Design and Evaluation of Novel Contrast Agents for Mr Molecular Imaging of Cancer

DESIGN AND EVALUATION OF NOVEL CONTRAST AGENTS FOR MR MOLECULAR IMAGING OF CANCER

ZHENG HAN

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy

Dissertation Advisor: Dr. Zheng-Rong Lu

Department of Biomedical Engineering

CASE WESTERN RESERVE UNIVERSITY

May 2017

CASE WESTERN RESERVE UNIVERSITY SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis/dissertation of: Zheng Han Candidate for the Doctor of Philosophy degree

Zheng-Rong Lu

Nicole F. Steinmetz (chair of the committee)

William P. Schiemann

Julian Kim

March 17, 2017

We also certify that written approval has been obtained for any proprietary material contained therein

Dedication

This dissertation is dedicated to my wife, Yuechen Wu and my parents Ziqing Du and Wenhan Han, who are offering me love and support.

iii

TABLE OF CONTENTS

Abstract ...... 1

Chapter 1 Diagnosis of Prostate and Breast cancer ...... 3

1.1 ABSTRACT...... 3

1.2 PROSTATE CANCER...... 3

1.2.1 Prostate cancer epidemiology and risk factors ...... 3

1.2.2 Current tools for prostate cancer diagnosis ...... 6

1.2.3 Prostate cancer biomarkers ...... 12

1.3 BREAST CANCER ...... 16

1.3.1 Breast cancer epidemiology and risk factors ...... 16

1.3.2 Current tools for breast cancer diagnosis ...... 18

1.4 TARGETING FIBRONECTIN FOR CANCER IMAGING DIAGNOSIS ...... 21

1.4.1 Fibronectin as a biomarker for cancer ...... 22

1.4.2 Targeting Fibronectin for Cancer Imaging ...... 35

1.5 PHAGE DISPLAY SELECTION OF LIGANDS FOR CANCER BIOMARKERS ...... 51

1.5.1 Principle of phage display ...... 51

1.5.2 Filamentous phage systems ...... 53

1.5.3 T7 phage library ...... 55

1.5.4 T4 bacteriophage system ...... 56

1.5.5 How to choose phage display system: our rationale ...... 57

1.6 DISCUSSION AND CONCLUSIONS ...... 59

Chapter 2 Phage Display Selection of Peptides Specific to EDB-FN ...... 78

2.1 ABSTRACT...... 78

2.2 INTRODUCTION ...... 79

2.3 MATERIALS AND METHODS ...... 81

2.4 RESULTS ...... 90

2.5 DISCUSSIONS AND CONCLUSIONS ...... 103

Chapter 3 Design and development of a MRI molecular imaging agent that targets EDB-FN for the differential diagnosis of prostate cancer ...... 110

3.1 DESIGN AND SYNTHETIC ROUTE ...... 110

3.2 MATERIALS AND METHODS ...... 111

3.3 RESULTS ...... 120

3.4 DISCUSSION ...... 137

Chapter 4 MRI differential diagnosis of breast cancer using a high-relaxivity

Gadolinium metallofullerene targeted to extradomain-B fibronectin ...... 145

4.1 ABSTRACT...... 145

4.2 INTRODUCTION ...... 146

4.3 MATERIAL AND METHODS ...... 148

4.4 RESULTS AND DISCUSSIONS ...... 155

Chapter 5 MR molecular imaging of triple-negative breast cancer and micrometastases ...... 169

5.1 ABSTRACT...... 169

5.2 INTRODUCTION ...... 170

5.3 MATERIALS AND METHODS ...... 173

5.4 RESULTS ...... 178

5.5 DISCUSSIONS ...... 188

Chapter 6 Future directions and concluding remarks ...... 198

6.1 SUMMARY OF CURRENT WORK ...... 198

6.2 NUCLEAR IMAGING BASED ON EDB-TARGETING PROBES ...... 199

6.2.1 Non-metallic elements ...... 199

6.2.2 Metallic elements ...... 201

6.3 DELIVERY THERAPEUTIC AGENTS BASED EDB-TARGETING STRATEGY ...... 203

6.3.1 Radiotherapy ...... 203

6.3.2 Delivery of anti-tumor agents ...... 204

LIST OF FIGURES

Chapter 1

Figure 1. Fibronectin structure and its binding sites to cell and other ECM molecules ...... 24

Figure 2. Upregulation of FN is crucial for cancer invasion and metastasis...... 27

Figure 3. Tumor uptake of 131I-L19SIP in a patient with metastatic prostate cancer...... 43

Figure 4. MRI contrast agents based on FN-targeting peptides...... 47

Figure 5. MRI Detection of breast cancer micrometastases using FibFN targeting contrast agent, CREKA-Tris(Gd-DOTA)3...... 49

Figure 6. Use of EDB-FN specific peptide, APTEDB, for development of imaging probes...... 50

Figure 7. Morphology of phages that have been engineered as phage display systems...... 54

Chapter 2

Figure 1. Construction of the EDB-expressing plasmid and SDS-PAGE assay of lysates from EDB-expressing E. coli strain BL21...... 91

Figure 2. Schematic illustration of the (A) synthesis of ZD2-Cy5 and (B)

MALDI-TOF mass spectrum of ZD2-Cy5...... 94

Figure 3. Characterization of the interaction between cyclic ZD2 and the EDB fragment using SPR ...... 94

Figure 4. ZD2-Cy5 binding in PC3 cells...... 96

vii

Figure 5. In vivo binding of ZD2-Cy5 in mice bearing PC3-GFP prostate cancer flank xenografts...... 99

Figure 6. Histological analysis of tissue sections of PC3 tumors from mice injected with ZD2-Cy5 or CERAK-Cy5...... 100

Figure 7. ZD2-Cy5 binding in human prostate sections...... 102

Chapter 3

Figure 1. Synthesis and characterization of ZD2-Gd(HP-DO3A)...... 122

Figure 2. In vitro complexation stability and transmetallation analysis of ZD2-

Gd(HP-DO3A)...... 123

Figure 3. Upregulated EDB-FN expression is a promising biomarker for high-risk prostate cancer ...... 126

Figure 4. The EDB-FN targeting contrast agent, ZD2-Gd(HP-DO3A), is capable of differentiating between PC3 and LNCaP tumors in T1-weighted MRI...... 129

Figure 5. Whole body signal changes in T1-weighted MRI with the contrast agents in tumor bearing mice...... 131

Figure 6. Contrast enhanced MRI with CREKA-Gd(HP-DO3A) in the PC3 and

LNCaP tumors...... 132

Figure 7. T1 maps and accumulation of ZD2-Gd(HP-DO3A) validate its specific binding in PC3 tumors...... 134

Figure 8. In vivo transmetallation and biodistribution of MRI contrast agents after intravenous administration...... 136

Chapter 4

viii

Figure 1. An EDB-FN-targeting gadofullerene for molecular MRI of breast cancer...... 160

Figure 2. Characterization of ZD2-Gd3N@C80...... 161

Figure 3. EDB-FN overexpression is a signature of aggressive breast cancer. .. 162

Figure 4. Contrast enhanced MRI with ZD2-Gd3N@C80 of MDA-MB-231 tumours and MCF-7 tumours in mice...... 164

Figure 5. Dynamic change of normalized signal intensities in muscle, heart, liver, kidney, and bladder in MDA-MB-231 and MCF-7 tumor models...... 166

Chapter 5

Figure 1. EDB-FN overexpression is a hallmark for 4T1 tumors with an aggressive phenotype...... 180

Figure 2. Evaluation of ZD2 in targeting 4T1 primary tumor by ex vivo fluorescence imaging and histological analysis...... 181

Figure 3. MRI of 4T1 primary tumors...... 183

Figure 4. Ex vivo imaging of organs containing 4T1 metastatic tumors at 5 h after injection of ZD2-Cy5.5 ...... 185

Figure 5. MRI of 4T1 metastatic tumor models ...... 186

Figure S1. Analysis of EDB mRNA in MDA-MB-231 with or without TGFβ treatment...... 192

Figure S2. Imaging of MDA-MB-231 primary tumor ...... 193

Figure S3. Imaging of MDA-MB-231 metastatic tumor...... 194

Chapter 6

Figure 1. Illustration of [18F]SFB and Al-18F based peptide labeling procedures.

201

Figure 2. Proposed 64Cu labeling procedure for ZD2...... 203

Figure 3. Structural illustration of APTEDB –modified liposomes for delivery of

Doxorubicin and EDB siRNA ...... 210

LIST OF TABLES

Chapter 1 Table 1. FN-targeting ligands and their application in cancer imaging and therapy

...... 36

Design and Evaluation of Novel Contrast Agents for MR Molecular

Imaging of Cancer

Abstract

ZHENG HAN

Magnetic resonance imaging (MRI) is a favorable imaging modality for cancer diagnosis due to its high soft tissue resolution and absence of ionizing radiation. However, most cancers are heterogeneous and existing MRI characteristics are insufficient to precisely delineate tumor aggressiveness for accurate prediction of clinical outcomes. MR molecular imaging aims at mining signatures that are reflections of cellular or subcellular processes in cancer, thereby uncovering hallmarks closely correlated to cancer proliferation, progression and metastasis. In this regard, contrast agents that differentially generate imaging contrast according to tumor aggressiveness is urgently needed.

This dissertation is focused on developing novel contrast agents capable of probing the aggressiveness of prostate and breast cancers. Prostate cancer and breast cancer represent the leading causes of morbidity and mortality in men and women, respectively. In addition to the relatively low detection sensitivity, diagnosis of both cancers are suffering from high false-positive rates using current

diagnostic techniques. In this work, extradomain-B fibronectin (EDB-FN), an abundant tumor microenvironment molecule exclusively expressed in aggressive tumors, is used as the biomarker. Peptide ligands were developed to specifically target EDB-FN, which were subsequently used to construct a library of gadolinium- based contrast agents. Our results indicate that the contrast agents developed in this work can potentially improve the sensitivity of MRI in detecting malignant cancers, lowering non-specific detection of indolent tumors and reducing gadolinium toxicity. Considering the huge unmet clinical need for accurate risk stratification and reducing overtreatment, this technology hold promise to be incorporated into clinical practice to achieve better diagnostic accuracy and treatment planning. This dissertation is divided into six chapters. Chapter 1 gives an overview of the clinical need for a non-invasive imaging modality for accurate stratification of prostate cancer and breast cancer. EDB-FN as a biomarker for cancer imaging is discussed.

Chapter 2 describes the screening of peptide ligands specific for EDB-FN. Chapter

3 and Chapter 4 describe molecular imaging of prostate cancer and breast cancer, respectively, based on EDB-targeting contrast agents. Chapter 5 demonstrates imaging of triple negative breast cancers and metastases with EDB-targeting contrast agents. Chapter 6 is a summary of current work and future directions.

Chapter 1 Diagnosis of Prostate and Breast cancer

1.1 Abstract

This chapter introduces current diagnostic strategies in prostate cancer and breast cancer, with emphasis on the strength and weakness of these strategies in stratifying cancer aggressiveness and predicting clinical outcomes. Strategies using fibronectin, particularly EDB-FN, as a cancer biomarker for cancer imaging are also reviewed. This provides the scientific background and motivation of the development of new imaging technologies that fulfil the need for a non-invasive tool that accurately determines cancer risk. This chapter also includes introduction to phage display technology, which we used for selection of peptide ligands targeted to EDB-FN.

1.2 Prostate cancer

1.2.1 Prostate cancer epidemiology and risk factors

In United States, prostate cancer (PCa) is the second most lethal cancer after lung cancer. Each year, more than 0.2 million men are diagnosed with PCa. There are currently more than 2 million men living with PCa in U.S. The lifetime risk of being diagnosed with PCa is high—17.6% (1 in 6) in Caucasians and 20.6% (1 in

5 men) in African American men (AAM) [1]. However, these rates vary significantly in the globe, with the lowest rates observed in Asian countries, specifically China and Japan, and the highest rates are observed in North America and Scandinavia. A number of factors contribute to these differences, including

genetic susceptibility, exposure to unknown external risk factors, or artificial reasons such as cancer registration and healthcare screening recommendations in each country.

Among several known PCa risk factors, the most extensively studied are race, ethnicity and age. AAM have the highest reported incidence of PCa in the US, with 2.5 times higher mortality rate than Caucasian men [1]. It is believed that nearly all men develop PCa eventually if they live long enough, and approximately

80% of men are detected with microscopic lesions of PCa after they reach eighth decade of life, regardless of ethnic groups they belong to [2].

Non-genetic factors such as socioeconomic status, literacy and access to healthcare are potential contributors to racial disparities with respect to stage at diagnosis and incidence of PCa. Environmental factors are also recognized to influences PCa incidence. For example, the Western nutrition/diet has been recognized as one of the main influences on increased risk of PCa [3, 4]. Excess intake of total dietary fat, consumption of red meat, and dairy products are classified as major dietary offenders [5]. Although the exact mechanism by which these dietary products influence PCa incidence remains clear, some studies implied the involvement of certain hormones in the induction of PCa [5]. High intake of saturated fatty acids contributes to elevated levels of circulating plasma insulin-like growth factor-1 (IGF-1), which is a potential contributor to cancer incidence.

Adipose tissue also actively secrete a variety of endocrine growth factors, which

regulate malignant transformation or cancer progression. In addition, heterocyclic amines, which are found in grilled and high temperature cooked meats, are also associated with increased risk of PCa development [6]. Dairy products contain high level of calcium, which suppresses serum vitamin D. But vitamin D has anti- proliferative and pro-apoptotic effect on a number of human cancer cell lines [7].

Following this, high calcium intake also exhibited correlation to for high grade, advanced and fatal PCa [8].

Increasing evidences have also pointed to inflammation as another cause of

PCa [9]. Prostate symptoms including prostatitis, chronic inflammation, and sexually transmitted disease (STDs) have been documented as precursors of PCa

[10]. While inflammation lies at the core of the immune response, inflammation can cause collateral damages and prostate cancer may be one of them. Histologic evidence of inflammation indicates induction of chronic immune-mediated oxidant stress, which may results in accelerated prostate carcinogenesis. A recent study discovered that inflammation was very common in prostate and 78% of men who were free from cancer showed signs of inflammation. An even higher chance of inflammation may be found in men with prostate disease (86%) and men with high- grade prostate cancer (88%) [11]. There are a number of players in the realm of prostate infection, including bacteria, protozoa and viruses. The bacterial infections, such as gonorrhea and chlamydia, have been linked to an increase in the risk of developing PCa, as has infection with protozoan Trichomonas vaginalis. The

relationship between sexually transmitted infections and PCa could possibly explain why PCa is more common in uncircumcised men [12]. In addition, just like how viral infection plays a role in a variety of cancers, viruses, such as Xenotropic murine leukienmia virus-related virus (XMRV), are thought to increase chances of

PCa because of altered expression of tumor suppressors or induction of protoncogenes following viral infections [13].

In addition, exposure to Agent Orange, radiation, and tobacco has been reported to correlate with PCa. Besides, effect of vasectomy, ejaculatory frequency, and infertility on PCa also has been under research.

1.2.2 Current tools for prostate cancer diagnosis

The triad of PSA testing, digital rectal examination (DRE) and transrectal ultrasonography (TRUS)-guided biopsy serve as the principal tools of appraising prostate diseases. The ensuing part of this section will focus on principles of each tool and discuss their strengths and limitations in giving an accurate PCa diagnosis.

1.2.2.1 PSA test

PSA test has been the most frequently used way for PCa detection since its introduction in 1986. PSA is a member of the tissue kallikrein family encoded by the hKLK3 gene. Abnormal KLK gene expression is commonly seen in various malignancies including prostate, ovarian, and breast cancers. Dysregulated KLK expression is associated with cancer cell growth and metastasis. Other than PSA,

KLK genes 2, 4, 5, 11, 14 and 15 are also considered as potential diagnostic and

prognostic biomarkers of prostate cancer. PSA is a serine protease that is involved in proteolysis of semenogelin I, II and fibronectin in freshly ejaculated semen, to liquefy seminal coagulum so that spermatozoa can be released. PSA is primarily produced by prostate ductal and acinar epithelial cells, with the highest concentration—106 times more than the serum PSA concentration—found in the lumen of the prostate gland. As integrity of basal cell layer, basement membrane, and endothelial cells are disrupted, PSA leaks into circulation, resulting in increased serum PSA levels [14].

The clinical use of PSA testing has led to increased detection of clinically localized prostate cancer. However, PSA testing is still under debate due to its poor specificity, which results in significant overdiagnosis and overtreatment. In 2012, the US Preventive Services Task Force recommended that the harm of PSA test may overweigh its benefit. In two large prospective, randomized trials investigating the effects of screening on prostate cancer mortality, the incidence of death per

10,000 person-year was 2.0 (50 deaths) in the screening group compared to 1.77

(44 deaths) in the control group. A conclusion was made that prostate cancer mortality was rare and did not differ significantly between the screening group and control group [15]. The decreased prostate cancer mortality in screened population should be weighed against adverse effect of overdiagnosis, overtreatment, harmed quality of life and cost-effectiveness.

Except for prostate cancer, an array of non-lethal conditions can also lead to elevated PSA level, including acute or chronic inflammation, urinary retention, benign prostatic hyperplasia (BPH), uretheral catherization, intravesical Bacilus

Calmette-Guerin instillation and prostatis. Prostate biopsy, ejaculation and prostate message can also increase PSA level. What makes diagnosis based on PSA testing more complicated is that age, race, and other factors also influence PSA level, and the presence of malignancy does not necessarily lead to high PSA. Therefore, a single PSA cut-off value cannot suffice to provide a high sensitivity and specificity.

The generally accepted PSA value cut-off point, 4.0 ng/mL, has a sensitivity of 20-

25% and specificity of 91%. Therefore, despite that a lot of patients are falsely diagnosed with prostate cancer, some prostate cancers cannot be sensitively detected [16]. In spite of this, in order to increase PSA detection sensitivity, there is a trend of using even lower cut-off value, with the sacrifice of specificity.

Several modifications have been developed to enhance the performance of

PSA, especially emphasizing on improving specificity in order to decrease false- positive results. These include age-specific PSA ranges, fPSA to total PSA ratio,

PSA density, PSA velocity, PSA doubling time (PSADT), and PSA half time.

However, these new PSA parameters either require validation in larger clinical trials, or ease of methodology to confirm their advantage over traditional PSA test.

In sum, despite its disadvantages, PSA testing is still the best available marker for detection, staging and follow-up of prostate cancer.

1.2.2.2 Gleason grading

Samples for grading are acquired by TRUS-guided biopsy, which is an invasive procedure that usually requires sedation and carries certain risks. Thus, to rule out temporary increases in PSA due to benign causes, it is recommended that biopsies be performed following a second PSA reading after an interval of a few weeks. In TRUS-guided biopsy, a probe (approximately the size of a thumb) with a retractable hollow needle for obtaining sample is inserted, in which only a small representation of the entire prostate is collected. There are chances that small but clinically significant areas of prostate cancer be missed. Further, the resulting specimens are fragile strips of tissue that are easily damaged, which further reduces the sensitivity of diagnosis. Highly variable and dependent on operator experience,

TRUS-guided biopsies have an overall sensitivity and specificity for detecting prostate cancer of 32% to 51%, respectively [17, 18]. The definitive result is often not guaranteed on initial biopsy, since the differentiation between benign and malignant tissue may not be confidently made, requiring a second biopsy at a later stage. The false negative rate in a single biopsy session is as high as 19-23% [18].

Biopsy samples are evaluated according to a system called "Gleason grading". The Gleason grading method, which has stood the test of time, was devised in the 1960s and 1970s by Dr. Donald F Gleason and members of the

Veterans Administration Cooperative Urological Research Group (VACURG).

This grading system relies entirely on the histologic pattern of arrangement of

carcinoma cells in H&E-stained sections. In a nutshell, the classification of cancer is based on categorization of histologic patterns ‘at relatively low magnification

(×10-40) by the extent of glandular differentiation and the pattern of growth of the tumor in the prostatic stroma’. Dr. Gleason consolidated nine observed growth pattern of cancer into five grades. The five basic grade patterns are used to yield a histologic score, which can range from 2 to 10, with addition of the primary grade pattern and secondary grade pattern. The primary pattern is the one that constitutes the largest area, by simple visual inspection, and the secondary pattern is the second most common pattern. In special cases, such as only one grade is in the tissue sample or the second grade is less than 3% of the total tumor, the primary grade is doubled to give the Gleason score. Essentially, the assignment of a Gleason score averages the primary and secondary grades.

Since Gleason grading system was devised in the late 1960s, medicine in general and prostate carcinoma in specific has seen dramatic changes. In the 1960s, digital rectal examination existed as the only screening tool for prostate cancer. In addition, most men (86%) involved in Gleason's studies possessed advanced diseases with either local cancer invasion out of the prostate or distant metastases.

Additionally, the method of obtaining prostate tissue was quite different from today's practice. In contrast to the current 18-core needle biopsy that obtains thin samples of prostate, biopsies in the old days involves the use of a couple of thick- gauged needle biopsies were directed into the area of abnormality. Consequently,

in Gleason's era, the grading of prostate cancer was not concerning about issues that involved in thin cores and multiple cores from different sites of the prostate. In addition, radical prostatectomy was not seen commonly and glands were not processed in their entirety or as extensively and systematically to the current degree.

The presence of multiple nodules and tertiary patterns in prostatectomy specimens causes issues with the use of the original Gleason system. Due to the aforementioned reasons, along with advances of immunohistochemical studies and emergence of new variant of prostate adenocarcinoma, Gleason grading has been adapting to changes in contemporary settings, which resulted in modifications to improve its accuracy. However, with these changes have come variations in applying the Gleason system among pathologists.

In sum, Gleason scores from biopsy samples reflect the pathology of only a few needle samples of tissue from the prostate gland and are thus inherently inaccurate. Small areas of high-grade tumor can easily be missed, leading to errors in the Gleason score assigned. In addition, prostate biopsy specimens are fragile and easily broken up either during the procedure or during preparation for analysis, which can make histological examination difficult. There is also inconsistency over the number of cores taken in a single biopsy session. Even though the adapted

Gleason grading system improved the consensus in certain extent, Gleason grading is still a subjective process. Thus, interobserver variance is inevitable. Besides, morphological arrangement of tumors does not directly correlate to the

aggressiveness of tumor. On the other hand, biopsy is still a relatively “blind” process since ultrasound gives no information on cancer malignancy. There are still chances that small but malignant regions will be missed, thus resulting in incorrect risk-evaluation of prostate condition.

1.2.2.3 Digital rectal examination

Digital rectal examination (DRE) is performed on all patients in whom prostate cancer is suspected. In current practice, DRE combined with PSA, serve as the first screening tools patients encounter. Although it is quick, simple and relatively painless and in experienced hands, DRE has a high specificity of 83.6% and a sensitivity of 53.2%, DRE is still associated with inaccuracies stemming from the fact that the tip of index finger only appreciates a small area of prostate and 30-

60% of tumors can be missed out using this method [19-21].

1.2.3 Prostate cancer biomarkers

The global oncology biomarker market is expected to reach 15,973 million

USD by 2020 (http://www.marketsandmarkets.com/Market-Reports/oncology- biomarkers-202.html). In prostate cancer, PSA testing constitutes a large segment of that market. The successor to PSA— or a panel of successors, presumably in conjugation with improved prostate imaging—holds promise to generate substantial profit for the industry. However, the real gain is incurred by the millions of men with elevated PSA levels undergoing biopsy to exclude cancer. Accurate

diagnosis needs to be made prior to biopsy. Biomarkers beyond PSA will likely change clinical management of prostate cancer dramatically.

Advances in technologies such as genomics, proteomics, and metabolomics, coupled with the enormous potential for clinical and commercial benefits of biomarker discovery, have led to an increase in biomarker discovery over recent years. With the issues involved in current prostate cancer screening becoming more prominent, the need for a better method of distinguishing prostate cancer from benign prostatic conditions also increases. Enabled by a more accurate biomarker, clinicians will miss fewer cases of prostate cancer, and patients with benign and indeterminate prostate pathologies would be spared the anxiety and discomfort from TRUS-guided biopsy. In addition, biomarkers that improve staging accuracy would reduce the possibility of tumor stage being upgraded or downgraded and may prove useful in detecting micrometastases. This is of particular significance in patients with non-organ-confined disease, who may be better treated with radiotherapy and hormonal therapy. On the other hand, biomarkers that detect lymph node invasion, when coupled with appropriate nomograms, could produce improved predictive and prognostic capabilities.

However, limited success was seen in translating theses new biomarkers into clinical utility. Many PCa biomarkers that promised to go beyond PSA have come and gone. In fact, the last PCa biomarker approved by the FDA was %fPSA in 1988. Currently, about 24 candidate biomarkers for PCa are currently under

study [22]. Two of those biomarkers, PCA3 and proPSA, have undergone full regulatory approval in Europe and are currently being reviewed by the U.S. FDA.

They both have been under development for more than 10 years and have received an extensive scientific underpinning. Here, two biomarkers, PCA3 and PSMA, will be introduced, with a special focus on their strength and limitations.

1.2.3.1 PCA3

Prostate cancer gene 3 (PCA3), also known as PCA3DD3 or DD3PCA3, was first reported in 1999 by Bussemakers and colleagues using differential display analysis [23]. In their study, mRNA expression patterns in benign vs. malignant prostate tissue were compared to identify unknown genes involved in prostate tumorigenesis. One of the clones, DD3 (differential display-3), was discovered to possess high expression level in prostate tumors and an apparent absence in benign tissue, which was further identified as PCA3DD3. Receiver-operating characteristic

(ROC) curve analysis on PCA3 in detecting prostate cancer yielded an AUC of 0.98, indicating a very high sensitivity and specificity for prostate cancer when isolated prostate tissue were examined directly [24].

Inspired by the promising results of PCA3 overexpression in prostate cancer tissues, researchers in Schalken laboratory at Radbound University assessed the feasibility of a noninvasive PCA3-based test for predicting biopsy outcome. Their method was based on the hypothesis that manipulation of the prostate would release cells into the urethra; therefore, sediments from urine collected following a digital

rectal exam could be used for PCA3 analysis. Using biopsy as the reference standard, ROC analysis yielded an AUC of 0.72 (95% CI: 0.58-0.85) using this non-invasive PCA3 analysis, which was a huge improvement compared to traditional PSA test [25]. This work represented the first study to demonstrate the potential of a quantitative urinary PCA3 test that aided in the prediction of biopsy outcome.

Currently, the function of PCA3 is under active research. The limitations of

PCA3 still reside in its presence as an mRNA analysis, which is relatively expensive and the preparation process may induce inaccuracy when non-ideal samples are acquired. However, as one of the most promising markers of PCa detection, PCA3 test, as a tool for accurate prostate cancer detection, deserves further attention and advancement into clinical use.

1.2.3.2 Prostate-specific membrane antigen

Prostate-specific membrane antigen (PSMA) is a membrane glycoprotein produced in high concentrations by epithelial cells of the prostate. In 1987,

Horoszewicz and colleagues originally defined PSMA as a protein recognized by

7E11-C5.3, a murine immunoglobulin G (IgG1) Mab specific to prostate cancer cells [26]. The expression of PSMA was verified by immunohistochemical analyses of frozen sections from a variety of normal and malignant human tissues and human cell lines. It was found that PSMA expression was confined within epithelial cells of the prostate, with no staining of stromal components. Prostate cancer tissue

exhibited relatively higher PSMA expression than normal of hyperplastic tissue.

Further studies from different institutions provided results similar to that of the original report [26]. Though as a prostate-specific molecule, PSMA is also present in non-prostatic tissues, including the small intestine, renal tubular cells, and salivary gland [27]. The expression in those tissues is nonetheless 100- to 1000- fold lower than that in PCa. .Elevated PSMA levels in aggressive tumors implicate its role in transformation or invasiveness of prostatic epithelial cells [28]. To date, little is known about how PSMA molecules are shed into circulation. The competitive enzyme-linked immunosorbent assay (ELISA) and western blot analysis are major approaches for the measurement of PSMA [29]. Nevertheless, higher PSMA levels do not correlate with high pathologic stages. The clinical utility of serum PSMA level for diagnosis is limited. So far, the most successful demonstration of the diagnostic value of PSMA is radio-immunoscintigraphy. The

111Indium-labeled monoclonal antibody (Mab) 7E11-C5, which recognizes PSMA on the prostatic cell surface, is branded as CYT-356 or ProstaScint® [30] for clinical testing.

1.3 Breast cancer

1.3.1 Breast cancer epidemiology and risk factors

Breast cancer constitutes approximately 30% of newly diagnosed cancers in American women. An estimated total of 234,190 new breast cases were diagnosed in 2015 globally. Breast cancer has the highest incidence rate among any

cancer among females in the majority of countries in the world [31], with the exception of several countries in Eastern and Western Africa, South America and part of Asia, where the incidence of cervical cancer or lung cancer is higher. While the overall number of new cases is similar in developed countries to less developed countries, incidence rates are about two times higher in the former after normalization to population size and age structure. A number of factors contribute to the regional variations in incidence rates. Women in developed countries have higher breast cancer risk because of higher obesity, alcohol consumption, fewer children, less breastfeed, etc. Higher public awareness and utilization of organized screening programs in developed countries also increase diagnostic rates of breast cancers, which may otherwise remain undetected in other countries. It cannot be neglected that overdiagnosis associated with current screening procedures results in non-specific detection of breast cancers as well.

Breast cancer risk rises markedly when women reach the age of 40. In the globe, 89% of breast cancers are diagnosed in patients over the age of 40. However, breast cancer in younger women are generally larger, less differentiated, and more likely to metastasize [32].

Owing to organized screening, 50-60% of cases diagnosed in developed countries are localized. In contrast, as few as 25% of breast cancers in less developed countries are diagnosed at an early stage [33].

1.3.2 Current tools for breast cancer diagnosis

1.3.2.1 Mammography

The aim of population screening is to detect breast cancers at an early stage in asymptomatic women to reduce related morbidity and mortality. Although evidences are lacking concerning the balance between the benefit and harms of breast cancer screening, it is generally believed that screening programs resulted in increased survival of patients. Currently, mammography is used for breast screening. Mammography is an x-ray exam of the breast that captures images of each breast at two different angles to detect signs of breast cancer in asymptomic women. Standard mammograms are visualized in large film sheet, while digital mammogram, also known as full-field digital mammography or FFDM, and 3D mammography are emerging and being well received by the public.

Mammography is efficient in detecting calcifications due to malignancies, such as ductal carcinoma in situ (DCIS). Some invasive cancers, which will or have already spread to secondary organs, often manifest as non-calcified masses, therefore being subtle or occult in mammography. Mammography is also less efficient in detecting lesions in dense breasts, particularly in young women, with a low sensitivity of 30% to 48% [34]. Although digital and 3D mammography appear to increase the detection sensitivity, the increase is quite marginal and benefit of mammographic screening in dense breasts far from satisfactory.

1.3.2.2 Ultrasound

Although ultrasound alone provides insufficient sensitivity for breast cancer, ultrasound has been used to supplement mammography to enhance depiction of small, node-negative breast cancers occult on mammography but can be felt in breast exam [35]. In particular, ultrasound has been considered a routine for young patients due to the low sensitivity of mammography [36]. The combination of ultrasound and mammography slightly increases sensitivity to about 50% [36].

Recent advances in ultrasonography include shear-wave (SW) elastography, which delineates the propagation speed of SWs within the breast masses as a reflection of the tissue stiffness. SW elastography provides both qualitative and quantitative information on elasticity of the breast tissues, therefore facilitating assessment of likelihood of malignancy and improving patient management [37]. Despite the improved sensitivity of this new technology, the combined mammography and ultrasound still fail to provide sufficient sensitivity for breast cancer.

1.3.2.3 MRI

MRI excels at producing detailed soft tissue contrast. Without the use of contrast agent, the contrast between tissues in breast depends on the magnetic environment of the hydrogen atoms in water and fat, which contributes to variations of the brightness of tissues in the images. Contrast-enhanced MRI utilizes a paramagnetic small molecular gadolinium-based contrast agent administered intravenously to afford a more accurate detection of cancers or other lesions. Since

enhanced lesion signal is difficult to separate from fat, subtraction images or/and fat suppression can be used. The enhancement dynamics and washout pattern in tissues are also interesting signatures of MRI that help the discrimination of slowly growing lesions or aggressive breast tumors. In this case, the pre-injection images with sequential sets of images after contrast injection are acquired (dynamic contrast-enhanced MRI, DCE-MRI). These techniques are currently widely employed for assessing asymptotic and symptomic breast diseases, and have demonstrated much higher sensitivity than mammography, ultrasound, and the combination or both. Because of the high sensitivity, MRI is used as a screening tool for breast cancer in women with high risk based on family history and a genetic predisposition, such as BRCA mutation [38]. It is reported that MRI increases the detection sensitivity to about 93% [38].

Despite its high sensitivity, MRI still lacks specificity for malignant lesions.

This is supported by evidences showing no effect of preoperative breast MRI on reducing re-excision rates, locoregional recurrence or disease-free survival [39].

Nonmass lesions represent a challenging subgroup causing a high proportion of false-positive diagnoses. According to a study by Pascal et al., nonmass lesions constitute about half of false-positive diagnoses [40]. Benign fibrocystic changes, such as fibroadenoma, sclerosing adenois, intraductal papillomas, stromal fibrosis, apocrine metaplasia, also may manifest as false-positive enhancement in MRI.

Hence, the US Preventative Task Force recommended caution about the use of MRI

for breast screening, and currently MRI screening is only applied to patients with high familial risk of breast cancers.

1.4 Targeting Fibronectin for Cancer Imaging Diagnosis

This section is adapted based on the following published paper:

Han Z, and Lu ZR. Targeting fibronectin for cancer imaging and therapy. Journal of Materials Chemistry B. (2017).

During cancer progression, the extracellular matrix (ECM) undergoes dramatic changes, in which fibronectin (FN) is prominently upregulated. FN serves as a central organizer of ECM molecules and mediates the crosstalk between the tumor microenvironment and cancer cells. Its upregulation is closely correlated with angiogenesis, cancer progression, metastasis, and drug resistance. Thus far,

FN-targeting imaging agents have been tested for nuclear imaging, MRI, and fluorescence imaging of cancer. This section summarizes current understanding on the role of FN in cancer, discusses the design and development of FN-targeting agents, and highlights applications of these FN-targeting agents in cancer imaging.

The findings discussed in this section inspired us to use EDB-FN as the biomarker for development of imaging probes for cancer imaging.

1.4.1 Fibronectin as a biomarker for cancer

Cancer is a genetic disease and stems from gene mutation within normal cells. Based on this notion, the strategies in designing cancer imaging agents and therapeutics generally have been focused on targeting the cancer related molecular signatures existing either intracellularly or on cell surface. However, it has been increasingly recognized that tumor microenvironment is a key determinant for cancer survival and progression. The extracellular matrix (ECM) of cancer is highly remodeled with a dramatic change in biochemical and physical properties. This change is accompanied by tumor-permissive immunological surroundings and recruitment of a plethora of stromal cells, including cancer-associated fibroblasts

(CAFs), endothelial cells, macrophages, etc.[41]. Cancer cells in the aggressive primary site are adept at exploiting the remodeled tumor microenvironment for outgrowth and metastasis [42]. Among the tumor microenvironmental cues investigated, FN stands as one of the lead cancer-related extracellular biomolecules.

As an abundant molecule in ECM, FN takes part in a variety of processes that promote cancer cell progression, by interacting with cells and other ECM molecules.

Over the past two decades, the role of FN in cancer has been recognized and FN- targeting strategies have been devised as promising cancer imaging approaches. In this article, we provide an overview of the interaction of FN with cancer cells and other ECM molecules, how this interaction correlates to cancer malignancy, and

FN-targeting strategies in cancer imaging, highlighting their application in different types of cancers.

FN is an abundant, high-molecular-weight, adhesive glycoprotein that exists in the ECM or body fluids[43]. FN takes the form of dimers, joined by a pair of disulfide bonds in the carboxylic termini, the structure of which is shown in

Figure 1. Each monomer of FN consists of three types of homologous repeats termed as type I, II and III domains. Although FN is encoded by a single gene, alternative splicing of pre-mRNA and posttranslational modifications result in the formation of cell- and tissue-specific FN isoforms. Splicing occurs at three sites, including the site located between III11 and III12 with the insertion of extradomain

A, the site located between III7 and III8 with the insertion of extradomain B, and various portions of the IIICS domain between III14 and III15 (Figure 1). Splicing events result in two main FN types, soluble and insoluble FNs. Soluble FN, also called plasma fibronectin (pFN), is produced by hepatocytes and distributes in plasma (~300 μg/mL) and other body fluids[43]. Extradomains are usually absent in soluble fibronectin. Insoluble FNs, on the other hand, are synthesized by a variety of cell types, including fibroblasts, muscle cells, endothelial cells and cancer cells.

It is the product of fibrillorgenesis, i.e. polymerization of FNs, which provides a multi-dimensional platform that interacts other ECM molecules and cell surface receptors [44].

Figure 1. Fibronectin structure and its binding sites to cell and other ECM molecules. Extradomains in insoluble FNs (extradomain-A, extradomain-B and

IIICS) are denoted by white blocks between other type III domains.

Two FN isoforms generated by splicing in type III repeats are termed as extradomain-A fibronectin (EDA-FN) and extradomain-B fibronectin (EDB-FN).

EDA- and EDB-FN are the most investigated FN isoforms for tumor-targeting strategies. EDA- and EDB-FN are expressed during embryogenesis, but their expressions are highly conservative in adult tissue except in wound healing and cancer, because of which they are called “oncofetal” fibronectins [45]. Single EDA- or EDB-FN null mice show fairly normal development but mice lacking both FN isoforms show vascular defects that results in embryonic lethality [46-48].

The critical role of FN in cancer lies in that FN serves as a central mediator of the crosstalk between cancer cells and other ECM molecules [49]. FN is a well- established hallmark of epithelial-to-mesenchymal transition (EMT). EMT

generates cells that possess more cancer-stem-cell-like phenotype [50], which frequently occur in the invasive front of aggressive tumors [51] (Figure 2A).

Cancer cells that undergo EMT upregulate FN along with other ECM molecules that constitute the remodeled ECM. As a result, FN-matrix assembly, angiogenesis and cell invasion are enhanced. Interestingly, secreted FN further stimulate EMT, for example, by sensitizing cancers to TGFβ induction, and trigger a variety of signaling pathways that in turn upregulate FN expression [52, 53] (Figure 2B).

This feedback loop once again reflects a close alliance of FN and cancer cells. As previously mentioned, insoluble FNs are formed as a result of fibrillorgenesis, which is mediated by integrin clustering [49] (Figure 2B). It is well known that the RGD sequence on fibronectin III10 is responsible for binding cell-surface integrins [54]. However, RGD motif alone showed a much lower affinity to its targets than larger protein fragments of FN or intact protein. This indicates that the area in vicinities of RGD, e.g. PHSRN sequence in III9 domain, contributes to the binding [54, 55]. Considering that insoluble FNs distinguish themselves from soluble FN by extradomains, it is natural to reason that these extradomains play indispensable role in fibrillorgenesis and the enhanced interaction between FN and cancer cells. It is suggested that the extradomain insertion in type III repeats results in a conformational change within FN that stabilizes head to tail dimerization of separate FN chains, which further links to FN matrix assembly [56]. Despite that the extradomain insertion does not alter integrin binding sites, the change in

conformation exposes of a pair of these sites on the same face of the macromolecules, facilitating its interaction with multiple receptors on the cell surface[55, 57]. This also enhances clustering of integrins, coupled to fibrillorgenesis and a series of intracellular signaling events [49, 57] (Figure 2B).

Besides, evidences indicate that the insertion of EDA domain may increase RGD exposure to integrins, resulting in enhanced interaction with cancer cells [58, 59].

Meanwhile, EDA domain itself also binds to α9β1 or α4β1 integrins [60]. In terms of

EDB-FN, although the receptor of EDB domain on cancer cell remains elusive, it is known that adhesion of EDB domain to cancer cell induces tyrosine phosphorylation of focal adhesion proteins, followed by activation of FAK tyrosine phosphorylation pathways [61].

Figure 2. Upregulation of FN is crucial for cancer invasion and metastasis. A.

FN is upregulated in the invasive front of primary tumor, which mainly comprises of cancer cells of mesenchymal phenotype. These mesenchymal cells also possess traits of cancer stem cells, and are prone to disseminate to secondary organs. FN and other ECM molecules form a unique tumor microenvironment that promotes the migration and extravasation of cancer cells. B. FN serves as a central organizer of ECM molecules and mediates the crosstalk between tumor microenvironment and cancer cells. In malignant cancer cells, FN assembles into FN matrix mediated by the clustering of integrins. FN matrix provides docking sites for collagen and fibrin. Clustering of integrin also activates intracellular signaling complexes that

stimulate the expression of a series of oncogenes. This also promotes the expression and secretion of more FNs, MMPs, and other tumor-promoting factors, such as

TGFβ. MMPs modulate the regulated degradation of FN, collagen, and fibrin matrix, and the degraded products are endocytosed with the aid of integrins. C. A receptive microenvironment characteristics of FN upregulation in the secondary organ is essential for the metastases formation. Extravasated cancer cells in the secondary organ are faced with two fates: dormancy and death if they are arrested in tissues that lack a favorable microenvironment, or proliferation and metastases if they are engrafted in pro-metastatic niche enriched with FN, collagen, fibrin, and etc. The cells may lose their mesenchymal traits for outgrowth. FN originated from primary tumor may be used to prime tissue where pro-metastatic niche is formed.

FN also serves as a scaffold that provides binding sites for a variety of ECM molecules, thus making it a central organizer of ECM (Figure 2B). One of the ECM molecules directly coupled to FN is fibrin. Fibronectin, specifically plasma fibronectin (pFN), is covalently linked to fibrin [62]. Crosslinking of FN with fibrin is a key event in clot formation. In the setting of cancer, even though pFN is reported to have no clinical correlation with cancer [63, 64], the presence of fibrin- fibronectin complexes (FibFN) is a characteristic of malignant cancers [65]. A study showed that pFN-deficient demonstrated decreased metastasis formation in the lung, indicating a pro-metastatic effects of blood clotting in vivo [66]. Further, thrombin antagonist hiridin would impair lung metastasis through inhibition of

platelet activation and fibrin formation. It is suggested that this impact of FibFN complex on metastases is mediated through fibronectin-αVβ3 integrin interaction, which is absent with sole existence of pFN. However, FibFN had no effect on tumor cell growth and initial tumor cell arrest, suggesting its primary role in clotted plasma to assist in cell extravasation [66].

Other ECM molecules actively involved in ECM remodeling include collagen and matrix metalloproteinases (MMPs). Collagen is the most abundant protein in mammals, and is prominently overexpressed in cancer [67, 68]. Recent studies reported the inductive function of fibronectin in collagen deposition and architecture regulation [68-70]. Accordingly, inhibition of FN matrix assembly would lead to abnormal collagen expression [44]. MMPs are essential for peri- tumor tissue degradation and cancer cell migration. Studies indicated a role of

EDA-FN in promoting expression of certain MMPs [71, 72]. In return, MMPs also regulates FN matrix assembly since FN is also a substrate of MMPs [73]. Even though formation of FN fibrils plays a central role in interacting with cells and other

ECM molecules, cell migration is facilitated by a regulated degradation of those fibrils. It is suggested that MT1-MMP degrades FN fibrils and degraded FN is endocytosed with the aid of β1 integrin [74] (Figure 2B). Thus, under FN regulation, these molecules work in concert to form a unique tumor microenvironment to promote tumor cell migration and invasion.

In addition to interacting with cancer cells, FN, especially EDB-FN, is thought to be involved in regulation of endothelial cells. Over the last two decades,

EDB-FN has been recognized as a biomarker for angiogenesis [75]. Angiogenesis is the sprouting of new blood vessel from preexisting ones. By providing oxygen and nutrition to the tumor mass, angiogenesis takes a pivotal part in sustained tumor growth in the primary site, and initiation of new metastasis in distant organs. It has been shown that EDB-FN excreted by cancer cells is able to induce endothelin-1

(ET-1) expression in endothelial cells, which promotes angiogenesis [75] (Figure

2B). In turn, EDB-FN upregulation in cancer and endothelial cells is stimulated by

ET-1[76]. This feedback loop augments angiogenesis and cancer progression [75].

Due to this reason, anti-angiogenic therapy based on silencing EDB-FN expression was developed, which can potentially be used to undermine tumor blood vessel formation [77].

Clinical evidences indicate that FN, especially onfFN, is overexpressed in various cancer types, including breast cancer [68, 78], prostate cancer [79, 80], bladder cancer [81], oral squamous cell carcinoma [82], head and neck squamous cell carcinoma[83], colorectal cancer[45], and lung cancer[84]. Thus, FN may serve as an omnipresent biomarker regardless of the origins of cancer cells. More importantly, upregulated FN expression is correlated with poor prognosis of the patients. For instance, in studying oral squamous cell carcinoma, Lyons et al. reported that strong onfFN expression was seen in 63% and 81% of cases with

cervical metastases and extracapsular lymph node spread, respectively [82]. In another study, Young et al. evaluated FN expression in invasive breast cancer tissue and a significant correlation was found between FN expression and pathologic tumor stage, pathologic lymph node stage and histologic grade [78]. A worse survival was also found for the patients with FN expression higher than those with negative expression of FN. Additionally, in a recent study, EDA-FN in urine was demonstrated to be a significant discriminator of bladder cancer patient survival[81]. All these evidences point towards the value of FN as a biomarker that relates to cancer malignancy and patient survival.

Metastasis, the spread of cells from the primary site of a solid tumor to distant sites, remains the major cause of cancer mortality [85]. Cancer metastasis is a multi-stage process that includes tumor cells escaping from main tumor mass, invading into blood and lymph system, surviving in circulation, extravasation, and colonization into secondary organs. Metastasis has an early onset in aggressive cancer type, and may occur before the formation of large solid tumors [86].

Disseminated cells are also responsible for recurrence years after dissection of the primary solid tumor [87]. Epithelial-to-mesenchymal transition (EMT) has been thought to play an important role in cancer metastasis [50]. During EMT, epithelial cells undergo a transition from a cell-cell contact to a cell-ECM interaction [88, 89], allowing mesenchymal cells to reach and squeeze through vessel wall and spread

(Figure 2A). In metastatic sites, FN upregulation is one of the earliest events

following cancer cell engraftment (Figure 2C). It has been proposed that disseminated tumor and its metastatic niche function in a “seed and soil” manner[90]. This concept is introduced that disseminated tumor cells relies on a receptive microenvironment for outgrowth. Failure of cancer cells to engraft in a hospitable environment would result in tumor sluggish growth or dormancy

(Figure 2C). FN overexpression is a common event in metastatic niche [91, 92].

Surprisingly, this upregulation is found to precede secondary tumor engraftment

[93]. FN at metastatic niche can trace back to primary tumors, implying that primary tumor secretes FN as a way to prime certain tissues for tumor engraftment

[94] (Figure 2C). On the other hand, the lack of FN in metastatic niche would lead to decreased metastases formation (Figure 2C). For example, a recent study demonstrated that silencing EDA-FN reduced metastasis of colon cancer cells [95].

In companion with EMT, mesenchymal-to-epithelial transition (MET) is also crucial for metastasis formation. While EMT helps the cancer cells to invade into blood vessel, suppress immune response, and going through extravasation, MET helps in sustained growth of cancer cells in later stage [96, 97]. However, MET doesn’t disqualify FN as a biomarker of cancer metastasis since sustained growth of new solid tumor requires angiogenesis to satisfy its nutrition need. As discussed in previous sections, EDB-FN is a well-known marker for angiogenesis. Together, the upregulation of FN can serve as a marker for localizing cancer metastases.

Cancer drug resistance imposes a major challenge to cancer therapy. It is believed that the generation of cancer stem cells (CSCs), primarily through EMT, is essentially responsible for drug resistance [98]. This endows cancer cells to go through adaptive changes so that they can survive from anticancer therapeutics, including radiation and chemotherapy. Hence, as an EMT marker, FN content can serve as a predictive marker of cancer resistance. High FN expression, as linked to

EMT, coincides with loss of epithelial targets such as EGFR and HER2, and emergence of mesenchymal markers [98]. Targeted chemotherapy therapy using epithelial targets would exert modest or no therapeutic effects in this case. It is also indicated that FN confers drug resistance through protecting cells from apoptosis by activating a number of intracellular signaling pathways [99-101]. The contribution of β1-intergrin mediated signaling pathway to drug resistance implies the importance of FN-integrin interaction in protecting cancer from chemotherapy[102]. Since FN isoforms, such as EDA- and EDB-FN enhance the

FN-integrin interaction, the existence of FN isoforms may compromise the therapeutic effect exerted by chemotherapy and radiotherapy [103]. Indeed, it has been demonstrated that the use of certain chemotherapy drugs, for example

Cetuximab, could induce fibronectin biosysnthesis, which in turn attenuates its cytotoxic and radiosensitizing potential [104]. In another study, cDNA microarray analysis indicated that the fibronectin-encoding gene, FN1, could be a predictive marker for resistance to radiation therapy in head and neck squamous cancer [105].

Inspired by this, attempts have been done to counteract this resistance by suppressing EDA-FN expression or function [106, 107], which enhances cancer radiosensitivity both in vivo and in vitro. Together, the correlation of FN with cancer resistance may provide a new avenue for predicting therapy response—a crucial step towards personalized therapy.

Understanding the role of FN in cancer is certainly helpful in the design of novel approaches for cancer diagnosis and therapy. Imaging of FN enables the non- invasive detection of cancer-related processes, such as angiogenesis, EMT, metastasis, and cancer resistance, permitting the accurate detection, diagnosis, risk- management of tumors, and non-invasive assessment of therapeutic responses.

Therapies directed towards FN can be used for delivery of cytotoxic drugs, cytokines, radioisotopes, etc. to malignant sites. A novel class of cancer therapeutics can also be developed to reverse FN’s tumor-promoting functions by downregulating FN expression, therefore disrupting a number of cancer-promoting processes, such as suffocating oxygen and nutrient supply to tumor through inhibiting angiogenesis, disrupting EMT, and degrading pre-metastatic niche to suppress metastasis. In the following section, we will summarize the advances in developing FN-targeting imaging probes and therapeutic agents and their application in cancer diagnosis and treatment.

1.4.2 Targeting Fibronectin for Cancer Imaging

In cancer diagnosis, accurate detection, localization, delineation, and risk- stratification of tumors are critical for clinical cancer management and treatment.

Cancer diagnostic imaging set force to provide a non-invasive method to fulfill these needs. Cancer diagnostic imaging has seen rapid advances with the evolution of imaging technologies in resolution, speed, and sensitivity [108]. Nevertheless, conventional anatomic imaging methods only rely on morphologic properties or changes for cancer diagnosis. Imaging of the molecular signatures of cancer has the ability to provide biochemical information for more accurate characterization and diagnosis. Substantial progresses have been made in developing contrast agents and probes for molecular imaging of cancer biomarkers and elucidating cellular or subcellular events non-invasively [108].

Biopsy is currently the gold-standard method for cancer risk-stratification.

But biopsy only examines limited sample regions in possible disease sites. Taking prostate cancer for example, even though saturation biopsy is used routinely for prostate cancer evaluation, the overall sensitivity and specificity with prostate biopsy are still very low, approximately 32% and 51%, respectively [109, 110]. In addition, as an invasive approach, biopsy is associated with bleeding, pain, anxiety, possible infections, and other unintended side effects. Molecular imaging is a non- invasive approach that can probe biomarker expression related to the tumor malignancy throughout the entire regions of interest. Therefore, molecular imaging

can provide more accurate detection and diagnosis of cancer based on complete characterization of the biomarker expression, facilitating decision-making in the intervention and management of the disease.

FN has been tested as one of the biomarkers for molecular imaging. Because of its central role in ECM modeling, FN is a viable target for cancer molecular imaging. Its abundant expression in the ECM of malignant cancer presents excellent accessibility of imaging probes to the molecular target for effective molecular imaging. Antibodies and peptides have been developed to target FN and its isoforms. Table 1 summarizes some of the reported targeting ligands and their applications in cancer imaging and therapy. Majority of these ligands have been developed to target onfFN because of specific high expression of the biomarker in cancer.

Table 1. FN-targeting ligands and their application in cancer imaging and therapy

Imaging agents Ligand Form Target Cargo Modality Application L19 mAb EDB 123I Nuclear Brain[111], lung[111], and colon[111] cancers 124I Nuclear Head and neck cancer[112], brain metastases[113] 131I Nuclear Prostate cancer[114]

99mTc Nuclear Teratocarcinoma[115] 76Br Nuclear Teratocarcinoma[116]

BC-1 mAb 7FNIII 99mTc Nuclear Glioma[117]

F8 mAb EDA

CLT1 peptide FibFN Gd-DTPA MRI Colon cancer[118], breast cancer[94] G2-(Gd-DOTA) MRI Prostate cancer[119]

(Gd-DOTA)4 MRI Prostate cancer[95,96,97] CLT2 peptide FibFN

CREKA peptide FibFN Tris(Gd-DOTA)4 MRI Breast cancer[93][120]

SPION MRI Breast cancer[121, 122] ZD2 peptide EDB Prostate cancer [123]

APTEDB peptide EDB SPION MRI Breast cancer[124]

Antibodies for onfFN isoforms were reported in early 1990s. EDB-FN targeting antibodies, such as BC-1, are generated by mice immunization with EDB-

FN [125, 126]. Their epitopes are on another FNIII domain unmasked by EDB insertion [127]. Phage display against bacteria expressed EDB protein from human scFv antibody library resulted in another EDB-specific antibody, named L19[128].

Several forms of antibodies containing the variable region of L19(scFv) were later derived, including L19(scFv)2, L19SIP (SIP: small immunoprotein), AP39

(complete IgG1) [129], and L19-His. L19SIP shows a better plasma stability, pharmacokinetics, and tumor accumulation (2-5 times increase over

L19(scFv))[129], and has been frequently used to construct imaging or therapeutic agents. AP39 and L19(scFv)2 have also been labeled with radionuclide for imaging or therapy purposes [129]. An antibody scFv named F8 against EDA-FN, identified using colony filtering screening has also been tested for cancer targeting [130].

Small peptides have been identified to target onfFN and FibFN in cancer.

Small peptides are advantageous in immunogenicity, versatility in chemical modification, and cost-effective production as compared to antibodies. Small peptides also possess rapid extravasation and high tissue penetrating ability in tumor targeting [131]. This enables them to target not only vascular FN, but also

FN matrix surrounding cancer cells. An EDB binding scaffold-like peptide,

APTEDB, is devised with phage display technology [132]. APTEDB consists of a stabilizing scaffold and two target binding regions. Taking advantage of the three- dimensional structure for optimal binding, APTEDB exhibited a high binding affinity

(65 nM) to EDB protein. A small cyclic nonapeptide ZD2 of the sequence of

CTVRTSADC was recently identified using peptide phage display to target EDB.

This peptide demonstrated excellent specific targeting to tumor in vivo[123].

Another class of FN-targeting ligands is the peptides that target fibronectin-fibrin complexes. Cyclic peptides, CGLIIQKNEC (CLT1) and CNAGESSKNC (CLT2),

and pentapeptide CREKA identified from phage display exhibit specific binding to

FibFN in tumor stroma[133].

These targeting agents have been used in the preparation of imaging agents targeting FN and its isoforms for cancer imaging. The effectiveness of the agents has been demonstrated in various tumor models. Some of the agents have been tested in human patients. Recent advances in molecular imaging of FN and onfFN are highlighted in the following section.

Nuclear imaging

Nuclear imaging is a highly sensitive molecular imaging modality and provides detection, localization, and quantification of biomarkers in cancer imaging.

Radioisotopes labeled with targeting ligands are commonly used as probes in nuclear imaging. Images of biomarker localization are acquired by recording the γ- rays emitted from the radionuclides of the imaging probes on gamma cameras.

Depending on dimensions of acquired image, nuclear imaging can be classified into gamma-ray scintigraphy, which produces two-dimensional images, and single- photon emission computed tomography (SPECT) and positron emission tomography (PET), which produce three-dimensional images.

The probes for nuclear imaging are often developed by chemical coupling of γ-ray emitting radionuclides with a short half-life to targeting ligands. The pharmacokinetics of labeled antibodies or peptides, along with radionuclide half- lives, should be considered together in the design of the imaging probes for optimal

detecting sensitivity. Antibodies with a larger size possess a relatively long pharmacokinetic half-life. A longer time is needed for the clearance of the antibodies from the circulation and background tissues for effective image contrast.

In this case, radioisotopes with long half-lives, such as 99mTc, 123I, 76Br, 124I with t1/2 = 6h, 13h, 16.2h, 4.18d, respectively, are needed to couple with antibodies.

Small peptides often exhibit relatively fast target binding and accumulation, and fast clearance from the circulation and background tissue. Therefore, the pharmacokinetic half-life of peptides, rather than the radionuclide half-life, is a major factor in the design of peptide radioisotope probes.

The potential of molecular imaging of EDB-FN with a radioactive probe in cancer imaging was first demonstrated with the BC-1 antibody [117]. BC-1 antibody was labeled with 99mTc for scintigraphy and SPECT [117]. The preliminary study of the probe 99mTc-BC-1 in 5 brain cancer patients showed specific tumor uptake and very low nonspecific uptake in the bone marrow, liver and spleen. The probe provided a more accurate tumor detection than nonspecific indicator, 99mTc-DTPA, and its tumor uptake strongly correlated with specific oncofetal fibronectin expression.

Significant progress has been made on molecular imaging of EDB-FN using the L19 antibody and its derivatives. In a clinical study, L19(scFv)2 antibody

123 123 labeled with I， I-L19(scFv)2, was tested in detecting multiple cancers, including brain, lung and colorectal cancer with immunoscintigraphy[111]. Sixteen

123 out of 20 patients injected with I-L19(scFv)2 showed positive cancer detection, while the 4 patients with negative scans had tumors that did not express EDB-FN.

In another study, the L19(scFv) variant, AP39, was labeled with 99mTc for imaging

F9 teratocarcinoma. 99mTc-AP39 showed favorable pharmacokinetic and tumor- targeting properties with low concentration of radioactivity in the blood. The probe also demonstrated rapid renal excretion and high in vivo stability [115]. L19SIP was also labeled with 76Br for a small-animal PET imaging of teratocarcinoma[116].

Although this probe demonstrated fast and specific tumor targeting, a major concern arose due to the slow renal clearance of this probe. Persistent radioactivity in blood and stomach suggests partial 76Br-L19SIP debromination in vivo, which led to lower target to non-target ratios. Later, 124I-L19SIP was synthesized and tested for PET tumor imaging in FaDu head and neck cancer model, in which a clear delineation of tumors as small as 50 mm3 was achieved [112]. Due to similar pharmacokinetics of 124I-L19SIP and 131I-L19SIP, PET with 124I-L19SIP was used to predict proper dose of 131I-L19SIP for immunotherapy in patients with brain metastasis to achieve the optimal delivery of radiation to the tumor while minimizing burden to the dose-limiting organs (bone red marrow and normal brain)[113]. More recently, 131I-L19SIP was tested in a prostate cancer patient with

2D-scintigraphy, as highlighted in Figure 3. A selective update of 131I-L19SIP was seen in metastatic prostate tumors, qualifying the use of EDB-FN as a promising target for pharmacodelivery of anticancer agents in prostate cancer [114].

There has been no report on using FN-targeting peptides for PET or SPECT detection of cancer. Small peptides are advantages over antibodies due to their rapid tumor accumulation and fast clearance in circulation and background. In this regard, high tumor-to-background ratio at earlier time points after injection can be achieved in nuclear imaging using radiolabeled peptides. Thus far, several small peptides have been conjugated to paramagnetic chelates to develop targeted MRI contrast agents(90-92) (see later). In fact, the peptide ligand conjugates can be readily radiolabeled with radioisotopes, such as 55Co, 64Cu, 67Cu, 47Sc, 68Ga, 99mTc, and etc., for PET or SPECT imaging. In addition, these peptides can also be labeled with 18F for PET based on recently developed strategies [134, 135].

Figure 3. Tumor uptake of 131I-L19SIP in a patient with metastatic prostate cancer. A. CT scans confirmed the presence of G3 prostate cancer in the sacrum, the lower spine, and his right lobe of liver as indicated with arrows. B. A diagnostic dose of 131I-L19SIP was administered intravenously and whole-body planar images (anterior, left; posterior, right) were taken at post injection. Uptake of 131I-L19SIP is indicated in orange color, which is present in the oscadrum, lower spin, liver and mediastinum 18 and 52 h post injection (p.i.; yellow arrows).

Nonspecific uptake to the oropharygea mucosal linings is indicated with red arrow. Adapted and reprinted with permission based on ref. 80.

Magnetic resonance imaging

Magnetic resonance imaging (MRI) offers higher spatial resolution images of soft tissues and presents no ionizing radiation to patients. The challenge with

MRI is its inherently low sensitivity in molecular imaging. However, effective MR molecular imaging of cancer can still be achieved by targeting highly upregulated

FN in tumor stoma, which will allow specific binding of sufficient targeted contrast agents to generate detectable MR signal. Despite the success of nuclear imaging using EDB-FN targeting antibodies, antibodies with large size may not be suitable for targeted MRI contrast agents, especially for gadolinium-based contrast agents, because they cannot be rapidly excreted after the diagnostic imaging, exposing patients to potential toxic effects. In contrast, small molecular targeted contrast agents developed on top of peptides are desired, since they undergo rapid extravasation and accumulation in tumor tissues and demonstrate fast clearance of unbound agent from the circulation.

Small peptide targeted MRI contrast agents have been constructed by conjugating Gd chelates to FibFN-targeting peptides for cancer imaging[119, 136,

137]. CLT1-(Gd-DTPA) exhibited significant tumor contrast enhancement for at least 60 min post injection in HT-29 human colon carcinoma xenografts (Figure

A)[118]. This agent was also effective for imaging of MDA-MB-231 breast

carcinoma in a mouse model[138], and resulted in significant tumor contrast enhancement at a dose of 0.05 mmol/kg for at least 60 min after injection. In another study, the macrocyclic Gd chelate, Gd-DOTA, was also conjugated to CLT1 peptide. Macrocyclic chelate Gd-DOTA and its derivatives have higher kinetic stability than Gd-DTPA against transmetallation and more complete excretion from the body. Multiple peptide molecules and chelates were conjugated onto a generation 2 dendrimeric nanoglobule to give CLT1-G2-(Gd-DOTA) and to improve the efficiency of targeting and contrast enhancement in MR molecular imaging. The nanoglobular agent contains 3 CLT1 peptides and 20 Gd-DOTA chelates on average, and has high T1 and T2 relaxivities of 11.6 and 15.7 mM-1s-1 per Gd at 1.5T, respectively (Figure B)[119]. It produced significant tumor contrast enhancement up to 50 min post-injection in a mouse orthotopic PC3 prostate tumor model. A smaller targeted contrast agent with one CLT1 peptide and 4 Gd-DOTA chelates CLT1-dL-(Gd-DOTA)4 was synthesized by conjugating the Gd-DOTA monoamide chelates to the CLT1 peptide via generation 1 lysine dendrimer to reduce the size of the targeted contrast agent and to facilitate rapid excretion after imaging (Figure C)[139-141]. Similarly, CLT1-dL-(Gd-DOTA)4 demonstrated a

-1 -1 high T1 relaxivity (10.1 mM s per Gd) and significant tumor contrast enhancement in orthotopic PC3 human prostate cancer model, in which a low dose of 0.03 mmol Gd/kg was used. CLT1-dL-(Gd-DOTA)4 showed rapid renal clearance of the unbound agent and reduced Gd accumulation in normal organs at

48h after injection. Using the same approach, CREKA-dL-Gd(DOTA)4 was constructed and used for prostate cancer imaging(93). CREKA peptide and its targeted contrast agents exhibited better water solubility than CLT1 peptide and targeted contrast agents. A tripod macrocyclic Gd(III) chelate CREKA-Tris-

Gd(DOTA)3 was also synthesized and tested for breast cancer imaging (Figure

D)[137]. Specific targeting of this agent to FibFN in tumor microenvironment resulted in greater contrast enhancement than commercialized contrast agent

ProHance®. A rapid clearance of the contrast agents via renal filtration was seen, resulting in a low tissue retention.

It is known that FN is highly expressed in the metastatic niches. Contrast enhanced MRI has a potential to image micrometastases and very small tumors in high spatial resolution by targeting the highly expressed FN with a targeted contrast agent. Zhou et al. recently demonstrated that CREKA-Tris(Gd-DOTA)3 was able to detect micrometastases of triple negative breast cancer in animal models [120].

As shown in this study, metastatic tumors had much higher FN expression than the primary tumor, enabling sufficient binding of the contrast agent in micrometastases for sensitive detection. By co-registering detected micrometastases in MRI with high-resolution fluorescence cryoimaging of the fluorescently labeled tumors, MRI with CREKA-Tris(Gd-DOTA)3 was demonstrated to detect micrometastases with size <0.5 mm in diameter at a sensitivity of 83%, extending the cancer detection limit of the current clinical imaging modalities. The targeted contrast agent has

shown the potential to facilitate early detection of high-risk breast cancer and micrometastases so that early treatment is possible.

Figure 4. MRI contrast agents based on FN-targeting peptides. Chemical structures of CLT1-Gd-DTPA (A), CLT1-G2-(Gd-DOTA) (B), CLT1-dL-(Gd-

DOTA)4 (C), and CREKA-Tris(Gd-DOTA)3 (D) are shown.

Figure 5. MRI Detection of breast cancer micrometastases using FibFN targeting contrast agent, CREKA-Tris(Gd-DOTA)3. A. Breast cancer metastasis is companied by TGFβ upregulation that induces FN expression in metastases. By targeting overexpressed FN, which forms complexes with Fibrin,

CREKA-Tris(Gd-DOTA) accumulates at high concentration in sites of metastases.

CREKA-Cy5 also accumulates in metastases, which enabled the use of fluorescence cryoimaging to validate MRI detected tumors. B. MRI images of breast cancer micrometastases with CREKA-Tris(Gd-DOTA)3. Images that are presented include MR images of coronal slices before and after CREKA-Tris(Gd-

DOTA)3 injection, the subtraction images of pre-injection from the post-injection images, and the enlarged subtraction MRI images of metastatic sites and their corresponding cryo-images pot-injection of CREKA-Cy5.0 (tumors are indicated by arrow; all scale bars, 1mm). Adapted and reprinted with permission based on ref.

86.

Besides Gd-based contrast agents, FN-targeting peptides, including

CREKA and APTEDB, have also been used to modify super paramagnetic iron oxide nanoparticles (SPIO) for T2*-weighted MR molecular imaging[121, 122, 124]. It has been shown that CREKA-SPIO could induce additional plasma clotting in tumor, producing binding sites for more nanoparticles. This clotting-based amplification greatly enhanced tumor imaging[121, 122, 124]. Preparation of the

APTEDB labeled SPION nanoparticles is depicted in Figure . APTEDB-TCL-SPIO

nanoparticles showed significant accumulation in breast tumor initiating cells,

NDY-1, with specific overexpression of EDB-FN. In contrast, APTEDB-TCL-SPIO, could not bind to non-breast tumor initiating cells, such MCF-7 cells, which had low EDB-FN expression[124].

Figure 6. Use of EDB-FN specific peptide, APTEDB, for development of imaging probes. A. functionalizing APTEDB for labeling nanoparticles. B. a schematic depiction of APTEDB-SPION nanoparticle for T2*-weighted MRI. Adapted and reprinted with permission based on ref. 89.

Other imaging modalities

FN-targeting antibodies and peptides can be readily modified with fluorescent dyes for optical imaging. Even though limited studies have been done

using FN-targeting optical imaging probe for in vivo cancer imaging[123, 142], fluorescent probes have been developed in most studies for validating tumor targeting efficiency of the ligands. They also have been tested for in vivo diagnosis of other diseases, including rheumatoid arthritis, atherosclerosis, etc.[143-145].

The limitation of optical imaging is the penetration of light in tissues. However, there is an increasing need for optical imaging probes in delineation of tumor margin during surgery. FN-targeting optical imaging probes could be used in image-guided dissection of tumors, meanwhile minimizing harm to healthy tissue.

1.5 Phage Display Selection of Ligands for Cancer Biomarkers

Section 1.4 reviewed advances in cancer imaging based on fibronectin- targeting strategies and demonstrated that fibronectin, particularly oncofetal fibronectin, is a promising biomarker, based on which potent imaging technology can be developed. A prerequisite to the development of molecular imaging probes is the discovery of targeting ligands specific to oncofetal fibronectin. This section introduces phage display technology for selection of ligands for cancer biomarkers.

This technology was used for selection of peptides specific to EDB-FN.

1.5.1 Principle of phage display

Selection of a ligand specific to a target usually entails a huge library of ligands to start with, followed by repeated selection for high affinity ligands.

However, batch synthesis of random ligands that fulfils the complexity of the library is the impractical. Further, the inability of chemical ligands to replicate itself

makes repeated selection problematic. Phage display is a chemical selection process inspired by the natural selection process in the living world. The key characteristics of phages, mutability and replicability (i.e., ability to make copies of themselves), make phage display selection a practical realization of the artificial chemical evolution. A chemical’s “fitness” in this artificial biosphere is defined by its ability to bind to its receptor. To fit in the machinery of phage system, ligands that are screened for can only encompasses enzymes that bind their substrates, hormone receptors that bind their hormones, antibodies that bind their antigens, etc, rather than molecules that cannot be encoded by DNA. Once appropriate selection conditions are devised, the population of ligands would evolve toward specific ligands for that receptor.

A phage library is a heterogeneous mixture of phage clones, each possessing a distinct foreign DNA insert and therefore displaying a different ligand on its surface. Replication of a phage carrying such peptide will generate a huge crop of phage progenies displaying the same peptide. Since ligands are displayed on phage surface to allow maximal accessibility to solvent, ligands often behaves essentially as it would if they were not attached to the virion surface. Another essential component of this system is an immobilized receptor to specifically capture ligands from a complex mixture of compounds. Amplification of captured phages by infecting them into fresh cells would yield a large crop of progeny phages with decreased complexity, which can serve as the input for another round of affinity

purification. Eventually, captured phages are cloned so that the displayed peptides responsible for binding can be studied individually. The amino acid sequence of peptide is easily obtained by determining the corresponding coding sequence in the viral DNA. A successful selection will dramatically shrink down the library complexity and yield consensus ligand sequences if multiple clones are characterized.

Current phage display systems can be categorized by the type of phage strains. Most common systems include filamentous phages, such as M13, and non- filamentous phages, such as T7 and T4 phages, the structural representations of which are illustrated in Figure 2. The following sections will discuss the merits and shortcomings of each system and current commercial phage display systems will also be introduced. This comparison leads to our selection of phage display system for screening EDB-FN targeting peptides.

1.5.2 Filamentous phage systems

Filamentous phages are flexible rods about 1 μm long and 6 nm in diameter, composed mainly of a tube of helically arranged molecules of the 50-residue major coat protein pVIII; there are 2700 copies in wild-type virions, encoded by a single phage gene VIII. At one tip of the particle, there are five copies each of the minor coat proteins pIII and pVI; minor coat proteins pVII and pIX are at the other tip.

The phages infect strains of E.coli that display a threadlike appendage called F pilus.

Infection is initiated by attachment of the N-terminus domain of pIII (about 200

amino acids) to the tip of the pilus. It is the end of the particle that enters the cell first. The coat protein then dissolve into the cell surface and the uncoated ssDNA concomitantly enters the cytoplasm. Thereafter, the ssDNA is complemented with host-synthesized complementary DNA, resulting in a double-stranded replicative form (RF). The RF then replicates and serve as templates for transcription of phage genes and synthesis of progeny ssDNAs. These ssDNA then assemble with coat protein synthesized by host system, which ultimately extrude through cell envelope.

In the process, host cells are not killed and are subject to further infection.

Figure 7. Morphology of phages that have been engineered as phage display systems. Phage M13, λ, T7 and T3 are included. Phage DNAs (ssDNA or dsDNA) are represented with orange color; phage capsid is represented in black color.

Inserted peptides are represented in green squares. The figure is adapted based on ref. [146]

Foreign ligands have been fused to three coat proteins of filamentous phages: pIII, pVIII and pVI. Point of infusion should make sure incorporated ligands are exposed to phage exterior, and inserted peptides do not affect phage capsid assembly. The most common type of phage-display constructs are “random” peptide libraries, an outgrowth of the synthetic “mimotope” strategy of Geysen and his co-workers [147]; In this case, the DNA inserts are derived from “degenerate” oligonucleotides, which are synthesized chemically by adding mixtures of nucleotides (rather than single nucleotides) to growing nucleotide chain. In the degenerate sequence NNKNNKNNKNNK, for example, each N is an equal mixture of A, G, C, and T; each K is an equal mixture of G and T; each NNK is a mixture of 32 triplets that include codons for all 20 natural amino acids; and the entire 12-base sequence is an equimolar mixture of over a million (324) different molecular species collectively encoding all 160,000 possible 4-residue peptides. A typical random peptide library has about a billion phage clones—enough to represent most of the 64 million possible 6-mers, but far too small to represent the

3×1019 possible 15-mers.

1.5.3 T7 phage library

Another type of phage library is based on T7 bacteriophage. T7 phage is a non-filamentous phage, thus possessing a much distinctive structure than

filamentous phages (Figure 2). Rather than being extruded from bacteria membrane and assembling viron particle right before secretion, T7 phages assemble themselves in cytoplasm and escape from host cell by lysis. One advantage of T7 based phage libraries is that T7 phage can possess more foreign ligands than filamentous phages. Peptide fusions can be made through pXA or pXB coat protein, yielding displayed peptides up to about 50 amino acids in size and in high copy number (415 per phage). Another advantage of T7 phage system is that, in contrary to filamentous phage, T7 phages are less susceptible to biases resulting from peptide censorship commonly seen in filamentous phage assembly, secretion or infection processes. Phage clones displaying peptides composed of amino acid residues incompatible with these processes tend to be retained and propagated in an M13 library, resulting in increased diversity.

1.5.4 T4 bacteriophage system

Development of T4-based phage display system is based on the fact that minor T4 fibrous protein fibritin (gpwac) could be elongated at the C terminus without impairing its proper folding and being integrated into phage capsid.

Another display site in T4-based system include fusion of foreign protein to the decorative protein Soc or Hoc. T4 bacteriophage system share the same advantage with T7 bacteriophage, and what makes it unique is the ability to be modified so that it contain extra component, such as GST purification tags, for a more simplified purification processes.

1.5.5 How to choose phage display system: our rationale

The first question to ask regarding the choice of a phage display system should be what type of ligands to screen for. The answer is up to the target and application of this ligand. The ligands can be categorize into antibodies or peptides.

In terms of antibodies, phage display technology provides an alternative approach to vaccination, in which an adjuvant of the target and “bait” compound is introduced to the host— normally rodents— to generate target specific monoclonal or polyclonal antibodies. However, vaccination-generated antibodies are full-sized antibodies, and to shrink the size of them would raise the risk of losing their affinity.

In contrary, phage display screening can be done on truncated antibodies, for example, small-immunoprotein (SIP), or scFV, which are sometimes preferable for their rapid clearance. Several phage display studies on these truncated antibodies have been reported to generate “magic bullets” against cancer. In this case, the phage system should be able to host larger antibodies incorporated on their capsid, which requires a careful choice of phage systems and infusion sites. However, antibody based probes stay longer in systemic circulation, and for imaging purposes, best contrast can only be achieved hours after administration. This property is especially not desired when it comes to MRI, in which pre-contrast and post- contrast are sequentially acquired to ensure consistent positions. Thus, peptide phage screening is used in a larger spectrum of studies. Peptides, usually with a length of 5-20 amino acids, can be displayed on most of the above introduced phage

systems. Resultant peptides can further be easily synthesized and be tested on their affinity and specificity to their targets. In overall, peptides with the advantages of fast-clearance, cost-effectiveness, and ease of chemical modification prompted us to screen for peptides that are specific to EDB-FN.

Currently, a number of commercial phage libraries have emerged.

Novagen@ offers T7Select@ phage display system. However, customers are required to construct random peptide DNA library to ligate with the provided phage vector. A series of M13 phage libraries with a variety of peptide lengths, including

Ph.D 7 (displaying 7-mer linear peptides), Ph.D C7C (displaying 9-mer cyclic peptide with two Cysteines on flanks), Ph.D 12 (12-mer linear peptides), and Ph.D

C12C (12-mer cyclic peptide constrained by two Cysteines on flanks) are offered by New England Biolabs (NEB). NEB’s Ph.D C7C was used for selection of EDB affinitive cyclic peptides. Since M13 phage does not lyse the bacteria, purification of phage particles after phage tittering is easier compared to other system. The in- house integration of peptide libraries account for another advantage of using M13 based system.

1.6 Discussion and Conclusions

Radical prostatectomy and mastectomy remain the most successful treatment modalities for patients exhibiting clinically localized prostate and breast cancer. Despite this, many patients that have undergone surgery go through biochemical recurrence. Recurrent tumors usually retaliate with resistance to radiotherapy and chemotherapy, as well as higher metastatic potential. Curing prostate and breast cancer requires understanding of the distinct biological events that differentiate malignant cancers from unlethal diseases. Identifying the tumors that actually pose a threat is urgently needed in order to focus therapy where it is needed. This would exempt millions of women and men from potentially harmful interventions. MRI stands as a promising modality for sensitive detection of breast cancers. However, what hinders the broader application of MRI is the lack of specificity for high-risk prostate and breast cancer. MR molecular imaging provides opportunity for increasing the clinical utility of MRI by delineating cellular or subcellular processes of cancer that are indistinguishable with conventional MRI.

Thus, the clinical management of prostate and breast cancer demands novel prognostic biomarkers capable of predicting biochemical recurrence to direct therapeutic interventions at earlier disease stages.

FN and its isoforms play a significant role in cancer progression and are promising targets of cancer. Antibodies and small peptides have been developed to target upregulated FN and its isoforms in malignant cancer to deliver imaging

agents and therapeutics for cancer detection and therapy. Imaging probes and contrast agents have been developed based on these ligands showing the promise for cancer molecular imaging in preclinical and clinical studies. The imaging agents targeting overexpressed FN in cancer have the potential to improve the accuracy for early detection and diagnosis of malignant tumor.

Therefore, the objective of this dissertation is to develop MR molecular imaging agents for cancer diagnosis based on FN-targeting strategies. Specifically, due to the higher specific of EDB-FN in cancer, and its conservative expression in all mammals, we chose EDB-FN as the target. The prerequisite of developing such agents is discovery of targeting ligands for EDB-FN. We favor peptides over antibodies because peptides are advantageous in their low immunogenicity, higher tumor penetration, fast clearance, and cost-effectiveness. The ensuing chapter is focused on our work in developing EDB-specific peptides, which lays the groundwork for further development of EDB-targeting MRI probes.

References

1. Grönberg, H., Prostate cancer epidemiology. The Lancet, 2003. 361(9360): p. 859-864.

2. Gudmundsson, J., et al., Genome-wide association study identifies a second prostate cancer susceptibility variant at 8q24. Nature genetics, 2007. 39(5): p. 631-637.

3. Stacewicz‐Sapuntzakis, M., et al., Correlations of dietary patterns with prostate health. Molecular nutrition & food research, 2008. 52(1): p. 114- 130.

4. Freedland, S.J., et al., Obesity is a significant risk factor for prostate cancer at the time of biopsy. Urology, 2008. 72(5): p. 1102-1105.

5. Allen, N., et al., Animal foods, protein, calcium and prostate cancer risk: the European Prospective Investigation into Cancer and Nutrition. British journal of cancer, 2008. 98(9): p. 1574-1581.

6. Chan, J.M., P.H. Gann, and E.L. Giovannucci, Role of diet in prostate cancer development and progression. Journal of Clinical Oncology, 2005. 23(32): p. 8152-8160.

7. Gao, X., M.P. LaValley, and K.L. Tucker, Prospective studies of dairy product and calcium intakes and prostate cancer risk: a meta-analysis. Journal of the National Cancer Institute, 2005. 97(23): p. 1768-1777.

8. Giovannucci, E., et al., Risk factors for prostate cancer incidence and progression in the health professionals follow‐up study. International Journal of cancer, 2007. 121(7): p. 1571-1578.

9. Weir, K., Microbiology: Inflammatory evidence. Nature, 2015. 528(7582): p. S130-S131.

10. Dennis, L.K., C.F. Lynch, and J.C. Torner, Epidemiologic association between prostatitis and prostate cancer. Urology, 2002. 60(1): p. 78-83.

11. Gurel, B., et al., Chronic inflammation in benign prostate tissue is associated with high-grade prostate cancer in the placebo arm of the prostate cancer prevention trial. Cancer Epidemiology Biomarkers & Prevention, 2014. 23(5): p. 847-856.

12. Ravich, A. and R. Ravich, Prophylaxis of cancer of the prostate, penis, and cervix by circumcision. New York State journal of medicine, 1951. 51(12): p. 1519.

13. Kim, S., et al., Integration site preference of xenotropic murine leukemia virus-related virus, a new human retrovirus associated with prostate cancer. Journal of virology, 2008. 82(20): p. 9964-9977.

14. Stamey, T.A., et al., Prostate-specific antigen as a serum marker for adenocarcinoma of the prostate. New England Journal of Medicine, 1987. 317(15): p. 909-916.

15. Andriole, G.L., et al., Prostate cancer screening in the randomized Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial: mortality results after 13 years of follow-up. J Natl Cancer Inst, 2012. 104(2): p. 125-32.

16. Thompson, I.M., et al., Operating characteristics of prostate-specific antigen in men with an initial PSA level of 3.0 ng/ml or lower. Jama, 2005. 294(1): p. 66-70.

17. Montet, X., et al., Multivalent effects of RGD peptides obtained by nanoparticle display. Journal of medicinal chemistry, 2006. 49(20): p. 6087-6093.

18. Huang, J.X., S.L. Bishop-Hurley, and M.A. Cooper, Development of anti- infectives using phage display: biological agents against bacteria, viruses, and parasites. Antimicrob Agents Chemother, 2012. 56(9): p. 4569-82.

19. Kugler, J., et al., Oligopeptide m13 phage display in pathogen research. Viruses, 2013. 5(10): p. 2531-45.

20. Tassa, C., et al., Binding affinity and kinetic analysis of targeted small molecule-modified nanoparticles. Bioconjug Chem, 2010. 21(1): p. 14-9.

21. Catalona, W., et al., Comparison of digital rectal examination and serum prostate specific antigen in the early detection of prostate cancer: results of a multicenter clinical trial of 6,630 men. The Journal of urology, 1994. 151(5): p. 1283-1290.

22. Sardana, G., B. Dowell, and E.P. Diamandis, Emerging biomarkers for the diagnosis and prognosis of prostate cancer. Clin Chem, 2008. 54(12): p. 1951-60.

23. Bussemakers, M.J., et al., DD3:: A New Prostate-specific Gene, Highly Overexpressed in Prostate Cancer. Cancer research, 1999. 59(23): p. 5975-5979.

24. de Kok, J.B., et al., DD3PCA3, a very sensitive and specific marker to detect prostate tumors. Cancer research, 2002. 62(9): p. 2695-2698.

25. Hessels, D., et al., DD3PCA3-based Molecular Urine Analysis for the Diagnosis of Prostate Cancer. European Urology, 2003. 44(1): p. 8-16.

26. Horoszewicz, J.S., E. Kawinski, and G. Murphy, Monoclonal antibodies to a new antigenic marker in epithelial prostatic cells and serum of prostatic cancer patients. Anticancer research, 1986. 7(5B): p. 927-935.

27. Trover, J.K., M.L. Beckett, and G.L. Wright, Detection and characterization of the prostate‐specific membrane antigen (PSMA) in tissue extracts and body fluids. International Journal of Cancer, 1995. 62(5): p. 552-558.

28. Gregorakis, A.K., E. Holmes, and G. Murphy. Prostate-specific membrane antigen: current and future utility. in Seminars in urologic oncology. 1998.

29. Murphy, G., et al., Comparison of prostate specific membrane antigen, and prostate specific antigen levels in prostatic cancer patients. Anticancer research, 1994. 15(4): p. 1473-1479.

30. Abdel-Nabi, H., et al. Monoclonal antibodies and radioimmunoconjugates in the diagnosis and treatment of prostate cancer. in Seminars in urology. 1992. Saunders.

31. Youlden, D.R., et al., The descriptive epidemiology of female breast cancer: An international comparison of screening, incidence, survival and mortality. Cancer Epidemiology, 2012. 36(3): p. 237-248.

32. Pollan, M., Epidemiology of breast cancer in young women. Breast Cancer Research And Treatment, 2010. 123: p. 3-6.

33. Shulman, L.N., et al., Breast cancer in developing countries: opportunities for improved survival. J Oncol, 2010. 2010: p. 595167.

34. Kolb, T.M., J. Lichy, and J.H. Newhouse, Comparison of the performance of screening mammography, physical examination, and breast US and evaluation of factors that influence them: An analysis of 27,825 patient evaluations. Radiology, 2002. 225(1): p. 165-175.

35. Kuhl, C.K., et al., Mammography, breast ultrasound, and magnetic resonance imaging for surveillance of women at high familial risk for breast cancer. Journal Of Clinical Oncology, 2005. 23(33): p. 8469-8476.

36. Giuliano, V. and C. Giuliano, Improved breast cancer detection in asymptomatic women using 3D-automated breast ultrasound in mammographically dense breasts. Clinical Imaging, 2013. 37(3): p. 480- 486.

37. Berg, W.A., et al., Shear-wave Elastography Improves the Specificity of Breast US: The BE1 Multinational Study of 939 Masses. Radiology, 2012. 262(2): p. 435-449.

38. Evans, D.G., et al., MRI breast screening in high-risk women: cancer detection and survival analysis (vol 145, pg 663, 2014). Breast Cancer Research And Treatment, 2014. 147(3): p. 691-691.

39. Sung, J.S., et al., Preoperative Breast MRI for Early-Stage Breast Cancer: Effect on Surgical and Long-Term Outcomes. American Journal Of Roentgenology, 2014. 202(6): p. 1376-1382.

40. Baltzer, P.A., et al., False-positive findings at contrast-enhanced breast MRI: a BI-RADS descriptor study. AJR Am J Roentgenol, 2010. 194(6): p. 1658-63.

41. Mao, Y., et al., Stromal cells in tumor microenvironment and breast cancer. Cancer Metastasis Rev, 2013. 32(1-2): p. 303-15.

42. Joyce, J.A. and J.W. Pollard, Microenvironmental regulation of metastasis. Nat Rev Cancer, 2009. 9(4): p. 239-52.

43. Hynes, R.O. and K.M. Yamada, Fibronectins: multifunctional modular glycoproteins. J Cell Biol, 1982. 95(2 Pt 1): p. 369-77.

44. Sottile, J. and D.C. Hocking, Fibronectin polymerization regulates the composition and stability of extracellular matrix fibrils and cell-matrix adhesions. Mol Biol Cell, 2002. 13(10): p. 3546-59.

45. Inufusa, H., et al., Localization of oncofetal and normal fibronectin in colorectal cancer. Correlation with histologic grade, liver metastasis, and prognosis. Cancer, 1995. 75(12): p. 2802-8.

46. Fukuda, T., et al., Mice lacking the EDB segment of fibronectin develop normally but exhibit reduced cell growth and fibronectin matrix assembly in vitro. Cancer Res, 2002. 62(19): p. 5603-10.

47. Tan, M.H., et al., Deletion of the alternatively spliced fibronectin EIIIA domain in mice reduces atherosclerosis. Blood, 2004. 104(1): p. 11-18.

48. Astrof, S., D. Crowley, and R.O. Hynes, Multiple cardiovascular defects caused by the absence of alternatively spliced segments of fibronectin. Dev Biol, 2007. 311(1): p. 11-24.

49. Wierzbicka-Patynowski, I. and J.E. Schwarzbauer, The ins and outs of fibronectin matrix assembly. J Cell Sci, 2003. 116(Pt 16): p. 3269-76.

50. Mani, S.A., et al., The epithelial-mesenchymal transition generates cells with properties of stem cells. Cell, 2008. 133(4): p. 704-15.

51. Christofori, G., New signals from the invasive front. Nature, 2006. 441(7092): p. 444-50.

52. Ding, Y., et al., Induction of epithelial-mesenchymal transition with O- glycosylated oncofetal fibronectin. FEBS Lett, 2012. 586(13): p. 1813-20.

53. Park, J. and J.E. Schwarzbauer, Mammary epithelial cell interactions with fibronectin stimulate epithelial-mesenchymal transition. Oncogene, 2014. 33(13): p. 1649-57.

54. Johansson, S., et al., Fibronectin-integrin interactions. Front Biosci, 1997. 2: p. d126-46.

55. Redick, S.D., et al., Defining fibronectin's cell adhesion synergy site by site-directed mutagenesis. J Cell Biol, 2000. 149(2): p. 521-7.

56. Schiefner, A., M. Gebauer, and A. Skerra, Extra-domain B in oncofetal fibronectin structurally promotes fibrillar head-to-tail dimerization of extracellular matrix protein. J Biol Chem, 2012. 287(21): p. 17578-88.

57. Mao, Y. and J.E. Schwarzbauer, Fibronectin fibrillogenesis, a cell- mediated matrix assembly process. Matrix Biol, 2005. 24(6): p. 389-99.

58. Manabe, R., et al., Modulation of cell-adhesive activity of fibronectin by the alternatively spliced EDA segment. J Cell Biol, 1997. 139(1): p. 295- 307.

59. Xia, P. and L.A. Culp, Adhesion Activity in Fibronectin's Alternatively Spliced Domain ED a (EIIIA) and Its Neighboring Type III Repeats: Oncogene-Dependent Regulation. Experimental cell research, 1994. 213(1): p. 253-265.

60. Liao, Y.F., et al., The EIIIA segment of fibronectin is a ligand for integrins alpha 9beta 1 and alpha 4beta 1 providing a novel mechanism for regulating cell adhesion by alternative splicing. J Biol Chem, 2002. 277(17): p. 14467-74.

61. Chen, W. and L.A. Culp, Adhesion to fibronectin's EDbdomain induces tyrosine phosphorylation of focal adhesion proteins in Balb/c 3T3 cells. Clinical & experimental metastasis, 1998. 16(1): p. 30-42.

62. Clark, R.A., et al., Fibronectin and fibrin provide a provisional matrix for epidermal cell migration during wound reepithelialization. J Invest Dermatol, 1982. 79(5): p. 264-9.

63. Zerlauth, G. and G. Wolf, Plasma Fibronectin as a Marker for Cancer And Other Diseases. American Journal Of Medicine, 1984. 77(4): p. 685- 689.

64. Choate, J.J. and D.F. Mosher, Fibronectin concentration in plasma of patients with breast cancer, colon cancer, and acute leukemia. Cancer, 1983. 51(6): p. 1142-7.

65. Wilhelm, O., et al., Fibrin-Fibronectin Compounds In Human Ovarian Tumor Ascites And Their Possible Relation To the Tumor Stroma. Cancer Research, 1988. 48(12): p. 3507-3514.

66. Malik, G., et al., Plasma fibronectin promotes lung metastasis by contributions to fibrin clots and tumor cell invasion. Cancer Res, 2010. 70(11): p. 4327-34.

67. Chen, P., M. Cescon, and P. Bonaldo, Collagen VI in cancer and its biological mechanisms. Trends Mol Med, 2013. 19(7): p. 410-7.

68. Ioachim, E., et al., Immunohistochemical expression of extracellular matrix components tenascin, fibronectin, collagen type IV and laminin in breast cancer: their prognostic value and role in tumour invasion and progression. European Journal Of Cancer, 2002. 38(18): p. 2362-2370.

69. Sottile, J., et al., Fibronectin-dependent collagen I deposition modulates the cell response to fibronectin. Am J Physiol Cell Physiol, 2007. 293(6): p. C1934-46.

70. Erat, M.C., et al., Identification and structural analysis of type I collagen sites in complex with fibronectin fragments. Proc Natl Acad Sci U S A, 2009. 106(11): p. 4195-200.

71. Okamura, Y., et al., The extra domain A of fibronectin activates Toll-like receptor 4. J Biol Chem, 2001. 276(13): p. 10229-33.

72. Saito, S., et al., The fibronectin extra domain A activates matrix metalloproteinase gene expression by an interleukin-1-dependent mechanism. J Biol Chem, 1999. 274(43): p. 30756-63.

73. Takino, T., et al., Membrane-type 1 matrix metalloproteinase regulates fibronectin assembly and N-cadherin adhesion. Biochem Biophys Res Commun, 2014. 450(2): p. 1016-20.

74. Shi, F. and J. Sottile, MT1-MMP regulates the turnover and endocytosis of extracellular matrix fibronectin. J Cell Sci, 2011. 124(Pt 23): p. 4039-50.

75. Khan, Z.A., et al., EDB fibronectin and angiogenesis -- a novel mechanistic pathway. Angiogenesis, 2005. 8(3): p. 183-96.

76. Marini, M., et al., Endothelin-1 induces increased fibronectin expression in human bronchial epithelial cells. Biochem Biophys Res Commun, 1996. 220(3): p. 896-9.

77. Cseh, B.F.-S., Samantha ; Grall, Dominique ; et al., Autocrine fibronectin directs matrix assembly and crosstalk between cell–matrix and cell–cell adhesion in vascular endothelial cells. JOURNAL OF CELL SCIENCE, 2010. 123(22): p. 3989-3999.

78. Bae, Y.K., et al., Fibronectin expression in carcinoma cells correlates with tumor aggressiveness and poor clinical outcome in patients with invasive breast cancer. Hum Pathol, 2013. 44(10): p. 2028-37.

79. Suer, S., et al., Tissue sialic acid and fibronectin levels in human prostatic cancer. Cancer Lett, 1996. 99(2): p. 135-7.

80. Albrecht, M., et al., Fibronectin in human prostatic cells in vivo and in vitro: expression, distribution, and pathological significance. Histochem Cell Biol, 1999. 112(1): p. 51-61.

81. Arnold, S.A., et al., Urinary oncofetal ED-A fibronectin correlates with poor prognosis in patients with bladder cancer. Clin Exp Metastasis, 2016. 33(1): p. 29-44.

82. Lyons, A.J., et al., Oncofetal fibronectin and oral squamous cell carcinoma. Br J Oral Maxillofac Surg, 2001. 39(6): p. 471-7.

83. Mhawech, P., et al., EB-D fibronectin expression in squamous cell carcinoma of the head and neck. Oral Oncol, 2005. 41(1): p. 82-8.

84. Khan, Z.A., et al., ED-B FIBRONECTIN IN NON–SMALL CELL LUNG CARCINOMA. Experimental lung research, 2005. 31(7): p. 701-711.

85. van Denderen, B.J. and E.W. Thompson, Cancer: The to and fro of tumour spread. Nature, 2013. 493(7433): p. 487-8.

86. Husemann, Y., et al., Systemic spread is an early step in breast cancer. Cancer Cell, 2008. 13(1): p. 58-68.

87. Rice, J., Metastasis: The rude awakening. Nature, 2012. 485(7400): p. S55-7.

88. Kalluri, R. and R.A. Weinberg, The basics of epithelial-mesenchymal transition. J Clin Invest, 2009. 119(6): p. 1420-8.

89. Kalluri, R., EMT: when epithelial cells decide to become mesenchymal- like cells. J Clin Invest, 2009. 119(6): p. 1417-9.

90. Psaila, B. and D. Lyden, The metastatic niche: adapting the foreign soil. Nat Rev Cancer, 2009. 9(4): p. 285-93.

91. Rybak, J.N., et al., The extra-domain A of fibronectin is a vascular marker of solid tumors and metastases. Cancer Res, 2007. 67(22): p. 10948-57.

92. Mastarone, D.J., et al., A modular system for the synthesis of multiplexed magnetic resonance probes. J Am Chem Soc, 2011. 133(14): p. 5329-37.

93. Kaplan, R.N., et al., VEGFR1-positive haematopoietic bone marrow progenitors initiate the pre-metastatic niche. Nature, 2005. 438(7069): p. 820-7.

94. Erler, J.T., et al., Hypoxia-induced lysyl oxidase is a critical mediator of bone marrow cell recruitment to form the premetastatic niche. Cancer Cell, 2009. 15(1): p. 35-44.

95. Ou, J., et al., Fibronectin extra domain A (EDA) sustains CD133(+)/CD44(+) subpopulation of colorectal cancer cells. Stem Cell Res, 2013. 11(2): p. 820-33.

96. Hugo, H., et al., Epithelial-mesenchymal and mesenchymal - Epithelial transitions in carcinoma progression. Journal Of Cellular Physiology, 2007. 213(2): p. 374-383.

97. Malanchi, I., et al., Interactions between cancer stem cells and their niche govern metastatic colonization. Nature, 2012. 481(7379): p. 85-9.

98. Singh, A. and J. Settleman, EMT, cancer stem cells and drug resistance: an emerging axis of evil in the war on cancer. Oncogene, 2010. 29(34): p. 4741-51.

99. Sethi, T., et al., Extracellular matrix proteins protect small cell lung cancer cells against apoptosis: a mechanism for small cell lung cancer growth and drug resistance in vivo. Nat Med, 1999. 5(6): p. 662-8.

100. Fornaro, M., et al., Fibronectin protects prostate cancer cells from tumor necrosis factor-alpha-induced apoptosis via the AKT/survivin pathway. J Biol Chem, 2003. 278(50): p. 50402-11.

101. Pontiggia, O., et al., The tumor microenvironment modulates tamoxifen resistance in breast cancer: a role for soluble stromal factors and fibronectin through beta1 integrin. Breast Cancer Res Treat, 2012. 133(2): p. 459-71.

102. Aoudjit, F. and K. Vuori, Integrin signaling in cancer cell survival and chemoresistance. Chemother Res Pract, 2012. 2012: p. 283181.

103. Tosoian, J.J., et al., Intermediate and Longer-Term Outcomes From a Prospective Active-Surveillance Program for Favorable-Risk Prostate Cancer. J Clin Oncol, 2015. 33(30): p. 3379-85.

104. Eke, I., et al., Cetuximab attenuates its cytotoxic and radiosensitizing potential by inducing fibronectin biosynthesis. Cancer Res, 2013. 73(19): p. 5869-79.

105. Jerhammar, F., et al., Fibronectin 1 is a potential biomarker for radioresistance in head and neck squamous cell carcinoma. Cancer Biology & Therapy, 2014. 10(12): p. 1244-1251.

106. Ou, J., et al., Silencing fibronectin extra domain A enhances radiosensitivity in nasopharyngeal carcinomas involving an FAK/Akt/JNK pathway. Int J Radiat Oncol Biol Phys, 2012. 82(4): p. e685-91.

107. Nam, J.M., et al., Breast cancer cells in three-dimensional culture display an enhanced radioresponse after coordinate targeting of integrin alpha5beta1 and fibronectin. Cancer Res, 2010. 70(13): p. 5238-48.

108. Weissleder, R. and U. Mahmood, Molecular imaging. Radiology, 2001. 219(2): p. 316-33.

109. Oon, S.F., et al., Biomarker research in prostate cancer--towards utility, not futility. Nat Rev Urol, 2011. 8(3): p. 131-8.

110. Hendrikx, A., L. Safarik, and P. Hammerer, TRUS and biopsy: practical aspects. European Urology, 2002. 41(6): p. 1-10.

111. Santimaria, M., et al., Immunoscintigraphic detection of the ED-B domain of fibronectin, a marker of angiogenesis, in patients with cancer. Clin Cancer Res, 2003. 9(2): p. 571-9.

112. Tijink, B.M., et al., 124I-L19-SIP for immuno-PET imaging of tumour vasculature and guidance of 131I-L19-SIP radioimmunotherapy. European journal of nuclear medicine and molecular imaging, 2009. 36(8): p. 1235-1244.

113. Poli, G.L., et al., Radretumab radioimmunotherapy in patients with brain metastasis: a 124I-L19SIP dosimetric PET study. Cancer Immunol Res, 2013. 1(2): p. 134-43.

114. Locher, R., et al., Abundant in vitro expression of the oncofetal ED-B- containing fibronectin translates into selective pharmacodelivery of (131)I-L19SIP in a prostate cancer patient. J Cancer Res Clin Oncol, 2014. 140(1): p. 35-43.

115. Berndorff, D., et al., Imaging of tumor angiogenesis using 99mTc-labeled human recombinant anti-ED-B fibronectin antibody fragments. Journal of nuclear medicine, 2006. 47(10): p. 1707-1716.

116. Rossin, R., et al., Small-animal PET of tumor angiogenesis using a (76)Br-labeled human recombinant antibody fragment to the ED-B domain of fibronectin. J Nucl Med, 2007. 48(7): p. 1172-9.

117. Mariani, G., et al., A pilot pharmacokinetic and immunoscintigraphic study with the technetium-99m-labeled monoclonal antibody BC-1 directed against oncofetal fibronectin in patients with brain tumors. Cancer, 1997. 80(12): p. 2484-2489.

118. Ye, F., et al., A peptide targeted contrast agent specific to fibrin- fibronectin complexes for cancer molecular imaging with MRI. Bioconjug Chem, 2008. 19(12): p. 2300-3.

119. Tan, M., et al., MR molecular imaging of prostate cancer with a peptide- targeted contrast agent in a mouse orthotopic prostate cancer model. Pharm Res, 2012. 29(4): p. 953-60.

120. Zhou, Z., et al., MRI detection of breast cancer micrometastases with a fibronectin-targeting contrast agent. Nat Commun, 2015. 6: p. 7984.

121. Simberg, D., et al., Biomimetic amplification of nanoparticle homing to tumors. Proc Natl Acad Sci U S A, 2007. 104(3): p. 932-6.

122. Peng, X.H., et al., Targeted magnetic iron oxide nanoparticles for tumor imaging and therapy. Int J Nanomedicine, 2008. 3(3): p. 311-21.

123. Han, Z., et al., EDB Fibronectin Specific Peptide for Prostate Cancer Targeting. Bioconjug Chem, 2015. 26(5): p. 830-8.

124. Sun, Y., et al., MRI of breast tumor initiating cells using the extra domain- B of fibronectin targeting nanoparticles. Theranostics, 2014. 4(8): p. 845- 57.

125. Carnemolla, B., et al., A tumor-associated fibronectin isoform generated by alternative splicing of messenger RNA precursors. J Cell Biol, 1989. 108(3): p. 1139-48.

126. Balza, E., et al., A novel human fibronectin cryptic sequence unmasked by the insertion of the angiogenesis-associated extra type III domain B. Int J Cancer, 2009. 125(4): p. 751-8.

127. Ventura, E., et al., Alternative splicing of the angiogenesis associated extra-domain B of fibronectin regulates the accessibility of the BC loop of the type III repeat 8. PLoS One, 2010. 5(2): p. e9145.

128. Pini, A., et al., Design and use of a phage display library - Human antibodies with subnanomolar affinity against a marker of angiogenesis eluted from a two-dimensional gel. Journal Of Biological Chemistry, 1998. 273(34): p. 21769-21776.

129. Borsi, L., et al., Selective targeting of tumoral vasculature: comparison of different formats of an antibody (L19) to the ED-B domain of fibronectin. Int J Cancer, 2002. 102(1): p. 75-85.

130. Villa, A., et al., A high-affinity human monoclonal antibody specific to the alternatively spliced EDA domain of fibronectin efficiently targets tumor neo-vasculature in vivo. Int J Cancer, 2008. 122(11): p. 2405-13.

131. Zahnd, C., et al., Efficient tumor targeting with high-affinity designed ankyrin repeat proteins: effects of affinity and molecular size. Cancer Res, 2010. 70(4): p. 1595-605.

132. Kim, S., et al., Bio-inspired design and potential biomedical applications of a novel class of high-affinity peptides. Angew Chem Int Ed Engl, 2012. 51(8): p. 1890-4.

133. Pilch, J., et al., Peptides selected for binding to clotted plasma accumulate in tumor stroma and wounds. Proc Natl Acad Sci U S A, 2006. 103(8): p. 2800-4.

134. Sutcliffe-Goulden, J.L., et al., Rapid solid phase synthesis and biodistribution of 18F-labelled linear peptides. Eur J Nucl Med Mol Imaging, 2002. 29(6): p. 754-9.

135. D'Souza, C.A., et al., High-yielding aqueous 18F-labeling of peptides via Al18F chelation. Bioconjug Chem, 2011. 22(9): p. 1793-803.

136. Tan, M.Q., et al., Synthesis and Evaluation of Nanoglobular Macrocyclic Mn(II) Chelate Conjugates as Non-Gadolinium(III) MRI Contrast Agents. Bioconjugate Chemistry, 2011. 22(5): p. 931-937.

137. Zhou, Z., et al., Peptide targeted tripod macrocyclic Gd(III) chelates for cancer molecular MRI. Biomaterials, 2013. 34(31): p. 7683-93.

138. Ye, F., et al., Evaluation of CLT1-(Gd-DTPA) for Cancer MR Molecular Imaging in a Mouse Breast Cancer Model. Bo pu xue za zhi= Chinese journal of microwave & radio-frequency spectroscopy/Zhongguo wu li xue hui bo pu xue zhuan ye wei yuan hui bian ji, 2011. 2(5): p. 325.

139. Wu, X., et al., MR molecular imaging of prostate cancer with a small molecular CLT1 peptide targeted contrast agent. J Vis Exp, 2013(79).

140. Wu, X., et al., Synthesis and evaluation of a peptide targeted small molecular Gd-DOTA monoamide conjugate for MR molecular imaging of prostate cancer. Bioconjug Chem, 2012. 23(8): p. 1548-56.

141. Wu, X., et al., Peptide targeted high-resolution molecular imaging of prostate cancer with MRI. Am J Nucl Med Mol Imaging, 2014. 4(6): p. 525-36.

142. Neri, D., et al., Targeting by affinity–matured recombinant antibody fragments of an angiogenesis associated fibronectin isoform. Nature biotechnology, 1997. 15(12): p. 1271-1275.

143. Graf, K., et al., Monitoring therapeutical intervention with ezetimibe using targeted near-infrared fluorescence imaging in experimental atherosclerosis. Molecular Imaging, 2008. 7(2): p. 68-76.

144. Vollmer, S., et al., Extra Domain B Fibronectin as a Target for Near- Infrared Fluorescence Imaging of Rheumatoid Arthritis Affected Joints In Vivo. Molecular Imaging, 2009. 8(6): p. 330-340.

145. Matter, C.M., et al., Molecular imaging of atherosclerotic plaques using a human antibody against the extra-domain B of fibronectin. Circ Res, 2004. 95(12): p. 1225-33.

146. Konthur, Z., J. Glökler, and K. Skriner, Protein Interaction Analysis: Phage Display, in Encyclopedic Reference of Genomics and Proteomics in Molecular Medicine. 2005, Springer. p. 1509-1514.

147. Smith, G.P., Filamentous fusion phage: novel expression vectors that display cloned antigens on the virion surface. Science, 1985. 228(4705): p. 1315-1317.

Chapter 2 Phage Display Selection of Peptides Specific to EDB-FN

This chapter is adapted based on the following published paper:

Han Z, Zhou Z, Shi X, Wang J, Wu X, Sun D, Chen Y, Zhu H, Magi-Galluzzi C, and Lu ZR. EDB Fibronectin Specific Peptide for Prostate Cancer Targeting.

Bioconjugate Chemistry. 26 (5), pp 830–838. (2015).

Chapter 1 reviewed advances in cancer imaging based on fibronectin- targeting strategies and demonstrated that fibronectin, particularly oncofetal fibronectin, is a promising biomarker, based on which potent imaging technology can be developed. A prerequisite to the development of molecular imaging probes is the discovery of targeting ligands specific to oncofetal fibronectin. This chapter focuses on the verification of EDB-FN as a biomarker for cancer (the verification is performed in prostate cancer), phage display for selection of EDB-FN targeting peptides and evaluation of these peptides in targeting prostate cancer using fluorescence imaging. The evaluation of the targeting ability of these peptides to prostate cancer and breast cancer will be demonstrated in Chapter 4 and Chapter 5, respectively, along with MRI imaging results.

2.1 Abstract

Extradomain-B fibronectin (EDB-FN), one of the oncofetal fibronectin

(onfFN) isoforms, is a high molecular weight glycoprotein that mediates cell adhesion and migration. The expression of EDB-FN is associated with a number of

cancer-related biological processes such as tumorigenesis, angiogenesis, and epithelial-to-mesenchymal transition (EMT). Here, we report the development of a small peptide specific to EDB-FN for targeting prostate cancer. A cyclic nonapeptide, CTVRTSADC (ZD2), was identified using peptide phage display. A

ZD2-Cy5 conjugate was synthesized to accomplish molecular imaging of prostate cancer in vitro and in vivo. ZD2-Cy5 demonstrated effective binding to upregulated EDB-FN secreted by TGF-β-induced PC3 cancer cells following EMT.

Following intravenous injections, the targeted fluorescent probe specifically bound to and delineated PC3-GFP prostate tumors in nude mice bearing the tumor xenografts. ZD2-Cy5 also showed stronger binding to human prostate tumor specimens with a higher Gleason score (GS9) compared to those with a lower score

(GS 7), with no binding in benign prostatic hyperplasia (BPH). Thus, the ZD2 peptide is a promising strategy for molecular imaging and targeted therapy of prostate cancer.

2.2 Introduction

Prostate cancer (PCa) is the second most lethal form of cancer affecting men in the United States [1]. The lifetime risk of being diagnosed with PCa is about 1 in

7, making PCa a potential epidemic health problem in men. Early detection and timely treatment is critical to improve the survival of patients diagnosed with high- risk PCa. Prostate-specific antigen (PSA) screening is routinely used in detecting

PCa, and has resulted in a significant decrease in PCa mortality due to early

treatment [2]. However, PSA screening for PCa detection remains controversial, because non-malignant conditions such as benign prostatic hyperplasia (BPH) can also present with elevated PSA levels [2, 3]. In fact, a significant proportion of PCa diagnosed through PSA screening are considered to be low risk and indolent. Even so, most patients with low risk PCa are prescribed with treatment to avoid potential undertreatment, leading to treatment-related long-term side effects. Out of concern that the risks of overtreatment may outweigh the benefits of PSA screening, the US

Preventive Services Task Force recommended against PSA screening in 2012, despite the significant decrease in PCa mortality rate since the introduction of PSA screening in the early 1990s [4, 5]. Consequently, there is an urgent demand for highly accurate diagnostic tools for the non-invasive detection of high-risk PCa.

Malignant tumors have a unique microenvironment, which facilitates cancer cell survival, proliferation, and metastasis. The extracellular matrix (ECM) of malignant tumors exhibits an abnormally high expression of cancer-related proteins.

For example, oncofetal fibronectin (onfFN), one of the most abundant ECM components [6], plays a key role in tumorigenesis [7, 8], angiogenesis [9, 10], and metastasis [11, 12]. Various human cancers [13-15], including PCa [4, 16-19] demonstrate the presence of onfFN. In addition, increased expression of onfFN isoforms, including extradomain-B fibronectin (EDB-FN), is inversely correlated with patient survival [13, 20, 21]. onfFN mediates cell migration and invasion, both of which are essential for the stroma-cancer interaction during the development of

PCa [15]. EDB-FN is a known marker of angiogenesis and epithelial-mesenchymal transition (EMT) [17], a process that initiates cancer metastasis [11] and generates invasive mesenchymal cells in high-grade prostate tumors [22, 23]. Thus, the elevated expression of EDB-FN is a promising biomarker for the detection and diagnosis of high-risk PCa. By virtue of its abundance in the tumor ECM and its accessibility to molecular probes and imaging agents, EDB-FN is also a suitable molecular target for molecular imaging, imaging-guided surgery, and tumor- specific drug delivery.

We set out to develop small peptide sequences that can specifically bind to

EDB-FN for PCa targeting and imaging. Although EDB-FN-specific antibodies have been developed in the past, small peptides are advantageous, due to their lack of immunogenicity, cost-efficient production, and relatively simple development for translational studies. In this report, we used phage display to identify a small peptide specific to EDB-FN. A fluorescent imaging probe of the peptide was synthesized for imaging the protein marker EDB-FN in PCa. The binding property of the peptide to EDB-FN was investigated in vitro and in vivo. In addition, the PCa targeting potential of the peptide was investigated in a mouse prostate tumor model and human PCa specimens of different grades.

2.3 Materials and Methods

Materials

All reagents were used without further purification unless otherwise stated.

Fmoc-protected amino acids and 2-chlorotrityl chloride resin were purchased from

Chem-Impex International, Inc. (Wood Dale, IL). Fmoc-12-amino-4,7,10- trioxadodecanoic acid was purchased from EMD Chemicals Inc. (Gibbstown, NJ).

Sulfo-Cy5.0 NHS ester was purchased from Lumiprobe (Hallandale Beach, FL).

Anhydrous N,N-diisopropylethyl amine (DIPEA) and N,N-dimethylformamide

(DMF) were purchased from Alfa Aesar (Ward Hill, MA). Trifluoroacetic acid

(TFA) was purchased from Oakwood Products, Inc. (West Columbia, SC).

FastDigest enzymes for plasmid construction were purchased from Fermentas

(Thermo Scientific Co., Rockford, IL). Anti-fibronectin monoclonal antibody (BC-

1; ab154210) and FITC-conjugated goat polyclonal anti-mouse Fc (ab97264) were purchased from Abcam (Cambridge, UK). Rhodamine-Red-X conjugated goat polyclonal anti-rabbit IgG (H+L) was obtained from Jackson ImmunoResearch Lab

(West Grove, PA).

EDB expression in E. coli

The coding sequence for EDB was optimized by GeneOptimizer software algorithm and further synthesized (GeneArt, Regensburg,Germany) before being cloned into pQE-T7-1 expression vector (Qiagen, Valencia, CA). NdeI and PstI restriction sites in the multicloning sites were used for insertion of EDB DNA segment, which resulted in the fusion of 10 His tags at the N-terminus of the EDB protein. The expression of EDB is regulated by a T7 promoter along with the

control of a lac operator, as shown in Fig.1A. Successful ligation was verified by sequencing the inserted DNA on the recombined plasmid. Primers used for EDB sequencing are: 5ʹ-GCAGCAGCCAACTCAGCT-3ʹ (forward), and 5ʹ-

CCTCTAGAAATAATTTTGTTTAACTTT-3ʹ (reverse). Sequencing was performed by CWRU Genomics Core. The recombined plasmid was transformed into E. coli strain BL21(DE3)-T1 (Sigma-Aldrich, St Louis, MO) for EDB production. EDB expression was induced by 1 mM IPTG at the mid-log phase of

BL21, followed by incubation for 3 h. The bacteria were collected by centrifugation at 4000×g for 5 min and lysed with lysozyme (Sigma Aldrich, St Louis, MO) as instructed by manufacturer. Purification of EDB using 10 His tags was carried out with Ni Sepharose 6 Fast Flow (GE healthcare, Waukesha, WI), followed by dialysis against water and lyophilization. The size and purity of the extracted EDB protein were determined by SDS-PAGE.

Phage screening

The Ph.D C7C library (New England Biolabs, Beverly, MA) was used to screen for EDB-specific cyclic nonapeptides. Candidate peptides were selected by panning for four rounds. In each round, purified EDB fragment (100 μg/mL) was immobilized by overnight coating on non-treated 96-well plates (Corning Costar,

Tewksbury, MA, USA) 4ºC. BSA (0.5%) was used to block non-specific binding

(1 h, room temperature) followed by incubating with phages for 1 h at room temperature. Extensive washing with PBST (0.1%, 0.3%, 0.5% BSA, respectively,

three times) was performed to remove non-binding phages before eluting the bound phages with 0.1 M glycine-HCl (pH 2.2) and neutralizing with Tris-HCl (pH 9.1).

The eluted phages were titered and amplified with E. coli (ER2758), according to the user’s manual. Amplified phages in the medium were purified by ultrafiltration and PEG/NaCl precipitation. At the end of round 4, properly diluted phages were cultured on LB/IPTG/Xgal plates and DNA from 29 random blue plaques was sequenced using supplied primers (New England Biolabs) along with the phage library. Peptide sequences were acquired after translating the corresponding DNA sequences.

Use of SPR for characterizing the interaction between cyclic ZD2 and EDB

ZD2 peptide, bearing the sequence CTVRTSADC, was synthesized using standard solid-phase synthesis from Fmoc-protected amino acids on a 2-chlorotrityl chloride resin. Cyclization of the peptide was carried out by exposing the peptide to air in 10% DMSO at pH 7.0. Purification of the cyclic peptide was done using

RP-HPLC followed by lyophilization. Elution buffer (PBS containing 500 mM imidazole) containing about 4 mg/mL EDB protein was desalted with Zebra Spin

Desalting Columns (Thermo Scientific, Rockford, IL) prior to immobilization. SPR was performed using Biacore T100 (GE healthcare, Waukesha, WI). For the affinity experiments, all the samples were in phosphate buffered saline (pH 7.4) containing 150 mM NaCl and 0.05% P20. The same buffer was also used as the running buffer. To immobilize the EDB protein, the active and reference flow cells

of a CM5 series S sensor ship (GE Healthcare Life Science, Ohio, USA) were activated with 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide hydrochloride

(EDC) and N-hydroxysuccinimide (NHS) for 7 min, according to the manufacturer’s instructions. EDB protein diluted to 25 μg/mL with acetate buffer

(pH 4.5) was injected over the active cell for 3 min at 10 μL/min, resulting in an immobilization level of 2000 RU. Both the reference flow cell and active flow cell were blocked with ethanolamine, pH 8.5 for 7 min. The system was then switched to running buffer and different concentrations of the peptide were injected for 120 s at 50 μL/min, with 5 min of dissociation. The binding data was analyzed using the Scatchard plot. Affinity was calculated by fitting data points into a linear trend line using Microsoft Office Excel software. Slope of the trend line represents -1/Kd.

Fluorescence imaging probe synthesis

ZD2 peptide was synthesized in solid phase as previously described. A short

PEG (Fmoc-12-amino-4,7,10-trioxadode-canoic acid) was conjugated to the deprotected amine group as a linker. Sulfo-Cy5.0 NHS ester was sequentially conjugated to form fluorescent ZD2 probe. The probe was treated with TFA solution (TFA 94%, 1, 2-ethanedithiol 2.5%, triisobutylsilane 2.5%, water 1.0%) and precipitated with cold ether. Cyclization of the peptide was carried out by exposing the peptide to air in 10% DMSO at pH 7.0. Purification of the cyclic peptide was done using RP-HPLC followed by lyophilization. The final product, denoted as ZD2-Cy5, was characterized by MALDI-TOF mass spectrometry.

In vitro target binding

PC3 cells were purchased from American Type Culture Collection (ATCC,

Manassas, VA,), and maintained in RPMI/10% FBS medium. The cells were transfected with lentivirus to express green fluorescent protein (GFP), as previously reported [24]. Glass-bottom dishes were pre-coated with a thin layer of matrigel for cell culture to facilitate the retention of cell-secreted protein on the glass surface.

To induce EMT, the PC3 cells were cultured in the presence of TGF-β (5 ng/mL) for 5 days. Induced cells were maintained in medium containing 200 nM ZD2-Cy5 for 1 h and peptide binding was monitored with confocal microscopy. Intense shaking and washing was avoided for successful retention of secreted EDB-FN on the plate. Non-induced PC3 cells were also incubated with ZD2-Cy5. Competitive study was performed by pre-incubating the cells with 20 μM non-labeled ZD2 peptide, followed by the addition of 200 nM ZD2-Cy5. Binding of 200 nM

CERAK-Cy5 on the induced PC3 cells was used as a control. Regions of interest

(ROIs) in three independent experiments were selected to compare the percent count of positively stained cells (cells with peptide-Cy5 binding in cell periphery) in different conditions. qRT-PCR

Total RNA was extracted from the cell samples using an RNeasy Plus Kit

(Qiagen). RNA was reverse-transcribed into cDNA using the High Capacity cDNA

Transcription Kit (Applied Biosystems, Foster City, CA). Semiquantitative real-

time PCR was carried out using a SYBR Green Master Mix (Life Technologies), according to the manufacturer’s recommendations. RNA expression for the individual genes examined was normalized to the corresponding GAPDH RNA signals. Both cDNA synthesis and real-time PCR were performed in the

Mastercycler realplex2 (VWR International, West Chester, PA). Relative mRNA expression levels were calculated using the 2-ΔΔCT method [25]. The following primer sequences were used: 5ʹ-GCAGCCCACAGTGGAGTAT-3ʹ (EDB forward), 5ʹ-GGAGCAAGGTTGATTTCTTT-3ʹ (EDB reverse); 5ʹ-

ACCCAGAAGACTGTGGATGG-3ʹ (GADPH forward), and 5ʹ-

TCTAGACGGCAGGTCAGGTC-3ʹ (GADPH reverse).

In vivo fluorescence imaging

NIH athymic nude male mice (4–5 weeks) were maintained at the Athymic

Animal Core Facility at Case Western Reserve University, according to the animal protocols approved by the Institutional Animal Care and Use Committee. Mice were subcutaneously injected with 50 μL cell suspension in matrigel (4 × 107 cells/mL) in the flanks. Two to three weeks after inoculation, the tumors reached an average size of 0.7 cm in diameter. The mice were intravenously injected with

ZD2-Cy5 or CERAK-Cy5 (10 nmol). For the in vivo competitive binding study, 1

μmol of cyclic ZD2 without Cy5 was co-injected with 10 nmol ZD2-Cy5. The targeted binding of ZD2-Cy5 to the prostate tumor was assessed in vivo by Maestro

FLEX In Vivo Imaging System (Perkin-Elmer, Waltham, MA) using a yellow filter

set (spectral range of 630-800 nm, 1000 ms exposure time) for Cy5 and a blue filter set (spectral range 500-720 nm, 400 ms exposure time) for GFP. During imaging, the imaging bed was kept at 37°C. Mice were anesthetized by isofluorane inhalation via a nose cone attached to the imaging bed. The mice were imaged over 3 h post- injection. After 5 h, the mice were sacrificed and the tumor and organs were imaged with Maestro FLEX In Vivo Imaging System. For analyzing probe binding in the tumor, regions of interest (ROIs) in live mouse images were selected from the flank tumor and leg muscle.

Immunostaining of mouse tissues

Tumors and organs collected from mice after fluorescence imaging were embedded in optimum cutter temperature compound (OCT) before being frozen at

-80°C. OCT blocks were cut at 5 μm in the dark. Cold acetone was used for fixing the sections. For assessing peptide distribution in different organs, sections from the organ were directly counter-stained with DAPI and a coverslip was placed. For immunostaining with anti-EDB antibody and anti-CD31 antibody, the sections were blocked with 0.5% BSA for 1 h at room temperature before applying antibody solutions. All the sections were examined by confocal laser scanning microscopy.

SDS-PAGE and western blotting analysis

Tissues from the mice were lysed with T-PER Tissue Protein Extraction

Reagent (Thermo Scientific, Rockford, IL) supplemented with PMSF (Sigma, St

Louis, USA) and protease inhibitors (Sigma, St Louis, USA), according to

manufacturer’s instructions. Protein concentration in the extracted lysates was measured with bicinchoninic acid (BCA) assay kit (Thermo Scientific, Rockford,

IL). Proteins (20 μg) were loaded onto 4–15 % Mini-protein TGX precast gel

(Biorad, Hercules, CA) in Tris/glycine/SDS buffer for electrophoresis. Bacterial lysates (20 μL) were directly loaded without the BCA assay. After SDS-PAGE, the protein bands were visualized by Coomassie blue staining after migration. For western blotting, the proteins in the SDS-PAGE gel were transferred onto PVDF membranes (Invitrogen, Carlsbad, CA). Membranes were blotted with 0.5% BSA for 1 h at room temperature and incubated with anti-EDB antibody (BC-1, 1:2000) for 1 h. The membranes were then washed and incubated with FITC-conjugated anti-mouse secondary antibody (1:1000) in 0.5% BSA. After extensive washing, a

Typhoon Phosphor imager (General Electric, Fairfield, CT) was used for processing the membrane.

Peptide and antibody binding in human prostate specimens

Human prostate sections were acquired from OriGene (Rockville, MD).

Frozen tissue sections (5 μm) imbedded in OCT were used for immunostaining and peptide staining. Paraffin embedded samples were de-paraffinized and processed with standard antigen retrieval methods. The sections were permeabilized and fixed with cold acetone followed by 0.5% BSA blocking for 1 h at room temperature.

ZD2-Cy5 (5 μM) was then incubated with the tissue sections. Slides were counter- stained with DAPI and a coverslip was placed using Prolong Gold regent

(Invitrogen, Carlsbad, CA) before imaging. Immunostaining was performed using similar protocol as described for the mouse tissues. The stained tissues were imaged on an Olympus FV1000 confocal laser scanning microscope.

Statistical analysis

All the data are presented as mean ± SD unless otherwise stated. When two groups were compared, the two-tailed Student’s t-test was used (p<0.05 was considered significant).

2.4 Results

An EDB-FN binding peptide

The EDB fragment is a type-III-homology repeat with a highly conserved sequence of 91 amino acids encoded by a single exon [13]. The plasmid construction scheme is shown in Fig.1A. EDB-FN was successfully expressed in E. coli and SDS-PAGE was performed to confirm the size and purity of the EDB fragment (Fig.1B). Four rounds of panning yielded an enriched phage library containing phages with high EDB binding ability. Out of the 29 selected phage clones, the peptide sequence CTVRTSADC appeared 5 times. This peptide was named ZD2 and identified as the lead peptide for targeting EDB-FN.

Figure 1. Construction of the EDB-expressing plasmid and SDS-PAGE assay of lysates from EDB-expressing E. coli strain BL21. (A) The DNA-encoding

EDB fragment was inserted under the control of T7 promoter, along with 10 His tags at the N-terminal. The Lac operator (lacO) allows for the control of EDB expression with IPTG induction. (B) Stained gel with the lanes labeled as follows:

M, protein ladder; 1, BL21 cell lysate before induction with IPTG; 2, 1.5 h post- induction; 3, 3 h post-induction; and 4, purified EDB solution from lysate acquired

3 h after induction.

Cyclic ZD2 peptide (CTVRTSADC) was synthesized using standard solid phase peptide chemistry. The peptide was then labeled with the fluorescent chromophore cyanine 5 (Cy5) through a short PEG linker to obtain the fluorescent peptide probe ZD2-Cy5 (Fig.2A). The MALDI-TOF mass spectrum of ZD2-Cy5

is shown in Fig.2B. Surface Plasma Resonance (SPR) was used to characterize the interaction between the cyclic ZD2 peptide and EDB protein. EDB was immobilized onto the chip surface, through the reaction of carboxyl groups on the chip with the amine group on EDB. Since EDB contains only one amine group at the N-terminus, this immobilization method results in a uniform steric orientation of EDB, thereby optimizing site exposure for ZD2 binding. Sensorgrams, as shown in Fig.3A, were generated by flowing ZD2 peptide solutions of 1.9 μM to 250 μM over the chip surface. Scatchard plot of saturation levels at different concentrations, as shown in Fig.3B, indicates that the binding response was not a simple Langmuir equilibrium binding isotherm [26]. Two binding sites were identified by fitting data from 1.9 μM to 7.8 μM or data from 15.6 μM to 250 μM. A tight binding site had an affinity of 11 μM, and a weak site was measured to have an affinity of 384 μM.

Considering the short length of ZD2, the binding to the tight binding site is biologically competent for in vivo targeting studies.

Figure 2. Schematic illustration of the (A) synthesis of ZD2-Cy5 and (B)

MALDI-TOF mass spectrum of ZD2-Cy5.

Figure 3. Characterization of the interaction between cyclic ZD2 and the EDB fragment using SPR. (A) Sensorgrams acquired by injecting different concentrations of ZD2 (denoted by different colors) onto the immobilized EDB

protein are shown. (B) Scatchard plot of the binding levels. The slope of the trend lines represents -1/Kd for two different binding sites.

In vitro binding of ZD2-Cy5 to EDB-FN secreted by TGF-β-induced PC3 prostate cancer cells

Elevated expression of onfFN is a marker of EMT in PCa cells. The treatment of PC3-GFP human PCa cells with TGF-β resulted in an elongated mesenchymal phenotype, as compared with cells without TGF-β treatment

(Fig.4A). Up-regulation of the mRNA expression level of EDB-FN was confirmed by quantitative RT-PCR (p<0.05) (Fig.4B). TGF-β induction leads to the binding of ZD2-Cy5 in the medium (0.25 μM) to the fibrillary network of EDB-FN secreted by the induced cells. As shown in Fig.4C, GFP signal delineates the boundary of the PC3 cells, while red fluorescence clearly reveals high accumulation of ZD2-

Cy5 at the cell periphery. Pre-incubation of the induced cells with medium containing 25 μM free ZD2 peptide blocked the binding of ZD2-Cy5 due to competitive binding. ZD2-Cy5 did not exhibit significant binding to non-induced

PC3 cells. In addition, the control probe CERAK-Cy5 (25 μM) showed negligible binding to induced PC3 cells (Fig.4C). Further quantitative analysis of the Cy5 fluorescence intensity of peptide binding was also performed to validate the enhanced binding of ZD2-Cy5 to the TGF-β-induced PC3 cells (Fig.4D). These

results demonstrate that ZD2-Cy5 specifically binds to EDB-FN secreted by TGF-

β-induced PC3 cells.

Figure 4. ZD2-Cy5 binding in PC3 cells. (A) Morphology of PC3 cells with and without TGF-β induction. Images were taken by phage-contrast microscopy at 10× and 40× magnification (inset). (B) qRT-PCR analysis of EDB expression in PC3 cells with or without TGF-β induction. (C) Representative confocal fluorescence images of ZD2-Cy5 binding in non-induced PC3 cells (control), induced PC3 cells

(induced), and induced cells pre-incubated with 25 μM free ZD2 peptide

(competitive). Binding of CERAK-Cy5 in induced PC3 cells was used as a negative control. Scale bar: 100 μm. Inset: magnified image shows binding of ZD2-Cy5 in non-induced and induced PC3 cells. GFP signal (green) delineates cell shape and

ZD2-Cy5 signal (red) is clearly visible at cell periphery. (D) Percentage count of positively stained cells (cells with peptide-Cy5 binding on cell periphery) in different conditions.

In vivo binding of ZD2-based imaging probes in mouse PC3 prostate tumor model

The in vivo targeting of ZD2 to prostate tumors was demonstrated by fluorescence imaging of mice bearing PC3-GFP flank tumor xenografts after intravenous injection of ZD2-Cy5 (Fig.5). At 1.5 h after injection, ZD2-Cy5 exhibited strong binding to the GFP-labeled prostate tumor, while no apparent accumulation was seen for the control CERAK-Cy5 in the tumor. At 3 h after injection, the overall Cy5 signal in mice decreased due to systemic clearance of the peptide probe, but the tumor-to-normal signal ratio (T/N ratio) remained unchanged for ZD2-Cy5. At the 3 h timepoint, the T/N ratio of ZD2-Cy5 was around 2, while that of CERAK-Cy5 was around 1 (Fig.5B). The mice were sacrificed 5 h after the injection, and the tumor and major organs were collected and imaged to further verify the in vivo targeting specificity of Cy5-labeled ZD2 in the tumors. As shown in Fig.5C, ZD2-Cy5 shows significant signal in the tumor, and comparatively lower signal in the normal organs and tissues, confirming specific in vivo targeting of the peptide to prostate tumor. The control CERAK-Cy5 showed little accumulation in both tumor and normal organs. In addition, an in vivo competitive study was

performed by co-injecting a mixture of 1 μM cyclic ZD2 and 10 nmol ZD2-Cy5, which resulted in decreased signal in the tumor, indicating that the binding between

ZD2 and the tumor is mediated through a specific site, i.e., EDB-FN. Western blot analysis of protein lysates from the organs showed that PC3 tumor expresses substantially more EDB-FN than the liver and lung (Fig.5D). Histological analysis of the tissue sections from the tumor-bearing mice injected with ZD2-Cy5 showed that the Cy5 signal is in the ECM of the tumor, which further confirms the tumor- specificity of ZD2-Cy5 (Fig.6A). No accumulation of ZD2-Cy5 was found in the liver or lung sections (Fig.6B). EDB-FN is a known biomarker of tumor angiogenesis and its expression is associated with angiogenesis [13]. The binding of ZD2-Cy5 in the prostate tumor sections was further correlated with immunofluorescence staining of EDB-FN and CD31 (Fig.6C). The overlap of ZD2-

Cy5 binding with immunostaining of both EDB-FN and CD31 confirmed specific targeting of ZD2-Cy5 to PCa.

Figure 5. In vivo binding of ZD2-Cy5 in mice bearing PC3-GFP prostate cancer flank xenografts. (A) Representative fluorescence images of PC3-bearing mouse at 1 min before injection (labeled as pre), 90 and 180 min after intravenous injection of 10 nmol ZD2-Cy5, non-specific control CERAK-Cy5, and a mixture of 1 μM cyclic ZD2 and 10 nmol ZD2-Cy5 (competitive). (B) Fluorescence intensity ratio between tumor and normal tissues (T/N ratio) from mice injected with ZD2-Cy5, CERAK-Cy5, or a mixture of 1 μM cyclic ZD2 and 10 nmol ZD2-

Cy5 (competitive). Images of tissues obtained 1 min before injection (pre), and 90 and 180 min after injection were analyzed (N=3). (C) Representative fluorescence

images of organs harvested from PC3-GFP tumor-bearing mouse 5 h after injection of ZD2-Cy5, CERAK-Cy5, or a mixture of 1 μM cyclic ZD2 and 10 nmol ZD2-

Cy5 (competitive). The organs are represented with numbers: 1, tumor; 2, liver; 3, muscle; 4, heart; 5, brain; 6, lung; and 7, spleen. (D) Western blot analysis of the liver, lung, and tumor protein lysates. Antibody BC-1 was used to characterize the

EDB-FN expression, and expression of β-actin was used as the loading control.

Figure 6. Histological analysis of tissue sections of PC3 tumors from mice injected with ZD2-Cy5 or CERAK-Cy5. (A) Representative fluorescence images of tumor sections from PC3 tumor xenografts injected with ZD2-Cy5 or CERAK-

Cy5. (B) Representative fluorescence images of the liver and lung sections from the same mice showing non-binding of ZD2-Cy5. Scale bar: 20 μm. (C) Correlation of ZD2 distribution with EDB-FN (BC-1) distribution and blood microvessel distribution (anti-CD31) in the PC3 tumor sections. Scale bar: 20 μm. Pseudo- colors are assigned as follows: red, peptide; yellow, antibody; and blue, nucleus.

100

Binding study in human prostate specimens of varying aggressiveness

Tumor targeting of the ZD2 peptide was further assessed in human prostate tumor sections with varying Gleason scores, while a human prostate BPH tissue section was used as control. As shown in Fig.7, ZD2-Cy5 significantly binds to prostate tumors (GS7 and GS9), but not to the BPH tissue sections. The fluorescence intensity of ZD2-Cy5 in the tumor sections was found to correlate with the Gleason score-based tumor aggressiveness, with stronger staining seen in both the stromal and glandular areas of specimens with a higher Gleason score. This finding is consistent with a previous study that shows an overexpression of EDB-

FN in prostate carcinoma compared with BPH [16, 27]. A similar trend was observed with BC-1 immunofluorescence staining. These results demonstrate that the specific binding of ZD2 to EDB-FN also leads to effective binding of ZD2 to high-risk human prostate tumors, with the potential to differentiate between the tumor aggressiveness.

101

Figure 7. ZD2-Cy5 binding in human prostate sections. Prostate cancer tissues with varying Gleason scores (GS7, 3+4, GS9, and 4+5) and BPH tissue were evaluated. H&E staining, immunofluorescence staining with BC-1, and ZD2-Cy5 staining are shown. Scale bar: 100 μm. Pseudo-colors in the confocal images are assigned as follows: red, peptide; green, BC-1; and blue, nucleus.

102

2.5 Discussions and conclusions

Specific targeting to cancer biomarkers is essential for the development of non-invasive imaging technology, so as to facilitate accurate detection and efficacious therapy of high-risk PCa. EDB-FN is one of the hallmarks of EMT and its expression is elevated in post-EMT PCa cells [16]. EDB-FN is also involved in tumor angiogenesis [22]. Consequently, we identified the cyclic nonapeptide ZD2 specific to EDB-FN using phage display. The binding specificity of this peptide was first verified using the fluorescent probe ZD2-Cy5 ex vivo in post-EMT PC3 prostate cancer cells. EMT induction of PC3 cells by TGF-β resulted in substantial up-regulation of EDB-FN. ZD2-Cy5 showed enhanced binding in post-EMT PC3 cells, but no binding in non-induced cells. The strong binding signal of ZD2-Cy5 was localized to the periphery of the induced PC3 cells, which is consistent with the fact that FN is an ECM protein. The tumor-specificity of the peptide was also demonstrated in mice bearing PC3-GFP prostate cancer xenografts and in human prostate tumor specimens with varying Gleason scores. Gleason score is the most commonly used pathological grading system for the clinical management of PCa.

Our results suggest that EDB-FN is a potential marker of high-risk PCa and the

ZD2 peptide is a viable probe for targeting this biomarker.

Currently, needle-biopsy Gleason scoring is routinely used in the risk- stratified management and therapeutic intervention of PCa. The goal of this management strategy is to minimize treatment-related risks in patients who do not

103

benefit from the treatment [28]. However, the accuracy of the diagnostic procedure is often compromised by the heterogeneity of PCa cells within the same prostate tissue and the inadequacy of prostate sampling from needle-biopsy. Consequently, a molecular imaging technology that can non-invasively map all areas of the prostate tissue in high-risk PCa is more advantageous over invasive biopsy and could provide more accurate differential diagnosis. Since EDB-FN is a molecular marker of PCa, angiogenesis, EMT, and is characteristic of high-risk PCa, the ZD2 peptide may open avenues in developing imaging agents for non-invasive molecular imaging and accurate diagnosis of PCa [29-31]. Although antibodies have been developed to target EDB-FN, small peptides are advantageous due to their lack of immunogenicity and cost effective scale-up for mass production.

We have shown that ZD2-Cy5 can be used to effectively visualize PCa by fluorescence imaging. This fluorescent peptide probe or a further optimized probe could potentially serve to delineate the margins of high-risk PCa during image- guided surgery. This would facilitate the complete resection of high-risk prostate tumors while minimizing unnecessary removal of healthy tissue [32]. It is also possible to use this peptide for the development of targeted imaging agents for clinical imaging modalities, including SPECT, PET, and MRI, for non-invasive detection and differential diagnosis of high-risk PCa in the clinic. Currently, we are exploring the potential of the ZD2 peptide in the synthesis of targeted MRI contrast agents and PET probes for non-invasive molecular imaging of PCa.

104

The ZD2 peptide could also be used in developing targeted drug delivery systems to deliver therapeutics, including chemotherapeutics and therapeutic nucleic acids, and to improve their efficacy. Because of the increased expression of

EDB-FN in the tumor ECM, the ZD2 peptide could facilitate rapid enrichment of the therapeutics in the tumor to improve tumor targeting efficiency, in turn minimizing systemic circulation of the therapeutics, reducing potential side effects, and enhancing uptake in cancer cells to improve therapeutic efficacy.

Using phage display, we have identified a peptide CTVRTSADC (ZD2), which is specific to EDB-FN, for targeting PCa. ZD2 can specifically bind to EDB-

FN produced by post-EMT PC3 prostate cancer cells, mouse prostate tumor xenografts, and to human prostate tissue specimens with high Gleason scores. The fluorescent ZD2 peptide probe can facilitate molecular imaging of PCa and image- guided surgery. The ZD2 peptide is a promising tool for the design of targeting agents for molecular imaging and targeted drug delivery in PCa diagnosis and therapy.

105

References

1. Siegel, R., D. Naishadham, and A. Jemal, Cancer statistics, 2012. CA: a cancer journal for clinicians, 2012. 62(1): p. 10-29.

2. Barry, M.J., Clinical practice. Prostate-specific-antigen testing for early diagnosis of prostate cancer. N Engl J Med, 2001. 344(18): p. 1373-7.

3. Dhanasekaran, S.M., et al., Delineation of prognostic biomarkers in prostate cancer. Nature, 2001. 412(6849): p. 822-6.

4. Melnikow, J., et al., Counterpoint: Randomized trials provide the strongest evidence for clinical guidelines: The US Preventive Services Task Force and Prostate Cancer Screening. Med Care. 51(4): p. 301-3.

5. Moyer, V.A., Screening for prostate cancer: U.S. Preventive Services Task Force recommendation statement. Ann Intern Med. 157(2): p. 120- 34.

6. Wernert, N., The multiple roles of tumour stroma. Virchows Arch, 1997. 430(6): p. 433-43.

7. Park, J. and J.E. Schwarzbauer, Mammary epithelial cell interactions with fibronectin stimulate epithelial-mesenchymal transition. Oncogene, 2014. 33(13): p. 1649-57.

8. Alisson-Silva, F., et al., Increase of O-glycosylated oncofetal fibronectin in high glucose-induced epithelial-mesenchymal transition of cultured human epithelial cells. PLoS One, 2013. 8(4): p. e60471.

9. Borsi, L., et al., Use of human recombinant antibodies to the marker of angiogenesis ED-B in cancer therapy. Tumori, 2001. 87(6): p. S8-S10.

106

10. Hajitou, A., R. Pasqualini, and W. Arap, Vascular targeting: recent advances and therapeutic perspectives. Trends Cardiovasc Med, 2006. 16(3): p. 80-8.

11. Mani, S.A., et al., The epithelial-mesenchymal transition generates cells with properties of stem cells. Cell, 2008. 133(4): p. 704-15.

12. Balanis, N., et al., Epithelial to mesenchymal transition promotes breast cancer progression via a fibronectin-dependent STAT3 signaling pathway. J Biol Chem, 2013. 288(25): p. 17954-67.

13. Kaspar, M., L. Zardi, and D. Neri, Fibronectin as target for tumor therapy. Int J Cancer, 2006. 118(6): p. 1331-9.

14. Ratliff, T.L., L.R. Kavoussi, and W.J. Catalona, Role of fibronectin in intravesical BCG therapy for superficial bladder cancer. J Urol, 1988. 139(2): p. 410-4.

15. Tuxhorn, J.A., G.E. Ayala, and D.R. Rowley, Reactive stroma in prostate cancer progression. J Urol, 2001. 166(6): p. 2472-83.

16. Albrecht, M., et al., Fibronectin in human prostatic cells in vivo and in vitro: expression, distribution, and pathological significance. Histochem Cell Biol, 1999. 112(1): p. 51-61.

17. Freire-de-Lima, L., et al., Involvement of O-glycosylation defining oncofetal fibronectin in epithelial-mesenchymal transition process. Proc Natl Acad Sci U S A, 2011. 108(43): p. 17690-5.

18. Suer, S., et al., Tissue sialic acid and fibronectin levels in human prostatic cancer. Cancer Lett, 1996. 99(2): p. 135-7.

19. Sonmez, H., et al., Tissue fibronectin levels of human prostatic cancer, as a tumor marker. Cancer Biochemistry Biophysics, 1995. 15(2): p. 107- 110.

107

20. Inufusa, H., et al., Localization of oncofetal and normal fibronectin in colorectal cancer. Correlation with histologic grade, liver metastasis, and prognosis. Cancer, 1995. 75(12): p. 2802-8.

21. Menzin, A.W., et al., Identification of oncofetal fibronectin in patients with advanced epithelial ovarian cancer: detection in ascitic fluid and localization to primary sites and metastatic implants. Cancer, 1998. 82(1): p. 152-8.

22. Barron, D.A. and D.R. Rowley, The reactive stroma microenvironment and prostate cancer progression. Endocr Relat Cancer, 2012. 19(6): p. R187-204.

23. Jaggi, M., et al., N-cadherin switching occurs in high Gleason grade prostate cancer. Prostate, 2006. 66(2): p. 193-9.

24. Tan, M., et al., MR molecular imaging of prostate cancer with a peptide- targeted contrast agent in a mouse orthotopic prostate cancer model. Pharmaceutical research, 2012. 29(4): p. 953-960.

25. Schmittgen, T.D. and K.J. Livak, Analyzing real-time PCR data by the comparative CT method. Nature protocols, 2008. 3(6): p. 1101-1108.

26. Wilkinson, K.D., Quantitative analysis of protein-protein interactions, in Protein-Protein Interactions. 2004, Springer. p. 15-31.

27. Jankovic, M.M. and M.M. Kosanovic, Fibronectin pattern in benign hyperplasia and cancer of the prostate. Dis Markers, 2008. 25(1): p. 49- 58.

28. Glass, A.S., K.C. Cary, and M.R. Cooperberg, Risk-based prostate cancer screening: who and how? Current urology reports, 2013. 14(3): p. 192- 198.

108

29. Locher, R., et al., Abundant in vitro expression of the oncofetal ED-B- containing fibronectin translates into selective pharmacodelivery of (131)I-L19SIP in a prostate cancer patient. J Cancer Res Clin Oncol, 2014. 140(1): p. 35-43.

30. Wu, X., et al., Peptide targeted high-resolution molecular imaging of prostate cancer with MRI. Am J Nucl Med Mol Imaging, 2014. 4(6): p. 525-36.

31. Wu, X., et al., Synthesis and evaluation of a peptide targeted small molecular Gd-DOTA monoamide conjugate for MR molecular imaging of prostate cancer. Bioconjug Chem. 23(8): p. 1548-56.

32. Vahrmeijer, A.L., et al., Image-guided cancer surgery using near-infrared fluorescence. Nat Rev Clin Oncol, 2013. 10(9): p. 507-18.

109

Chapter 3 Design and development of a MRI molecular imaging agent that targets EDB-FN for the differential diagnosis of prostate cancer

This chapter is adapted based on the following published paper:

Han Z, Li Y, Roelle S, Zhou Z, Liu Y, Sabatelle R, DeSanto A, Yu X, Zhu H, Magi-

Galluzzi C, and Lu ZR. A Targeted Contrast Agent Specific to an Oncoprotein in

Tumor Microenvironment with the Potential for Detection and Risk Stratification of Prostate Cancer with MRI. Bioconjugate Chemistry. (2017).

3.1 Design and synthetic route

Magnetic resonance imaging (MRI) offers higher spatial resolution images of soft tissues and causes no ionizing radiation to patients. The challenge with MRI is its inherently low sensitivity in molecular imaging. However, effective MR molecular imaging of cancer can still be achieved by targeting highly upregulated oncofetal FN in tumor stoma, which will allow specific binding of sufficient targeted contrast agents to generate detectable MR signal. Despite the success of nuclear imaging using antibodies targeting EDB-FN, large size of antibodies may not be suitable for targeted MRI contrast agents, especially for gadolinium-based contrast agents because agents with large sizes cannot be rapidly excreted after the diagnostic imaging. Small molecular contrast agents with small peptides can address the limitations of the antibodies in MRI. Therefore, the rationale behind designing an oncofetal-FN-targeting contrast agent is that the contrast agent should

110

be small enough to penetrate into tumor and provide sufficient contrast. In addition, the complexation stability and biodistribution of the contrast agent should be carefully considered due to toxic effect of Gd. Previously, our lab has demonstrated the use of CLT1-dL-(Gd-DOTA)4 [1-3] for prostate cancer imaging. This contrast agent was able to accumulate in PC3 tumors by targeting Fibronectin-Fibrin complex, a biomarker has also been widely used for cancer imaging. In this study, the EDB-FN targeting peptide, ZD2, was modified with Gd(HP-DO3A), to yield a small contrast agent, ZD2-Gd(HP-DO3A). This agent evaluated for stability and efficacy in differential diagnosis of LNCaP and PC3 tumors.

3.2 Materials and Methods

Materials and Cell culture

All reagents for chemical synthesis were purchased from Sigma Aldrich unless stated otherwise. PC3 and LNCaP cells were purchased from American Type

Culture Collection (ATCC, Manassas, VA, USA), and grown in RPMI medium

(Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 10% Fetal

Bovine Serum (FBS) and 1% Penn/Strep at 37°C in 5% CO2. To construct GFP- expressing cell lines, cells were transfected with lentivirus as previously reported

[4]. Matrigel 3D culture was formed following the 3D “on-top” protocol [5]. Briefly, a glass-bottom plate was coated with a thick layer of freshly thawed matrigel, followed by solidifying the matrigel at 37C for 15-30 min. Cells were then plated onto the coated plate with medium containing 5% matrigel. The culture was

111

maintained for at least 4-6 days and then ZD2-Cy5.5 (250 nM) was added to the culture medium. After 1 h, cells in the 3D matrigel were imaged with confocal laser fluorescence microscopy.

Animals

Male athymic nude mice were purchased from Case Comprehensive Cancer

Center (Cleveland, OH, USA) and housed in the Animal Core Facility at Case

Western Reserve University. All animal experiments were performed in accordance with the animal protocol approved by the CWRU Institutional Animal Care and

Use Committee. Athymic nude male mice were subcutaneously injected with 100

μL cell suspension (4×107 cells/mL) in matrigel matrix (Corning Bioscience,

Corning, NY) to initiate tumors. Mice with tumors of 5-8 mm in diameter were used for imaging studies.

Quantitative PCR analysis of cancer cells

Total RNA were isolated from cells using RNeasy Plus Kit and reverse- transcribed into cDNA using high capacity cDNA Transcription Kit.

Semiquantitative real-time PCR was performed using a SYBR Green Master Mix

(Life Technologies, Carlsbad, CA). Primers used in this study were as follows:

EDB: forward: 5ʹ-CCTGGAGTACAATGTCAGTG-3ʹ; reverse: 5ʹ-

GGTGGAGCCCAGGTGACA-3ʹ. β-actin: forward, 5'-

GTTGTCGACGACGAGCG-3ʹ; reverse, 5'-AGCACAGAGCCTCGCCTTT-3ʹ.

112

PSMA: forward, 5ʹ-AACTGGACCCCAGGTCTGGA-3ʹ; reverse, 5ʹ-

GAGGATTTTATAAACCACCCGAA-3ʹ; EGFR: forward, 5ʹ-

GGAGAACTGCCAGAAACTGACC-3ʹ; reverse, 5ʹ-

GCCTGCAGCACACTGGTTG-3ʹ; E-cadherin: forward, 5ʹ-

TGCCCAGAAAATGAAAAAGG-3ʹ; reverse, 5ʹ-

GTGTATGTGGCAATGCGTTC-3ʹ; N-cadherin: forward, 5ʹ-

ACAGTGGCCACCTACAAAGG-3ʹ; reverse, 5ʹ-

CCGAGATGGGGTTGATAATG-3ʹ; vimentin: forward, 5ʹ-

GAGAACTTTGCCGTTGAAGC-3ʹ; reverse, 5ʹ-

GCTTCCTGTAGGTGGCAATC-3ʹ; Real-time PCR were carried out on the

Mastercycler realplex2. The relative mRNA expression levels were calculated using 2-ΔΔCT method, and normalized to the corresponding β-actin levels.

Western Blot

Tumors harvested from different mouse organs and normal murine tissues were homogenized in 200-500 μL T-PER buffer (Thermo Fisher

Scientific, Rockford, IL, USA) mixed with protease inhibitor cocktail (Sigma-

Aldrich, St. Louis, MO, USA) and the resulting lysates were centrifuged for 10 min at 10,000 g at 4°C. The supernatants were collected and protein content was determined by BCA assay (Bio-rad). Protein extracts (20 μg) were subjected to

SDS-PAGE, followed by blotting onto a PVDF membrane. The blots were washed and incubated with 1:1000 anti-EDB-FN BC-1 antibody (Abcam, Cambridge, MA,

113

USA). HRP-conjugated anti-mouse IgG antibody was used as secondary antibody.

Anti-β-actin antibody was used as loading control. The blots were developed using the ChemiDoc XRS System (Bio-rad).

MRI contrast agent synthesis

The ZD2 peptide sequence, TVRTSAD, was synthesized in solid phase using Fmoc chemistry. A short PEG spacer (Fmoc-12-amino-4,7,10-trioxadodecanoic acid), and 5-hexynoic acid, were also sequentially conjugated in solid phase followed by trifluoroacetic acid (TFA) treatment, which yielded ZD2-PEG- propargyl. Azido-Gd(HP-DO3A) was synthesized as reported previously [6]. Click chemistry reaction between ZD2-PEG-propargyl (1 Eq) and azido-Gd(HP-DO3A)

(1.1 Eq) was performed in a 2:1 mixture of t-butanol and water under nitrogen following addition of [Cu(MeCN)4]PF6 (0.02 Eq) and TBTA (Tris[(1-benzyl-1H-

1,2,3-triazol-4-yl)methyl]amine, 0.02 Eq). The final product, named ZD2-Gd(HP-

DO3A), was purified by high performance liquid chromatography (HPLC) on an

Agilent 1100 HPLC system equipped with a semi-preparative C18 column. The gradient of HPLC was 100% water for 10 min and 0-20% acetonitrile in water for another 20 min and 50-100% acetonitrile in water for 5 min. The Gd(III) content was measured by inductively coupled plasma optical emission spectroscopy (ICP-

OES Optima 3100XL, Perkin-Elmer, Norwalk, CT). ZD2-Gd(HP-DO3A) was characterized by matrix-assisted laser desorption/ionization time-of-flight

(MALDI-TOF) mass spectrometry on a Voyager DE-STR spectrometer

114

(PerSeptive BioSystems) in linear mode with R 2,5-dihydroxybenzoic acid as a matrix (M+1: 1600.2 Da; 1601.79 Da (calc.), Fig. S1B). HPLC was used to characterize the purity of the product (Fig. S1C) using the same gradient for production. The fibrin-FN clot targeting contrast agent, CREKA-Gd(HP-DO3A), was synthesized with a similar approach. It was characterized by MALDI-TOF mass spectrometry (M+1: 1457.87 Da; 1459.56 Da (calc.)). The binding affinity of the contrast agent to EDB protein was measured using surface plasma resonance

(SPR) according to a previous protocol [7]. Briefly, different concentrations of the contrast agent were injected to a CM5 series S chip surface (GE Healthcare Life

Science, Ohio, USA) coated with immobilized EDB proteins. Sensorgrams generated from the binding between contrast agents and EDB were recorded with

Biacore T100 (GE healthcare, Waukesha, WI), and analyzed using Scatchard plot.

Affinity was calculated by fitting data into a linear line, with the slope of the line representing -1/Kd.

Fluorescence probe synthesis

Synthesis of the fluorescent probe, ZD2-PEG-Cy5.5 was achieved by conjugating ZD2-PEG on resin to Cy5.5 NHS ester (Lumiprobe, Hallandale Beach,

FL, USA), followed by TFA treatment. The product was precipitated in cold ether and freeze dried and characterized by MALDI-TOF mass spectrometry (M=1:

1473.76 Da; 1472.78 Da (calc.)). Concentration of ZD2-PEG-Cy5.5 in PBS was quantified by measuring the absorbance at 450 nm.

115

Relaxivity measurement at 7T

To accurately capture T1 maps, a fast T1 mapping method was developed.

This accelerated multi-slice T1-mapping method combines a saturation recovery look-locker sequence [8, 9] and spiral k-space sampling [10]. The NMR tubes filled with 3.12 - 50 μM ZD2-Gd(HP-DO3A) were positioned in a mouse coil and scanned with T1- weighted spin echo sequence, cartesian T1-mapping sequence, and spiral T1-mapping sequence. Axial images were acquired. CNR values from the T1- weighted sequence were calculated from signal from the samples subtracting surrounding water signal, and divided by noise. T1 values of the samples were reconstructed by fitting signal change during the acquisition time to the adapted T1- relaxation curve.

In vitro transmetallation

Aqueous solutions (0.1 mL, 2 mM-Gd) of ZD2-Gd(HP-DO3A) and

ProHance were mixed with 0.9 mL human serum and incubated at room temperature for 2 h. The plasma mixtures were transferred to CF-10 centrifugal filters (molecular weight cut-off 10 KDa) and centrifuged at 4,000 rpm and 25°C for 150 min. During centrifugation, the solution in the upper reservoir was agitated using a pipette every 20 min for the plasma mixtures. The content of metal ions in both the upper reservoir and the filtrates was determined by ICP-OES after appropriate dilution. The degree of transmetallation of the contrast agents with

Zn(II) or Cu(II) ions in the plasma was evaluated using the percentage of Zn(II) or

116

Cu(II) ions filtered through the membrane, calculated as Zn(Cu)% = (concentration of Zn(II) or Cu(II) in the filtrates)/(total Zn(II) or Cu(II) concentration before centrifugal filtration) × 100. Human serum was used a control.

In vivo transmetallation

Balb/c mice (Charles River Laboratories, Wilmington, MA, USA) were randomly divided into groups of ZD2-Gd(HP-DO3A) and ProHance (n = 5) and placed in the metabolism cages, 48 hours prior to injection. After acclimatization, the mice were injected with either ZD2-Gd(HP-DO3A) or ProHance at a dose of

0.1 mmol-Gd/kg via tail vein. Urine samples were collected at 12 h pre-injection, and then 8, 24, 48, and 72 h post-injection from the metabolic box. The collected urine samples were centrifuged at 4,000 rpm for 15 min. The concentration of

Gd(III), Zn(II), Cu(II), and Ca(II) in the supernatant of the urine samples was determined by ICP-OES after appropriate dilutions.

Biodistribution

PC3 tumor-bearing athymic nude mice (sixteen mice) were randomly divided into 4 groups (four mice each group). Then two of the groups were injected with either ZD2-Gd(HP-DO3A) or ProHance at a dose of 0.1 mmol-Gd/kg via tail vein. The animals were sacrificed at 2 and 7 days after injection. The blood and tissue samples, including brain, femur, heart, lung, liver, muscle, spleen, and kidney,

117

were collected and weighed. The tissue samples were then cut into small pieces and mixed with ultra-pure nitric acid (1.0 mL, 70%, EMD, Gibbstown, NJ). The tissue samples were liquefied within 2 weeks and the solution was transferred to a centrifuge tube and centrifuged at 14,000 rpm for 15 min. The supernatant (0.2 mL) was diluted with 9.8 mL deionized water and further centrifuged at 14,000 rpm for

15 min. The Gd(III) concentration in the final supernatant was measured by ICP-

OES. The average Gd(III) content in each organ or tissue was calculated from the measured Gd(III). In addition, 8 LNCaP tumor-bearing mice were also tested with similar injections of ZD2-Gd(HP-DO3A) and ProHance (n = 4). The Gd content in different organ/tissues was calculated as the percentage of injected dose per gram of organ/tissues (% Dose/g).

Histological analysis

After fluorescence imaging of the organs, they were embedded in O.C.T and sectioned at 5 μm thickness, and mounted on to glass slides. The slides were fixed and permeabilized with cold acetone, followed by blocking with 1% BSA for

1 hour. DAPI or EDB-FN specific antibody, BC-1 (Abcam, Cambridge, MA), were applied subsequently for staining. Unbound antibodies were washed with TBS-T

(0.1%). Alexa Fluor 594-conjugated goat anti-mouse IgG H&L (Abcam) was used as the secondary antibody. Prolong Gold anti-fade solution (Invitrogen, Grand

118

Island, NY), was used to cover the slides. The stained tissue sections were imaged using confocal laser scanning microscopy.

Ex vivo fluorescence imaging

The targeted binding of ZD2-Cy5.5 to the prostate tumor was assessed ex vivo using Maestro FLEX In Vivo Imaging System (Caliper Life Sciences,

Hopkinton, MA) using a red filter set (spectral range of 630-910 nm, 1000 ms exposure time). ZD2-Cy5.5 (10 nmol) was administered to study the in vivo distribution of ZD2-based probes. At 3 h post-injection, the mice were sacrificed to image the tumors and the organs.

In vivo MRI

All MRI experiments were performed on a horizontal 7T Bruker scanner

(Bruker Biospin Co., Billerica, MA). Imaging experiments were performed when the tumor size reached 5-8 mm in diameter. ZD2-Gd(HP-DO3A) was administered intravenously at the dose of 0.1 mmol/kg after acquiring pre-contrast images. Post- contrast images were obtained at 10 min interval up to 30 min. The axial slices of mouse at the tumor location were acquired using T1-weighted spin echo sequence with the following parameters: field of view (FOV): 3 cm; slice thickness: 1.2 mm; interslice distance: 1.2 mm; TR: 500 ms, TE: 8.1 ms; flip angle: 80°; average: 2; matrix size: 128 × 128. Contrast-to-noise ratio (CNR) at each time point was calculated by measuring the signal intensity ratio between tumor areas and muscle

119

areas, scaled to image noise. ProHance® (gadoteridol) was used as a control in the imaging experiments.

T1 mapping

The following imaging parameters were used for T1 mapping methods: flip angle 10°; echo time 2.09 ms; slice thickness 1 mm; number of average 1; field of view 3 × 3 cm2; matrix size 128 × 128. For each slice, 50 images that covered 5 s of the saturation recovery curve were acquired within an interval of 100 ms. Proton density (M0) images were acquired with repetition time (TR) of 2s. Cartesian

MRSLL images were also acquired as a validation of spiral MRSLL method. Center

64 k-space lines were acquired in Cartesian method as described previously [9].

Other acquisition parameters were the same as spiral MRSLL.

3.3 Results

Contrast agent synthesis and characterization

To verify the effectiveness of molecular MRI of EDB-FN in differential diagnostic imaging and characterization the aggressiveness of prostate cancer, we first designed and synthesized a small molecular targeted MRI contrast agent ZD2-

Gd(HP-DO3A), by conjugating EDB-FN-targeting peptide ZD2 [7] to a clinical macrocyclic agent ProHance® [Gd(HP-DO3A)] (Figure 1A). Identity and purity of the final product were confirmed with MALDI-TOF mass spectra (Figure 1B) and HPLC (Figure 1C). ZD2-Gd(HP-DO3A) exhibited a binding affinity of 1.7

µM (Figure 1D and 1E) to the EDB fragment, which is sufficient for reversible

120

binding, so as to produce detectable signal enhancement of the molecular target in contrast enhanced MRI, as well as to facilitate rapid clearance from the body post- imaging and minimize potential side-effects. ZD2-Gd(HP-DO3A) demonstrated high chelation stability, comparable to its clinical analog ProHance, with no evident transmetallation during 24 hours of incubation with endogenous metal ions Cu2+,

Ca2+, and Zn2+ in PBS (pH = 7.4) (Figure 2A). We next compared the stability of

ZD2-Gd(HP-DO3A) with ProHance and the linear contrast agents, MultiHance® and OmniScan®, against transmetallation in blood plasma containing Cu2+, Ca2+, and Zn2+ ions. As expected, ZD2-Gd(HP-DO3A) had the same high stability as

ProHance over both the linear agents (Figure 2B). This superior chelation stability is critical to minimizing potential side effects of the targeted contrast agent for its clinical application.

121

Figure 1. Synthesis and characterization of ZD2-Gd(HP-DO3A). A, synthesis route, B, MALDI-TOF mass spectrum (detected molecular weight: 1600.2 Da; calculated molecular weight: 1601.79 Da), and C, reversed-phase high performance liquid chromatography (RP-HPLC) analysis of ZD2-Gd(HP-DO3A). D,

Sensorgram measured by surface plasma resonance (SPR) analysis of varying concentrations of ZD2-Gd(DO3A). E, Plot of response/concentration versus response for the Sensorgram shown in (c) yields the binding affinity of 1.7 µM

122

between ZD2-Gd(HP-DO3A) and EDB protein (organ trend line). Data that represent a weak binding site was also shown (blue trend line). F, Plot of 1/T1 measured by T1 mapping in phantom containing increasing concentration of ZD2-

Gd(HP-DO3A) (abbreviated as ZD2) or ProHance (abbreviated as Pro.) in PC3

-1 -1 tumor lysates. Fitting of the data yielded r1 relaxivities of 4.12 mM s for ZD2-

Gd(HP-DO3A) and 3.17 mM-1s-1 for ProHance.

Figure 2. In vitro complexation stability and transmetallation analysis of ZD2-

Gd(HP-DO3A). A, relative complexed Gd content assayed during 200-min incubation of ZD2-Gd(HP-DO3A) in phosphate buffer saline (PBS) with and without metal ions (Cu, Ca, and Zn). No decrease in Gd content is observed, demonstrating a high chelation stability. B, In vitro transmetallation analysis of

MultiHance, OmniScan, ProHance (abbreviated as Pro.), and ZD2-Gd(HP-DO3A)

123

(abbreviated as ZD2) in mouse serum. Percentage of transmetallation is indicated by the increase in measured Cu, Ca, or Zn ions in the free form.

EDB-FN upregulation is associated with the metastatic phenotype of prostate cancers

Unlike LNCaP cells, PC3 cells exhibit typical characteristics of post epithelial-to-mesenchymal transition (EMT) phenotype, including loss of E- cadherin and upregulation of N-cadherin and vimentin (Figure 3A) [11]. This EMT signature is strongly associated with enhanced cancer cell migration, metastasis, and drug resistance [12]. Importantly, EDB-FN expression is significantly upregulated in PC3 cells as compared to LNCaP cells (Figure 3A and 3B), indicating that EDB-FN is a molecular marker of high-risk prostate cancer [7, 13,

14]. It is interesting to note that the expression of epidermal growth factor receptor

(EGFR) and prostate-specific membrane antigen (PSMA) did not correlate with the metastatic potential of the two cell lines. When cultured on top of a thick layer of matrigel, LNCaP and PC3 cells displayed distinctive spheroid-forming abilities

(Figure 3C). Only PC3 cells were able to penetrate into the matrix and form 3D spheroids, resembling tumors cultured in a native microenvironment, while the

LNCaP cells formed cell clusters that distributed primarily on the matrigel surface.

We next investigated whether the EDB-FN targeting peptide, ZD2, could differentiate between GFP-labeled PC3 and LNCaP cells in 3D culture. Cy5.5-

124

labeled ZD2 peptide (ZD2-Cy5.5) was able to bind EDB-FN secreted by the PC3 spheroids, highlighted by the dispersed red fluorescence signal from the spheroids, whereas no peptide binding was detected on the LNCaP cell clusters (Figure 3D).

This specific targeting of ZD2 to aggressive prostate cancer was also validated by examining the Cy5.5 signal in the tumor and organs dissected from mice bearing

PC3 and LNCaP prostate tumor xenografts, 3 hours after intravenous injection of

ZD2-Cy5.5. As expected, only the PC3, but not the LNCaP tumors, showed an enhanced fluorescence signal from ZD2-Cy5.5, when compared to the respective normal mouse organs (Figure 3E). Microscopically, ZD2 was found to colocalize with the EDB-FN fibril network in the PC3 tumor sections, whereas EDB-FN and

ZD2-Cy5.5 were undetectable in the LNCaP tumor sections (Figure 3F). The difference in the expression and distribution of EDB-FN in these two tumors was also verified by immunohistochemical (IHC) staining (Figure 3G), indicating that upregulated EDB-FN is a promising molecular target for differential diagnosis of aggressive prostate cancer and the ZD2 peptide is an effective targeting agent for this marker.

125

Figure 3. Upregulated EDB-FN expression is a promising biomarker for high- risk prostate cancer. A, comparison of the relative mRNA expression of prostate cancer biomarkers, including EDB-FN (EDB), Vimentin, E-Cadherin (E-Cad), N-

Cadherin (N-Cad), epidermal growth factor receptor (EGFR), and prostate-specific membrane antigen (PSMA) between PC3 and LNCaP cells (n = 3, unpaired two- tailed t-test, *: P<0.05, **: P<0.01, ***: P<0.001). All gene expression was normalized to β-actin mRNA levels. B, western blot analysis of EDB-FN expression in PC3 and LNCaP tumors. Actin expression was used as a loading

126

control. C, representative phase contrast images of LNCaP and PC3 cells grown in

3D matrigel. Scale bar: 40 μm. D, representative fluorescence images of 3D cultures of LNCaP and PC3 cells incubated with 250 nM ZD2-Cy5.5, using confocal microscopy. Colors: green, GFP; red: ZD2-Cy5.5. Inset: enlarged image of the 3D spheres. E, ex vivo fluorescent Cy5.5 and bright field (BF) images of tumors and organs from PC3 and LNCaP mouse models, 3 h post-injection of 10 nmol ZD2-Cy5.5. Numbers denote: 1, lung; 2, tumor; 3, spleen; 4, muscle; 5, brain;

6, heart; 7, kidney; 8, liver. F, confocal fluorescence microscopy images of PC3 and LNCaP tumor sections stained with ZD2-Cy5.5 and anti-EDB-FN antibody.

Colors: blue, DAPI; red, ZD2-Cy5.5; yellow, EDB-FN. Overlay: addition of DAPI,

ZD2-Cy5.5, and EDB-FN channels. Scale bar: 20 μm. G, immunohistochemical

(IHC) staining for EDB-FN and H&E staining on PC3 and LNCaP tumor sections.

The brown color in IHC staining indicates EDB-FN distribution only in PC3 but not in the LNCaP sections. Scale bar: 20 μm.

Differential imaging based on T1-weighted MR molecular imaging

We next determined the effectiveness of contrast enhanced MRI with ZD2-

Gd(HP-DO3A) in differential imaging of prostate cancer in mice bearing subcutaneous PC3 and LNCaP prostate tumor xenografts. The T1 relaxivity of ZD2-

-1 -1 Gd(HP-DO3A) was 3.7 mM s at 7T, as calculated from the measured T1 values of a multi-compartment phantom containing different concentrations of ZD2-

127

Gd(HP-DO3A) (Fig. 4B). T1-weighted MR images of the tumor models were obtained before and at different time points after intravenous administration of 0.1 mmol/kg ZD2-Gd(HP-DO3A). Strong signal enhancement was observed in the

PC3 tumors with ZD2-Gd(HP-DO3A) and remained prominent for at least 30 min post-injection (Figure 4C and Figure 5). In contrast, the LNCaP tumors showed a significantly lower signal enhancement with ZD2-Gd(HP-DO3A) (Figure 4C and

Figure 5). The non-targeted clinical control ProHance only resulted in modest contrast enhancement in both the PC3 and LNCaP tumors (Figure 4C). As shown in Figure 4D, the contrast-to-noise ratio (CNR) in the PC3 tumors showed an over two-fold increase at 10 min after ZD2-GD(HP-DO3A) injection, and peaked at 30 min (CNR = 23.7±1.68; n=5; SEM), which is an over five-fold increase due to contrast agent accumulation in tumor and clearance in the normal tissues. The CNR in LNCaP tumors only increased about 10~60% during the 30 min after ZD2-

Gd(HP-DO3A) injection (CNR = 6~9), and was more than two times lower than that in the PC3 ones (CNR = 12~25) (P < 0.01 for 10 min and 20 min post-injection;

P < 0.001 for 30 min post-injection). ProHance resulted in lower CNR in PC3 tumors (CNR = 7~11; P < 0.05 at all time points compared to the ZD2-Gd(HP-

DO3A) group), which is not significantly different from that in LNCaP tumors

(CNR= 7~10) (P > 0.05) (Figure 4D). ZD2-Gd(HP-DO3A) did not produce significant contrast enhancement in other organs and normal tissues similar as the clinical control ProHance (Figure 5). Further, co-injection of excess free ZD2

128

peptide (0.5 mmol/kg) significantly reduced signal enhancement and CNR in the

PC3 tumors with ZD2-Gd(HP-DO3A) (0.1 mmol/kg), due to competitive binding of the free peptide to EDB-FN in the PC3 tumors (Figure 4D). These results suggest that molecular MRI of EDB-FN with ZD2-Gd(HP-DO3A) can effectively and specifically differentiate high-risk prostate tumors from low-risk ones.

Figure 4. The EDB-FN targeting contrast agent, ZD2-Gd(HP-DO3A), is capable of differentiating between PC3 and LNCaP tumors in T1-weighted

MRI. A, chemical structure of ZD2-Gd(HP-DO3A). B, plot of 1/T1 value versus

129

concentration of ZD2-Gd(HP-DO3A) for calculating r1 relaxivity in PBS at 7T.

Inset: T1 color-coded maps of phantoms containing ZD2-Gd(HP-DO3A) solution at different concentrations. Numbers close to each phantom denote: 1, 50 μM; 2,

25 μM; 3, 12.5 μM; 4, 6.25 μM; 5, 3.12 μM. C, Axial MRI images of PC3 and

LNCaP tumor models at the indicated tumor positions acquired with a T1-weighted sequence. Images of mice at pre-contrast (abbreviated as pre) and at 10 min, 20 min, and 30 min post-injection are shown. The names of the contrast agents are abbreviated as: ZD2, ZD2-Gd(HP-DO3A); Pro., ProHance. Competitive: injection of 0.1 mmol/kg ZD2-Gd(HP-DO3A) mixed with 0.5 mmol/kg ZD2 peptide in mice with PC3 tumors. D, change in contrast-to-noise ratio of tumors in the experiments shown in C. Data represent the mean±s.e.m. of 5 mice in all experimental groups except for the competitive group (n = 3) (unpaired two-tailed t-test: P<0.05 for comparison of ZD2-PC3 vs. all other groups at 10 min, 20 min and 30 min).

Legends: ZD2-PC3, PC3 tumor model injected with ZD2-Gd(HP-DO3A); ZD2-

LNCaP, LNCaP tumor model injected with ZD2-Gd(HP-DO3A); Pro.-PC3, PC3 tumor model injected with ProHance; Pro.-LNCaP, LNCaP tumor model injected with ProHance.

130

Figure 5. Whole body signal changes in T1-weighted MRI with the contrast agents in tumor bearing mice. Change in normalized signal intensity (SI) in different tissues after contrast agent injection, as measured from T1-weighted three- dimensional MRI images. Pre-injection signals were used to normalize the post- injection signals. (*, P < 0.05 for comparison of ZD2-PC3 vs. all other groups).

Legends: ZD2-PC3, PC3 tumor model injected with ZD2-Gd(HP-DO3A); ZD2-

LNCaP, LNCaP tumor model injected with ZD2-Gd(HP-DO3A); Pro.-PC3, PC3 tumor model injected with ProHance; Pro.-LNCaP, LNCaP tumor model injected with ProHance.

Previous reports have shown signal enhancement of pentapeptide CREKA

(Cys-Arg-Glu-Lys-Ala) targeted imaging agents specific to fibrin-FN clots in cancer imaging [15-18]. We synthesized CREKA-Gd(HP-DO3A) by conjugating

131

CREKA to Gd(HP-DO3A) to assess its potential in differential MRI of the prostate tumors and to validate the effectiveness of ZD2-Gd(HP-DO3A). Although

CREKA-Gd(HP-DO3A) generated significant contrast enhancement in the periphery of both the tumors, no significant difference was observed in signal enhancement between the PC3 and LNCaP tumors (Figure 6).

Figure 6. Contrast enhanced MRI with CREKA-Gd(HP-DO3A) in the PC3 and LNCaP tumors. A, axial images of PC3 and LNCaP tumor locations. B,

Quantification of the change in contrast-to-noise ratio in the PC3 and LNCaP tumors up to 30 min after CREKA-Gd(HP-DO3A) injection. No significant difference is seen between the two groups.

T1-mapping analysis

We also acquired quantitative T1 maps of the tumors before and at 30 min after the injection of ZD2-Gd(HP-DO3A) and ProHance. ZD2-Gd(HP-DO3A)

132

resulted in the T1 decrease from 2.75±0.05 s to 1.81±0.07 s and from 3.21±0.12 s to 2.54±0.13 s for PC3 and LNCaP tumors, respectively (Figure 7A). The maps of the changes in relaxation rate (ΔR1) before and after the contrast agent injection revealed the highest relaxation rate change in the PC3 tumors among all cases, indicating the highest accumulation of the targeted contrast agent only in the PC tumors (Figure 7A and 7B). Based on the relaxivity of the contrast agents measured after incubation with tumor homogenates (Fig. 1F and 1G), the average calculated concentration was 46.7±6.2 and 20.2±2.2 µM for ZD2-Gd(HP-DO3A), and 28.5±4.9 and 32.5±6.4 µM for ProHance in the PC3 and LNCaP tumors, respectively. At the end of the T1 map acquisitions, the mice were sacrificed and the Gd concentration in their tumors was measured by inductively coupled plasma optical emission spectrometry (ICP-OES). The measured Gd concentration was slightly lower than (but still comparable to) the values obtained by the T1 mapping, possibly because of the concentration changes occurring during the time lag between T1 mapping and animal sacrificing. Nevertheless, ZD2-Gd(HP-DO3A) had significantly higher concentration in the PC3 tumors than in the LNCaP tumors and that of ProHance in both the tumors (Fig. 7C), further validating the targeting specificity of ZD2-Gd(HP-DO3A) to the metastatic PC3 tumors.

133

Figure 7. T1 maps and accumulation of ZD2-Gd(HP-DO3A) validate its specific binding in PC3 tumors. A, T1 and ΔR1 maps showing the tumor T1 and

ΔR1 values in PC3 or LNCaP tumor models injected with ZD2-Gd(HP-DO3A)

(abbreviated as ZD2) or ProHance (abbreviated as Pro.) at pre-contrast (pre) or 30 min post-injection. Images are displayed as overlay of tumor color-coded maps with axial T1-weighted images. B, quantification of average ΔR1 after contrast injection in the groups shown in A (n = 4; *, P < 0.05, **, P < 0.01). C, comparison of contrast agent accumulation in the tumors, as measured by ICP-OES at 30 min after contrast agent injection (unpaired two-tailed t-test: n = 4; *, P < 0.05; NS: not significant).

134

In vivo chelation stability and biodistribution of the targeted contrast agent

Safety is the primary concern for clinical application of gadolinium-based

MRI contrast agents. The release of free Gd(III) ions from contrast agents and long- term tissue retention may cause toxic side-effects [19, 20]. To address this concern, we determined the in vivo chelation stability and clearance of ZD2-Gd(HP-DO3A), which are important safety parameters. Similar to ProHance, ZD2-Gd(HP-DO3A) underwent a rapid clearance through renal filtration with over 95% of the injected dose excreted in urine within 24 hours post-injection (Figure 8A). In vivo chelation stability was determined by measuring the urine concentration of Ca(II), Cu(II), and Zn(II), in comparison with ProHance and OmniScan. ZD2-Gd(HP-DO3A) exhibited the same chelation stability as ProHance, without any significant increase in the urine concentration of these ions post-injection, when compared to pre- injection ion concentrations (P > 0.05). The linear contrast agent Omniscan® resulted in significant increase in Zn(II) concentration (P < 0.05). ZD2-Gd(HP-

DO3A) exhibited the same low tissue retention as ProHance (Figure 8B). The

Gd(III) concentration was below the detection limit of ICP-OES in most of the organs, including the brain, heart, and skin. These results suggest that ZD2-Gd(HP-

DO3A) has the same in vivo chelation stability and minimal tissue retention as well as the same safety profile compared to ProHance, one of the most stable MRI contrast agents.

135

Figure 8. In vivo transmetallation and biodistribution of MRI contrast agents after intravenous administration. A, Gd content in urine before and at 8 h and 24 h after ZD2-Gd(HP-DO3A), ProHance, MultiHance, or OmniScan injections. B,

136

Gd biodistribution in mouse tissues at 1 week after injection of ZD2-Gd(HP-DO3A)

(n = 4) or ProHance (n = 3). Gd content is represented as the ratio of dose injected to the weight of the tissue. Inset: biodistribution data shown with a shorter-scale on the Y-axis. No significant difference is seen between the two groups.

3.4 Discussion

Molecular imaging of the protein markers associated with tumor aggressiveness has the potential to provide accurate and non-invasive detection and characterization of high-risk tumors. Oncofetal FN subtypes, including EDB-FN, are the well-documented markers for EMT, which involves in cancer invasion, metastasis, and drug resistance. Their expression levels are inversely correlated with the survival of cancer patients. These properties of oncofetal FN imply its potential for cancer differential diagnosis and characterization with imaging [21,

22]. Consistent to the reports in the literature, EDB-FN was highly expressed by the aggressive and androgen insensitive PC3 prostate cancer cells, not by slow- growing and androgen sensitive LNCaP cells. The expression of EGFR and PSMA, the protein markers commonly investigated for prostate cancer imaging, did not show obvious correlation with aggressiveness of the cell lines. EDB-FN seems to be a relevant and promising protein target for characterization of the aggressiveness of prostate cancer.

137

High-resolution MR molecular imaging has the ability to detect and characterize aggressive tumors in the entire prostate at an early stage. However, clinical application of MR molecular imaging is limited by its low sensitivity. We have shown that this limitation can be overcome by targeting the cancer-related protein markers abundant in the ECM of aggressive tumors [16]. Previously, we showed that molecular MRI of fibrin-FN clots was able to detect tumors, including micrometastasis [4, 16, 17]. However, the clots were mainly formed in the angiogenic areas of the tumor and not a specific target for differentiating two tumor models. The dynamic contrast enhanced MRI revealed that a sufficient amount of

ZD2-Gd(HP-DO3A) rapidly bound to abundant EDB-FN in aggressive PC3 tumors to produce strong signal enhancement, which is not evident in slow growing LNCaP tumors and normal tissues. The clearance of the unbound targeted agent from the non-specific tissues gradually increased the CNR only in the PC3 tumors in the first

30 minutes of contrast administration. Differential characterization of the tumor models was also confirmed by quantitative T1 mapping and quantification of the contrast accumulation in the tumors. The use of an excess of free ZD2 peptide in competitive MRI blocked the binding of the targeted agent and reduced tumor enhancement in the PC3 tumors, which validated its specificity to the protein marker. These results demonstrated the effectiveness and specificity of MR molecular imaging with ZD2-Gd(HP-DO3A) in detection and characterization of high-risk prostate cancers of high expression of EDB-FN.

138

In this study, we adopted a Look-Locker sequence for fast T1 mapping [9], and further accelerated the imaging with spiral trajectory. This method shortened the acquisition time to about 3 min 44 s using both Look-Locker sequence and M0 determination. With M0 determination done once in pre-contrast acquisition, a temporal resolution of 2 min 40 s with a voxel size of 0.23 × 0.23 × 1 mm3 was achieved for post-contrast acquisition. This method is inherently more tolerant to

B1 inhomogeneity and sensitive to contrast agent induced T1 change. The concentration of the targeted contrast agent in the different tumors determined by the T1 mapping was comparable to the Gd concentrations quantified using ICP-

OES as well as consistent to the protein expression levels revealed by the immunohistochemical staining (Figure 1). Rapid T1 mapping could be a valuable imaging tool to provide non-invasive quantitative measurement of the protein marker with the targeted contrast agent for quantitative characterization of prostate cancers.

Safety is a critical parameter for clinical application of any gadolinium based MRI contrast agent. Stability and complete excretion of the agents are essential to minimize any potential toxic side effects. The reported toxic side effects are generally associated with slow excretion and poor chelation stability against transmetallation of some of the gadolinium based contrast agents, mainly DTPA based linear chelates. Macrocyclic MRI contrast agents generally have high chelation stability and good safety profile. We selected one of the most stable

139

macrocyclic clinical contrast agents, Gd(HP-DO3A), in the design of the small peptide targeted contrast agents. The targeted contrast agent possessed the same in vitro and in vivo chelation stability as the clinical agent and the same minimal tissue accumulation. The results indicate that the targeted agent ZD2-Gd(HP-DO3A) could have the same safety profile as the clinical contrast agent.

Because the EDB fragment is completely conserved in all mammalian species [14, 23], MR molecular imaging of EDB-FN with ZD2-Gd(HP-DO3A) could be readily translated into clinical application in human patients and implemented in the existing clinical protocols of contrast enhanced MRI. Strong signal enhancement was observed in high-risk tumors in first 10 minutes post- injection before strong signal enhancement was observed in the bladder due to clearance of the unbound agent. This may minimize potential interference of the bladder signal enhancement on diagnostic imaging of the prostate. Nevertheless, the bladder enhancement in MRI may not be as a significant issue as in other molecular imaging modalities such as PET and SPECT, due to its high spatial resolution. This molecular imaging technology could be applied to detect, localize, and characterize high-rick tumors in the entire prostate after the initial PSA screening. Accurate detection, localization, and stratification of high-risk tumors would facilitate the earliest possible clinical interventions for the tumors, while sparing patients with indolent tumors from radical procedures. This strategy could also be used for non-invasive active surveillance of the indolent tumors. Further

140

comprehensive preclinical and clinical assessment of the safety and effectiveness of the targeted agent and MR molecular imaging is needed to warrant the clinical application of the imaging technology.

Collectively, our data show the effectiveness of MR molecular imaging of

EDB-FN with ZD2-Gd(HP-DO3A) in detection and characterization of aggressive prostate cancer in mouse models. The targeted contrast agent exhibits the same high in vitro and in vivo chelation stability and minimal body accumulation as a clinical contrast agent, which bode well a good safety profile for clinical translation. MR molecular imaging with the targeted contrast agent has the promise to improve the accuracy of the detection and diagnosis of high-risk prostate cancer and to minimize overdiagnosis and overtreatment in clinical management of the disease. In addition,

EDB-FN is also highly expressed in other types of aggressive cancers, potentially opening up avenues for the use of the imaging technology in early detection and differential diagnosis as well as image-guided therapy of a broad spectrum of cancers.

141

References

1. Wu, X., et al., MR molecular imaging of prostate cancer with a small molecular CLT1 peptide targeted contrast agent. J Vis Exp, 2013(79).

2. Wu, X.M., et al., Synthesis and Evaluation of a Peptide Targeted Small Molecular Gd-DOTA Monoamide Conjugate for MR Molecular Imaging of Prostate Cancer. Bioconjugate Chemistry, 2012. 23(8): p. 1548-1556.

3. Wu, X., et al., Peptide targeted high-resolution molecular imaging of prostate cancer with MRI. American journal of nuclear medicine and molecular imaging, 2014. 4(6): p. 525.

4. Tan, M., et al., MR molecular imaging of prostate cancer with a peptide- targeted contrast agent in a mouse orthotopic prostate cancer model. Pharm Res, 2012. 29(4): p. 953-60.

5. Lee, G.Y., et al., Three-dimensional culture models of normal and malignant breast epithelial cells. Nat Methods, 2007. 4(4): p. 359-65.

6. Mastarone, D.J., et al., A modular system for the synthesis of multiplexed magnetic resonance probes. J Am Chem Soc, 2011. 133(14): p. 5329-37.

7. Han, Z., et al., EDB Fibronectin Specific Peptide for Prostate Cancer Targeting. Bioconjug Chem, 2015. 26(5): p. 830-8.

8. Li, W., M. Griswold, and X. Yu, Rapid T1 mapping of mouse myocardium with saturation recovery Look-Locker method. Magn Reson Med, 2010. 64(5): p. 1296-303.

9. Jiang, K., et al., Rapid multislice T1 mapping of mouse myocardium: Application to quantification of manganese uptake in alpha-Dystrobrevin knockout mice. Magn Reson Med, 2015. 74(5): p. 1370-9.

142

10. Ahn, C.B., J.H. Kim, and Z.H. Cho, High-speed spiral-scan echo planar NMR imaging-I. IEEE Trans Med Imaging, 1986. 5(1): p. 2-7.

11. Freire-de-Lima, L., et al., Involvement of O-glycosylation defining oncofetal fibronectin in epithelial-mesenchymal transition process. Proc Natl Acad Sci U S A, 2011. 108(43): p. 17690-5.

12. Polyak, K. and R.A. Weinberg, Transitions between epithelial and mesenchymal states: acquisition of malignant and stem cell traits. Nat Rev Cancer, 2009. 9(4): p. 265-73.

13. Albrecht, M., et al., Fibronectin in human prostatic cells in vivo and in vitro: expression, distribution, and pathological significance. Histochem Cell Biol, 1999. 112(1): p. 51-61.

14. Locher, R., et al., Abundant in vitro expression of the oncofetal ED-B- containing fibronectin translates into selective pharmacodelivery of (131)I-L19SIP in a prostate cancer patient. J Cancer Res Clin Oncol, 2014. 140(1): p. 35-43.

15. Agemy, L., et al., Nanoparticle-induced vascular blockade in human prostate cancer. Blood, 2010. 116(15): p. 2847-56.

16. Zhou, Z., et al., MRI detection of breast cancer micrometastases with a fibronectin-targeting contrast agent. Nat Commun, 2015. 6: p. 7984.

17. Wu, X., et al., Peptide targeted high-resolution molecular imaging of prostate cancer with MRI. Am J Nucl Med Mol Imaging, 2014. 4(6): p. 525-36.

18. Zhou, Z., et al., Peptide targeted tripod macrocyclic Gd(III) chelates for cancer molecular MRI. Biomaterials, 2013. 34(31): p. 7683-93.

143

19. Zhou, Z. and Z.R. Lu, Gadolinium-based contrast agents for magnetic resonance cancer imaging. Wiley Interdiscip Rev Nanomed Nanobiotechnol, 2013. 5(1): p. 1-18.

20. Runge, V.M., Safety of the Gadolinium-Based Contrast Agents for Magnetic Resonance Imaging, Focusing in Part on Their Accumulation in the Brain and Especially the Dentate Nucleus. Invest Radiol, 2016. 51(5): p. 273-9.

21. Castellani, P., et al., Differentiation between high- and low-grade astrocytoma using a human recombinant antibody to the extra domain-B of fibronectin. American Journal Of Pathology, 2002. 161(5): p. 1695- 1700.

22. Santimaria, M., et al., Immunoscintigraphic detection of the ED-B domain of fibronectin, a marker of angiogenesis, in patients with cancer. Clin Cancer Res, 2003. 9(2): p. 571-9.

23. Kaspar, M., L. Zardi, and D. Neri, Fibronectin as target for tumor therapy. Int J Cancer, 2006. 118(6): p. 1331-9.

144

Chapter 4 MRI differential diagnosis of breast cancer using a high-relaxivity

Gadolinium metallofullerene targeted to extradomain-B fibronectin

This chapter is adapted based on a paper under submission:

Han Z, Wu X, Roelle S, Chen C, Schiemann W and Lu ZR. Targeted gadofullerene with superior relaxivity for sensitive molecular MRI of breast cancer. Nature communications. (2017).

4.1 Abstract

Magnetic resonance imaging (MRI) has yet to be used in breast cancer diagnosis due to its inherent low sensitivity and inability to differentiate high-risk breast cancer from low-risk diseases. In this regard, development of high-relaxivity contrast agent that generates prominent contrast in aggressive tumours in an urgent need. Here we report functionalization of gadolinium (Gd) containing metallofullerenes with an extradomain-B fibronectin (EDB-FN) targeting peptide,

ZD2, to yield a high-relaxivity tumour-targeting contrast agent. Due to the abundant expression of EDB-FN in high-risk breast cancer, this contrast agent was shown to generate significant contrast in a high-risk triple negative tumor model, MDA-MB-

231, but not in low-risk tumour model, MCF-7, at a low dose of 0.005 mmol/kg. In addition, this contrast agent demonstrated stable in vivo and in vitro stability, and fast clearance. Our data suggests that MRI with this high-relaxivity contrast agent

145

was able to lower the dose for breast cancer MRI, meanwhile enabling risk- stratification of breast cancers.

4.2 Introduction

Breast cancer constitutes approximately 30% of newly diagnosed cancers in American women. However, roughly 15-30% of tumours detected by screening are unlikely to cause trouble if left untreated. Thus, identifying the tumours that actually pose a threat is urgently needed in order to focus therapy where it is needed.

This would exempt millions of women from potentially harmful interventions.

Being a non-invasive method, breast imaging poses itself as a tempting strategy to determine the breast cancer risk. Current modalities of breast imaging include mammogram, ultrasound, and magnetic resonance imaging (MRI), among which

MRI stands as the modality that generates no ionizing radiation, and provide high soft-tissue resolution. However, what hinder the broader application of MRI are its inherent low sensitivity and lack of specificity for high-risk breast cancer.

Our strategy for breast cancer differential diagnosis involves design of an extradomain-B fibronectin (EDB-FN) targeting high-relaxivity MRI contrast agent.

EDB-FN is a biomarker highly expressed in malignant breast cancer, with relatively lower expression in benign cancer. Contrast agents that target EDB-FN tend to accumulate in higher concentration at sites of malignant cancer, thus producing more contrast enhancement. Previously, an EDB-FN targeting peptide, ZD2, was discovered in our lab. Built on ZD2, a Gadolinium (Gd) containing probe can be

146

developed. Gadofullerenes are interesting candidates of high-relaxivity contrast agents. The high-relaxivity of this type of contrast agents stem from the fast proton exchange, and increased correlation time. However, unmodified Gadofullerenes have poor water-solubility, because of which a number of approaches have been devised to improve their solubility. This is also an essential step towards endowing these particles for targeting purposes. Previously, an oxidization approach based on acyl peroxide has been developed, which yields hydroxylated and caboxylated

Gadofullerenes. This potentiated a facile method to functionalize Gadofullerenes with targeting moieties based on the carboxyl acids.

Herein, we report a facile method to functionalize Gd3N@C80 to display maleimido groups, through which thiol-containing moieties, for example, peptides with Cysteines could be easily conjugated. In our attempt to functionalize

Gadofullerenes to targeting EDB-FN that are abundantly expressed in high-risk tumor, we conjugated ZD2 peptides on Gadofullerenes. This resulted in a contrast agent with a high r1 relaxivity of 71.6 mM-1s-1. This contrast agent was used for

MRI of two tumor models, MDA-MB-231 and MCF-7, which represented high- risk and low-risk breast cancers, respectively. A substantially lower dose of 0.005 mmol Gd/kg, as opposed to the clinical dose of 0.1 mmol/kg, was used for imaging.

In addition, we characterized the stability and biodistribution of this contrast to demonstrate its safety for clinical use. This novel high-relaxivity contrast agent is

147

of great diagnostic and prognostic potential to improve the clinical management of breast cancer.

4.3 Material and methods

Materials

Succinic anhydride and o-dichlorobenzene were purchased from Sigma

Aldrich (Saint Louis, MO, USA). Mono-Fmoc ethylene diamine was purchased from Combi-Blocks (San Diego, CA, USA). N,N,N′,N′-Tetramethyl-O-(1H- benzotriazol-1-yl)uronium hexafluorophosphate, O-(Benzotriazol-1-yl)-

N,N,N′,N′-tetramethyluronium hexafluorophosphate (HBTU) and all amino acids for peptide synthesis were purchased from Anaspec (Fremont, CA, USA).

N,N-Diisopropylethylamine (DIPEA) was acquired from MP Biomedicals, LLS

(Solon, OH, USA. Dimethylformamide (DMF), dichloromethane (DCM) and piperidine were purchased from Fisher Scientific (Pittsburgh, PA, USA).

Gd3N@C80 were purchased from SES Research (Houston, TX, USA). Maleimide- opfp (pentafluorophenyl 3-(2,5-dioxo-2,5-dihydro-1H-pyrrol-1-yl)propanoate) was synthesized according to a previous protocol[1].

Methods

Synthesis of succinic acid acyl peroxide (1). At 0 oC, succinic anhydride

(1.5 g, 15 mmol) was added dropwise to hydrogen peroxide and the mixture was stirred for 30 min. The product was washed with pure water and filtered, followed by lyophilization for further use.

148

Oxidation of Gd3N@C80. Succinic acid acyl peroxide (8.0 mg, 5 equiv) was added to Gd3N@C80 (10 mg) in 10 mL o-dichlorobenzene. The resultant solution was deaerated by flashing with nitrogen and heated at 84 ºC for 48 h.

Additional succinic acid acyl peroxide (8.0 mg each time) was added at an interval of 12 h during the reaction. A brown sludge precipitated from the solution at the end of the reaction. Then, 20 mL 0.2 M NaOH aqueous solution was added to extract the water-soluble product. The top aqueous layer was deep brown and the bottom layer was colorless. The top layer was concentrated, purified with a PD10 column, and lyophilized to yield Gd3N@C80(OH)18(COOH)4 (2) (yield, 74%).

MALDI-TOF (m/Z, [M+1]+): 2221(obsd); 2220 (calc).

Synthesis of Gd3N@C80(OH)18(MAL)4. Gd3N@C80(OH)18(COOH)4 (6 mg) and mono-Fmoc ethylene diamine (5 equiv) were dissolved in DMF, and

HBTU (5 equiv) and DIPEA (5 equiv) then were added. The reaction was stirred at room temperature for 2 hours. Then piperidine/DMF (20%, v/v) was used to remove the protecting group of Fmoc. Afterwards, the product was precipitated in cold ether to obtain Gd3N@C80(OH)18(NH2)4 (3) of brown color. (yield, 63%).

Gd3N@C80(OH)18(NH2)4 (3 mg) was dissolved in DMF (5 mL) and maleimido- opfp (20 mg) was added. The reaction continued for 30 min before precipitating in cold-ether to give Gd3N@C80(OH)18(MAL)4 (4) (yield, 95%).

Synthesis of ZD2-Gd3N@C80. ZD2-Cys peptide (sequence: Thr-Val-Arg-

Thr-Ser-Ala-Asp-Cys) was synthesized in solid phase using standard Fmoc-

149

chemistry. ZD2-Cys and Gd3N@C80(OH)18(MAL)4 (ZD2/MAL = 2, molar ratio) were dissolved in pure water and stirred for 30 min. Then the solution was concentrated and purified with PD10 column. The final product ZD2-Gd3N@C80 was collected and lyophilized (yield, 82%). MALDI-TOF ([M+1]+: m/z = 3927

(obsd); 3912 (calc. for Gd3N@C80(OH)18(MAL)4-Cys-ZD2).

Synthesis of ZD2-Cy5.5. ZD2 peptide (sequence: Thr-Val-Arg-Thr-Ser-

Ala-Asp) were synthesized in solid phase using standard Fmoc-chemistry. After

Fmoc removal with 10% piperidine, Fmoc-NH-(PEG)2-COOH (Anaspec. Inc,

Fremont, CA, USA) was added to the peptide sequence. After Fmoc removal with

10% piperidine, the resin was washed with DMF/DCM and air-dried, and 10 mg of the dried resin then was swelled in DCM for 1 h, followed by reacting with 3 mg of Cy5.5-NHS ester (Lumiprobe Corporation, Hallandale Beach, FL, USA) in presence of 5 μL DIPEA. Reaction was stirred overnight at room temperature.

Excess Cy5.5-NHS ester was removed by filtration and washing with DMF/DCM

10 min for 3 times. Peptides were cleaved off resin using TIPS, and precipitated in cold ether. The products were separated from ether by centrifugation at 4000 × g.

The final product was characterized by MALDI-TOF mass spectrometry ([M+1]+: m/z = 1473.76 (obsd); 1472.78 (calc.)). The product was lyophilized and reconstituted in 500 μL PBS. The concentration of the solution was characterized by measuring absorbance at 650 nm.

150

Relaxivity measurement. Lyophilized ZD2-Gd3N@C80 was reconstituted to a series dilution of 0.0625 to 0.5 μM in water. The solutions were pipetted in

NMR tubes (500 μL in each tube), and placed in a relaxometor (Bruker) at 1.5T. T1 and T2 values of each solution were measured. The r1 and r2 relaxivities of the contrast agent was calculated as the slop of the plot of 1/T1 and 1/T2 relaxation rates against the concentrations.

Cell culture. MDA-MB-231 and MCF-7 cells were acquired from

American Type Culture Collection (ATCC, Rockville, MD, USA). Both cell lines were maintained in Dulbecco Modified Eagle Medium (DMEM) supplemented with 10% fetal bovine serum (FBS) and 5% penicillin/streptomycin (pen/strep).

Three-dimensional culture of cells was achieved by an “on-top” matrigel method reported previously[2]. Briefly, on a glass-bottom plate a thick layer of matrigel was coated, followed by plating MDA-MB-231 or MCF-7 on top of the matrigel.

Once 3D sphere or clusters formed, ZD2-Cy5.5 was added to the medium to the final concentration of 250 nM. Binding of ZD2-Cy5.5 to the 3D spheres was evaluated using a confocal laser scanning microscope (Olympus Corporation,

Tokyo, Japan) after culturing for 1 h.

Animal tumor models. Female Balb/c athymic mice were purchased from the Case Comprehensive Cancer Center and housed in the Case Center Imaging

Research. All experiments regarding animals were carried out according a protocol approved by the IACUC of Case Western Reserve University. To initiate MDA-

151

MB-231 and MCF-7 tumor xenografts, cells that were cultured in 150 mm dishes were trypsinized and centrifuged. Cell pellets from each dish were suspected in 100

μL PBS and reconstituted with Corning Matrigel Matrix High Concentration

(Corning, Corning, NY, USA) into the concentration of 2 ×107 cells/mL on ice. The suspended cells in 100 μL matrigel solution were injected subcutaneously on the flank of mice (4-6 weeks) using a 19-gauge needle. After injection, a plug in flank was formed due to matrigel gelling. Tumors were allowed to grow for at least a month before tumor size reached 7-10 mm in diameter.

Fluorescence imaging. To determine the distribution of ZD2-Cy5.5 in the major organs and tumors, 10 nmol ZD2-Cy5.5 was injected in tumor-bearing mice through a tail vein. At 3 hours after injection, mice were sacrificed. Tumors and organs were collected and imaged with CSi Maestro imaging system (Woburn, MA,

USA) using the deep red filter sets.

MR imaging. MR images of tumor bearing mice were acquired on an

Aspect M3 small animal MRI scanner (1 Tesla). Mice were placed on a holder with the temperature maintained at 37 °C, with isoflurane/oxygen mixture supplied to the mice through a nose cone. A thin catheter filled with PBS was connected to a tail vein of the mice. After mice were placed in the coil, a pilot scan was performed to adjust mice to the proper location in the coil. An axial T1 weighted sequence (TR

= 500 ms; TE = 9 ms; Flip angle = 90°; Field of view = 3 cm × 3 cm; Matrix size =

128 × 128 × 8; slice thickness = 2 mm; inter-slice distance = 1 mm) was then used

152

to acquire the images of the tumors before and 10 min, 20 min and 30 min after contrast injection. Images were exported into DICOM data, which were then processed and analyzed using Matlab (Natick, MA, USA). The contrast-to-noise ratio of tumors in the images were calculated as the difference between tumor mean intensity minus muscle mean intensity, divided by the noise. Three-dimensional images of mice were acquired using a gradient echo T1-weighted sequence with the following parameters: TR = 17 ms; TE = 6 ms; Flip angle = 15°; Field of view =

3.5 cm × 8 cm; Matrix size = 128 × 512 × 16; slice thickness = 1.5 mm. Analysis of change in signal intensity in muscle, heart, liver, kidney and bladder was performed using Matlab.

qRT-PCR. RNA was harvested from MDA-MB-231 and MCF-7 cells using the RNeasy Plus Mini Kit (Cat. # 74134, Qiagen, Hamburg, Germany) and reverse transcription was carried out with the miScript II RT Kit (Cat. # 218161,

Qiagen, Hamburg, Germany) according to manufacturer’s protocols. Quantitative real-time polymerase chain reaction (qRT-PCR) was performed using SYBR Green

PCR Master Mix (Cat. # 4309155, Applied BioSystems, Foster City, CA, USA) and the Eppendorf RealPlex Thermocycler (Eppendorf, Hauppauge, NY, USA).

Cycle threshold (Ct) values were evaluated by the RealPlex 2.2 software system

(Eppendorf, Hauppauge, NY, USA). Expression levels of human EDB-FN were analyzed in triplicate by the ΔΔCt method and normalized to expression levels of

β-actin. Change in gene expression was determined with the 2-ΔΔCt equation and

153

significance was found when p ≤ 0.05. Primers were obtained from Invitrogen

(Carlsbad, CA, USA) and sequences are as follows: human EDB-FN forward: 5’-

CCTGGAGTACAATGTCAGTG-3’, human EDB-FN reverse: 5’-

GGTGGAGCCCAGGTGACA-3’, human β-actin forward: 5’-

GTTGTCGACGACGAGCG-3’, human β-actin reverse: 5’-

AGCACAGAGCCTCGCCTTT-3’.

Biodistribution. The mice injected with the contrast agents were sacrificed one week post-injection. Tissue samples were collected, weighed, and digested by

1.00 mL ultrapure nitric acid (EMD Millipore, Billerica, MA, USA) for 7 days. The digested sample (0.50 mL) was diluted to 5.00 mL with ultrapure water (Milli-Q,

EMD Millipore, Billerica, MA, USA). The solution was centrifuged and filtered using a 0.45 µm filter and the concentration of Gd(III) ions was determined using inductively coupled plasma-optical emission spectrometry (ICP-OES) on a 730-ES

ICP-OES system (Agilent Technologies, Santa Clara, CA, USA). Samples were measured at three different wavelengths for Gd at 336.224, 342.246, 358.496 nm and the results were averaged across wavelengths. Intensities were evaluated by

ICP Expert II v. 2.0.2 software and were related to concentration by a calibration curve. A standard calibration curve was developed from a blank and seven standards from a stock solution of 1000 ppm Gd in 3% nitric acid (Ricca Chemical

Company, Arlington, TX, USA) and diluted with 2% nitric acid.

154

Statistical analysis. All experiments were performed in triplicates unless stated otherwise. Data were represented as mean ± s.e.m. Analysis of differences between two groups was performed using Student’s t-test, and the difference were considered significant if p<0.05.

4.4 Results and Discussions

In this work, we designed a high-relaxivity targeted contrast agent by conjugating a small peptide ZD2 (Thr-Val-Arg-Thr-Ser-Ala-Asp) to hydroxylated

Gd3N@C80. ZD2 specifically binds to EDB-FN highly expressed in many types of aggressive human cancer[3]. Gd3N@C80 was first oxidized with acryl peroxide and NaOH to yield carboxyl and hydroxyl groups on the cage surface (Figure 1).

The oxidized product Gd3N@C80(OH)18(COOH)4 was characterized by MALDI-

TOF mass spectrometry (m/z = 2221 Da, Figure 2A). Carboxyl groups on

Gd3N@C80(OH)18(COOH)4 were then converted into amines, followed by reaction with maleimido-opfp, yielding maleimido-Gd3N@C80 (Figure 1) for conjugation with thiol-bearing ZD2 peptide (Cys-Thr-Val-Arg-Thr-Ser-Ala-Asp) (Figure 1).

One ZD2 peptide was conjugated to each Gd3N@C80 based on MALDI-TOF mass spectrum (m/z = 3927 Da, Figure 2B). The final product ZD2-Gd3N@C80 had good water solubility, which is essential for in vivo use as a contrast agent. ZD2-

Gd3N@C80 had a diameter of approximately 1 nm on average as determined by transmission electron microscopy (TEM) (Figure 2C) and dynamic light scattering

(DLS) (Figure 2D), which is much smaller than the renal filtration threshold.

155

The r1 and r2 relaxivity of ZD2-Gd3@NC80 was measured at 1.5 Tesla. It

-1 -1 -1 -1 had a high r1 and r2 relaxivities of 74.6 mM s and 114.9 mM s per Gd(III) ion

-1 -1 -1 -1 or 223.8 mM s and 344.7 mM s per molecule, respectively (Figure 2E). The r1 relaxivity was almost 20 times of that of clinical GBCA, Gd-DTPA and Gd(HP-

DO3A). Such a superior relaxivity is critical to increase the sensitivity contrast enhanced MRI for molecular imaging at low contrast agent doses.

Extradomain-B fibronectin (EDB-FN) is identified as a promising target for molecular MRI of aggressive tumors with ZD2-Gd3N@C80. EDB-FN is a biomarker of tumour angiogenesis[4] and epithelial-to-mesenchymal transition

(EMT)[5], abundant in the microenvironment of malignant cancers, which contribute to cancer proliferation and invasion. Clinical evidence shows that EDB-

FN overexpression is associated with the poor prognosis of a variety of cancers[6,

7]. EDB-FN expression was determined in MDA-MB-231 triple negative breast cancer (TNBC) cells and MCF-7 estrogen receptor (ER) positive breast cancer cells.

The cellular EDB-FN mRNA level and tumor EDB-FN protein level in MDA-MB-

231 model were significantly higher than that in MCF-7 model (Figure 3A & B).

In a matrigel-based three-dimensional (3D) system, MDA-MB-231 cells were able to form large spherical structures, whereas MCF-7 only formed smaller cell clusters because of their limited matrigel invasion (Figure 3C). Meanwhile, MDA-MB-231 spheres secreted a large amount of EDB-FN, which were illuminated by ZD2-Cy5.5 under confocal microscopy. Very limited ZD2-Cy5.5 binding was seen for MCF-7

156

cells (Figure 3C). To validate the targeting specificity of ZD2 peptide, ZD2-Cy5.5 was injected intravenously into the MDA-MB-231 and MCF-7 tumour-bearing mice at the dose of 0.5 µmol/kg. ZD-Cy5.5 had a substantially greater binding to

MDA-MB-231 tumours than normal tissues and organs at 3 hours after injection, while no significant binding was seen in the MCF-7 tumour (Figure 3D). EDB-FN expression and ZD2-Cy5.5 binding in two tumours were further verified by immunofluorescence staining of EDB-FN and Cy5.5 fluorescence imaging, respectively (Figure 3E). MDA-MB-231 tumour section was rich in EDB-FN expression and Cy5.5 fluorescence. The co-localization of EDB-FN expression and

ZD2-Cy5.5 signal validated the specific binding of ZD2-Cy5.5 EDB-FN highly expressed in MDA-MB-231 tumours. Microscopically, MCF-7 tumor displayed a lower cell density as compared to the more aggressive MDA-MB-231 TNBC tumors (Figure 3F).

We next tested molecular MRI with ZD2-Gd3N@C80 for detection and differentiation of the highly aggressive MDA-MB-231 TBNC tumors from MCF-

7 tumors in animal models. MR image acquisition was performed with the mice bearing MDA-MB-231 and MCF-7 tumour models at 1 Tesla before and after intravenous injection of ZD2-Gd3N@C80 at a very dose of 1.67 µmol/kg or 5

µmol-Gd/kg, a dose that is 20 times less than the dose of the clinical GBCA.

Significant tumour enhancement was observed in MDA-MB-231 tumours, while little enhancement was observed in MCF-7 tumours or both tumors injected with

157

non-targeted Gd3N@C80 (Figure 4A & B). Co-injection of 1.67 µmol/kg of ZD2-

Gd3N@C80 with 25 µmol/kg ZD2 peptide significantly reduced the signal enhancement in MDA-MB-231 tumours due to the competitive binding to the target.

Quantitative analysis revealed that ZD2-Gd3N@C80 produced 39 to 45% increase of CNR in MDA-MB-231 tumours (Figure 4B). There were no significant differences of signal enhancement in the normal tissues before and after contrast injection in all tested groups, except for the kidneys and bladder, which are organs for clearance of the contrast agents (Figure 5). Increased signal intensity in the bladder indicates the ready excretion of unbound ZD2-Gd3N@C80 via renal filtration. Both ZD2-Gd3N@C80 and Gd3N@C80 had low tissue accumulation at

1 week post-injection, suggesting complete excretion of the contrast agents after the MRI studies (Figure 4C). The results indicate that ZD2-Gd3N@C80 is a safe high-relaxivity targeted MRI contrast agent for sensitive molecular MRI of aggressive tumors.

Despite efforts to increase the water-solubility of gadofullerenes to potentiate them for in vivo use [8-10], little has been done in endowing tumour- homing abilities to these gadofullerenes. Here we report a facile method for functionalizing gadofullerenes with an EDB-FN targeting peptide, ZD2. The resultant contrast agent, Gd3N@C80-ZD2, had a high r1 relaxivity, which enabled the use of a much lower dose (0.005 mmol Gd/kg) to generate sufficient tumour contrast enhancement in MRI. In contrast to macromolecular dendrimer-based

158

contrast agents, these targeted gadofullerenes offer high relaxivities without a large molecular weight [11, 12]. The smaller size of Gd3N@C80-ZD2 also ensures fast clearance and fast tumour accumulation, which is a desirable property for increasing turnover rate in MRI and minimizing long-term Gd accumulation in tissues. Compared to mammography and ultrasound, which are mere reflections of anatomical or physical properties of tumours, Gd3N@C80-ZD2, as a molecular imaging agent, extends the ability of MRI to evaluating EDB-FN expression in breast cancer. Our data suggested that EDB-targeting gadofullerenes generated high contrast enhancement in MDA-MB-231 tumour models, a triple negative breast cancer (TNBC) that is commonly recognized to induce the poor clinical outcome of patients. We also showed that Gd3N@C80-ZD2 did not result in signal enhancement in nonmetastatic MCF-7 tumours, which represent a large portion of slow-growing breast lesions that are unlikely to progress to advanced stages. This property of Gd3N@C80-ZD2 is desirable for reducing false-positive detection of benign breast lesions, such as fibrocystic diseases, which are indistinguishable from breast cancers by current MRI technology [13]. Currently, breast MRI is currently only recommended for patients with >20% greater life-time risk out of the concern for MRI’s high false-positive rate [14, 15]. By stratifying the risk of breast cancer aggressiveness, our technology can help identify aggressive tumours to focus therapy where it is urgently needed, meanwhile exempting millions of women from

159

potential harmful interventions [16]. We think this contrast agent worthy of further testing in clinical arena to validate its value in breast cancer differential diagnosis.

Figure 1. An EDB-FN-targeting gadofullerene for molecular MRI of breast cancer. A, Synthesis scheme of ZD2-Gd3N@C80. Cyan, Gd; blue, nitrogen; red, oxygen; gray, hydrogen. B, Illustration of tumor targeting with ZD2-Gd3N@C80 for detection and characterization of breast cancer in mouse models. MCF-7 and

MDA-MB-231 cells were used to construct high-risk breast cancer and low-risk breast cancer, respectively. Intravenous injection of the EDB-FN-targeting agent,

Gd3N@C80-ZD2, results in different binding levels corresponding to the EDB-FN

160

expression and tumor aggressiveness for tumour detection and characterization with molecular MRI.

Figure 2. Characterization of ZD2-Gd3N@C80. A,B, MALDI-TOF mass spectra of Gd3N@C80(OH)18(COOH)4 (a), and ZD2-Gd3N@C80 (b). C,D, Plots of 1/T1 and 1/T2 versus contrast agent concentrations for calculation of r1 and r2 relaxivities

161

-1 -1 -1 -1 of Gd3N@C80(OH)18(COOH)4 (c) (r1 = 57.1 mM s ; r2 = 98.5 mM s ), and ZD2-

-1 -1 -1 -1 Gd3N@C80 (d) (r1 = 74.6 mM s ; r2 = 114.9 mM s ). E,F, TEM images (e) and

DLS size distribution (f) of ZD2-Gd3N@C80. Scale bar: 5 nm.

Figure 3. EDB-FN overexpression is a signature of aggressive breast cancer.

A, RT-PCR analysis of EDB-FN mRNA levels in MCF-7 and MDA-MB-231 cells, showing the higher EDB-FN expression in MDA-MB-231 cells (n=3; two-tailed t-

162

test: P < 0.05). The level of EDB-FN mRNA in MCF-7 cells were used to normalize the data. B, Western blotting analysis of EDB-FN expression in MCF-7 and MDA-

MB-231 tumours. β-actin was used as a loading control. C, Representative fluorescence images of ZD2-Cy5.5 (red) binding, bright field images of 3D culture of MCF-7 and MDA-MB-231 cells, and the overlay of fluorescence images with bright field images. D, Ex vivo fluorescence images of tumour and organs tissues collected from mice bearing MCF-7 and MDA-MB-231 tumours at 3 h after injection of 10 nmol ZD2-Cy5.5. Numbers in the images indicate the following tissues: 1, tumour; 2, muscle; 3, spleen; 4, liver; 5, brain; 6, heart; 7, kidney; 8, lung.

E, Analysis of EDB-FN expression (yellow) and ZD2-Cy5.5 (red) binding in MCF-

7 and MDA-MB-231 tumour sections. DAPI (blue) was used for staining nucleus.

Scale bar: 25 µm. F, H&E staining showing the morphology of MCF-7 and MDA-

MB-231 tumour sections. Scale bar: 50 µm. Inset: enlarged images of tumor sections (Scale bar: 10 µm).

163

Figure 4. Contrast enhanced MRI with ZD2-Gd3N@C80 of MDA-MB-231 tumours and MCF-7 tumours in mice. A, Representative axial T1-weighted 2D spin-echo MRI images of MDA-MB-231 or MCF-7 tumours in mice. Images were acquired before and at 10, 20 and 30 min after injection of ZD2-Gd3N@C80 and

Gd3N@C80(OH)18(COOH)4 at a dose of 1.67 µmol, or a mixture of 25 µmol/kg free ZD2 and 1.67 µmol ZD2-Gd3N@C80 (competitive group). Tumour locations are indicated by white arrow heads. B, Analysis of percent increase of tumor contrast-to-noise ratio (CNR) from images acquired in groups indicated in (A) (n=4 for MDA-MB-231 tumours and n=3 for MCF-7 tumours. *: P < 0.05 for comparison of the increased CNR ratio of ZD2-Gd3N@C80 in MDA-MB-231 group with that in all other groups). C, Gd biodistribution at 1 week after injection of ZD2-Gd3N@C80 or Gd3N@C80(OH)18(COOH)4 in MDA-MB-231 tumour

164

models. There was no statistical difference of the retention of the contrast agents in all tested tissues (n=3).

165

Figure 5. Dynamic change of normalized signal intensities in muscle, heart, liver, kidney, and bladder in MDA-MB-231 and MCF-7 tumor models. The mice were injected with 0.005 mmol Gd/kg Gd3N@C80-ZD2, Gd3N@C80, or the mixture of 0.025 mmol/kg free ZD2 and 0.005 mmol Gd/kg Gd3N@C80

(competitive) (n=3).

166

References

1. Zhou, Z.X., et al., Peptide targeted tripod macrocyclic Gd(III) chelates for cancer molecular MRI. Biomaterials, 2013. 34(31): p. 7683-7693.

2. Lee, G.Y., et al., Three-dimensional culture models of normal and malignant breast epithelial cells. Nature methods, 2007. 4(4): p. 359-365.

3. Han, Z., et al., EDB Fibronectin Specific Peptide for Prostate Cancer Targeting. Bioconjugate chemistry, 2015. 26(5): p. 830-838.

4. Khan, Z.A., et al., EDB fibronectin and angiogenesis–a novel mechanistic pathway. Angiogenesis, 2005. 8(3): p. 183-196.

5. Freire-de-Lima, L., et al., Involvement of O-glycosylation defining oncofetal fibronectin in epithelial-mesenchymal transition process. Proceedings of the National Academy of Sciences, 2011. 108(43): p. 17690-17695.

6. Mhawech, P., et al., EB-D fibronectin expression in squamous cell carcinoma of the head and neck. Oral Oncology, 2005. 41(1): p. 82-88.

7. Khan, Z.A., et al., ED-B fibronectin in non-small cell lung carcinoma. Experimental Lung Research, 2005. 31(7): p. 701-711.

8. Feng, Y.Q., et al., A highly soluble gadofullerene salt and its magnetic properties. Dalton Transactions, 2015. 44(17): p. 7781-7784.

9. Zhang, J., et al., Gd3N@C84(OH)x: a new egg-shaped metallofullerene magnetic resonance imaging contrast agent. J Am Chem Soc, 2014. 136(6): p. 2630-6.

167

10. Shu, C.Y., et al., Aggregation studies of the water-soluble gadofullerene magnetic resonance imaging contrast agent: [Gd@C82O6(OH)(16)(NHCH2CH2COOH)(8)](x). Journal Of Physical Chemistry B, 2006. 110(31): p. 15597-15601.

11. Raymond, K.N., Next generation, high relaxivity gadolinium MRI agents. Abstracts Of Papers Of the American Chemical Society, 2004. 227: p. U145-U146.

12. Longmire, M.R., et al., Dendrimers as high relaxivity MR contrast agents. Wiley Interdisciplinary Reviews-Nanomedicine And Nanobiotechnology, 2014. 6(2): p. 155-162.

13. Baltzer, P.A., et al., False-positive findings at contrast-enhanced breast MRI: a BI-RADS descriptor study. AJR Am J Roentgenol, 2010. 194(6): p. 1658-63.

14. Saslow, D., et al., American Cancer Society guidelines for breast screening with MRI as an adjunct to mammography. CA: a cancer journal for clinicians, 2007. 57(2): p. 75-89.

15. Henteleff, H.J.A., et al., Comparative Effectiveness of MRI in Breast Cancer (COMICE) trial: a randomized controlled trial. Journal Of the American College Of Surgeons, 2012. 215(6): p. 894-896.

16. Woolston, C., Breast cancer: 4 big questions. Nature, 2015. 527(7578): p. S120-S120.

168

Chapter 5 MR molecular imaging of triple-negative breast cancer and micrometastases

This chapter is adapted based on the paper in submission:

Han Z, Zhou Z and Lu ZR. MR molecular imaging of triple-negative breast cancer and metastases by targeting an oncoprotein in the tumor microenvironment

5.1 Abstract

In this chapter, non-invasive MRI detection of TNBC primary and metastatic tumors with ZD2-Gd(HP-DO3A) is demonstrated. We constructed primary and metastatic triple negative breast cancer (TNBC) mouse models using

4T1 and MDA-MB-231 cells, which are murine and human TNBC cells, respectively. We show that EDB-FN is abundantly expressed in the ECM of primary TNBC tumors; in metastatic tumors, EDB-FN expression is even higher, which allows for accumulation of imaging probes at a higher locoregional concentration. In T1-weighted MRI, ZD2-Gd(HP-DO3A) generated superior contrast enhancement in primary TNBC tumors than the non-targeting clinical agent ProHance, during 30 min after contrast injection. Moreover, ZD2-Gd(HP-

DO3A) also induced a marked increase in contrast-to-noise ratio (CNR) of TNBC metastases, enabling sensitive localization and delineation of metastases that appeared occult in non-contrast-enhanced or ProHance-enhanced MRI. These findings potentiate the use of ZD2-Gd(HP-DO3A) for MR molecular imaging of

169

TNBC primary and metastatic tumors to improve the clinical management of these high-risk breast cancers.

5.2 Introduction

Triple-negative breast cancers (TNBCs) are a subgroup of breast cancers that lack estrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER2) expression. TNBCs represent 15-20% of breast cancer cases worldwide and generally appear less differentiated with a high metastatic potential [1]. TNBCs are characterized by resistance to chemotherapy and resultant metastasis is primarily responsible for cancer mortality [2-4]. The early and sensitive detection of locoregional and distant metastasis of TNBCs is critical to clinical management of breast cancer and improving clinical outcomes.

Current clinical imaging modalities, including mammography and ultrasonography, lack sensitivity and specificity in detecting TNBCs and their micrometastases [5].

Magnetic resonance imaging (MRI) provides superior soft tissue contrast and have demonstrated much higher sensitivity in detecting breast tumors that are occult in mammography and ultrasound [6]. Small molecular Gd(III) chelates are routinely used as contrast agents in MRI. However, currently available contrast agents have no specificity for TNBCs and fail to provide sufficient contrast in TNBCs and metastases. Novel contrast agents capable of differentially enhance TNBCs and their micrometastases in MRI are urgently needed.

170

Development of targeted contrast agents for TNBCs is challenging as

TNBCs are characterized by loss of cell markers commonly used for cancer targeting [7, 8]. TNBCs are also heterogeneous in nature, making it difficult to find a common cell marker to track all cancers [9]. Therefore, instead of targeting cell biomarkers, our strategy is to target the tumor microenvironment, on which cancer substantially relies for growth and dissemination. A receptive microenvironment in distant organs is also crucial for the outgrowth of engrafted cancer cells. MRI with an imaging probe specific to unique signatures in this tumor microenvironment has great potential in improving efficiency in detection of TNBCs and their metastases at an early stage. Fibronectin is one of the most abundant extracellular matrix (ECM) proteins. Fibronectin serves as a central organizer of a variety of ECM components and modulate ECM remodeling during cancer progression [10]. Previously, we have shown that a fibronectin-targeting contrast agent, CREKA-Tris(Gd-DOTA)3, was able to induce significant contrast enhancement in breast micrometastases less than 0.5 mm in diameter [11]. Here we used a contrast agent specific to an isoform of fibronectin, extradomain-B fibronectin (EDB-FN), for enhanced detection sensitivity and specificity. EDB-FN is exclusively expressed in angiogenesis, wound healing, and cancer. EDB-FN overexpression is also a result of epithelial- to-mesenchymal transition (EMT) [12], which is associated with generation of cancer with high metastatic potential. Previously, we have shown that the EDB-

FN-targeting small molecular contrast agent, ZD2-Gd(HP-DO3A), was able to

171

differentiate highly metastatic PC3 tumors and slow-growing LNCaP tumors non- invasively in MRI [13]. In this study, the murine TNBC cell line, 4T1 [14, 15], and human TNBC cell line, MDA-MB-231 were used to construct primary and metastatic TNBC mouse models. We demonstrated the potential of ZD2-Gd(HP-

DO3A) in delineating TNBC primary tumors and distant metastases. Clinical application of this contrast agent could facilitate accurate staging, treatment planning, and prognosis of breast cancer patients.

172

5.3 Materials and methods

Materials and cell culture

All reagents used were of the highest grade commercially available.

Reagents for chemical synthesis were purchased from Sigma-Aldrich (Saint Louis,

MO, USA) unless otherwise stated. The fluorescence imaging probe, ZD2-Cy5.5, and MRI contrast agent, ZD2-Gd(HP-DO3A), were synthesized and characterized as previously described [13]. The firefly luciferase-expressing cell line, 4T1-GFP-

Luc, breast cancer cells was acquired from PerkinElmer (Waltham, MA, USA) and cultured in Dulbecco’s Modified Eagle Medium (DMEM) supplemented with 10%

FBS and 1% Penicillin-Streptomycin (5,000 U/mL, Thermofisher Scientific,

Waltham, MA, USA). MDA-MB-231 cells were acquired from American Type

Culture Collection (ATCC, Rockville, MD, USA). To stably express firefly luciferase, MDA-MB-231 cells were transfected with pNifty-CMV-luciferase, followed by selection with Zeocin (500 μg/mL). 4T1-GFP-Luc and MDA-MB-231-

Luc cells were cultured in RPMI-1640 medium supplemented with 10% FBS and

1% Penicillin-Streptomycin (5,000 U/mL, Thermo fisher Scientific). Cells were cultured in an incubator maintained at 37 °C and 5% CO2.

Semiquantitative real-time PCR

4T1-GFP-luc or MDA-MB-231-luc cells were seeded onto 6-well plates overnight. TGFβ1 (5 ng/mL, Abcam, Cambridge, MA, USA) was used to treat the

4T1-GFP-Luc and MDA-MB-231-luc cells for 3 days. The total RNA was isolated

173

using the RNeasy Plus Kit (Qiagen, Valencia, CA, USA). Reverse transcription was performed using the High Capacity cDNA Transcription Kit (Applied Biosystems,

Waltham, MA, USA). Semiquantitative real-time PCR was conducted with the

SYBR Green Master Mix (Applied Biosystems, San Diego, CA, USA) and the

Master ep realplex2 (Eppendorf, Hamburg, Germany) according to manufacturer’s recommendations. Quantification of the relative expression of EDB mRNA was performed using the 2-ΔΔCt method and the level of GAPDH mRNA as internal control. Primers used in this study included: 5’- CCTGGAGTACAATGTCAGTG

-3’ (forward) and 5’- GGTGGAGCCCAGGTGACA -3’ (reverse) for human EDB;

5’- CCTGGAGTACAATGTCAGTG -3’ (forward) and 5’-

GGTGGAGCCCAGGTGACA-3’ (reverse) for mouse EDB; 5‘-

ACCCAGAAGACTGTGGATGG-3’ (forward) and 5’-

TCTAGACGGCAGGTCAGGTC -3’ (reverse) for human GAPDH; 5’-

TCCATGACAACTTTGGTATTCGT-3’ (forward) and 5’-

AGTAGAGGCAGGGATGATGTT -3’ (reverse) for mouse GAPDH.

Western Blot

The 4T1 primary and metastatic tumor tissues (30-100 mg) were acquired during a previous study using orthotopic 4T1 tumor models, which were homogenized in 200-500 μL T-PER buffer (Thermo Fisher Scientific) supplemented with the protease inhibitor cocktail (Sigma-Aldrich) and PMSF

(Phenylmethanesulfonyl fluoride) (Sigma-Aldrich). After centrifugation at 10,000

174

g for 10 min at 4 °C, the supernatant of the lysates was collected and the protein concentration quantified using BCA assay (Biorad, Hercules, CA, USA). Proteins of 25 μg were resolved using SDS-PAGE, transferred to polyvinylidene difluoride

(PVDF) membranes (Invitrogen, Carlsbad, CA, USA), and incubated with the anti-

EDB-FN antibody (BC-1, Abcam, Hercules, CA). After washing with TBST, fluorescein-conjugated anti-mouse secondary antibody was applied. For visualization of β-actin, the fluorescein-conjugated anti-β-actin antibody was used.

The Typhoon trio scanner (GE healthcare) was used for visualization of EDB-FN and β-actin bands using the channel for fluorescein.

3D culture and peptide binding study

The 3D culture of 4T1 and MDA-MB-231 cells was used to evaluate binding of the EDB-targeting probe to breast cancer cells within a tumor- microenvironment. The 3D matrigel culture was prepared as described previously.

Briefly, matrigel (3.6-4.4 mg/mL) diluted with PBS was pipetted onto a glass bottom plate and allowed to gel at 37 °C for 30 min. A thick matrigel layer was formed. Cells were then seeded into the plate. After 5 days, fluorescent dye-labeled peptides (200 nM) were added to the medium and incubated with the 3D spheres for 1 h. Confocal laser scanning microscopy (Olympus FV1000, Japan), was used to image the 3D spheres. Intensive shaking was avoided during peptide binding and imaging.

Primary and metastatic breast cancer models

175

4T1 breast cancer models were constructed by injecting 3 million cells into the mammary fat pad of Balb/c mice, 4-6 weeks of age. MDA-MB-231 models were constructed by injecting 3 million cells subcutaneously into the flank of mice.

The 4T1 metastatic tumor models were constructed by injecting 1 million cells into the left ventricle of heart in PBS. Bioluminescence imaging was used to monitor the growth of metastatic tumors in whole body after i.p. injection of D-Luciferin

(Gold Biotechnology, St Louis, MO, USA). At about 3 weeks after injection, mice were used for MRI.

Ex vivo fluorescence imaging

MDA-MB-231 and 4T1 tumor-bearing mice were subjected to injection of

10 nmol ZD2-Cy5.5 or CREKA-Cy5.5. At 3h post-injection, mice were sacrificed using cervical dislocation, with tumor and normal tissues dissected. Images of the organs were examined using the Maestro Imaging System (Caliper life Sciences,

Waltham, MA, USA) using the yellow filter set and exposure time of 1000 ms. GFP signal from the tumor was recorded using the blue filter set and exposure time of

500 ms.

Histological analysis

Following the ex vivo fluorescence imaging, tumors and normal tissues were embedded in optimal cutting temperature compound (OCT) and kept frozen in -80 °C until further use. Tissue slides were prepared after cryo-sectioning the tissues at 5 μm. Tissue slices were fixed with 5% PFA, permeabilized with Triton-

176

X and blocked with 1% BSA. BC-1 antibodies were then used to incubate the tissue for 1 h. The ProLong Gold mounting medium was then used to mount the coverslip.

MRI

The Bruker Biospec 7T MRI scanner (Bruker Corp., Billerica, MA, USA) equipped with a volume radio frequency coil was used for in vivo MRI. Mice bearing 4T1 and MDA-MB-231 tumors were anaesthetized with 2% isoflurane. A

30-gauge needle connected with a 1.6-m tubing was fixed into the tail vein. Mice were then placed in the magnet and kept under anesthesia with 1.5% isoflurane.

Body temperature was maintained at 36 °C by blowing warm air into the magnet.

For primary tumor, an axial T1-weighted multi-slice multi-echo sequence was used with the following parameters: field of view (FOV): 3 cm; slice thickness: 1.2 mm; interslice distance: 1.2 mm; TR: 500 ms, TE: 8.1 ms; flip angle: 90°; average: 2; matrix size: 128 × 128. A dose of 0.1 mmol/kg was injected through the tail vein.

For imaging of metastases imaging, a high-resolution fat suppression 3D T1- weighted FLASH sequence with respiratory gating was used. The parameters were as follows: TR: 25 ms, TE: 2.8 ms, average 3, flip angle: 15 °, in-plane FOV: 6cm, slab thickness: 18-mm, resolution: 0.1172 × 0.0976 × 0.562 mm, scan duration with respiratory gating: 20 min. A dose of 0.2 mmol/kg was used. The CNR of tumors in the MR images was calculated using the following equation: CNR = (Stumor-

Snormal)/(σ), where Stumor and Snormal denote the signal in tumor and its surrounding

177

normal tissue, respectively, and σ is the standard deviation of noise estimated from the background air.

5.4 Results

EDB upregulation is a hallmark of TNBC breast cancer with high metastatic potential

To validate EDB as a molecular target for imaging, we first investigated whether EDB upregulation correlates to the malignancy of breast cancer.

Transforming growth factor beta (TGFβ) induction is known to drive the generation of breast cancer cells with higher metastatic potential. As shown in Figure 1a and

S1, EDB mRNA levels was higher in TGFβ treated 4T1 and MDA-MB-231 cells, which is consistent with the role of EDB-FN as a marker of EMT [12]. We also showed that EDB-FN is highly expressed in 4T1 tumor tissues (Figure 1b).

Importantly, metastatic 4T1 tumors demonstrated even higher EDB-FN expression than that of primary tumors. Negligible EDB-FN expression was seen in brain, lung and liver tissues. By targeting EDB proteins in tumor microenvironment, this upregulation in metastatic sites could potentially increase locoregional probe concentration, which enhances detection sensitivity. Moreover, we conducted binding studies on 3D culture of 4T1 cells. The 3D culture of 4T1 cells shown in

Figure 1c is a close mimic of the micrometastases in secondary organs. The Cy5.5 labeled ZD2 peptide bound to not only the entire 3D spheres in matrigel, but also the periphery due to EDB secretion in the periphery of spheres. The binding of

178

CREKA-Cy5.5 is used as a control. CREKA is pentapeptide targeted to the fibronectin-fibrin complex in the microenvironment in the aggressive tumor [16].

Here we show that CREKA-Cy5.5 did not bind to 3D culture of 4T1, indicating that the no fibronectin-fibrin complexes were formed by in 3D spheres. This supports the advantage of using EDB-targeted probes for imaging metastases as

EDB-FN is highly secreted by metastases, independent other components of tumor stroma. Imaging technologies based on EDB-targeting could presumably lead to higher detection sensitivity of micrometastases.

179

Figure 1. EDB-FN overexpression is a hallmark for 4T1 tumors with an aggressive phenotype. a. Analysis of EDB mRNA levels in 4T1 cells with or without TGFβ treatment. The level of GAPDH was used as an internal control. Data are normalized to the EDB mRNA level of non-treated 4T1 cells. b. Western blot images showing the relatively higher EDB-FN expression in 4T1 metastatic tumors in comparison to that in primary tumor and normal tissues. c. Binding of Cy5.5 labeled ZD2 and CREKA on 3D matrigel culture of 4T1 cells. Peptide binding is shown in red.

MR molecular imaging of EDB-FN increases contrast enhancement of TNBC primary tumors

The EDB-targeting MRI contrast agent, ZD2-Gd(HP-DO3A), was used to test its ability in imaging 4T1 primary xenografts. First, the distribution of ZD2-

Cy5.5 in 4T1 primary tumor model was characterized by ex vivo fluorescence imaging (Figure 2a). At 3 h post-injection, tumors and other normal tissues were harvested and examined for Cy5.5 fluorescence. It is clear that substantially higher signal can be seen in tumors, while signal in other normal tissues was relatively lower, suggesting that the EDB-targeting probes preferentially accumulated in 4T1 tumors. Histoimmunological analysis of those tumors indicated that ZD2-Cy5.5 signal precisely localized with EDB expression, as probed by anti-EDB antibodies, which distributed in the ECM of tumor in a fibrillary pattern (Figure 2b). Moreover,

180

fluorescence imaging of MDA-MB-231 primary tumors also confirmed a specific accumulation of ZD2-Cy5.5 in the tumors (Figure S2a).

Figure 2. Evaluation of ZD2 in targeting 4T1 primary tumor by ex vivo fluorescence imaging and histological analysis. a. Ex vivo fluorescence imaging of tumor and organs harvested from a 4T1 primary tumor bearing mouse at 4 h after injection of ZD2-Cy5.5. b. Histological analysis of the binding of peptides and

EDB-FN distribution. Merged channel of ZD2 and EDB-FN showed the colocalization of ZD2-Cy5.5 and EDB-FN.

Inspired by this, we proceeded to test the efficacy of ZD2-Gd(HP-DO3A) in MR molecular imaging of 4T1 and MDA-MB-231 primary tumors. ZD2-Gd(HP-

DO3A) was administered at a dose of 0.1 mmol/kg through the tail vein. Axial images of tumor were acquired in T1-weighted MRI before and during 30 min post-

181

injection. As shown in Figure 3a, no apparent contrast could be observed between

4T1 tumors and normal tissues in pre-injection images. At 10 min post-injection, prominent contrast enhancement could be identified in the periphery of 4T1 tumors due to probe accumulation. As contrast agents diffused into the tumor and signal from surrounding normal tissues cleared out, the CNR of tumor remained high at

20 min and 30 min post-injection (Figure 3a and 3b). Analysis of the contrast-to- noise ratio (CNR) indicated that the CNR of tumor increased over 200% after injection. Comparatively, administration of the nonspecific contrast agent,

ProHance, only resulted in less than 50% increase in tumor CNR, which decreased quickly thereafter (Figure 3b). Consistently, we also demonstrated that ZD2-

Gd(HP-DO3A) induced a significant contrast enhancement (an 100% to 300% increase in CNR) in the MDA-MB-231 xenograft (Figure S2b and S2c). Limited

CNR increase was observed in MDA-MB-231 tumors using ProHance as the contrast agent. Collectively, these data demonstrated the superior efficacy of ZD2-

Gd(HP-DO3A) in generating contrast enhancement in primary TNBCs, compared to ProHance. Based on this, we continued to assess the feasibility of MRI detection of TNBC micrometastases using ZD2-Gd(HP-DO3A).

182

Figure 3. MRI of 4T1 primary tumors. a. Representative axial MRI images acquired from 4T1 tumor bearing mice injected with ZD2-Gd(HP-DO3A) or

ProHance. Tumor locations are denoted with arrowheads. b. CNR analysis of tumor.

MR molecular imaging of EDB-FN enables detection of TNBC micrometastases

The 4T1 metastatic tumor model was constructed by intracardiac injection of 4T1-GFP-Luc cells. Intracardial injection of 4T1-GFP-Luc cells in the left ventricle, rather than intravenously injection, effectively bypassed retention of cancer cells by lung veins, leading to widespread metastases formation in various organs. Bioluminescence imaging (BLI) was used to monitor the tumor growth

(Figure 5a). At two to three weeks post-injection, ex vivo fluorescence imaging was first performed to evaluate targeting of ZD2-Cy5.5 to metastases.

Representative images of the correlation of 4T1 micrometastases manifested by

GFP and ZD2-Cy5.5 signal are shown in Figure 4. Brain and lung micrometastases

183

was prominently highlighted by the EDB-targeting fluorescent probes (Figure 4a).

Large lymph node and adrenal gland metastases were also significantly illuminated, compared to the muscle (Figure 4b). Encouraged by this, we conducted whole- body high-resolution MRI with i.v. injection of 0.2 mmol/kg ZD2-Gd(HP-DO3A).

The high resolution T1-weigthed sequence were completed before injection and initiated at 3 min after injection, which lasted 20 min for each. Figure 5b shows representative MRI images of metastatic tumors in lymph nodes and adrenal glands.

Despite their small size, these tumors were dramatically enhanced after ZD2-

Gd(HP-DO3A) injection, and correlation of tumor locations between MRI and BLI was possible (Figure 5b). Subtraction of post-contrast by pre-contrast images clearly highlighted the location and size of these tumors (Figure 5b). Consistently, a marked increase in CNR of these identified tumors could be observed (Figure

5c). Whereas, metastatic tumors showed no distinguishable contrast in neither non- contrast-enhanced MRI or ProHance-enhanced MRI (Figure 5b). No statistically significant difference in CNR could be seen for pre-injection and post-injection tumor CNR using ProHance as the contrast agent (Figure 5c). Additionally, we explored the EDB-FN-targeting probes in detecting MDA-MB-231 metastases. We first verified that ex vivo fluorescence imaging with the EDB-targeted fluorescent probe successfully detected a lymph node metastatic tumor in the armpit (Figure

S3a). In MRI experiments, we used a mouse with metastatic tumors developed around the abdominal region as identified by BLI (Figure S3b). Imaging of this

184

mouse with ZD2-Gd(HP-DO3A) generated significant signal enhancement in these metastatic tumors, whereas no clear tumor delineation could be observed using

ProHance (Figure S3b). Taken together, these data support that ZD2-Gd(HP-

DO3A) could also be used for non-invasive detection of TNBC micrometastases in

MRI by inducing a higher contrast enhancement in these micrometastases.

Figure 4. Ex vivo imaging of organs containing 4T1 metastatic tumors at 5 h after injection of ZD2-Cy5.5. a. Images of brain and lung with GFP denoting the tumor location, ZD2-Cy5.5 signaling demonstrating the specific targeting of the

185

probe. b. Images of muscle, lymph node, and adrenal gland with 4T1 metastatic tumors.

Figure 5. MRI of 4T1 metastatic tumor models. a. Illustration of animal handling and MRI procedures. Mice were injected with 4T1-GFP-luc cells via the left ventricle of the heart. During 2-3 weeks, BLI was used to track the growth of metastatic tumors throughout the body. MRI was performed before and after i.v. injection of 0.2 mmol/kg ZD2-Gd(HP-DO3A), followed by inspection of

186

subtraction images and CNR of tumors. b. Representative BLI and MRI images of mice. Images that covered tumors in the adrenal glands and lymph nodes were presented to compare the efficiency of ZD2-Gd(HP-DO3A) and ProHance in highlighting 4T1 metastatic tumors.

187

5.5 Discussions

Here we report MR molecular imaging of murine and human TNBCs with the EDB-targeting contrast agent ZD2-Gd(HP-DO3A). The contrast agent has a small size, which allows rapid clearance of unbound Gd-based contrast agent from the body via renal filtration [13]. The smaller size of ZD2-Gd(HP-DO3A) could facilitate extravasation, tumor penetration, thus rendering higher efficiency in imaging of TNBCs and metastases embedded in normal tissues. We first demonstrated that ZD2-Gd(HP-DO3A) provided superior contrast enhancement in

TNBC primary tumors compared to the clinical contrast agent ProHance. EDB-FN is abundantly expressed in the tumor microenvironement, which is more accessible to intravenously injected contrast agent than cell-surface or intracellular biomarkers.

The specific expression of EDB-FN in tumors minimizes non-specific enhancement in normal tissues, leading to clear delineation of tumor margins.

Currently, MRI detection of TNBC is based on features such as mass-type lesion, smooth mass margin, high rim enhancement, etc., which are not definitive markers that differentiate TNBCs from low-risk tumors [3, 17, 18]. This EDB-FN-targeting contrast agent enhances the capability of MRI in identifying TNBCs by inducing tumor contrast enhancement specifically in aggressive tumors. This could serve as an efficient method for early detection of these high-risk tumors. Preoperative MR examination with this contrast agent could possibly increase accuracy of TNBC

188

localization to ensure complete dissection of tumors, meanwhile minimizing unnecessary removal of healthy tissues during surgery.

We next explored the feasibility of ZD2-Gd(HP-DO3A) in detecting TNBC micrometastases. Currently, metastatic status is evaluated, predominantly by biopsy, to determine cancer stages, perform treatment planning and evaluate treatment response. Breast cancers with metastases require treatment far more complicated than that for locoregional cancers. Unfortunately, current imaging technologies have limited efficacy in detecting breast cancer metastases. Here we show that whole-body MRI with ZD2-Gd(HP-DO3A) prominently delineated metastases. This finding supports the essential role of EDB for the outgrowth of engrafted secondary tumors. Accurate staging could potentially be enabled by this technology by non-invasively examining the presence of metastases in lymph node, adrenal gland, bone and other organs, exempting patients of painful and risky biopsies that exacerbate quality of life. Early and accurate localization of metastases also enable evaluation of treatment response and detection of recurrent tumors, allowing for timely adjustment of treatment strategies. Further, diagnostic strategies capable of detecting and localizing these micrometastases early is the prerequisite for the design of effective therapies against these micrometastases. The inability to localize metastases leaves chemotherapy the only feasible therapeutic option. Based on the successful imaging of metastases in MRI with ZD2-Gd(HP-

189

DO3A), further pursuits of therapeutic strategies for metastatic tumors may be practical.

Currently, MRI is emerging as a favorable diagnostic tool in breast cancer due to its higher sensitivity than mammography, ultrasound, and the combination or both. Despite its high sensitivity, MRI has not demonstrated effectiveness in detecting metastatic tumors. The limited specificity of MRI to aggressive breast cancers led to the use of MRI only as a screening tool in women with high risk based on family history and a genetic predisposition, such as BRCA mutation [19].

Our previous study indicated that by using this EDB-targeted agent, non-invasive determination of prostate cancers aggressiveness could be achieved in MRI. In this study, substantially higher contrast enhancement was seen in high-risk PC3 tumors than low-risk LNCaP tumors. This is due to the specific and abundant expression of EDB-FN in cancers undergone EMT. TNBCs represent post-EMT breast cancers.

Our technology overcomes the difficulty of imaging post-EMT cancers by targeting tumor microenvironment EMT signatures. Distant metastases, as the result of EMT, were also successfully detected. Thus far, rare studies have been performed on the use of MRI in imaging detection of breast cancer metastases, and these reports are mostly focused on imaging of axillary lymph node, brain and bone metastases [15,

20, 21]. Together, our data are consistent with evidences showing that EDB-FN is an attractive target for development of imaging probes for a large array of cancers

[22-24]. We think this EDB-FN-targetig contrast agent has a great potential in

190

extending the ability of clinical breast MRI in sensitive and specific detection of

TNBCs, meanwhile opening a new avenue to detection and treatment of breast cancer metastases.

191

Supplementary information

Figure S1. Analysis of EDB mRNA in MDA-MB-231 with or without TGFβ treatment. The level of GAPDH was used as an internal control. Data are normalized to the EDB mRNA level in non-treated MDA-MB-231 cells.

192

Figure S2. Imaging of MDA-MB-231 primary tumor. a. Ex vivo fluorescence imaging of tumor and organs harvested from MDA-MB-231 primary tumor models. b. Representative axial MRI images of MDA-MB-231 primary tumor models injected with ZD2-Gd(HP-DO3A) or ProHance. c. CNR analysis of MRI of MDA-

MB-231 tumor models.

193

Figure S3. Imaging of MDA-MB-231 metastatic tumor. a. Ex vivo fluorescence imaging of MDA-MB-231 metastatic tumor models at 5 h after injection. b. MRI of MDA-MB-231 metastatic tumor mice. BLI images showed tumor locations viewed at two angles. The same mice were imaged with both ZD2-Gd(HP-DO3A) and ProHance.

194

References

1. Haffty, B.G., et al., Locoregional relapse and distant metastasis in conservatively managed triple negative early-stage breast cancer. J Clin Oncol, 2006. 24(36): p. 5652-7.

2. Foulkes, W.D., I.E. Smith, and J.S. Reis-Filho, Triple-negative breast cancer. N Engl J Med, 2010. 363(20): p. 1938-48.

3. Gluz, O., et al., Triple-negative breast cancer-current status and future directions. Annals Of Oncology, 2009. 20(12): p. 1913-1927.

4. Hosseini, H., et al., Early dissemination seeds metastasis in breast cancer. Nature, 2016. 540(7634): p. 552.

5. Hamaoka, T., et al., Bone imaging in metastatic breast cancer. Journal Of Clinical Oncology, 2004. 22(14): p. 2942-2953.

6. Lee, C.H., et al., Breast Cancer Screening With Imaging: Recommendations From the Society of Breast Imaging and the ACR on the Use of Mammography, Breast MRI, Breast Ultrasound, and Other Technologies for the Detection of Clinically Occult Breast Cancer. Journal Of the American College Of Radiology, 2010. 7(1): p. 18-27.

7. Rakha, E.A., et al., Prognostic markers in triple-negative breast cancer. Cancer, 2007. 109(1): p. 25-32.

8. Baum, R.P., et al., Molecular imaging of HER2-expressing malignant tumors in breast cancer patients using synthetic 111In-or 68Ga-labeled affibody molecules. Journal of nuclear medicine, 2010. 51(6): p. 892-897.

195

9. Metzger, O., et al., Dissecting the Heterogeneity of Triple-Negative Breast Cancer. Journal Of Clinical Oncology, 2012. 30(15): p. 1879-1887.

10. Han, Z. and Z.-R. Lu, Targeting fibronectin for cancer imaging and therapy. Journal of Materials Chemistry B, 2017.

11. Zhou, Z., et al., MRI detection of breast cancer micrometastases with a fibronectin-targeting contrast agent. Nature Communications, 2015. 6: p. 7984.

12. Petrini, I., et al., ED-B fibronectin expression is a marker of epithelial- mesenchymal transition in translational oncology. Oncotarget, 2017. 8(3): p. 4914-4921.

13. Han, Z., et al., Targeted Contrast Agent Specific to an Oncoprotein in Tumor Microenvironment with the Potential for Detection and Risk Stratification of Prostate Cancer with MRI. Bioconjug Chem, 2017.

14. Zhang, Y., et al., A novel role of hematopoietic CCL5 in promoting triple- negative mammary tumor progression by regulating generation of myeloid-derived suppressor cells. Cell Res, 2013. 23(3): p. 394-408.

15. Hamilton, A.M. and P.J. Foster, In vivo magnetic resonance imaging investigating the development of experimental brain metastases due to triple negative breast cancer. Clin Exp Metastasis, 2017.

16. Zhou, Z., et al., Peptide targeted tripod macrocyclic Gd(III) chelates for cancer molecular MRI. Biomaterials, 2013. 34(31): p. 7683-93.

17. Uematsu, T., M. Kasami, and S. Yuen, Triple-Negative Breast Cancer: Correlation between MR Imaging and Pathologic Findings 1. Radiology, 2009. 250(3): p. 638-647.

18. Chen, J.H., et al., Triple-negative breast cancer: MRI features in 29 patients. Ann Oncol, 2007. 18(12): p. 2042-3.

196

19. Evans, D.G., et al., MRI breast screening in high-risk women: cancer detection and survival analysis (vol 145, pg 663, 2014). Breast Cancer Research And Treatment, 2014. 147(3): p. 691-691.

20. Kim, E.J., et al., Diagnostic value of breast MRI for predicting metastatic axillary lymph nodes in breast cancer patients: diffusion-weighted MRI and conventional MRI. Magnetic Resonance Imaging, 2014. 32(10): p. 1230-1236.

21. Serres, S., et al., Molecular MRI enables early and sensitive detection of brain metastases. Proceedings Of the National Academy Of Sciences Of the United States Of America, 2012. 109(17): p. 6674-6679.

22. Berndorff, D., et al., Imaging of tumor angiogenesis using 99mTc-labeled human recombinant anti-ED-B fibronectin antibody fragments. Journal of nuclear medicine, 2006. 47(10): p. 1707-1716.

23. Santimaria, M., et al., Immunoscintigraphic detection of the ED-B domain of fibronectin, a marker of angiogenesis, in patients with cancer. Clinical Cancer Research, 2003. 9(2): p. 571-579.

24. Locher, R., et al., Abundant in vitro expression of the oncofetal ED-B- containing fibronectin translates into selective pharmacodelivery of (131)I-L19SIP in a prostate cancer patient. J Cancer Res Clin Oncol, 2014. 140(1): p. 35-43.

197

Chapter 6 Future directions and concluding remarks

This chapter is partially reproduced based on the following published paper:

Zheng Han, Zheng-Rong Lu. Targeting Fibronectin for Cancer Imaging and

Therapy. 2016. Journal of Materials Chemistry B.

6.1 Summary of current work

In this thesis, we initiated our study based on the urgent need for a non- invasive approach for early and accurate diagnosis of prostate and breast cancer.

Considering the critical role of EMT in generating tumors with higher metastatic potential and poor clinical outcome, we chose EDB-FN, a tumor ECM molecule highly expressed during EMT, as the biomarker for development of tumor-homing peptides. The nonapeptide, ZD2, was discovered, which was further used for synthesizing MRI probes that target EDB-FN. Those MRI probes demonstrated sensitive detection of prostate and breast cancer with high aggressiveness, whereas generating significantly lower contrast in unlethal tumors. Accurate and sensitive localization of metastatic tumors is also possible using these EDB-targeting contrast agents. Therefore, we believe that when incorporated in current MRI procedures, these agents possess great clinical potential in improving the screening, preoperative diagnosis, surgery guidance, active surveillance, and response monitoring of both cancers.

198

ZD2 can be used for other imaging modalities, such as PET, and cancer therapies by delivering therapeutic agents, etc. The ensuing sections will discuss current progresses in aforementioned directions and propose studies that can be conducted based on existing data.

6.2 Nuclear imaging based on EDB-targeting probes

Nuclear imaging is advantageous over MRI in its high sensitivity and administration of much less doses of imaging agents. Therefore, development of imaging agents for nuclear imaging is worth investigation. As discussed in Chapter

2, FN-targeting strategies have been exploited for developing nuclear imaging probes for cancers. However, all reported nuclides are antibody-based and EDB- targeting peptides have not yet been used for synthesizing nuclides. Small peptides are the preferred agents over antibodies due to the smaller size, high tumor-to- background ratios and rapid blood clearance. EDB-targeting peptides, such as ZD2 discovered in our study, is suitable for conjugating of radioisotopes for nuclear imaging purposes. Isotopes for PET can be classified mainly as radioactive nonmetallic elements, such as 18F, 11C, 15O, and 13N, and metallic elements, such as

64Cu, 89Zr, and 68Ga.

6.2.1 Non-metallic elements

The half-lives of 18F, 11C, 15O, and 13N are 110 min, 20.4 min, 2.04 min and

9.97 min, respectively. The short half-lives of 11C, 15O, and 13N make them

199

unsuitable for practical use. Hence, this section focuses on the chemistry for 18F labeling of peptides. In addition to the longer half-life, 18F has a low positron energy of 0.64 MeV, giving low radiation dose to patients and a short range in tissue (2.3 mm). Therefore, 18F-probes afford a higher imaging resolution in PET. The disadvantage of 18F labeling is the laborious and time-consuming preparation of the

18F labeling precursors. Techniques that allow fast and efficient labeling of 18F have recently been devised [1].

The essential component for 18F-labeling of peptides is the bifunctional ligands suitable for attaching both radionuclides and peptides. Traditionally, freshly prepared 18F is used to displace a leaving group on the prosthetic molecule, which was then attached to peptide via oxime formation, acylation, alkylation, maleimide/thiol coupling, etc. The most popular chemistry of this route is based on

N-Succinimidyl 4-[18F](fluoromethyl)benzoate [(18F)SFB] (Figure 1A). Details concerning available prosthetic groups and applications can be referred in several reviews [1-5]. Unfortunately, the 18F-prosthetic group conjugation is still time- consuming, cumbersome and requires HPLC purification, which hinders the development of new PET probes for medical use. Recently, the formation of aluminum-fluoride complexes have be exploited for 18F labeling of peptide. The

18F-aluminum complexes form rapidly and tightly, allowing for rapid preparation of stable 18F labeled peptides [6]. In this setting, Al3+ is first complexed with a chelator, in which pentadenate chelator is used, leaving one binding site in Al3+ for

200

fluorine (Figure 1B). The NOTA ligand is known to form stable complexes with

Al3+, and therefore received extensive testing. This route requires fewer steps and time before the probe is ready for injection. It also demands less or no purification.

Thus, this 18F-Al chemistry is most suitable for 18F-labeling of ZD2 for PET imaging of EDB-FN expression in cancer.

Figure 1. Illustration of [18F]SFB and Al-18F based peptide labeling procedures.

6.2.2 Metallic elements

Compared to non-metallic isotopes, nonmetallic isotopes have much longer half-lives (89Zr: 3.3 d; 64Cu: 12.7 h; 68Ge: 270 d). Radiolabeling procedure of

201

those metallic isotopes is predominantly a complexation process, which only takes less than an hour. 68Ge is the parent isotope of 68Ga. The long half-life of 68Ge makes it attractive for preclinical and clinical applications since once 68Ge is generated, the generator containing 68Ge can be eluted several times a day, significantly lowering the cost for each experiment. If synthesized in large quantities, 68Ge can sustain almost for a year. 64Cu is also frequently used for PET imaging, and its short half-life is desired to reduce radiation dose to patients. 89Zr is also useful for imaging in patients. The choice of metallic isotopes is based on the half-life of isotopes and the pharmacokinetics of the targeting ligand.

Considering the small size of peptides, 64Cu is presumably an ideal isotope for peptide labeling [7].

As introduced in Chapter 4, the synthesis of ZD2-HP-DO3A can be easily adapted to afford 64Cu labeling for PET imaging. In this case, Cu chelation is performed after the conjugation of ZD2 and HP-DO3A. However, click chemistry is not applicable here since click chemistry entails the presence of Cu(I) as a catalyst, which can also be chelated. Therefore, another synthesis approach can be devised, as shown in Figure 2. All steps prior to Cu chelating are designed for solid phase synthesis, allowing for high-yield and easy purification. Complexation between 64Cu and ZD2-HP-DO3A can be performed in PBS for 30 min at 45 °C and the final solution is immediately ready for injection.

202

Figure 2. Proposed 64Cu labeling procedure for ZD2.

6.3 Delivery therapeutic agents based EDB-targeting strategy

6.3.1 Radiotherapy

Cancer therapy can be achieved by replacing the γ-emitters in FN targeting imaging probes with high-energy radioisotopes. This strategy embraces the benefit of selective targeting and decreased systemic toxicity. The first reported FN targeting radiation therapy was 125I labeled BC-1 antibody (named as 125I-BC-1), which accumulated favorably in human tumor implants[8]. Later studies directed towards using L19SIP antibody due to its specific targeting to EDB domain.

Administration of 131I-L19SIP resulted in selective uptake in SW1222 and LS174T colorectal tumors, which led to tumor growth retardation and prolonged survival of

203

mice[9]. 131I-L19SIP was also tested in treating FaDu and HNX-OE head and neck squamous cell carcinoma (HNSCC) model, in which a combination of 131I-L19SIP and anti-EGFR mAb cetuximab achieved a significant therapeutic effect[10], and in imaging and treating Hodgkin and non-Hodgkin lymphoma patients[11]. 131I-

L19SIP radioimmunotherapy induced a sustained partial response in 2 relapsed

Hodgekin lymphoma patients.

Inspired by those studies, ZD2 can be radiolabeled with radiotherapy ligands to increase radiation dose to tumors, while reducing dose to normal tissue.

The potential radioisotopes for labeling peptides include 111In, 131I, 177Lu and 99mTc.

For 131I labeling, the standard iodogen method, which involves radioiodination using Na131I in presence of oxidants such as chloramines-T or tetrachlorodiphenylglycouril, requires Tyrosine residues in peptides for the substitution of the iodine ortho to hydroxyl groups. Therefore, Tyrosine needs to be added to ZD2, presumably after a spacer. Since 111In, 177Lu and 99mTc are metallic isotopes, they can be used for radiolabeling of ZD2 using similar method proposed in section 6.2.2. It can be expected that ZD2 could facilitate higher tumor uptake of the radioligands and minimize non-specific accumulation in normal tissue, potentially improving the clinical outcome of cancer patients.

6.3.2 Delivery of anti-tumor agents

FN-targeting ligands have been used to deliver chemotherapy drugs to FN expressing tumors. The specific delivery of these cytotoxic drugs to tumor could

204

make the best use of the drug’s anti-tumor effect, while minimizing harm to normal cells. APTEDB was conjugated to doxorubicin-containing liposomes for targeted drug delivery (Figure 3). As reported, APTEDB targeted liposomes specifically accumulates in glioma, resulting in 55% decrease in tumor size as compared to 20% decrease with free doxorubicin[12]. APTEDB was also tested in guiding PEG-PLA nanoparticles loaded with paclitaxel (PTX) for treatment of glioma [13]. The EDB targeting nanoparticles loaded with PTX resulted in enhanced tumor regression than the free drug and a prolonged survival. Direct conjugation of APTEDB to docetaxel (DTX) also yielded an improved delivery efficiency of docetaxel in glioma, with enhanced therapeutic effects [14]. CLT1 was used to modify PEG-

PLA nanoparticles to delivery PTX to glioma [15]. By targeting overexpressed

FibFN in tumor, the CLT1 modified PTX-loaded nanoparticles exhibited favorable nanoparticle penetration into the core of glioma spheroid and therefore induced more inhibitive effects on glioma growth. Similarly, CREKA was recently used to modify polyamidoamine (PAMAM) dendrimer[16]. The modified dendrimer was able to penetrate glioblastoma (GBM) tissue and enhance the retention effect, which could be potentially used to deliver chemotherapy drugs.

In another study, ZD2 was conjugated to DOX, yielding an amphiphilic compound that self-assembles into nanoparticles[17]. These nanoparticles targeted efficiently to solid PC3 tumors, and disassembled in the high thiol environment in

205

tumor to release DOX. As a result, an enhancement tumor-inhibiting effect was achieved.

FN-targeting ligands have also been used for targeted delivery of cytokines, including interleukin-2 (IL-2), interleukin-12 (IL-12), and TNFα. Cytokines are able to activate a wide range of immunological cells, which include cytotoxic T cells, natural killer cells, and lymphokine activated killer cells. However, therapeutic efficacy of cytokines is limited by its short blood half-life and severe toxicity related to vascular leak syndrome at high doses. The genetic fusion of scFv

L19 and IL-2 resulted in L19-IL-2, which enabled specific delivery of IL-2 to F9 murine teratocarcinoma so that an efficacious concentration can be achieved in tumor without causing severe systemic toxicity [18]. Concomitant with inhibited tumor growth, tumors treated with L19-IL2 were rich in accumulation of CD8+ cytotoxic T lymphocytes, CD4+ cells, CD11b+ cells (macrophages and natural killer cells), symphonizing with the function of IL2 [18]. Besides, L19-IL-2 has been tested in preclinical studies to treat other types of cancers, including lymphoma[19] and pancreatic cancer [20], in which L19-IL2 consistently showed enhanced therapeutic potency. L19-IL2 is branded as DARLEUKIN® for clinical use, and is mainly used to treat melanoma, head and neck cancer, and lymphoma in combination with L19-TNF, dacarbazine or rituximab. Recent clinical phase II evaluation of L19-IL2 in treating melanoma patients showed a promising result in terms of reducing metastasis and extending patient survival [21, 22]. The combined

206

treatment of L19-IL2 and dacarbazine was reasonably tolerated, and all side effects encountered in the study were manageable and reversible. Similarly, an EDA-FN targeting antibody, F8, was used to deliver IL-2 to a variety of cancers, including renal cell carcinoma [23], melanoma [24], colon cancer, lymphoma, and teratocarcinoma [25]. L19 was also used to modify IL-12, yielding L19-IL-12, to treat mouse lung metastases and aggressive murine tumors [26]. Enhanced anti- tumor effect was seen with the increased tumor infiltration with lymphocytes, macrophages and natural killer cells. Humanized BC-1 antibody, huBC1, was also developed to modify IL-12 for tumor therapy [104]. One cycle of treatment with huBC1-IL12 resulted in tumor suppression in PC3mm2 (human prostate cancer),

A431 (human melanoma), and HT29 (human colon cancer) subcutaneous tumor models, and PC3mm2 lung metastasis model [104]. A phase I study was conducted using huBC1-IL12, and demonstrated that huBC1-IL12 was well tolerated at the dose of 15 μg/kg weekly. Stable disease was seen in 46% of the patients [27, 28].

Tumor necrosis factor (TNF) is another cancer-related cytokine modified with FN-targeting strategies. An antibody-cytokine fusion of L19 and TNF, L19-

TNF, was constructed for cancer therapy. Initial effort with L19-TNF therapy in murine models induced a retardation of tumor growth, but no curative effect was observed [29]. However, the combination of L19-IL2 and L19-TNF therapy showed a synergetic effect to eradicate F9 teratocarcinomas grafted in immunocompetent mice [30]. Also, tumor development was delayed when the

207

cured mice were again challenged with tumor cells, indicating an induction of anti- tumor vaccination effect of the combination therapy. Treatment with L19-TNF in combination with melphalan induced tumor regression and long lasting tumor rejection in 83% of BALB/c mice with WEHI-164 fibrosarcoma and 33% of animals with C51 colon carcinoma [31]. Similar results were seen for treatment of sarcomas and melanoma [32, 33]. Encouraged by these preclinical trials, a phase

I/II first-in-human trial was conducted L19-TNF monotherapy. Dose escalation clinical trials showed little toxic effects in patients with solid metastatic cancers in the dose range of 1.3-14 μg/kg [34]. In another study, L19-TNF was tested in treating melanoma patients [35]. With only 6.25% of the dose approved for TNF,

L19-TNF induced objective responses in 89% of patients, including a complete response in 5/10 patients. The response was durable at 12 months in four patients.

Recently, results on a phase II study that evaluated effects of L19-IL2/L19-TNF combination therapy on stage II or stage IVM1a melanoma patients were reported

[36]. In this study, intralesional administration of L19-IL2 and L19-TNF resulted in complete responses in 32 melanoma lesions in 20 efficacy-evaluated patients, with only mild side effects limited to injection site reactions. This intralesional administration of L19-IL2 and L19-TNF is a simple and effective method for local control of inoperable melanoma lesions. Currently, this approach has been approved by German and Italian authorities for the phase III clinical trial.

208

Thus far, no studies have been reported on the use of FN-targeting peptides for delivery of cytokines. Provided the small size of peptides, the biological activity of cytokines can be best retained. Simple chemical conjugation or genetic fusion between peptides and cytokines can be used.

Considering the role of onfFN in cancer, attempts have been made to silence the expression of onfFN in vivo for cancer therapy. One approach toward suppressing onfFN expression is RNA interference (RNAi) technology [37-39].

RNAi therapy requires safe and efficient delivery of siRNA to tumor sites with high onfFN expression. Advances in developing siRNA delivery systems made this a possibility [40, 41]. Studies have been done using APTEDB as targeting moiety, with

EDB siRNA encapsulated in liposome for treatment of a high-risk breast cancer model derived from breast cancer stem cells (BCSC), as illustrated in Figure 3.

This design enabled simultaneous EDB-FN targeting and siRNA therapy for BCSC tumors with high EDB-FN expression [42]. However, EDB as a target for gene silencing therapy has to be validated in clinical settings. In addition, the bottleneck of development of efficient gene therapy technology is delivery vehicles. It can be envisioned that with recent advances in delivery vehicles [40], EDB silencing, presumably through siRNAs, can potentially provide a novel route for suppressing tumor progression, overcoming cancer resistance, and improving clinical outcome.

209

Figure 3. Structural illustration of APTEDB –modified liposomes for delivery of Doxorubicin and EDB siRNA. Adapted and reprinted with permission based on ref. 117 and 138.

210

References

1. Okarvi, S.M., Recent progress in fluorine-18 labelled peptide radiopharmaceuticals. Eur J Nucl Med, 2001. 28(7): p. 929-38.

2. E Olberg, D. and O. K Hjelstuen, Labeling strategies of peptides with 18F for positron emission tomography. Current topics in medicinal chemistry, 2010. 10(16): p. 1669-1679.

3. Ametamey, S.M., Molecular imaging with PET (vol 108, pg 1501, 2008). Chemical Reviews, 2008. 108(9): p. 4036-4036.

4. Miller, P.W., et al., Synthesis of 11C, 18F, 15O, and 13N radiolabels for positron emission tomography. Angew Chem Int Ed Engl, 2008. 47(47): p. 8998-9033.

5. Schirrmacher, R., C. Wangler, and E. Schirrmacher, Recent developments and trends in F-18-radiochemistry: Syntheses and applications. Mini- Reviews In Organic Chemistry, 2007. 4(4): p. 317-329.

6. McBride, W.J., R.M. Sharkey, and D.M. Goldenberg, Radiofluorination using aluminum-fluoride (AI(18)F). Abstracts Of Papers Of the American Chemical Society, 2014. 248.

7. Baranyai, Z., et al., A shortcut to high-affinity Ga-68 and Cu-64 radiopharmaceuticals: one-pot click chemistry trimerisation on the TRAP platform. Dalton Transactions, 2015. 44(24): p. 11137-11146.

8. Mariani, G., et al., Tumor targeting potential of the monoclonal antibody BC-1 against oncofetal fibronectin in nude mice bearing human tumor implants. Cancer, 1997. 80(12): p. 2378-2384.

211

9. El-Emir, E., et al., Characterisation and radioimmunotherapy of L19-SIP, an anti-angiogenic antibody against the extra domain B of fibronectin, in colorectal tumour models. Br J Cancer, 2007. 96(12): p. 1862-70.

10. Tijink, B.M., et al., Radioimmunotherapy of head and neck cancer xenografts using 131I-labeled antibody L19-SIP for selective targeting of tumor vasculature. Journal of Nuclear Medicine, 2006. 47(7): p. 1127- 1135.

11. Sauer, S., et al., Expression of the oncofetal ED-B–containing fibronectin isoform in hematologic tumors enables ED-B–targeted 131I-L19SIP radioimmunotherapy in Hodgkin lymphoma patients. Blood, 2009. 113(10): p. 2265-2274.

12. Saw, P.E., et al., Aptide-conjugated liposome targeting tumor-associated fibronectin for glioma therapy. Journal Of Materials Chemistry B, 2013. 1(37): p. 4723-4726.

13. Gu, G., et al., PEG-PLA nanoparticles modified with APTEDB peptide for enhanced anti-angiogenic and anti-glioma therapy. Biomaterials, 2014. 35(28): p. 8215-26.

14. Kim, H., et al., Synthesis and therapeutic evaluation of an aptide- docetaxel conjugate targeting tumor-associated fibronectin. J Control Release, 2014. 178: p. 118-24.

15. Zhang, B., et al., Targeting fibronectins of glioma extracellular matrix by CLT1 peptide-conjugated nanoparticles. Biomaterials, 2014. 35(13): p. 4088-98.

16. Zhao, J., et al., CREKA peptide-conjugated dendrimer nanoparticles for glioblastoma multiforme delivery. J Colloid Interface Sci, 2015. 450: p. 396-403.

212

17. Han, G., et al., Self-assembled nanoparticles covalently consisting of doxorubicin and EDB fibronectin specific peptide for solid tumour treatment. RSC Advances, 2016. 6(81): p. 77321-77328.

18. Carnemolla, B., et al., Enhancement of the antitumor properties of interleukin-2 by its targeted delivery to the tumor blood vessel extracellular matrix. Blood, 2002. 99(5): p. 1659-65.

19. Schliemann, C., et al., Complete eradication of human B-cell lymphoma xenografts using rituximab in combination with the immunocytokine L19- IL2. Blood, 2009. 113(10): p. 2275-83.

20. Wagner, K., et al., The targeted immunocytokine L19-IL2 efficiently inhibits the growth of orthotopic pancreatic cancer. Clin Cancer Res, 2008. 14(15): p. 4951-60.

21. Weide, B., et al., Intralesional treatment of stage III metastatic melanoma patients with L19-IL2 results in sustained clinical and systemic immunologic responses. Cancer Immunol Res, 2014. 2(7): p. 668-78.

22. Eigentler, T.K., et al., A dose-escalation and signal-generating study of the immunocytokine L19-IL2 in combination with dacarbazine for the therapy of patients with metastatic melanoma. Clin Cancer Res, 2011. 17(24): p. 7732-42.

23. Frey, K., et al., The immunocytokine F8-IL2 improves the therapeutic performance of sunitinib in a mouse model of renal cell carcinoma. J Urol, 2010. 184(6): p. 2540-8.

24. Moschetta, M., et al., Paclitaxel enhances therapeutic efficacy of the F8- IL2 immunocytokine to EDA-fibronectin-positive metastatic human melanoma xenografts. Cancer Res, 2012. 72(7): p. 1814-24.

213

25. Hemmerle, T. and D. Neri, The antibody-based targeted delivery of interleukin-4 and 12 to the tumor neovasculature eradicates tumors in three mouse models of cancer. Int J Cancer, 2014. 134(2): p. 467-77.

26. Halin, C., et al., Enhancement of the antitumor activity of interleukin-12 by targeted delivery to neovasculature. Nat Biotechnol, 2002. 20(3): p. 264-9.

27. Lo, K.M., et al., huBC1-IL12, an immunocytokine which targets EDB- containing oncofetal fibronectin in tumors and tumor vasculature, shows potent anti-tumor activity in human tumor models. Cancer Immunol Immunother, 2007. 56(4): p. 447-57.

28. Rudman, S.M., et al., A phase 1 study of AS1409, a novel antibody- cytokine fusion protein, in patients with malignant melanoma or renal cell carcinoma. Clin Cancer Res, 2011. 17(7): p. 1998-2005.

29. Borsi, L., et al., Selective targeted delivery of TNFalpha to tumor blood vessels. Blood, 2003. 102(13): p. 4384-92.

30. Halin, C., et al., Synergistic therapeutic effects of a tumor targeting antibody fragment, fused to interleukin 12 and to tumor necrosis factor alpha. Cancer Research, 2003. 63(12): p. 3202-3210.

31. Balza, E., et al., Targeted delivery of tumor necrosis factor-alpha to tumor vessels induces a therapeutic T cell-mediated immune response that protects the host against syngeneic tumors of different histologic origin. Clin Cancer Res, 2006. 12(8): p. 2575-82.

32. Pretto, F., et al., Preclinical evaluation of IL2-based immunocytokines supports their use in combination with dacarbazine, paclitaxel and TNF- based immunotherapy. Cancer Immunol Immunother, 2014. 63(9): p. 901- 10.

214

33. Hemmerle, T., et al., The antibody-based targeted delivery of TNF in combination with doxorubicin eradicates sarcomas in mice and confers protective immunity. Br J Cancer, 2013. 109(5): p. 1206-13.

34. Spitaleri, G., et al., Phase I/II study of the tumour-targeting human monoclonal antibody-cytokine fusion protein L19-TNF in patients with advanced solid tumours. J Cancer Res Clin Oncol, 2013. 139(3): p. 447- 55.

35. Papadia, F., et al., Isolated limb perfusion with the tumor-targeting human monoclonal antibody-cytokine fusion protein L19-TNF plus melphalan and mild hyperthermia in patients with locally advanced extremity melanoma. J Surg Oncol, 2013. 107(2): p. 173-9.

36. Danielli, R., et al., Intralesional administration of L19-IL2/L19-TNF in stage III or stage IVM1a melanoma patients: results of a phase II study. Cancer Immunol Immunother, 2015. 64(8): p. 999-1009.

37. Cseh, B.F.-S., Samantha ; Grall, Dominique ; et al., Autocrine fibronectin directs matrix assembly and crosstalk between cell–matrix and cell–cell adhesion in vascular endothelial cells. JOURNAL OF CELL SCIENCE, 2010. 123(22): p. 3989-3999.

38. Khan, Z.A., et al., EDB fibronectin and angiogenesis -- a novel mechanistic pathway. Angiogenesis, 2005. 8(3): p. 183-96.

39. Freire-de-Lima, L., et al., Involvement of O-glycosylation defining oncofetal fibronectin in epithelial-mesenchymal transition process. Proc Natl Acad Sci U S A, 2011. 108(43): p. 17690-5.

40. Gujrati, M., A. Vaidya, and Z.R. Lu, Multifunctional pH-Sensitive Amino Lipids for siRNA Delivery. Bioconjug Chem, 2016. 27(1): p. 19-35.

215

41. Bobbin, M.L. and J.J. Rossi, RNA Interference (RNAi)-Based Therapeutics: Delivering on the Promise? Annu Rev Pharmacol Toxicol, 2016. 56: p. 103-22.

42. Sun, Y., et al., Targeted Therapy for Breast Cancer Stem Cells by Liposomal Delivery of siRNA against Fibronectin EDB. Adv Healthc Mater, 2015. 4(11): p. 1675-80.

216