Investigating the Role of CD109 in Pancreatic Ductal Adenocarcinoma

The Texas Medical Center Library DigitalCommons@TMC

The University of Texas MD Anderson Cancer Center UTHealth Graduate School of The University of Texas MD Anderson Cancer Biomedical Sciences Dissertations and Theses Center UTHealth Graduate School of (Open Access) Biomedical Sciences

8-2019

Investigating the role of CD109 in Pancreatic Ductal Adenocarcinoma

MennatAllah Shaheen

Follow this and additional works at: https://digitalcommons.library.tmc.edu/utgsbs_dissertations

Part of the Biology Commons, Genetics Commons, Genomics Commons, and the Medicine and Health Sciences Commons

Recommended Citation Shaheen, MennatAllah, "Investigating the role of CD109 in Pancreatic Ductal Adenocarcinoma" (2019). The University of Texas MD Anderson Cancer Center UTHealth Graduate School of Biomedical Sciences Dissertations and Theses (Open Access). 971. https://digitalcommons.library.tmc.edu/utgsbs_dissertations/971

This Thesis (MS) is brought to you for free and open access by the The University of Texas MD Anderson Cancer Center UTHealth Graduate School of Biomedical Sciences at DigitalCommons@TMC. It has been accepted for inclusion in The University of Texas MD Anderson Cancer Center UTHealth Graduate School of Biomedical Sciences Dissertations and Theses (Open Access) by an authorized administrator of DigitalCommons@TMC. For more information, please contact [email protected]. INVESTIGATING THE ROLE OF CD109 IN PANCREATIC DUCTAL ADENOCARCINOMA

By:

MennatAllah Shaheen, B.Sc.

Approval Page APPROVED:

______Giulio F. Draetta M.D. Ph.D. Advisory Professor

______Swathi Arur Ph.D.

______Richard R. Behringer Ph.D.

______George T. Eisenhoffer Jr. Ph.D.

______Haoqiang Ying Ph.D.

APPROVED:

______

Dean, The University of Texas

MD Anderson Cancer Center UTHealth Graduate School of Biomedical Sciences INVESTIGATING THE ROLE OF CD109 IN PANCREATIC DUCTAL

ADENOCARCINOMA

Title Page A

Thesis

Presented to the Faculty of

The University of Texas

MD Anderson Cancer Center UTHealth

Graduate School of Biomedical Sciences

In Partial Fulfillment of the Requirements

of the Degree of

MASTER OF SCIENCE

By:

MennatAllah Shaheen, B.Sc.

Houston, Texas

August, 2018

Abstract

Investigating the Role of CD109 in Pancreatic Ductal Adenocarcinoma

MennatAllah Shaheen, M.S.

Advisory Professor: Giulio F. Draetta, M.D., Ph.D.

Pancreatic Ductal Adenocarcinoma (PDAC) is the 3rd leading cause of cancer death in the US. We performed loss of function genomic screening on a cohort of four patient derived PDAC cell populations and our data shows a cell surface receptor CD109 to be a common vulnerability, the biologic role of which in PDAC is yet unstudied and largely unknown. We hypothesized that CD109 expression provides PDAC cells with a survival advantage, and promotes cancer progression through activation of downstream signaling. We believe therefore that targeting CD109 could improve PDAC patients’ survival. Here we report that CD109 plays a role in cell proliferation, viability, and clonogenicity in vitro. We also find that it promotes tumor formation and progression in nude mice, therefore decreasing their survival. We revealed an association between

CD109 expression and YAP/TAZ signaling through RPPA and RNA Sequencing data.

This data establishes CD109 as a cell surface protein exclusively expressed in PDAC rather than healthy pancreatic tissue, demonstrating pro-oncogenic behavior and tumor initiation potential in vitro and in vivo. This helps us understand more about PDAC and provides insights into a relatively unknown protein with a therapeutic potential.

iii

Acknowledgements

I feel I can’t thank my mentor Dr Giulio Draetta enough for giving me the opportunity to be a part of his laboratory, to be under his kind knowing supervision, and most importantly making me feel welcomed that I am part of the lab family. It proved to be the best environment to learn, wrok, and make progress.

Whole hearted thanks to my advisory committee, Dr swathi Arur, Dr Richard Behringer, Dr

George Eisenhoffer, and Dr Haoqiang Yin, who were extremely supportive throughout my masters work, and until the very end were things were crazy. Their time, feedback, expertise, and flexibility were critical to the success of this project.

Thank you to Dr Draetta’s lab, especially Johnathon Rose who essentially taught me from scratch and directly mentored me even though he had his own projects to attend to. Thank you

Sanjana Srinivisan, and Melinda Seoung, for your help, support, and friendship. I am very happy to know you all.

Finally, thank you Dr Wantong Yao. You have been the voice of logic, and practicality. I am honored you have been on my oral defense committee.

I believe the best way to repay mentors is by benefiting from what they taught you, and passing on this knowledge. This way the circle of giving goes on forever.

Dedication

First, and foremost, thank You Allah for blessing me with this experience. I pray my knowledge will eventually benefit patients and can help make lives better.

This work, small as it is, is dedicated to my family, my parents Ahmed Shaheen, and Sahar

Sayed, and my siblings, Tuqa, Ziad, and Zeinab, who have sacrificed beyond imagination for what is best for us. I can write hundreds of pages expressing how thankful I am but it will never do justice to your efforts. Without you we wouldn’t be where we are.

My utmost love, and gratefulness, to my dear husband, Mustafa Hussein, who believed in me so much as to cross oceans and continents so I can become a better person. He made this an enjoyable adventure with his unconditional support, and love through good and hard times. I hope I made you proud, and I ma forever thankful to you.

Thank you to my friend, Menna Khedr for providing encouragement and advice from day one to graduation. You don’t know how much I appreciate it.

Finally, thanks to my Reema. You are our little family away from home. Our spark of happiness, and joy. And our motivation to be better people to be your parents.

Table of Contents

Approval Page ...... i

Title Page ...... ii

Abstract ...... iii

Acknowledgements ...... iv

Dedication ...... v

Table of Contents ...... vi

List of Figures ...... ix

List of Tables ...... x

1. Introduction: ...... 1

1.1 Loss of Function Genomic Screening ...... 1

1.2 CD109: A History ...... 2

1.3 CD109: Structure ...... 3

1.4 CD109 in Physiological Conditions ...... 5

1.5 CD109 in Cancer ...... 5

1.6 CD109 Mechanism of Action ...... 7

1.7 Project Summary and Hypothesis ...... 8

2. Materials and Methods ...... 9

2.1 Cell Lines ...... 9

2.2 Plasmid Production ...... 11

2.3 Lentivirus Production ...... 11

2.4 Rescue Model ...... 12

2.5 Inducible system ...... 12

2.6 Western Blot ...... 13

2.7 Colony Formation Assay ...... 13

2.8 Cell Cycle Analysis ...... 14

2.9 Apoptosis Assay ...... 14

2.10 Viability Assay ...... 14

2.11 Migration Assay ...... 14

2.12 Staining for CD109 at the Cell Surface ...... 15

2.13 In Vivo Tumorigenesis Assay ...... 15

2.14 Functional Proteomics by Reverse Phase Protein Assay (RPPA) ...... 15

2.15 RNA Sequencing ...... 16

2.16 Quantification and Statistical Analysis ...... 16

3 Results ...... 19

3.3 No effect on Migration in Vitro ...... 33

3.4 In Vivo Phenotype Tumor Initiation Potential ...... 34

3.5 CD109 is Differentially Expressed on the Cell Surface ...... 37

3.6 CD109 Mechanism of Action Involves Some Known Interactors ...... 43

4. Discussion and Future Directions ...... 47

4.1 Discussion ...... 47

vii

4.2 Future Directions ...... 50

Bibliography ...... 51

Vita ...... 57

viii

List of Figures

Figure 1.1 CD109 Structure and Post-translational Journey……………………………….4

Figure 1.2 CD109 is significantly overexpressed in PDAC compared to normal pancreas tissue………………………………………………………………………………………….…6

Figure 3.1 Results from Loss of Function (LOF) screens……………………...………….19

Figure 3.2 Colony Formation Assay…………………………………………………………23

Figure 3.3 Mutant CD109 is resistant to shCD109-3 knockdown………………………..25

Figure 3.4 Colony Formation Assay Partial Phenotype Rescue...... …...26

Figure 3.5 Viability Assay Phenotype and Rescue………………………………………...27

Figure 3.6 Cell Cycle Analysis and Apoptosis……………………………………………...29

Figure 3.7 In Vitro Migration Assay …………………………………………………………32

Figure 3.8 In Vivo Tumorigenesis Assay…………………………………………………....34

Figure 3.9 Titration of PE conjugated CD109 Antibody……………………………………37

Figure 3.10 Distribution of CD109 and CD133 in the Four Models………………………38

Figure 3.11 CD109 KD in HPNE……………………………………………………..………40

Figure 3.12 Reverse Phase Protein Array (RPPA)……………………………….…….….42

Figure 3.13 RNA Sequencing……………………………………………………………..…43

List of Tables

Table 2.1 Antibodies…………………………………………………………………………16

Table 2.2 Important Reagents and Kit……………………………………………………..17

1. Introduction:

Pancreatic Ductal Adenocarcinoma (PDAC) is the most common type of pancreatic cancer (90%) and the fourth leading cause of cancer related deaths in the world[1]. The most effective therapeutic approach is surgical resection; however, most patients are diagnosed with late-stage, unresectable disease. PDACs are highly recalcitrant tumors that respond poorly to standard-of-care chemotherapeutic regimens, and little progress has been made with the addition of targeted agents or immunotherapies, resulting in a 5-year survival rate of only 8% [2]. At least 90% of

PDACs are driven by mutant KRAS, which has made quenching the downstream effector proteins of this oncogene an attractive therapeutic target [3]. Unfortunately, to date, these approaches have been unsuccessful in the clinic, and there is an urgent need to identify new therapeutic targets that may disrupt cancer-essential, KRAS-driven functions in PDAC [4].

1.1 Loss of Function Genomic Screening

Our laboratory conducted loss of function (LOF) genomic screens in four early passage

PDAC models derived from patients at MD Anderson in both in vivo and in vitro settings.

The purpose of these loss-of-function RNAi screens was to functionally inform on a set of PDAC-prioritized genes located at the extracellular face of the plasma membrane

(the cell “surfaceome”). The screens were performed using a pooled barcoded lentiviral library on cell lines generated from patient-derived xenografts (PDXs). Timing and library coverage conditions were optimized. PDX models were established prior to screening, and were chosen to cover the molecular spectrum of PDAC from Classical to

Basal-like subtypes and were based on empty barcode library assays, i.e. barcode-only

1 libraries of the same size. A platform was optimized for two-step PCR-based barcode isolation for both in vitro and orthotopic in vivo screening. Functional targets were defined in the context of paired 2D and orthotopic in vivo conditions over at least two time points. The last time point (28 days) was used for the target described. This provides a look into how surface associated targets differ across the PDAC spectrum, and also which targets seem to be persistent. Targets that are persistent and unstudied, in the context of PDAC, provide a straightforward avenue to leverage for vulnerability and mechanistic insight. Redundant siRNA Analysis (RSA) which is a score based on fold change depletion relative to the reference barcode population, was used to calculate hit priority. The quality of the data is based on control separation, library coverage and biological replicate reproducibility. CD109 is one of the few targets that scored across all the models in the RNAi screens.

1.2 CD109: A History

In 1991 in an experiment to discover new antigens on the surface of primitive hematopoietic stem cells, CD109 was first recognized by monoclonal antibodies as a

170 KD glycoprotein expressed in CD34+ acute myeloid leukemia cell line, KG1a. In that study, CD109 was reported to mark a more primitive hematopoietic stem cell subset, its expression becoming more limited as these cells differentiate, eventually becoming undetectable in mature resting blood cells, and only expressed in a restricted pattern on the surface of activated platelets, activated T-lymphocytes and endothelial cells[5]. After that CD109 role in blood cells became an interesting question. Studies related CD109 to the α2 – macroglobulin/ complement family of thioester proteins and showed that it carries the human platelet antigen 15 (HPA-15 or Gov-a/b) which is

2 implicated in neonatal alloimmune thrombocytopenia (NAT), and post-transfusion purpura, but beyond that its physiological action in blood remained unclear[6-10].

1.3 CD109: Structure

CD109 protein consists of 1445 aa. It has a leader signaling peptide on the N-terminal, a GPI-anchor on the C-terminal as well as a thioester and a furinase cleavage sites

(figure1.1(a)). After translation, CD109 protein enters the endoplasmic reticulum for posttranslational modifications where it gets glycosylated and linked to glycosylphosphatidylinositol (GPI). Then CD109 is cleaved at the furinase site into two fragments in the golgi apparatus, and exported together to the cell surface as a complex, where it can also detach and be secreted in the medium (figure1.1 (b)).[11]

(a)

(b)

Figure 1.1 CD109 Structure and Posttranslational Journey

(a) Diagrammatic representation of CD109 protein.

(b) Sequence of events from CD109 RNA translation to surface expression and

extracellular secretion. (Hagiwara et al. 2010)

1.4 CD109 in Physiological Conditions

CD109 expression in normal tissues seems to be restricted to: myoepithelial cells of the mammary, lacrimal, salivary, and bronchial glands, bronchial epithelia and epidermis, seminiferous tubules of the testis, basal cells of the prostate, and osteoblasts and osteoclasts. Yet its function in each organ or tissue is not clear [12-14]. One study attempted a CD109 knockout mouse model (CD109-/-) and they found the mice to be viable but with abnormal skin and hair growth. The group later re-analyzed these mice by investigating their bones after reports of CD109 expression on pre-osteoclasts and its role in osteoclast formation in vitro. Indeed, they found that CD109 deficient mice have significantly less bone volume and more bone turnover compared to wild type mice, resulting in an osteoporosis-like effect. Both studies imply that CD109 might have a role in keratinocyte differentiation and normal bone turn over in humans[13, 14].

1.5 CD109 in Cancer

Much earlier than the CD109-/- mouse studies, reports on CD109 overexpression in cancer were published. CD109 is involved in glioma, glioblastoma, lung and breast cancer where it associates with a more mesenchymal form of disease, chemoresistance, poor prognosis and seems to factor in tumor initiation. CD109 overexpression is also detected in various sarcomas, and squamous cell carcinomas including the uterus, and esophagus, hepatocellular carcinoma, bladder urothelial carcinoma, and lately, pancreatic adenocarcinoma joins the list. A study using glycoprotein surface capture technique detected significant difference between CD109 expression in PDAC and in normal pancreas (figure1.2)[15]. Another study aimed at

5 developing specific monoclonal antibody against CD109 and the antibody could detect

CD109 in 96% of their cohort of 65 PDAC samples[16]. CD109 has been studied in

PDAC for a while now but its role in PDAC remains largely unknown. CD109 expression seems to be significantly higher in higher grade glioma, breast ductal carcinoma, non- small cell lung carcinoma, hepatocellular carcinoma, and soft tissue sarcomas, with the exception of urothelial carcinoma where higher expression of CD109 corresponds to lower grade disease [15-37]. Notably, CD109 is conserved across species, and is rarely found mutated in cancer according to TCGA data.

PDAC cells PDAC cells with weak CD109 with moderate CD109

PDAC cells Normal Duct Figurewith 1.2:stro n Cg D109CD10 9 is significantly(n e overexpressedgative) in PDAC compared to normal pancreas tissue

IHC scores are calculated according to the staining intensity as well as the percentage of cells that stain positive for CD109 as determined by a pathologist (Figure adapted from Huan R. et al 2014).

1.6 CD109 Mechanism of Action

Aside from all that is unknown about CD109, there seems to be significant conflict on its pathways of action. Initially a study showed CD109 downregulating TGF-β in keratinocytes in vitro, however the CD109-/- mouse didn’t show significant difference in smad2 phosphorylation, and instead stat3 phosphorylation increased [11, 13, 38, 39]. A few years later another group studying fibrosis and wound healing used transgenic mice overexpressing CD109 in the epidermis and noticed that they have better wound healing through TGF-β suppression [40-42]. A third group used CD109 deficient mice found that they have reduced skin tumorigenesis also by antagonizing TGF-β[43]. On the other hand, in lung adenocarcinoma a team using KP mice demonstrated lower lung metastases in CD109 knockdown cells through the enhancement of JAK/STAT3 pathway [34]. These different findings suggest that the role of CD109 might be tissue and context dependent.

1.7 Project Summary and Hypothesis

CD109 has been considerably studied both in physiological and pathological implications. Its controversial findings and association of CD109 with cancer initiation and/or progression has found fresh interest in the research community. My interest in

CD109 began from unbiased screening as well as the fact that it is overexpressed in pancreatic adenocarcinoma but its role in PDAC is virtually unstudied and unknown.

Based on our screen data, we hypothesized that CD109 expression is significantly increasing PDAC cells survival. We wanted to first investigate the phenotype of CD109 deficiency in PDAC both in vitro and in vivo, to evaluate its tumor initiation property and that the underlying molecular mechanisms are consistent with what was previously mentioned in the literature.

2. Materials and Methods

2.1 Cell Lines

PDAC models used are four human cell lines of early passaged xenografts derived from

PDAC patients in MD Anderson. The four models are PATC124 (Pancreatic

Adenocarcinoma Tumor Cell line), PATC69, PATC53, and PATC153. PATC53 is from a liver metastasis, and the rest are from the primary tumor. PATC53 and PATC153 are more mesenchymal like, while PATC124 and PATC69 are quasi-mesenchymal.

PATC124, PATC69 and PATC153 were maintained using DME/F12 growth medium

(HyClone) while PATC53 was maintained under Dulbecco's Modified Eagle's Medium

(DMEM) (HyClone). Growth media was supplemented with 10% Fetal Bovine Serum

(heat inactivated) (FBS-HI) (Gibco) and 1% Penicillin-Streptomycin (Pen-Strep) (Gibco).

Cell lines were subcultured when they reach 80%-90% confluency, and growth media changed twice a week.

Cas9 cell lines of PATC124, 69, 53, and 153 were established and kept under similar conditions to their parent cell line in addition to 1:100 blasticidin to select for the cas9 population.

Two hTERT HPNE cell lines, P52, and P51which are human pancreatic ductal cell lines were maintained under complete growth medium recommended by the ATCC website as follows:

• 75% DMEM without glucose (with additional 2 mM L-glutamine and 1.5 g/L

sodium bicarbonate)

• 25% Medium M3 Base

• To make the complete growth medium, add the following components to the

base medium:

• fetal bovine serum 5% (final conc.)

• 10 ng/ml human recombinant EGF

• 5.5 mM D-glucose (1g/L)

• 750 ng/ml puromycin

293T (ATCC® CRL-3216™) human embryonic kidney cell line was used for the purpose of lentiviral production and maintained under complete growth medium DMEM supplemented with 10% FBS-HI and 1% Pen-Strep. Cells are subcultured at about 60-

70% confluency and media changed twice a week.

The subculturing technique used for all cell lines is as follows:

• Remove and discard culture medium.

• Briefly rinse the cell layer with Dulbecco's phosphate-buffered saline (D-PBS).

• Add 0.25% Trypsin-EDTA solution and observe cells until cell layer is dispersed.

• Add complete growth medium and aspirate cells by gently pipetting.

• Transfer cell suspension to a centrifuge tube and spin at approximately 1100 rpm

for 4 minutes. Discard supernatant.

• Resuspend the cell pellet in fresh growth medium. Add appropriate aliquots of

the cell suspension to new culture vessels.

• Incubate cultures at 37°C.

2.2 Plasmid Production pLKO.1 vectors encoding the short hairpin RNAs (shRNAs) targeting CD109 mRNA

(shCD109-1, shCD109-2, shCD109-3, shCD109-4), shRNA targeting Luciferase as a negative control (shLuc-1), as well as single guide RNAs (sgRNAs) against CD109 gene (sgCD109-1, sgCD109-2), and sgRNAs against Luciferase as a negative control

(sgNC-1) were designed and produced by IACS facilities. Plasmids carry ampicillin resistance gene.

Plasmids were used to transform NEB® 5-alpha Competent E. coli (High Efficiency) according to the High Efficiency Transformation Protocol (C2987H/C2987I). After transformation colonies are selected on 1:1000 Ampicillin LB plates. Two different colonies are picked and incubated in 50 ml LB with 1:1000 ampicillin for 12-14 hours at

37°C under rotation. Plasmids are then collected using the QIAGEN Plasmid Plus Midi

Kit and according to the protocol, quantified using the NanoDropTM (Thermofisher), and finally stored in -20°C.

2.3 Lentivirus Production

We used a lentivirus delivery system to introduce shRNAs for CD109 gene knockdown.

293T cells were co-transfected with the pLKO.1 vector encoding the shRNA and the helper plasmids for virus production (psPAX2and pMGD2) using Lipofectamine 3000 according to the Lipofectamine 3000 protocol by Invitrogen.

Lentivirus was collected 72 hr after transfection and is then either spun down in the ultracentrifuge resuspended in 100 ul PBS, and stored at -80°C or aliquoted, stored at

4°C and used within one week.

The lentiviral backbone has puromycin resistance gene and the PATC cell lines were titrated against a range of puromycin concentrations to determine the optimum selection concentration which was 1:5000. The cells were also titrated against the lentivirus to optimize virus concentration used for transduction. Lentivirus infection was performed using 1:1000 polybrene and after 24 hours cells were washed and selected in medium with 1:5000 puromycin.

2.4 Rescue Model

We established a CD109 overexpressing line in PATC153 which originally has relatively low endogenous CD109 expression. We transduced PATC153 with a gateway adapted entry vector containing resistant CD109 ORF (referred to here as pHAGE CD109m).

The ORF is resistant to one of the hairpins (shCD109-3) and was developed by site- directed mutagenesis (PCR-based strategy), and multiple synonymous mutations across the binding site were included to reduce the binding efficiency of the shRNA. The gateway is based on the pHAGE Gateway system. A separate PATC153 is transfected with pHAGE-GFP as a control for the transduction process.

2.5 Inducible system

We attempted to produce conditional cell lines using a doxycycline inducible RNAi one vector system pGLTR-X (Addgene) targeting CD109. It was a tetracycline on system

(rtTA). A multitude of doxycycline concentrations from 1:1000 to 1:100, and sampling timepoints from 24 hr to 144 hours after the addition of doxycycline were tested, however the results were less than satisfactory. Inducible CD109 knockdown was not observed at all with ishCD109-2 construct and minimally achieved with ishCD109-3 after 72 hours post doxycycline. The cells were seeded under doxycycline for colony

12 formation and viability assays, compared to cells seeded without doxycycline and there was no significant difference. We finally resorted to the constitutional knockdown system.

2.6 Western Blot

Proteins were extracted from collected cell pellets in the presence of protease and phosphatase inhibitors, and quantified using the Bradford method. Western blot analysis was then performed using the standard protocols.

We used 20 ug protein for loading, run the gel at 90 v, and used wet transfer with nitrocellulose membrane. After the transfer the membrane was then blocked in 5% milk suspended in 1% TBST for one hour at least.

Primary antibody is anti CD109 mouse monoclonal antibody (1:100 in 5% milk TBST), incubated with the membrane at 4°C overnight, then washed with 1% TBST three times before incubating with LICOR secondary antibody mouse 800 nm (1:10000 in Odyssey buffer). After three more washings with 1% TBST the membrane is scanned with

LICOR.

2.7 Colony Formation Assay

Cell lines were transduced with each of the CD109 shRNAs as well as the Luciferase shRNA, and left to select under puromycin until the puromycin control plate is wiped.

After selection cells are collected and counted with the hemocytometer. For each condition in triplicates 1500 cells are seeded per well in a 6 well plate, and left to form colonies for 3 weeks. Colonies are then fixed with 6.0% v/v paraformaldehyde, stained with 0.5% w/v crystal violet and visually compared and/or counted for quantification.

2.8 Cell Cycle Analysis

After selection PATC cells were incubated with BrdU and cell cycle assay was performed as described in the cell cycle analysis protocol using the BD PharmingenTM

BrdU Flow Kit. Cell cycle analysis was performed by the South Campus Flow Cytometry

Facility. For each measurement at least 10000 events were captured. Data analysis was performed using Flowjo software program.

2.9 Apoptosis Assay

After selection we followed the FITC Annexin V and Propidium Iodide Apoptosis

Detection Kit I protocol described by BD PharmingenTM. Uninfected sample was exposed to heat shock as a positive control for apoptosis, and untreated sample as a negative control for apoptosis.

2.10 Viability Assay

After selection cells from each condition were collected and counted. 2000 cells times five replicates per condition were seeded in 96 well plate and the CellTiter-Glo®

Luminescent Cell Viability Assay protocol was followed.

2.11 Migration Assay

After selection PATC124 cells were starved for 24 hours in serum free DME/F-12 medium, harvested using 0.05% Trypsin and proceeded as described in QCMTM 24-

Well Colorimetric Cell Migration Assay protocol. We compared the migration observed with untreated PATC124 and the negative control against the CD109 knockdown conditions in the presence and absence of serum in the bottom chamber.

2.12 Staining for CD109 at the Cell Surface

We followed the Flow Cytometry Protocol for Staining Membrane-associated Proteins in

Suspended Cells described by R&D Systems. We used human PE-conjugated anti-

CD109 by R&D for staining and performed a titration experiment to optimize antibody concentration and ensure we capture all CD109 signal available.

2.13 In Vivo Tumorigenesis Assay

This assay was performed on PATC124. Cells were transduced with either shCD109-2, shCD109-3, or shLuc-1. After selection cells were collected and counted. One million cells times five replicates of each condition was resuspended in 100 ul PBS plus 100 ul matrigel and injected subcutaneously in 14-weeks old female nude mice that were randomly assigned into three experimental groups of five mice each. We followed up the formation of tumors in each condition as well as the size and survival over a period of 8 weeks. Animal experiments were performed at MD Anderson Cancer Center

Animal Facility and followed approved Institutional Animal Care and Use Committee protocol 884-RN02.

2.14 Functional Proteomics by Reverse Phase Protein Assay (RPPA)

To compare CD109 overexpression against endogenous CD109 we transduced

PATC153 pHAGE-GFP and PATC153 pHAGE-CD109m with the two working CD109 working hairpins as well as the negative control hairpin. After selection three biological replicates of each condition were collected and pelleted over a timecourse (48, 96, and

144 hr). Pellets were stored in -80°C until sample submission to the MD Anderson

RPPA Core Fcaility.

Statistical analysis of the normalized log2 intensities was conducted with the Limma package in R using the lmFit and eBayes function.

Hairpins shCD109-3 for pHAGE-CD109m vs pHAGE-GFP were compared and differential proteins with p < 0.01 are reported (rescue protein expression in the context of shCD109-3)

2.15 RNA Sequencing

We stained three biological replicates of untreated PATC124 cells for CD109 at the surface and sorted CD109 negative and CD109 positive populations using FACS. RNA from each sorted sample was extracted using the QIAGEN RNeasy Midi Kit and following described kit protocol. After elution RNA content was calculated using

Nanodrop. Samples were submitted for Sanger sequencing at the MD Anderson

Sequencing and Microarray Facility (SMF).

Cd109 positive and negative populations were analyzed with DESeq using raw counts.

Significantly deregulated genes were identified using an adjusted p value of 0.05 and an absolute log2foldchange greater than 1. GSEA preranked test was run on the t scores obtained from the DESeq analysis.

2.16 Quantification and Statistical Analysis

P value was calculated in Microsoft Excel using a two-tailed Student t test. Cell culture experiments were done with an n ≥ 3. In vivo studies have been performed using n = 5 mice/condition. Error bars in graphs represent the mean ± SD.

Table 2.1 Antibodies

Name Company Notes

CD109 (C-9) mouse Used for western blotting in monoclonal antibody Santa Cruz Biotechnology 1:100 concentration.

β-Actin mouse monoclonal antibody Santa Cruz 1:10000

Used for staining cell

surface CD109 and for

Human PE-conjugated anti- FACS analysis and sorting.

CD109 R&D (5 ul per 250000 cells)

Used for staining cell

surface CD133 and for

Human APC-conjugated FACS analysis. (5 ul per anti-CD133 R&D 250000 cells)

IgG control for PE-

Mouse IgG2A Control PE conjugated CD109. (5 ul

Conjugated R&D per 250000 cells)

Mouse IgG2b kappa IgG control for APC- isotype control APC conjugated CD133. (5 ul conjugated Invitrogen per 250000 cells)

Fc Block (5 ul per 1 Million

Human TruStain FcXTM Biolegend cells)

Table 2.2 Important Reagents and Kits

Name Company

Lipofectamine 3000 Invitrogen

Accutase Stem Cell Technologies

Medium M3 Base Incell Corp

3 Results

3.1 CD109 is a Common Vulnerability Among the PDAC Models

Loss of function genomic screens are an unbiased method of target discovery. Results from our in vitro and in vivo loss of function screens performed on four different patient derived PDAC cell populations show that CD109 is one of only five hits/vulnerabilities that were common between the eight screens. Interestingly it scores alongside ERBB2,

ERBB3 (HER/EGFR family) which drive tumor growth in a significant percentage of pancreatic cancers [44] (figure3.1)

The first thing we noticed from the screens is the differential vulnerability of CD109 across the four models (figure3.1). We wanted to see if the degree of vulnerability correlates with the amount of CD109 protein in each model but it doesn’t quite align as seen from western blot (figure3.1).

(a Surfaceome Screens: PATC69 in vivo vs. in vitro (Day 28) Surfaceome Screens: PATC124 in vivo vs. in vitro (Day 28) RSA Comparison RSA Comparison ) 0 LUC 0

ZDHHC5 ATP1A1 CNIH4 EFNA5 CSF1 -1 MSLN TSPAN1 -1 ITGA3 TMPRSS4 CADM1 EPSTI1 LUC TFRC PVRL3

screens PTPRA

screens PLAUR EPHB4 ERBB3 TMEM248 -2 F11R ANPEP F11R MET ANPEPTMEM248 in vivo in vivo TSPAN15 -2 RSA (LOGP)

RSA (LOGP) LAPTM4B CLDN1 CD109 ITGAV CTTN GPR87 ERBB3 PATC69

CDH1 PATC124 STX6 SLC39A4 -3 ERBB2 SLCO4A1 IGF1R SEMA4B SPINT2 -6 ATP6AP2 -3 CD109 -9 -12 -15 -15 -12 -9 -6 -3 -2 -1 0 -3 -2 -1 0 PATC69 in vitro screens PATC124 in vitro screens RSA (LOGP) RSA (LOGP)

Surfaceome Screens: PATC53 in vivo vs. in vitro (Day 28) Surfaceome Screens: PATC153 in vivo vs. in vitro (Day 28) RSA Comparison RSA Comparison 0 0 LUC PLAUR ITGB5 ATP1A1 LUC SLC20A1

AMIGO2 BLCAP SLC12A7 -1 -1 PVRL2 CXADR PVRL2 TMEM183A ITGB5 LAPTM4B TFRC CD109 SLCO4A1 ITGB1 HYAL2 screens screens ATP6AP2 SLC39A4 CD109 ITGAV SCAMP3 ERBB2 ITFG1 KCNK6AMIGO2 -2 F11R -2 YIF1B TM9SF4 PROCR TM9SF4 CLDN1 ATP6AP2 in vivo FAM171A2 in vivo ANPEP TPBG

RSA (LOGP) ERBB2 HYAL2 TTYH3 RSA (LOGP) EPHB4 ITGB1 EPHB4 PROM1 CSF1 ANTXR2 ITGAV SLC20A1 PATC53 FAM171A2 PATC153 -3 DSC2 -3 CMTM1 EPSTI1 -4 -5 -4 -6 -5 -7 -8 -6 -8 -7 -6 -5 -4 -3 -2 -1 0 -6 -5 -4 -3 -2 -1 0 PATC53 in vitro screens PATC153 in vitro screens RSA (LOGP) RSA (LOGP)

(b) (c)

4 3 9 3 2 5 6 5 1 1 C C C C T T T T A A A

(d) A P P P P

CD109 150 kDa

37 kDa GADPH

Figure 3.1 Results from LOF screens on PATC124, PATC 69, PATC 53, and

PATC153 in vitro and in vivo

(a) Representation of the in vivo and in vitro results for each model as an RSA score.

Genes that score within the horizontal rectangle area are a vulnerability in vivo, and the genes scoring within the vertical rectangle are a vulnerability in vitro. The circle represents CD109 position in each plot.

(b) and (c) Table and Vinn diagram representation of the common hits between models

(d) Western blot showing different levels of CD109 expression between the four PDAC models.

3.2 CD109 influence on PDAC cell proliferation in Vitro

Clonogenic assays indicate the ability of a cell to divide and produce a colony in vitro.

We assayed colony formation in all four models using 3 different working hairpins against CD109 and comparing to the negative control hairpin. Knockdown was assessed using western blot, which shows that shCD109-2, and shCD109-3 hairpins are much more efficient at eliminating CD109 than shCD109-4. All four models show reduction in colony formation capacity that correlates to the level of knockdown of

CD109 (figure 3.2). Based on these results I focused on PATC124 since it had the highest expression levels of CD109 and is the most vulnerable to its depletion. On the other hand PATC153 exhibits the lowest expressing and least dependency on CD109. I also excluded shCD109-4 because of its inefficient CD109 knockdown and only used shCD109-2 (most efficient CD109 knockdown), and shCD109-3 (second most efficient) besides the negative control hairpin shLuc-1. The shCD109-4 hairpin mainly served as proof of correlation of phenotype to the level of CD109 knockdown.

We used PATC153 pHAGE-CD109m line which expresses a mutant form of CD109 resistant to shCD109-3 and compared its colony formation phenotype against PATC153 pHAGE-GFP control. This approach showed a partial rescue of phenotype that was statistically significant (figure 3.3 and 3.4).

Next I wanted to further assess cell ability to proliferate in the absence of CD109, this time using viability assay. I used CellTiter-Glo® which is a luminogenic ATP assay and again CD109 deficient cells are less viable, the phenotype is less pronounced in

PATC153 than PATC124, and is rescued in the PATC153 pHAGE-CD109m line (figure

3.5).

Having validated that there is a proliferation phenotype attributed to CD109 in our PDAC models, we proceeded to evaluate cell cycle progression and apoptosis in PATC124 and PATC153. Both models seem to stall in G1 phase of the cell cycle in the CD109 knockdown conditions, but I observed no difference in the rate of apoptosis (figure 3.6).

d 3 e - t 9 c e 0 f (a) 1 n I D - C n h o PATC124 s N

- 1

9 - 0 c 1 u L D h C s h

s 2 3 4 - - -

9 9 9 1 - 0 0 0

c 1 1 1 2 u - D D D 9 L C C C 0 (b) h s h h h 1 s s s D C h s 250 CD109 1K5D0a KDa

β-Actin 42 KDa

d 3 e -

s PATC69 N

- 1 9 - 0 c 1 u D L h C s h s

2 3 4 - - -

9 9 9 1 0 0 0 -

1 1 1 c 2 D D u D - 9 L C C C 0 h h h (d) h s s s 1 s D C

h 250 s CD109 1K5D0a KDa

β-Actin 42 KDa

d e - t 9 c e 0 f 1 n I D

(e) - C n

h o

s 3 N PATC153

- 9 - 0 c 1 u D L

h C s 1

h s 4

4 3 2 - - - 1 - 9 9 9 - c 0 0 0 9 u 1 1 1 0 (f) L 1 D D D h s C C D C h h h C s s s

h s 2 CD109 150 KDa

42 β-Actin KDa

(g)

d e - t 9 c e 0 f 1 n I D - C n

h o s 3 N PATC53 - 9 - 0 c 1 u D L

h C s 1

h s 4 - - - 9 9 9 0 0 0 - 1 1 1 c - D D D u 9 C 0 C C L

1 h h h s 4 s 3 s 2 s 1 D

C (h)

h 250 s 2 CD109 1K5D0a KDa

42 β-Actin KDa

Figure 3.2 Colony Formation Assay

Scan of 6-well plates showing the effect of CD109 depletion on each of PATC124, PATC69, PATC153, and PATC53 using 3 different shRNAs against CD109 gene, along with western blot demonstrating the level of knockdown.

(a) PATC124 colony formation assay results.

(b) Western blot evaluating CD109 knockdown in PATC124 using the 3 hairpins and

control.

(d) Western blot evaluating CD109 knockdown in PATC69 using the 3 hairpins and

control.

(e) PATC153 colony formation assay results.

(f) Western blot evaluating CD109 knockdown in PATC153 using the 3 hairpins and

control.

(g) PATC53 colony formation assay results.

(h) Western blot evaluating CD109 knockdown in PATC53 using the 3 hairpins and

control.

Rescue Synonymous mutation: shCD109-3

resistance (CD109m)

PATC153 PATC153 pHAGE-GFP pHAGE-CD109m ½

N N N s o s o s 3 s o h n h h h n

n C C - C C - - D I I s s I D D D n n n h h 1 f 1 1 1 f f L L e 0 e e 0 0 0 u c u 9 c c 9 9 9 t c c t t - e - e - - - e 3 - 2 d 2 1 1 d d *

Figure 3.3 Mutant CD109 is resistant to shCD109-3 knockdown

PATC153 pHAGE-CD109m is a CD109 overexpressing line which expresses a

resistant ORF to shCD109-3 due to multiple synonymous mutations across the binding

site.

The western blot confirms overexpression and resistance to shCD109-3. PATC153

pHAGE-GFP is a control for the transduction process.

(a)

Non-Infect ed shCD109-2

shLuc-1 shCD109-3* PATC153 pHAGE-GFP

shCD109-2 shCD109-3*

Non-Infected shCD109-2

shCD109-3* PATC153 shLuc-1 pHAGE-CD109m

shCD109-2 shCD109-3*

(b)

180 161.5 160 140 120 100 77.5 80 60 40 20 0 C3 Cm C3 GFP

Figure 3.4 Colony Formation Partial Phenotype Rescue in PATC153

(a) A scan of the wells shows partial rescue of the colony formation phenotype observed with shCD109-3. (b) Colonies were counted for both PATC153 pHAGE-CD109m shCD109-3 knockdown wells (C3 Cm) and PATC153 pHAGE-GFP shCD109-3 wells (C3 GFP). Statistically significant difference using paired t test (p=0.001).

120 (a) 120

100 100

80 80

60 60

40 40

20 20

0 0 UT L1 C2 C3 UT L1 C2 C3 PATC124 PATC153

PATC153 - CD109m Rescue Viability (%) vs. Time

1.6 1.4 (b)

GFP) GFP) 1.2 - 1 0.8 0.6 0.4 0.2 0 Infected (pHAGE Infected 1 1 2 3 2 3 ------Viability (%) Viability shLuc shLuc Infected Infected Infected - - shCD109 shCD109 shCD109 shCD109 Non Non

Relative to Non to Relative pHAGE-GFP pHAGE-CD109m

72 hours 120 hours

Figure 3.5 Viability Assay Phenotype and Rescue

(a) Bar charts representing the relative viability as a percentage of untreated (non infected) condition in PATC124 and PATC153. (b) Viability phenotype rescue in PATC153 pHAGE-CD109m.

Untreated UT, shLuc-1 L1, shCD109-2 C2, shCD109-3 C3

Non-Infected shLuc-1 shCD109-2 shCD109-3

d 2 3 e - t - c 9 9

e 0 0 1 Non-Infected Heat Shock Positive Control f 1 - 1 n c I D D - u C n C L h o h h s s s N

CD10 9

GADP H

shLuc-1 shCD109-2 shCD109-3 Non-Infected shLuc-1 shCD109-2 shCD109-3

2 3 - -

1 Non-Infected Heat Shock Positive Control - Infected - shCD109 shCD109 shLuc Non

CD109

GADPH

shLuc-1 shCD109-2 shCD109-3 Figure 3.6 Cell Cycle Analysis and Apoptosis in PATC124 and PATC153

CD109 knockdown in PATC124 and PATC153 results in stalling in the G1 phase of the cell cycle, and doesn’t affect apoptosis in vitro.

3.3 No effect on Migration in Vitro

CD109 in glioma seems to promote invasion and metastasis. To test this idea in our

models I performed a migration in vitro experiment using trans-well technique an, but

we didn’t observe any significant difference (Figure 3.7).

120 140

100 120 100 80 80 60 60 40 40

20 20

0 0 UT L1 C2 C3 UT L1 C2 C3

PATC124 PATC153

Figure 3.7 Migration Assay in PATC124 and PATC153

No significant difference

3.4 In Vivo Phenotype Tumor Initiation Potential

Now it was evident CD109 deficiency reduces PDAC cell viability in vitro by stalling cell cycle in the G1 phase, rather than direct induction of cell death. Sometimes an in vitro phenotype doesn’t hold true in vivo which is why it was important to see if the absence of CD109 significantly interferes with PDAC tumor formation and growth. We attempted an inducible CD109 knockdown system and tested it in vitro in PATC124, but it didn’t result in sufficient knockdown of CD109 to produce phenotype (figure 3.8 (a) and (b)).

We used the constitutive CD109 knockdown to inject nude mice randomly assigned to shLuc-1, shCD109-2, or shCD109-3 groups. Each group consisted of 5 mice and we observed that four out of five mice in the negative control developed tumors by week 5, and all four reached ≥100 mm3 tumors and were euthanized by week 8. In the shCD109-2 group none of the mice formed tumors, and in shCD109-3 three out of five mice developed tumors, but only one of them reached the 100 mm3 threshold and was euthanized (figure 3.8 (c)).

(a) (b)

3 120 hr 120 3 3 72 hr 72 3 3 120 hr 120 3 - 3 72 72 3 2 120 hr 120 2 - - 2 72 hr 72 2 2 120 hr 120 2 - - 2 72 hr 72 2

- - - ishLuc-1 1 72 hr 72 1 hr 72 1 1 120 hr 120 1 1 120 hr 120 1 - - - - shCD109 ishCD109 shCD109 ishCD109 ishCD109 shLuc shCD109 shCD109 Untreated ishCD109 ishLuc ishLuc shLuc β-Actin ishCD109-2

CD109 ishCD109-3 PATC124

(c)

Survival proportions: Survival of Three groups

100 shLuc-1 shCD109-2 shCD109-3

50 Percent survival

0 0 20 40 60 80 Days

Figure 3.8 In Vivo Tumorigenesis

(a) Western blot showing PATC124 inducible CD109 knockdown compared to

constitutive CD109 knockdown at 72 and 120 hr. (b) colony formation assay using

inducible CD109 system in PATC124. (c) Kaplan-Meier plot showing the survival

rate of mice after being injected with either PATC124 shLUC-1 cells, shCD109-2

cells, or shCD109-3 cells.

3.5 CD109 is Differentially Expressed on the Cell Surface

In glioma CD109 was shown to mark a subset of cancer stem-like cells that were separate from CD133 cancer stem-like cells. To check these results in PDAC I stained all four models PATC124, PATC69, PATC53, and PATC153 cells for both CD109 and

CD133 and analyzed them by flow cytometry. Interestingly not all the cells within each model were expressing cell surface CD109 and CD133 but it seemed there were two separate subpopulations expressing either one. Also the percentage of CD109 positive cells in the four models corresponded to the level of expression of that model, where

PATC124 has almost 60% CD109 positive cells, and PATC153 has less than 3%

CD109 positive cells (figure 3.10). We sorted CD109 positive and CD109 negative cells to check protein levels using western blot and the CD109 positive population has much higher CD109 content as expected, but at the same time CD109 negative cells still showed a faint band of CD109 protein, knowing that the ratio of antibody to number of cells has been titrated and so this is not due to an under saturation of the cells with the antibody (figure 3.9). We sorted and recultured CD109 positive and CD109 negative cells separately for eight days which corresponds to eight cell doubling times, then reanalyzed the percentage of CD109 positive and negative in each. As seen in figure

3.10 (d) both populations come back to the original CD109 positive to negative ratio.

The finding that we have about 40% of CD109 negative population in PATC124 and still demonstrate such a profound cell proliferation phenotype suggests that either CD109 positive cells have a direct influence on the surrounding cells or that CD109 is being expressed internally in CD109 negative cells and still playing a role in regulating cell proliferation. In order to rule out the possibility of off-target effects from the two working

37 hairpins I checked CD109 expression in normal pancreatic cell lines hTERT HPNE P52 and P51. As seen from figure 3.11 there is almost no CD109 expression detected with these lines, so we transduced HPNE 52 with shLuc-1, shCD109-2, and shCD109-3, and seeded them to evaluate colony formation which showed no significant difference.

Unstained IgG control – 1.25*105

0.1% 0.2%

CD109 PE – CD109 PE – CD109 PE –

53.2% 36.8% 53.6%

Figure 3.9 Titration of CD109 PE conjugated antibody against the number of cells

The amount of antibody used is kept constant, and used to stain 500000, 250000, or

125000 cells. Saturation is achieved at 250000 cells.

(a) PATC69 PATC124 PATC53 PATC153

3 9 3 - 2 + 5 6 5 9 1 9

1 0 C C 0 C C 1 1 T T T T D A A D A A P P C P P C (b) (c)

CD109 150 kDa

37 kDa GADPH

(d) 64.8 57.8 % %

CD109 positive CD109 negative

Figure 3.10 Distribution of CD109 and CD133 positive cells in the four models

(a) FACS analysis of each model according to CD109 and CD133 staining. Green =

Untreated, Red = CD109 + CD133, Blue = IgG2A (CD109 antibody control),

Orange = IgGB2 (CD133 antibody control )

(b) Western blot showing PATC124 CD109 positive against CD109 negative cells.

(d) FACS analysis of sorted CD109 positive and CD109 negative cells after four

doublings.

(a) (b)

shLuc-1 CD109

shCD109-2 β-Actin

1 2

5 5 4 P 2 P

1 E E C N N T P P shCD109-3 A P H H

hTERT HPNE P52

Figure 3.11 CD109 knockdown in HPNE

(a) Colony formation assay in HPNE P52 shows no difference when treated with

shCD109-2, shCD109-3, than with shLuc-1.

(b) Western blot detecting CD109 in PATC124 and not in HPNE P52, or P51.

3.6 CD109 Mechanism of Action Involves Some Known Interactors

We used reverse phase protein lysate array (RPPA) to compare the signaling between knockdown and rescue of CD109 in PATC153 overexpressing line p HAGE-CD109m and the control PATC153 pHAGE-GFP. The top scoring proteins which were enriched in the CD109 overexpressing condition included pYAP, total YAP, pRb, pHER3, and others as seen in the table from figure 3.12.

We also looked into differential mRNA expression between freshly sorted CD109 positive and CD109 negative cells from three different PATC124 samples(figure 3.13

(a)). CD109 levels were confirmed by western blot. RNA sequencing data is still under further analysis and validation but we were able to detect enrichment of YAP signature in the CD109 positive group.

(a) (b)

shLuc-1 shCD109-2 shCD109-3 PATC153 RPPA - 144 hours

Differential Signaling between Knockdown and Rescue of CD109

48 hours 96 hours 96 hours 144 hours 48 hours

Gene (CD109m/GFP) log2FC Fold Change Untreated 48 hours 96 hours 144 hours 144 hours YAP_pS127-R-V 1.08 2.12 Jagged1-R-V 0.96 1.95 Rb_pS807_S811-R-V 0.49 1.40 CD109 Src-M-V 0.45 1.37 YAP-R-C 0.41 1.32 CDK1-R-C 0.18 1.13 Connexin-43-R-C 0.16 1.11 Β-Actin HER3_pY1289-R-C 0.13 1.09 WIPI2-R-C 0.13 1.09 Stat5a-R-V 0.10 1.07 RBM15-R-V 0.08 1.06 shLuc-1 shCD109-2 shCD109-3 MSI2-R-C 0.07 1.05 HES1-R-V 0.06 1.04

Myt1-R-C 0.04 1.03 E-Cadherin-R-V 0.03 1.02

TUFM-R-V 0.01 1.01 hours 48 hours 96 hours 144 hours 96 hours 48 hours 48 hours 144 hours 96 hours Untreated 144

CD109

Β-Actin

Figure 3.12 RPPA Knockdown –PATC153 (GFP/CD109m)

(a) Table arranging the top scoring proteins from the RPPA results in order of highest fold change. (b) Western blot confirming CD109 knockdown in submitted samples.

(a)

(b) (c)

(d)

Figure 3.13 RNA Sequencing Analysis of PATC124 CD109 positive and CD109

negative cell populations

(a) Flow cytometry sorting data showing the chosen separation between CD109

positive and negative signals in three different samples.

(b) Western blot of the sorted populations.

(d) Enrichment plots of some of the captured signaling.

4. Discussion and Future Directions

4.1 Discussion

Pancreatic adenocarcinoma is considered the most fatal common cancer nowadays.

PDAC is mostly diagnosed at an advanced stage when the disease has already metastasized and surgical resection is no longer an effective option. It is largely resistant to chemotherapy and has remained within the same 8% 5-year survival rate for decades now despite nonstop research efforts. In this study I investigate CD109 cell surface protein based on our preliminary data from loss of function LOF RNAi screen findings on PDAC models. CD109 has been implicated in various types of cancers but it had not been studied in the context of PDAC yet. I focused on validating and specifying a phenotype of CD109 deficiency in PDAC cells in vitro using an shRNA CD109 knockdown system. Our results show significant reduction in colony formation capacity and viability across the four PDAC models in the absence of CD109 expression, a reduction that corresponded to the actual level of CD109 knockdown. A partial rescue of both the colony formation phenotype and the viability phenotype further support this observation. To further validate the specificity of the phenotype to CD109 we transduced a normal pancreatic ductal cell line hTERT HPNE P52 which doesn’t express CD109 protein with our set of CD109 targeting hairpins shCD109-2, and shCD109-3, and shLuc-1 as a negative control. When we assayed the cells for colony formation we couldn’t detect any significant difference. Together this data suggests that

CD109 is influencing the survival of PDAC cells in our models.

Next, we questioned whether this reduction is a result of induced cell death or not, so we evaluated cell cycle progression, and the rate of apoptosis in the CD109 knockdown

47 condition compared to the negative control. Cell cycle appeared to be stalling in the G1 phase, while there seems to be no effect on apoptosis. This means that CD109 is not directly causing cells to die, and might be supporting the TGF-β pathway theory.

Previous studies on glioma and lung cancer have shown that CD109 is pro-metastatic.

We tested this in our PDAC models in vitro using migration assays and we couldn’t capture a significant difference on that front. This pro-metastatic property might not be significant in PDAC [24, 34, 36, 37].

A recent study in glioma has also revealed that CD109 is marking a certain population of cancer stem like cells which is mutually exclusive with CD133 positive cancer stem like cells[37]. We wanted to check if this applies to our models and found that there is a specific percentage of CD109 expressing cells in each PDAC model, that correlates to the amount of CD109 protein detected on western blot, but when we sorted the CD109 negative populations and ran them on western blot we were able to detect a small level of CD109 protein present. So even though CD109 is not detected on the surface of these cells it is still minimally expressed. This might explain why we didn’t see this

CD109 negative population in our previous assays. We decided to sort the positive and negative populations of the highest CD109 expressing line PATC124 and reculture them separately. After about four doubling times, both CD109+ and CD109- populations eventually produce almost the same ratio of CD109+ to CD109- as their parent line.

This means that CD109 protein is not exclusively present at the surface of PDAC, but can also be found in the cytoplasm of the cells, and in this case cells that are negatively staining for cell surface CD109 are still capable of CD109 expression and might have intracellular CD109 that is functionally active. It also implies the presence of a critical

48 homeostasis of CD109 expression allowing PDAC cells to control the amount and percentage of CD109. This is in contrast to what was published recently about the higher hierarchy of CD109+ cells in glioma and the inability of CD109- cells to give rise to CD109+, while the opposite is possible [37].

We have also attempted an in vivo validation of the phenotype, by injecting nude mice with either CD109 knockdown cells or the negative control and noticed that none of the mice injected with shCD109-2 knockdown cells formed tumors, compared to 80% of the negative control group. This experiment is still ongoing in order to determine survival differences between the two groups and to obtain IHC data.

Finally we wanted to investigate the interactors and pathways CD109 might be acting through in PDAC. There has been conflicting data on this aspect in the literature, suggesting that signaling downstream of CD109 are probably context and/or site dependent. We performed two different experiments to answer this question, first was to look at the differential signaling between knockdown and rescue of CD109 protein through RPPA which detects differences in protein expression. Our data notably show significant changes in YAP and pYAP. CD109 involvement in the YAP/TAZ pathway in glioma has also been recently reported, and supports the tumor initiation property of

CD109.

The second experiment was to sort CD109 positive and CD109 negative cells and compare their differential mRNA expression, which this time was able to capture TGF-β enrichment as well as YAP and pRb.

4.2 Future Directions

Although we were able to validate the pro-oncogenic effect of CD109 expression in

PDAC cells, and show an insight into possible downstream signaling, further work needs to be done and is actually ongoing to specify CD109 interactors in PDAC. We have successfully established a flag tagged CD109 that is expressed on the surface and we are currently in the process of doing a pilot Co-Immunoprecipitation/MS experiment to get a more solid idea on what proteins are directly binding to CD109.

We also want to test the translational significance of CD109 as a therapeutic target.

Less than a year ago, a group reported failure of CD109 to internalize following monoclonal antibody binding[16]. We noticed, however, the association of CD109 with

HER-3 in our RPPA data as well as data from the screens. There is an exciting possibility that designing a bi-functional antibody targeting both CD109 and a relevant cell surface receptor such as HER-3 could enhance the internalization of both and consequently reduce PDAC cell survival.

Finally, the role of CD109 as a secreted protein as well as a cell surface marker is too important to ignore, especially in the case of PDAC. Further investigations are warranted to provide insight on the timing and stage at which CD109 is detectable or overexpressed in PDAC as well as in other CD109 overexpressing tumors and the potential of CD109 as a new biomarker for these tumors.

Bibliography

1. Ryan DP, Hong TS, Bardeesy N: Pancreatic adenocarcinoma. N Engl J Med 2014,

371(22):2140-2141.

2. Adamska A, Domenichini A, Falasca M: Pancreatic Ductal Adenocarcinoma: Current

and Evolving Therapies. Int J Mol Sci 2017, 18(7).

3. Chiaravalli M, Reni M, O'Reilly EM: Pancreatic ductal adenocarcinoma: State-of-the-

art 2017 and new therapeutic strategies. Cancer Treat Rev 2017, 60:32-43.

4. McGuigan A, Kelly P, Turkington RC, Jones C, Coleman HG, McCain RS: Pancreatic

cancer: A review of clinical diagnosis, epidemiology, treatment and outcomes.

World J Gastroenterol 2018, 24(43):4846-4861.

5. Sutherland DR, Yeo E, Ryan A, Mills GB, Bailey D, Baker MA: Identification of a cell-

surface antigen associated with activated T lymphoblasts and activated platelets.

Blood 1991, 77(1):84-93.

6. Smith JW, Hayward CP, Horsewood P, Warkentin TE, Denomme GA, Kelton JG:

Characterization and localization of the Gova/b alloantigens to the

glycosylphosphatidylinositol-anchored protein CDw109 on human platelets. Blood

1995, 86(7):2807-2814.

7. Lin M, Sutherland DR, Horsfall W, Totty N, Yeo E, Nayar R, Wu XF, Schuh AC: Cell

surface antigen CD109 is a novel member of the alpha(2) macroglobulin/C3, C4,

C5 family of thioester-containing proteins. Blood 2002, 99(5):1683-1691.

8. Schuh AC, Watkins NA, Nguyen Q, Harmer NJ, Lin M, Prosper JY, Campbell K,

Sutherland DR, Metcalfe P, Horsfall W et al: A tyrosine703serine polymorphism of

CD109 defines the Gov platelet alloantigens. Blood 2002, 99(5):1692-1698.

9. Solomon KR, Sharma P, Chan M, Morrison PT, Finberg RW: CD109 represents a

novel branch of the alpha2-macroglobulin/complement gene family. Gene 2004,

327(2):171-183.

10. Ertel K, Al-Tawil M, Santoso S, Kroll H: Relevance of the HPA-15 (Gov)

polymorphism on CD109 in alloimmune thrombocytopenic syndromes. Transfusion

2005, 45(3):366-373.

11. Hagiwara S, Murakumo Y, Mii S, Shigetomi T, Yamamoto N, Furue H, Ueda M,

Takahashi M: Processing of CD109 by furin and its role in the regulation of TGF-

beta signaling. Oncogene 2010, 29(15):2181-2191.

12. Hasegawa M, Hagiwara S, Sato T, Jijiwa M, Murakumo Y, Maeda M, Moritani S,

Ichihara S, Takahashi M: CD109, a new marker for myoepithelial cells of mammary,

salivary, and lacrimal glands and prostate basal cells. Pathol Int 2007, 57(5):245-

250.

13. Mii S, Murakumo Y, Asai N, Jijiwa M, Hagiwara S, Kato T, Asai M, Enomoto A, Ushida

K, Sobue S et al: Epidermal hyperplasia and appendage abnormalities in mice

lacking CD109. Am J Pathol 2012, 181(4):1180-1189.

14. Mii S, Hoshino A, Enomoto A, Murakumo Y, Ito M, Yamaguchi A, Takahashi M: CD109

deficiency induces osteopenia with an osteoporosis-like phenotype in vivo. Genes

Cells 2018, 23(7):590-598.

15. Haun RS, Fan CY, Mackintosh SG, Zhao H, Tackett AJ: CD109 Overexpression in

Pancreatic Cancer Identified by Cell-Surface Glycoprotein Capture. J Proteomics

Bioinform 2014, Suppl 10:S10003.

16. Arias-Pinilla GA, Dalgleish AG, Mudan S, Bagwan I, Walker AJ, Modjtahedi H:

Development of novel monoclonal antibodies against CD109 overexpressed in

human pancreatic cancer. Oncotarget 2018, 9(28):19994-20007.

17. Hashimoto M, Ichihara M, Watanabe T, Kawai K, Koshikawa K, Yuasa N, Takahashi T,

Yatabe Y, Murakumo Y, Zhang JM et al: Expression of CD109 in human cancer.

Oncogene 2004, 23(20):3716-3720.

18. Sato T, Murakumo Y, Hagiwara S, Jijiwa M, Suzuki C, Yatabe Y, Takahashi M: High-

level expression of CD109 is frequently detected in lung squamous cell

carcinomas. Pathol Int 2007, 57(11):719-724.

19. Hagiwara S, Murakumo Y, Sato T, Shigetomi T, Mitsudo K, Tohnai I, Ueda M, Takahashi

M: Up-regulation of CD109 expression is associated with carcinogenesis of the

squamous epithelium of the oral cavity. Cancer Sci 2008, 99(10):1916-1923.

20. Hasegawa M, Moritani S, Murakumo Y, Sato T, Hagiwara S, Suzuki C, Mii S, Jijiwa M,

Enomoto A, Asai N et al: CD109 expression in basal-like breast carcinoma. Pathol Int

2008, 58(5):288-294.

21. Hagikura M, Murakumo Y, Hasegawa M, Jijiwa M, Hagiwara S, Mii S, Hagikura S,

Matsukawa Y, Yoshino Y, Hattori R et al: Correlation of pathological grade and tumor

stage of urothelial carcinomas with CD109 expression. Pathol Int 2010, 60(11):735-

743.

22. Ohshima Y, Yajima I, Kumasaka MY, Yanagishita T, Watanabe D, Takahashi M, Inoue

Y, Ihn H, Matsumoto Y, Kato M: CD109 expression levels in malignant melanoma. J

Dermatol Sci 2010, 57(2):140-142.

23. Yi JM, Dhir M, Van Neste L, Downing SR, Jeschke J, Glockner SC, de Freitas Calmon

M, Hooker CM, Funes JM, Boshoff C et al: Genomic and epigenomic integration

identifies a prognostic signature in colon cancer. Clin Cancer Res 2011, 17(6):1535-

1545.

24. Cuppini L, Calleri A, Bruzzone MG, Prodi E, Anghileri E, Pellegatta S, Mancuso P,

Porrati P, Di Stefano AL, Ceroni M et al: Prognostic value of CD109+ circulating

endothelial cells in recurrent glioblastomas treated with bevacizumab and

irinotecan. PLoS One 2013, 8(9):e74345.

25. Emori M, Tsukahara T, Murase M, Kano M, Murata K, Takahashi A, Kubo T, Asanuma

H, Yasuda K, Kochin V et al: High expression of CD109 antigen regulates the

phenotype of cancer stem-like cells/cancer-initiating cells in the novel epithelioid

sarcoma cell line ESX and is related to poor prognosis of soft tissue sarcoma.

PLoS One 2013, 8(12):e84187.

26. Ozbay PO, Ekinci T, Yigit S, Yavuzcan A, Uysal S, Soylu F, Cakalagaoglu F:

Investigation of prognostic significance of CD109 expression in women with

vulvar squamous cell carcinoma. Onco Targets Ther 2013, 6:621-627.

27. Dong F, Wang Y, Li L, Wang Y, Liu X, Liu J: CD109 expression is increased in

cutaneous squamous cell carcinoma. J Dermatol 2014, 41(10):947-949.

28. Tao J, Li H, Li Q, Yang Y: CD109 is a potential target for triple-negative breast

cancer. Tumour Biol 2014, 35(12):12083-12090.

29. Dong F, Liu F, Yan S, Liu X, Jiang Z, Liu J: Elevated Expression of CD109 in

Esophageal Squamous Cell Carcinoma. Pathol Oncol Res 2015, 21(4):1273-1275.

30. Emori M, Tsukahara T, Murata K, Sugita S, Sonoda T, Kaya M, Soma T, Sasaki M,

Nagoya S, Hasegawa T et al: Prognostic impact of CD109 expression in

myxofibrosarcoma. J Surg Oncol 2015, 111(8):975-979.

31. Moffitt RA, Marayati R, Flate EL, Volmar KE, Loeza SG, Hoadley KA, Rashid NU,

Williams LA, Eaton SC, Chung AH et al: Virtual microdissection identifies distinct

tumor- and stroma-specific subtypes of pancreatic ductal adenocarcinoma. Nat

Genet 2015, 47(10):1168-1178.

32. Jia W, Ren C, Wang L, Zhu B, Jia W, Gao M, Zeng F, Zeng L, Xia X, Zhang X et al:

CD109 is identified as a potential nasopharyngeal carcinoma biomarker using

aptamer selected by cell-SELEX. Oncotarget 2016, 7(34):55328-55342.

33. Zong G, Xu Z, Zhang S, Shen Y, Qiu H, Zhu G, He S, Tao T, Chen X: CD109 Mediates

Cell Survival in Hepatocellular Carcinoma Cells. Dig Dis Sci 2016, 61(8):2303-2314.

34. Chuang CH, Greenside PG, Rogers ZN, Brady JJ, Yang D, Ma RK, Caswell DR, Chiou

SH, Winters AF, Gruner BM et al: Molecular definition of a metastatic lung cancer

state reveals a targetable CD109-Janus kinase-Stat axis. Nat Med 2017, 23(3):291-

300.

35. Dong F, Wang J, Xu Y, Cheng Z, Chen X, Wang Y, Liu J: CD109 expression is

upregulated in penile squamous cell carcinoma. Oncol Lett 2017, 14(5):6012-6016.

36. Shiraki Y, Mii S, Enomoto A, Momota H, Han YP, Kato T, Ushida K, Kato A, Asai N,

Murakumo Y et al: Significance of perivascular tumour cells defined by CD109

expression in progression of glioma. J Pathol 2017, 243(4):468-480.

37. Minata M, Audia A, Shi J, Lu S, Bernstock J, Pavlyukov MS, Das A, Kim SH, Shin YJ,

Lee Y et al: Phenotypic Plasticity of Invasive Edge Glioma Stem-like Cells in

Response to Ionizing Radiation. Cell Rep 2019, 26(7):1893-1905 e1897.

38. Bizet AA, Liu K, Tran-Khanh N, Saksena A, Vorstenbosch J, Finnson KW, Buschmann

MD, Philip A: The TGF-beta co-receptor, CD109, promotes internalization and

degradation of TGF-beta receptors. Biochim Biophys Acta 2011, 1813(5):742-753.

39. Zhang JM, Murakumo Y, Hagiwara S, Jiang P, Mii S, Kalyoncu E, Saito S, Suzuki C,

Sakurai Y, Numata Y et al: CD109 attenuates TGF-beta1 signaling and enhances

EGF signaling in SK-MG-1 human glioblastoma cells. Biochem Biophys Res

Commun 2015, 459(2):252-258.

40. Vorstenbosch J, Al-Ajmi H, Winocour S, Trzeciak A, Lessard L, Philip A: CD109

overexpression ameliorates skin fibrosis in a mouse model of bleomycin-induced

scleroderma. Arthritis Rheum 2013, 65(5):1378-1383.

41. Vorstenbosch J, Gallant-Behm C, Trzeciak A, Roy S, Mustoe T, Philip A: Transgenic

mice overexpressing CD109 in the epidermis display decreased inflammation and

granulation tissue and improved collagen architecture during wound healing.

Wound Repair Regen 2013, 21(2):235-246.

42. Winocour S, Vorstenbosch J, Trzeciak A, Lessard L, Philip A: CD109, a novel TGF-beta

antagonist, decreases fibrotic responses in a hypoxic wound model. Exp Dermatol

2014, 23(7):475-479.

43. Sunagawa M, Mii S, Enomoto A, Kato T, Murakumo Y, Shiraki Y, Asai N, Asai M, Nagino

M, Takahashi M: Suppression of skin tumorigenesis in CD109-deficient mice.

Oncotarget 2016, 7(50):82836-82850.

44. Thomas G, Chardes T, Gaborit N, Mollevi C, Leconet W, Robert B, Radosevic-Robin N,

Penault-Llorca F, Gongora C, Colombo PE et al: HER3 as biomarker and therapeutic

target in pancreatic cancer: new insights in pertuzumab therapy in preclinical

models. Oncotarget 2014, 5(16):7138-7148.

Vita

MennatAllah Shaheen was born in Giza, Egypt on July 16, 1991, the daughter of Sahar

Sayed and Ahmed Shaheen. After completing her work at Baraem Misr Language

School, Giza, Egypt in 2008, she entered Cairo University in Giza, Egypt. She received the degree of Bachelor of Pharmaceutical Sciences in May, 2013. For the next four years she worked as a pharmacist in El Ezaby Pharmacies in Egypt. In August of 2017 she entered The University of Texas MD Anderson Cancer Center UTHealth Graduate

School of Biomedical Sciences.

Permanent address:

193 B, Pyramids Gardens

Giza, Egypt 12522