however, this patient had already progressed on other treatments and therefore had aggressive and wide-spread disease (8). These studies in humans have very few patients (3 for the everolimus study); more large-scale research is needed on mTOR and MEK inhibitors in angiosarcoma, studies best-achieved in a model system due to the rarity of the disease in humans.
Current models for angiosarcoma are lacking. Lack of knowledge concerning angiosarcoma biology has resulted in limited development of models for the disease.
9 There are a few cellular models that have been used to study angiosarcoma.
One of these transforms murine endothelial cells with SV40 large T antigen and expression of H-RAS. These cells form tumors when injected into nude mice; however, the tumors have not been classified by a pathologist, nor have diagnostic stains of these tumors been published to confirm that the tumors are angiosarcomas
(43). Further, these cells have not been published in mice with an intact immune system which is an important consideration in cancer biology as the immune system can aid in finding and killing abnormal tumor cells (44, 45). Cell lines have been generated from human tumors as well (46, 47). While cell lines are useful models for disease, they also have drawbacks: they tend to have greater copy number alterations than the original tumor they came from (48-50) and they are not in the same tumor microenvironment, which can affect sensitivity to drugs (51). Further, these angiosarcoma cell lines are not available for purchase, making them difficult to obtain; it is at the discretion of the groups that developed them whether they wish to distribute them. Using patient-derived cells for xenograft models have drawbacks as well: there can be changes in the prevalence of mutations of xenografts compared with the original tumor (52) and the human cells must be injected into immunocompromised mice so that the immune system does not recognize the foreign cells and kill them (53). However, only one of the two angiosarcoma cell lines can be grown in mice (46, 47).
There is one genetically engineered mouse model for angiosarcoma reported.
However, this Ink4a/Arf-/- mouse develops angiosarcomas at a low rate of 30% (54).
They also develop other types of sarcomas, making it a difficult model to work with,
10 especially for testing therapeutics (55, 56). A genetically engineered mouse model for angiosarcoma is necessary to provide a model system with the advantages of a competent immune system, tumors can be induced that recapitulate signaling aberrations found in human disease in the putative cell of origin, and develops tumors with high penetrance and short latency to facilitate the study of tumorigenesis and therapeutic interventions.
Phosphatases play an important role in angiosarcoma. Recently, the protein tyrosine phosphatase PTPRB has been implicated in the formation of angiosarcoma. Behjati et al performed whole genome sequencing on three angiosarcomas and found that two of them harbored mutations in PTPRB. More samples were then analyzed either by whole exome sequencing or targeted sequencing. These efforts uncovered 14 mutations in PTPRB in 10/39 angiosarcomas. Mutations in PTPRB have been corroborated in the study by Murali, although at different frequencies, probably due to the small sample sizes in each study (32). PTPRB is found only in vascular endothelium; in vitro inhibition of this protein results in increased angiogenesis. It is thought that these mutations in angiosarcomas are driver mutations (36). It is responsible for negatively regulating the RTKs VEGFR2 and TIE2 (tyrosine kinase with Ig and EGF (epithelial growth factor) homology domains). Loss of PTPRB results in increased activity of the MAPK pathway, leading to increased endothelial cell proliferation (57). Ptprb null mice die in utero around E10.5 due to vascular defects (58). These data indicate that phosphatases regulating RTKs in the
11 endothelium play important tumor suppressive roles in the formation of
angiosarcoma.
There is a clear need for novel models for angiosarcoma which recapitulate human
disease in order to explore the role of phosphatases in these tumors and to test new
therapeutic strategies to improve the currently dismal outcomes for these patients.
Phosphatases and PTPN12
Phosphatases are essential for regulating signaling through
dephosphorylation, antagonizing kinase activity. This dephosphorylation is capable
of negatively or positively regulating protein activity (59). There are phosphatases
that dephosphorylate specific residues: some specific for dephosphorylating
tyrosine residues, some for serine/threonine residues, and dual specificity
phosphatases that can dephosphorylate all three. Phosphatases themselves are
regulated; the non-receptor protein tyrosine phosphatases (PTPs) can be regulated
through phosphorylation, post-translational modifications, and microRNAs.
Receptor tyrosine phosphatases can also be regulated through dimerization and
ligand binding (59).
Phosphatases can be oncogenes and tumor suppressors. Phosphatases are relatively
understudied compared with their kinase counterparts. Phosphatases have been
implicated as both oncogenes and tumor suppressors in a variety of cancers
12 including esophageal, gastric, colorectal, pancreatic, renal, bladder, prostate,
endometrial, cervical, ovarian, lung, breast, glioblastoma, and sarcomas (60, 61).
Oncogenic PTPs can be disregulated through gene amplifications or
mutations, leading to enhanced growth and survival of cancer cells (62). Small
molecule inhibitors could be used to inhibit these oncogenic PTPs. SHP2 (SH2
containing phosphatase 2, encoded by the gene PTPN11) is an oncogenic PTP that
has recently been of interest as a target for inhibition. Several inhibitors have been
investigated: Cryptotanshinone, derived from the plant Salvia miltiorrhiza Bunge
(Danshen), inhibits the mutant form of SHP2 by altering access to the catalytic cleft
in an irreversible manner; HLP46, an inhibitor derived from edible mushrooms, is a
noncompetitive inhibitor of SHP2 that has off-target effects; and SHP099, a SHP2
specific allosteric inhibitor that holds SHP2 in its inactive conformation. All three of
these inhibitors are capable of reducing cell proliferation in vitro; however,
Cryptotanshinone may also target SHP1 and HLP46 can exert its effects through
other targets as well. SHP099 proved to be quite specific for SHP2 and also was
tested in vivo in a xenograft model with positive results (63-65). These data prove
that while finding compounds capable of specifically targeting one PTP is difficult,
targeting oncogenic PTPs for cancer therapy is not only possible but effective.
Tumor suppressor PTPs can undergo chromosomal losses and deactivating
mutations as well, however, these lead to loss of function (62). These modes of
inactivation can be treated using gene therapy. In this treatment, the phosphatase
can be placed into a viral vector that has been engineered to remove the portions
that cause disease. The viral vector is then administered, resulting in viral
13 replication of the phosphatase, introducing a normal, functional version of the
mutated tumor suppressor. Gene therapy has not yet been perfected and comes
with several risks, including potential to regain disease-causing properties, fatal
immune-response generation, and possible integration before an oncogene, which
can result in upregulation of the oncogene. Despite these potential pitfalls, there are
numerous clinical trials ongoing in cancer testing gene therapy as a possible
therapeutic, often inserting an active p53, although none to our knowledge for
phosphatases (clinicaltrials.gov).
Tumor suppressor PTPs can also be altered by epigenetic modifications and
microRNAs. DNA methylation at promoter regions results in suppression of gene
transcription (62). DNA hypomethylating agents are being explored to restore the
transcription of these genes; one drug, decitabine, has already been FDA approved
for use in myelodysplastic syndromes, which can progress to acute myeloid
leukemia (66). A newer possibility is silencing through microRNAs, which are
capable of targeting the mRNA of tumor suppressors for degradation. These can be
inhibited with the use of oligonucleotides, microRNA sponges, or microRNA masks
(62).
PTPN12. PTPN12, protein tyrosine phosphatase nonreceptor 12, is a 120 kDa
protein that is ubiquitously expressed in human and mouse tissues (67-69).
Cysteine 231 is known to be the residue responsible for the phosphatase catalytic
activity of PTPN12 (70). It interacts with RTKs as well as many proteins involved in
cell adhesion and motility. This protein is also known as PTP-PEST due to the PEST
14 domain found at the C-terminus (67). The PEST sequence (proline, glutamic acid, serine, and threonine-rich domain) is thought to be a degradation signal that results in a short protein half-life (71). This protein is important developmentally as well as in cancers.
PTPN12 plays roles in RTK signaling and migration. Some downstream effectors of
PTPN12 have been uncovered including EGFR, PDGFRβ (platelet derived growth factor receptor beta), HER2, SHC (Src homology 2 domain containing), GRB2
(growth factor receptor bound protein 2), SRC (a proto-oncogene), FAK (focal adhesion kinase), PYK2 (protein tyrosine kinase 2 beta), paxillin, and p130CAS (or
BCAR1, breast cancer anti-estrogen resistance 1) (72-79). Regulation of these proteins by PTPN12 typically results in negative regulation of the MAPK pathway or regulation of cell migration.
Dephosphorylation of many of these proteins by PTPN12 leads to attenuation of the MAPK pathway (Figure 1.3). PTPN12 regulates RTK signaling, both directly and indirectly. The RTKs affected by PTPN12 dephosphorylation are
EGFR, HER2, and PDGFRβ (72, 75, 80). The result of this regulation is a decrease in signaling through the MAPK pathway; when PTPN12 is decreased or mutated this results in increased activation of the MAPK pathway. PTPN12 interacts with the adaptor protein GRB2, bringing PTPN12 in to contact with EGFR. Initially it was shown that although PTPN12 was brought to EGFR it did not dephosphorylate EGFR
(72). Later studies have shown that in that in some contexts PTPN12 does indeed dephosphorylate EGFR (80). Downstream of the RTKs, PTPN12 has been shown to
15 act on other proteins that are involved in the MAPK pathway. PTPN12 has been shown to interact with SHC, leading to regulation of the RAS-MAPK pathway (70,
81). Increases in MAPK signaling occurs in approximately one third of cancers and can lead to enhanced proliferation, migration, and angiogenesis (82, 83).
FAK and PYK2 are proteins belonging to the same family that are regulated by PTPN12 (Figure 1.3). PYK2 and FAK are involved in cell adhesion and migration
(84, 85). The interaction with PYK2 and FAK can be indirect, mediated by an association of PTPN12 and paxillin (76, 86). FAK phosphorylation is increased in
PTPN12 null fibroblast cells (77). Another study demonstrated a direct interaction between PTPN12 and FAK mediated by phosphorylation of PTPN12 by MAPK, allowing PTPN12 to interact with PIN1. This interaction leads to the dephosphorylation of FAK causing increased tumor cell migration, invasion, and metastasis (87). Interestingly, PYK2 actually immunoprecipitates more strongly with PTPN12 and is more dephosphorylated by PTPN12 than FAK is in HEK293T
(human embryonic kidney) cells (88). FAK and PYK2 are regulated by PTPN12 in B and T cells as well (70, 89).
PTPN12 also regulates many proteins involved in cytoskeletal organization, leading to regulation of motility and migration including paxillin, HIC-5 (a focal adhesion protein), leupaxin, p130CAS, PSTPIP (proline-serine-threonine phosphatase interacting protein 1), filamin-A, and p50CSK (C-terminal SRC kinase)
(69, 78, 90-94). Paxillin and p130CAS are the best studied of these proteins (Figure
1.3). PTPN12 interacts with these proteins by its C-terminal domain leading to dephosphorylation and defects in cell adhesion and migration (73, 78, 79). These
16 proteins can also be regulated indirectly by PTPN12 through FAK and PYK2 (88,
95). As mentioned previously, PTPN12 mainly exerts its effects on FAK through a direct interaction with paxillin (76). p130CAS directly interacts with PTPN12 as demonstrated by immunoprecipitation using the substrate-trapping mutant of
PTPN12 (73, 96). In COS-1 fibroblast cells PTPN12 relocalized to the membrane periphery following stimulation by integrins, a characteristic of proteins involved in cell migration. Wound healing assays and chamber mobility assays demonstrated that fibroblast cells isolated from Ptpn12 null mouse embryos migrated more slowly than cells expressing Ptpn12. Vinculin staining of these cells showed that cells lacking Ptpn12 had significantly more focal adhesions, indicating that their migration defects could be due to disregulation of focal adhesion assembly/disassembly. Loss of Ptpn12 also caused the cells to attach more readily to fibronectin. P130CAS, FAK, and paxillin were all hyper-phosphorylated in cells lacking Ptpn12 (77). The interaction between paxillin and PTPN12 was shown to be essential for cell spreading and migration in MEF cells using a paxillin mutant incapable of interacting with PTPN12 (97). In A20 cells, PTPN12 associates with
SHC, paxillin, CSK, and p130CAS (70). PTPN12 has recently been shown to form a complex with an ephrin receptor, EPHA3, along with paxillin and FAK (98, 99). In
HEK293T cells, overexpression of PTP-PEST results in an increase of EPHA3 endocytosis in a tyrosine phosphorylation dependent manner. Overexpression of
PTP-PEST can also affect migration through its regulation of EPHA3 by reducing both the extension and rate of retraction of filopodia-like structures in response to
17 EphrinA5 stimulation (99). Ephrin signaling is important for adhesion as well as cytoskeletal remodeling (100).
PTPN12 has also been implicated in SRC signaling. Recently, PTPN12 was shown to regulate SKAP-Hom (SRC kinase-associated phosphoprotein homologue)
(Figure 1.3). This protein was identified as a potential substrate of PTPN12 in a yeast two-hybrid screen. Immunoprecipitations using the substrate-trapping mutant of PTPN12 confirmed that PTPN12 bound SKAP-Hom. SKAP-Hom-/- MEFs display decreased mobility by trans-well migration assays. SKAP-Hom mutants lacking the domain necessary for interaction with PTPN12 had increased mobility, indicating that SKAP-Hom dephosphorylation by PTPN12 was negatively regulating cell migration (101).
PTPN12 is also involved in cell migration and adhesion through regulation of
RAC/RHO (GTPases), although its role appears to be context dependent (Figure 1.3).
Overexpression of PTPN12 leads to inhibited cell motility in fibroblasts and CHOK1
(Chinese hamster ovarian) cells; fibroblasts from Ptpn12 null mice also display decreased motility (77, 102, 103). Catalytic activity was necessary for inhibition of motility in the CHOK1 cells as demonstrated using expression of the catalytically inactive C231S mutant. In CHOK1 cells overexpression of Ptpn12 exerts its effects on cell adhesion and motility by decreasing integrin-dependent and growth factor- dependent activation of RAC1 (103). Ptpn12 null fibroblast cells also failed to activate RHOA in response to adhesion to fibronectin. This study demonstrated a direct interaction between PTPN12 and VAV2 (a guanine nucleotide exchange
18 factor, or GEF, responsible for converting RAC from its inactive GDP bound form to the active GTP bound form) important for RAC1 activity, as well as between PTPN12 and p190RHOGAP (a GTPase activating protein, or GAP, responsible for removing the GTP from RHO to be replaced with a GDP) important for RHOA activity. PTPN12 was important for tail retraction and membrane protrusions in cell migration through regulation of these proteins (104). Using a paxillin mutant that abrogates binding to PTPN12, Jamieson et al. showed that the interaction between paxillin and
PTPN12 was important for suppressing RAC1 activity as well (97). PTPN12 also binds and dephosphorylates RHOGDI1 (RHO GDP dissociation inhibitor, holds RHO in its active form) in mouse embryonic fibroblast (MEF) cells. Silencing PTPN12 in astrocytes led to an increase in PAK (serine/threonine p21 activating kinase) phosphorylation, a sign of increased RAC1 activity (105). Therefore, in various different contexts, loss of PTPN12 leads to an increase in RHOGDI1 and RAC1 activity and a decrease in RHOA activity, leading to defects in cell motility.
19
Figure 1.3: Schematic of PTPN12 signaling. Proteins depicted in dark blue bind to PTPN12.
PTPN12 can be regulated by phosphorylation, methylation, and microRNAs. PTPN12
can be regulated in several different ways. PTPN12 is phosphorylated on both Ser39
and Ser435. However, only the phosphorylation on Ser39 is regulatory. This
phosphorylation is conducted by protein kinase A (PKA) and C (PKC), which leads to
repression of the enzymatic activity of PTPN12 (106). Recently, protein
phosphatase 1 alpha (PP1α) was shown to be responsible for dephosphorylating
PTPN12 at Ser39 as well (107). Interestingly, mutation of this residue alters binding
to SKAP-Hom but not to p130CAS, paxillin, or SHC, indicating that this regulation
may specifically regulate some functions of PTPN12 but not others (101).
PTPN12 can be regulated through methylation as well. PTPN12 has a CpG
island in its promoter region that can be methylated, leading to repression of gene
20 transcription (Cancer genome workbench). In triple negative (ER negative, PR
negative, HER2 negative) breast cancer cell lines, treatment with a DNA
methyltransferase (the enzyme responsible for methylating DNA) inhibitor resulted
in an increase in PTPN12 mRNA. Methylation-specific polymerase chain reaction
(PCR) confirmed that PTPN12 is methylated in these cell lines (108). Another study
in breast cancer confirmed that CpG promoter methylation correlated strongly with
decreased PTPN12 expression in both cell lines and patient samples (109).
PTPN12 has several predicted microRNA (miR) binding sites through which
it could be negatively regulated. MiRs are short RNA sequences that target mRNA to
prevent protein translation and are often dysregulated in cancers (110-112).
TargetScan identifies miR-124, 506, 194, 200, 429, 548, and 490-3p as having
potential target sites on PTPN12. Some of these sites have been validated. In ovarian
cancer cells, miR-194 can negatively regulate PTPN12 expression (113). miR-124
expression can lead to decreased levels of PTPN12 in breast cancer (80).
PTPN12 is important for the development of vasculature. PTPN12 has been shown to
play several critical developmental roles. Ptpn12 null mice die in utero due to
defects in mesenchyme formation, neuroepithelial development, impaired
vasculogenesis, and failure to develop liver organs between E9.5-10.5 (114). Ptpn12
has been deleted conditionally in mice as well using an endothelial specific TIE2
promoter. These mice die embryonically as well, between E10.5-11.5, indicating that
the vascular deficiencies are severe enough alone to result in embryonic lethality
(115). This protein has also been knocked down in zebrafish leading to disorganized
21 vasculature; vessels failed to perform properly. This resulted in decreased blood flow due to the impaired vasculature (116). Interestingly, there is not much information on germline mutations of PTPN12 in people. There is one report of mosaic expression of PTPN12 in which PTPN12 was absent in approximately 74% of lymphocytes. This resulted in severe vascular malformations, specifically a ventricular septal defect and an interrupted aortic arch (116). This highlights the role of PTPN12 in the vasculature across species.
Based on these studies it is clear that PTPN12 plays an important role in formation of the vasculature; therefore, its role in endothelial cell signaling has been of interest. In order to circumvent embryonic lethality due to loss of this gene, inducible mouse models have been created. Souza et al. developed a Ptpn12fl/fl mouse under an ubiquitin C (UBC) Cre-ERT2 promoter so that the gene would be deleted upon exposure to tamoxifen. Endothelial cells were isolated from these mice that had been fed tamoxifen and grown in culture. These cells did not have any
PTPN12 protein by Western blot; they did, however, express normal markers of endothelial cells such as CD31, vascular endothelial (VE)-cadherin, (intercellular adhesion molecule) ICAM-1, ICAM-2, and β-integrin, which suggests normal differentiation of endothelial cells. TIE2-Cre mice were used to conditionally delete
Ptpn12; embryos examined expressed these markers of endothelial cells as well, providing more evidence that PTPN12 is not necessary for endothelial cell differentiation. These cultures also grew at the same rate as cultures isolated from animals that had not been exposed to tamoxifen and therefore had normal amounts of PTPN12, indicating that PTPN12 loss has no effect on proliferation. Cells lacking
22 PTPN12 had a decreased ability to adhere to fibronectin or collagen, indicating a role for PTPN12 in integrin-mediated adhesion. Further, lack of PTPN12 resulted in decreased the ability of endothelial cells to migrate both by Transwell and wound healing assays, a process essential for vascular development. Several known targets of PTPN12 were examined to determine whether they were regulated in this context: CAS, paxillin, and PYK2 all showed increased phosphorylation in the cells lacking PTPN12 while there was no effect seen on FAK activity (115). Clearly
PTPN12 is important for normal endothelial cell function.
PTPN12 protein expression is often decreased in cancer. Decreased expression of
PTPN12 protein has been linked to several cancers including breast cancer, hepatocellular carcinoma, oral squamous cell carcinoma, colon cancer, esophageal cancer, non-small cell lung cancer, nasopharyngeal carcinoma, and glioblastoma.
This decrease in protein levels attributes to increased cell invasiveness, aggressiveness of disease, and activation of the MAPK pathway (61, 80, 117-122).
In triple negative breast cancer, loss of PTPN12 can substitute for amplifications of receptor tyrosine kinases. Instead of HER2 amplifications, these tumors were shown to have a decrease in the amount of PTPN12 protein, leading to activation of the MAPK pathway. PTPN12 was also shown to regulate both EGFR and platelet derived growth factor receptor beta (PDGFR-β). Loss of PTPN12 expression also strongly correlated with the triple negative breast cancer type; PTPN12 mutation or decreased protein expression and amplification of HER2 tend to be mutually exclusive. miR-124 was found to be elevated in these tumors, which could
23 be a cause of PTPN12 suppression. Loss of PTPN12 expression resulted in decreased colony formation in vitro and increased metastasis to the lung in vivo. Restoration of
PTPN12 expression was able to ameliorate these characteristics. Inhibition of multiple receptor tyrosine kinases was also an effective therapy, indicating that
PTPN12 is exerting its effects on multiple RTKs leading to cancer progression (80).
Another potential mechanism for repression of PTPN12 is through promoter methylation in triple negative breast cancer (108). Another study by Li et al. focused on examining Ptpn12 in an HER2 dependent genetically engineered mouse model for breast cancer. Mice lacking Ptpn12 developed tumors earlier; these tumors were more numerous and grew larger than in mice with Ptpn12. Mice with Ptpn12 deletion also exhibited more frequent metastases of these tumors to the lung. Cells from these tumors also displayed increased migration in vitro. Tumors from mice lacking Ptpn12 also exhibited markers of epithelial to mesenchymal transition as well as features of the more aggressive basal-type breast cancer. These tumors had increased phosphorylation of PYK2, CAS, and paxillin as a result of loss of Ptpn12; changes that result in increased cell migration. Indeed, inhibition with a PYK2 inhibitor resulted in decreased migration as a result of decreased phosphorylation of PYK2 and paxillin (123). Response to neoadjuvant chemotherapy was better in patients with high expression of PTPN12 (23.2%) compared with those with low expression of PTPN12 (5.2%), indicating that PTPN12 levels can predict outcome and should be taken into account when determining treatment course (124).
Further, lower levels of PTPN12 correlated with larger tumor size, higher tumor stage, and increased likelihood of metastasis. These patients demonstrated a
24 statistically significant poorer survival than those with high levels of PTPN12 (109,
125).
PTPN12 loss in hepatocellular carcinoma (HCC), non-small cell lung cancer
(NSCLC), esophageal squamous cell cancer (ESCC), and nasopharyngeal carcinoma
(NPC) has detrimental effects. In hepatocellular carcinoma (HCC), tumors demonstrated significantly lower expression of PTPN12 compared with normal adjacent tissue by IHC (117). By Western blot and IHC, PTPN12 protein expression was shown to be lower in NSCLC samples than in normal adjacent tissue (121). In
ESCC, PTPN12 levels were lower than adjacent normal tissue by Western blot.
However, when IHC was performed, only about 40% of samples had low levels of
PTPN12 (119). Low levels of PTPN12 correlated with higher tumor grade, tumor recurrence, and tumor metastasis. Low expression of PTPN12 resulted in significantly shortened disease-free survival and overall survival due to this increase in aggressiveness and invasiveness (117, 119, 121).
The role of PTPN12 has been examined in colon cancer cell lines. Knockdown of PTPN12 in a nonaggressive cell line that expressed endogenous PTPN12 resulted in an increase in cell migration. Conversely, overexpressing PTPN12 in an aggressive and highly metastatic colon cancer cell line that had very low levels of endogenous PTPN12 resulted in a decrease in cell migration. These results were shown to be due to the catalytic activity of PTPN12 using the phosphatase-dead mutant. Further, adherens junction assembly was impaired as a result of increased
Rac1 activity and decreased RHOA activity in these cells (118). Another study in colon cancer cell lines found several different transcript lengths of PTPN12 by PCR
25 (polymerase chain reaction); however, the consequences of these variants were not explored in this study (126). One study found rare PTPN12 mutations in patient samples as well (127). However, another study demonstrated that the loss of
PTPN12 is not consistently seen in colon cancers (128). This discrepancy may be due to small sample sizes or rarity of PTPN12 mutations. PTPN12 expression may also be lost in other manners besides mutation.
PTPN12 has also been implicated in prostate cancer. In PC-3 cells, a metastatic prostate cancer cell line, PTPN12 forms a complex with leupaxin (a member of the paxillin family), SRC, and PYK2. Overexpression of PTPN12 in these cells resulted in a decrease in cell migration, potentially through dephosphorylation of PYK2 (92).
In ovarian cancer PTPN12 mRNA is decreased compared with normal ovarian tissue. This correlates strongly with an overexpression of miR-194 in ovarian cancer samples. Overexpression of miR-194 increased the ability of ovarian cells to migrate, a similar phenotype to that expected of PTPN12 depletion. The predicted target sequence for miR-194 on PTPN12 was cloned into a luciferase reporter vector either wild type of with a mutation and cotransfected with miR-194.
Cells transfected with the wild type predicted sequence experienced a marked decrease in luciferase activity when miR-194 was overexpressed while there was no change in the luciferase activity in the cells with the mutated predicted sequence, indicating that miR-194 is negatively regulating PTPN12 (113).
Unfortunately, no studies have been performed investigating PTPN12 in angiosarcoma. Sequencing efforts by Behjati et al found mutations in another
26 phosphatase, PTPRB, but did not report mutations in PTPN12 (36). However, as described above, there are numerous other ways in which PTPN12 protein may be decreased in cancers. Mutations in the PTPN12 gene in the cancers described above were rare; however, decreased expression of PTPN12 protein was important for tumor growth and correlated strongly in various cancer types with decreased survival, indicating that PTPN12 is being down-regulated through a method besides genetic mutation. Therefore, although PTPN12 mutations have not been found in angiosarcoma, it is more likely that the protein is down-regulated in anther manner in these tumors.
The mTOR and MAPK pathways
The mTOR and MAPK pathways are responsible for controlling cell survival, proliferation, and migration; loss of control of these mechanisms through aberrant activation of these pathways can lead to cancer. Indeed, the mTOR and MAPK pathways are aberrantly activated in many forms of cancer (129, 130). This can occur through activating mutations of oncogenes or deactivating mutations of tumor suppressors. As there is much cross-talk between the two pathways, they are often activated simultaneously.
The mTOR and MAPK pathway components. The mTOR pathway is downstream of
RTKs. Phosphorylation of RTKs after ligand-binding leads to phosphorylation of
PI3K (alternatively RAS-GTP can activate PI3K), which in turn recruits AKT (or PKB, protein kinase B) to the plasma membrane and leads to the phosphorylation and
27 activation of AKT. This step is negatively regulated by PTEN, which dephosphorylates PIP3 (phosphatidylinositol 3,4,5 trisphosphate) to PIP2
(phosphatidylinositol 4,5, bisphosphate), decreasing the ability of AKT to localize to the plasma membrane and therefore decreasing its phosphorylation. AKT then phosphorylates and deactivates TSC1/2 (tuberous sclerosis complex ½), proteins involved in the negative regulation of mTORC1 (mTOR complex 1), which is a complex comprised of several proteins including (but not limited to) mTOR, Raptor, and DEPTOR. This complex can then phosphorylate 4EBP1 and S6K (S6 kinase), which then phosphorylates S6. S6 and 4EBP1 function as transcription factors; once phosphorylated transcription of proteins involved in cell survival and proliferation are transcribed. mTOR is also part of another complex, mTORC2, consisting of several proteins including (but not limited to) mTOR, DEPTOR, and Rictor, which can phosphorylate AKT, leading to activation of the pathway (Figure 1.4) (131, 132).
The MAPK pathway is also downstream of RTKs. RTKs are activated by binding of ligand, leading to recruitment of GRB2 to the RTK. GRB2 then recruits the guanine nucleotide exchange factor (GEF) SOS, which positively regulates RAS.
Activated RAS can then activate RAF, which activates MEK, which activates MAPK, which can then phosphorylate other proteins (such as RSK-ribosomal S6 kinase) or translocate to the nucleus and act as a transcription factor (133). Activated MAPK and one of its substrates, RSK, can phosphorylate and de-activate TSC2, a negative regulator of the mTOR pathway (Figure 1.4) (134, 135).
28
Figure 1.4: Schematic of mTOR and MAPK pathways. Proteins commonly inactivated by mutation in cancers and known tumor suppressors are indicated by light blue. Proteins commonly amplified or activated by mutation in cancers are indicated in dark blue.
Mutations in the mTOR and MAPK pathways are common and often result in activation of both pathways. Many cancers exhibit mutations that lead to activation of these pathways. Common mutations include activating mutations in RTKs,
PIK3CA, RAS genes, and RAF genes or inactivating mutations of PTEN, TSC1/2, and
NF1 (130, 136, 137). Oftentimes these mutations can affect both pathways. RAS (a
GTPase, a protein active when bound to GTP and inactive when bound to GDP) not only is able to activate PI3K but can also signal through RAF to activate MAPK (138).
Therefore, activating mutations of RAS can activate both the mTOR and MAPK pathways. Inactivating mutation of NF1 (a RAS-GAP, GTPase activating protein, responsible for stimulating hydrolysis of GTP to GDP) also leads to an increase of
RAS-GTP, the active form of RAS, activating both pathways (139). Further, activation
29 of MAPK can lead to phosphorylation and inactivation of TSC1/2, leading to activation of the mTOR pathway (134, 135). Activation of both pathways is common, either through mutation of a gene in a protein shared between the pathways or through activation of one pathway which leads to activation of the other through the various nodes of cross-talk.
Targeted therapeutics directed at one pathway have had little success while inhibition of both appears promising. As activation of these pathways is a common occurrence in cancer, they have been the interest of targeted therapeutics. Rapamycin and its derivatives (rapalogs) function by binding to FKBP12 (12kDa FK506 binding protein); this complex binds and inhibits mTORC1 but not mTORC2. However, some reports show that in some cell types (endothelial cells among them), prolonged exposure to rapamycin does inhibit mTORC2 by binding free mTOR and preventing assembly of mTORC2 (140-142). Rapalogs have been disappointing in the clinic, potentially due to the inability to inhibit mTORC2, however, they do have positive effects in renal cell carcinoma, lymphoma, and Kaposi sarcoma, although even these fail to meet objective response criteria (either partial or complete response) (143).
More recently, ATP-competitive inhibitors of mTOR have been described; these inhibitors are capable of inhibiting both mTORC1 and mTORC2 (144). However, as the ATP-binding pockets of kinases are relatively conserved, achieving target specificity with ATP-competitive inhibitors is difficult; indeed, these ATP- competitive mTOR inhibitors do have an inhibitory effect on other kinases (145,
146). These drugs prove efficacious in mouse models and cell lines and can even
30 overcome rapamycin resistance in some cases, thought to be due to the additional inhibition of mTORC1 and the PI3K pathway (147). As these drugs are newer, there is not as much data available on their effectiveness in people; however, one phase I clinical trial completed on AZD8055 showed disappointing results: while the drug was well-tolerated, no clinical responses were seen in the 49 patients on study
(148).
MEK inhibitors are typically non-competitive inhibitors with the respect to the ATP binding site; rather, they are allosteric inhibitors. This provides the benefit of making them more selective than inhibitors that are competitive with the ATP- binding pocket. These drugs tested in mouse models and cell lines prove effective; in the clinic they have also experienced some success (149, 150). A phase I trial of selumetinib (AZD6244) in solid tumors demonstrated that the drug is tolerable and efficacious; 49% (19/39) had stable disease, nine of these maintained stable disease for at least five months. This treatment was especially effective in tumors exhibiting mutations in genes within the MAPK pathway, either KRAS, NRAS, or BRAF (151).
Melanoma is one form of cancer that is often driven by mutations in the MAPK pathway, specifically in BRAF and NRAS (152). Trametinib was the first MEK inhibitor FDA-approved for treatment of metastatic melanoma (153). A phase I trial with trametinib found an objective response rate of 10% for all patients; however, the objective response rate for BRAF-mutant melanoma was 33% (154). A phase II study in BRAF mutant metastatic melanoma specifically was equally positive: 2% had a complete response, 23% had partial responses, and 51% had stable disease
(155). Phase III study of trametinib demonstrated longer progression free survival
31 with use of trametinib compared with chemotherapy; these data led to the FDA approval of trametinib for the treatment of BRAF mutant melanoma (156).
Inhibition of either pathway alone can result in an increase in activation of the other, leading to resistance to therapy. Treatment with mTOR inhibitors, both the rapalogs and the ATP-competitive inhibitors, can result in an increase in the
MAPK pathway in cancer patients, mouse models of cancer, and cancer cell lines, leading to resistance. Addition of MEK inhibitors results in better growth inhibition in cell lines and mouse models (157-162). This effect is seen in HUVECs as well; addition of a MEK inhibitor was able to better inhibit endothelial cell proliferation.
In a mouse xenograft model for colon cancer addition of both inhibitors in combination was able to reduce tumor angiogenesis (163). Conversely, inhibition of
MEK can result in an increase in the PI3K/mTOR pathway (164). Resistance to MEK inhibition in BRAF mutant melanoma can be attributed to an increase in the
PI3K/mTOR pathway (165). Addition of AKT or mTOR inhibition was able to overcome this resistance in melanoma cell lines (166). These data indicate that inhibition of both pathways is more efficacious than inhibition of either pathway alone.
It should be noted that combinations of MEK and mTOR inhibitors have been studied in early phase clinical trials (167, 168). Both trials were designed to achieve doses as close to the recommended single agent dose as possible. One trial was a phase I dose-finding study of the combination of trametinib and the rapalog everolimus and was deemed to have failed due to the inability of patients to tolerate pre-defined optimal doses (168). However, no pharmacodynamic analyses of tumor
32 tissue were conducted to look for target inhibition. Importantly, no dose-limiting toxicities were reported at the starting dose level which corresponded to 25% of the single agent dose for trametinib and 50% for everolimus; however, some adverse effects were still seen (168). In the second trial, which was a phase II trial of the
MEK inhibitor, selumetinib given at 67% of the single agent dose and the rapalog, temsirolimus given at 100% (subsequently reduced to 80%), less than 10% of patients who received the combined treatment withdrew from the study due to toxicity (167). Again no pharmacodynamic studies in tumor tissue were conducted in this trial. As several studies have suggested that inhibition of both pathways is highly effective and can be tolerated in mice, further work should be performed to find a viable combination of drugs that effectively inhibits the targets with minimal adverse effects. More research is needed to find a tolerable, effective means of inhibiting these two pathways simultaneously in cancer patients to achieve better outcomes.
Rationale: Initially we set out to examine the role of PTPN12 in glioblastoma, as recurrent rearrangements were recently described (61). Loss of PTPN12 protein expression in breast cancer samples occurs exclusively in samples that do not display amplification of RTKs, leading to activation of the MAPK pathway, effectively substituting for RTK amplification (80). RTK amplification occurs in around 59% of glioblastoma; we therefore began with the theory that loss of PTPN12 expression would substitute for RTK amplification in the formation of glioblastoma (169). In order to test this, we utilized a previously described mouse model for glioblastoma
33 in which Trp53 and Pten are deleted using a GFAP-CreER driver. We chose this model because the most common secondary mutations that occur in the brain tumors in these mice are RTK amplifications, namely Met, Pdgfra, and Egfr (170).
Therefore, we proposed that additional deletion of Ptpn12 in this model would substitute for RTK amplifications and these secondary mutations would not be present. However, we find that these mice develop angiosarcoma prior to brain tumor formation. We therefore hypothesize that deletion of these three genes in combination results in a relevant murine model for angiosarcoma in which the mTOR and MAPK pathways are important drivers of disease.
The formation of angiosarcoma was an unexpected but exciting finding as currently there are no described genetically engineered mouse models for angiosarcoma that can be used to test novel therapeutics. This finding can be explained by several factors. The GFAP-CreER driver expresses cre in a subset of endothelial cells (our data and (171)), the putative cell of origin for angiosarcoma.
The genes/proteins we manipulated are also implicated in angiosarcoma formation;
TP53 mutations and decreased expression of PTEN protein can be found in angiosarcoma patient samples (13, 32, 33, 36, 37). The involvement of PTPN12 in angiosarcoma is a novel finding; however, mutations in another PTP, PTPRB, have been described in angiosarcoma, often simultaneously with TP53 mutations (32,
36). Interestingly, deletion of Ptpn12 with either Trp53 or Pten is sufficient to form angiosarcoma. Although PTPN12 mutations have not been described in angiosarcoma, evidence from other cancers demonstrates that loss of PTPN12 protein expression occurs more often than genetic mutations (80, 108, 117, 119,
34 121). Both PTPN12 and PTPRB are responsible for the negative regulation of RTKs, leading to attenuation of the MAPK pathway (57, 80).
Importantly, activation of the mTOR and MAPK pathways is seen in angiosarcoma patient samples, often concurrently (13, 32, 35-37). These pathways are known to drive tumor growth (129, 130). PI3K/mTOR inhibitors are effective in angiosarcoma cells; trials of mTOR inhibitors demonstrated efficacy in angiosarcoma patients as well, although the number of patients involved is small
(three with everolimus, one with ridaforolimus) (40-42, 172). Inhibition of MEK in canine angiosarcomas results in tumor regression; MEK inhibition combined with mTOR inhibition results in cell death in canine angiosarcoma cell lines (38, 39).
Inhibition of one pathway often results in recurrence which can be attributed to activation of the other pathway; inhibition of both pathways is more effective (157,
159-166). Our murine model provides an opportunity to determine the importance of these two pathways in angiosarcoma as all genotypes that develop this tumor display activation of both pathways. We have chosen to pursue the triple knockout model as almost 100% of mice develop these tumors in a short, consistent time frame and are therefore a convenient model in which to test therapeutics. We propose that inhibition of both pathways in our murine model will provide a novel therapeutic strategy for angiosarcoma, a disease with poor outcomes badly in need of new treatment options.
35
36 Chapter 2: Development and Characterization of a Novel Genetically
Engineered Mouse Model for Angiosarcoma
Excerpts in preparation for submission
Combined mTOR and MEK Inhibition is an Effective Therapy in a Novel Mouse Model for Angiosarcoma Michelle L Chadwick1,2; Adam Lane2; Dana Thomas2; Amanda R Smith2; Angela R White2; Yuxin Feng2; Elisa Boscolo2; Yi Zheng1,2; Denise Adams3; Anita Gupta4; Andre Veillette5; Lionel ML Chow1,2
1Department of Cancer and Cell Biology, University of Cincinnati, Cincinnati, OH 2Cancer and Blood Diseases Institute, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH 3Vascular Anomalies Center, Boston Children’s Hospital, Boston, MA 4Department of Pathology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH 5Institut de Recherches Cliniques de Montréal, Montréal, Canada
37 Introduction
Protein tyrosine phosphatases are often altered in different forms of cancer, either through mutations, hypermethylation, or miRNA depletion. However, compared to their kinase counterparts they are relatively under-studied (60).
PTPN12 is a PTP that has been shown to dephosphorylate RTKs as well as proteins involved in cell migration, specifically HER2, EGFR, PDGFR-β, SHC, p130Cas, and paxillin (73, 75, 76, 80). Developmentally, Ptpn12 expression is critical as knockout mice die around E10.5 due in part to impaired embryonic vascularization (114).
PTPN12 has been shown to regulate p130CAS, paxillin, and PYK2 in endothelial cells in mice, and loss-of-function leads to defects in adhesion, spreading, and migration
(115). Several studies have proposed a tumor suppressive role for PTPN12 in various cancers including breast, prostate, colon, kidney, melanoma, and esophageal carcinoma, often suppressing migration and invasion of tumor cells (92, 118, 119,
173, 174). In breast cancer, PTPN12 loss resulted in increased phosphorylation of p130CAS, PYK2, paxillin, and p70S6 kinase, and while no effect on proliferation was noted, a block in apoptosis and increased migration and invasion were described
(123). Furthermore, in triple negative breast cancer, loss of PTPN12 has been shown to substitute for amplification of RTKs, and leads to downstream activation of the
MAPK pathway. PTPN12 null breast cancer cells were also found to have enhanced metastatic potential (80). Similarly, decreased expression of PTPN12 in colon cancer cell lines resulted in increased cellular migration (118). Collectively, these data suggest that PTPN12 plays an important role in endothelial cells/vascularization as well as disease progression in a variety of cancers.
38 Angiosarcoma is a rare malignancy of the vascular endothelium accounting for approximately 2% of all soft tissue sarcomas. The approach to treatment is multimodal and includes complete surgical resection where possible followed by chemotherapy and radiation therapy in cases of metastatic disease (1, 2). Despite aggressive therapy, 5-year overall survival varies from 31% to 43% and is significantly worse with metastatic disease (3-7). Due to the rarity of the disease and a lack of available model systems, relatively little is known about the molecular pathogenesis of angiosarcoma. In order to develop novel targeted therapeutic regimens for this disease, a greater understanding of the signaling pathways driving tumor growth and models with which to interrogate these pathways are needed.
Although only a few reports have described the driver mutations leading to angiosarcoma development, these reveal the presence of recurrent pathway aberrations. Specifically, mutations leading to activation of the PI3K/mTOR and
MAPK pathways have been reported. One study found activating mutations in
PIK3CA in 40% of tumors (13) while other investigators describe elevated levels of phosphorylated S6 in 100% (175) and phosphorylated 4EBP1 in 88% of angiosarcomas (176), both of which are downstream effectors of mTOR.
Importantly, one study demonstrated that treatment with the mTOR inhibitor rapamycin decreases cell proliferation in vitro and delayed tumor growth in vivo using a xenograft model (175). Interestingly, a trial of the mTOR inhibitor everolimus in patients with recurrent soft tissue sarcomas reported that the progression free rate was highest in angiosarcoma patients compared to patients with other high-grade sarcomas (41). Seki et al. observed a partial response of an
39 angiosarcoma patient treated with ridaforolimus, another mTOR inhibitor (172).
Mutations in KRAS have been found in 13-60% of angiosarcomas as well, capable of leading to activation of both the PI3K/mTOR and MAPK pathways (36, 177-179). A recent deep sequencing study investigated mutations in the RAS/MAPK pathway comprehensively and found that 53% of angiosarcomas contained hotspot mutations in KRAS, HRAS, NRAS, BRAF, or MAPK1 (32). To our knowledge, there are no clinical trials of BRAF or MEK inhibitors in patients with angiosarcoma, although one case report described a child with angiosarcoma containing a KRAS mutation who did not respond to the MEK inhibitor trametinib (8). TP53 mutations have been noted in various studies (32, 33, 36, 180-182) with the frequency of mutations ranging from 35% to 52% of samples (32, 181). Moreover, evidence for the involvement of p53 comes from Li-Fraumeni patients who are at increased risk for angiosarcoma (183). Recently, a genome sequencing effort discovered recurrent mutations in PTPRB, a vascular endothelial cell-specific PTP that regulates RTKs leading to negative regulation of the MAPK pathway, in angiosarcoma (36, 57). TP53 mutations often co-occur with these PTPRB mutations (36). While little is known about the role that RTKs themselves play in the pathogenesis of angiosarcoma,
VEGFR2 expression correlates with better prognosis, leading to the hypothesis that
VEGFR2 is not involved in tumorigenesis but rather serves as a marker for a more differentiated tumor (18, 182).
The goal of these studies is to assess the role of PTPN12 in cancer with the long-term goal of finding novel therapeutics targeted at PTPN12 substrates. We have added deletion of Ptpn12 to a previously established mouse model to
40 determine its effects so that Pten, Trp53, and Ptpn12 are deleted using a GFAP-CreER driver (170). We hypothesize that the mice with addition of Ptpn12 deletion will develop more aggressive tumors at an earlier time frame due to its role in cell migration and invasion and regulation of the MAPK pathway. Using the GFAP-CreER driver these genes are deleted in a subset of vascular endothelial cells (171).
Deletion of these three genes in concert in these endothelial cells results in angiosarcoma with complete penetrance and short latency; double knockouts
Pten;Ptpn12 and Trp53;Ptpn12 also form angiosarcomas, although with incomplete penetrance and longer latency. PTEN expression is decreased and TP53 mutations are found in human angiosarcoma; we propose that Ptpn12 deletion substitutes for the PTPRB mutations that are seen in humans as we predict that they play similar roles. We find that loss of Ptpn12 is important for the development of this phenotype as Trp53;Pten double knockout mice do not develop this particular disease while all genetic combinations in which Ptpn12 is deleted do develop angiosarcomas. We obtained patient samples as well in order to compare our model to human disease and find that our model is similar to that of humans, both histologically and in signaling pathways elevated. This is the first genetically engineered mouse model for angiosarcoma that can be exploited to develop novel therapeutic strategies for a currently lethal and under-studied disease.
41 Results
Loss of Pten, Trp53, and Ptpn12 results in angiosarcoma. We engineered mice that inducibly delete Pten, Trp53, and Ptpn12 using a GFAP-CreER mouse driver line.
Tamoxifen is administered for three consecutive days between P28 and P44 a time frame chosen to circumvent developmental defects (Figure 2.1A). We analyzed cohorts of mice with all combinations of single, double and triple gene deletions and found that mice in three of the cohorts (Trp53;Ptpn12, Pten;Ptpn12, and
Trp53;Pten;Ptpn12 knockout ((KO)) developed multiple dark cutaneous lesions
(Figure 2.1B), which were easily detectable on the FVB albino background (Figure
2.1C). Double KO mice for Pten and Trp53 develop other subcutaneous tumors in addition to high grade glioma, but none develop cutaneous vascular lesions (data not shown and (170)). Triple KO mice develop tumors with higher penetrance and shorter latency than either of the two double KO genotypes with nearly 100% of mice developing angiosarcomas (Figure 2.1B). These tumors typically occur subcutaneously, typically on the extremities, perianally, and on the face, common locations for human tumors as well. Visceral tumors were found in the mediastinum although this occurred in a small percentage of mice (Figure 2.1C). Histologic evaluation of the lesions revealed aggressive vascular lesions with anastomosing vascular channels, nuclear heterotypia and prominent mitotic figures (Figure 2.1D).
The tumor cells stained strongly for CD31 (a marker of vascular endothelium) and were negative for PROX1 (which marks lymphatic cells), identifying them as angiosarcoma. Indeed the resemblance of the mouse lesion histology to human angiosarcoma was striking (Figure 2.1D). PCR amplification of tumor DNA
42 confirmed Cre-mediated recombination of Pten, Trp53, and Ptpn12 (Supplemental
Figure 2.1A).
In order to ensure that the GFAP-CreER mouse line is driving tumors of endothelial origin we performed co-immunofluorescence (IF) using antibodies directed against β-galactosidase (which was included as a marker gene on the bicistronic CreER transgene) and CD31. We find that a small percentage of endothelial cells also express β-galactosidase, indicating that CreER is expressed in these cells (Supplemental Figure 2.1B). Moreover, we crossed the GFAP-CreER mouse line with ROSA-Tomato reporter mice and demonstrated Cre-mediated recombination in some CD31-positive cells of the skin (Supplemental Figure 2.1C).
Lastly we used an endothelial cell-specific Cre driver, TIE2-CreER and found that the triple KO combination also developed angiosarcomas, albeit with a broader tissue distribution when compared to the GFAP-CreER mice (Supplemental Figure 2.1D,E).
Tumors in the GFAP-CreER mice are generally cutaneous and occur most frequently on the head, extremities, and perianally. When tumors from all three genetic combinations were compared, all were found to have a similar immunophenotype with respect to CD31 and PROX1 expression and all demonstrated a high proliferative index as indicated by KI67 staining (Figure 2.1D) The triple KO and
Trp53; Ptpn12 double KO tumors were classified as high-grade angiosarcoma due to the prominent nuclear pleomorphism observed while tumors from the Pten; Ptpn12 double KO mice were classified as low-grade.
43 Activation of the PI3K/mTOR and MAPK pathways in murine angiosarcoma. In order to investigate pathways involved in the pathogenesis of murine angiosarcoma we first characterized the expression of known endothelial cell markers in the tumors.
CD31 and VEGFR1 are expressed in angiosarcomas at levels comparable to that of
CD31+ endothelial cells isolated from mouse lung, while VEGFR2, VEGFR3, and VE-
Cadherin were relatively overexpressed when compared to the control lung endothelial cells (Figure 2.2A). These results underscore the vascular nature of these tumors. We then focused on the analysis of the PI3K/mTOR and MAPK pathways as these are known to be regulated respectively by two of the targeted genes, Pten and Ptpn12. Western blot analysis revealed that phospho-4EBP1 is elevated in all tumors including those that were Pten wild-type, suggesting that activation of the mTOR pathway and 4EBP1 in particular is important for angiosarcoma growth (Figure 2.2B). Phospho-MEK and phospho-MAPK are also elevated across all genotypes as expected since loss of Ptpn12 was common to all tumors. In agreement, we also demonstrated by immunohistochemistry high levels of phospho-S6 and phospho-MAPK in tumors of all genotypes (Figure 2.2C). As expected, PTEN was absent in the Pten;Ptpn12 double KO and triple KO tumors; however, surprisingly, very low levels were also detected in the Trp53;Ptpn12 KO tumors (Figure 2.2A and 2.2C).
We were particularly interested to identify the target(s) of PTPN12 activity critical for angiosarcoma formation. We first took a candidate approach to test whether RTKs involved in endothelial cell signaling and PTPN12 effectors are phosphorylated in angiosarcomas. Tyrosine phosphorylated proteins were
44 immunoprecipitated from triple KO angiosarcoma tumor lysates and blotted for the
VEGF receptors (Supplemental Figure 2.2A-G). Receptor phosphorylation was either not detected in anti-phosphotyrosine immunoprecipitates (VEGFR1), detected at similar levels compared to control lung tissue (VEGFR2) or tyrosine phosphorylation was not detected in the reciprocal receptor immunoprecipitation experiment (VEGFR3). Therefore we decided to use an RTK antibody array to screen for hyperphosphorylated receptor tyrosine kinases in murine angiosarcoma.
We found that PDGFR-β was phosphorylated in all three angiosarcoma models and was particularly prominent in Ptpn12; Pten double KO tumors in which this was the only RTK elevated in the array (Supplemental Figure 2.2H, Supplemental Table 2.1).
This finding was verified by anti-phosphotyrosine immunoprecipitation of lysates from triple KO tumors followed by western blotting for PDGFR-β as well as by the reciprocal immunoprecipitation experiment (Figure 2.2D). Furthermore, the presence of high levels of PDGFR-β expression was confirmed by IHC in all three models (Figure 2.2E). These findings are interesting as PDGFR-β has previously been shown to be a target of PTPN12 (75, 80).
Activation of the PI3K/MTOR and MAPK pathways in human tumors of vascular origin. Five snap frozen aliquots of angiosarcomas were obtained from the biobank at Cincinnati Children’s Hospital Medical Center. Signaling in these tumors was compared to tumors from all three mouse genotypes as well as to HUVECs with and without stimulation by VEGFA (Figure 2.5A). VEGFR2 was expressed in the human angiosarcoma samples in three of the tumors while VEGFR1 and PDGFR-β were
45 detected in four each. Both the mTOR and MAPK pathways were activated as demonstrated by phospho-4EBP1 and phospho-MAPK respectively (Figure 2.5A).
Anti-phosphotyrosine western blotting demonstrated an overlapping pattern of tyrosine phosphorylated proteins in the human angiosarcoma samples compared to tumors from our mouse models (Supplemental Figure 2.7).
A panel of human tumors of vascular origin were also analyzed by IHC
(Figure 2.5B). The endothelial origin of all of the tumors was confirmed by showing immunoreactivity for CD31. Interestingly, PTEN expression was largely absent in high-grade tumors (angiosarcoma and lymphangiosarcoma) while phosphorylation of S6 and MAPK was present in all tumor types. Further, expression of the PTPN12 protein was also very low in this panel of tumors when compared to normal blood vessels. PDGFR-β was not detectable by IHC in the angiosarcoma sample but was found in other tumors of vascular origin. Collectively, these results suggest that at least a subset of human angiosarcomas could benefit from treatment that we find to be effective against our murine angiosarcoma model. Furthermore, other tumors of endothelial cell origin have similar pathway activation and may also respond to this treatment regimen.
46 Discussion
We developed a novel and robust genetically engineered mouse model for angiosarcoma through the combined deletion of Pten, Trp53, and Ptpn12 using a
GFAP-CreER driver. We confirmed a vascular origin for these tumors in several ways. As our construct contains a β-galactosidase reporter gene, the presence of cre in a small subset of endothelial cells was confirmed through staining with a β- galactosidase antibody. We also bred these mice with a ROSA-Tomato reporter mouse in which cells with active cre fluoresce red; indeed we see a small population of endothelial cells that fluoresce, indicating that our genes are being deleted in these cells. This is confirmed through PCR for the recombined genes; we find that all three genes are recombined in the tumors. We further recapitulated the formation of angiosarcoma using a more specific vascular endothelial driver, TIE2-CreER.
While use of the GFAP-CreER driver targeted a small subset of endothelial cells, allowing growth of a few angiosarcoma tumors per mouse, the TIE2-CreER model deleted our genes of interest in a much larger target population, leading to the formation of numerous small internal tumors which caused the mice to succumb to disease before the tumors grew large. The mice with the GFAP-CreER driver model is therefore more useful for study as the tumors are external and grow to larger sizes, both of which make the tumors easier to measure and harvest, important for studies of disease pathology and testing of therapeutics. Interestingly, although this driver is primarily used to target genes in the brain, these mice do not develop brain tumors that interfere with the current study. Some mice do develop brain tumors but they are typically quite small (microscopic, not visible to the eye upon
47 dissection) and do not evince any symptoms expected of mice with brain tumors.
The Trp53;Pten double KO mice do typically die of brain tumors around 210 days of age (170); however, the triple KO mice die earlier than this and therefore have not yet developed large brain tumors. PTPN12 appears to be a much more potent tumor suppressor for angiosarcoma than for glioblastoma. Therefore, based on these data, the GFAP-CreER model deletes our genes of interest in a small subset of endothelial cells and results in angiosarcoma tumors that are amenable to study due to their size and location.
Conditional deletion of all three genes results in aggressive cutaneous tumors in mice that develop rapidly and with virtually 100% penetrance. Deletion of Ptpn12 in combination with either Trp53 or Pten resulted in angiosarcoma as well, although with incomplete penetrance and increased latency, while mice with intact Ptpn12 do not form angiosarcomas at all. This highlights the importance of loss of Ptpn12 in the formation of angiosarcoma in these mice. However, this also precludes us from completing our goal of studying PTPN12 in cancer; we have no tumors to compare with as the mice with normal PTPN12 expression do not develop angiosarcoma.
Therefore, other avenues of interrogating the role of PTPN12 in this disease must be explored; development of a cell line from these murine tumors in which we could re-express PTPN12 in order to assess its effects would provide us with the means for determining this more effectively. Interestingly, mice in which only Ptpn12 was deleted appeared normal. This may indicate that loss of Ptpn12 is involved in either angiosarcoma progression or disease initiation but that perhaps a second hit is required in order for angiosarcomas to form. These tumors stain strongly for the
48 endothelial marker CD31 but negatively for the lymphatic marker PROX1 by IHC, indicating that these tumors are of vascular endothelial origin and confirming the angiosarcoma diagnosis. The tumors also mimic other characteristics of angiosarcoma including anastomosing channels (irregular, connecting vascular channels), high levels of nuclear pleomorphism, and high proliferative index. These murine tumors have been reviewed by a pathologist with confirmation of high- grade angiosarcoma in the triple and Trp53;Ptpn12 KO mice while the Pten;Ptpn12
KO mice were diagnosed as low-grade angiosarcoma. Loss of p53 may be important for tumors to progress to high-grade by evading the damage-sensing pathway and the cell death it triggers. Overall, the triple KO model provides a viable model for study of disease pathogenesis and therapeutic studies due to its similarity to human disease, the high penetrance, short latency, and development of large, measurable tumors.
TP53 and PTEN aberrations have been previously associated with angiosarcoma while the involvement of PTPN12 is a novel finding of this study. TP53 mutations have been reported in human tumors (32, 36). PTEN mutations are common in vascular abnormalities and decreased protein expression is seen in angiosarcoma specifically (37). There are no reports of PTPN12 mutations in angiosarcoma to date. However, there are reports of mutations in another PTP,
PTPRB, in 26% of angiosarcomas (36). This phosphatase plays a similar role to
PTPN12, dephosphorylating RTKs leading to negative regulation of the MAPK pathway (57). Therefore, we propose that in our model the loss of Ptpn12 is providing the same effects as loss of PTPRB in human tumors. Further, this indicates
49 that the deregulation of RTKs leading to the activation of the MAPK pathway is important for the formation of angiosarcoma in both humans and mice. The lack of
PTPN12 mutations seen in human angiosarcoma may be because the gene is not mutated but rather the protein expression is decreased, either through microRNA degradation or epigenetic regulation, as is the case in many other forms of cancer
(80, 109, 117, 119, 121, 125). TP53 mutations and PTPRB mutations can occur in the same tumor; 2/6 tumors with PTPRB mutations also had TP53 mutations (36). Also, the mTOR and MAPK pathways are often elevated in the same tumors; 11/17 tumors with active MAPK also had mTOR activation; we see activation of both the mTOR and MAPK pathways simultaneously in the human samples we have obtained as well as in our mouse model, indicating that our model accurately represents human disease (13). Further, we see decreased expression of PTEN and PTPN12 in the same tumor in the human samples we obtained, indicating that not only is
PTPN12 expression decreased in human angiosarcoma but that down-regulation of both proteins can occur together. As these aberrations occur together in human disease, it is appropriate that our mouse model delete multiple genes to achieve the same effects. More research is needed to determine how PTPN12 is downregulated in human angiosarcoma. If expression of PTPN12 is due to miRs or methylation it may be possible to inhibit these functions to restore PTPN12 expression.
As stated previously, loss of Ptpn12 is important for the formation of the murine angiosarcoma as only the genetic combinations that involve loss of Ptpn12 develop these tumors. These results suggest that the PTPN12 protein is a tumor suppressor for endothelial cells and that it regulates pathways that govern
50 malignant progression to angiosarcoma. We analyzed the activation status of known
PTPN12 substrates and RTKs expressed in vascular endothelial cells and several were found to be tyrosine phosphorylated in angiosarcoma. However, PDGFR-β was the only substrate whose phosphorylation was reliably elevated when compared to control tissue and it was the only RTK that was phosphorylated in angiosarcomas from all three genetic crosses. While PDGFR-β is predominantly a marker of pericytes (184), there are several reports suggesting that its expression can be induced in endothelial progenitor cells particularly in the context of proliferation in response to vascular injury (185-187). We further find expression of PDGFR-β in our human samples; however, phosphorylation levels of this receptor were unable to be evaluated, potentially due to the quality of the samples or the sensitivity of the assays used. Additional studies will focus on elucidating other critical substrates of
PTPN12 in angiosarcoma which could lead to the identification of novel targets for therapy.
The contribution of RTK signaling to the pathogenesis of angiosarcoma was suggested with the finding of amplifications and mutations in the genes encoding the VEGFR1, VEGFR2, and VEGFR3 (18, 188, 189). The description of PTPRB mutations in angiosarcoma not only reconfirms the importance of RTK signaling in this tumor, but also raises the possibility that a broader range of RTKs expressed (or mis-expressed) in endothelial cells may be de-repressed and contribute to oncogenesis. In our murine tumors we confirm presence of VEGFR2 and VEGFR3 protein; however, immunoprecipitations show that these proteins are not phosphorylated as strongly in the murine angiosarcomas as in controls. We can find
51 no reports assessing phosphorylation levels of these VEGFRs in human tumors; however, overexpression of VEGFR2 actually correlates with better prognosis (18).
Further, use of VEGF inhibitors in patients have shown little efficacy; trials with bevacizumab show little benefit compared with paclitaxel alone (19, 23). Therefore, our data showing low levels of phosphorylation of VEGFR2 supports the data thus far in humans; perhaps rather than participating in tumor growth through VEGFR2 phosphorylation and signaling, VEGFR2 expression in these tumors acts as a marker for a more-differentiated, less aggressive tumor (18). We confirmed that PDGFR-β is phosphorylated in these murine angiosarcomas, however, activation of other RTKs was also suggested by our screening RTK array experiment, especially in the triple
KO tumors (Supplemental Figure 2.2H). These findings have important implications in the treatment of patients with angiosarcoma. Contemporary clinical trials have focused on targeting RTKs involved in angiogenesis with monoclonal antibodies such as bevacizumab (19) or targeted kinase inhibitors such as sorafenib and pazopanib (20, 21, 190). Overall, the clinical benefit of these treatments has been short-lived with progression free survival (PFS) ranging from 1.8 to 3.8 months in these studies. This is comparable to the PFS of 4 months reported in a phase II trial of paclitaxel alone (23). Furthermore, in a randomized phase II trial of paclitaxel compared to paclitaxel and bevacizumab, no improvement in PFS was seen with the addition of the antiangiogenic agent to chemotherapy (22). The lack of overall response in these clinical trials may therefore be due to redundancy of RTKs expressed and activated in angiosarcoma and/or direct mutational activation of downstream effectors that are not dependent on RTK activity. Our work suggests
52 that a more robust and durable response to therapy may be obtained by targeting two primary downstream effector pathways of RTKs, specifically the PI3K/mTOR and MAPK pathways. We investigated the activation status of the PI3K/mTOR and
MAPK pathways in murine angiosarcoma and found both to be active in all three genetic combinations. Moreover, in concordance with other studies, these pathways were also found to be activated in human angiosarcoma as well as other vascular tumors. Targeting of these two pathways may provide a viable therapeutic option for angiosarcoma patients and may improve their currently dismal outcomes.
53 Methods
Mice: Trp53 floxed mice were obtained from Mouse Models of Human Cancer
Consortium Repository (191), Pten floxed mice from Dr. Mak at the University of
Toronto (192), Ptpn12 floxed mice from Dr. Veillette at the Institut de Recherches
Cliniques de Montréal (89), and TIE2-CreER mice from Dr. Zheng at Cincinnati
Children’s Hospital (unpublished). Mice of all genetic combinations of the genes of interest were developed, 7 lines in total. A stock solution of Tamoxifen (Sigma) was made at 20mg/ml in corn oil and filtered. Mice were injected with Tamoxifen at
9mg/40g body weight between 28 and 42 days of age to circumvent developmental defects for 3 consecutive days. Mice were observed until moribund or reaching 18 months of age. Upon dissection, mice were perfused with 12 ml PBS. Tissues were both frozen and fixed in 4% paraformaldehyde (PFA). Frozen tissues were used for
RNA, DNA, and protein analyses while the PFA fixed tissues were processed by pathology at CCHMC and used for IHC.
Western Blot: 10 ug protein were loaded on to Invitrogen NuPAGETM 4-12% Bis-Tris
Midi Gels (WG1402BX10). Gels were run at 150 V for 1.5 hours. Running buffer was
NuPAGETM MES SDS Running Buffer (NP0002). Gels were transferred to PVDF membrane at 100 V for 1.5 hours. Following transfer membranes were blocked in
5% milk for 1 hour at room temp. Antibodies used include pS6 (#2211), S6 (#2217), pAKT S473 (#9271), AKT (#4685), p4EBP1 (#9459), 4EBP1 (#9452), pMAPK
(#9102), MAPK (#9101), PTEN (#9559), FAK (#3285), VEGFR2 (#2479), Shc
(#2432), PDGFRβ (#3169). Antibodies from Cell Signaling are all used at 1:1000.
54 Other antibodies include actin (Sigma, #A5441, 1:10000), CD31 (Abcam, #ab28364,
1:1000), 4G10 (Millipore, #05-321, 1:1000), PTPN12 (Abcam, ab76492, 1:1000), pFAK (Sigma, f7926, 1:1000), VEGFR1 (Abcam, ab32152, 1:1000), VEGFR3 (Thermo,
PA5-16871, 1:1000), paxillin (BD, #610619, 1:1000). Primary antibodies were diluted in 5% BSA. Secondary antibody (rabbit: NA934), mouse: NA931 from GE) was incubated at 1:5000 in 5% milk for 1 hour at room. Blots were imaged on the
UVP BioSpectrum® Imaging System (M-26XV) and quantification was performed using their software.
Endothelial cell isolation: 200ul/sample of Sheep anti-Rat magnetic Dynabeads
(Thermofisher, #11035) were washed 3x with PBS+0.1% BSA. 5ul of CD31 (BD
Pharmingen, #553370) per 50ul beads were added in 200ul PBS + 0.1% BSA per sample and incubated at 4C overnight. Supernatant was removed and samples added. Samples were obtained from lungs of adult FVB mice. The lungs were removed, minced, and incubated in DMEM media (Gibco, #12634-010) with collagenase 1 (Gibco, #17100-017) at 2mg/ml for 1 hour at 37C. After this hour, the tissue was passed through a 70 micron filter (greiner bio-one, #542070) and spun down. The cell pellet was resuspended in 2ml PBS+0.1% BSA and added to 40ul of the CD31 conjugated Dynabeads. These were rotated for 15 minutes at room temperature and then washed 3x in PBS+0.1% BSA followed by protein isolation.
Statistics: Statistics on quantifications of Western blots are two-tailed t tests performed in GraphPad. * P≤0.05, ** P≤0.01, *** P≤0.001
55
Immunohistochemistry: Samples for immunohistochemistry (IHC) were fixed in 4%
PFA overnight and processed by the Pathology Core at CCHMC. After embedding,
5um thick sections were made. Slides were placed in xylene washes followed by rehydration in a series of ethanol washes. Antigen retrieval was then performed using citrate buffer. A peroxide quench was then performed using 0.6% H2O2.
Tissues were blocked for 1 hour at room temp in serum. In addition to the antibodies used above for Western blotting, KI67 (Leica, NCL-Ki67p, 1:4000) and
Cleaved Caspase-3 (BD, 559565, 1:500) were used. Phospho-MAPK 1:400, PTEN
1:1000, pS6 1:500, PDGFRβ 1:200.
Immunoprecipitation: Protocol can be found on novex by life technologies’TM website. 4G10 antibody (used at 0.06 ug/ul beads) was conjugated to Dynabeads®
Protein A (Thermo Fisher Scientific). Tyrosine-phosphorylated proteins were pulled down from 500ug of protein from CD31+ lung endothelial cells, whole lung, and tumor lysate using the pY-1000 antibody from Cell Signaling (#8954). These samples were run on Western as described above. Blots were then probed using
4G10 phosphotyrosine antibody and total antibody for the VEGF receptors and other known effectors of PTPN12 as described above.
RTK Array: RTK array was purchased from R&D systems (#ARY014). Instructions included were followed. 500 ug of protein were used on each blot.
56 Study Approval: Use of human material for this study was approved by the institutional review board at CCHMC.
Author Contributions
MC is the primary author and contributed experimentally, with data interpretation, and writing the manuscript along with LC. DT, AS, AW provided help with experiments and mouse work. AG provided pathological interpretations. YZ and AV provided the TIE2-CreER mice and Ptpn12fl/fl mice respectively. DA and EB provided valuable intellectual input and suggestions.
57 Figures
Figure 2.1
t for 3 h A eig B t w ctionsa y s ay fen injeve d xi o ccumb am ns of disease T g e su consecuti ic 9mg/kg per 40g bodirst si M F to disease
0 28 42 100 200 Time (Days)
C b c d D
Human Triple KO Trp53;Ptpn12 KO Pten;Ptpn12 KO E
a & H
e E & H
h f 1
g 3 D C
a b c d 1 X O R P
e f g h 7 6 I K
Figure 2.1: Ptpn12 deletion leads to angiosarcoma in mice
(A) Schematic for development of triple knockout angiosarcoma mouse model. (B)
Kaplan Meier curve demonstrating incidence of angiosarcoma in the seven mouse
58 lines generated. Ptpn12 KO n=42, Trp53 KO n=9, Pten KO n=8, Pten;Ptpn12 KO n=10,
Trp53;Ptpn12 KO n=13, Trp53;Pten KO n=10, triple KO n=30. (C) Pathology and histology of murine angiosarcomas. Blue arrows point to angiosarcomas. Histology below is letter-matched to the gross pictures above (a-h). Scale bar in (h) is 50 microns and represents the scale for (a-h). (D) Histology of human angiosarcoma found in bone compared to the three genotypes of mice that develop angiosarcomas.
Scale bar in the first row is 100 microns and applies to the first row, scale bar in the second row is 50 microns and applies to all remaining pictures.
59 Figure 2.2
A te B te a O a K O O O K K K
ung Isol tpn12 ung Isol L P tpn12 O L tpn12 tpn12 O P K P P K ung rp53; ten; riple p53; ten; iple HUVEC L CD31+ T P T HUVEC Lung CD31+ Tr P Tr VEGFA - + VEGFA - +
PTPN12 pMAPK PTEN VEGFR1 MAPK VEGFR2 p4EBP1
4EBP1 VEGFR3 Actin CD31 VE Cadherin
Actin
C D IP: pTYR PDGFRβ PDGFRβ Triple KO Trp53;Ptpn12 KO Pten;Ptpn12 KO ung ung ung umor umor umor nput nput T T T L I L I L N E T P
Blot: PDGFRβ pTYR PDGFRβ
-S6
P E Triple KO Trp53;Ptpn12 KO Pten;Ptpn12 KO
APK
M
-
P PDGFRβ
Figure 2.2: mTOR and MAPK pathways are activated in murine angiosarcoma
(A) Western blot analyzing different endothelial cell markers in HUVECs untreated or treated with VEGF, total lung, endothelial cells isolated from lung using CD31 beads, tumors from both Trp53; Ptpn12 and Pten; Ptpn12 double KO mice, and tumors from triple KO mice. (B) Western blot using the same samples as in (A) but investigating signaling pathways. (C) IHC corresponding to the Western blot in (B).
Scale bar in the first row is 50 micrometers and applies to all pictures. (D)
60 Immunoprecipitation using Py-1000 antibody from Cell Signaling to isolate proteins that are phosphorylated on tyrosine residues. Membranes were then blotted for candidate RTKs. Total antibody was used for immunoprecipitation as well followed by staining with 4G10, another phospho-tyrosine antibody. (E) Tumors were stained by IHC for PDGFR-β. Scale bars are 50 microns.
61
Figure 2.3
coma A O r K KO iosa ng tpn12 O A P Ptpn12 K HUVEC Trp53; Pten; Triple Human VEGF - +
VEGFR2
VEGFR1
PDGFRβ
pMAPK
MAPK
p4EBP1
4EBP1
Actin
B rcoma oma giosa rc gioendothelioma gioma iosa rmal ng No A Heman Heman Lymphan H&E CD31 PTEN APK M p pS6 PTPN12
PDGFRβ
62 Figure 2.3: Murine angiosarcomas closely mimic human disease
(A) Western blot analysis of HUVECs compared with all three mouse genotypes of angiosarcoma and human angiosarcoma. (B) IHC analysis of normal skin blood vasculature compared with angiosarcoma and other tumors of endothelial origin.
Scale bar in lymphangiosarcoma H&E is 50 microns and applies to all images in this figure.
63 Supplemental Figure 2.1
A io 1 io 2 io 3 io 1 io 2 io 3 io 1 io 2 io 3 g g g g g g tion tion tion g g g n n n n n n a a a n n n A A A
+ - A A A A A A + - + - en tpn12 t ombin ombin ombin P P c c c e e e R R p53 R B GFAP-Cre + GFAP-Cre -
Green: bgal Red: CD31 Blue: DAPI
C GFAP-Cre + GFAP-Cre -
Green: CD31 Red:Tomato Blue: DAPI
D
E H&E KI67 CD31 PTEN PTPN12
Supplemental Figure 2.1: Mouse model deletes genes in a subset of endothelial
cells leading to angiosarcoma
64 (A) PCR for the recombination products of Ptpn12, Pten, and Trp53 show recombination of all three genes in the triple KO tumors. (B) β-gal (green) co- stained with CD31 (red) shows some overlap in GFAP-Cre + mice but not in Cre- mice, indicating that cre is active in a subset of endothelial cells. (C) Generation of
GFAP-Cre Tomato mice demonstrate Tomato activity (red) in endothelial cells
(CD31, green), indicating once again that cre is active in cells of the vasculature. (D)
Picture of angiosarcomas in triple KO mice driven by the TIE2 promoter. The tumors are generally found in the peritoneum and lungs. (E) Diagnostic stains of TIE2 Cre driven tumors.
65 Supplemental Figure 2.2
Supplemental Figure 2.2: PDGFR-β is phosphorylated in angiosarcoma while the VEGF receptors are not
Immunoprecipitations were performed with the PY1000 antibody followed by blotting with VEGFR1 (A), VEGFR2 (B), and VEGFR3 (C). Immunoprecipitations were then performed with total antibody for VEGFR2 (D) and VEGFR3 (E) followed
66 by blotting with the 4G10 phosphotyrosine antibody. These blots were also blotted with total antibody for VEGFR2 (F) and VEGFR3 (G) to verify that the pulldown worked. (H) RTK array was peformed. PDGFR-β spots are within the red boxes.
67 Supplemental Figure 2.3
oma rc O K O K iosa g n tpn12 O A P tpn12 K P HUVEC ten; riple VEGF - + Trp53; P T Human
260
140
70
50
40 35
25
Supplemental Figure 2.3: Proteins around 40 and 70kd are phosphorylated in
human and murine angiosarcomas
Western blot that was probed using the 4G10 phosphotyrosine antibody in HUVECs
as well as the three mouse genotypes of angiosarcoma, and human angiosarcoma.
68 Supplemental Table 2.1
Supplemental Table 2.1: Positions of RTK array spots
RTKs and positions of spots on the RTK array (R&D Systems).
69 Chapter 3: Development of an Angiosarcoma Cell Line from Murine Tumors
70 Introduction
We have developed a mouse model for angiosarcoma that deletes Pten,
Trp53, and Ptpn12 in a subset of endothelial cells. Deletion of Ptpn12 is critical for formation of this particular tumor type. PTPN12 is involved in dephosphorylating receptor tyrosine kinases (RTKs) as well as playing roles in cell motility and migration (72, 75, 80). Importantly, this protein is a known tumor suppressor in a number of cancers, including breast, colon, non-small cell lung cancer, esophageal squamous cell carcinoma, prostate, hepatocellular carcinoma, nasopharyngeal carcinoma, and glioblastoma (61, 80, 117-122). Although the status of PTPN12 has not been studied in angiosarcoma, another protein tyrosine phosphatase, PTPRB, has been found to be mutated in a number of human angiosarcoma tumors. This
PTP acts by dephosphorylating RTKs specifically in the endothelium, playing a role similar to that of PTPN12 (36). Phosphatases appear to be important for the development of this disease and should therefore be studied more thoroughly. Our mouse model proves ineffective for the study of the role of PTPN12 as only mice with Ptpn12 deletion develop angiosarcoma; there is no Ptpn12 wildtype tumor to compare these two. Therefore, another model must be developed in order to determine the role of PTPN12 in the context of angiosarcoma.
Tissue culture is a useful model for analyzing protein-protein interactions, migration, and analyzing signaling pathways in a time-efficient manner. Also, performing experiments in vitro can provide results in a more rapid time frame.
However, cell culture is often contested as putting cells in a dish can alter their properties (such as cell morphology, proliferation, differentiation, apoptosis, and
71 gene/protein expression) since the conditions are not physiologically relevant
(193). These changes are often blamed for the failure of drugs in vivo that appeared promising in these 2D culture systems, leading to the argument that more physiologically relevant systems must be created (194, 195). In order to create more relevant conditions that mimic the tumor microenvironment, co-culture systems can be used where tumor cells as are co-cultured with stromal cells (196).
Sphere cultures are also used to create an environment more similar to that of tumor cells: the core of these spheres often receives less oxygen, creating a hypoxic environment, while the cells around the outside of the spheres are exposed to more of the media and therefore are proliferative, similar to what occurs in tumors (197-
199). This is useful for cancer research especially as resistance to chemotherapeutics is thought to be a result of persistence of a cancer stem cell population (200).
Although we have a mouse model in which to study the disease, there are many studies that would be more efficiently performed via tissue culture. The role of PTPN12 appears to be context dependent: in normal cells, either over- or under- expression of this protein seems to result in decreased cell migration. However, in cancer cells, knockdown of the protein typically results in increased migration and invasion. Therefore, it is important not just to study the role of PTPN12 in normal endothelial cells but also in endothelial tumor cells. We would like to utilize cell culture to find the substrates of PTPN12 in this context. Using cells lines generated from our mouse model, it would be possible to interrogate known substrates of
PTPN12 as well as perform pull-downs and mass spectrometry to find new
72 substrates. The effect of PTPN12 on migration and invasion could also be better studied using a cell culture system. Currently there are no publicly available angiosarcoma cell lines, although two groups have been successful in culturing these tumors from humans (46, 47). Unfortunately, culturing endothelial cells from mice is notoriously difficult. Here we describe our attempts at culturing tumors from our mouse model.
73 Results
Cells in monolayer culture grow very slowly and do not appear to be angiosarcoma tumor cells. We had extreme difficulty in culturing these murine angiosarcoma cells, initially with problems growing any cells at all and then with growing cells that do not match our tumor cells utilizing different markers. Our cell culture trials are summarized in Table 3.1. Initial conditions for a monolayer culture were tested.
First, we utilized the base digestion media (DMEM/F12 1:1 with 1M HEPES) plus
0.05% Trypsin and 0.1% Collagenase I. Cells were plated onto plates coated with 1% gelatin (sigma Aldrich, #G9391-100G). The media used was the DMEM/F12 1:1 with
1M HEPES with the addition of endothelial cell growth supplement (ECGS), which was used at 5ug/mL, as well as Pen/Strep and Glutamax. After several days we noted that there were very few live cells present in our culture. We hypothesized that the issue was too little ECGS and increased the ECGS to the maximum recommended, 150 ug/mL. However, another attempt with this new media condition proved unsuccessful as well due to too few live cells (Figure 3.1). We proposed this was due to low cell counts following digestion.
We next tested new digestion conditions to compare the base digestion media plus 1) 0.05% Trypsin and 0.1% Collagenase I, 2) 0.1% Collagenase I, 3) 0.1%
Collagenase II (Gibco, #17101015), or 4) 0.1% Collagenase I and II. These tests demonstrated that digestion with just 0.1% Collagenase I in the base digestion media resulted in the most live cells following digestion (Table 3.2). While more cells were plated (90,000 tumor cells), many of the cells still do not look healthy and appear black in phase contrast images (Figure 3.2A). At this point the cells were
74 heterogeneous. The plates had tracks of cells running through them, patterning in a manner that appears similar to stromal cells running through tumors, and grew in clusters over the course of several months (Figure 3.2B).
We therefore decided to select for a population of cells that would consist of more tumor cells and fewer infiltrating cell types. To do this, we performed an endothelial cell isolation using CD31 conjugated magnetic beads, CD146 conjugated beads (miltenyi, #130-092-007), or serial plating to remove the fibroblast cell population. Serial plating was performed across three uncoated plates for 10 minutes each. The fibroblasts should adhere to these plates rapidly, allowing other cells to be transferred from well to well. We find that use of the CD31 coated beads resulted in the best removal of the contaminating fibroblast cells. Further, we obtain
>70% purity of endothelial cells when we perform these isolations (Supplemental
Figure 3.1A-B). These endothelial cells also exhibit recombination of Ptpn12, Pten, and Trp53, indicating that they are tumor cells (Supplemental Figure 3.1C).
Comparison of CD31 and CD146 antibodies for endothelial cell isolation by Western blot demonstrated that CD31 isolation resulted in better overall protein recovery as well as maintaining signaling found in the total tumor. Not only did the CD31 isolation result in higher levels of overall protein isolated, it also resulted in signaling similar to those of the tumor cells. CD31 isolates from tumor cells demonstrated similar levels of actin, PTEN, and VEGFR3 when compared to lysate directly from the tumors. The CD31 isolates had increased VEGFR1, indicating that the endothelial-derived tumor cells isolated have higher expression of VEGFR1 than is found in the tumor as a whole. VE-Cadherin and VEGFR2 are expressed at lower
75 levels in the CD31 isolates than in the total tumor. Finally, phospho-4EBP1 is found at slightly lower levels in the CD31 isolates than in the total tumor lysates
(Supplemental Figure 3.1D). These cells grew better than previous conditions; however, the tumor cells were adhered so strongly to the plate that 0.25% Trypsin for 10 minutes would not remove them. Use of 2.5% Trypsin yielded similar results.
In order to culture cells that are more easily removed from the plate we decided to test different matrices with which to coat the plates. We chose to test collagen (thermofisher, #A1048301), fibronectin (Millipore, #FC010-10MG), and gelatin. We found that fibronectin promoted better cell growth (data not shown); however, the cells still adhered strongly to the dish. We obtained a culturing protocol for human umbilical vein endothelial cells (HUVECs) from Dr. Boscolo in which cells were cultured on fibronectin coated plates with EGM2 (endothelial cell growth media)-LONZA media bullet kit (#CC-3156). Therefore, we decided to test our media and the LONZA media to determine whether we could achieve better cell growth. Here we find that the LONZA media resulted in better cell growth (data not shown). Unfortunately, we find that the cells that grow in these conditions do not demonstrate recombination of Pten, Trp53, or Ptpn12 (Figure 3.3).
Finally, as we had noticed in Figure 3.2 that the cells appeared to grow irregularly with different cell types running through the cultures, we considered whether the tumor cells need other cell types to grow. In order to test this hypothesis, we began by plating cells directly following digestion, CD31 positive cells obtained from CD31 isolations, and the flow-through cells from the CD31 isolations in separate dishes. The media from the flow-through cells was then
76 collected and put on the cells generated from the total tumor and the CD31 positive cells to determine whether the flow-through population of cells from the CD31 isolations were secreting factors necessary for tumor growth (Figure 3.4A). The cells in the flow through plate look like fibroblast cells from their elongated cell appearance. We find that this rapidly speeds the growth of the cells. Cells were removed from plates by scraping as the cells still adhere too strongly to be removed by trypsin. Cells were re-sorted after two weeks in culture (Figure 3.4B). Cells were split into a chamber slide for staining after about 3.5 weeks in culture. The CD31 positive cells had low expression of PTEN as would be expected of our tumor cells while the flow through cells and total tumor cells expressed higher levels. KI67 levels were relatively high as well, indicating that the cells are proliferating in all three groups. However, we find that these cells express high levels of PTPN12, which we would not expect from our tumor cells. Further, they have a very low level of CD31 positivity, indicating that they are probably not the cells we were hoping to isolate and grow (Figure 3.4C).
Cells grown in sphere culture do not exhibit recombination of genes. We decided to try to culture these murine angiosarcoma cells as spheres. Sphere cultures are better- able to maintain endogenous signaling and grow as a suspension culture, thus solving the problem of the cells adhering too strongly to the plate. We used a protocol already in place in the laboratory for the generation of neurospheres. We did this using the 0.1% Collagenase I digestion media found to work best for tumor digestion during the monolayer experiments. When this single-cell suspension was
77 placed in serum-free media, spheres began to form. Interestingly, the cells grew fairly well as spheres (Figure 3.5A). After 1 month in culture, the spheres were harvested, fixed, and suspended in histogel. These samples were then processed and paraffin-embedded. By H&E these spheres look promising; they look very similar to the triple KO tumors (Figure 3.5B). They are also negative for PTEN and GFAP but positive for CD31 by IHC as expected for these tumors (Figure 3.5C-E). KI67 was also detected by IHC, confirming that these cells grow well in vitro (Figure 3.5F).
However, by PCR these cells do not exhibit recombination of the three genes that lead to angiosarcoma formation in the mice (Figure 3.5G).
Cells lose characteristic signaling following digestion. As we experienced difficulty in obtaining cells using both the monolayer and sphere conditions we decided to test more carefully the steps that these two methods had in common. In order to test this, we performed a CD31 cell isolation as we did for the culture conditions. We took samples at each point in the process: the original tissue (lysate), the tumor cells following digestion and passage through a 70 micron filter (post-digest), the CD31 positive fraction, and the cells not selected in the CD31 selection (flow through, or
FT). HUVECs were used as controls. Comparison of both tumor and lung lysates with the cells following digestion demonstrates a marked decrease in every protein examined, PTPN12, PTEN, VEGFR2, VEGFR1, VE-Cadherin, CD31, phospho-MAPK, total MAPK, phospho-4EBP1 and total 4EBP1. The samples following this step, the
CD31 positive fraction and flow through cells, show similar decreases in signaling
78 compared with the original tissue samples (Figure 3.6). Therefore, better digestion methods may be necessary to obtain optimal purification of tumor cells for culture.
79
Discussion
Here we describe our attempts at culturing cells from our murine angiosarcoma mouse model. To do this we tested two different culture types: sphere and monolayer. We adapted a sphere-forming protocol already in use in the lab to develop neurospheres and adapted a monolayer protocol from a paper previously published by Veillette et al (115). Our first issue was a lack of viable cells within the cultures. To rectify this, we tested several different digestion techniques and found that using Collagenase I at 0.01% was the best for obtaining live cells.
These cells, however, were rapidly overgrown with fibroblasts. In order to remove this unwanted cell type, we tested three sorting conditions: magnetic sorting using CD31 conjugated beads, CD146 conjugated beads, or serial plating. We find that the CD31 conjugated beads results in the best sorting conditions to remove contaminating cell populations. We also find that the signaling is very similar to the overall tumor population by Western blot with several interesting changes. First, there is an increase in the VEGFR1 in the CD31 isolated tumor cells when compared with the overall tumor cells. We propose that this is due to the loss of infiltrating cell types that do not express VEGFR1 that may be diluting the signaling of the protein when analyzed by Western blot. We also see a decrease in VE-Cadherin and VEGFR2.
VE-Cadherin, although it is an endothelial cell marker, is typically expressed at low to no levels in angiosarcoma specimens from patients (201). Interestingly, VEGFR2 is expressed at varying levels in angiosarcoma in humans: higher levels of VEGFR2 correlates with better prognosis. It is hypothesized that the presence of VEGFR2 is
80 not important for signaling but rather is a marker for a better-differentiated, less aggressive tumor (18). Therefore, low levels of VEGFR2 present in our CD31 positive tumor population could be explained by the fact that our murine tumors have been classified as high-grade and are quite aggressive. Ultimately, we determined that CD31 isolation was best for isolation of endothelial tumor cells as well as to rid the cultures of fibroblast cells.
These cells, unfortunately, were not easily removed from the plates by trypsin, making them difficult to split. We tested whether this could be due to the matrix used to coat the plates. Previously we had been using 0.1% gelatin; we tested this condition along with collagen coated plates, fibronectin coated plates, and uncoated plates. Using these matrices we found that fibronectin coated plates increased cell growth and slightly increased the cells ability to be detached with trypsin, although none of the conditions completely rectified the situation.
Interestingly, we observed that these cells grew not in a monolayer sheet but rather formed clusters and had tracks of cells running through them similar to stromal cells running through tumors. We decided to test whether the tumor cells might need a factor secreted by other cell types to grow in culture. To test this we plated cells from the total tumor along with cells that were positively selected for CD31.
While this did help the cells to grow more quickly, the cells continued to adhere strongly to the plates.
Finally, we had noticed that the monolayer cells tended to grow in a heterogenous population with stromal-looking cells running throughout the culture.
Thus, we hypothesized that cell growth was enhanced by some property(ies) of
81 these different cell types. We chose to test this by performing a CD31 cell sort and plating cells from the total tumor immediately following digestion, CD31 positive cells, and also the flow-through cells from the CD31 isolation. We then used the media from the flow-through cells to feed the total tumor cells and the CD31 positive cells. This tested whether the flow-through cells were secreting a factor(s) that would mediate tumor cell growth. This test did, in fact, help the cells grow faster.
Despite these efforts, the cells were never the cells that we had set out to culture, the murine angiosarcoma cells when analyzed by immunocytochemistry or recombination PCR. We analyzed the cells at various points with recombination
PCRs to check for recombination of Trp53, Pten, and Ptpn12. None of our conditions yielded cells with these genes recombined. Further, staining of these cells showed little CD31 positivity while showing high positivity for PTEN and PTPN12, which we would expect to be low or absent due to recombination.
As these tumors have many stromal cells running through them, we would like to further investigate whether the stromal cells are necessary for growth in culture. In the future we would like to test this using a layer of fibroblast feeder cells. Further, we find that the signature of the cells has changed following digestion.
We would like to further test digestion techniques in order to find one in which the cells more closely resemble the original tumor post-digestion. Some potential options include using a larger cell strainer after digestion as the tumor cells are much larger than regular endothelial cells. Another possibility is to decrease the concentration of collagenase when performing the digestion or perform a purely
82 manual tumor disassociation. Finally, it would be prudent to characterize more thoroughly the tumor cells in order to ensure that the CD31 positive cells are tumor cells we should be isolating. As angiosarcoma cells become less differentiated it’s possible that they stop expressing CD31; indeed, not every tumor cell within the tumor expresses CD31 by IHC. Therefore, it’s possible that by utilizing a CD31 isolation technique we are selecting for more differentiated, less aggressive tumor cells that grow more slowly in culture. Another avenue to explore is to propagate these cells in vivo. Changes in gene/protein expression that can occur in culture is often restored when the cells are put back in vivo (194, 202). It would be interesting to implant the cells grown in culture into a mouse in order to determine whether
CD31 is perhaps one of the proteins whose expression is decreased upon growth in vitro; if this is the case we would expect that implantation in vivo would restore
CD31 levels in these cells.
There are other cells that could be used for these studies as well. The best cells to obtain are the cells cultured from primary human tumors; two of these exist, however, only one can be grown in mice (46, 47). If these are unavailable, endothelial colony forming cells could be obtained, which are thought to be circulating endothelial progenitor cells (203). Immortalized human dermal microvessel endothelial cells may even provide some insight into the role of PTPN12 in endothelial tumor-forming cells (204). Finally, the ATCC offers a sarcoma cell panel; these may be beneficial, although not a perfect model as angiosarcoma is not among them.
83 Overall, while we did not achieve our goal of developing a model system in which to explore the role of PTPN12 in angiosarcoma, we have made progress towards achieving it in the future. Elucidating the role of PTPN12 in these tumors is important as it plays a major part in the formation of our murine angiosarcomas.
Finding the substrates of PTPN12 in this context could be important for development of future targeted therapeutics.
84 Methods
Cell Preparation: 200ul/sample of Sheep anti-Rat magnetic Dynabeads
(Thermofisher, #11035) were washed 3x with PBS+0.1% BSA. 5ul of CD31 (BD
Pharmingen, #553370) per 50ul beads were added in 200ul PBS + 0.1% BSA per sample and incubated at 4C overnight. Supernatant was removed and samples added. Samples were obtained from lungs of adult FVB mice or angiosarcoma tumors. The tissues were removed, minced, and incubated in DMEM/F12 1:1 (Gibco,
#12634-010) media with collagenase 1 (Gibco, #17100-017) at 2mg/ml for 1 hour at 37C. After this hour, the tissue was passed through a 70 micron filter (greiner bio- one, #542070) and spun down. Cells were either plated at this point or continued on for endothelial cell isolation. Cells for endothelial cell isolation were resuspended in
2ml PBS+0.1% BSA and added to 40ul of the CD31 conjugated Dynabeads. These were rotated for 15 minutes at room temperature and then washed 3x in PBS+0.1%
BSA followed by plating.
Digestion media base: DMEM/F12 1:1 with 1M HEPES.
Media for monolayer: DMEM/F12 1:1 with 20% FBS, 5mL glutamax, 5mL Pen/Strep,
150ug/ml endothelial cell growth supplement (ECGS) (BD Biosciences, #354006), and 100ug/mL heparin. ECGS and heparin was added to the cells every other day.
Media for spheres: Cells were plated on low-adhesive plates. Serum-free media consisting of DMEM/F12 1:1, 1% N2, 2% B-27, 5 mL glutamax, 5 mL
85 Penicillin/Streptomycin (Pen/Strep), and 20ng of EGF (Millipore, #01-101) and FGF
(R&D systems, #233-FB-025) each. The EGF and FGF were added every other day to the cells.
Cell pellet processing: Cells were spun down at 1000 x g for 10 minutes in a 15 mL conical vial. Supernatant was removed and 1mL 4% paraformaldehyde was added to fix the cells and incubated for 10 minutes at room temperature. Cells were centrifuged as before. Just enough warmed histogel was added to cover the cell pellet. This was cooled in ice for 20 minutes. The hardened histogel pellet was then removed from the 15 mL conical using a spatula. The histogel pellet was placed in a cassette and taken to Cincinnati Children’s Hospital Medical Center (CCHMC) pathology core for processing. Histogel pellets were embedded in paraffin following processing. (Histogel and protocol provided by CCHMC Pathology).
Immunofluorescence: Day 1: Slides were de-paraffinized and hydrated utilizing a series of xylene washes (3x5 minutes) followed by ethanol (EtOH) washes (100%
EtOH 2x2 min, 95% EtOH 1x2 min, 70% EtOH 1x1 min, 50% EtOH 1x1 min, 20%
EtOH 1x1 min, running water 1x4 min). Antigen retrieval was then performed on the slides using a citrate buffer. After 30 minutes, the slides were blocked with 10% goat serum for one hour at room temperature. Primary antibodies were diluted and left overnight at 4C. Anti-CD31 antibody was used from Abcam (#28364, 1:100).
Day 2: Slides were washed in TBS 3x5 min and secondary antibody conjugated with a fluorescent tag was incubated for 1 hour at room temperature. Slides were washed
86 3x5 minutes in TBS followed by rinsing in water. Slides were dried and coverslipped with VectaShield with DAPI.
Immunohistochemistry: Day 1: Slides were de-paraffinized and hydrated utilizing a series of xylene washes (3x5 minutes) followed by ethanol (EtOH) washes (100%
EtOH 2x2 min, 95% EtOH 1x2 min, 70% EtOH 1x1 min, 50% EtOH 1x1 min, 20%
EtOH 1x1 min, running water 1x4 min). Antigen retrieval was then performed on the slides using a citrate buffer. After 30 minutes, the slides were washed with TBS
1x5 min followed by a peroxide quench for 30 minutes. The slides were washed in
TBS 2x5 min and in TBS-T 1x5 min. The slides were blocked with 10% goat serum for one hour at room temperature. Primary antibodies were diluted and left overnight at 4C. Primary antibodies: CD31 (abcam, #28364, 1:200), KI67 (cell signaling (CS), #9129, 1:200), PTEN (CS, #9559, 1:2000), PTPN12 (abcam, ab76492,
1:1000), GFAP (Sigma, G3893, 1:1000). Day 2: Slides were washed in TBS-T 3x5 min and secondary antibody was incubated for 1 hour at room temperature. Slides were washed 3x5 minutes in TBS-T followed ABC for 1 hour at room temperature. Slides were washed 3x5 min in TBS-T. DAB was applied for 10 minutes. Slides were then washed in running water for 5 minutes. Slides were incubated in hematoxylin for 5 minutes followed by a 5 minute wash in running water. Slides were then dipped in acetic acid 10x followed by dipping in water 10x. Slides were place in bluing solution for 1 minute followed by 1 minute of water. The slides were then dehydrated using a series of EtOH washes (20% 1x1 min, 50% 1x1 min, 70%
87 1x1min, 95% 1x2 min, 100% 2x2 min) followed by xylene (3x5 min). Slides were then coverslipped with Permount.
Western blot: 10 ug protein were loaded on to Invitrogen NuPAGETM 4-12% Bis-Tris
Midi Gels (WG1402BX10). Gels were run at 150 V for 1.5 hours. Running buffer was
NuPAGETM MES SDS Running Buffer (NP0002). Gels were transferred to PVDF membrane at 100 V for 1.5 hours. Following transfer membranes were blocked in
5% milk for 1 hour at room temp. Antibodies used include Cell Signaling antibodies p4EBP1 (#9459), 4EBP1 (#9452), pMAPK (#9102), MAPK (#9101), PTEN (#9559),
VEGFR2 (#2479). Antibodies from Cell Signaling are all used at 1:1000. Other antibodies include actin (Sigma, #A5441, 1:10000), CD31 (Abcam, #ab28364,
1:1000), PTPN12 (Abcam, ab76492, 1:1000), VEGFR1 (Abcam, ab32152, 1:1000),
VEGFR3 (Thermo, PA5-16871, 1:1000). Primary antibodies were diluted in 5% BSA.
Secondary antibody (rabbit: NA934), mouse: NA931 from GE) was incubated at
1:5000 in 5% milk for 1 hour at room. Blots were imaged on the UVP BioSpectrum®
Imaging System (M-26XV) and quantification was performed using their software.
88 Figures
Figure 3.1
Monolayer
Figure 3.1: Cells grown in monolayer conditions grow slowly
Cells were dissociated and plated using the methods described above. The media used included 150ug/mL ECGS. Images were taken two weeks after the cells were put in culture. All cells obtained from the digestion were used, approximately 4,000.
89 Figure 3.2
A 8/25 8/28 8/31 umor T
B 2 months in culture 3 months in culture umor T
Figure 3.2: Cells grow best with Collagenase I, but do not grow in a homogeneous sheet
(A) Tumor cells digested with the base digestion media plus only 0.1% Collagenase I were put into culture on 8/19. Many of the cells are black and do not look healthy.
(B) After several months in culture the cells begin to grow in different patterns reminiscent of stromal cells growing throughout a tumor.
90
Figure 3.3
er y ol ol r a r t t n n er o t onol o a C C M - + W
Trp53
Pten
Ptpn12
Recombination
Figure 3.3: Monolayer cells do not have Trp53, Pten, or Ptpn12 recombined
PCR was used to test whether cells obtained from the monolayer cell cultures had
recombination of the three genes that we have deleted in our mouse model.
91 Figure 3.4
A CD31 -
CD31 +
Media Collection
Media used on total tumor and CD31+ cells
B 1/25 2/3 Pre stim Begin Stim 2/5 2/12 2/18
CD31 + umor T otal T
Split cells Cells moved to Cells moved to
e Re-sort CD31 chamber slide chamber slide v
ti for for CD31 + a total tumor Neg
C CD31 KI67 PTEN PTPN12
CD31 + umor T otal T e v ti a Neg
Figure 3.4: Stimulation experiment resulted in better growth but the cells do not express characteristic markers
92 (A) Schematic of stimulation experiment for cell culture. Cells were either plated immediately following digestion or after CD31 sorting. CD31 positive and CD31 negative fractions were plated separately. Media from the CD31 negative fraction was then used on the cells from the original tumor and the CD31 positive fraction.
(B) Phase contrast images of the CD31 positive fraction, total tumor cells, and flow through fraction throughout the 3.5 weeks in culture. (C) Immunocytochemical staining of the cells from each fraction for CD31, KI67, PTEN, and PTPN12.
93 Figure 3.5
A B Phase Contrast H&E
C D
PTEN CD31
E F GFAP KI67
G er y ol ol a r r e t t r n n er o o t onol C C a Sphe M - + W p53 en t P tpn12 P
Figure 3.5: Sphere cultures express expected markers but not recombination by PCR
94 (A) Sphere formation can be seen by phase contrast. (B) H&E of processed, paraffin embedded spheres. (C) PTEN staining of spheres. (D) CD31 stain of spheres. (E)
GFAP staining of spheres. (F) KI67 staining of spheres. (G) PCR for recombination of
Trp53, Pten, and Ptpn12.
95 Figure 3.6
e digest t e t- digest sa t t- y sa L y L
umor FT ung FT umor CD31+ ung CD31+ umor ung umor pos ung pos HUVEC T L T L T L T L VEGF - +
PTPN12
PTEN
VEGFR2
VEGFR1
VE Cadherin
CD31
pMAPK
MAPK
p4EBP1
4EBP1
Actin
Figure 3.6: Digestion results in a change in cell signaling
Western blotting of samples taken after dissection (lysate), after digestion and passage through a filter (post-digest), CD31 positive cells after CD31 isolation, flow- through (FT) cells following CD31 isolation.
96 Table 3.1
Selection Digestion Media Matrix Problems
DMEM/F12 20% FBS 0.1% collagenase I glutamax none gelatin Few live cells 0.5% trypsin pen/strep 5ug ECGS heparin DMEM/F12 20% FBS No growth- 0.1% collagenase I glutamax none gelatin digestion too 0.5% trypsin pen/strep harsh? 150ug ECGS heparin DMEM/F12 0.1% Collagenase I OR 20% FBS 0.1% Collagenase II OR glutamax fibroblast none gelatin 0.1% Collagenase I and pen/strep contamination II 150ug ECGS heparin DMEM/F12 20% FBS Growth very CD31 beads glutamax 0.1% collagenase I gelatin slow, cells serial plating pen/strep stuck to plate 150ug ECGS heparin DMEM/F12 20% FBS gelatin glutamax Growth still CD31 beads 0.1% collagenase I fibronectin pen/strep slow collagen 150ug ECGS heparin Cells not DMEM/F12 CD31 beads 0.1% collagenase I fibronectin angiosarcoma Lonza cells
Table 3.1: Summary of monolayer cell culture trials
97 Table 3.1 outlines the order in which different cell culture trials were performed and why they were unsuccessful. Best conditions are written in red text.
98 Table 3.2
0.05% Trypsin + 0.1% Collagenase 0.1% Collagenase 0.1% Collagenase 0.1% Collagenase I II I + Collagenase II I Tumor - 190,000 130,000 1000
Lung 4000 260,000 3000 0
Table 3.2: Digestion with 0.1% Collagenase I results in most live cells
Four different digestion conditions were tested followed by counting of live cells with Trypan blue after passage through the filter.
99 Supplemental Figure 3.1
A Lung Tumor
Red: CD31 Blue: DAPI
t r t o r ol S o issue ol r S r t e T t n r n er o er o t f t o C a B C e f otal C B A T - + W
Ptpn12
Pten
Trp53
tes Recombination es a tes at es a at es at ung Isol s L ung Isol Tumor Isol Ly D L Tumor Isol
HUVEC CD146 CD31 CD146 CD31 Tumor VEGF - +
PTEN
PTPN12
VEGFR1
VEGFR2
VEGFR3
VE-Cadherin
p4EBP1
4EBP1
Actin
100 Supplemental Figure 3.1: CD31 cell sorting results in a 70-80% pure population of endothelial cells
(A) Immunofluorescence staining of cells isolated using the endothelial cell
isolation technique described in the methods. Red staining represents CD31
positive cells while blue staining, DAPI, shows all cell nuclei. (B) Quantification
of the cells stained in panel A. Similar purity is obtained in both lung and tumor
samples. (C) PCR for recombination products of Trp53, Pten, and Ptpn12 in the
tumor prior to CD31 isolation, after digestion (before sort), and after CD31
sorting. (D) Western blotting comparing two forms of endothelial cell isolation,
CD146 beads and CD31 beads. Isolations were performed in lung and tumor.
Tumor isolates were compared to the original tumor tissue. HUVECs that were
or were not stimulated with VEGF were used as a control.
101 Chapter 4: Combined Inhibition of MEK and mTOR Results in
Sustained Tumor Suppression
Excerpts in preparation for submission
Combined mTOR and MEK Inhibition is an Effective Therapy in a Novel Mouse Model for Angiosarcoma Michelle L Chadwick1,2; Adam Lane2; Dana Thomas2; Amanda R Smith2; Angela R White2; Yuxin Feng2; Elisa Boscolo2; Yi Zheng1,2; Denise Adams3; Anita Gupta4; Andre Veillette5; Lionel ML Chow1,2
1Department of Cancer and Cell Biology, University of Cincinnati, Cincinnati, OH 2Cancer and Blood Diseases Institute, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH 3Vascular Anomalies Center, Boston Children’s Hospital, Boston, MA 4Department of Pathology, Cincinnati Children’s Hospital Medical Center, Cincinnati, OH 5Institut de Recherches Cliniques de Montréal, Montréal, Canada
102 Introduction
Angiosarcoma is a rare malignancy of the endothelium that does not have many therapeutic options. The most effective treatment is surgical resection; however, many tumors are not resectable due to their location or metastases. For these patients, treatment options include taxanes and anthracyclines; however, prognosis remains poor with median five-year survival ranging from 31-43% and median time to progression at 5-7.6 months with paclitaxel treatment (1-7, 15, 16).
Clearly, novel treatment strategies are necessary for this disease.
The mTOR and MAPK pathways are activated in these tumors; often, they are both activated within the same tumor (13, 32). Angiosarcoma patient samples show decreased expression of PTEN or mutations in PIK3CA, which can lead to activation of the mTOR pathway (36, 37). Mutations in PTPRB, RAF genes, MAPK1, and NF1 can lead to activation of the MAPK pathway (32, 36). Mutations in these genes can be found in over half of angiosarcoma samples (32). Indeed, activation of these pathways are prominent in angiosarcoma; one study found elevation of the mTOR pathway in 100% of 59 tumor samples while another found phosphorylation of S6 and 4EBP1 in 42% of samples (13, 175). Importantly, 65% of the samples that had activation of the mTOR pathway had concurrent activation of the MAPK pathway
(13).
Inhibition of the mTOR pathway alone is somewhat effective in these patients; a clinical trial of everolimus demonstrated the best responses in angiosarcoma patients. However, this study only had three angiosarcoma patients enrolled (41). Use of MEK inhibitors in xenografts of a canine angiosarcoma cell line
103 resulted in decreased tumor growth (39). Combined inhibition of mTOR and MEK in these same xenografts proved even more effective (38). Indeed, inhibition of one pathway often leads to resistance due to increased activity of the other (157-166).
Therefore, inhibition of both pathways simultaneously may provide clinical benefit to angiosarcoma patients who have either mTOR, MAPK, or both pathways activated in their tumors.
Our mouse model utilizes deletion of Pten, whose expression is decreased in human tumors, and Ptpn12, a phosphatase with similar functions to PTPRB, which is mutated in human tumors, along with Trp53 deletion (a deletion that often co- occurs with PTPRB mutations in humans) (36). We have further shown activation of both the mTOR and MAPK pathways in our mouse model similar to human samples.
Therefore, our mouse model provides a good system in which to test therapeutics for this disease. The goal of these studies is to determine whether inhibition of the mTOR pathway, the MAPK pathway, or both could provide a viable therapeutic option for treatment of angiosarcoma patients. We propose that inhibition of both pathways will result in the best repression of tumor growth, resulting in improved outcomes. Our long-term goal is to utilize the information gained from these studies in order to find a better treatment strategy for patients with angiosarcoma.
104 Results
The mTOR pathway can be effectively inhibited in angiosarcoma and results in decreased tumor growth. We have previously demonstrated activation of the mTOR and MAPK pathways in our murine angiosarcoma model, pathways that have been shown to be elevated in patient samples as well. In order to determine whether these pathways are important for the growth of murine angiosarcomas we treated mice with small-molecule targeted inhibitors. Rapamycin and trametinib (a MEK inhibitor) were selected to inhibit the mTOR and MAPK pathways respectively as these are both selective for their targets and FDA-approved for cancer therapy. To determine the lowest biochemically effective dose of each drug, we treated mice whose tumor measured at least 5mm in its longest diameter for five days.
Five doses of rapamycin were tested in vivo: 25mg/kg, 10mg/kg, 5mg/kg,
3mg/kg, and 1mg/kg. Rapamycin effectively inhibited the mTOR pathway down to a dose of 3mg/kg as assessed by phosphorylation of AKT, S6, and 4EBP1 by Western blot (Figure 4.1A-C) and IHC (Figure 4.3E-F). Phosphorylation of AKT at S473 is mediated by mTORC2, not mTORC1, and is therefore not expected to be inhibited by rapamycin, which is classically though to inhibit only mTORC1. However, in some cell lines such as endothelial cells prolonged exposure to rapamycin can suppress phospho-AKT, as we see in our five day treatments (140, 142). At the lowest dose tested, 1mg/kg, phosphorylation of AKT and 4EBP1 was not significantly diminished; however, S6 phosphorylation remained low. A significant biological effect of treatment with rapamycin was also noted as tumor cell proliferation was decreased and apoptosis, as measured by Cleaved Caspase-3 was increased (Figure
105 4.1E). Moreover all treatment doses tested resulted in regression of tumor size over the five day period of observations (Figure 4.1G). Importantly, treatment with rapamycin appeared to result in an increase in phosphorylation of MAPK suggesting that the tumor cells were attempting to compensate for the loss of proliferative signals from mTORC1 inhibition (Figure 4.1A,D,E).
Treatment with a MEK inhibitor results in suppression of the MAPK pathway and decreased tumor growth. The MEK inhibitor, trametinib was tested at three different doses in vivo: 3mg/kg, 2mg/kg, and 1mg/kg. We found that phosphorylation of
MAPK is effectively inhibited at a dose of 2 mg/kg or greater while partial inhibition was seen at 1mg/kg by Western blot (Figure 4.2A-B) and IHC (Figure 4.2C). All treatment doses significantly decreased proliferation as measured by KI67 and increased apoptosis as detected by Cleaved Caspase-3 (Figure 4.2C-D). As was observed with rapamycin treatment, trametinib treatment at all doses was able to shrink established tumors over the course of five days (Figure 4.2E). Interestingly, the effect on mTOR pathway signaling with MEK inhibition varied at each node. AKT phosphorylation was slightly increased while S6 phosphorylation was diminished and that of 4EBP1 remained relatively unaffected (Figure 4.2A,C).
Combination treatment with mTOR and MEK inhibitors results in inhibition of both pathways and decreased tumor size. We tested combined treatment of both rapamycin and trametinib in vivo compared to either drug alone at a dose of 1mg/kg for 5 days so that we could detect if synergistic or additive effects were present.
106 Dual treatment with these inhibitors resulted in a more profound suppression of signaling in the PI3K/mTOR and MAPK pathways than was seen with either treatment alone at the same dose (Figure 4.3A). In particular, phospho-4EBP1 was reduced by the dual treatment compared to treatment with rapamycin alone and phospho-MAPK was also diminished compared to treatment with trametinib alone
(Supplemental Figure 4.1A,B). Importantly the compensatory hyperphosphorylation of MAPK induced by treatment with rapamycin alone was completely reversed by the addition of trametinib (Supplemental Figure 4.1B). Similarly, increased phosphorylation of both MEK and AKT induced by trametinib was reversed by the addition of rapamycin (Figure 4.3A and Supplemental Figure 4.1C). Despite the enhanced biochemical effect on the pathways induced by dual inhibitor treatment, proliferation was similar when compared to either treatment alone (Figure 4.3B,C) and cell death as measured by TUNEL staining was also similar to rapamycin treatment alone (Figure 4.3B,D). Again, there was a significant regression of tumors over the five day treatment period when compared to vehicle treated animals, but this effect was similar to that of either drug treatment alone (Figure 4.3E).
Long-term treatment with mTOR and MEK inhibitors is synergistic and results in a sustained response. In view of the robust biochemical response to rapamycin and trametinib and evidence for tumor regression after only five days of treatment, a long-term treatment and survival study was carried out. Triple KO mice were observed until a single angiosarcoma reached a calculated volume of 100mm3. They were then assigned to one of four treatment arms: vehicle, rapamycin, trametinib, or
107 dual therapy using rapamycin and trametinib at 1mg/kg each. All treatments were administered daily until the animal developed morbidity from tumor growth or from the treatment, or until they reached 140 days of treatment whichever came first. As was observed during five-day treatments of mice, all drug treatments were initially effective at inducing tumor regression when compared to the vehicle control (Figure 4.4A). However, most of the tumors treated with rapamycin and every tumor treated with trametinib regrew aggressively between 40 to 80 days of treatment (Figure 4.4A,B). In contrast, mice treated with both inhibitors experienced tumor regression which was sustained over the 140 day treatment period (Figure 4.4A). None of the dual drug treated mice succumbed to tumor growth while receiving treatment, however several animals did die of presumed toxicity of the combined inhibitors (Figure 4.4B). As a consequence, overall survival of dual inhibitor treated mice was not improved compared to either inhibitor alone
(Supplemental Figure 4.2A). Further, reducing the dose to 0.5mg/kg for each drug did not diminish the toxicities seen (Supplemental Figure 4.2B). Full necropsy of these animals did not reveal the presence of metastatic disease or any other tumor growth. Importantly, angiosarcomas recurred in mice upon withdrawal of the drugs after 140 days of treatment indicating the persistence of viable tumor cells (Figure
4.4A). Response to inhibitor treatment was also assessed by waterfall plot (Figure
4.4C). This analysis demonstrated that the best responses were achieved in mice who received both rapamycin and trametinib concurrently. Remarkably, 14 of the
15 tumors treated with both inhibitors achieved a partial response (defined as a decrease in tumor volume by at least 50%). This rate of response was significantly
108 greater than that seen in vehicle or trametinib treated mice, and approached statistical significance when compared to rapamycin treatment (Table 4.1).
To determine the cause of tumor recurrence after prolonged treatment with either inhibitor alone, tumors were assessed for pathway activation at various time points indicated in Figure 4.4A (a-e). Tumors that recurred while on rapamycin treatment maintained complete repression of phospho-S6; similarly, tumors that recurred while on treatment with trametinib exhibited low levels of phospho-MAPK
(Figure 4.4D, Supplemental Figure 4.3A,B). These results suggest that tumor recurrence is not due loss of the ability of rapamycin or trametinib to suppress their molecular targets, but rather to the development of parallel pathways that support renewed tumor growth. In fact, IHC for KI67 on recurrent tumors revealed brisk levels of proliferation similar to levels seen in vehicle treated tumors while apoptosis remained elevated in treated tumors as detected by TUNEL staining
(Figure 4.4D, Supplemental Figure 4.3C,D). In tumors treated for five days with rapamycin or trametinib, hyperactivation of the uninhibited pathway was noted
(Figure 4.3A and Supplemental Figure 4.1B,C). In the long-term treatments, some tumors continued to exhibit similar hyperphosphorylation; however, this was not consistently seen in all tumors. As PTPN12 regulates at the RTK level, it’s possible that other effectors downstream of the receptors could be elevated to drive tumor growth. Therefore, we assessed another RTK pathway, the STAT3 (signal transducer and activator of transcription 3) pathway, to determine whether it could be contributing to proliferative signals leading to tumor recurrence. Neither rapamycin nor trametinib treated tumors were found to have consistently elevated
109 levels of phospho-STAT3 Y705 (Supplemental Figure 4.3A,B). We also assessed focal adhesion kinase (FAK), a protein regulated by PTPN12 that can be involved in MAPK signaling, to determine whether activation of this protein could be driving tumor growth in trametinib treated mice. However, recurring tumors in trametinib treated mice did not have an increase in FAK phosphorylation (Supplemental Figure 4.3B).
The presence of a stem cell population has been implicated in mechanisms of resistance in other tumors; we therefore used SOX2 staining as a stem cell marker to determine whether the resistant tumors had an increase in stem cells (205).
However, this is not the case in these tumors as we do not see positive staining for
SOX2 (Supplemental Figure 4.3E). While other growth promoting pathways may be involved in tumor regrowth in mice treated with a single agent, it is possible that tumor proliferation is driven by a subtle shift in signaling towards the mTOR pathway in tumors treated with trametinib and towards the MAPK pathway in tumors treated with rapamycin, emphasizing the importance of simultaneous inhibition of both pathways in order to achieve a sustained response.
110 Discussion
As the mTOR and MAPK pathways are activated in both murine and human angiosarcomas, we hypothesized that these pathways are critical for tumor growth.
In order to assess the importance of these pathways, we investigated the effect of inhibition of these pathways separately and in combination in vivo in our mouse model using targeted mTOR and MEK inhibitors. We selected rapamycin as the mTOR inhibitor as it is selective for mTOR; this is important to prove that anti- tumor effects seen are due to inhibition of mTOR and not off-target effects. Further, rapamycin is well-characterized in mice and humans and is well-tolerated; our laboratory has performed studies using rapamycin at 25mg/kg in mice long-term with no side effects (unpublished data). Inhibitors of MAPK itself are very new and have not been tested extensively in vivo (206). Therefore, MEK1/2 inhibitors are typically used to inhibit MAPK as MEK is directly upstream of MAPK and has no other known substrates besides MAPK. We have selected trametinib as this drug is selective for MEK1/2 and is the only FDA-approved MEK inhibitor for cancer treatment (for BRAF mutant melanoma specifically). Trametinib is thought to have fewer side-effects than other MEK inhibitors such as PD0325901 as it does not penetrate the brain and therefore does not exhibit the neurological side-effects seen with PD0325901 (207). Further, it is being tested in a phase I/IIa clinical trial for use in pediatric cancers currently, including pediatric sarcomas (clinicaltrials.gov,
NCT02124772).
First we determined that administration of these drugs at doses relevant to clinical applications effectively inhibited the intended targets. We find that while
111 these drugs do inhibit the intended pathways, they have the unexpected effect of increasing signaling of the reciprocal pathway; treatment with both drugs abrogated this effect. We were surprised that five-day courses of either drug alone or in combination resulted in significant regression of tumor volume. As these results were promising we decided to use these drugs long-term to determine whether they represent a viable treatment option for angiosarcoma patients who desperately need new therapeutic options.
Although initially these monotherapies decreased tumor volume, long-term treatment proved that these tumors progressed while still undergoing therapy.
However, we find that these tumors are not re-activating the inhibited pathways; mTOR remains inhibited in the rapamycin-treated tumors and MAPK remains inhibited in the trametinib-treated tumors. Recently a study was done finding that angiosarcoma patient samples contained a population of stem cells (208). As tumor resistance to therapy can be attributed to the survival of a population of cancer stem cells we performed SOX2 staining to determine whether this was the case in our therapy-resistant murine angiosarcomas (205). However, we see no staining within the tumor cells, indicating that this is likely not the mechanism of resistance.
However, testing of other stem cell markers to verify this finding is prudent. As
PTPN12 regulates the RTKs, it is possible that other pathways downstream of these
RTKs may become elevated, resulting in resistance to therapy. Indeed, resistance to trametinib can be attributed to STAT3 activation in other cancers (209). However, we do not see elevation of this pathway compared with vehicle, indicating that
STAT3 signaling is not responsible for tumor growth. We therefore propose that the
112 increase in signaling of the reciprocal pathways that we see in the five-day treatments is responsible for tumor recurrence. This could be further tested by adding the inhibitor for the reciprocal pathway after the tumors become resistant to monotherapy. Indeed, we do find that combined therapy resulted in a more profound and sustained tumor regression than either drug alone.
A significant limitation of our study is the mortality of mice due to toxicity from the combined agents. To mitigate these effects, we defined the minimally effective biochemical dose of each agent independently, then decreased each drug by one dose level in combination (1 mg/kg each). We conducted a further in vivo dose de-escalation study to 0.5 mg/kg of each drug in combination which is 25% and 17% respectively of the minimally effective biochemical dose of trametinib and rapamycin in this model. Remarkably, this combination also resulted in sustained anti-tumor activity comparable to higher doses, but still caused treatment limiting toxicity in mice. This toxicity issue could be overcome in a number of ways. First, there are many different mTOR and MEK inhibitors; perhaps a different drug combination could achieve inhibition of both pathways with less toxicity. There are reports of RAD001 (a rapalog) or rapamycin and PD0325901 (a MEK inhibitor) in mouse models for prostate cancer where no toxicities were reported at biochemically effective doses (157, 158). Temsirolimus has been used in combination in vivo with PD0325901 in canine hemangiosarcoma xenografts; while this decreased tumor growth with no known toxicities, the study performed was only three weeks long (38). Further, it’s possible that a second-generation ATP- competitive mTOR inhibitor may prove more efficacious, either alone or in
113 combination with MAPK inhibitors. A combination of ATP-competitive mTOR inhibitor AZD8055 with the MEK inhibitor selumetinib was well-tolerated and effective in nude xenograft mouse models for lung and colon cancer. However, these mice were only treated for 20 days (160). However, due to their increased effects on other kinases compared with rapalogs, it would be more difficult to determine whether any effects seen were due specifically to inhibition of mTOR or due to off- target effects of the drugs. Finally, adjusting the dosing schedule may prove efficacious. The study previously mentioned testing RAD001 and PD0352901 in mice dosed every other day; the studies with rapamycin and PD0352901 and temsirolimus and PD0352901 were dosed daily for five days followed by two days off (38, 157, 158). The half-life of rapamycin in mice when injected intraperitoneally has been found to be around 15 hours; the half-life of trametinib in mice dosed by oral gavage is calculated to be 33 hours (207, 210). Therefore, it is possible that dosing with trametinib every other day may be sufficient to maintain repression of the MAPK pathway while decreasing toxicities.
As discussed previously, combinations of MEK and mTOR inhibitors have been studied in early phase clinical trials (167, 168). One phase I trial studied the combination of trametinib and everolimus; this study failed due to the large number of adverse effects of this combination (168). A phase II trial of selumtinib and temsirolimus found few toxicities; however, selumetinib was used significantly below its single agent dose (167). No pharmacodynamic studies in tumor tissue were conducted in either trial. The dramatic response to combined in vivo mTOR and MEK inhibition which we have observed in our genetically engineered animal
114 model for angiosarcoma should renew efforts to design clinical trials for patients with this disease. We suggest that these trials should incorporate tumor pharmacodynamic analyses in order to determine target inhibition, a dose de- escalation scheme, and adequate supportive care to mitigate the incidence of mucosal inflammation and stomatitis which were the most frequent grade 3 toxicities noted in previous clinical trials of combined therapy. Developing a tolerable combination therapy is important not just for treatment of angiosarcoma, in which the mTOR and MAPK pathways are frequently elevated simultaneously, but also for the other forms of cancer that have activation of both pathways.
115 Methods
Mice: Trp53 floxed mice were obtained from Mouse Models of Human Cancer
Consortium Repository (191), Pten floxed mice from Dr. Mak at the University of
Toronto (192), Ptpn12 floxed mice from Dr. Veillette at the Institut de Recherches
Cliniques de Montréal (89), and TIE2-CreER mice from Dr. Zheng at Cincinnati
Children’s Hospital (unpublished). Mice of all genetic combinations of the genes of interest were developed, 7 lines in total. A stock solution of Tamoxifen (Sigma) was made at 20mg/ml in corn oil and filtered. Mice were injected with Tamoxifen at
9mg/40g body weight between 28 and 42 days of age to circumvent developmental defects for 3 consecutive days. Mice were observed until moribund or reaching 18 months of age. Upon dissection, mice were perfused with 12 ml PBS. Tissues were both frozen and fixed in 4% paraformaldehyde (PFA). Frozen tissues were used for
RNA, DNA, and protein analyses while the PFA fixed tissues were processed by pathology at CCHMC and used for IHC.
Drug Testing: For rapamycin (LC labs) dose 25mg/kg, a stock solution was made at
50mg/ml in dimethyl sulfoxide (DMSO). Prior to treatment this solution was diluted to a working solution of 2.5 mg/ml in 5.2% Tween 80 in ddH2O. The 10, 5, 3, and
1mg/kg doses were made from a stock solution of 10mg/ml in DMSO and diluted to working solutions of 1, 0.5, 0.3, and 0.1 mg/ml respectively in 5.2% Tween 80. For trametinib (LC labs) doses 3, 2, and 1mg/kg a stock solution was made at 10mg/ml in DMSO and diluted prior to use to the working solutions of 0.3, 0.2, and 0.1 mg/ml respectively in 5.2% Tween 80. Stock solutions were stored at -20. These were
116 aliquoted and diluted fresh daily to make the working solutions. Mice were treated daily for 5 days once they had a tumor reaching 5mm in size in one direction with
100ul/10g body weight of working solutions to achieve the desired doses.
Rapamycin was dosed through intraperitoneal (IP) injection while trametinib was administered via oral gavage. Vehicle solutions (DMSO with no drug) for each drug were prepared and administered in the same manner as the drug treatments. Mice treated with rapamycin were euthanized 2 hours after the last dose, trametinib treated mice 4 hours after the last dose, and dual treated mice 3 hours after the last dose. Samples were both frozen and PFA-fixed.
Drug Survival Study: Mice were treated daily once tumors reached 100mm3 with either vehicle, rapamycin, trametinib, or a combination of both drugs prepared at the 1mg/kg dose as stated above. Stock solutions of drugs were prepared and aliquoted as above biweekly. Tumors were measured by calipers 3 times weekly during treatment and for one week prior to treatment. Mice were weighed once weekly throughout treatment. Endpoints included tumor ulceration, dehydration, tumor burden, and inability to reach food and water. Both frozen and PFA-fixed samples were collected.
Western Blot: 10 ug protein were loaded on to Invitrogen NuPAGETM 4-12% Bis-Tris
Midi Gels (WG1402BX10). Gels were run at 150 V for 1.5 hours. Running buffer was
NuPAGETM MES SDS Running Buffer (NP0002). Gels were transferred to PVDF membrane at 100 V for 1.5 hours. Following transfer membranes were blocked in
117 5% milk for 1 hour at room temp. Antibodies used include pS6 (#2211), S6 (#2217), pAKT S473 (#9271), AKT (#4685), p4EBP1 (#9459), 4EBP1 (#9452), pMAPK
(#9102), MAPK (#9101), PTEN (#9559), FAK (#3285), VEGFR2 (#2479), Shc
(#2432), PDGFRβ (#3169). Antibodies from Cell Signaling are all used at 1:1000.
Other antibodies include actin (Sigma, #A5441, 1:10000), CD31 (Abcam, #ab28364,
1:1000), 4G10 (Millipore, #05-321, 1:1000), PTPN12 (Abcam, ab76492, 1:1000), pFAK (Sigma, f7926, 1:1000), VEGFR1 (Abcam, ab32152, 1:1000), VEGFR3 (Thermo,
PA5-16871, 1:1000), paxillin (BD, #610619, 1:1000). Primary antibodies were diluted in 5% BSA. Secondary antibody (rabbit: NA934), mouse: NA931 from GE) was incubated at 1:5000 in 5% milk for 1 hour at room. Blots were imaged on the
UVP BioSpectrum® Imaging System (M-26XV) and quantification was performed using their software.
Statistics: Statistics on quantifications of Western blots and 5-day treatment tumor volumes are two-tailed t tests performed in GraphPad. Fisher’s Exact Test was performed to compare response rate of different treatment groups following long- term treatment. * P≤0.05, ** P≤0.01, *** P≤0.001
Immunohistochemistry: Samples for immunohistochemistry (IHC) were fixed in 4%
PFA overnight and processed by the Pathology Core at CCHMC. After embedding,
5um thick sections were made. Slides were placed in xylene washes followed by rehydration in a series of ethanol washes. Antigen retrieval was then performed using citrate buffer. A peroxide quench was then performed using 0.6% H2O2.
118 Tissues were blocked for 1 hour at room temp in serum. In addition to the antibodies used above for Western blotting, KI67 (Leica, NCL-Ki67p, 1:4000) and
Cleaved Caspase-3 (BD, 559565, 1:500) were used. Phospho-MAPK 1:400, PTEN
1:1000, pS6 1:500, PDGFRβ 1:200.
Author Contributions
MC is the primary author and contributed experimentally, with data interpretation, and writing the manuscript along with LC. DT, AS, AW provided help with experiments and mouse work. AL contributed to the statistical analyses, especially in designing the mouse survival studies. AV provided the Ptpn12fl/fl mice. DA and EB provided valuable intellectual input and suggestions.
119 Figures
Figure 4.1
A Rapamycin Vehicle 25 mg/kg 10 mg/kg 5 mg/kg 3 mg/kg 1 mg/kg
pAKT S473
AKT
p4EBP1
4EBP1
pMAPK
MAPK
Actin
B C D ** * *
E F Vehicle 25 mg/kg 10 mg/kg 5 mg/kg 3 mg/kg 1 mg/kg
* -S6 P APK M - P -3 e ed v a aspas Cle C KI67
G
**
Figure 4.1: Rapamycin is effective down to 1mg/kg
120 (A) Western blot of vehicle compared with rapamycin after five days of treatment at
25mg/kg, 10mg/kg, 5mg/kg, 3mg/kg, and 1mg/kg. (B-D) Quantification of (A). (E)
IHC of the same samples as (A) demonstrating activity of mTOR, MAPK, apoptosis, and proliferation. Scale bar in pS6 for 1mg/kg represents 50 microns and applies to images in this figure. (F) Quantification of proliferation from (E). (G) Change in tumor volume of the different drug concentrations over the course of five days of treatment. Vehicle n=6, 25mg/kg n=8, 10mg/kg n=3, 5mg/kg n=3, 3mg/kg n=3,
1mg/kg n=3. Two-tailed t tests were used for all statistics in this figure. * P≤0.05, **
P≤0.01, *** P≤0.001
121 Figure 4.2
A C Trametinib Vehicle 3 mg/kg 2 mg/kg 1 mg/kg Vehicle 3 mg/kg 2 mg/kg 1 mg/kg pMAPK APK M - MAPK P
pAKT -S6 P AKT ed -3 v p
p4EBP1 a as Cle C 4EBP1 67 Actin I K
B D
** *
E
**
Figure 4.2: Trametinib is effective down to 1mg/kg
(A) Western blot analysis of vehicle compared with Trametinib after five days of
treatment at 3mg/kg, 2mg/kg, and 1mg/kg. (B) Quantification of MAPK from the
122 Western blot in (A). n=3 for all treatments. (C) IHC data from the same samples as in
(A) determining activity of the mTOR and MAPK pathways as well as apoptosis and proliferation. (D) Quantification of proliferation from the IHC in (C). Two-tailed t tests were used. (E) Comparison of change in tumor volumes between the different drug concentrations and the vehicle control. Vehicle n=7, 3mg/kg n=8, 2mg/kg n=3,
1mg/kg n=3. * P≤0.05, ** P≤0.01, *** P≤0.001
123 Figure 4.3
A C 1mg/kg Vehicle Rapa Tram Dual * pAKT S473
AKT pMEK MEK
pMAPK
MAPK
p4EBP1
4EBP1 D Actin
B Rapamycin Trametinib Dual Vehicle 1mg/kg 1mg/kg 1mg/kg 67 I K TUNEL d 3 e - v p a s
e E a l C C 6 S p ** K P MA p
Figure 4.3: Five-day treatment with trametinib, rapamycin, or a combination of both results in disease regression
124 (A) Western blot comparing all treatments at 1mg/kg and the vehicle. (B) IHC demonstrating proliferation, apoptosis, mTOR, and MAPK signaling after five days of treatment with either vehicle, rapamycin, trametinib, or both inhibitors in combination. Scale bar KI67 dual 1mg/kg picture is 50 micrometers and applies to all images in this figure. (C) Quantification of KI67 staining from (B). (D)
Quantification of TUNEL staining from (B). (E) Comparison of change in tumor volume over the course of the five day treatments. Two-tailed t tests were performed for all statistics. * P≤0.05, ** P≤0.01
125 Figure 4.4
A
End treatment
a c b d e
Day 20 40 60 80 100 120 140 Treatment Number of Mice Alive Vehicle 4 3 0 0 0 0 0 aRapamycin 12 12 10 6 5 3 1 B Trame nib 11 7 2 0 0 0 0 C Dual 11 6 5 4 4 4 3 B
*** ** ***
E D Cleaved pS6 pMAPK KI67 Caspase-3 TUNEL a e l c i h e V n i t b D c n y a t m s i a s p e a R R b t c i n n i a t t e s i s m e a r R T d l a u D
t e t n - s e o m P t l a a e u r T D
Figure 4.4: Long-term treatment using both rapamycin and trametinib results in sustainable disease regression
(A) Comparison of change in tumor volume of long-term treatment. Bold arrow represents when mice were taken off treatment at 140 days. Regular arrows represent the time points at which samples were analyzed. (B) Disease-related mortality shown by Kaplan Meier. Stars beside legend represent statistically
126 significant difference compared with dual treatment. (C) Waterfall plot showing worst to best responses. (D) mTOR, MAPK, proliferation, and apoptosis analysis by
IHC. Lowercase letters are the time-points that the tissue was harvested as delineated in (A). Scale bar in vehicle cleaved caspase-3 represents 50 microns and applies to all images in this figure.
127 Table 4.1
Response to Treatment
PR (50% Decrease) Dual 14 1 Dual vs Veh 0.0002 Rapamycin 6 6 Dual vs Rapa 0.054 Trametinib 5 7 Dual vs Tram 0.009 Vehicle 0 6 Rapa vs Veh 0.043 Rapa vs Tram 0.685 Tram vs Veh 0.1141 Table 4.1: Statistical analysis of waterfall data Waterfall data was analyzed by fisher exact test. 128 Supplemental Figure 4.1 A C ** * * B * ** 129 Supplemental Figure 4.1: Combination therapy results in best inhibition of mTOR and MAPK signaling pathways Quantification and statistical significance for the Western blot in Figure 4.3. (A) We see statistical significance between vehicle and dual treatment as well as rapamycin and dual treatment in phospho-4EBP1. (B) There is a significant increase in phospho-AKT S473 between the vehicle and trametinib treatment a decrease between trametinib and dual treatment. (C) Significance is seen between vehicle and trametinib or dual and between rapamycin and dual in phospho-MAPK levels. * P≤0.05, ** P≤0.01, *** P≤0.001 130 Supplemental Figure 4.2 A B Day 20 40 60 80 100 120 140 Treatment Number of Mice Alive 1mg/kg 11 6 5 4 4 4 3 0.5mg/kg 9 7 2 1 1 1 1 Supplemental Figure 4.2: Dual treatment results in toxicities, even at a dose of 0.5mg/kg of each drug 131 (A) Overall survival of mice on long-term drug treatments at 1mg/kg. Dual treated mice die largely due to drug toxicities rather than disease progression. (B) Tumor volumes of survival study with addition of 0.5 mg/kg dual dose group. 132 Supplemental Figure 4.3 A Rapamycin 1mg/kg B Trametinib 1mg/kg Vehicle 5 Day Recurrence Vehicle 5 day Recurrence PTEN pSTAT3 Y705 pSTAT3Y705 STAT3 pFAK STAT3 FAK pAKT S473 pAKT AKT AKT pMAPK pMAPK MAPK MAPK p4EBP1 p4EBP1 4EBP1 4EBP1 Actin Actin C D E Rapamycin Trametinib Positive Control Vehicle Resistant Resistant X2 O S Supplemental Figure 4.3: Therapeutic resistance is not due to re-activation of inhibited pathways 133 (A) Western blot of rapamycin resistant samples compared with vehicle or five day treatments. (B) Western blot of trametinib resistant samples compared with vehicle or five day treatments. (C) Quantification of TUNEL staining from Figure 4.4D. (D) Quantification of KI67 staining from Figure 4.4D. (E) SOX2 staining of resistant angiosarcoma tumors compared with a SOX2 positive murine brain tumor. 134 Chapter 5: Discussion and Future Directions 135 Development of a Novel Mouse Model for Angiosarcoma Currently little is known about the pathology of angiosarcoma. Five-year survival for these patients remains poor at around 31-43%; this number is even lower for patients with unresectable or metastatic disease (1-7). Gaining knowledge of the mechanisms leading to tumor formation and disease progression is critical in order to develop better therapeutics for these patients. We have developed a genetically engineered mouse model for angiosarcoma that comprises deletion of Trp53, Pten, and Ptpn12 using a GFAP-CreER construct to restrict expression of the floxed alleles. We show that this allows for the deletion of our target genes in a subset of endothelial cells. This GFAP driver is more conducive to the study of angiosarcoma than a cre driver that targets all endothelial cells such as TIE2 as it results in fewer tumors formed. These GFAP tumors can then grow to a size that is more amenable for study. Use of the TIE2-CreER driver results in numerous small tumors that result in death due to disease although no one tumor grows large enough to be measured. Further, the GFAP-CreER mice develop tumors primarily subcutaneously which makes tumor measurement possible. The result of the loss of these three genes in the endothelium results in an angiosarcoma that closely recapitulates human disease. These murine angiosarcomas typically present subcutaneously on the head and limbs, although other locations have been noted more infrequently. Histologically, they also mimic human disease in that they have large, atypical nuclei, anastomosing channels, and stain strongly for CD31 but negatively for PROX1 by IHC. Triple KO and Trp53;Ptpn12 tumors have been classified as high grade while the Pten;Ptpn12 tumors are low grade. 136 Involvement of these or similar genes can be found in the development of human angiosarcoma. PTEN protein expression is decreased in a number of angiosarcoma samples (37). Decreased PTEN is associated with activation of the mTOR pathway, a common event in angiosarcoma. TP53 mutations are seen in angiosarcoma patient samples as well, often in conjunction PTPRB mutations, a PTP that plays a role similar to that of PTPN12 (36). Both phosphatases are responsible for dephosphorylating RTKs, leading to the negative regulation of the MAPK pathway. However, the involvement of PTPN12 in angioarcoma formation is a novel finding of these studies and has not been analyzed previously in patient samples. Mutations in this gene have not been noted; however, only a small number of samples have been fully sequenced to date. Further, PTPN12 isn’t often mutated in the cancers that it has been associated with; rather, PTPN12 protein expression is typically decreased (80, 113, 117, 119, 121). As our studies indicate that PTPN12 loss is important for the development of angiosarcoma (as only genetic combinations with loss of Ptpn12 develop this particular tumor), it is important to assess the status of PTPN12 in human tumors in the future. Our data indicate that PTPN12 is expressed at a low level in patient samples compared with normal skin endothelial controls; however, this should be expanded to include more samples. Also, how PTPN12 is down-regulated in these tumors has not been assessed. PTPN12 can be regulated by microRNAs; the status of these microRNAs should be analyzed in angiosarcoma samples to determine whether this is responsible for decreased PTPN12 protein expression. DNA methylation of the CpG island found in the promoter region of PTPN12 should be examined as well. DNA methylation leads 137 to repression of gene expression; this can be reversed through the use of inhibitors, often targeting DNA methyltransferases, the enzyme responsible for the addition of the methyl groups to the DNA. These are interesting questions to answer as, if PTPN12 is regulated by one of these mechanisms in these tumors, it may be possible to target the microRNA(s) or protein(s) responsible in order to restore PTPN12 levels. Patient samples have demonstrated elevation of both the mTOR and MAPK pathways through various means. Decreased protein expression of PTEN has been noted as well as activating mutations in PIK3CA, both of which lead to activation of the mTOR pathway (36, 37). Further, mutations in different genes involved in the MAPK pathway were found in 53% of samples, leading to activation of the MAPK pathway (32). Finally, 42% of samples demonstrated activation of the mTOR pathway; 65% of these also had activation of the MAPK pathway (13). These two pathways are known drivers of tumor growth in other cancers (130, 136, 137). Therefore, it is important that a mouse model for angiosarcoma demonstrates activation of these pathways. Indeed, we have shown that our murine model does, in fact, display activation of both of these pathways, due to the inactivation of Pten and Ptpn12 in our model. Therefore, as our mouse model displays elevation of the two pathways most often implicated in angiosarcoma, it represents a model for angiosarcoma that resembles human disease. As our murine model closely recapitulates human disease histologically, displays similar pathway activation of the mTOR and MAPK pathways, and develops angiosarcoma with high penetrance and short latency, it represents a good model 138 for future study of the disease as well as therapeutic strategies. Therefore, the similarities with human disease should be further characterized. RNA-sequencing should be performed on endothelial cell isolates, human angiosarcoma patient samples, and our murine angiosarcoma samples to analyze gene expression levels. This would give us several key pieces of information. It would tell us how similar the gene expression levels in our mouse model are to human samples, allowing us to better assess how similar our mouse model is to human disease. It would also allow us to determine which pathways are most important for angiosarcoma development by finding pathways that are elevated or decreased in both human and murine angiosarcoma compared with normal endothelial cells. This information may provide novel pathways to pursue for therapeutic targeting. Although our murine model closely mimics human disease, it is not a perfect model when current knowledge of angiosarcoma biology is considered. Our model uses a GFAP-CreER driver, which we find targets a small subset of endothelial cells but is traditionally used to target glial cells in the brain. We would like to test other cre drivers that are better-characterized to target cells of the endothelium, the putative cell of origin for angiosarcoma, such as the SCL-CreER driver (211). We have also used deletion of Pten and Ptpn12, two genes that have not been shown to be mutated in human disease. However, levels of both proteins are decreased within the tumors compared with normal endothelial controls; we have therefore used gene deletion to attain the same effect as currently this is the best way to decrease protein expression in a murine model (our data and (37)). As we were conducting our experiments, new data came to light due to a sequencing effort 139 by Behjati et al. This data found novel mutations in several genes in angiosarcoma, namely PTPRB and PLCG1 (36). This led us to propose that in our mouse model Ptpn12 is playing a role similar to that of PTPRB in human disease. Given this new data, mouse models should be created to explore the roles of PTPRB and PLCG1 in angiosarcoma. To interrogate PTPRB in angiosarcoma, we would like to create a Ptprbfl/fl model (as germline deletion results in embryonic lethality) and combine it with the Trp53 and/or Pten deletion as we did with Ptpn12 to compare it with our current model (58). A mouse model combining Ptprb deletion with Trp53 deletion may provide the more biologically relevant model of angiosarcoma as these mutations often co-occur in human patient samples (36). The role of activating mutations in PLCG1 should also be explored in a similar manner. Further, mutations in PTEN do not often occur in angiosarcoma (rather, protein expression of PTEN is often decreased); therefore, a model closer to human disease would utilize activating mutation of Pik3ca as this gene is often mutated in patient samples (36). To create these genetic changes we would like to use the newer, more time-efficient, and cost-effective method CRISPR/CAS9, a bacterial system that can be used to create site-directed double-stranded breaks in DNA into which a construct containing loxP sites and specific mutations can be inserted (212-214). The involvement of these genetic mutations in angiosarcoma could also be tested using in vitro systems such as endothelial colony forming cells or HUVECs to determine whether they are capable of transforming cells. Using cell culture we could knockdown or overexpress the genes that are mutated in patient samples followed by injection into mice to determine whether they are capable of forming 140 tumors. Alternatively, colony-forming assays can be performed in vitro with HUVECs. These experiments can be performed more rapidly than development of novel genetically engineered mouse models. These genetic mutations could result in tumors that are more like human disease as they utilize the actual mutations found in patient samples. We can use these tools to better assess the roles of these genes in angiosarcoma; we can determine whether mutation of one of these genes is sufficient to transform endothelial cells or form angiosarcomas. Creation of other mouse models is important to explore the necessity of activation of the mTOR and MAPK pathways for the formation of angiosarcoma and to further test our treatment strategy of combined inhibition of these pathways to determine whether this treatment results in tumor regression in other models. Development of Cell Culture System to Study Role of PTPN12 As we expect PTPN12 is playing an important role in these tumors, we would like to better define its role in this context. We have discussed previously that the functions of PTPN12 appear to be context-dependent; in normal cells loss of PTPN12 often results in decreased migration while in tumor cells this loss often results in an increase in migration. Development of a cell culture from the murine angiosarcomas would provide a good way to study the effect of PTPN12 on migration. PTPN12 could be re-expressed in the murine angiosarcoma cells; the migration of the PTPN12-/- tumor cells could then be compared with the cells with exogenously expressed PTPN12. Other studies could be done using these cells to 141 determine the substrates of PTPN12. We could compare phosphorylation levels of known PTPN12 substrates between the PTPN12-/- cells and cells with PTPN12 re- expressed; PTPN12 substrates should be more phosphorylated in the PTPN12-/- cells. We could also perform immunoprecipitations using an anti-phosphotyrosine antibody in both cell culture systems (PTPN12-/- and PTPN12 re-expressed). Mass spectrometry could then be performed on the bands that are stronger in the PTPN12-/- cells to potentially find new targets of PTPN12. We therefore set out to culture our murine angiosarcoma cells in order to better explore the proteins interacting with and regulated by PTPN12 as well as the role PTPN12 is playing in migration in these tumors. We were, unfortunately, not successful in our attempts; primary mouse endothelial cells are notoriously difficult to culture (personal communications). However, valuable information was gained that will help us with future culture trials. Our final Western blot (Figure 3.6) demonstrated that our digestion condition yielded differences in protein expression compared with the original tumor sample. Therefore, our failure to culture tumor cells may be due to digestion conditions. We should begin by attempting digestion using a solely mechanical digestion. We should also try using a larger filter size (perhaps 100um) following digestion as we could be selecting out the larger, proliferative tumor cells by using a 70um filter. These conditions may result in better selection for tumor cells that are viable for cell culture. We would like to continue our cell culture trials using a fibroblast feeder layer, often used for culturing primary cells. This may be important for our tumor 142 cells as our tumors have many stromal cells throughout the tumors and we had difficulty in ridding our culture of fibroblasts. These cells may be important for the growth of the tumor cells, either through secretion of important growth factors or simply through contact with the tumor cells. Finally, there are other cell culture systems that we could use to study PTPN12. However, as the role of PTPN12 is context-dependent, they may not be the perfect system. There are two human angiosarcoma cell lines described in the literature; it may be possible to obtain one of these. However, it is at the discretion of the laboratories that developed them. Only one of these cell lines develops tumors when implanted into mice; this would be the more ideal cell line to obtain for these studies (46, 47). An immortalized cell line capable of forming tumors in mice may be used, potentially human dermal microvessel endothelial cells. Another possibility is to use endothelial colony forming cells, which are isolated circulating endothelial cells that are progenitor-like (203). Finally, the ATCC offers a sarcoma panel that could be used for these studies. However, angiosarcoma is not among the sarcomas in this panel, therefore, it is not the ideal system in which to test the role of PTPN12. Finding a viable cell culture option is necessary to determine the role of PTPN12 in angiosarcoma as it is important for the development of the murine tumors. Determining the substrates of PTPN12 in this context may help provide new therapeutic targets for a disease that sorely needs novel therapies. Inhibition of mTOR and MEK is an Effective Therapy for Murine Angiosarcoma 143 We find that the mTOR and MAPK pathways are elevated in these tumors compared with endothelial cell isolates and lung controls, as they are in human patient samples (13, 32, 36). We know that Pten loss results in activation of the mTOR pathway and we propose that Ptpn12 loss is resulting in activation of the MAPK pathway, as it has been shown to do previously (80). We sought to determine whether these pathways were important for tumor growth and whether inhibition of one or both could provide a viable therapeutic option for these patients. Interestingly, mTOR inhibitors have been tested with some efficacy in angiosarcoma patients (40-42). MEK inhibitors have proved efficacious in canine angiosarcoma cells as well, either alone or in combination with mTOR inbihitors (38, 39). However, there are very few studies performed on this particular patient population; the clinical trials are often for all sarcoma types and typically have low- enrollment of angiosarcoma patients. In one study, however, use of everolimus resulted in the best responses in 2/3 of the angiosarcoma patients (41). One pediatric patient with a RAS mutation was treated with a combination of mTOR and MEK inhibitors with little effect; however, this treatment was not attempted until after failure of multiple other therapies (8). Research has shown that inhibition of one pathway often results in resistance due to activation of the other pathway (157- 166). Therefore, inhibiting both pathways simultaneously could result in the best responses. However, trials with inhibitors of both pathways often results in toxicities (167, 168). Research is needed to determine an effective combination of drugs to inhibit both pathways. 144 Further exploration is necessary into whether this treatment could be effective for angiosarcoma patients with activation of the mTOR and MAPK pathways. To test the effectiveness of inhibition of these pathways in our mouse model we used targeted inhibitors against mTOR (rapamycin) and MEK (trametinib). We find that use of either drug alone does not result in sustainable tumor suppression; however, use of both drugs simultaneously does result in regression of the tumors that is sustainable for the duration of our treatment (140 days). Initially either drug alone was capable of causing tumor regression; however, eventually most of the tumors became resistant (Figure 4.4B). We have evaluated mechanisms of resistance such as STAT3 activation (seen in trametinib resistance) and an increase in the stem cell population (205, 209). Neither of these mechanisms appear to be the case as STAT3 phosphorylation is not elevated in resistant tumors compared with vehicle control and there appear to be no stem cells present within the tumors as evidenced by SOX2 staining. Therefore, we propose the resistance seen is due to the increase in activation of the reciprocal pathway as is seen in our five day treatments (Figure 4.3A). Further testing of this could be done by allowing the tumors to reach resistance on monotherapy and then adding the other drug to determine whether this can rescue the resistance. Dual therapy with both rapamycin and trametinib resulted in sustainable tumor regression until the end of study at 140 days. However, when the animals were removed from study, their tumors began to grow back, indicating that residual disease is present. Future studies will define this population of cells in order to develop a treatment regimen capable of killing all tumor cell populations within the 145 tumor for best disease regression. Another issue with dual treatment was toxicity; although the tumors regressed and did not regrow, this effect did not translate into a survival advantage due to loss of mice from drug toxicity. We even tried to decrease the dose of both drugs from 1mg/kg to 0.5mg/kg but still witnessed toxicity with this treatment. Therefore, we propose that use of different drugs may be able to inhibit the pathways while decreasing toxicity, preferably other rapalogs as they are more specific for targeting mTOR. However, ATP-competitive mTOR inhibitors could be tested as well. Other studies have demonstrated efficacy of mTOR and MAPK inhibitors for the treatment of cancers in mice; however, these treatments were not every day and did not last as long as our study. We could also try changing our dosing schedule to determine whether this would be more tolerable. It is important to find a drug combination or dosing schedule that is tolerated as we have shown that dual therapy is very effective for treatment of angiosarcoma. 146 Summary This dissertation has characterized a novel mouse model for angiosarcoma. We have shown that this model is representative of human disease. We describe attempts to culture these cells in order to delineate the role of PTPN12 in these tumors as loss of PTPN12 is critical for angiosarcoma formation in our model. We further define two pathways that we expect to be elevated based on the genes we have deleted. We utilized targeted therapeutics to demonstrate that these pathways are, in fact, important for tumor growth. These pathways are elevated in human angiosarcoma; therefore, inhibition of these two pathways may provide a novel therapeutic intervention for people with angiosarcoma. Acknowledgements We would like to acknowledge our funding sources, Children’s Hospital Research Foundation, Cancer-Free Kids, Sarcoma Foundation of America, and Canadian Institutes of Health Research MOP-133679. We would also like to thank Jillian Goines in the Boscolo lab for her help trouble shooting techniques and for lending us reagents. 147 References 1. Ravi V, and Patel S. Vascular sarcomas. Current oncology reports. 2013;15(4):347-55. 2. Young RJ, Brown NJ, Reed MW, Hughes D, and Woll PJ. Angiosarcoma. The Lancet Oncology. 2010;11(10):983-91. 3. Fayette J, Martin E, Piperno-Neumann S, Le Cesne A, Robert C, Bonvalot S, Ranchere D, Pouillart P, Coindre JM, and Blay JY. Angiosarcomas, a heterogeneous group of sarcomas with specific behavior depending on primary site: a retrospective study of 161 cases. Annals of oncology : official journal of the European Society for Medical Oncology / ESMO. 2007;18(12):2030-6. 4. Buehler D, Rice SR, Moody JS, Rush P, Hafez GR, Attia S, Longley BJ, and Kozak KR. Angiosarcoma outcomes and prognostic factors: a 25-year single institution experience. American journal of clinical oncology. 2014;37(5):473- 9. 5. Mark RJ, Poen JC, Tran LM, Fu YS, and Juillard GF. Angiosarcoma. A report of 67 patients and a review of the literature. Cancer. 1996;77(11):2400-6. 6. Fury MG, Antonescu CR, Van Zee KJ, Brennan MF, and Maki RG. A 14-year retrospective review of angiosarcoma: clinical characteristics, prognostic factors, and treatment outcomes with surgery and chemotherapy. Cancer journal. 2005;11(3):241-7. 7. Ferrari A, Casanova M, Bisogno G, Cecchetto G, Meazza C, Gandola L, Garaventa A, Mattke A, Treuner J, and Carli M. Malignant vascular tumors in children and adolescents: a report from the Italian and German Soft Tissue Sarcoma Cooperative Group. Medical and pediatric oncology. 2002;39(2):109- 14. 8. Jeng MR, Fuh B, Blatt J, Gupta A, Merrow AC, Hammill A, and Adams D. Malignant transformation of infantile hemangioma to angiosarcoma: response to chemotherapy with bevacizumab. Pediatric blood & cancer. 2014;61(11):2115-7. 9. Fujii T, Zen Y, Sato Y, Sasaki M, Enomae M, Minato H, Masuda S, Uehara T, Katsuyama T, and Nakanuma Y. Podoplanin is a useful diagnostic marker for epithelioid hemangioendothelioma of the liver. Modern pathology : an official journal of the United States and Canadian Academy of Pathology, Inc. 2008;21(2):125-30. 10. Antonescu C. Malignant vascular tumors--an update. Modern pathology : an official journal of the United States and Canadian Academy of Pathology, Inc. 2014;27 Suppl 1(S30-8. 11. Miettinen M, and Wang ZF. Prox1 transcription factor as a marker for vascular tumors-evaluation of 314 vascular endothelial and 1086 nonvascular tumors. The American journal of surgical pathology. 2012;36(3):351-9. 12. Guadagnolo BA, Zagars GK, Araujo D, Ravi V, Shellenberger TD, and Sturgis EM. Outcomes after definitive treatment for cutaneous angiosarcoma of the face and scalp. Head & neck. 2011;33(5):661-7. 148 13. Italiano A, Chen CL, Thomas R, Breen M, Bonnet F, Sevenet N, Longy M, Maki RG, Coindre JM, and Antonescu CR. Alterations of the p53 and PIK3CA/AKT/mTOR pathways in angiosarcomas: a pattern distinct from other sarcomas with complex genomics. Cancer. 2012;118(23):5878-87. 14. Nagano T, Yamada Y, Ikeda T, Kanki H, Kamo T, and Nishigori C. Docetaxel: a therapeutic option in the treatment of cutaneous angiosarcoma: report of 9 patients. Cancer. 2007;110(3):648-51. 15. Schlemmer M, Reichardt P, Verweij J, Hartmann JT, Judson I, Thyss A, Hogendoorn PC, Marreaud S, Van Glabbeke M, and Blay JY. Paclitaxel in patients with advanced angiosarcomas of soft tissue: a retrospective study of the EORTC soft tissue and bone sarcoma group. European journal of cancer. 2008;44(16):2433-6. 16. Fata F, O'Reilly E, Ilson D, Pfister D, Leffel D, Kelsen DP, Schwartz GK, and Casper ES. Paclitaxel in the treatment of patients with angiosarcoma of the scalp or face. Cancer. 1999;86(10):2034-7. 17. Perez MC, Padhya TA, Messina JL, Jackson RS, Gonzalez RJ, Bui MM, Letson GD, Cruse CW, Lavey RS, Cheong D, et al. Cutaneous angiosarcoma: a single- institution experience. Annals of surgical oncology. 2013;20(11):3391-7. 18. Itakura E, Yamamoto H, Oda Y, and Tsuneyoshi M. Detection and characterization of vascular endothelial growth factors and their receptors in a series of angiosarcomas. Journal of surgical oncology. 2008;97(1):74-81. 19. Agulnik M, Yarber JL, Okuno SH, von Mehren M, Jovanovic BD, Brockstein BE, Evens AM, and Benjamin RS. An open-label, multicenter, phase II study of bevacizumab for the treatment of angiosarcoma and epithelioid hemangioendotheliomas. Annals of oncology : official journal of the European Society for Medical Oncology / ESMO. 2013;24(1):257-63. 20. Maki RG, D'Adamo DR, Keohan ML, Saulle M, Schuetze SM, Undevia SD, Livingston MB, Cooney MM, Hensley ML, Mita MM, et al. Phase II study of sorafenib in patients with metastatic or recurrent sarcomas. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2009;27(19):3133-40. 21. Ray-Coquard I, Italiano A, Bompas E, Le Cesne A, Robin YM, Chevreau C, Bay JO, Bousquet G, Piperno-Neumann S, Isambert N, et al. Sorafenib for patients with advanced angiosarcoma: a phase II Trial from the French Sarcoma Group (GSF/GETO). The oncologist. 2012;17(2):260-6. 22. Ray-Coquard IL, Domont J, Tresch-Bruneel E, Bompas E, Cassier PA, Mir O, Piperno-Neumann S, Italiano A, Chevreau C, Cupissol D, et al. Paclitaxel Given Once Per Week With or Without Bevacizumab in Patients With Advanced Angiosarcoma: A Randomized Phase II Trial. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2015;33(25):2797-802. 23. Penel N, Bui BN, Bay JO, Cupissol D, Ray-Coquard I, Piperno-Neumann S, Kerbrat P, Fournier C, Taieb S, Jimenez M, et al. Phase II trial of weekly paclitaxel for unresectable angiosarcoma: the ANGIOTAX Study. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2008;26(32):5269-74. 149 24. Yudoh K, Kanamori M, Ohmori K, Yasuda T, Aoki M, and Kimura T. Concentration of vascular endothelial growth factor in the tumour tissue as a prognostic factor of soft tissue sarcomas. British journal of cancer. 2001;84(12):1610-5. 25. Mercurio AM, Bachelder RE, Bates RC, and Chung J. Autocrine signaling in carcinoma: VEGF and the alpha6beta4 integrin. Seminars in cancer biology. 2004;14(2):115-22. 26. Vogelstein B, and Kinzler KW. Cancer genes and the pathways they control. Nature medicine. 2004;10(8):789-99. 27. Vogelstein B, Lane D, and Levine AJ. Surfing the p53 network. Nature. 2000;408(6810):307-10. 28. Vousden KH, and Lane DP. p53 in health and disease. Nature reviews Molecular cell biology. 2007;8(4):275-83. 29. Hinds PW, Finlay CA, Quartin RS, Baker SJ, Fearon ER, Vogelstein B, and Levine AJ. Mutant p53 DNA clones from human colon carcinomas cooperate with ras in transforming primary rat cells: a comparison of the "hot spot" mutant phenotypes. Cell growth & differentiation : the molecular biology journal of the American Association for Cancer Research. 1990;1(12):571-80. 30. Reich NC, Oren M, and Levine AJ. Two distinct mechanisms regulate the levels of a cellular tumor antigen, p53. Molecular and cellular biology. 1983;3(12):2143-50. 31. Oren M, Maltzman W, and Levine AJ. Post-translational regulation of the 54K cellular tumor antigen in normal and transformed cells. Molecular and cellular biology. 1981;1(2):101-10. 32. Murali R, Chandramohan R, Moller I, Scholz SL, Berger M, Huberman K, Viale A, Pirun M, Socci ND, Bouvier N, et al. Targeted massively parallel sequencing of angiosarcomas reveals frequent activation of the mitogen activated protein kinase pathway. Oncotarget. 2015;6(34):36041-52. 33. Weihrauch M, Markwarth A, Lehnert G, Wittekind C, Wrbitzky R, and Tannapfel A. Abnormalities of the ARF-p53 pathway in primary angiosarcomas of the liver. Human pathology. 2002;33(9):884-92. 34. Mendoza MC, Er EE, and Blenis J. The Ras-ERK and PI3K-mTOR pathways: cross-talk and compensation. Trends in biochemical sciences. 2011;36(6):320- 8. 35. Du W, Gerald D, Perruzzi CA, Rodriguez-Waitkus P, Enayati L, Krishnan B, Edmonds J, Hochman ML, Lev DC, and Phung TL. Vascular tumors have increased p70 S6-kinase activation and are inhibited by topical rapamycin. Lab Invest. 2013;93(10):1115-27. 36. Behjati S, Tarpey PS, Sheldon H, Martincorena I, Van Loo P, Gundem G, Wedge DC, Ramakrishna M, Cooke SL, Pillay N, et al. Recurrent PTPRB and PLCG1 mutations in angiosarcoma. Nature genetics. 2014;46(4):376-9. 37. Verbeke SL, Bertoni F, Bacchini P, Oosting J, Sciot R, Krenacs T, and Bovee JV. Active TGF-beta signaling and decreased expression of PTEN separates angiosarcoma of bone from its soft tissue counterpart. Modern pathology : an official journal of the United States and Canadian Academy of Pathology, Inc. 2013;26(9):1211-21. 150 38. Andersen NJ, Boguslawski EB, Kuk CY, Chambers CM, and Duesbery NS. Combined inhibition of MEK and mTOR has a synergic effect on angiosarcoma tumorgrafts. International journal of oncology. 2015;47(1):71- 80. 39. Andersen NJ, Nickoloff BJ, Dykema KJ, Boguslawski EA, Krivochenitser RI, Froman RE, Dawes MJ, Baker LH, Thomas DG, Kamstock DA, et al. Pharmacologic inhibition of MEK signaling prevents growth of canine hemangiosarcoma. Molecular cancer therapeutics. 2013;12(9):1701-14. 40. Wada M, Horinaka M, Yasuda S, Masuzawa M, Sakai T, and Katoh N. PDK1 is a potential therapeutic target against angiosarcoma cells. Journal of dermatological science. 2015;78(1):44-50. 41. Yoo C, Lee J, Rha SY, Park KH, Kim TM, Kim YJ, Lee HJ, Lee KH, and Ahn JH. Multicenter phase II study of everolimus in patients with metastatic or recurrent bone and soft-tissue sarcomas after failure of anthracycline and ifosfamide. Investigational new drugs. 2013;31(6):1602-8. 42. Bundscherer A, Vogt T, Kohl G, Landthaler M, and Hafner C. Antiproliferative effects of rapamycin and celecoxib in angiosarcoma cell lines. Anticancer research. 2010;30(10):4017-23. 43. Arbiser JL, Moses MA, Fernandez CA, Ghiso N, Cao Y, Klauber N, Frank D, Brownlee M, Flynn E, Parangi S, et al. Oncogenic H-ras stimulates tumor angiogenesis by two distinct pathways. Proceedings of the National Academy of Sciences of the United States of America. 1997;94(3):861-6. 44. Richmond A, and Su Y. Mouse xenograft models vs GEM models for human cancer therapeutics. Disease models & mechanisms. 2008;1(2-3):78-82. 45. Dranoff G. Cytokines in cancer pathogenesis and cancer therapy. Nature reviews Cancer. 2004;4(1):11-22. 46. Masuzawa M, Fujimura T, Hamada Y, Fujita Y, Hara H, Nishiyama S, Katsuoka K, Tamauchi H, and Sakurai Y. Establishment of a human hemangiosarcoma cell line (ISO-HAS). International journal of cancer. 1999;81(2):305-8. 47. Krump-Konvalinkova V, Bittinger F, Olert J, Brauninger W, Brunner J, and Kirkpatrick CJ. Establishment and characterization of an angiosarcoma- derived cell line, AS-M. Endothelium : journal of endothelial cell research. 2003;10(6):319-28. 48. Neve RM, Chin K, Fridlyand J, Yeh J, Baehner FL, Fevr T, Clark L, Bayani N, Coppe JP, Tong F, et al. A collection of breast cancer cell lines for the study of functionally distinct cancer subtypes. Cancer Cell. 2006;10(6):515-27. 49. Domcke S, Sinha R, Levine DA, Sander C, and Schultz N. Evaluating cell lines as tumour models by comparison of genomic profiles. Nature communications. 2013;4(2126. 50. Li H, Wawrose JS, Gooding WE, Garraway LA, Lui VW, Peyser ND, and Grandis JR. Genomic analysis of head and neck squamous cell carcinoma cell lines and human tumors: a rational approach to preclinical model selection. Molecular cancer research : MCR. 2014;12(4):571-82. 51. McMillin DW, Delmore J, Weisberg E, Negri JM, Geer DC, Klippel S, Mitsiades N, Schlossman RL, Munshi NC, Kung AL, et al. Tumor cell-specific 151 bioluminescence platform to identify stroma-induced changes to anticancer drug activity. Nature medicine. 2010;16(4):483-9. 52. Ding L, Ellis MJ, Li S, Larson DE, Chen K, Wallis JW, Harris CC, McLellan MD, Fulton RS, Fulton LL, et al. Genome remodelling in a basal-like breast cancer metastasis and xenograft. Nature. 2010;464(7291):999-1005. 53. Morton CL, and Houghton PJ. Establishment of human tumor xenografts in immunodeficient mice. Nature protocols. 2007;2(2):247-50. 54. Yang J, Kantrow S, Sai J, Hawkins OE, Boothby M, Ayers GD, Young ED, Demicco EG, Lazar AJ, Lev D, et al. INK4a/ARF [corrected] inactivation with activation of the NF-kappaB/IL-6 pathway is sufficient to drive the development and growth of angiosarcoma. Cancer Res. 2012;72(18):4682-95. 55. Serrano M, Lee H, Chin L, Cordon-Cardo C, Beach D, and DePinho RA. Role of the INK4a locus in tumor suppression and cell mortality. Cell. 1996;85(1):27- 37. 56. Sharpless NE, Ramsey MR, Balasubramanian P, Castrillon DH, and DePinho RA. The differential impact of p16(INK4a) or p19(ARF) deficiency on cell growth and tumorigenesis. Oncogene. 2004;23(2):379-85. 57. Winderlich M, Keller L, Cagna G, Broermann A, Kamenyeva O, Kiefer F, Deutsch U, Nottebaum AF, and Vestweber D. VE-PTP controls blood vessel development by balancing Tie-2 activity. The Journal of cell biology. 2009;185(4):657-71. 58. Dominguez MG, Hughes VC, Pan L, Simmons M, Daly C, Anderson K, Noguera- Troise I, Murphy AJ, Valenzuela DM, Davis S, et al. Vascular endothelial tyrosine phosphatase (VE-PTP)-null mice undergo vasculogenesis but die embryonically because of defects in angiogenesis. Proceedings of the National Academy of Sciences of the United States of America. 2007;104(9):3243-8. 59. Barford D, Das AK, and Egloff MP. The structure and mechanism of protein phosphatases: insights into catalysis and regulation. Annual review of biophysics and biomolecular structure. 1998;27(133-64. 60. Stebbing J, Lit LC, Zhang H, Darrington RS, Melaiu O, Rudraraju B, and Giamas G. The regulatory roles of phosphatases in cancer. Oncogene. 2014;33(8):939-53. 61. Ritz A, Paris PL, Ittmann MM, Collins C, and Raphael BJ. Detection of recurrent rearrangement breakpoints from copy number data. BMC bioinformatics. 2011;12(114. 62. Motiwala T, and Jacob ST. Role of protein tyrosine phosphatases in cancer. Progress in nucleic acid research and molecular biology. 2006;81(297-329. 63. Liu W, Yu B, Xu G, Xu WR, Loh ML, Tang LD, and Qu CK. Identification of cryptotanshinone as an inhibitor of oncogenic protein tyrosine phosphatase SHP2 (PTPN11). Journal of medicinal chemistry. 2013;56(18):7212-21. 64. Chen C, Liang F, Chen B, Sun Z, Xue T, Yang R, and Luo D. Identification of demethylincisterol A3 as a selective inhibitor of protein tyrosine phosphatase Shp2. European journal of pharmacology. 2017;795(124-33. 65. Chen YN, LaMarche MJ, Chan HM, Fekkes P, Garcia-Fortanet J, Acker MG, Antonakos B, Chen CH, Chen Z, Cooke VG, et al. Allosteric inhibition of SHP2 152 phosphatase inhibits cancers driven by receptor tyrosine kinases. Nature. 2016;535(7610):148-52. 66. Saba HI. Decitabine in the treatment of myelodysplastic syndromes. Therapeutics and clinical risk management. 2007;3(5):807-17. 67. Yang Q, Co D, Sommercorn J, and Tonks NK. Cloning and expression of PTP- PEST. A novel, human, nontransmembrane protein tyrosine phosphatase. The Journal of biological chemistry. 1993;268(9):6622-8. 68. Charest A, Wagner J, Shen SH, and Tremblay ML. Murine protein tyrosine phosphatase-PEST, a stable cytosolic protein tyrosine phosphatase. The Biochemical journal. 1995;308 ( Pt 2)(425-32. 69. Davidson D, Cloutier JF, Gregorieff A, and Veillette A. Inhibitory tyrosine protein kinase p50csk is associated with protein-tyrosine phosphatase PTP- PEST in hemopoietic and non-hemopoietic cells. The Journal of biological chemistry. 1997;272(37):23455-62. 70. Davidson D, and Veillette A. PTP-PEST, a scaffold protein tyrosine phosphatase, negatively regulates lymphocyte activation by targeting a unique set of substrates. The EMBO journal. 2001;20(13):3414-26. 71. Rogers S, Wells R, and Rechsteiner M. Amino acid sequences common to rapidly degraded proteins: the PEST hypothesis. Science. 1986;234(4774):364-8. 72. Charest A, Wagner J, Kwan M, and Tremblay ML. Coupling of the murine protein tyrosine phosphatase PEST to the epidermal growth factor (EGF) receptor through a Src homology 3 (SH3) domain-mediated association with Grb2. Oncogene. 1997;14(14):1643-51. 73. Garton AJ, Flint AJ, and Tonks NK. Identification of p130(cas) as a substrate for the cytosolic protein tyrosine phosphatase PTP-PEST. Molecular and cellular biology. 1996;16(11):6408-18. 74. Charest A, Wagner J, Jacob S, McGlade CJ, and Tremblay ML. Phosphotyrosine-independent binding of SHC to the NPLH sequence of murine protein-tyrosine phosphatase-PEST. Evidence for extended phosphotyrosine binding/phosphotyrosine interaction domain recognition specificity. The Journal of biological chemistry. 1996;271(14):8424-9. 75. Markova B, Herrlich P, Ronnstrand L, and Bohmer FD. Identification of protein tyrosine phosphatases associating with the PDGF receptor. Biochemistry. 2003;42(9):2691-9. 76. Shen Y, Schneider G, Cloutier JF, Veillette A, and Schaller MD. Direct association of protein-tyrosine phosphatase PTP-PEST with paxillin. The Journal of biological chemistry. 1998;273(11):6474-81. 77. Angers-Loustau A, Cote JF, Charest A, Dowbenko D, Spencer S, Lasky LA, and Tremblay ML. Protein tyrosine phosphatase-PEST regulates focal adhesion disassembly, migration, and cytokinesis in fibroblasts. The Journal of cell biology. 1999;144(5):1019-31. 78. Garton AJ, Burnham MR, Bouton AH, and Tonks NK. Association of PTP-PEST with the SH3 domain of p130cas; a novel mechanism of protein tyrosine phosphatase substrate recognition. Oncogene. 1997;15(8):877-85. 153 79. Shen Y, Lyons P, Cooley M, Davidson D, Veillette A, Salgia R, Griffin JD, and Schaller MD. The noncatalytic domain of protein-tyrosine phosphatase-PEST targets paxillin for dephosphorylation in vivo. The Journal of biological chemistry. 2000;275(2):1405-13. 80. Sun T, Aceto N, Meerbrey KL, Kessler JD, Zhou C, Migliaccio I, Nguyen DX, Pavlova NN, Botero M, Huang J, et al. Activation of multiple proto-oncogenic tyrosine kinases in breast cancer via loss of the PTPN12 phosphatase. Cell. 2011;144(5):703-18. 81. Faisal A, el-Shemerly M, Hess D, and Nagamine Y. Serine/threonine phosphorylation of ShcA. Regulation of protein-tyrosine phosphatase-pest binding and involvement in insulin signaling. The Journal of biological chemistry. 2002;277(33):30144-52. 82. Dunn KL, Espino PS, Drobic B, He S, and Davie JR. The Ras-MAPK signal transduction pathway, cancer and chromatin remodeling. Biochemistry and cell biology = Biochimie et biologie cellulaire. 2005;83(1):1-14. 83. Yoon S, and Seger R. The extracellular signal-regulated kinase: multiple substrates regulate diverse cellular functions. Growth factors. 2006;24(1):21- 44. 84. Parsons JT. Focal adhesion kinase: the first ten years. Journal of cell science. 2003;116(Pt 8):1409-16. 85. Richardson A, and Parsons T. A mechanism for regulation of the adhesion- associated proteintyrosine kinase pp125FAK. Nature. 1996;380(6574):538- 40. 86. Wade R, Bohl J, and Vande Pol S. Paxillin null embryonic stem cells are impaired in cell spreading and tyrosine phosphorylation of focal adhesion kinase. Oncogene. 2002;21(1):96-107. 87. Zheng Y, Yang W, Xia Y, Hawke D, Liu DX, and Lu Z. Ras-induced and extracellular signal-regulated kinase 1 and 2 phosphorylation-dependent isomerization of protein tyrosine phosphatase (PTP)-PEST by PIN1 promotes FAK dephosphorylation by PTP-PEST. Molecular and cellular biology. 2011;31(21):4258-69. 88. Lyons PD, Dunty JM, Schaefer EM, and Schaller MD. Inhibition of the catalytic activity of cell adhesion kinase beta by protein-tyrosine phosphatase-PEST- mediated dephosphorylation. The Journal of biological chemistry. 2001;276(26):24422-31. 89. Davidson D, Shi X, Zhong MC, Rhee I, and Veillette A. The Phosphatase PTP- PEST Promotes Secondary T Cell Responses by Dephosphorylating the Protein Tyrosine Kinase Pyk2. Immunity. 2010;33(2):167-80. 90. Cote JF, Turner CE, and Tremblay ML. Intact LIM 3 and LIM 4 domains of paxillin are required for the association to a novel polyproline region (Pro 2) of protein-tyrosine phosphatase-PEST. The Journal of biological chemistry. 1999;274(29):20550-60. 91. Nishiya N, Iwabuchi Y, Shibanuma M, Cote JF, Tremblay ML, and Nose K. Hic- 5, a paxillin homologue, binds to the protein-tyrosine phosphatase PEST (PTP-PEST) through its LIM 3 domain. The Journal of biological chemistry. 1999;274(14):9847-53. 154 92. Sahu SN, Nunez S, Bai G, and Gupta A. Interaction of Pyk2 and PTP-PEST with leupaxin in prostate cancer cells. American journal of physiology Cell physiology. 2007;292(6):C2288-96. 93. Cote JF, Chung PL, Theberge JF, Halle M, Spencer S, Lasky LA, and Tremblay ML. PSTPIP is a substrate of PTP-PEST and serves as a scaffold guiding PTP- PEST toward a specific dephosphorylation of WASP. The Journal of biological chemistry. 2002;277(4):2973-86. 94. Playford MP, Lyons PD, Sastry SK, and Schaller MD. Identification of a filamin docking site on PTP-PEST. The Journal of biological chemistry. 2006;281(45):34104-12. 95. Guan JL. Role of focal adhesion kinase in integrin signaling. The international journal of biochemistry & cell biology. 1997;29(8-9):1085-96. 96. Cote JF, Charest A, Wagner J, and Tremblay ML. Combination of gene targeting and substrate trapping to identify substrates of protein tyrosine phosphatases using PTP-PEST as a model. Biochemistry. 1998;37(38):13128- 37. 97. Jamieson JS, Tumbarello DA, Halle M, Brown MC, Tremblay ML, and Turner CE. Paxillin is essential for PTP-PEST-dependent regulation of cell spreading and motility: a role for paxillin kinase linker. Journal of cell science. 2005;118(Pt 24):5835-47. 98. Wimmer-Kleikamp SH, Nievergall E, Gegenbauer K, Adikari S, Mansour M, Yeadon T, Boyd AW, Patani NR, and Lackmann M. Elevated protein tyrosine phosphatase activity provokes Eph/ephrin-facilitated adhesion of pre-B leukemia cells. Blood. 2008;112(3):721-32. 99. Mansour M, Nievergall E, Gegenbauer K, Llerena C, Atapattu L, Halle M, Tremblay ML, Janes PW, and Lackmann M. PTP-PEST controls EphA3 activation and ephrin-induced cytoskeletal remodelling. Journal of cell science. 2016;129(2):277-89. 100. Nievergall E, Lackmann M, and Janes PW. Eph-dependent cell-cell adhesion and segregation in development and cancer. Cellular and molecular life sciences : CMLS. 2012;69(11):1813-42. 101. Ayoub E, Hall A, Scott AM, Chagnon MJ, Miquel G, Halle M, Noda M, Bikfalvi A, and Tremblay ML. Regulation of the Src kinase-associated phosphoprotein 55 homologue by the protein tyrosine phosphatase PTP-PEST in the control of cell motility. The Journal of biological chemistry. 2013;288(36):25739-48. 102. Garton AJ, and Tonks NK. Regulation of fibroblast motility by the protein tyrosine phosphatase PTP-PEST. The Journal of biological chemistry. 1999;274(6):3811-8. 103. Sastry SK, Lyons PD, Schaller MD, and Burridge K. PTP-PEST controls motility through regulation of Rac1. Journal of cell science. 2002;115(Pt 22):4305-16. 104. Sastry SK, Rajfur Z, Liu BP, Cote JF, Tremblay ML, and Burridge K. PTP-PEST couples membrane protrusion and tail retraction via VAV2 and p190RhoGAP. The Journal of biological chemistry. 2006;281(17):11627-36. 105. Lee HS, Cheerathodi M, Chaki SP, Reyes SB, Zheng Y, Lu Z, Paidassi H, DerMardirossian C, Lacy-Hulbert A, Rivera GM, et al. Protein tyrosine phosphatase-PEST and beta8 integrin regulate spatiotemporal patterns of 155 RhoGDI1 activation in migrating cells. Molecular and cellular biology. 2015;35(8):1401-13. 106. Garton AJ, and Tonks NK. PTP-PEST: a protein tyrosine phosphatase regulated by serine phosphorylation. The EMBO journal. 1994;13(16):3763- 71. 107. Nakamura K, Palmer HE, Ozawa T, and Mashima K. Protein phosphatase 1alpha associates with protein tyrosine phosphatase-PEST inducing dephosphorylation of phospho-serine 39. Journal of biochemistry. 2010;147(4):493-500. 108. Thummuri D, Kumar S, Surapaneni SK, and Tikoo K. Epigenetic regulation of protein tyrosine phosphatase PTPN12 in triple-negative breast cancer. Life sciences. 2015;130(73-80. 109. Xunyi Y, Zhentao Y, Dandan J, and Funian L. Clinicopathological significance of PTPN12 expression in human breast cancer. Brazilian journal of medical and biological research = Revista brasileira de pesquisas medicas e biologicas. 2012;45(12):1334-40. 110. Macfarlane LA, and Murphy PR. MicroRNA: Biogenesis, Function and Role in Cancer. Current genomics. 2010;11(7):537-61. 111. Garzon R, Fabbri M, Cimmino A, Calin GA, and Croce CM. MicroRNA expression and function in cancer. Trends in molecular medicine. 2006;12(12):580-7. 112. Volinia S, Calin GA, Liu CG, Ambs S, Cimmino A, Petrocca F, Visone R, Iorio M, Roldo C, Ferracin M, et al. A microRNA expression signature of human solid tumors defines cancer gene targets. Proceedings of the National Academy of Sciences of the United States of America. 2006;103(7):2257-61. 113. Liang T, Li L, Cheng Y, Ren C, and Zhang G. MicroRNA-194 promotes the growth, migration, and invasion of ovarian carcinoma cells by targeting protein tyrosine phosphatase nonreceptor type 12. OncoTargets and therapy. 2016;9(4307-15. 114. Sirois J, Cote JF, Charest A, Uetani N, Bourdeau A, Duncan SA, Daniels E, and Tremblay ML. Essential function of PTP-PEST during mouse embryonic vascularization, mesenchyme formation, neurogenesis and early liver development. Mechanisms of development. 2006;123(12):869-80. 115. Souza CM, Davidson D, Rhee I, Gratton JP, Davis EC, and Veillette A. The phosphatase PTP-PEST/PTPN12 regulates endothelial cell migration and adhesion, but not permeability, and controls vascular development and embryonic viability. The Journal of biological chemistry. 2012;287(51):43180-90. 116. Duffy EA, Pretorius PR, Lerach S, Lohr JL, Hirsch B, Souza CM, Veillette A, and Schimmenti LA. Mosaic partial deletion of PTPN12 in a child with interrupted aortic arch type A. American journal of medical genetics Part A. 2015;167A(11):2674-83. 117. Luo RZ, Cai PQ, Li M, Fu J, Zhang ZY, Chen JW, Cao Y, Yun JP, Xie D, and Cai MY. Decreased expression of PTPN12 correlates with tumor recurrence and poor survival of patients with hepatocellular carcinoma. PLoS One. 2014;9(1):e85592. 156 118. Espejo R, Rengifo-Cam W, Schaller MD, Evers BM, and Sastry SK. PTP-PEST controls motility, adherens junction assembly, and Rho GTPase activity in colon cancer cells. American journal of physiology Cell physiology. 2010;299(2):C454-63. 119. Cao X, Li Y, Luo RZ, He LR, Yang J, Zeng MS, and Wen ZS. Tyrosine-protein phosphatase nonreceptor type 12 is a novel prognostic biomarker for esophageal squamous cell carcinoma. The Annals of thoracic surgery. 2012;93(5):1674-80. 120. Su Z, Tian H, Song HQ, Zhang R, Deng AM, and Liu HW. PTPN12 inhibits oral squamous epithelial carcinoma cell proliferation and invasion and can be used as a prognostic marker. Medical oncology. 2013;30(3):618. 121. Cao X, Chen YZ, Luo RZ, Zhang L, Zhang SL, Zeng J, Jiang YC, Han YJ, and Wen ZS. Tyrosine-protein phosphatase non-receptor type 12 expression is a good prognostic factor in resectable non-small cell lung cancer. Oncotarget. 2015;6(13):11704-13. 122. Zhang XK, Xu M, Chen JW, Zhou F, Ling YH, Zhu CM, Yun JP, Cai MY, and Luo RZ. The prognostic significance of tyrosine-protein phosphatase nonreceptor type 12 expression in nasopharyngeal carcinoma. Tumour biology : the journal of the International Society for Oncodevelopmental Biology and Medicine. 2015;36(7):5201-8. 123. Li J, Davidson D, Martins Souza C, Zhong MC, Wu N, Park M, Muller WJ, and Veillette A. Loss of PTPN12 Stimulates Progression of ErbB2-Dependent Breast Cancer by Enhancing Cell Survival, Migration, and Epithelial-to- Mesenchymal Transition. Molecular and cellular biology. 2015;35(23):4069- 82. 124. Wang YY, Liu H, Mao XY, Jin F, Ma B, Jiang JY, and Cao Y. Identifying the role of PTPN12 expression in predicting the efficacy of capecitabine to neoadjuvant chemotherapy in breast cancer treatment. European review for medical and pharmacological sciences. 2016;20(16):3400-9. 125. Wu MQ, Hu P, Gao J, Wei WD, Xiao XS, Tang HL, Li X, Ge QD, Jia WH, Liu RB, et al. Low expression of tyrosine-protein phosphatase nonreceptor type 12 is associated with lymph node metastasis and poor prognosis in operable triple-negative breast cancer. Asian Pacific journal of cancer prevention : APJCP. 2013;14(1):287-92. 126. Takekawa M, Itoh F, Hinoda Y, Adachi M, Ariyama T, Inazawa J, Imai K, and Yachi A. Chromosomal localization of the protein tyrosine phosphatase G1 gene and characterization of the aberrant transcripts in human colon cancer cells. FEBS letters. 1994;339(3):222-8. 127. de Voer RM, Hahn MM, Weren RD, Mensenkamp AR, Gilissen C, van Zelst- Stams WA, Spruijt L, Kets CM, Zhang J, Venselaar H, et al. Identification of Novel Candidate Genes for Early-Onset Colorectal Cancer Susceptibility. PLoS genetics. 2016;12(2):e1005880. 128. Broderick P, Dobbins SE, Chubb D, Kinnersley B, Dunlop MG, Tomlinson I, and Houlston RS. Validation of Recently Proposed Colorectal Cancer Susceptibility Gene Variants in an Analysis of Families and Patients-a Systematic Review. Gastroenterology. 2017;152(1):75-7 e4. 157 129. McCormick F. Signalling networks that cause cancer. Trends in cell biology. 1999;9(12):M53-6. 130. Luo J, Manning BD, and Cantley LC. Targeting the PI3K-Akt pathway in human cancer: rationale and promise. Cancer Cell. 2003;4(4):257-62. 131. Laplante M, and Sabatini DM. mTOR signaling in growth control and disease. Cell. 2012;149(2):274-93. 132. Sarbassov DD, Ali SM, and Sabatini DM. Growing roles for the mTOR pathway. Current opinion in cell biology. 2005;17(6):596-603. 133. McKay MM, and Morrison DK. Integrating signals from RTKs to ERK/MAPK. Oncogene. 2007;26(22):3113-21. 134. Zoncu R, Efeyan A, and Sabatini DM. mTOR: from growth signal integration to cancer, diabetes and ageing. Nature reviews Molecular cell biology. 2011;12(1):21-35. 135. Roux PP, Ballif BA, Anjum R, Gygi SP, and Blenis J. Tumor-promoting phorbol esters and activated Ras inactivate the tuberous sclerosis tumor suppressor complex via p90 ribosomal S6 kinase. Proceedings of the National Academy of Sciences of the United States of America. 2004;101(37):13489-94. 136. Bader AG, Kang S, Zhao L, and Vogt PK. Oncogenic PI3K deregulates transcription and translation. Nature reviews Cancer. 2005;5(12):921-9. 137. Wullschleger S, Loewith R, and Hall MN. TOR signaling in growth and metabolism. Cell. 2006;124(3):471-84. 138. Castellano E, and Downward J. RAS Interaction with PI3K: More Than Just Another Effector Pathway. Genes & cancer. 2011;2(3):261-74. 139. Cichowski K, and Jacks T. NF1 tumor suppressor gene function: narrowing the GAP. Cell. 2001;104(4):593-604. 140. Sarbassov DD, Ali SM, Sengupta S, Sheen JH, Hsu PP, Bagley AF, Markhard AL, and Sabatini DM. Prolonged rapamycin treatment inhibits mTORC2 assembly and Akt/PKB. Molecular cell. 2006;22(2):159-68. 141. Wang X, Yue P, Kim YA, Fu H, Khuri FR, and Sun SY. Enhancing mammalian target of rapamycin (mTOR)-targeted cancer therapy by preventing mTOR/raptor inhibition-initiated, mTOR/rictor-independent Akt activation. Cancer Res. 2008;68(18):7409-18. 142. Boscolo E, Limaye N, Huang L, Kang KT, Soblet J, Uebelhoer M, Mendola A, Natynki M, Seront E, Dupont S, et al. Rapamycin improves TIE2-mutated venous malformation in murine model and human subjects. J Clin Invest. 2015;125(9):3491-504. 143. Meric-Bernstam F, and Gonzalez-Angulo AM. Targeting the mTOR signaling network for cancer therapy. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2009;27(13):2278-87. 144. Liu Q, Kang SA, Thoreen CC, Hur W, Wang J, Chang JW, Markhard A, Zhang J, Sim T, Sabatini DM, et al. Development of ATP-competitive mTOR inhibitors. Methods in molecular biology. 2012;821(447-60. 145. Bain J, Plater L, Elliott M, Shpiro N, Hastie CJ, McLauchlan H, Klevernic I, Arthur JS, Alessi DR, and Cohen P. The selectivity of protein kinase inhibitors: a further update. The Biochemical journal. 2007;408(3):297-315. 158 146. Liu Q, Kirubakaran S, Hur W, Niepel M, Westover K, Thoreen CC, Wang J, Ni J, Patricelli MP, Vogel K, et al. Kinome-wide selectivity profiling of ATP- competitive mammalian target of rapamycin (mTOR) inhibitors and characterization of their binding kinetics. The Journal of biological chemistry. 2012;287(13):9742-52. 147. Sun SY. mTOR kinase inhibitors as potential cancer therapeutic drugs. Cancer letters. 2013;340(1):1-8. 148. Naing A, Aghajanian C, Raymond E, Olmos D, Schwartz G, Oelmann E, Grinsted L, Burke W, Taylor R, Kaye S, et al. Safety, tolerability, pharmacokinetics and pharmacodynamics of AZD8055 in advanced solid tumours and lymphoma. British journal of cancer. 2012;107(7):1093-9. 149. Wang D, Boerner SA, Winkler JD, and LoRusso PM. Clinical experience of MEK inhibitors in cancer therapy. Biochimica et biophysica acta. 2007;1773(8):1248-55. 150. Sebolt-Leopold JS, and Herrera R. Targeting the mitogen-activated protein kinase cascade to treat cancer. Nature reviews Cancer. 2004;4(12):937-47. 151. Adjei AA, Cohen RB, Franklin W, Morris C, Wilson D, Molina JR, Hanson LJ, Gore L, Chow L, Leong S, et al. Phase I pharmacokinetic and pharmacodynamic study of the oral, small-molecule mitogen-activated protein kinase kinase 1/2 inhibitor AZD6244 (ARRY-142886) in patients with advanced cancers. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2008;26(13):2139-46. 152. Hocker T, and Tsao H. Ultraviolet radiation and melanoma: a systematic review and analysis of reported sequence variants. Human mutation. 2007;28(6):578-88. 153. Grimaldi AM, Simeone E, and Ascierto PA. The role of MEK inhibitors in the treatment of metastatic melanoma. Current opinion in oncology. 2014;26(2):196-203. 154. Infante JR, Fecher LA, Falchook GS, Nallapareddy S, Gordon MS, Becerra C, DeMarini DJ, Cox DS, Xu Y, Morris SR, et al. Safety, pharmacokinetic, pharmacodynamic, and efficacy data for the oral MEK inhibitor trametinib: a phase 1 dose-escalation trial. The Lancet Oncology. 2012;13(8):773-81. 155. Kim KB, Kefford R, Pavlick AC, Infante JR, Ribas A, Sosman JA, Fecher LA, Millward M, McArthur GA, Hwu P, et al. Phase II study of the MEK1/MEK2 inhibitor Trametinib in patients with metastatic BRAF-mutant cutaneous melanoma previously treated with or without a BRAF inhibitor. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2013;31(4):482-9. 156. Ascierto PA, Minor D, Ribas A, Lebbe C, O'Hagan A, Arya N, Guckert M, Schadendorf D, Kefford RF, Grob JJ, et al. Phase II trial (BREAK-2) of the BRAF inhibitor dabrafenib (GSK2118436) in patients with metastatic melanoma. Journal of clinical oncology : official journal of the American Society of Clinical Oncology. 2013;31(26):3205-11. 157. Carracedo A, Ma L, Teruya-Feldstein J, Rojo F, Salmena L, Alimonti A, Egia A, Sasaki AT, Thomas G, Kozma SC, et al. Inhibition of mTORC1 leads to MAPK 159 pathway activation through a PI3K-dependent feedback loop in human cancer. J Clin Invest. 2008;118(9):3065-74. 158. Kinkade CW, Castillo-Martin M, Puzio-Kuter A, Yan J, Foster TH, Gao H, Sun Y, Ouyang X, Gerald WL, Cordon-Cardo C, et al. Targeting AKT/mTOR and ERK MAPK signaling inhibits hormone-refractory prostate cancer in a preclinical mouse model. J Clin Invest. 2008;118(9):3051-64. 159. Engelman JA, Chen L, Tan X, Crosby K, Guimaraes AR, Upadhyay R, Maira M, McNamara K, Perera SA, Song Y, et al. Effective use of PI3K and MEK inhibitors to treat mutant Kras G12D and PIK3CA H1047R murine lung cancers. Nature medicine. 2008;14(12):1351-6. 160. Holt SV, Logie A, Davies BR, Alferez D, Runswick S, Fenton S, Chresta CM, Gu Y, Zhang J, Wu YL, et al. Enhanced apoptosis and tumor growth suppression elicited by combination of MEK (selumetinib) and mTOR kinase inhibitors (AZD8055). Cancer Res. 2012;72(7):1804-13. 161. Hoang B, Benavides A, Shi Y, Yang Y, Frost P, Gera J, and Lichtenstein A. The PP242 mammalian target of rapamycin (mTOR) inhibitor activates extracellular signal-regulated kinase (ERK) in multiple myeloma cells via a target of rapamycin complex 1 (TORC1)/eukaryotic translation initiation factor 4E (eIF-4E)/RAF pathway and activation is a mechanism of resistance. The Journal of biological chemistry. 2012;287(26):21796-805. 162. Blaser B, Waselle L, Dormond-Meuwly A, Dufour M, Roulin D, Demartines N, and Dormond O. Antitumor activities of ATP-competitive inhibitors of mTOR in colon cancer cells. BMC cancer. 2012;12(86. 163. Dormond-Meuwly A, Roulin D, Dufour M, Benoit M, Demartines N, and Dormond O. The inhibition of MAPK potentiates the anti-angiogenic efficacy of mTOR inhibitors. Biochemical and biophysical research communications. 2011;407(4):714-9. 164. Turke AB, Song Y, Costa C, Cook R, Arteaga CL, Asara JM, and Engelman JA. MEK inhibition leads to PI3K/AKT activation by relieving a negative feedback on ERBB receptors. Cancer Res. 2012;72(13):3228-37. 165. Catalanotti F, Solit DB, Pulitzer MP, Berger MF, Scott SN, Iyriboz T, Lacouture ME, Panageas KS, Wolchok JD, Carvajal RD, et al. Phase II trial of MEK inhibitor selumetinib (AZD6244, ARRY-142886) in patients with BRAFV600E/K-mutated melanoma. Clinical cancer research : an official journal of the American Association for Cancer Research. 2013;19(8):2257-64. 166. Gopal YN, Deng W, Woodman SE, Komurov K, Ram P, Smith PD, and Davies MA. Basal and treatment-induced activation of AKT mediates resistance to cell death by AZD6244 (ARRY-142886) in Braf-mutant human cutaneous melanoma cells. Cancer Res. 2010;70(21):8736-47. 167. Eroglu Z, Tawbi HA, Hu J, Guan M, Frankel PH, Ruel NH, Wilczynski S, Christensen S, Gandara DR, and Chow WA. A randomised phase II trial of selumetinib vs selumetinib plus temsirolimus for soft-tissue sarcomas. British journal of cancer. 2015;112(10):1644-51. 168. Tolcher AW, Bendell JC, Papadopoulos KP, Burris HA, 3rd, Patnaik A, Jones SF, Rasco D, Cox DS, Durante M, Bellew KM, et al. A phase IB trial of the oral MEK inhibitor trametinib (GSK1120212) in combination with everolimus in 160 patients with advanced solid tumors. Annals of oncology : official journal of the European Society for Medical Oncology / ESMO. 2015;26(1):58-64. 169. Chow LM, and Baker SJ. Capturing the molecular and biological diversity of high-grade astrocytoma in genetically engineered mouse models. Oncotarget. 2012;3(1):67-77. 170. Chow LM, Endersby R, Zhu X, Rankin S, Qu C, Zhang J, Broniscer A, Ellison DW, and Baker SJ. Cooperativity within and among Pten, p53, and Rb pathways induces high-grade astrocytoma in adult brain. Cancer Cell. 2011;19(3):305-16. 171. Heffner CS, Herbert Pratt C, Babiuk RP, Sharma Y, Rockwood SF, Donahue LR, Eppig JT, and Murray SA. Supporting conditional mouse mutagenesis with a comprehensive cre characterization resource. Nature communications. 2012;3(1218. 172. Seki Y, Yamamoto N, Tamura Y, Goto Y, Shibata T, Tanioka M, Asahina H, Nokihara H, Yamada Y, Shimamoto T, et al. Phase I study for ridaforolimus, an oral mTOR inhibitor, in Japanese patients with advanced solid tumors. Cancer chemotherapy and pharmacology. 2012;69(4):1099-105. 173. Streit S, Ruhe JE, Knyazev P, Knyazeva T, Iacobelli S, Peter S, Hoefler H, and Ullrich A. PTP-PEST phosphatase variations in human cancer. Cancer genetics and cytogenetics. 2006;170(1):48-53. 174. Koroknai V, Ecsedi S, Vizkeleti L, Kiss T, Szasz I, Lukacs A, Papp O, Adany R, and Balazs M. Genomic profiling of invasive melanoma cell lines by array comparative genomic hybridization. Melanoma research. 2016;26(2):100-7. 175. Du W, Gerald D, Perruzzi CA, Rodriguez-Waitkus P, Enayati L, Krishnan B, Edmonds J, Hochman ML, Lev DC, and Phung TL. Vascular tumors have increased p70 S6-kinase activation and are inhibited by topical rapamycin. Lab Invest. 2013;93(10):1115-27. 176. Lahat G, Dhuka AR, Hallevi H, Xiao L, Zou C, Smith KD, Phung TL, Pollock RE, Benjamin R, Hunt KK, et al. Angiosarcoma: clinical and molecular insights. Annals of surgery. 2010;251(6):1098-106. 177. Przygodzki RM, Finkelstein SD, Keohavong P, Zhu D, Bakker A, Swalsky PA, Soini Y, Ishak KG, and Bennett WP. Sporadic and Thorotrast-induced angiosarcomas of the liver manifest frequent and multiple point mutations in K-ras-2. Lab Invest. 1997;76(1):153-9. 178. Boivin-Angele S, Lefrancois L, Froment O, Spiethoff A, Bogdanffy MS, Wegener K, Wesch H, Barbin A, Bancel B, Trepo C, et al. Ras gene mutations in vinyl chloride-induced liver tumours are carcinogen-specific but vary with cell type and species. International journal of cancer. 2000;85(2):223-7. 179. Garcia JM, Gonzalez R, Silva JM, Dominguez G, Vegazo IS, Gamallo C, Provencio M, Espana P, and Bonilla F. Mutational status of K-ras and TP53 genes in primary sarcomas of the heart. British journal of cancer. 2000;82(6):1183-5. 180. Hollstein M, Marion MJ, Lehman T, Welsh J, Harris CC, Martel-Planche G, Kusters I, and Montesano R. p53 mutations at A:T base pairs in angiosarcomas of vinyl chloride-exposed factory workers. Carcinogenesis. 1994;15(1):1-3. 161 181. Naka N, Tomita Y, Nakanishi H, Araki N, Hongyo T, Ochi T, and Aozasa K. Mutations of p53 tumor-suppressor gene in angiosarcoma. International journal of cancer. 1997;71(6):952-5. 182. Zietz C, Rossle M, Haas C, Sendelhofert A, Hirschmann A, Sturzl M, and Lohrs U. MDM-2 oncoprotein overexpression, p53 gene mutation, and VEGF up- regulation in angiosarcomas. The American journal of pathology. 1998;153(5):1425-33. 183. Calvete O, Martinez P, Garcia-Pavia P, Benitez-Buelga C, Paumard-Hernandez B, Fernandez V, Dominguez F, Salas C, Romero-Laorden N, Garcia-Donas J, et al. A mutation in the POT1 gene is responsible for cardiac angiosarcoma in TP53-negative Li-Fraumeni-like families. Nature communications. 2015;6(8383. 184. Hellstrom M, Kalen M, Lindahl P, Abramsson A, and Betsholtz C. Role of PDGF-B and PDGFR-beta in recruitment of vascular smooth muscle cells and pericytes during embryonic blood vessel formation in the mouse. Development. 1999;126(14):3047-55. 185. Guo S, Yu L, Cheng Y, Li C, Zhang J, An J, Wang H, Yan B, Zhan T, Cao Y, et al. PDGFRbeta triggered by bFGF promotes the proliferation and migration of endothelial progenitor cells via p-ERK signalling. Cell Biol Int. 2012;36(10):945-50. 186. Wang H, Yin Y, Li W, Zhao X, Yu Y, Zhu J, Qin Z, Wang Q, Wang K, Lu W, et al. Over-expression of PDGFR-beta promotes PDGF-induced proliferation, migration, and angiogenesis of EPCs through PI3K/Akt signaling pathway. PLoS One. 2012;7(2):e30503. 187. Wyler von Ballmoos M, Yang Z, Volzmann J, Baumgartner I, Kalka C, and Di Santo S. Endothelial progenitor cells induce a phenotype shift in differentiated endothelial cells towards PDGF/PDGFRbeta axis-mediated angiogenesis. PLoS One. 2010;5(11):e14107. 188. Antonescu CR, Yoshida A, Guo T, Chang NE, Zhang L, Agaram NP, Qin LX, Brennan MF, Singer S, and Maki RG. KDR activating mutations in human angiosarcomas are sensitive to specific kinase inhibitors. Cancer Res. 2009;69(18):7175-9. 189. Guo T, Zhang L, Chang NE, Singer S, Maki RG, and Antonescu CR. Consistent MYC and FLT4 gene amplification in radiation-induced angiosarcoma but not in other radiation-associated atypical vascular lesions. Genes Chromosomes Cancer. 2011;50(1):25-33. 190. Kollar A, Jones RL, Stacchiotti S, Gelderblom H, Guida M, Grignani G, Steeghs N, Safwat A, Katz D, Duffaud F, et al. Pazopanib in advanced vascular sarcomas: an EORTC Soft Tissue and Bone Sarcoma Group (STBSG) retrospective analysis. Acta Oncol. 2016:1-5. 191. Marino S, Vooijs M, van der Gulden H, Jonkers J, and Berns A. Induction of medulloblastomas in p53-null mutant mice by somatic inactivation of Rb in the external granular layer cells of the cerebellum. Gene Dev. 2000;14(8):994-1004. 192. Backman SA, Stambolic V, Suzuki A, Haight J, Elia A, Pretorius J, Tsao MS, Shannon P, Bolon B, Ivy GO, et al. Deletion of Pten in mouse brain causes 162 seizures, ataxia and defects in soma size resembling Lhermitte-Duclos disease. Nature genetics. 2001;29(4):396-403. 193. Tibbitt MW, and Anseth KS. Hydrogels as extracellular matrix mimics for 3D cell culture. Biotechnology and bioengineering. 2009;103(4):655-63. 194. Birgersdotter A, Sandberg R, and Ernberg I. Gene expression perturbation in vitro--a growing case for three-dimensional (3D) culture systems. Seminars in cancer biology. 2005;15(5):405-12. 195. Bhadriraju K, and Chen CS. Engineering cellular microenvironments to improve cell-based drug testing. Drug discovery today. 2002;7(11):612-20. 196. Goodspeed A, Heiser LM, Gray JW, and Costello JC. Tumor-Derived Cell Lines as Molecular Models of Cancer Pharmacogenomics. Molecular cancer research : MCR. 2016;14(1):3-13. 197. Kim JB. Three-dimensional tissue culture models in cancer biology. Seminars in cancer biology. 2005;15(5):365-77. 198. Khaitan D, Chandna S, Arya MB, and Dwarakanath BS. Establishment and characterization of multicellular spheroids from a human glioma cell line; Implications for tumor therapy. Journal of translational medicine. 2006;4(12. 199. Mehta G, Hsiao AY, Ingram M, Luker GD, and Takayama S. Opportunities and challenges for use of tumor spheroids as models to test drug delivery and efficacy. Journal of controlled release : official journal of the Controlled Release Society. 2012;164(2):192-204. 200. Pastrana E, Silva-Vargas V, and Doetsch F. Eyes wide open: a critical review of sphere-formation as an assay for stem cells. Cell stem cell. 2011;8(5):486-98. 201. Zanetta L, Corada M, Grazia Lampugnani M, Zanetti A, Breviario F, Moons L, Carmeliet P, Pepper MS, and Dejana E. Downregulation of vascular endothelial-cadherin expression is associated with an increase in vascular tumor growth and hemorrhagic complications. Thrombosis and haemostasis. 2005;93(6):1041-6. 202. Kenny PA, Lee GY, Myers CA, Neve RM, Semeiks JR, Spellman PT, Lorenz K, Lee EH, Barcellos-Hoff MH, Petersen OW, et al. The morphologies of breast cancer cell lines in three-dimensional assays correlate with their profiles of gene expression. Molecular oncology. 2007;1(1):84-96. 203. Melero-Martin JM, Khan ZA, Picard A, Wu X, Paruchuri S, and Bischoff J. In vivo vasculogenic potential of human blood-derived endothelial progenitor cells. Blood. 2007;109(11):4761-8. 204. Jiang M, Min Y, Debusk L, Fernandez S, Strand DW, Hayward SW, and Lin PC. Spontaneous immortalization of human dermal microvascular endothelial cells. World journal of stem cells. 2010;2(5):114-20. 205. Dean M, Fojo T, and Bates S. Tumour stem cells and drug resistance. Nature reviews Cancer. 2005;5(4):275-84. 206. Samatar AA, and Poulikakos PI. Targeting RAS-ERK signalling in cancer: promises and challenges. Nature reviews Drug discovery. 2014;13(12):928- 42. 207. Gilmartin AG, Bleam MR, Groy A, Moss KG, Minthorn EA, Kulkarni SG, Rominger CM, Erskine S, Fisher KE, Yang J, et al. GSK1120212 (JTP-74057) is an inhibitor of MEK activity and activation with favorable pharmacokinetic 163 properties for sustained in vivo pathway inhibition. Clinical cancer research : an official journal of the American Association for Cancer Research. 2011;17(5):989-1000. 208. Amaya CN, and Bryan BA. Enrichment of the embryonic stem cell reprogramming factors Oct4, Nanog, Myc, and Sox2 in benign and malignant vascular tumors. BMC clinical pathology. 2015;15(18. 209. Zhao C, Xiao H, Wu X, Li C, Liang G, Yang S, and Lin J. Rational combination of MEK inhibitor and the STAT3 pathway modulator for the therapy in K-Ras mutated pancreatic and colon cancer cells. Oncotarget. 2015;6(16):14472-87. 210. Arriola Apelo SI, Neuman JC, Baar EL, Syed FA, Cummings NE, Brar HK, Pumper CP, Kimple ME, and Lamming DW. Alternative rapamycin treatment regimens mitigate the impact of rapamycin on glucose homeostasis and the immune system. Aging cell. 2016;15(1):28-38. 211. Gothert JR, Gustin SE, van Eekelen JA, Schmidt U, Hall MA, Jane SM, Green AR, Gottgens B, Izon DJ, and Begley CG. Genetically tagging endothelial cells in vivo: bone marrow-derived cells do not contribute to tumor endothelium. Blood. 2004;104(6):1769-77. 212. Dow LE, Fisher J, O'Rourke KP, Muley A, Kastenhuber ER, Livshits G, Tschaharganeh DF, Socci ND, and Lowe SW. Inducible in vivo genome editing with CRISPR-Cas9. Nature biotechnology. 2015;33(4):390-4. 213. Singh P, Schimenti JC, and Bolcun-Filas E. A mouse geneticist's practical guide to CRISPR applications. Genetics. 2015;199(1):1-15. 214. Yang H, Wang H, Shivalila CS, Cheng AW, Shi L, and Jaenisch R. One-step generation of mice carrying reporter and conditional alleles by CRISPR/Cas- mediated genome engineering. Cell. 2013;154(6):1370-9. 164