EGFL6 promotes endometrial cancer cell migration and proliferation.

OBJECTIVE: EGFL6, a growth factor produced by adipocytes, is upregulated in and implicated in the tumorigenesis of multiple tumor types. Given the strong link between obesity and endometrial cancer, we sought to determine the impact of EGFL6 on endometrial cancer. METHODS: EGFL6 expression in endometrial cancer and correlation with patient outcomes was evaluated in the human protein atlas and TCGA. EGFL6 treatment, expression upregulation, and shRNA knockdown were used to evaluate the impact of EGFL6 on the proliferation and migration of 3 endometrial cancer cell lines in vitro. Similarly, the impact of EGFL6 expression and knockdown on tumor growth was evaluated. Western blotting was used to evaluate the impact of EGFL6 on MAPK phosphorylation. RESULTS: EGFL6 is upregulated in endometrial cancer, primarily in cony-number high tumors. High tumor endometrial cancer expression of EGFL6 predicts poor patient prognosis. We find that EGFL6 acts to activate the MAPK pathway increasing cellular proliferation and migration. In xenograft models, EGFL6 overexpression increases endometrial cancer tumor growth while EGFL6 knockdown decreases endometrial cancer tumor growth. CONCLUSIONS: EGFL6 is a marker of poor prognosis endometrial cancers, driving cancer cell proliferation and growth. As such EGFL6 represents a potential therapeutic target in endometrial cancer.

Generation and characterization of humanized affinity-matured EGFL6 antibodies for ovarian cancer therapy.

OBJECTIVES: Epidermal growth factor EGF-like domain multiple-6 (EGFL6) is highly expressed in high grade serous ovarian cancer and promotes both endothelial cell proliferation/angiogenesis and cancer cell proliferation/metastasis. As such it has been implicated as a therapeutic target. As a secreted factor, EGFL6 is a candidate for antibody therapy. The objectives of this study were to create and validate humanized affinity-matured EGFL6 neutralizing antibodies for clinical development. METHODS: A selected murine EGFL6 antibody was humanized using CDR grafting to create 26 variant humanized antibodies. These were screened and the lead candidate was affinity matured. Seven humanized affinity-matured EGFL6 antibodies were screened for their ability to block EGFL6 activity on cancer cells in vitro, two of which were selected and tested their therapeutic activity in vivo. RESULTS: Humanized affinity matured antibodies demonstrated high affinity for EGFL6 (150 pM to 2.67 nM). We found that several humanized affinity-matured EGFL6 antibodies specifically bound to recombinant, and native human EGFL6. Two lead antibodies were able to inhibit EGFL6-mediated (i) cancer cell migration, (ii) proliferation, and (iii) increase in ERK phosphorylation in cancer cells in vitro. Both lead antibodies restricted growth of an EGFL6 expressing ovarian cancer patient derived xenograft. Analysis of treated human tumor xenografts indicated that anti-EGFL6 therapy suppressed angiogenesis, inhibited tumor cell proliferation, and promoted tumor cell apoptosis. CONCLUSIONS: Our studies confirm the ability of these humanized affinity-matured antibodies to neutralize EGFL6 and acting as a therapeutic to restrict cancer growth. This work supports the development of these antibody for first-in-human clinical trials.

Identifying an ovarian cancer cell hierarchy regulated by bone morphogenetic protein 2.

Whether human cancer follows a hierarchical or stochastic model of differentiation is controversial. Furthermore, the factors that regulate cancer stem-like cell (CSC) differentiation potential are largely unknown. We used a novel microfluidic single-cell culture method to directly observe the differentiation capacity of four heterogeneous ovarian cancer cell populations defined by the expression of the CSC markers aldehyde dehydrogenase (ALDH) and CD133. We evaluated 3,692 progeny from 2,833 cells. We found that only ALDH(+)CD133(+) cells could generate all four ALDH(+/-)CD133(+/-) cell populations and identified a clear branched differentiation hierarchy. We also observed a single putative stochastic event. Within the hierarchy of cells, bone morphologenetic protein 2 (BMP2) is preferentially expressed in ALDH(-)CD133(-) cells. BMP2 promotes ALDH(+)CD133(+) cell expansion while suppressing the proliferation of ALDH(-)CD133(-) cells. As such, BMP2 suppressed bulk cancer cell growth in vitro but increased tumor initiation rates, tumor growth, and chemotherapy resistance in vivo whereas BMP2 knockdown reduced CSC numbers, in vivo growth, and chemoresistance. These data suggest a hierarchical differentiation pattern in which BMP2 acts as a feedback mechanism promoting ovarian CSC expansion and suppressing progenitor proliferation. These results explain why BMP2 suppresses growth in vitro and promotes growth in vivo. Together, our results support BMP2 as a therapeutic target in ovarian cancer.

Egfl6 promotes ovarian cancer progression by enhancing the immunosuppressive functions of tumor-associated myeloid cells.

Tumor-associated macrophages (TAMs) and myeloid-derived suppressor cells (MDSCs) play a critical role in resistance to immunotherapy. In this study, we identified epidermal growth factor-like 6 (Egfl6) as a new regulator of myeloid cell functions. Our analyses indicated that Egfl6, via binding with ß3 integrins and activation of p38 and SYK signaling, acts as a chemotactic factor for myeloid cells migration and promotes their differentiation towards an immunosuppressive state. In syngeneic mouse models of ovarian cancer (OvCa), tumor expression of Egfl6 increased the intra-tumoral accumulation of polymorphonuclear (PMN) MDSCs and TAMs and their expression of immunosuppressive factors, including CXCL2, IL-10 and PD-L1. Consistent with this, in an immune 'hot' tumor model, Egfl6 expression eliminated response to a-PD-L1 therapy, while Egfl6 neutralizing antibody decreased the accumulation of tumor-infiltrating CD206+ TAMs and PMN-MDSCs and restored the efficacy of a-PD-L1 therapy. Supporting a role in human tumors, in human OvCa tissue samples, areas of high EGFL6 expression co-localized with myeloid cell infiltration. scRNAseq analyses revealed a correlation between EGFL6 and immune cell expression of immunosuppressive factors. Our data provide mechanistic insights into the onco-immunologic functions of EGFL6 in mediating tumor immune suppression and identified EGFL6 as a potential novel therapeutic target to enhance immunotherapy in OvCa patients.

A Novel Humanized Immune Stroma PDX Cancer Model for Therapeutic Studies.

Standard preclinical human tumor models lack a human tumor stroma. However, as stroma contributes to therapeutic resistance, the lack of human stroma may make current models less stringent for testing new therapies. To address this, using patient-derived tumor cells, patient derived cancer-associated mesenchymal stem/progenitor cells, and human endothelial cells, we created a Human Stroma-Patient Derived Xenograft (HS-PDX) tumor model. HS-PDX, compared to the standard PDX model, demonstrate greater resistance to targeted therapy and chemotherapy, and better reflect patient response to therapy. Furthermore, HS-PDX can be grown in mice with humanized bone marrow to create humanized immune stroma patient-derived xenograft (HIS-PDX) models. The HIS-PDX model contains human connective tissues, vascular and immune cell infiltrates. RNA sequencing analysis demonstrated a 94-96% correlation with primary human tumor. Using this model, we demonstrate the impact of human tumor stroma on response to CAR-T cell therapy and immune checkpoint inhibitor therapy. We show an immunosuppressive role for human tumor stroma and that this model can be used to identify immunotherapeutic combinations to overcome stromally mediated immunosuppression. Combined, our data confirm a critical role for human stoma in therapeutic response and indicate that HIS-PDX can be an important tool for preclinical drug testing. Statement of Significance: We developed a tumor model with human stromal, vascular, and immune cells. This model mirrors patient response to chemotherapy, targeted therapy, and immunotherapy, and can be used to study therapy resistance.

Targeting Therapeutic Resistance and Multinucleate Giant Cells in CCNE1-Amplified HR-Proficient Ovarian Cancer.

Approximately 20% of high-grade serous ovarian cancers (HGSOC) have CCNE1 amplification. CCNE1-amplified tumors are homologous recombination (HR) proficient and resistant to standard therapies. Therapy resistance is associated with increased numbers of polyploid giant cancer cells (PGCC). We sought to identify new therapeutic approaches for patients with CCNE1-amplified tumors. Using TCGA data, we find that the mTOR, HR, and DNA checkpoint pathways are enriched in CCNE1-amplified ovarian cancers. Furthermore, Interactome Mapping Analysis linked the mTOR activity with upregulation of HR and DNA checkpoint pathways. Indeed, we find that mTOR inhibitors (mTORi) downregulate HR/checkpoint genes in CCNE1-amplified tumors. As CCNE1-amplified tumors are dependent on the HR pathway for viability, mTORi proved selectively effective in CCNE1-amplified tumors. Similarly, via downregulation of HR genes, mTORi increased CCNE1-amplifed HGSOC response to PARPi. In contrast, overexpression of HR/checkpoint proteins (RAD51 or ATR), induced resistance to mTORi. In vivo, mTORi alone potently reduced CCNE1-amplified tumor growth and the combination of mTORi and PARPi increased response and tumor eradication. Tumors treated with mTORi demonstrated a significant reduction in ALDH+ PGCCs. Finally, as a proof of principle, we identified three patients with CCNE1 amplified tumors who were treated with an mTORi. All three obtained clinical benefits from the therapy. Our studies and clinical experience indicate mTORi are a potential therapeutic approach for patients with CCNE1-amplified tumors.

MicroRNA-122 inhibits tumorigenic properties of hepatocellular carcinoma cells and sensitizes these cells to sorafenib.

MicroRNAs are negative regulators of protein coding genes. The liver-specific microRNA-122 (miR-122) is frequently suppressed in primary hepatocellular carcinomas (HCCs). In situ hybridization demonstrated that miR-122 is abundantly expressed in hepatocytes but barely detectable in primary human HCCs. Ectopic expression of miR-122 in nonexpressing HepG2, Hep3B, and SK-Hep-1 cells reversed their tumorigenic properties such as growth, replication potential, clonogenic survival, anchorage-independent growth, migration, invasion, and tumor formation in nude mice. Further, miR-122-expressing HCC cells retained an epithelial phenotype that correlated with reduced Vimentin expression. ADAM10 (a distintegrin and metalloprotease family 10), serum response factor (SRF), and insulin-like growth factor 1 receptor (Igf1R) that promote tumorigenesis were validated as targets of miR-122 and were repressed by the microRNA. Conversely, depletion of the endogenous miR-122 in Huh-7 cells facilitated their tumorigenic properties with concomitant up-regulation of these targets. Expression of SRF or Igf1R partially reversed tumor suppressor function of miR-122. Further, miR-122 impeded angiogenic properties of endothelial cells in vitro. Notably, ADAM10, SRF, and Igf1R were up-regulated in primary human HCCs compared with the matching liver tissue. Co-labeling studies demonstrated exclusive localization of miR-122 in the benign livers, whereas SRF predominantly expressed in HCC. More importantly, growth and clonogenic survival of miR-122-expressing HCC cells were significantly reduced upon treatment with sorafenib, a multi-kinase inhibitor clinically effective against HCC. Collectively, these results suggest that the loss of multifunctional miR-122 contributes to the malignant phenotype of HCC cells, and miR-122 mimetic alone or in combination with anticancer drugs can be a promising therapeutic regimen against liver cancer.

An evolving paradigm of cancer stem cell hierarchies: therapeutic implications.

Over a decade of research has confirmed the critical role of cancer stem-like cells (CSCs) in tumor initiation, chemoresistance, and metastasis. Increasingly, CSC hierarchies have begun to be defined with some recurring themes. This includes evidence that these hierarchies are 'flexible,' with both cell state transitions and dedifferentiation events possible. These findings pose therapeutic hurdles and opportunities. Here, we review cancer stem cell hierarchies and their interactions with the tumor microenvironment. We also discuss the current therapeutic approaches designed to target CSC hierarchies and initial clinical trial results for CSC targeting agents. While cancer stem cell targeted therapies are still in their infancy, we are beginning to see encouraging results that suggest a positive outlook for CSC-targeting approaches.

NFATC4 promotes quiescence and chemotherapy resistance in ovarian cancer.

Development of chemotherapy resistance is a major problem in ovarian cancer. One understudied mechanism of chemoresistance is the induction of quiescence, a reversible nonproliferative state. Unfortunately, little is known about regulators of quiescence. Here, we identify the master transcription factor nuclear factor of activated T cells cytoplasmic 4 (NFATC4) as a regulator of quiescence in ovarian cancer. NFATC4 is enriched in ovarian cancer stem-like cells and correlates with decreased proliferation and poor prognosis. Treatment of cancer cells with cisplatin resulted in NFATC4 nuclear translocation and activation of the NFATC4 pathway, while inhibition of the pathway increased chemotherapy response. Induction of NFATC4 activity resulted in a marked decrease in proliferation, G0 cell cycle arrest, and chemotherapy resistance, both in vitro and in vivo. Finally, NFATC4 drove a quiescent phenotype in part via downregulation of MYC. Together, these data identify NFATC4 as a driver of quiescence and a potential new target to combat chemoresistance in ovarian cancer.

CD105 Is Expressed in Ovarian Cancer Precursor Lesions and Is Required for Metastasis to the Ovary.

: Most high-grade serous ovarian cancers (HGSCs) initiate from the fallopian tube epithelium and then metastasize to the ovary and throughout the abdomen. Genomic analyses suggest that most HGSCs seed the ovary prior to abdominal dissemination. Similarly, animal models support a critical role for the ovary in driving abdominal dissemination. Thus, HGSC cell recruitment to the ovary appears to be a critical component of HGSC cell metastasis. We sought to identify factors driving HGSC recruitment to the ovary. We identified CD105 (endoglin, or ENG, a TGF- receptor family member) as a mediator of HGSC cell ovarian recruitment. We found that CD105 was expressed on both serous tubal intraepithelial carcinoma (STIC) cells (STICs-HGSC precursors in the fallopian tube epithelium) and HGSC cells. Using data from The Cancer Genome Atlas (TCGA) and the Cancer Cell Line Encyclopedia (CCLE), we showed that high CD105 expression by HGSC cells correlated with a metastatic signature. Furthermore, intravenous injection of CD105(+) HGSC tumor cells, but not CD105(-), resulted in ovarian-specific metastasis and abdominal dissemination of disease. CD105 knockdown or blockade with a clinically relevant CD105-neutralizing mAb (TRC105), inhibited HGSC metastasis, reduced ascites, and impeded growth of abdominal tumor nodules, thereby improving overall survival in animal models of ovarian cancer. CD105 knockdown was associated with a reduction in TGF-signaling. Together, our data support CD105 as a critical mediator of ovarian cancer spread to the ovary and implicate it as a potential therapeutic target.

5-Aza-deoxycytidine induces selective degradation of DNA methyltransferase 1 by a proteasomal pathway that requires the KEN box, bromo-adjacent homology domain, and nuclear localization signal.

5-Azacytidine- and 5-aza-deoxycytidine (5-aza-CdR)-mediated reactivation of tumor suppressor genes silenced by promoter methylation has provided an alternate approach in cancer therapy. Despite the importance of epigenetic therapy, the mechanism of action of DNA-hypomethylating agents in vivo has not been completely elucidated. Here we report that among three functional DNA methyltransferases (DNMT1, DNMT3A, and DNMT3B), the maintenance methyltransferase, DNMT1, was rapidly degraded by the proteasomal pathway upon treatment of cells with these drugs. The 5-aza-CdR-induced degradation, which occurs in the nucleus, could be blocked by proteasomal inhibitors and required a functional ubiquitin-activating enzyme. The drug-induced degradation occurred even in the absence of DNA replication. Treatment of cells with other nucleoside analogs modified at C-5, 5-fluorodeoxyuridine and 5-fluorocytidine, did not induce the degradation of DNMT1. Mutation of cysteine at the catalytic site of Dnmt1 (involved in the formation of a covalent intermediate with cytidine in DNA) to serine (CS) did not impede 5-aza-CdR-induced degradation. Neither the wild type nor the catalytic site mutant of Dnmt3a or Dnmt3b was sensitive to 5-aza-CdR-mediated degradation. These results indicate that covalent bond formation between the enzyme and 5-aza-CdR-incorporated DNA is not essential for enzyme degradation. Mutation of the conserved KEN box, a targeting signal for proteasomal degradation, to AAA increased the basal level of Dnmt1 and blocked its degradation by 5-aza-CdR. Deletion of the catalytic domain increased the expression of Dnmt1 but did not confer resistance to 5-aza-CdR-induced degradation. Both the nuclear localization signal and the bromo-adjacent homology domain were essential for nuclear localization and for the 5-aza-CdR-mediated degradation of Dnmt1. Polyubiquitination of Dnmt1 in vivo and its stabilization upon treatment of cells with a proteasomal inhibitor indicate that the level of Dnmt1 is controlled by ubiquitin-dependent proteasomal degradation. Overexpression of the substrate recognition component, Cdh1 but not Cdc20, of APC (anaphase-promoting complex)/cyclosome ubiquitin ligase reduced the level of Dnmt1 in both untreated and 5-aza-CdR-treated cells. In contrast, the depletion of Cdh1 with small interfering RNA increased the basal level of DNMT1 that blocked 5-aza-CdR-induced degradation. Dnmt1 interacted with Cdh1 and colocalized in the nucleus at discrete foci. Both Dnmt1 and Cdh1 were phosphorylated in vivo, but only Cdh1 was significantly dephosphorylated upon 5-aza-CdR treatment, suggesting its involvement in initiating the proteasomal degradation of DNMT1. These results demonstrate a unique mechanism for the selective degradation of DNMT1, the maintenance DNA methyltransferase, by well-known DNA-hypomethylating agents.

DNA methyltransferase 3b regulates nerve growth factor-induced differentiation of PC12 cells by recruiting histone deacetylase 2.

To elucidate the role of epigenetic reprogramming in cell- or tissue-specific differentiation, we explored the role of DNA methyltransferases (Dnmts) in the nerve growth factor (NGF)-induced differentiation of PC12 (pheochromocytoma) cells into neuronal cells. The mRNA and protein levels of de novo methyltransferase Dnmt3b increased, whereas those of Dnmt3a and Dnmt1 decreased, during NGF-induced neurite outgrowth. Dnmt3b localized in the nucleus, as well as in the growing neurites. When the expression of Dnmt3b was inhibited by antisense or small interfering RNA, PC12 cells continued to proliferate and failed to generate neurites. Cells depleted of Dnmt3b were unable to exit the cell cycle even after 6 days of NGF treatment. Furthermore, this failure in differentiation correlated with significant attenuation in tyrosine phosphorylation of TrkA (a marker for NGF-induced differentiation) and reduced the expression of neuronal markers, Hu antigen, and MAP2. The methyl-CpG content of the PC12 genome or the methylation status of repetitive elements was not significantly altered after differentiation and was not affected by Dnmt3b depletion. This was consistent with the ability of the catalytic-site mutant of Dnmt3b to induce differentiation in Dnmt3b-depleted cells after NGF treatment. The Dnmt3b-mediated differentiation was attributed to its N-terminal domain, which recruits histone deacetylase 2 (Hdac2), as demonstrated by (i) impeding of differentiation by the Hdac inhibitors, (ii) facilitation of the differentiation process by overexpression of the N-terminal domain of Dnmt3b, (iii) higher Hdac activity associated with Dnmt3b after NGF treatment, and (iv) coimmunoprecipitation and cosedimentation of Dnmt3b specifically with Hdac2 in a glycerol density gradient. These data indicate a novel role of Dnmt3b in neuronal differentiation.

Inhibitors of histone deacetylase and DNA methyltransferase synergistically activate the methylated metallothionein I promoter by activating the transcription factor MTF-1 and forming an open chromatin structure.

Inhibitors of DNA methyltransferase (Dnmt) and histone deacetylases (HDAC) synergistically activate the methylated metallothionein I gene (MT-I) promoter in mouse lymphosarcoma cells. The cooperative effect of these two classes of inhibitors on MT-I promoter activity was robust following demethylation of only a few CpG dinucleotides by brief exposure to 5-azacytidine (5-AzaC) but persisted even after prolonged treatment with the nucleoside analog. HDAC inhibitors (trichostatin A [TSA] and depsipeptide) either alone or in combination with 5-AzaC did not facilitate demethylation of the MT-I promoter. Treatment of cells with HDAC inhibitors increased accumulation of multiply acetylated forms of H3 and H4 histones that remained unaffected after treatment with 5-AzaC. Chromatin immunoprecipitation (ChIP) assay showed increased association of acetylated histone H4 and lysine 9 (K9)-acetyl H3 with the MT-I promoter after treatment with TSA, which was not affected following treatment with 5-AzaC. In contrast, the association of K9-methyl histone H3 with the MT-I promoter decreased significantly after treatment with 5-AzaC and TSA. ChIP assay with antibodies specific for methyl-CpG binding proteins (MBDs) demonstrated that only methyl-CpG binding protein 2 (MeCP2) was associated with the MT-I promoter, which was significantly enhanced after TSA treatment. Association of histone deacetylase 1 (HDAC1) with the promoter decreased after treatment with TSA or 5-AzaC and was abolished after treatment with both inhibitors. Among the DNA methyltransferases, both Dnmt1 and Dnmt3a were associated with the MT-I promoter in the lymphosarcoma cells, and association of Dnmt1 decreased with time after treatment with 5-AzaC. Treatment of these cells with HDAC inhibitors also increased expression of the MTF-1 (metal transcription factor-1) gene as well as its DNA binding activity. In vivo genomic footprinting studies demonstrated increased occupancy of MTF-1 to metal response elements of the MT-I promoter after treatment with both inhibitors. Analysis of the promoter by mapping with restriction enzymes in vivo showed that the MT-I promoter attained a more open chromatin structure after combined treatment with 5-AzaC and TSA as opposed to treatment with either agent alone. These results implicate involvement of multifarious factors including modified histones, MBDs, and Dnmts in silencing the methylated MT-I promoter in lymphosarcoma cells. The synergistic activation of this promoter by these two types of inhibitors is due to demethylation of the promoter and altered association of different factors that leads to reorganization of the chromatin and the resultant increase in accessibility of the promoter to the activated transcription factor MTF-1.

Physical and functional interaction of DNA methyltransferase 3A with Mbd3 and Brg1 in mouse lymphosarcoma cells.

Dnmt3a and Dnmt3b are de novo DNA methyltransferases that also act as transcriptional repressors independent of methyltransferase activity. To elucidate the underlying mechanism of transcriptional repression, Dnmt3a was purified from mouse lymphosarcoma cells (P1798) by extensive fractionation on five different chromatographic matrices followed by glycerol density gradient centrifugation. Liquid chromatography electrospray tandem mass spectrometry analysis of Dnmt3a-associated polypeptides identified the methyl CpG binding protein Mbd3, histone deacetylase 1(Hdac1), and components of Brg1 complex (Brg1, Baf155, and Baf57) in the purified preparation. Association of Dnmt3a with Mbd3 and Brg1 was confirmed by coimmunoprecipitation and coimmunolocalization studies. Glutathione S-transferase pulldown assay showed that the NH2-terminal ATRX homology domain of Dnmt3a interacts with the methyl CpG binding domain of Mbd3 and with both bromo and ATPase domains of Brg1. Chromatin immunoprecipitation assay revealed that all three proteins are associated with transcriptionally silent methylated metallothionein (MT-I) promoter in the mouse lymphosarcoma cells. To understand the functional significance of their association with the promoter, their role on the MT-I promoter activity was analyzed by transient transfection assay. The results showed that Mbd3 and Dnmt3a specifically inhibited the methylated promoter, and the catalytic activity of Dnmt3a was dispensable for the suppression. In contrast, the wild-type but not the ATPase-inactive mutant of Brg1 suppressed MT-I promoter irrespective of its methylation status, implicating involvement of ATP-dependent chromatin remodeling in the process. Coexpression of two of the three interacting proteins at a time augmented their repressor function. This study shows physical and functional interaction of Dnmt3a with components of nucleosome remodeling machinery.

Sampling from single-cell observations to predict tumor cell growth in-vitro and in-vivo.

Cancer stem-like cells (CSCs) are a topic of increasing importance in cancer research, but are difficult to study due to their rarity and ability to rapidly divide to produce non-self-cells. We developed a simple model to describe transitions between aldehyde dehydrogenase (ALDH) positive CSCs and ALDH(-) bulk ovarian cancer cells. Microfluidics device-isolated single cell experiments demonstrated that ALDH+ cells were more proliferative than ALDH(-) cells. Based on our model we used ALDH+ and ALDH(-) cell division and proliferation properties to develop an empiric sampling algorithm and predict growth rate and CSC proportion for both ovarian cancer cell line and primary ovarian cancer cells, in-vitro and in-vivo. In both cell line and primary ovarian cancer cells, the algorithm predictions demonstrated a high correlation with observed ovarian cancer cell proliferation and CSC proportion. High correlation was maintained even in the presence of the EGF-like domain multiple 6 (EGFL6), a growth factor which changes ALDH+ cell asymmetric division rates and thereby tumor growth rates. Thus, based on sampling from the heterogeneity of in-vitro cell growth and division characteristics of a few hundred single cells, the simple algorithm described here provides rapid and inexpensive means to generate predictions that correlate with in-vivo tumor growth.