Molecular Mechanisms of Neuropilin-Ligand Binding

University of Kentucky UKnowledge

Theses and Dissertations--Molecular and Cellular Biochemistry Molecular and Cellular Biochemistry

2014

Matthew W. Parker University of Kentucky, [email protected]

Right click to open a feedback form in a new tab to let us know how this document benefits ou.y

Recommended Citation Parker, Matthew W., "Molecular Mechanisms of Neuropilin-Ligand Binding" (2014). Theses and Dissertations--Molecular and Cellular Biochemistry. 15. https://uknowledge.uky.edu/biochem_etds/15

This Doctoral Dissertation is brought to you for free and open access by the Molecular and Cellular Biochemistry at UKnowledge. It has been accepted for inclusion in Theses and Dissertations--Molecular and Cellular Biochemistry by an authorized administrator of UKnowledge. For more information, please contact [email protected]. STUDENT AGREEMENT:

I represent that my thesis or dissertation and abstract are my original work. Proper attribution has been given to all outside sources. I understand that I am solely responsible for obtaining any needed copyright permissions. I have obtained needed written permission statement(s) from the owner(s) of each third-party copyrighted matter to be included in my work, allowing electronic distribution (if such use is not permitted by the fair use doctrine) which will be submitted to UKnowledge as Additional File.

I hereby grant to The University of Kentucky and its agents the irrevocable, non-exclusive, and royalty-free license to archive and make accessible my work in whole or in part in all forms of media, now or hereafter known. I agree that the document mentioned above may be made available immediately for worldwide access unless an embargo applies.

I retain all other ownership rights to the copyright of my work. I also retain the right to use in future works (such as articles or books) all or part of my work. I understand that I am free to register the copyright to my work.

REVIEW, APPROVAL AND ACCEPTANCE

The document mentioned above has been reviewed and accepted by the student’s advisor, on behalf of the advisory committee, and by the Director of Graduate Studies (DGS), on behalf of the program; we verify that this is the final, approved version of the student’s thesis including all changes required by the advisory committee. The undersigned agree to abide by the statements above.

Matthew W. Parker, Student

Dr. Craig W. Vander Kooi, Major Professor

Dr. Michael D. Mendenhall, Director of Graduate Studies

MOLECULAR MECHANISMS OF NEUROPILIN-LIGAND BINDING

DISSERTATION

A dissertation submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy in the College of Medicine at the University of Kentucky

By Matthew W. Parker Lexington, Kentucky

Director: Dr. Craig W. Vander Kooi Associate Professor of Molecular and Cellular Biochemistry Lexington, Kentucky 2014

ABSTRACT OF DISSERTATION

MOLECULAR MECHANISMS OF NEUROPILIN-LIGAND BINDING

Neuropilin (Nrp) is an essential cell surface receptor with dual functionality in the cardiovascular and nervous systems. The first identified Nrp-ligand family was the Semaphorin-3 (Sema3) family of axon repulsion molecules. Subsequently, Nrp was found to serve as a receptor for the vascular endothelial growth factor (VEGF) family of pro-angiogenic cytokines. In addition to its physiological role, VEGF signaling via Nrp directly contributes to cancer stemness, growth, and metastasis. Thus, the Nrp/VEGF signaling axis is a promising anti-cancer therapeutic target. Interestingly, it has recently been shown that Sema3 and VEGF are functionally opposed to one another, with Sema3 possessing potent endogenous anti- angiogenic activity and VEGF serving as an attractive cue for neuronal axons. We hypothesized that direct competition for an overlapping binding site within the Nrp extracellular domain may explain the observed functional competition between VEGF and Sema3. To test this hypothesis we have separately investigated the mechanisms of VEGF and Sema3 binding to Nrp. Utilizing structural biology coupled with biophysics and biochemistry we have identified both distinct and common mechanisms that facilitate the interaction between Nrp and these two ligand families. Specifically, we have identified an Nrp binding pocket to which these ligands competitively bind. The Sema3 family uniquely requires proteolytic activation in order to engage this overlapping binding site. These findings provide critical mechanistic insight into VEGF and Sema3 mediated physiology. Additionally, these data have informed the development of small molecules, peptides, and soluble receptor fragments that function as potent and selective inhibitors of VEGF/Nrp binding with exciting therapeutic potential for treating cancer.

Keywords: Neuropilin (Nrp), Vascular Endothelial Growth Factor (VEGF), Semaphorin-3 (Sema3), angiogenesis, axon guidance

Matthew W. Parker

______Date

MOLECULAR MECHANISMS OF NEUROPILIN-LIGAND BINDING

Matthew W. Parker

Craig W. Vander Kooi Director of Dissertation

Michael Mendenhall Director of Graduate Studies

______Date

ACKNOWLEDGEMENTS

The following dissertation benefited from the insights and direction of many people. First, I am thankful for the mentorship provided by my Dissertation Advisor,

Dr. Craig. Vander Kooi, who created a truly superb environment for my growth as a scientist. Additionally, each member of my Dissertation Committee, Dr. Matthew

Gentry, Dr. Andrew Morris, and Dr. Luke Bradley, has contributed in a unique way to the success of this dissertation. Together they provided invaluable advice in regards to my project and career, reviewed and edited fellowship proposals, provided letters of collaboration, and contributed letters of recommendation on my behalf. The opportunities I have leaving graduate school would not have been possible without the genuine and continuous support of my mentor and committee.

I would also like to thank my outside examiner, Dr. Alan Daugherty, for his willingness to serve in this capacity and also for the career advice he has provided over the past couple of years.

iii TABLE OF CONTENTS

Acknowledgements iii List of Figures vii List of Tables viii Chapter 1: Background and Introduction 1 Nrp Architecture 1 Nrp Function in Sema3 Dependent Axon Guidance 3 Nrp Function in VEGF Dependent Angiogenesis 6 8 Ligand Mediated Cross-Talk 9 Nrp in Tumor Angiogenesis 11 Nrp Tumor Cell Expression and Autocrine Activation 11 General Modes of Nrp Action 12 Nrp Functions by Physically Organizing the Full Signaling Complex 13 Nrp Trafficking 15 The Nrp Intracellular Domain 15 Nrp Coupling with Integrins 17 Scope of Dissertation 20 Chapter 2: Materials and Methods 22 Protein expression and purification 22 Differential scanning fluorimetry (DSF) 24 Structure determination 24 Affinity pull-down 26 Fluorescence Aniosotropy 26 In vitro plate binding assay 27 Isothermal titration calorimetry 28 Western blot analysis of Sema3F furin processing 29 AP-VEGF-A Recovery Assay 30 Circular Dichroism 31 Chapter 3: Furin processing of Sema3F determines its anti-angiogenic activity by regulating direct binding and competition for NRP 32 Introduction 32 Results 35 Sema3F is proteolytically processed in its C-terminal domain 35 C-terminal processing of Sema3F regulates interaction with Nrp 38 Binding of C-furSema to Nrp1 39 Inhibition of VEGF binding 43 Discussion 47 Chapter 4: Effect of C-terminal sequence on competitive Sema3 binding to Nrp1 52 Introduction 52 Results 55

iv C-1 sequence variation critically affects peptide potency 55 Potency is correlated with the propensity of the C-1 residue to adopt an extended conformation 58 Preferential backbone conformation minimizes the entropic cost associated with binding 62 Arginine is conserved at the C-1 position in Sema3F to maintain efficient proteolytic maturation 65 C-1 variation in the Sema3 family alters Nrp1 binding 68 Discussion 72 Chapter 5: Structural basis for nrp2 function as a VEGF-C receptor 75 INTRODUCTION 75 RESULTS 79 Structural basis for proteolytic-dependent VEGF-C binding to Nrp2 79 Characterization of the VEGF-C/Nrp2 interaction 83 Occluding the Nrp2 interloop cleft abolishes binding 86 87 A dimeric soluble Nrp2 splice form 88 s9Nrp2B is a uniquely potent inhibitor of VEGF-C/Nrp2 binding 91 93 DISCUSSION 95 Chapter 6: Structural basis for selective VEGF-A binding to nrp1 99 Introduction 99 Results 102 Physical basis for VEGF-A and Nrp1 binding 102 Isoform specific binding of VEGF-A to Nrp1 105 106 Exon eight encoded residues are essential for high-affinity VEGF-A binding to Nrp1 107 Exon seven residues directly physically engage the L1 loop of Nrp1 111 Selective VEGF-A binding to Nrp1 113 Discussion 117 Chapter 7: Mechanism of selective VEGF-A binding by nrp1 reveals a basis for specific ligand inhibition 122 Introduction 122 Results 126 Soluble Nrp1 selectively inhibits VEGF-A 126 Identification of the regions of Nrp1 that confer selective VEGF-A binding 128 Nrp2Chimera inhibits VEGF-A binding 132 Nrp1 and Nrp2 equivalently inhibit Sema3F binding 134 Nrp2 and Nrp1Chimera relieve C-furSema mediated inhibition of AP-VEGF-A binding to Nrp1 137 Discussion 139

v Chapter 8: Microplate-based screening for small molecule inhibitors of nrp2/VEGF-C interactions 141 Chapter 9: Discussion 151 Future Directions of Nrp Research 151 Nrp Function as an Essential Co-receptor 151 Physical basis for Sema3 function via Nrp receptors 153 Nrp Inhibition in Spinal Cord Injury 155 Physical basis for VEGF function via Nrp receptors 157 Nrp Inhibition in Tumor Angiogenesis 159 References 165 VITA 190

vi LIST OF FIGURES

Figure 1.1: Nrp function is essential for Sema3 dependent axon guidance...... 5 Figure 1.2: Nrp function is essential for VEGF dependent angiogenesis...... 8 Figure 1.3: Cross-talk between Nrp ligands allows coordinated regulation of neuronal and vascular tissues...... 10 Figure 1.4: Nrp utilizes common mechanisms to regulate diverse signaling pathways. .. 19 Figure 3.1: Furin processing of Sema3F...... 36 Figure 3.2: Interaction of the C-terminus of Sema3F with Nrp2 and Nrp1...... 41 Figure 3.3: C-furSema potently inhibits VEGF-A binding to Nrp...... 45 Figure 3.4: Model for the mechanism of anti-angiogenic activity of Sema3F...... 48 Figure 4.1: Inhibitory potency of C-1 peptide variants...... 57 Figure 4.2: The C-1 residue of Nrp1 ligands adopts an extended backbone conformation...... 60 Figure 4.3: Analysis of peptide binding by ITC...... 64 Figure 4.4: Mutation of the Sema3F C-1 residue alters processing efficiency...... 67 Figure 4.5: Furin proteolysis of Sema3 generates divergent C-terminal sequences...... 69 Figure 4.6: Sema3 family variation in potency is explained by the C-1 amino acid...... 71 Figure 5.1: Crystal structure of the VEGF-C/Nrp2 complex reveals the basis for proteolytic-dependent binding...... 82 Figure 5.2: Mechanism of VEGF-C binding to Nrp2...... 85 Figure 5.3: Crystal structure and VEGF-C binding properties of Nrp2-T319R...... 87 B Figure 5.4: s9Nrp2 unexpectedly forms a disulfide-linked dimer...... 90 B Figure 5.5: Crystal structure and inhibitory properties of s9Nrp2 ...... 93 Figure 6.1: The crystal structure of the VEGF-A HBD in complex with Nrp1...... 104 Figure 6.2: VEGF-A164 binds to Nrp1 with high-affinity...... 106 Figure 6.3: The exon eight encoded C-terminal arginine of VEGF-A mediates high- affinity Nrp1 binding...... 109 Figure 6.4: E154 of VEGF-A164 exon seven contributes to Nrp1 binding...... 112 Figure 6.5: Exon seven encoded residues of VEGF-A164 are responsible for Nrp1 binding selectivity...... 115 Figure 6.6: The HBD of VEGF-A is responsible for selective binding to the Nrp1 b1 domain...... 118 Figure 7.1: Nrp1 selectively inhibits VEGF-A binding...... 127 Figure 7.2: Nrp1 residues mediate specific VEGF-A binding...... 130 Figure 7.3: NrpChimera molecules exhibit reversed VEGF-A specificity...... 133 Figure 7.4: Nrp inhibits Sema3F binding through interaction with the C-terminal basic domain...... 136 Figure 7.5: Nrp2 and Nrp1Chimera preferentially sequester Sema3F...... 138 Figure 8.1: Assay optimization...... 145 Figure 8.2: Screen for small molecule inhibitors...... 149

vii LIST OF TABLES

Table 5.1: Data collection and refinement statistics for Chapter 5 ...... 98 Table 6.1: Data collection and refinement statistics for Chapter 6 ...... 121

viii CHAPTER 1: BACKGROUND AND INTRODUCTION

Nrp Architecture

Neuropilin (Nrp) was first identified in the optic tectum of Xenopus and referred to as A5-antigen (Takagi et al., 1991; Takagi et al., 1987). There are two conserved Nrp family members in vertebrates, Nrp1 and Nrp2, which share the same overall domain structure and are 44% identical at the amino acid level (Chen et al., 1997; Kolodkin et al., 1997). Nrp family members are type I transmembrane proteins that possess an N-terminal signal peptide, two Ca2+-binding

C1r/C1s/Uegf/Bmp1 (CUB) domains (a1a2) (Gaboriaud et al., 2011), two coagulation factor V/VIII-like discoidin domains (b1b2) (Fuentes-Prior et al., 2002;

Kiedzierska et al., 2007), a Meprin/A5-antigen/ptp-Mu (MAM) domain (c), a single transmembrane helix, and a short intracellular domain. The N-terminal four extracellular domains are necessary and sufficient for ligand binding. The a1 domain has been shown to interact with the sema domain of Sema3 ligands (Gu et al., 2002). The b1 domain contains a specific C-terminal arginine binding pocket and is essential for binding to the C-terminal domain of both vascular endothelial growth factor (VEGF) and Semaphorin-3 (Sema3) ligands (Parker et al., 2010;

Parker et al., 2012c; Starzec et al., 2007; Teesalu et al., 2009; Vander Kooi et al.,

2007; von Wronski et al., 2006). Additionally, the b1 and b2 domains physically interact to form an extended patch of basic residues that is involved in binding heparin, a highly sulfated member of the glycosaminoglycan (GAG) family of carbohydrates (Vander Kooi et al., 2007). The c domain is dispensable for ligand

1 binding but essential for ligand-dependent signaling (Nakamura et al., 1998). The transmembrane helix possesses a conserved GXXXG repeat and shows strong inherent dimerization potential (Roth et al., 2008). The intracellular domain of Nrp interacts with PSD-95/Dlg/ZO-1 (PDZ)-domain proteins, such as GAIP Interacting

Protein, C-terminus (GIPC), via its C-terminal residues (Cai and Reed, 1999).

Deletion of the Nrp1 intracellular domain does not result in embryonic lethality, as seen with the whole gene, but instead causes defects in vascular patterning

(Fantin et al., 2011).

There are alternative splice forms of Nrp1 and Nrp2 that have been shown to have functional implications. The most commonly identified alternative splice forms of both Nrp1 and Nrp2 result from intron inclusion in the region between b2 and c domains (Gagnon et al., 2000; Rossignol et al., 2000). The included intron contains a stop-codon resulting in production of secreted soluble proteins that preserve ligand binding and thus function as endogenous pathway inhibitors.

Alternative splicing of Nrp2 also produces two species with divergent C-terminal intracellular domains. While multiple splice forms of both Nrp receptors exist, the functional significance of these divergent sequences is largely unknown. In addition to splice variants, there are a number of post-translational modifications that alter Nrp function. Nrp1 and Nrp2 can be modified by both N- and O-linked glycans. Nrp1 has been shown to be modified by heparan- or chondroitin-sulfate which results in increased ligand binding (Shintani et al., 2006). Thus, this modification may poise or pre-activate Nrp for signaling. In contrast, Nrp2 is

2 modified by polysialic acid resulting in enhanced dendritic cell migration (Curreli et al., 2007; Rey-Gallardo et al., 2010).

Nrp Function in Sema3 Dependent Axon Guidance

Nrp was initially identified and functionally characterized as a receptor for the Sema3 family ligands in neurons. Neurons function in cooperative neural networks connected by dendrites and axons. The number and type of axonal connections are tightly regulated by guidance cues. One family of essential axon guidance molecules are the Semaphorins. There are seven Sema3 family members (Sema3A-G) that are secreted, diffuse through tissues, and provide guidance cues (Kolodkin et al., 1993). Indeed, axon guidance cues mediated by

Sema3 family members are essential during development for correct neuronal patterning in dorsal root ganglia, facial, vagal, olfactory-sensory, cortical, hippocampal, and cerebellar nerves, along with others (Koncina et al., 2007).

Sema3 family members bind Nrp receptors on the cell surface (He and

Tessier-Lavigne, 1997; Kolodkin et al., 1997) (Figure 1.1). The N-terminal a1 domain of Nrp1 and Nrp2 selectively binds the sema domain of different Sema3 family members (Chen et al., 1998; Koppel et al., 1997; Merte et al., 2010).

Notably, Sema3A specifically signals via Nrp1 (He and Tessier-Lavigne, 1997;

Kolodkin et al., 1997) and Sema3F via Nrp2 (Giger et al., 1998). The Nrp b1 domain allows high-affinity non-selective binding to the Sema3 C-terminal domain

(Giger et al., 1998; Nakamura et al., 1998; Parker et al., 2010). Following Sema3 binding by Nrp, Plexin family signaling receptors (PlexinA1-4) are then engaged and activated to directly guide axonal growth (Puschel, 2002; Takahashi et al.,

3 1999). Nrp functions by coupling specific high-affinity Sema3 binding to Plexin- dependent regulation of cytoskeleton dynamics in the axonal growth cone

(Takahashi et al., 1999). Collapse of the actin cytoskeleton results in axon repulsion and this appears to represent the most common modality for Nrp- dependent Sema3 function. However, Sema3 signaling can also be attractive rather than repulsive, such as when cGMP levels are high (Song et al., 1998).

Sema3 functionality also drives the pathological repulsion of axons following spinal cord injury where fibroblasts, which invade the glial scar, secrete members of the

Sema3 family and prevent regrowth of axons. Thus, inhibition of the Sema3/Nrp signaling axis represents a promising therapeutic modality for functional recovery following spinal cord injury (De Winter et al., 2002; Fawcett, 2006; Pasterkamp et al., 1999).

4 Semaphorin3 Family Sema3A-G

Sema

PSI

y Plexin Family it ic if Plexin A1-4 c e p S Ig

-S-S- Basic Sema Neuropilin Family ty ni ffi A Nrp1-2 h- ig PSI1 H a2 IPT1 b1 PSI2

I a1 S s P n i

a b2 &

m T o P D c I

GAP

Physiology: Pathology: Neuronal Axon Guidance Spinal Cord Injury

Figure 1.1: Nrp function is essential for Sema3 dependent axon guidance.

Nrps specifically bind to Sema3 family members through a bivalent binding mechanism allowing both specific and high-affinity binding. Engagement and activation of Plexin family receptors activates signaling in both physiological axon guidance and pathological spinal cord injury. Structures of proteins or homologues with known structures were utilized including: Semaphorin N-terminus: PDB=1OLZ

(Love et al., 2003); Nrp N-terminal domains: PDB=2QQK (Appleton et al., 2007);

Plexin extracellular domain N-terminus: PDB=3OKT (Janssen et al., 2010); Plexin intracellular domain: PDB=3IG3 (He et al., 2009).

5 Nrp Function in VEGF Dependent Angiogenesis

Angiogenesis is the process by which new vasculature is formed by branching or splitting from existing vasculature. Physiological angiogenesis is critical during development, for homeostatic maintenance of vasculature, and wound healing. Among the most potent pro-angiogenic cytokine families is the

VEGF family. VEGF-A was the first identified family member (Ferrara and Henzel,

1989; Senger et al., 1983), which also includes VEGF-B, -C, D, and placental growth factor (PlGF) (Ferrara et al., 2003). The different family members have both unique and partially overlapping functions. VEGF-A, -B and PlGF regulate hemangiogenesis by Nrp1 (Migdal et al., 1998; Soker et al., 1998; Zhang et al.,

2009). VEGF-C and -D regulate lymphangiogenesis by Nrp2 (Mattila et al., 2002).

Also, VEGF-B has recently been shown to control lipid transport in endothelial cells by Nrp1 (Hagberg et al., 2010). To add to the complexity, VEGF-A, -B, and PlGF all have numerous alternative splice forms with varying physical and functional properties (Leung et al., 1989; Nowak et al., 2008) whereas VEGF-C and -D require proteolytic processing from a pre-protein for activation (Joukov et al.,

1997). Nrp1 was initially identified as a VEGF-A165 splice-form specific receptor

(Soker et al., 1998). More recent work has demonstrated that Nrp1 can also bind other VEGF-A isoforms (Pan et al., 2007b; von Wronski et al., 2007) yet uniquely and specifically physically engages VEGF-A165 (Parker et al., 2012c).

VEGF-A is secreted by an initiating tissue, typically in response to hypoxia, and diffuses to nearby vessels where it binds to two families of essential endothelial cell-surface receptors: VEGF-receptor (VEGFR) tyrosine kinases

6 (Ferrara et al., 2003) and Nrps (Migdal et al., 1998; Soker et al., 1998) (Figure 1.2).

VEGF-A binding causes initiation of the angiogenic cascade, which involves activation of endothelial cells, basement membrane remodeling, proliferation of endothelial cells, and directional guidance leading to the growth of new vasculature towards the initiating tissue. Importantly, angiogenic signaling is only accomplished through the coordinated activity of VEGF, VEGFR, and Nrp

(Klagsbrun et al., 2002). While VEGF binding to VEGFR weakly activates its intracellular kinase activity, Nrp is required for strong and sustained kinase activation leading to initiation of the pro-angiogenic cascade (Becker et al., 2005;

Soker et al., 2002). Nrp1 null mice die in utero due to defective vasculature formation, thus emphasizing the critical role of Nrp in angiogenesis (Kitsukawa et al., 1997). Importantly, abnormalities associated with loss of Nrp1 are caused by defective endothelial cell migration rather than proliferation (Jones et al., 2008).

7 VEGF Family VEGF-A-D, PlGF

Cystine Knot

VEGFR Family Basic VEGFR1-3

Ig1

ty i y n t i f i f ic f Neuropilin Family A i - c h e Ig2 g p i Nrp1-2 S H & a2 Ig3 b1

Ig4-6 a1 b2 c Ig7

Kinase

Physiology: Pathology: Angiogenesis Tumorigenesis Angiogenesis/Autocrine/Stemness Figure 1.2: Nrp function is essential for VEGF dependent angiogenesis.

Nrps specifically bind to the C-terminal basic domain of VEGF family members.

Cooperative binding of VEGF by VEGFR and Nrp activates the angiogenic cascade necessary for developmental and homeostatic angiogenesis and also pathological signaling associated with tumorigenesis and other types of aberrant signaling. Structures of proteins with known structures were utilized including:

VEGF-A cystine knot domain: PDB=2VPF (Muller et al., 1997); VEGF-A basic domain: PDB=4DEQ (Parker et al., 2012c); Nrp N-terminal domains: PDB=2QQK

(Appleton et al., 2007); VEGFR2 Ig-like domain 2-3: PDB=2X1X (Leppanen et al.,

2010); VEGFR2 Ig-like domain 7: PDB=3KVQ (Yang et al., 2010); VEGFR2 intracellular domain: PDB=1VR2 (McTigue et al., 1999).

8 Ligand Mediated Cross-Talk

Intriguingly, it has been noted that blood vessels and nerves utilize similar signals and principles for guidance (Adams and Eichmann, 2010; Arese et al.,

2011; Carmeliet, 2003). Indeed, recent studies have identified an embryonic stem cell-derived population of cells that differentiate towards either vascular or neuronal cell-type depending on the microenvironment (Gualandris et al., 2009).

Nrp receptors are expressed in nerves and endothelial cells, and both cell types are responsive to secreted ligands of the Sema3 and VEGF families (Erskine et al., 2011; Miao et al., 1999; Schwarz et al., 2004). Not only are both cell types responsive to each family of ligands, but the different ligands can compete for Nrp binding (Figure 1.3).

Figure 1.3: Cross-talk between Nrp ligands allows coordinated regulation of neuronal and vascular tissues.

Both neurons and endothelial cells express Nrp which can respond to either

Sema3 or VEGF family guidance cues. Regulation of competitive Nrp binding between different ligands allows for an additional level of dominant control of Nrp function.

10 Nrp in Tumor Angiogenesis

Nrp function is important not only for physiological angiogenesis but also in

VEGF-dependent pathological hyper-vascularization in tumor angiogenesis (Bagri et al., 2009). It has long been known that the extent of tumor vascularization is correlated to tumor size (Algire and H.W., 1945) and that diffusible factors are able to induce neovascularization in solid tumors (Greenblatt and Shubi, 1968). Anti- angiogenesis therapies (Folkman, 1971) are now used against a range of solid tumors. Avastin, a monoclonal antibody targeting VEGF-A, has been among the most successful angiogenesis inhibitors (Ferrara et al., 2005). While providing significant benefit it has been noted that when patients relapse, those that received anti-angiogenesis treatment often have more aggressive and metastatic tumors

(Ebos et al., 2009; Paez-Ribes et al., 2009). Aberrant Nrp expression and function promotes tumorigenesis and metastasis in vivo in a variety of solid tumors

(Bielenberg et al., 2006; Ellis, 2006; Guttmann-Raviv et al., 2006; Klagsbrun et al.,

2002). Surprisingly, in addition to expression in tumor vasculature, direct Nrp overexpression in malignant cells has been widely reported.

Nrp Tumor Cell Expression and Autocrine Activation

The initial identification of Nrp as a receptor for VEGF noted its expression in both endothelial and tumor cells (Soker et al., 1998). Since then, aberrant Nrp expression in a wide variety of malignant cells has been observed (Banerjee et al.,

2000; Kreuter et al., 2006; Miao et al., 2000; Stephenson et al., 2002; Vacca et al.,

2006). The function of Nrp in tumor cells, and its contribution to tumorigenesis, is the source of considerable interest. Recently, Nrp expression was demonstrated

11 to be critical for autocrine activation of tumor cells, influencing carcinoma survival

(Bachelder et al., 2001), growth (Cao et al., 2012), and migration (Bachelder et al.,

2003; Bagci et al., 2009). Nrp-dependent autocrine activation provides a basis for understanding why Nrp expression and function directly correlates to both tumor aggressiveness (Goel et al., 2013) and the metastatic potential of a variety of solid tumors (Latil et al., 2000; Ochiumi et al., 2006). Further, Nrp function in tumor cells has been tied to dedifferentiation and stemness (Cao et al., 2008). It has recently been shown that Nrp1 directly contributes to the self-renewal of cancer stem cells in skin cancer (Beck et al., 2011). Indeed, Nrp1 deletion blocked the ability of

VEGF to promote cancer stem cell self-renewal. Further, an autocrine activation pathway involving Nrp1 was shown to contribute to stem-like cell viability and tumor growth in glioma (Hamerlik et al., 2012). The multi-functional role for Nrp in tumor initiation and progression has motivated research focusing on developing novel Nrp inhibitory modalities. Additionally, as the genetic profile of various tumors is characterized it will be interesting to determine whether there are specific gain- of-function mutations within the Nrp gene that further contribute to its tumorigenic activity.

General Modes of Nrp Action

Nrps are normally expressed in a variety of tissue types including the previously discussed nerves and vascular tissues along with immune and hematopoietic cells (Tordjman et al., 2002; Yamada et al., 2003). In addition to the critical role for Nrp in the Sema3 and VEGF pathways, interactions with a number of other ligands and receptors have been reported. Other ligands

12 identified include members of the fibroblast growth factor (FGF) family, platelet- derived growth factor (PDGF), transforming growth factor β1 (TGF- β1), and hepatocyte growth factor/scatter factor (HGF/SF) functioning via receptors in the

FGFR, PDGFR, TβRs and c-Met families, respectively (Banerjee et al., 2006; Cao et al., 2010a; Cao et al., 2010b; Glinka and Prud'homme, 2008; Goel et al., 2013;

Holmes and Zachary, 2008; Hu et al., 2007; Matsushita et al., 2007; West et al.,

2005). In light of the diverse expression profile and functional interactions, determining the general mechanism(s) of Nrp action is critical to further our understanding of physiological and pathological functions of Nrp.

Nrp Functions by Physically Organizing the Full Signaling Complex

As a primary function, Nrps promote pathway activation by recruiting ligands to the cell surface through high-affinity interactions (Figure 1.4, A). This is particularly important in Sema3 binding, which is the only diffusible Sema family and requires Nrp for cell-surface binding (Zhou et al., 2008). Additionally, Nrps show selective binding to different members within the VEGF and Sema3 families.

The basis for specific Nrp1-VEGF-A interactions in hemangiogenesis has been described biochemically and structurally (Geretti et al., 2007; Parker et al., 2012b), and other family members are actively being pursued.

Ligand binding is followed by assembly of the active signaling complex, where Nrp functions to promote and stabilize the active signaling holoenzyme

(Shraga-Heled et al., 2007; Soker et al., 2002; Takahashi et al., 1999) (Figure 1.4,

B). Assembly of the active signaling complex requires both receptor recruitment and binding associated conformational changes, providing the physical

13 mechanism required for signal transduction. Importantly, Nrp ligands are dimeric, with conserved intermolecular disulfide bonds. Thus, the minimum signaling holocomplex is thought to involve a dimeric ligand bound to two Nrp molecules and two signaling receptors. Energetically, it appears that highly favorable ligand binding drives the formation of specific homo- and hetero-typic receptor contacts that do not occur spontaneously (Brozzo et al., 2012; Fuh et al., 2000). This is consistent with optimal receptor function, with physical mechanisms strongly inhibiting spontaneous ligand-independent activation.

Understanding Nrp function in holocomplex assembly requires knowledge of the structure and interactions of the receptor both before and after ligand binding. The a2b1b2 domains of Nrp have been reported to adopt a stably associated trimeric core with loosely tethered a1 domain (Appleton et al., 2007).

The c/MAM domain, which is dispensable for ligand binding but essential for ligand-induced signaling (Nakamura et al., 1998), has been reported to form homo- and hetero-dimers between Nrp1 and Nrp2 (Chen et al., 1998). However, experiments with purified protein have reported that there is no self-association or dimerization propensity observable for the Nrp ectodomain, leaving the structure and physical coupling of the MAM domain within Nrp an open question (Appleton et al., 2007).

GAG binding also plays a critical role in Nrp-dependent signaling. GAGs are a diverse family of naturally occurring sulfated polysaccharides that orchestrate numerous processes including the function of endothelial cells in angiogenesis

(Stringer, 2006). GAG binding dramatically enhances Nrp-ligand interactions and

14 drives dimerization of Nrp (Fuh et al., 2000; Vander Kooi et al., 2007). Thus, GAG binding likely plays a dual role by bridging and stabilizing the dimer holocomplex assembly.

Intriguingly, a number of receptors have been shown to associate with Nrp in a ligand independent fashion. A direct and specific interaction has been demonstrated between Nrp1 and VEGFR1 (Fuh et al., 2000). Nrp and Plexin family receptors have also been shown to associate in a ligand independent manner (Takahashi and Strittmatter, 2001). In these cases, rather than having sequential receptor binding, the pre-associated receptor complex is poised to bind and signal and only requires conformational changes (Figure 1.4, C).

Nrp Trafficking

Following ligand binding and assembly of the active signaling complex, Nrp has been demonstrated to have a critical function in receptor trafficking (Figure

1.4, D). Intriguingly, it was shown that Sema3 and VEGF induce Nrp1 endocytosis via distinct pathways (Salikhova et al., 2008). Further, Nrp promotes the partitioning of VEGFR into vesicles that are recycled back to the cell surface, whereas in the absence of Nrp, VEGFR2 is targeted for degradation instead of recycling (Ballmer-Hofer et al., 2011). The intracellular GIPC-binding domain of

Nrp was demonstrated to be essential for the observed VEGFR recycling, suggesting the involvement of a Nrp-associated cytoplasmic protein.

The Nrp Intracellular Domain

The discovery of the Nrp-binding PDZ-protein GIPC provided a mechanism by which Nrp may directly couple to intracellular adaptors and pathways (Cai and

15 Reed, 1999) (Figure 1.4, E). The critical portion of the Nrp intracellular domain is the three C-terminal amino acids, conserved in Nrp1 and Nrp2, that constitute a

PDZ-binding motif and allow binding to PDZ-domain proteins. Increasing evidence implicates the intracellular domain of Nrp in a variety of biological processes. Mice expressing a mutant Nrp receptor with deleted cytoplasmic domain were generated (Fantin et al., 2011). While capable of binding ligand, these Nrp receptors are decoupled from signal transduction via its PDZ-binding motif and

GIPC and result in vascular abnormalities distinct from those of full Nrp1 knockout.

Angiogenesis was unaffected but vascular patterning was disrupted as evidenced by frequent crossing of arteries and veins in the retina. A similar approach was applied in zebrafish, demonstrating that knockdown of Nrp1 does disrupt angiogenesis and only full length Nrp, not a Nrp1 construct lacking the three C- terminal amino acids, is able to rescue disrupted blood vessel growth (Wang et al.,

2006). Further, deletion of the PDZ-binding C-terminal amino acids of Nrp1 disrupted stable association of the full angiogenic signaling complex with VEGFR-

2 (Prahst et al., 2008). There is also growing evidence for Nrp function in the absence of other recognized receptor families. Specifically, Nrp is capable of mediating adhesiveness (Murga et al., 2005) and migration (Wang et al., 2003) of endothelial cells in the absence of VEGFR2, functions that require the C-terminal

PDZ-binding motif of the Nrp intracellular domain.

Interaction with a common intracellular adaptor, such as GIPC, suggests that Nrp function in physiologically diverse pathways may be due to coupling to basic cellular signaling pathways. Intriguingly, the integrin subunits α6 and α5 have

16 been reported to interact with GIPC via their C-terminus (Tani and Mercurio, 2001),

(El Mourabit et al., 2002).

Nrp Coupling with Integrins

Integrins mediate interactions with the extracellular matrix and as a result are important for cellular adhesion and migration. Indeed, the biological function first associated with Nrp1 was its ability to mediate heterophilic cellular adhesion

(Takagi et al., 1995). A variety of studies have now demonstrated that both Nrp1 and Nrp2 physically and functionally associate with integrins (Fukasawa et al.,

2007; Goel et al., 2012; Serini et al., 2003) (Figure 1.4, F).

Exogenous Sema3 was able to inhibit the adhesion of endothelial cells to the integrin ligands fibronectin and vitronectin, while VEGF enhanced the adhesive strength of endothelial cells (Serini et al., 2003). Furthermore, a recent study demonstrated that Nrp1 was necessary for endothelial cell adhesion to fibronectin

(Valdembri et al., 2009). Surprisingly, rescue of fibronectin binding following Nrp1 knockdown was only seen using vectors carrying full-length Nrp but not Nrp constructs lacking the C-terminal PDZ-binding motif, suggesting coupling between integrin and PDZ-domain proteins. αvβ3 integrin has been shown to interact with

Nrp1 in a VEGF-dependent fashion and this serves to sequester Nrp1 from the active VEGF/VEGFR2 signaling complex thereby limiting angiogenesis (Robinson et al., 2009). The function of integrins in controlling cellular migration led to investigation of the interplay between Nrp1 and integrin β1, demonstrating that blockade of either Nrp1 or β1 resulted in reduced invasiveness and adhesion in a pancreatic cancer cell line (Fukasawa et al., 2007). Recently, an important specific

17 role for Nrp2 in integrin-mediated adhesion has also been defined. Nrp2 makes specific interactions with integrin α6β1 at focal adhesion sites. Depleting Nrp2 expression in breast carcinoma cells inhibited focal adhesion assembly of these cells and regulated the downstream targets of α6β1 integrin that ultimately modulate laminin adhesion (Goel et al., 2012). These data demonstrate that Nrp and integrin receptors function cooperatively to regulate cellular adhesion and there is considerable interest in understanding the precise nature of the physical coupling and extent of biological function.

18 Ligand

A. ECM

S Neuropilin i g

i I

n I

n n

t t

e e

R g

C. g

e r

c i

n n

p β

r F.

Endocytic Recycling PDZ Domain

D. E. Lysosomal Degradation

Figure 1.4: Nrp utilizes common mechanisms to regulate diverse signaling pathways.

A) Specific high-affinity ligand binding initiates Nrp mediated signaling. B)

Engagement and organization of an active signaling complex is accomplished through specific receptor-ligand and co-receptor contacts. C) Ligand-independent receptor association forms a pre-activated receptor complex poised for signaling.

D) Nrp regulates dynamic trafficking of signaling complexes. E) Binding of intracellular PDZ-domain proteins, including GIPC, allow direct coupling between

Nrp ligand binding and signaling functions. F) Direct engagement of integrins allows Nrp to couple molecular events with cellular cues.

19 Scope of Dissertation

As previously discussed, the b1 domain of Nrp is responsible for binding the C- terminal domain of both VEGF and Sema3. Three coagulation-factor loops of the b1 domain form a conserved binding pocket specific for C-terminal arginine containing ligands (Lee et al., 2003; Parker et al., 2012c; Vander Kooi et al., 2007).

All VEGF family members possess a C-terminal arginine, which is necessary for binding to Nrp. In contrast, no Sema3 family members possess a C-terminal arginine. However, all Sema3 family members contain furin-like protease RXXR consensus sites in their C-terminal domain (Adams et al., 1997). The interplay of

VEGF and Sema3 is important for diseases associated with angiogenesis.

Aberrant hypo-vascularization was recently demonstrated in the retina where expression of Sema3A prevents revascularization of ischemic neural tissue (Joyal et al., 2011). In contrast, aberrant hyper-vascularization is associated with VEGF upregulation or Sema3 downregulation in pathological angiogenesis of solid tumors (Neufeld et al., 2005; Sakurai et al., 2012; Vacca et al., 2006).

In order to gain an understanding of the complex physiological interplay between VEGF and Sema3, this dissertation aimed to understand both the distinct and overlapping mechanisms utilized by Nrp to bind these two ligand families. In

Chapters 3-4 we describe the molecular basis for Sema3 binding to Nrp, which is regulated by C-terminal proteolysis (Parker et al., 2010; Parker et al., 2013). In

Chapters 5-6 we investigate the physical mechanisms of Nrp1/VEGF-A and

Nrp2/VEGF-C binding using structural biology and biochemistry (Parker et al.,

2012c). Finally, we merge this knowledge in the concluding Chapters (Chapters 7-

20 8) to investigate the use of peptides, soluble receptor fragments, and small molecules as inhibitors of Nrp-ligand binding that hold significant therapeutic promise (Parker and Vander Kooi, 2014; Parker et al., 2012b).

Reprinted in full or part with permission from: Function of Members of the Neuropilin Family as Essential Pleiotropic Cell Surface Receptors. Matthew W. Parker, Hou-Fu Guo, Xiaobo Li, Andrew D. Linkugel, and Craig W. Vander Kooi, Biochemistry 2012 51 (47), 9437-9446. Copyright 2012 American Chemical Society.

21 CHAPTER 2: MATERIALS AND METHODS

Protein expression and purification

Human Nrp2-b1b2 (residues 276-595), human Nrp1-b1b2 (274-584), a fusion of human Nrp1-b1 (274-429) linked to the VEGF-A HBD (115-165) with an intervening Sal1 restriction site and 3X(GS), human Nrp2-b1 (residues 276-430), human Nrp2-T319R (residues 276-595 with T319R mutation), human Nrp1-

B T316R, s9Nrp2 (residues 275-547 plus the unique intron-encoded C-terminal insertion: VGCSWRPL) and the Nrp2-b1b2/VEGF-C fusion were expressed as hexa-histidine fusion proteins from pET28b in E. coli strain Rosetta Gami-2(DE3)

(Novagen/EMD Millipore, Billerica, MA). Nrp2-T319R and Nrp1-T316R were generated by megaprimer mutagenesis (Vander Kooi, 2013). To generate the

Nrp2/VEGF-C fusion, the five C-terminal residues of VEGF-C (223-SIIRR-227) were appended to the C-terminus of Nrp2-b1b2 (276-595) by incorporating the

VEGF-C sequence into the reverse primer during cloning. Likewise, generation of

B s9Nrp2 involved primer incorporation of the sequence VGCSWRPL C-terminal to residues 275-547. All constructs and mutations were sequence verified. Cells were grown in terrific broth and cold shocked at an OD600 = 1.5. Following induction with 1 M IPTG, cells were grown overnight at 16°C. Harvested cells were lysed by sonication and lysozyme treatment and proteins were purified via immobilized metal ion affinity chromatography (IMAC) (HIS-Select HF Nickel Affinity Gel,

Sigma-Aldrich, St. Louis, MO) and eluted with 300 mM Imidazole, 200 mM NaCl.

Nrp2-b1b2, Nrp2-T319R, and the fusion protein were subsequently subjected to heparin affinity chromatography (HiTrap Heparin HP, GE Healthcare Life

22 B Sciences, Pittsburgh, PA). Nrp2-b1 and s9Nrp2 , which were incubated for 48 hours to facilitate production of the disulfide-linked dimer, were further purified by size exclusion chromatography using a Superdex75 HiLoad 16/60 column (GE

Healthcare Life Sciences) equilibrated and run with 20 mM Tris pH 7.5, 100 mM

NaCl.

Alkaline-phosphatase tagged mouse VEGF-A164 (residues 27-190) (AP-VEGF-

A164), AP-VEGF-A164 mutants, human AP-VEGF-A165 (residues 27-191), mouse

AP-VEGF-C (residues 108-223), AP-VEGF-C R223E, human AP-Sema3F

(residues 20-779), and the Ig-Basic domain of human Sema3F (residues 605-779) were expressed in Chinese Hamster Ovary (CHO) cells via large-scale transient transfection from pAPtag-5 vector (GenHunter, Nashville, TN)(Parker et al., 2010).

The C-terminal Nrp binding region of human Sema3F (residues 605-785 (Gu et al., 2002)) was produced as a C-terminal or an N-terminal Human Growth

Hormone (Hgh) fusion from the pLexM vector (Aricescu et al., 2006; Leahy et al.,

2000). Protein was produced from CHO, furin deficient FD11, and furin overexpressing cells (Gordon et al., 1995). Protein was also produced in COS-7 cells in the absence and presence of Dec-RVKR-CMK and D-poly-Arg-NH2 furin inhibitors (Calbiochem, San Diego, CA). Cells were maintained in a-MEM supplemented with 5% FBS. For protein expression, cells were transferred to

Hybridoma-SFM media (Invitrogen, Carlsbad, CA) when they reached 80% confluence and transfected with PEI-MAX (Polysciences Inc., Warrington, PA) using 1mg DNA/mL media and a 3:1 PEI:DNA ratio.

23 Differential scanning fluorimetry (DSF)

Peptides corresponding to processed (RQVHSIIRR) and unprocessed

(RQVHSIIRRSLPA) VEGF-C were produced with an N-terminal tryptophan for quantitation by UV280 absorbance (LifeTein LLC, Hillsborough, NJ). Peptides were resuspended and combined with 2 μM of Nrp2-b1b2 and 5x SYPRO Orange

Protein Gel Stain (Life Technologies, Grand Island, NY) in PBS. Nrp2-b1b2 melting was monitored on a CFX96 Real-Time PCR system (BioRad, Hercules, CA) from

20 to 90°C at a rate of 1°C/50s with fluorescent readings taken every 1°C.

Structure determination

For crystallization, the His-tag was removed from all proteins after IMAC purification by overnight incubation with thrombin. Purified Nrp2-b1b2-VEGF-C fusion in 250 mM NaCl, 10 mM Tris pH 7.5, Nrp2-T319R in 100 mM NaCl, 5 mM

B Tris pH 7.5, and s9Nrp2 in 100 mM NaCl, 20 mM Tris pH 7.5 were concentrated to 2.0 mg/mL, 2.1 mg/mL and 3.5 mg/mL, respectively, and crystals were grown by hanging-drop vapor-diffusion experiments at RT. Crystals of the fusion protein were obtained in two weeks in 0.1 M MES pH 6.5, 0.5 M Ammonium Sulfate. Nrp2-

T319R crystals were obtained in five days in 0.1 M HEPES pH 7, 18% (w/v) PEG

B 12000 and crystals of s9Nrp2 were grown in 10% PEG 1000/10% PEG 8000. The

Nrp-VEGF-A HBD fusion protein was concentrated to 5mg/mL and mixed in a 3:1 ratio of protein to mother liquor containing 1.5M Na/K phosphate pH=6.5. Crystals formed in 1-2 weeks at 23°C. Crystals were passed through mother liquor supplemented with 10% glycerol and then flash frozen in liquid nitrogen. Diffraction data were collected at the SER-CAT 22-ID and 22-BM beamlines of the Advanced

24 Photon Source, Argonne National Laboratories. Data were processed using

HKL2000 (Otwinowski and Minor, 1997) (Table 1).

The structure of the Nrp2-b1b2-VEGF-C fusion and Nrp2-T319R were solved by molecular replacement with wild-type Nrp2-b1b2 (PDB code = 2QQJ) used as a search model. The b1 domain residues of 2QQJ (residues 276-427)

B were used as a search model for the s9Nrp2 molecular replacement solution. An initial molecular replacement solution for the Nrp-VEGF-A HBD fusion protein was obtained using PHASER with Nrp1-b1 (PDB=2QQI) as the search model (McCoy et al., 2007). Clear electron density for the VEGF-A HBD was observed and manually built.

Following molecular replacement, iterative modeling building and refinement was done using COOT (Emsley et al., 2010) and Refmac5 (Murshudov,

1997) to generate a final refined model (Table 1). To characterize intermolecular interactions, the structures were analyzed using the PISA interaction server

(http://www.ebi.ac.uk/ msd-srv/prot_int/pistart.html) and Ligplot+ (Laskowski and

Swindells, 2011). Protein geometry was analyzed using MolProbity (Chen et al.,

2010) and Superpose was used to calculate a difference density matrix (DDM) for the apo-Nrp2-b1b2 and complex structure (Maiti et al., 2004). Graphics were prepared with Pymol molecular graphics software (www.pymol.org).

The atomic coordinates and structure factors have been deposited in the

Protein Data Bank (PDB), www.pdb.org.

25 Affinity pull-down

Nrp2 was coupled to AffiGel (Bio-Rad, Hercules, CA) according to manufacturer’s recommendation at 5 mg protein/mL resin. N-terminally tagged Sema3F Ig-basic was expressed in Cos-7 cells with and without furin inhibitors. 200mL of conditioned media was diluted to 1 mL with Buffer A (20 mM Tris, pH=7.5, 100 mM

NaCl) and incubated with 100 mL of Nrp2 affinity resin for thirty minutes. Resin was washed three times with Buffer A, and then eluted using 1 M NaCl. Eluted protein was resolved using SDS-PAGE and visualized by western blot.

Fluorescence Aniosotropy

C-furSema was synthesized with an N-terminal Fluorescein isothiocyanate (FITC)

(Genscript). The peptide was resuspended and buffer exchanged into Buffer A.

1.8 mM FITC C-furSema was combined with increasing concentrations of Nrp2 and Nrp1. Fluorescence anisotropy was measured at 23C using a SpectraMax

M5 (Molecular Devices, Sunnyvale, CA) with excitation at 485 nm, emission at 525 nm, and an emission filter at 515 nm). Anisotropy was calculated from the average of three independent samples at each point using the experimentally determined

G-factor of 1.113. Dissociation constants (Kd) were calculated by fitting the data with Kaleida-Graph (Synergy Software, Reading, PA) using a single site model:

where ro is the initial anisotropy and ra is the difference in anisotropy between bound and free species.

26 In vitro plate binding assay

Nrp1 affinity plates were produced by diluting Nrp1-b1b2 to either 50, 25, or 2.5 ng/μl in 50 mM Na2CO3 pH 10.4 and 100 μl was added to 96 well protein high- binding plates (Costar, 9018), incubated for 1 hr at 37 °C, washed 5x 100 μl with

PBS-T (PBS, 0.1% Tween20), and stored at 4 °C. Peptides were synthesized using solid phase synthesis and purified to >95% purity. The well characterized

Nrp inhibitory peptide ATWLPPR was used as a positive control (Sigma-Genosys,

St. Louis, MO). Two dimeric disulfide linked peptides of the C-terminal region of

Sema3F were synthesized, oxidized to produce the natural intramolecular disulfide, and purified (Genscript, Piscataway, NJ). One peptide, C-Sema, corresponds to the final 46 residues of Sema3F

(GLIHQYCQGYWRHVPPSPREAPGAPRSPEPDQKKPRNRRHHPPDT) while the second peptide, C-furSema, is 40 residues and corresponds to the furin cleaved species (GLIHQYCQGYWRHVPPSPREA PGAPRSPEPQDQKKPRNRR). Two peptide libraries, one in which the C-1 residue of the S3F C-terminus

(WDQKKPRNXR) was substituted for all twenty amino acids and another that corresponded to the last nine residues of all Sema3 family members, were synthesized and purified by HPLC, with an average purity of 90% (NEO Group,

Cambridge, MA). An N-terminal tryptophan was added to aid in quantitation.

Peptides were resuspended in binding buffer and combined with AP-VEGF-A at peptide concentrations ranging from 500 μM to 230 nM and a final AP-VEGF-A activity of 1 μmol of pNPP hydrolyzed/min per μl. The pre-mixed AP-VEGF-

A/peptide solution was then added to Nrp1 affinity plates and incubated for 1 hr at

27 25 °C. Wells were washed with 100 μl of PBS-T three times using an EL404 plate washer (BioTek, Winooski, VT) and an additional 100 μl of PBS-T was added to the plate and incubated for 5 min. The wash solution was then removed and 100

μl of 1X alkaline phosphatase substrate was added(Jardin et al., 2008). After 25 minutes the reaction was quenched by adding 100 μl of 0.5N NaOH and the plate was read at 405 nm on a SpectraMax M5 plate reader (Molecular Devices,

Sunnyvale, CA). Retained AP-VEGF-A binding was plotted against peptide concentration to generate an inhibition curve that was fit with a four-parameter sigmoidal dose response curve to determine the half-maximal inhibitory concentration (IC50). Results are reported for each peptide as the mean IC50 ± standard deviation of two independent experiments.

For quantitative binding experiments, the absolute concentration of AP-

VEGF-A164 and AP-VEGF-A120 was determined using the quantitative MMV00 enzyme linked immunosorbent assay (ELISA) (R&D Systems). Binding of AP-

VEGF-A164 and AP-VEGF-A120 to Nrp1 or Nrp2 was measured as a function of retained AP activity. Binding curves were fit using a one-site specific binding mode to determine Kd (Graphpad Prism). Kd error is presented as the 95% confidence interval.

Isothermal titration calorimetry

Peptides with sequence WDQKKPRNRR (S3F-RR) and WDQKKPRNPR (S3F-

PR) were synthesized by solid-phase synthesis and purified to >95% purity

(LifeTein, South Plainfield, NJ). Purified Nrp1-b1b2 and the S3F peptides were exhaustively dialyzed against ITC buffer (10 mM phosphate, 238 mM NaCl, 2.7

28 mM KCl, pH 7.4). Following dialysis, Nrp1 was concentrated to 40 μM and S3F peptides were diluted to 390 μM as measured by OD280 (NanoDrop 1000, Thermo

Scientific).

Binding between Nrp1 and the S3F peptides was measured using a VP-ITC

Microcalorimeter (MicroCal, GE Healthcare) and data was processed with Origin software (OriginLab, Northampton, MA). Nrp1-b1b2 was transferred into the sample cell and the syringe was loaded with S3F. Sample measurements were made at 30 °C and the system was set to provide a reference power of 10 μcal/sec.

Thirty 10 μl injections were made, spaced 250 sec. apart. To determine the heat of ligand dilution, S3F-RR and S3F-PR were separately injected into ITC buffer utilizing the same parameters as the experimental runs. The heat of ligand dilution was subtracted from the heat of binding to generate a binding isotherm that was fit with a one-site binding model in Origin, allowing determination of the association constant (Ka), enthalpy (ΔH), and entropy (ΔS) of the interaction. Ka was used to calculate Kd (Kd=1/Ka). ITC data was collected in quadruplicate for each peptide with independent preparations of Nrp1-b1b2.

Western blot analysis of Sema3F furin processing

The Ig-basic domains of human Sema3F (residues 605-785, S3F-Hgh), as well as the single point mutant, R778P (R778P-Hgh), were cloned into the pLexM vector

(Aricescu et al., 2006) for expression with an N-terminal PTP-α signal sequence and C-terminal Hgh fusion (Leahy et al., 2000). Protein was expressed by transient transfection from CHO, furin deficient FD11, and furin overexpressing cells

(Gordon et al., 1995). Western blot analysis was performed on conditioned

29 medium from all cell types to detect the Ig-basic-Hgh fusion protein (unprocessed) or free Hgh (processed) as previously described (Parker et al., 2010). An anti-Hgh polyclonal primary antibody (1:10000 dilution, RD1-HGHabrX1, Fitzgerald

Industries, Acton, MA) and anti-rabbit HRP secondary antibody (1:20000 dilution, sC-2301, Santa Cruz Biotechnology, Santa Cruz, CA) were utilized and the blot was developed using SuperSignal West Pico (Pierce Biotechnology, Rockford, IL).

Image Lab software (Bio-Rad Laboratories, Hercules, CA) was used to calculate the percent of processed Ig-basic-Hgh relative to total protein.

Prism (Graphpad Software, La Jolla, CA) was used for analyzing and graphing data and for calculating the statistical significance between sets of data as determined by a student’s t-test. Molecular graphics were generated using

Pymol (The PyMOL Molecular Graphics System, Version 1.5.0.2 Schrödinger,

LLC).

AP-VEGF-A Recovery Assay

To measure the preferential binding of Nrp1 and Nrp2 to Sema3F or VEGF-A the potential of Nrp to promote the recovery of C-furSema [19] mediated inhibition of

AP-VEGF-A binding to Nrp1 was assessed. Nrp1 and Nrp2 were combined with both AP-VEGF-A (155 μmole p-NPP hydrolyzed/min/L) and C-furSema3F (45nM, the concentration at which ≈90% inhibition is achieved) in incubation buffer and added to low density Nrp1 affinity plates for 1hr at 25C. Plates were washed and retained AP activity quantitated as described above. Retained AP-VEGF-A was reported versus Nrp concentration as the percent of binding observed relative to uninhibited AP-VEGF-A (no C-furSema).

30 Circular Dichroism

Circular dichroism (CD) spectra were collected using a Jasco J-810

Spectropolarimeter. Wild-type and chimeric proteins were added to a 0.1-cm- pathlength cuvette at 10μM in PBS pH 7.4. Spectra were recorded at a speed of

20nm/min using the average of three accumulations over a range of 205-245 nm.

Data are reported as per residue molar ellipticity.

31 CHAPTER 3: FURIN PROCESSING OF SEMA3F DETERMINES ITS ANTI-

ANGIOGENIC ACTIVITY BY REGULATING DIRECT BINDING AND

COMPETITION FOR NRP

Introduction

Vertebrates employ a wide array of secreted growth factors and cell surface receptors to regulate the growth and guidance of axons. The semaphorins represent one of the largest families of cytokines that directly guide axon growth

(Koncina et al., 2007; Zhou et al., 2008). There are five recognized families of semaphorins in vertebrates, including the Sema3 family, all six members of which are secreted and able to diffuse through tissues (Kolodkin et al., 1993). Nrp directly binds to most Sema3 family members and is essential for axonal guidance (He and Tessier-Lavigne, 1997; Nakamura et al., 1998).

Nrp interacts with members of the Sema3 family of ligands and functions together with plexin family receptors in Sema3 mediated axon guidance (Fujisawa,

2004; Takahashi et al., 1999). Nrp also interacts with the VEGF family of ligands and functions together with VEGFR family receptors in VEGF mediated angiogenesis (Klagsbrun et al., 2002; Soker et al., 1998). Higher eukaryotes possess two Nrp family members, Nrp1 and Nrp2, which share 44% amino-acid sequence identity (Kolodkin et al., 1997). They both function in Sema3 and VEGF signaling but differ in their substrate specificity among ligands and receptors, as well as specific control of protein expression and recycling (Chen et al., 1997). In vivo, Sema3F functions via Nrp2 to control axon guidance both in the CNS and peripheral nervous system (Sahay et al., 2003).

32 The coagulation factor domains of Nrp, b1 and b2, contain the high-affinity binding site for both VEGF and the C-terminal domain of Sema3 (Giger et al., 1998;

Mamluk et al., 2002). Because Sema3 and VEGF share an overlapping binding site within the b1 domain of Nrp, the role of Nrp in mediating interplay between the two seemingly distinct pathways of VEGF dependent angiogenesis and Sema3 dependent axon guidance is the source of considerable interest. However, there are conflicting reports as to the role and extent of ligand competition for Nrp binding.

A number of researchers have observed direct competition between VEGF and Sema3 (Gu et al., 2002; Miao et al., 1999; Narazaki and Tosato, 2006). This is consistent with both VEGF and Sema3 families possessing a highly basic C- terminal domain that interacts with the b1 domain of Nrp. Additionally, multiple

Sema3 family members have been shown to have potent anti-angiogenic activity in vivo (Adams et al., 1997; Christensen et al., 2005; Kessler et al., 2004).

Surprisingly, a number of other researchers have recently reported that there is no competition between VEGF and Sema3 (Appleton et al., 2007; Vieira et al., 2007). Previous studies defined the critical importance of a C-terminal arginine residue in the binding of both VEGF and inhibitory peptides to Nrp

(Starzec et al., 2007; Teesalu et al., 2009; Vander Kooi et al., 2007; von Wronski et al., 2006). The observed lack of ligand competition for Nrp is consistent with the fact that no Sema3 family members possess a C-terminal arginine, and it has been suggested that two distinct surfaces in the b1 domain of Nrp may be employed for ligand binding.

33 These conflicting reports suggest that a critical mechanistic feature of Nrp ligand binding is not understood. This has motivated studies to determine the physical mechanism for the interaction of Nrp and Sema3 and the basis for

Sema3’s observed anti-angiogenic activity. The immediate C-terminus of Sema3 is not well conserved and does not contain a C-terminal arginine residue (Figure

3.1, A). However, a region just upstream of the C-termini is, in fact, highly conserved and has been shown to be a cleavage site for the furin family of pro- protein convertases. Proteolytic processing in Sema3’s C-terminal domain has been shown to regulate the anti-angiogenic potency of several Sema3s, which has been suggested to involve Nrp binding (Adams et al., 1997; Christensen et al.,

2005; Potiron et al., 2009). We hypothesized that proteolytic activation of the C- terminus of Sema3 may be critical for regulating interaction with Nrp. We demonstrate that Sema3F is proteolytically processed at its C-terminus. This processing is essential for the production of a mature C-terminal region of Sema3F that can physically interact with Nrp. Further, the mature form of Sema3 potently blocks VEGF binding to Nrp. These data demonstrate that mature Sema3 and

VEGF ligands do compete for binding to the overlapping binding site in the b1 domain of Nrp, but that furin processing of Sema3 is essential for its physical interaction and anti-angiogenic potency. These findings resolve conflicting results in the literature by providing a physical basis for understanding the regulation of

Sema3 interaction with Nrp. Further, these results open new avenues to understand the cross-talk between neuronal and vascular guidance through ligand competition for a shared co-receptor.

34 Results

Sema3F is proteolytically processed in its C-terminal domain

Sema3F possesses an RXRR consensus furin-like protease recognition sequence at its C-terminus. This region is highly conserved in five of the Sema3 family members (Figure 3.1, A). To test whether Sema3F is proteolytically processed in its C-terminus, we expressed the C-terminal Nrp binding region of

Sema3F (Ig-basic) with a C-terminal human growth hormone (Hgh) fusion. The construct was expressed in CHO cells, FD11 CHO cells lacking furin activity, and

CHO cells overexpressing furin (Gordon et al., 1995). Wild-type CHO cells expressed a mixture of processed and unprocessed forms of Sema3F (Figure 3.1,

B). FD11 cells produced solely the unprocessed form of Sema3F, whereas furin overexpressing cells produced only the processed form (Figure 3.1, B). To further confirm that the observed proteolytic processing is the result of furin activity, the construct was expressed in COS-7 cells. The protein was found to be >95% processed (Figure 3.1, C). The observed proteolytic processing is fully blocked by addition of furin inhibitors (Figure 3.1, C). Thus, as has been previously observed with other Sema3 family members (Adams et al., 1997; Christensen et al., 2005),

Sema3F is proteolytically processed in its C-terminal basic domain.

Figure 3.1: Furin processing of Sema3F.

A) Sema3 family members contain a conserved furin recognition sequence in their

C-terminus. B) A mixture of unprocessed (100 kDa) and processed (24 kDa) forms of Sema3F-Hgh fusion are observed when overexpressed in CHO cells. Furin deficient (FD11) cells produce only the unprocessed form, whereas furin overexpressing cells produce only the processed form. C) Sema3F-Hgh expressed in COS-7 cells is nearly completely processed, whereas addition of furin inhibitors produces almost complete reversal to the unprocessed form. Protein

36 molecular weights were estimated using the Magic Mark XP molecular weight standard (MW) (Invitrogen, Carlsbad, CA).

37 C-terminal processing of Sema3F regulates interaction with Nrp

A specific interaction has been shown between the C-terminus of Sema3F and core ligand binding domains (b1b2) of Nrp2 (Giger et al., 1998). To determine the effect of the observed proteolytic processing, we tested the ability of the unprocessed and processed forms of Sema3F to interact with Nrp, utilizing a Nrp2 affinity pull-down of the C-terminus of Sema3F. In order to test the effect of the C- terminal sequence, a construct was produced with a native C-terminal sequence and the Hgh attached to the N-terminus. Furin cleavage would remove only six residues at the C-terminus and so, as expected, no difference in apparent molecular weight is observed between N-terminally tagged protein expressed in

Cos-7 cells in the absence or presence of furin inhibitors. However, a dramatic difference is observed in their ability to interact with Nrp2. The processed form of

Sema3F, expressed from COS-7 cells in the absence of furin inhibitors, shows a robust interaction with the Nrp2 affinity resin (Figure 3.2, A). This result is consistent with previous reports describing the domain specific interaction between the C-terminal Ig-basic domain of Sema3F and Nrp2 (Giger et al., 1998).

In contrast, the unprocessed form, expressed from Cos-7 cells in the presence of furin inhibitors, shows little if any ability to interact with the Nrp2 affinity resin

(Figure 3.2, A). This result suggests that the mechanism underlying the profound physiological effect of furin processing of Sema3 may be direct regulation of the physical interaction between the C-terminus of Sema3 and the b1 domain of Nrp.

To more fully characterize the interaction of the C-terminus of Sema3 with

Nrp, we produced peptides from the C-terminus of Sema3F that include the C-

38 terminal intermolecular disulfide and basic domain. This allowed production of a pure, chemically defined species corresponding to the unprocessed (C-Sema, the final 46 residues of Sema3F) and processed (C-furSema, the same with the final six residues removed, thus possessing a C-terminal arginine) forms of Sema3F.

To quantitatively characterize the interaction of the two proteins, the binding of FITC labeled C-furSema to Nrp2 was measured using fluorescence anisotropy.

Incubation with Nrp2 resulted in a significant increase in anisotropy consistent with a decrease in the rotational diffusion of the bound complex. The observed binding was well fit with a single site binding model (R2=0.99) and allowed determination of the dissociation constant Kd=4.1±0.5 mM (Figure 3.2, B).

Based on these data, we conclude that furin mediated activation of Sema3F is critical for the physical interaction of the C-terminus of Sema3F with Nrp2. This likely represents an important regulatory mechanism since the furin family of pro- protein convertases is differentially expressed in the developing brain (Zheng et al., 1994), suggesting that processing of Sema3 is spatially regulated.

Binding of C-furSema to Nrp1

Sema3F functions with Nrp2 in axon guidance, but functionally blocks

VEGF-A binding to Nrp1 to block angiogenesis. To determine the basis for

Sema3F anti-angiogenic activity, we first tested if C-furSema was also able to bind to Nrp1. The binding of FITC C-furSema to Nrp1 was determined using fluorescence anisotropy as with Nrp2. Binding was again well fit with a single site

2 binding model (R =0.99) and a dissociation constant Kd=2.2±0.2 μM (Figure 3.2,

39 C). Thus, C-furSema can bind both to Nrp2 and Nrp1, the latter with slightly higher affinity.

40 A Furin Inhibitors

Load − Elute Load + Elute

B 0.16

0.15

0.14

0.13

0.12 Aniosotropy 0.11

0.1 K =4.1±0.5mM d 0.09 0 5 10 15 20 [Nrp-2] (uM) C 0.18

0.17

0.16

0.15

0.14

0.13

Aniosotropy 0.12

0.11

0.1 K =2.2±0.2mM d 0.09 0 5 10 15 20 [Nrp-1] (uM)

Figure 3.2: Interaction of the C-terminus of Sema3F with Nrp2 and Nrp1.

A) Affinity purification of the Hgh-Sema3F C-terminus using Nrp2 affinity resin demonstrates that only the protein with a furin processed C-terminus is efficiently

41 pulled down. B) FITC C-furSema shows a significant increase in anisotropy when bound to Nrp2, which is fit well with a single site binding curve. C) Nrp1 also binds to FITC C-furSema with slightly higher affinity.

42 Inhibition of VEGF binding

To test whether the observed anti-angiogenic activity of Sema3F is due to direct competition with VEGF-A for binding to Nrp1, we developed a novel inhibitory assay. This assay measures the ability of Sema3F to compete with

VEGF-A for binding to the core ligand binding domains (b1b2) of Nrp1. Nrp1 was adsorbed to 96-well plates to which AP-tagged VEGF-A binds specifically. VEGF-

A binding could be competitively blocked using anti-angiogenic inhibitory peptides or other blocking reagents. Bound AP-VEGF-A was quantitatively determined using a colorometric p-NPP based assay (Figure 3.3, A).

C-furSema, representing the processed form of Sema3F, was able to fully inhibit the binding of AP-VEGF-A to Nrp. C-furSema was found to be a very potent

2 inhibitor with an IC50=46 ± 3 nM (R =0.9999) (Figure 3.3, B, green). This demonstrates that, in fact, Sema3F and VEGF-A do directly compete for binding to the core ligand binding domains of Nrp, explaining the anti-angiogenic potency of Sema3F in vivo.

To determine the effect of proteolytic activation of Sema3F on anti- angiogenic potency, we tested the ability of C-Sema to competitively block VEGF-

A binding to Nrp. C-Sema, representing the unprocessed form of Sema3F, showed no inhibition of VEGF-A binding even at high concentrations (Figure 3.4,

B, red).

These results underline the essential importance of furin processing of

Sema3, and provide a mechanism for the observed anti-angiogenic potency of

Sema3F. In summary, we demonstrate that furin processing produces a form of

43 Sema3F that binds to the core ligand binding domains of Nrp and directly competes with VEGF-A for receptor binding. Since processing occurs within the

Sema3F basic domain, proteolysis may also regulate Sema3’s ability to interact with the ECM, such as is seen for VEGF-A. Indeed, specific glycosaminoglycans, components of the ECM, have been shown to interact with Sema3, VEGF, and

Nrp and are known to be important mediators of angiogenesis and neuronal axon guidance (Dick et al., 2013; Stringer, 2006; Vander Kooi et al., 2007).

Figure 3.3: C-furSema potently inhibits VEGF-A binding to Nrp.

A) Design of a novel plate based inhibition assay measuring the displacement of

AP-tagged VEGF-A from Nrp1 b1b2 coated plates with increasing concentration of peptide. B) C-furSema potently inhibits the binding of AP-VEGF-A to Nrp1 with an IC50=45nM (green). C-Sema shows no inhibitor potency even at high

45 concentrations (red). Each point is the average of three independent samples with error bars representing +/- one standard deviation.

46 Discussion

We demonstrate that Sema3F is proteolytically processed at its C-terminus.

This processing is essential for the interaction of Sema3F with the core ligand binding domains of Nrp. Our data provides a physical explanation for this, since furin processing liberates a C-terminal arginine. Possession of a C-terminal arginine has been demonstrated to be critical for the interaction of both VEGF and peptide inhibitors with Nrp (Starzec et al., 2007; Vander Kooi et al., 2007). Further, the b1 domain of Nrp is utilized for binding both VEGF and Sema3 families of Nrp ligands, yet the nature of the different ligand interactions with and competition for

Nrp has been unclear. We demonstrate that the two classes of ligands directly compete for Nrp binding, but only when Sema3 is processed. A C-terminal peptide representing the proteolytically processed form of Sema3F potently blocks the binding of VEGF to Nrp, explaining the anti-angiogenic activity of Sema3F (Figure

3.4).

Figure 3.4: Model for the mechanism of anti-angiogenic activity of Sema3F.

Our data demonstrates that furin dependent activation of the C-terminus of

Sema3F is essential for direct interaction with Nrp and anti-angiogenic activity via competition with VEGF-A for binding to the b1 domain of Nrp.

48 Understanding the mechanistic basis for the interaction of the C-terminus of Sema3 with Nrp also provides a simple yet elegant explanation for the divergent literature reports regarding the competition of VEGF and Sema3 ligands for Nrp binding. C-terminal fusions, such as Fc or AP, are often used in the expression and purification of Sema3 family members. This includes commercially available

Sema3s, which are expressed and purified using a C-terminal Fc-fusion (R & D

Systems). These proteins represent solely the unprocessed form of Sema3 and, as expected from our studies, are unable or have dramatically decreased ability to compete with VEGF for binding to Nrp. When using a C-terminal tag, such as AP, for quantitation but not purification, the protein produced will likely be a mixture of the processed and unprocessed forms. When using an N-terminal tag, care should be taken since proteolytic processing of the C-terminus does not produce an appreciable shift in molecular weight and thus a mixture of processed forms will be produced unless furin activity is specifically inhibited or enhanced. It is interesting to note that Sema3F produced in wild-type CHO cells has a larger percentage of unprocessed protein, whereas that produced in COS-7 cells is almost completely processed. In vitro, the ratio of processed to unprocessed Sema3 will be highly dependent on the cell type used to express the protein. In vivo, cell and tissue- specific proteolytic processing of Sema3 family members may well represent an important mechanism controlling the production of anti-angiogenic Sema3s.

The role of proteolytic processing of the C-terminus of Sema3F is the simplest case for the Sema3 family, since it possesses a single furin consensus site in its C-terminus. Other family members have additional furin-like consensus

49 sites. For instance, it has been demonstrated experimentally that Sema3A can be processed at three different sites in its final forty-five residues (Adams et al., 1997).

It will be interesting to determine if proteolytic processing at these different sites produces proteins with differing anti-angiogenic potency.

Paradoxically, it has been shown that furin processing of Sema3B can inactivate its anti-angiogenic potency (Varshavsky et al., 2008). However, the observed cleavage is at a known site in the middle of the Sema3 gene, upstream of the Ig domain, and removes the entire Ig-basic region, including the cysteine that forms the critical intermolecular disulfide essential for Sema3 function

(Kolodkin et al., 1993). Thus, proteolytic processing of Sema3 can either activate or inactivate its anti-angiogenic potency, depending on the site of proteolysis.

The difference between the observed direct binding affinity of C-furSema to

Nrp and its inhibitory potency in blocking VEGF binding to Nrp is of interest. The observed binding affinity of C-furSema for Nrp (2.2 mM) is comparable to those reported for VEGF binding to Nrp1 (2 mM) by surface plasmon resonance with low

Nrp density (Fuh et al., 2000). However, VEGF binding to Nrp has been shown to be highly dependent on Nrp density, with four fold higher density of Nrp leading to a twenty-fold increase in apparent affinity (Fuh et al., 2000). The higher potency of inhibition observed in our plate-based inhibitory assay is consistent with this result, since maximal amounts of Nrp are coupled to the plate. C-furSema’s inhibitory potency in the plate based assay is thus likely due to avidity affects of the dimeric ligand binding to two Nrp molecules, and accurately reproduces the

50 inhibitory potency of previously reported peptides measured both in vitro and in situ.

These data establish Sema3 C-terminal proteolysis as an essential mechanism regulating Sema3 binding to the Nrp coagulation factor domains.

Interestingly, C-terminal proteolysis across the Sema3 family is predicted to generate ligands with divergent C-terminal sequence. Thus, one exciting area for future research is aimed at determining whether furin proteolysis can liberate specific sequence elements upstream of the non-variant C-terminal arginine as a mechanism to modulate Sema3/Nrp affinity.

Reprinted in full or part with permission from: Furin processing of semaphorin 3F determines its anti-angiogenic activity by regulating direct binding and competition for neuropilin. Matthew W. Parker, Lance M. Hellman, Ping Xu, Michael G. Fried, and Craig W. Vander Kooi, Biochemistry 2010 49 (19), 4068- 4075. Copyright 2010 American Chemical Society.

51 CHAPTER 4: EFFECT OF C-TERMINAL SEQUENCE ON COMPETITIVE

SEMA3 BINDING TO NRP1

Introduction

The Nrp family of type I transmembrane receptors coordinate critical signaling events in the cardiovascular and nervous system where they are essential in development, homeostasis, and pathogenesis. The two families of canonical Nrp ligands are the VEGF family of pro-angiogenic cytokines and the

Sema3 family of axon guidance molecules (rev. in (Parker et al., 2012a)). VEGF signaling via Nrp and a VEGFR receptor tyrosine kinase family member is essential for physiological and pathological angiogenesis and plays a causative role in tumorigenesis (rev. in (Carmeliet, 2005; Ellis, 2006)) and wet macular degeneration (Cui et al., 2003; Rosenfeld et al., 2006). Additionally, it has recently been shown that Nrp is critical for autocrine cancer stem cell activation and maintenance (Beck et al., 2011; Goel et al., 2013). Sema3 signaling via Nrp and

Plexin receptors mediates physiological axon guidance and contributes to pathological axon repulsion following CNS injury (rev. in (Hota and Buck, 2012;

Niclou et al., 2006)).

Because of its role in multiple pathological conditions, Nrp represents an attractive therapeutic target. Peptides (Parker et al., 2010; Starzec et al., 2006; von Wronski et al., 2006), peptidomimetics (Jarvis et al., 2010), soluble receptor fragments (Gagnon et al., 2000), and monoclonal antibodies (Appleton et al., 2007;

Liang et al., 2007; Pan et al., 2007a) have all been explored as Nrp binding and blocking molecules. Peptides are the best-studied class of Nrp targeting and

52 modulating molecules and have been developed not only for their competitive ligand binding activity but also for other diverse purposes including in vivo Nrp diagnostic imaging (Benachour et al., 2012; Perret et al., 2004) and for cargo targeting to Nrp-expressing tumors (Roth et al., 2012; Teesalu et al., 2009). Nrp ligand-blocking peptides include sequences derived from endogenous Nrp ligands

(Jia et al., 2006; Parker et al., 2010), the naturally occurring immunostimulatory peptide Tuftsin (von Wronski et al., 2006), and phage-display derived peptides

(Starzec et al., 2006; Teesalu et al., 2009). Initial mechanistic and developmental work has provided critical insights into Nrp ligand binding, but additional insights are needed to produce peptides that are optimized for potency, selectivity, and stability.

Biochemical and structural approaches have demonstrated that all known

Nrp ligands require a C-terminal arginine (CR) for binding to a conserved pocket in the Nrp b1 domain (Delcombel et al., 2013; Parker et al., 2010; Parker et al., 2012c;

Vander Kooi et al., 2007). Alternative splicing generates a CR in many VEGF families, including VEGF-A (rev. in (Harper and Bates, 2008)) and VEGF-B

(Olofsson et al., 1996b), but proteolytic maturation is required in the case of VEGF-

C and VEGF-D (Karpanen et al., 2006). Similarly, Sema3 family members require proteolytic activation by furin-family proteases to liberate a CR (Adams et al., 1997;

Christensen et al., 2005; Varshavsky et al., 2008) that then allows them to function as endogenous competitive inhibitors of Nrp, as demonstrated in Chapter 3 (Gu et al., 2002; Miao et al., 1999; Parker et al., 2010). Indeed, relative levels of VEGF and Sema3 family members have been shown to critically contribute to

53 tumorigenesis (Bender and Mac Gabhann, 2013; Serini et al., 2013). Furin family proteases cleave substrates following an arginine residue (Hosaka et al., 1991).

There are between one and three canonical RXXR furin cleavage sites in the C- terminal basic domain of Sema3 family members, producing Sema3 ligands with alternative forms of the C-terminal domain (Adams et al., 1997).

All known peptide inhibitors of Nrp also contain a CR and target the conserved Nrp-b1 pocket, binding in a mode analogous to that of Nrp ligands (Jia et al., 2006; Starzec et al., 2007; Teesalu et al., 2009). Recently, the structural basis for CR dependent Nrp binding has been described. Crystal structures of the

VEGF-A heparin binding domain (HBD) (Parker et al., 2012c) and Tuftsin (Vander

Kooi et al., 2007) in complex with Nrp1 revealed a shared mode of receptor engagement and have provided critical insight into the physical basis for Nrp binding. Two residues of the ligand contribute to Nrp-b1 binding. The CR is critical for Nrp binding and mediates the majority of the interface via divalent engagement of both the CR side chain and carboxylate at the C-terminus (Parker et al., 2012c).

The third-to-last residue (denoted as residue-“C-2” hereafter) mediates the other interaction, with the C-2 backbone carbonyl forming a hydrogen bond with the aromatic hydroxyl of Nrp1-Y297. This interaction is also critical since mutation of

Y297 results in loss of ligand binding (Herzog et al., 2011). That this interaction critically depends on a backbone hydrogen bond is supported by the observation that for ATWLPPR, Nrp binding is C-2 sequence-independent but truncation smaller than a tetrapeptide eliminated activity (Starzec et al., 2007). While a CR residue is critical for all peptide inhibitors of Nrp, no upstream consensus has been

54 identified. This led us to investigate the sequence-dependence for Nrp-ligand binding and inhibition.

To determine the role of residues upstream of the CR, we studied the sequence dependence of Nrp binding and inhibition of Sema3F derived peptides.

We found that the C-1 residue serves the critical role of positioning the CR and C-

2 residues to allow concurrent Nrp binding. A peptide library with substitution of all 20 natural amino acids at the C-1 position revealed that residues that naturally adopt an extended conformation enhance inhibitory potency by six-fold. A C-1 proline produced the most potent Nrp binding peptide, which we demonstrate is due to a specific reduction in the entropic cost of binding. We further demonstrate that there is an inverse relationship between furin processing of Sema3 and inhibitory potency across the Sema3 family. These data provide critical insight into the mechanism of Nrp ligand binding and potent inhibition, and describe a novel mechanism for regulated Sema3 furin processing and Nrp receptor engagement.

Results

C-1 sequence variation critically affects peptide potency

To determine the contribution of the C-1 residue to Nrp binding and inhibitory potency, a peptide library derived from the C-terminal domain of Sema3F

(WDQKKPRNRR) was synthesized with all twenty natural amino acids substituted at the C-1 position. The library was tested for the ability to inhibit alkaline phosphatase (AP)-tagged VEGF-A binding to Nrp1 utilizing an in vitro plate-binding assay. Retained AP-VEGF-A was measured as a function of peptide concentration. For all peptides, full AP-VEGF-A inhibition was achieved and the

55 concentration of peptide resulting in half AP-VEGF-A inhibition (IC50) was determined. Significant differences in inhibitory potency were observed, with wild type Sema3F (S3F-RR, black line, IC50 = 11 μM) showing intermediate potency between the peptides with maximum and minimum potency, C-1 proline (S3F-PR, green line, IC50 = 4.7 μM) and C-1 aspartate (S3F-DR, red line, IC50 = 28 μM) peptides, respectively (Figure 4.1, A).

Sequence alteration at the C-1 residue resulted in a significant variation in peptide potency (Figure 4.1, B), indicating that this position plays a significant role in modulating Nrp binding. The most potent inhibitor was the C-1 proline peptide, which was six times more potent than the aspartate variant. Analysis of previously published peptide inhibitors reveals a preference for proline at the C-1 position.

Modified Tuftsin (TKPPR) and the phage display derived peptide ATWLPPR possess a C-1 proline. The prototypic CendR peptide identified by phage display,

RPARPAR, does not have this sequence element. However, an analysis of the other CendR peptides that were initially identified reveals that 30% of all peptides contained a C-1 proline, in striking contrast to the 5% theoretical representation if there were no selective pressure at the C-1 position (Haspel et al., 2011).

Collectively these data reveal a significant role for the C-1 position in determining competitive binding to Nrp.

Figure 4.1: Inhibitory potency of C-1 peptide variants.

(A) Peptides were titrated to determine their inhibitory potency versus AP-VEGF-

A. Shown are representative inhibition curves for wild type peptide (S3F-RR, black line) and the variants with maximum and minimum potency, S3F-PR (green line) and S3F-DR (red line), respectively. (B) Table of all C-1 variants with their respective IC50. Values are reported as the mean ± standard deviation

57 Potency is correlated with the propensity of the C-1 residue to adopt an extended conformation

Intriguingly, the identity and chemical properties of the C-1 residue, including charge and hydrophobicity, seemed to have little systematic impact on potency. Structural analysis indicates that the C-1 side-chain is dynamic and shows little if any direct interaction with Nrp. Instead, we hypothesized that the C-

1 residue critically tethers the C-2/Nrp1-Y297 hydrogen bond with the CR mediated salt-bridge and hydrogen bond network. Using structures of Nrp1 complexes, we measured the backbone torsion angles of the C-1 Tuftsin (PDB=2ORZ) and VEGF-

A (PDB=4DEQ) residue to determine whether a specific orientation is required to tether the adjacent CR and C-2 interaction interfaces. Although VEGF-A contains a C-1 arginine and Tuftsin a C-1 proline, residues with markedly different physiochemical properties, their backbone conformations were very similar (Figure

4.2, A). Analysis of the backbone dihedral angles demonstrate that they lie within the extended region of the Ramachandran plot with Φ,Ψ=-89°,108° and -85°,158°, respectively (Figure 4.2, B).

The extended conformation seen for both proline and arginine indicates that residues with an inherent propensity to adopt an extended conformation could, when present at the C-1 position, enhance Nrp binding. Upon analysis of all variants, it was striking to observe that proline and β-branched amino acids

(isoleucine, valine and threonine) produced the most potent C-1 variants (Figure

4.1, B). We examined the relationship between the IC50 for each C-1 variant and the Chou-Fasman β-sheet propensity scale for each amino acid (Fasman, 1989)

58 to determine whether these properties are correlated (Figure 4.2, C). Indeed, potency and β-sheet propensity showed a highly significant correlation (correlation coefficient (r) = -0.74, p = 0.0003). These data support a correlation between competitive potency and preferred C-1 backbone conformation.

Figure 4.2: The C-1 residue of Nrp1 ligands adopts an extended backbone conformation.

(A) Superimposition of the product bound structures of VEGF-A (green)

(PDB=4DEQ) and Tuftsin (gold) (PDB=2ORZ) in complex with Nrp1 (PDB=2ORZ).

There are two distinct interaction interfaces, one of which is mediated by the CR and the other by the C-2 carbonyl of the ligand. The C-1 residue serves the critical role of tethering the adjacent interacting residues and correctly orienting them with respect to one another within the binding pocket. (B) Ramachandran plot of the Φ and Ψ angles of the C-1 VEGF-A (C-1 R) and Tuftsin (C-1 P) residue reveals that they adopt an extended conformation. (C) Plotting the IC50 for each C-1 variant against the β-sheet propensity of each amino acid reveals a correlation between potency and the inherent preference of amino acids to adopt an extended

60 conformation. Proline was excluded from the fit due to its inability to conform to the backbone hydrogen bonding pattern present in β-sheets and, therefore, its low β- sheet propensity.

61 Preferential backbone conformation minimizes the entropic cost associated with binding

Thus, we hypothesized that peptides with a C-1 residue that inherently favors an extended conformation would have increased potency due to decreased entropic cost associated with Nrp1 binding. Therefore, we measured the thermodynamic parameters of S3F-RR (WDQKKPRNRR) and S3F-PR

(WDQKKPRNPR) binding to Nrp1 using isothermal titration calorimetry (ITC).

With Nrp1 in the sample cell, peptides were titrated to saturation and the resulting binding isotherms were fit with a one-site binding model, allowing determination of thermodynamic binding parameters. S3F-RR (Figure 4.3, A) and

S3F-PR (Figure 4.3, B) fit to a 1:1 stoichiometry, consistent with a single CR ligand

4 -1 binding to monomeric Nrp1. S3F-RR bound to Nrp1 with a Ka=3.0 x 10 M (Kd =

4 33 μM) and S3F-PR bound with significantly higher affinity, with a Ka=10.1 x 10

-1 M (Kd = 10 μM). The binding enhancement for S3F-PR relative to S3F-RR is consistent between direct binding and inhibitor potency (3.0-fold and 2.3-fold, respectively). Binding of the less potent wild-type S3F-RR was actually more favorable enthalpically (ΔH = -18.5 and -17.0 kcal/mole, for S3F-RR and S3F-PR, respectively, p = 0.04). However, there was a significantly larger loss of entropy upon S3F-RR binding as compared to S3F-PR binding (TΔS = -12.2 and -10.0 kcal/mole for S3F-RR and S3F-PR, respectively, p = 0.009). Thus, the difference in entropy accounts for the preferential binding potency of S3F-PR.

The measured thermodynamic parameters of binding represent the combined changes for both Nrp1 and the peptide upon binding. No significant

62 differences in the orientation of the bound product in the Nrp b1 binding site (Figure

4.2, A), conformation of the Nrp1 binding pocket itself (R.M.S.D. = 0.6Å for C over residue range 274-429), or protonation state have been indicated. Thus, the difference in entropy between S3F-RR and S3F-PR binding can be best interpreted in terms of conformational bias in the free ligand. Indeed, it was recently suggested that peptide backbone rigidity is able to enhance Nrp binding (Zanuy et al., 2013). These data directly support the hypothesis that proline substitution at the C-1 position enhances Nrp binding by preferentially positioning the free peptide backbone in a ligand bound-like extended conformation that allows engagement of the C-2 and CR interactions. With proline at the C-1 position the backbone is pre-organized for optimal receptor engagement.

Figure 4.3: Analysis of peptide binding by ITC.

Using ITC the thermodynamic parameters of Nrp1 binding were measured for wild type peptide (S3F-RR, A) and the C-1 proline variant (S3F-PR, B). A representative binding isotherm is shown for each peptide. Values are reported as the mean ± standard deviation of four independent experiments.

64 Arginine is conserved at the C-1 position in Sema3F to maintain efficient proteolytic maturation

Although a C-1 proline residue confers an advantage for Nrp1 binding, the majority of furin consensus sequences present in Sema3 family members possess a C-1 arginine residue. This opens the question of why arginine, as opposed to proline, is preferred at the C-1 position of Sema3 family members. The Sema3 family of ligands require processing by furin family proteases for biological function(Adams et al., 1997). This processing event liberates a CR and is required for Nrp1 binding (Parker et al., 2010). We hypothesized that the C-1 residue may directly affect furin-dependent proteolysis.

To test this hypothesis, we expressed the Sema3F Ig-basic domains with a

C-terminal human growth hormone (Hgh) fusion as either the wild type sequence

(S3F-Hgh) or with the C-1 residue mutated to proline (R778P-Hgh) (Figure 4.4, A).

S3F-Hgh and R778P-Hgh were secreted at similar levels, but processing by furin was significantly different (Figure 4.4, B). When expressed in wild type CHO cells, over 40% of S3F-Hgh protein was cleaved but only 10% of R778P-Hgh could be processed. No processing was observed for either S3F-Hgh or R778P-Hgh when expressed in CHO cells deficient in furin activity (FD11) (Gordon et al., 1995), demonstrating that the observed processing is due to furin activity. In contrast, in

CHO cells stably overexpressing furin (Gordon et al., 1995), the majority (70%) of secreted S3F-Hgh was furin processed, while 30% of R778P-Hgh was processed.

These data demonstrate that a C-terminal di-arginine motif allows optimal proteolysis by furin family proteases and that variation at the C-1 position alters

65 proteolytic efficiency. It is notable that R778P-Hgh is partially processed and that this processing increases in cells with higher levels of furin activity. Thus, tissue specific differences in furin activity or differential expression of different proprotein convertase (PPC) family members may serve as a regulatory mechanism underlying activating proteolysis of the different furin sites.

Figure 4.4: Mutation of the Sema3F C-1 residue alters processing efficiency.

(A) The Ig-basic domains of Sema3F were expressed as a C-terminal Hgh fusion.

In addition to wild type protein (S3F-Hgh), a mutant with the C-1 arginine residue of the furin consensus site mutated to proline (R778P-Hgh) was expressed. The arrow indicates the site of furin proteolysis. (B) S3F-Hgh and R778P-Hgh were expressed in CHO, furin deficient (FD11), and furin over expressing (Furin) cells and the efficiency of processing was measured by blotting for Hgh. S3F-Hgh was processed in CHO and Furin cells, as detected by the presence of the processed form, but significantly less processing was seen for R778P-Hgh in these cell types.

Neither construct was processed in FD11 cells.

67 C-1 variation in the Sema3 family alters Nrp1 binding

Sema3F contains a single furin proteolysis site in its C-terminus. This site has a strong furin consensus (KXRXRR) and is conserved in five of the seven

Sema3 family members (Figure 4.5, A, site 2). While it is clear that this extended basic sequence serves as an ideal furin substrate, different Sema3 family members possess up to two other furin sites that have a conserved dibasic RXXR motif (Molloy et al., 1992; Watanabe et al., 1993) and have been demonstrated to be processed (Adams et al., 1997; Christensen et al., 2005; Varshavsky et al.,

2008). For all family members, cleavage at the first and second furin site results in a C-1 arginine, while processing at the third site results in a C-1 P residue (Figure

4.5, B). By aligning the terminal four amino acids of all Sema3 family members, the differential usage of arginine and proline at the C-1 position is clearly illustrated

(Figure 4.5, C) indicating two distinct classes of furin cleavage sites.

Figure 4.5: Furin proteolysis of Sema3 generates divergent C-terminal sequences.

(A) The Sema3 family of ligands are furin processed at multiple sites within their

C-terminus that result in either a C-terminal arginine-arginine motif (site 1 and 2) or a proline-arginine motif (site 3). (B) Peptides corresponding to all furin cleavage sites (black arrow) within the basic domain of Sema3 family members were synthesized with a leading tryptophan (grey). Sema3 variants are labeled according to family member (A-G) and cleavage site number (1-3). (C) An alignment of the four C-terminal residues of all Sema3 peptides illustrating the either-or preference for proline and arginine at the C-1 position. The height of the amino acids at each position represents their relative conservation within the alignment.

69 We tested the VEGF inhibitory potency of peptides representing all forms of each member of the Sema3 family and found very significant differences.

Greater than a 40-fold difference was observed between the different Sema3 family members and an 18-fold difference between different sites in the same

Sema3 family member (Sema3A) (Figure 4.6, A). Sema3E.3 was the most potent inhibitor with an IC50 = 0.76 μM and Sema3A.2 was the weakest inhibitor with an

IC50 = 34 μM. Notably, all peptides with a C-1 proline were more potent than all peptides with a C-1 arginine (Figure 4.6, A). Indeed, when grouped by their C-1 residue (Figure 4.6, B), peptides with a C-1 arginine had a mean potency of 19 ±

11 μM, as opposed to peptides with a C-1 proline that had a mean potency of 2.4

± 1.4 μM (p = 0.02). Otherwise, no correlation between potency and sequence, equivalent to that for the C-1 residue, was observed at other upstream positions.

This does not rule out additional secondary effects, since a range of IC50 values are observed within the two C-1 groups. For instance, it is interesting to note that furin processing at site 2, the most conserved in the Sema3 family, always produces a C-2 asparagine (Figure 4.5, B), and these peptides showed the weakest VEGF inhibitory potency (Figure 4.6, A). A strongly negative preference for aspartate/asparagine is observed in the C-1 position (Figure 4.2, C), and it may be that the sidechain of an asparagine residue in either the C-1 or C-2 position competes with Nrp for mainchain interactions, thereby lowering affinity.

Figure 4.6: Sema3 family variation in potency is explained by the C-1 amino acid.

(A) The ability of all Sema3 ligand variants to inhibit VEGF-A binding to Nrp1 was measured. The data are reported as the mean IC50 ± standard deviation. (B) The

IC50 for each peptide was plotted against the amino acid present at the C-1 position. Peptides with a C-1 proline (average IC50 = 2.4 ± 1.4 μM) were significantly more potent than those with a C-1 arginine (average IC50 = 19 ± 11

μM). (C) The C-1 residue inversely effects furin consensus strength and Nrp affinity.

71 Discussion

Cumulatively, these data demonstrate that there is an inverse relationship between furin processing and Nrp binding. While a C-1 arginine is best suited for furin processing, a C-1 proline is best suited for Nrp1 engagement. Intriguingly, while the majority of furin sites in the different Sema3 family members have a C-1 arginine, those without an arginine exclusively possess a proline residue at this position, suggesting a possible tissue-specific regulatory mechanism of Sema3 activity.

While furin processing is critical, a differential physiological role and how processing at the distinct sites is regulated is currently unknown. These data indicate that there are two physically distinct classes of non-equivalent furin sites in the C-terminal domain of Sema3 family members that are differentiated by their

C-1 residue. Indeed, it has been demonstrated within the context of full-length protein that Sema3A cleaved solely at site 3 (Sema3A.3) shows maximal potency and that when processing at this site is abolished, function is markedly reduced in situ (Adams et al., 1997). Additionally, conformational flexibility may be important for productive assembly of the Sema3 signaling complex, as suggested by the recent structural characterization of the Nrp1/Sema3A/PlexinA2 complex (Janssen et al., 2012). Differential furin processing could modulate the interdomain distance between the Sema3 sema domain and basic domain CR, which engage Nrp a1 and b1 domains, respectively (Parker et al., 2012a; Parker et al., 2012b). It is interesting to note that Sema3A.3 is the most potent sequence as well as the most

C-terminal furin site opening the possibility for coupling between binding site

72 sequence and spacing. Thus, in physiological context, processing at the different sites can produce different amplitude of Sema3 signaling, thereby providing a mechanism for control of Sema3 signaling by regulation of both expression and selective processing.

These data demonstrate that the C-1 position of Nrp binding peptides and ligands is critical for regulating both potency and processing. This provides insight into the mechanism of potent Nrp-binding and inhibition and lays the groundwork for continued improvements to the potency and specificity of peptide inhibitors of

Nrp. Further, by studying the C-1 residue in the context of the C-terminus of

Sema3 we have gained an understanding of physiological Sema3 ligand function.

Interestingly, the Sema3 basic domain contains three non-equivalent furin processing sites defined by possessing either a C-1 Arg or Pro. These data define an inverse relationship between processing and potency, providing an important new mechanism for post-translational regulation of Sema3 activity. It was previously thought that furin processing simply functioned as a binary activation mechanism. However, our data indicates that the coupled relationship between differential furin processing and Nrp engagement allows production of Sema3 forms with a range of biological activities (Figure 4.6, C). This provides insight into the mechanistic basis for functional differences for Sema3 both in situ and in vivo.

Reprinted in full or part with permission from: Effect of C-terminal sequence on competitive semaphorin binding to neuropilin-1. Matthew W. Parker, Andrew D. Linkugel, and Craig W. Vander Kooi, Journal of Molecular Biology 2013 425 (22), 4405-4414. Copyright 2013 Elsevier.

74 CHAPTER 5: STRUCTURAL BASIS FOR NRP2 FUNCTION AS A VEGF-C

RECEPTOR

INTRODUCTION

The VEGF family of cytokines are critical regulators of endothelial cell function. There are five VEGF family members: VEGF-A, -B, -C, -D and placental growth factor (PlGF). Of these five, VEGF-C and VEGF-D selectively control lymphangiogenesis. While they show partially overlapping biological activity and physical properties, VEGF-C is essential for viability while VEGF-D is not (Baldwin et al., 2005; Karkkainen et al., 2004). Endothelial cells of homozygous VEGF-C knockout mice do not sprout to form lymphatic vessels, which results in an alymphatic embryo and embryonic lethality (Karkkainen et al., 2004).

Overexpression of VEGF-C results in selective induction of lymphatic but not vascular endothelial cell proliferation and lymphatic vessel enlargement (Jeltsch et al., 1997). In addition to its critical physiological role, VEGF-C signaling is also important for pathological lymphangiogenesis that is associated with both aberrant loss of function in lymphedema (Saaristo et al., 2002) or gain of function in tumorigenesis and metastasis (Caunt et al., 2008; Ellis, 2006; Stacker et al., 2002).

VEGF-C signals via the coordinated activity of two families of endothelial cell surface receptors, the VEGFR family of receptor tyrosine kinases (RTK) (rev. in (Stuttfeld and Ballmer-Hofer, 2009)) and the Nrp family of co-receptors (rev. in

(Parker et al., 2012a)). VEGF-C function is specifically mediated through VEGFR-

2/3 (Joukov et al., 1996; Kukk et al., 1996; Lymboussaki et al., 1999) and Nrp2

(Karkkainen et al., 2001; Xu et al., 2010), with VEGF-C capable of simultaneously

75 engaging both families of receptors (Favier et al., 2006). VEGFR-2/3 have dual functionality in both angiogenesis and lymphangiogenesis (rev. in (Lohela et al.,

2009)). In contrast, Nrp2 knockout mice display normal angiogenesis but abnormal lymphatic vessel development (Yuan et al., 2002), similar to the tissue specific function observed in the VEGF-C knockout (Karkkainen et al., 2004).

Intriguingly, it has also been demonstrated that Nrp2 can function in VEGF-C signaling independent of its role as a co-receptor for VEGFR (Caunt et al., 2008).

Each member of the VEGF family of ligands is produced in multiple forms by either alternative splicing (e.g. VEGF-A, -B, and PlGF) or proteolytic processing

(e.g. VEGF-C and -D) (Holmes and Zachary, 2005). In all cases, an invariant core cystine-knot domain, which specifically interacts with VEGFR, is combined with a variable C-terminal domain. VEGF-C is synthesized as a pro-protein with N- and

C-terminal domains flanking the central core cystine-knot domain. Prior to secretion, the C-terminal propeptide is cleaved followed by extracellular cleavage of the N-terminus (Joukov et al., 1997). These processing events critically alter both the physiological and pathological bioactivity of VEGF-C. The mature dual processed VEGF-C shows dramatically enhanced stimulatory activity in situ

(Joukov et al., 1997; McColl et al., 2003) and loss of C-terminal processing ablates function in vivo (Khatib et al., 2010). However, the physical basis for the enhanced activity of the mature form of VEGF-C remains unclear and has been connected to different properties including differential receptor binding and interactions with heparin/extracellular matrix (ECM) (Harris et al., 2013; Joukov et al., 1997;

76 Karpanen et al., 2006). The role of VEGF-C proteolytic maturation in regulating

Nrp2 binding is unknown.

The structural basis for VEGF-C binding to VEGFR-2/3 has recently been elucidated and was shown to involve the invariant cystine-knot domain of VEGF-

C binding to the N-terminal domains of VEGFR-2 and VEGFR-3 (Leppanen et al.,

2010; Leppanen et al., 2013). However, the structural basis for VEGF-C binding to Nrp2 remains to be determined. Alternative splicing and proteolysis modify the

C-terminal variable region of VEGF and regulate Nrp binding (Makinen et al., 1999;

Parker et al., 2012c; Soker et al., 1998). It has been demonstrated that Nrp1 binds the C-terminal basic domain of Sema3 and VEGF family of ligands (He and

Tessier-Lavigne, 1997; Soker et al., 1996) utilizing a binding pocket for ligands that contain a C-terminal arginine (Parker et al., 2010; Parker et al., 2012c; Vander

Kooi et al., 2007; von Wronski et al., 2006). As demonstrated in Chapters 3 and 4, the Sema3 family of ligands undergoes furin-dependent proteolytic maturation within their C-terminal domain, a process that liberates an extended basic sequence and directly regulates bioactivity and Nrp binding (Adams et al., 1997;

Parker et al., 2010; Parker et al., 2013).

Nrp2-dependent VEGF-C signaling is important in a variety of tumors and overexpression of these factors is correlated with advanced stage disease and poor prognosis (Ellis, 2006; Stacker et al., 2002). Thus, specific Nrp2/VEGF-C inhibitors are of clinical interest. Soluble receptor fragments are common endogenous inhibitors (Albuquerque et al., 2009; Ambati et al., 2006; Kendall and

Thomas, 1993; Rose-John and Heinrich, 1994). They can also be engineered for

77 use clinically, including VEGF-trap (Aflibercept), a chimeric VEGFR-1/2-Fc fusion, which is an inhibitor of VEGF-A (Holash et al., 2002). A soluble splice form of

Nrp2, s9Nrp2, has been identified at the transcript level (Rossignol et al., 2000). s9Nrp2 is produced by intro inclusion, which contains an in-frame stop codon. This stop codon is located prior to the transmembrane domain, and is thus predicted to produce a secreted form of Nrp2. Interestingly, the insertion occurs in the middle of the second coagulation factor domain (b2), rather than in an interdomain region.

The two Nrp2 coagulation factor domains (b1b2) form an integral unit (Appleton et al., 2007) and thus the nature of the production and function of s9Nrp2 is unclear.

Further, domains b1b2 of Nrp2 have been demonstrated to bind VEGF-C

(Karpanen et al., 2006), bringing into question whether this soluble splice form contains the structural requirements necessary to bind and sequester its ligands.

Here we demonstrate that removal of the VEGF-C C-terminal propeptide directly regulates binding to Nrp2. The structure of the mature VEGF-C C-terminus in complex with Nrp2 demonstrates that a cryptic Nrp2-binding motif is liberated upon C-terminal processing. This offers the first structural insight into the physical basis for VEGF-C binding to Nrp2, showing that the proteolytically liberated C- terminal arginine of VEGF-C directly binds the Nrp2 b1 domain. Mutagenesis of both VEGF-C and Nrp2 confirms the critical nature of the VEGF-C C-terminal sequence in Nrp2-b1 binding. Understanding the physical interactions underlying

VEGF-C/Nrp2 binding led us to consider mechanisms for VEGF-C inhibition. The secreted Nrp2 splice form, s9Nrp2, contains an intact Nrp2 b1 domain but a subsequent stop codon, and we assessed its function as a pathway specific

78 inhibitor. Strikingly, this soluble receptor forms a disulfide-linked dimer with two tightly integrated b1 domains and functions as a potent inhibitor of VEGF-C binding to Nrp2.

RESULTS

Structural basis for proteolytic-dependent VEGF-C binding to Nrp2

VEGF-C is synthesized as a pro-protein with N and C-terminal propeptides.

Removal of the VEGF-C C-terminal propeptide critically regulates its bioactivity.

C-terminal processing of VEGF-C liberates a polypeptide stretch rich in basic amino acids that terminates with a di-arginine sequence (Figure 5.1, A), a structural motif conserved across the VEGF and Sema3 family of ligands and known to be important for Nrp1 binding. Thus, we hypothesized that processing of

VEGF-C may directly regulate a physical interaction with Nrp2. To test this hypothesis, we produced peptides corresponding to the unprocessed (215-

RQVHSIIRRSLPA-227) and processed (215-RQVHSIIRR-223) VEGF-C C- terminus and measured the ability of each peptide to bind Nrp2 domains b1b2 using a differential scanning fluorimetry (DSF) thermal shift assay (Figure 5.1, B).

Processed VEGF-C significantly stabilized Nrp2-b1b2 (Tm 48.8°C ± 0.06°C to

50.3°C ± 0.05°C), while unprocessed VEGF-C showed no effect (Tm 48.4°C ±

0.04°C). Further, the processed VEGF-C peptide showed dose-dependent saturable binding to Nrp2-b1b2 with an apparent dissociation constant (Kd) = 199

μM ± 71 μM (Figure 5.1, C). These data demonstrate that C-terminal proteolytic maturation directly regulates VEGF-C binding to Nrp2.

79 To define the physical basis for proteolytic-dependent binding of VEGF-C to Nrp2, we determined the crystal structure of the processed VEGF-C C-terminus in complex with Nrp2 domains b1b2. The C-terminal five amino acids of mature

VEGF-C (219-SIIRR-223), which are strictly conserved across species and also with VEGF-D, were fused to the C-terminus of human Nrp2 domains b1b2

(residues 276-595). The fusion protein was expressed in E. coli, purified, and crystallized. The structure was solved by molecular replacement and was refined to a resolution of 1.9 Å (Figure 5.1, D, Table 1). Clear electron density for the

VEGF-C-encoded region was observed and manually built. There were two molecules in the asymmetric unit oriented in an anti-parallel fashion. Both molecules demonstrated specific binding of the VEGF-C encoded residues via an intermolecular interaction with a symmetry-related molecule.

Analysis of the structure reveals that VEGF-C (green) engages a binding pocket formed by the Nrp2-b1 (blue) coagulation factor loops (Figure 5.1, D).

Indeed, this interloop cleft uniquely accommodates the C-terminal residue of processed VEGF-C (Figure 5.1, E). The free carboxy-terminus of VEGF-C is integrated into the binding pocket through interactions with residues from the third coagulation factor loop (L3) of Nrp2-b1, which form a wall at one side of the binding pocket (“C-wall”). Specifically, an extensive hydrogen bond network forms between the VEGF-C free C-terminal carboxylate and the side chains of the C-wall residues S349, T352, and Y356 (Figure 5.1, E). Importantly, the position of the C- wall would preclude binding of the unprocessed protein, providing a physical

80 mechanism for the observed proteolytic-dependent binding of VEGF-C to Nrp2- b1b2.

81 A N-Propeptide VEGF-C C-Propeptide D E C-wall VEGF-C L3 SIIRRSLPAT L2 VEGF-C C-terminus B C 51.00 L1 100 No Peptide T352 Unprocessed L5 Nrp2 Processed 75 50.25 b1 S349

50 (°C)

m 49.50 25 T Nrp2 b2 Y356 Normalized RFU 0 48.75

44 46 48 50 52 54 0.0 0.2 0.4 0.6 0.8 1.0 Temperature (°C) [Peptide] (mM) Figure 5.1: Crystal structure of the VEGF-C/Nrp2 complex reveals the basis for proteolytic-dependent binding.

(A) Organization of the VEGF-C pro-protein and site of C-terminal processing

(black arrow). (B) Peptides corresponding to processed (green) and unprocessed

(black) VEGF-C were assayed for the ability to bind Nrp2-b1b2 as measured by differential scanning fluorimetry (DSF) thermal shift assays. Peptides were added to Nrp2-b1b2 to a final concentration of 0.5 mM and melting was monitored between 20 and 90°C. All samples were measured in triplicate and a representative melting curve is shown for each. (C) Processed VEGF-C dose- dependently enhances the Nrp2-b1b2 Tm. Error bars indicate the standard deviation of three measurements. (D) Structure of Nrp2-b1b2 (blue) in complex with the C-terminus of VEGF-C (green). (E) Cross-section of the Nrp2 binding pocket demonstrates that the free carboxy terminus of VEGF-C is buried against the Nrp2 C-wall, which is formed by the third coagulation factor loop.

82 Characterization of the VEGF-C/Nrp2 interaction

The clear electron density present at the Nrp2-b1 interloop cleft permitted modeling of both the VEGF-C polypeptide and interfacing solvent that bridge the two molecules (Figure 5.2, A). Analysis of the VEGF-C/Nrp2 interface reveals direct interactions between VEGF-C and residues within the L1, L5, and L3 loops of Nrp2-b1 (Figure 5.2, B). In addition to the hydrogen bond network formed between the VEGF-C free carboxy-terminus and the Nrp2 L3 loop, the side chain of the VEGF-C C-terminal arginine, R223, forms extensive interactions with the

Nrp2 binding pocket. The gaunidinium of VEGF-C R223 forms a salt-bridge with the Nrp2-b1 L5 loop residue D323. Additionally, the aliphatic portion of the R223 side chain displays extensive van der Waals interactions with two tyrosine residues of Nrp2-b1 that demarcate the sides of the binding pocket, Y299 (L1 loop) and

Y356 (L3 loop). In addition to interactions mediated by VEGF-C R223, there is a hydrogen bond between the backbone carbonyl of I221 and the aromatic hydroxyl of Nrp2 Y299.

While protein-protein binding is primarily mediated by direct interactions between polypeptide chains, interfacing solvent also plays a critical role in stabilizing protein-protein complexes (Janin, 1999; Karplus and Faerman, 1994).

Three water molecules, two of which bridge the interaction between VEGF-C and

Nrp2, are observed in the binding site. One solvent molecule facilitates a water- mediated hydrogen bond between the side chain hydroxyl of Nrp2 T319, located at the base of the binding pocket, and the side chain gaunidinium of VEGF-C R223

(Figure 5.2, B). Likewise, a second solvent molecule bridges the side chain

83 carboxylate of Nrp2 E351 and the free carboxylate of VEGF-C. These solvent- mediated interactions appear to further stabilize the position of the VEGF-C C- terminus within the Nrp2-b1 binding pocket.

To confirm the critical role of the VEGF-C C-terminus we mutated the C- terminal arginine of VEGF-C to glutamate (R223E) and compared the ability of alkaline phosphatase (AP)-tagged VEGF-C and VEGF-C R223E to bind Nrp2- b1b2 affinity plates (Figure 5.2, C). Robust binding was observed between AP-

VEGF-C and Nrp2-b1b2 but R223E binding was reduced by >95%. These data demonstrate that the C-terminal arginine of mature VEGF-C is necessary for high- affinity Nrp2-b1b2 binding, and confirm the importance of C-terminal propeptide processing within VEGF-C to produce a C-terminal arginine that allows avid engagement of Nrp2.

84 A B C R222 100 T352 E351 (%) E351 I221 2.69 75 L3 S349 2.91 S349 Y299 2.51 Y356 50 I220 2.78 2.82 Y356 L1 2.68 25 R223 2.66 T352 VEGF-C Bound 0 3.05 2.75 D323 3.04 C E Y299 - 7 F 2 T319 G 2 L5 E R T319 V D323 T W

Figure 5.2: Mechanism of VEGF-C binding to Nrp2.

(A) Zoom of the intermolecular interface between Nrp2 (blue) and VEGF-C (green) with the 2Fo-Fc electron density map for VEGF-C contoured at 1.0σ. Interfacing water is shown as grey spheres. (B) Ligplot+ generated representation of the interaction between VEGF-C (green) and Nrp2 (blue). Bond distances (Å) are labeled in black and water is shown as grey spheres. (C) Nrp2 binding was compared between VEGF-C and VEGF-C R223E. Binding was analyzed in triplicate for each construct and the data is plotted as the mean ± standard deviation.

85 Occluding the Nrp2 interloop cleft abolishes binding

The structure of VEGF-C in complex with Nrp2 reveals a critical role for the

Nrp2-b1 interloop cleft, which forms the VEGF-C binding pocket. To confirm that the Nrp2 binding pocket is responsible for VEGF-C binding, we carried out site- directed mutagenesis against Nrp2-b1b2 to generate a construct with an occluded binding pocket. Specifically, the residue at the base of the Nrp2-b1 interloop cleft,

T319, was mutated to arginine (Nrp2-T319R). We determined the crystal structure of Nrp2-T319R to a resolution of 2.4 Å (Figure 5.3, A, Table 1). The R319 side- chain showed clear electron density extending into the interloop cleft between the two binding pocket tyrosines, Y299 and Y356 (Figure 5.3, B). Superposing the

VEGF-C/Nrp2 complex onto Nrp2-T319R demonstrates that the binding site occupied by VEGF-C is occluded in the Nrp2 mutant (Figure 5.3, C). The Nrp2-

T319R mutant was then used to analyze the contribution of the interloop cleft to

VEGF-C binding. We compared the binding of VEGF-C to Nrp2-b1b2 and Nrp2-

T319R (Figure 5.3, D). While robust binding was observed between AP-VEGF-C and Nrp2-b1b2, binding to Nrp2-T319R was completely abolished. These data confirm that the interloop cleft, formed by the Nrp2-b1 coagulation factor loops, forms a structure that uniquely accommodates the C-terminus of VEGF-C to mediate binding of the C-terminally processed ligand.

86 A B E351 T352 L3 S349

R319 Y356 L1 Y299

D323 C D 100

(%) 75

VEGF-C Bound 0 2 R p 9 r 1 N 3 T

Figure 5.3: Crystal structure and VEGF-C binding properties of Nrp2-T319R.

(A) Structure of Nrp2-T319R with the stick representation for T319R shown in red.

(B) Zoom of the Nrp2-T319R binding pocket. The blue mesh illustrates the 2Fo-Fc electron density map for R319 contoured at 1.0σ. (C) Superimposition of VEGF-C

(green) onto the structure of Nrp2-T319R demonstrates that the binding pocket normally occupied by VEGF-C is blocked in the mutant. (D) VEGF-C binding was compared between Nrp2-b1b2 and Nrp2-T319R. Binding was analyzed in triplicate for each construct and the data is plotted as the mean ± standard deviation.

87 A dimeric soluble Nrp2 splice form

Based on the specific binding of VEGF-C to the Nrp2 b1 domain, we hypothesized that the previously identified splice form of Nrp2, s9Nrp2, could function as a selective inhibitor of VEGF-C. s9Nrp2 is an alternative Nrp2 splice form that arises from intron inclusion in the b2 domain. An in-frame stop-codon encoded within the intron is predicted to result in termination of translation prior to the transmembrane domain, and thus production of a secreted Nrp2 receptor.

Given that the b1 domain of Nrp2 is solely responsible for VEGF-C binding, we hypothesized that s9Nrp2 may be able to effectively sequester VEGF-C, thereby functioning as an inhibitor. However, it is unknown whether the s9Nrp2 transcript produces a functional protein, since s9Nrp2 retains residues coding only a portion of the b2 domain (114 of 159 residues). Indeed, s9Nrp2 lacks the coding region for three of the eight core β-strands that normally integrate to form the distorted jelly- roll fold that typifies the b1 and b2 domains of Nrp. Additionally, it was unknown whether s9Nrp2 could accommodate the loss of the b2 C-terminal capping cysteine and the introduction of an alternative cysteine encoded by the intronic sequence.

To investigate the physical and functional activity of s9Nrp2, we expressed and

B purified the coagulation factor domain residues of s9Nrp2 (Figure 5.4, A, s9Nrp2 : residues 275-547, plus the unique intron encoded sequence VGCSWRPL).

B B Analysis of s9Nrp2 by reducing SDS-PAGE revealed that purified s9Nrp2 , while running with a larger apparent MW than Nrp2-b1 alone, was smaller than expected from its primary sequence (Figure 5.4, B, observed MW = 22 kDa, expected MW

B = 34 kDa). Mass spectrometry confirmed that s9Nrp2 is an essentially

88 homogeneous single species with MW = 22,775 Da ± 20 Da. These data, together

B with the observed intact N-terminal His-tag, indicate that s9Nrp2 is cleaved C-

B terminal to E457 (predicted MW = 22,792 Da). Thus, the proteolyzed s9Nrp2 contains only a single cysteine residue from the b2 domain (C434), which normally forms an intradomain disulfide. Surprisingly, under non-reducing conditions,

B s9Nrp2 ran with an apparent MW = 38 kDa, indicating the formation of a disulfide- linked intermolecular dimer via the free b2 domain cysteine (Figure 5.4, B). The difference in oligomeric state was evident from size exclusion chromatography

(SEC) (Figure 5.4, C). Nrp2-b1 eluted off SEC with an apparent MW = 16 kDa

B (grey line), while s9Nrp2 had an apparent MW = 38 kDa (black line), consistent with the SDS-PAGE analysis.

B Figure 5.4: s9Nrp2 unexpectedly forms a disulfide-linked dimer.

(A) Domain organization of Nrp2 and the protein fragment utilized for our studies,

B s9Nrp2 . (B) Non-reducing and reducing SDS-PAGE analysis of Nrp2-b1 and

B B s9Nrp2 . (C) The oligomeric state of s9Nrp2 (black line) was analyzed by size- exclusion chromatography. Nrp2-b1 was run as a reference (grey line).

B s9Nrp2 is a uniquely potent inhibitor of VEGF-C/Nrp2 binding

B To understand the structural arrangement of the s9Nrp2 dimer, we

B determined the crystal structure of s9Nrp2 to a resolution of 2.4 Å (Figure 5.5, A,

Table 5.1). Continuous electron density was observed from F275 through S453, consistent with the mass spectrometry defined C-terminus. A single dimer was present in the asymmetric unit, with the base of each b1 domain apposed to the other, thus forming an extended anti-parallel dimer. The orientation of the dimer is stabilized by both the intermolecular disulfide and, unexpectedly, a unique dimeric helical bundle formed by residues from the b1-b2 linker and b2 domain (Figure 5.5,

B). The residues that form this unique helix (residues 428-453) display dramatic structural reorganization relative to that observed in the intact b2-domain where they form an extended sheet and loop motif (Figure 5.5, C). The C-terminal helix runs approximately 20° off parallel from the base of the b1 domain, an angle that is maintained by a cluster of hydrophobic residues at the hinge region between the helix and domain b1. The helix both caps the b1 domain and mediates the

B intermolecular interaction interface with the other monomer of the s9Nrp2 dimer.

The intermolecular interface is composed of both helix-helix interactions, which are mostly hydrophobic in nature (Figure 5B), and helix-b1 interactions, which are mostly hydrophilic in nature.

B The two binding pockets within the s9Nrp2 dimer are positioned 71 Å apart, suggesting that it could simultaneously engage both subunits of the VEGF-C dimer, which is 68 Å wide (Leppanen et al., 2010). Thus, we hypothesized that co-

91 B engagement of both VEGF-C monomers by s9Nrp2 would allow the dimer to function as a uniquely potent inhibitor of VEGF-C/Nrp2 binding. To test this hypothesis, we compared the inhibitory potency of ATWLPPR, an optimized peptide inhibitor of Nrp that functions by competitive binding (Parker and Vander

B Kooi, 2014; Starzec et al., 2007), with Nrp2-b1 and s9Nrp2 , both of which function as soluble competitors through sequestration of VEGF-C (Figure 5.5, D).

ATWLPPR showed dose dependent inhibition of VEGF-C binding to Nrp2 with an

IC50 = 10 M (black line), consistent with its modest reported potency. Next, we examined the ability of Nrp2-b1 to inhibit binding (grey line). Nrp2-b1 sequestered

VEGF-C with improved potency compared to the peptide inhibitor, with an IC50 =

1.5 M. As expected for a monomeric competitive inhibitor, the Hill slope was approximately -1 (ATWLPPR = -1.08 and Nrp2-b1 = -0.97). These data are consistent with independent engagement of each VEGF-C monomer by a single

B Nrp2-b1. Next, we measured the inhibitory potency of s9Nrp2 (orange line).

B Strikingly, s9Nrp2 potently sequestered VEGF-C with an IC50 = 250 nM, a significant improvement in potency from both the peptide inhibitor and Nrp2-b1.

B Additionally, the Hill slope for s9Nrp2 was -1.5. Thus, the enhanced potency of

B s9Nrp2 is due to its ability to synergistically sequester the VEGF-C dimer through simultaneous and cooperative engagement of the two VEGF-C monomers. These

B data demonstrate that s9Nrp2 is a potent inhibitor of VEGF-C and represents an exciting avenue for production of a lymphangiogenic inhibitor

92 A B

C434

L438 90˚

L441

I445 C D ATWLPPR 100 Nrp2-b1

(%) B s9Nrp2 75

BoundVEGF-C 0 s Nrp2B 9 -8 -7 -6 -5 -4 -3 WT Nrp2 10 10 10 10 10 10 [Inhibitor] (M)

B Figure 5.5: Crystal structure and inhibitory properties of s9Nrp2 .

B (A) Crystal structure of the s9Nrp2 dimer (Chain A: light orange; Chain B: dark orange). The Nrp2-b1 binding pockets are labeled with black arrows. (B) Zoom of the dimeric helical bundle with the 2Fo-Fc electron density map contoured at 1σ.

(C) The residues of the Nrp2 b1-b2 linker and b2 domain show a dramatic structural reorganization from an extended loop in the b1b2 sequence (blue) to an

B extended helix in the s9Nrp2 dimer (orange). (D) The inhibitory potency of

B ATWLPPR, Nrp2-b1, and s9Nrp2 against blocking VEGF-C/Nrp2 binding was measured. ATWLPPR inhibited binding with an IC50 50] = -4.98 ±

0.03), Nrp2-b1 inhibited binding with an IC50 50] = -5.82 ± 0.09),

B and s9Nrp2 inhibited binding with an IC50 = 250 nM (log[IC50] = -6.60 ± 0.08).

93 Inhibition was measured in triplicate and the data is plotted as the mean ± standard deviation.

94 DISCUSSION

Structural characterization of the mechanism for VEGF-C binding to Nrp2 represents an important step for understanding the physiological and pathological activity of VEGF-C. These data also inform the rationale design of specific VEGF-

B C/D antagonists, including s9Nrp2 , which potently inhibits VEGF-C/Nrp2 binding and represents a potential therapeutic avenue. Collectively, these results have important implications for interpreting both the aberrant loss and gain of function in the VEGF-C/Nrp2 signaling axis that critically underlies a number of disease states.

With complementary biochemical and structural approaches we show that

VEGF-C C-terminal proteolysis is required for Nrp2 binding. The requirement for proteolytic processing is determined by the position of the Nrp2 C-wall, formed by the L3 coagulation factor loop residues, which specifically engages the VEGF-C free carboxy-terminus, precluding binding of unprocessed protein. These results provide critical insight for interpreting the altered in vitro and in vivo functionality of alternative VEGF-C forms. While both N- and C-terminal processing regulate

VEGF-C activity (Joukov et al., 1997; McColl et al., 2003), processing at these sites is not functionally equivalent. Indeed, loss of C-terminal processing is uniquely detrimental, fully ablating VEGF-C function in vivo (Khatib et al., 2010), which we demonstrate blocks Nrp2 binding.

Determining the structural basis for VEGF-C signaling via Nrp2 informs ongoing studies to describe the effect of signaling deficiency on human disease.

Deficient VEGF-C signaling via Nrp2 has significant implications for both primary

95 and secondary lymphedema. Mutations in both VEGFR-3 (Karkkainen et al.,

2000) and VEGF-C (Gordon et al., 2013) have been demonstrated to underlie hereditary lymphedema and Nrp2 has been identified as an additional candidate gene (Ferrell et al., 2008; Karkkainen et al., 2001) Additionally, both VEGF-C and

Nrp2 have recently been identified as candidate genes for the development of secondary lymphedema following surgery in breast cancer (Miaskowski et al.,

2013). The structural insights gleaned from the VEGF-C/Nrp2 complex also provide an important molecular basis for interpreting emerging exome sequencing data that has identified Nrp2 variants in close proximity to the ligand binding interface. Intriguingly, a stringent examination of exome sequencing data has reported both common and rare Nrp2 variants in human populations (Tennessen et al., 2012). Several of these variants are located in the coagulation factor loops of Nrp2-b1, the region to which VEGF-C binds. Specifically, there are two reported variants in the L5 loop (N321I and L322M) that are located proximal to the critical salt-bridge formed by D323, and two in the L3 loop (Q353H and N354K). The structural data presented here provides a rationale for examining specific coagulation loop variants for loss of function on both a physical and functional level.

These results provide critical insight into the physical basis for VEGF-C binding to its receptor, Nrp2. We demonstrate that the C-terminal residues of

VEGF-C, particularly the VEGF-C C-terminal arginine (R223), engage a binding pocket within the Nrp2-b1 domain. Interestingly, previous studies have implicated a non-C-terminal domain within VEGF-A that is responsible for Nrp1 binding. This

96 motivates studies aimed at understanding the differences between VEGF-A and

VEGF-C engagement of their cognate receptors, Nrp1 and Nrp2, respectively.

97 Table 5.1: Data collection and refinement statistics B Construct: Nrp2-VEGF-C Nrp2-T319R s9Nrp2 Data Collection Beamline APS 22-ID APS 22-BM APS 22-ID Wavelength 1.0000 1.0000 1.0000 Space group P21 P212121 P21212 Cell dimensions (Å) 41.05, 120.81, 34.90, 70.76, 69.36, 91.39, 69.84 122.97 67.33 Cell dimensions (°) 90.0, 103.29, 90.0, 90.0, 90.0 90.0, 90.0, 90.0 90.0 Unique reflections 44,081 12,223 16,303 Completeness (%) 90.6(82.0) 96.4(83.2) 94.1(79.8) Resolution (Å) 1.95(2.02-1.95) 2.40(2.49-2.40) 2.40(2.49-2.40) Rmerge (%) 9.9(46.6) 8.0(29.2) 9.9(32.7) Redundancy 5.1(4.2) 6.8(5.9) 4.4(4.1) I/σ(I) 13.1(3.0) 29.4(5.1) 12.3(3.2) Refinement Resolution limits (Å) 20.00(1.95) 20.00(2.40) 20.00(2.40) No. reflections/no. to compute Rfree 41,511/2140 11,490/586 15,439/821 R(Rfree) 21.0(24.1) 20.1(25.5) 21.0(26.4) No. protein residues 632 313 361 No. solvent/ion molecules 333 123 107 RMSD Bond, Å 0.006 0.008 0.006 RMSD Angle, ° 1.11 1.19 1.04 Protein geometry Ramachandran outlier/favored (%) 0/96.7 0/96.1 0/96.7 Residues with bad bonds/angles 0/0 0/0 0/0 Rotamer outliers 0 0 0 Table 5.1: Data collection and refinement statistics for Chapter 5

98 CHAPTER 6: STRUCTURAL BASIS FOR SELECTIVE VEGF-A BINDING TO

NRP1

Introduction

Nrp1 is essential for VEGF dependent angiogenesis (rev. in (Pellet-Many et al., 2008)). The importance of Nrp1 function in vivo is illustrated by Nrp1 null mice, which show embryonic lethality due to cardiovascular defects (Kawasaki et al.,

1999). In angiogenesis, Nrp1 functions with the VEGFR family of receptor tyrosine kinases. Nrp1 is necessary for high affinity ligand binding to the cell surface and specifically promotes and stabilizes the active angiogenic signaling complex involving VEGF-A, VEGFR-2, and Nrp1 (rev. in (Grunewald et al., 2010)).

The VEGF-A gene is encoded by nine exons. A cystine knot domain, encoded by exons 1-5, is retained in all VEGF-A isoforms. This domain is essential for signaling, mediating homo-dimerization, and direct interaction with VEGFR

(Keyt et al., 1996). Alternative splicing of the remaining introns produces VEGF-A molecules with varying activity, extracellular matrix binding and diffusibility (Houck et al., 1992). It has long been recognized that the most potent stimulator of angiogenesis is VEGF-A164/5, named for the total number of amino acid residues in mouse and human proteins, respectively. VEGF-A164 possesses a heparin binding domain (HBD) encoded by exons seven and eight (Fairbrother et al., 1998;

Leung et al., 1989). It has been demonstrated that Nrp1 binds to the VEGF-A HBD

(Soker et al., 1996) via its b1 coagulation factor domain (Gu et al., 2002; Mamluk et al., 2002; Soker et al., 1998). However, the nature and extent of the interaction is not clear and has been the source of considerable study.

99 It was initially thought that exon seven encoded residues represented the

Nrp-binding region of VEGF-A, thus explaining the significant differences in the biological potency of different VEGF-A isoforms (Soker et al., 1998). In contrast to

VEGF-A164, VEGF-A120 differs by exclusion of exon seven. The clear biological role of exon seven encoded residues in determining the ability of the cytokine to activate endothelial cells has been demonstrated in situ and in vivo (Keyt et al.,

1996; Soker et al., 1997). However, subsequent studies demonstrated that the conserved exon eight encoded C-terminus of VEGF-A controls Nrp binding. All pro-angiogenic VEGF-A isoforms retain exon eight while an inhibitory VEGF splice form, VEGF-A165b, replaces exon eight with exon nine encoded residues (Bates et al., 2002). Nrp binding was isolated to the C-terminal portion of the VEGF-A HBD and a critical role was established for exon eight encoded residues (Jia et al.,

2006). Further, Tuftsin, an immuno-stimulatory peptide mimic of VEGF-A exon eight, was found to inhibit VEGF-A binding to Nrp1, while not affecting VEGF-A binding to VEGFR-2 (von Wronski et al., 2006). It was further shown that Nrp possesses a specific C-terminal arginine binding pocket located in the Nrp1-b1 domain (Vander Kooi et al., 2007). All known Nrp binding proteins and peptides have been shown to posses a C-terminal arginine (Parker et al., 2010; Starzec et al., 2007; Teesalu et al., 2009; Vander Kooi et al., 2007). Indeed, it was convincingly demonstrated that VEGF-A165 and VEGF-A121 both bind Nrp1, although with different kinetics and affinity (Pan et al., 2007b). This has left the physical role of exon seven in receptor binding now unclear.

100 There are two Nrp genes in higher vertebrates, Nrp1 and Nrp2, which share

44% identity in their primary sequence and have the same overall domain organization (Kolodkin et al., 1997). Although Nrp1 and Nrp2 are structurally related, they facilitate activation of functionally distinct pathways utilizing different members of the VEGF family. Nrp1 primarily mediates VEGF-A dependent angiogenesis (Soker et al., 1998) whereas Nrp2 primarily mediates VEGF-C dependent lymphangiogenesis (Xu et al., 2010). Nrps involvement in multiple physiological processes poses the unique challenge of isolating activation events to prevent inadvertent crosstalk. Differential ligand binding and temporal and tissue specific expression are important regulatory mechanisms controlling Nrp function (rev. in (Staton et al., 2007)). Differential ligand binding has been shown to be critical for the specific binding of the VEGF family members to different

VEGFRs. For example, VEGF-A binds to VEGFR-1 and VEGFR-2, but with approximately fifty-fold higher affinity for VEGFR-1 (Waltenberger et al., 1994).

Although VEGF binding to VEGFR has been characterized, the nature and basis for specific ligand binding to Nrp has not been determined.

In order to elucidate the molecular basis for the potent and specific binding of VEGF-A164 to Nrp, we have determined the structure of the VEGF-A HBD bound to Nrp1. This structure reveals an intermolecular interface with contributions from residues encoded by both exons seven and eight. We characterize these interactions and show that the exon eight encoded region determines high-affinity interaction with Nrp1. The exon seven encoded region is found to uniquely physically engage Nrp1 and contribute to binding. Strikingly, the exon seven

101 mediated interaction is shown to determine the selective binding of VEGF-A164 to

Nrp1. These results define the unique physical mechanism underlying VEGF-A binding to Nrp.

Results

Physical basis for VEGF-A and Nrp1 binding

To understand the basis for Nrp binding of VEGF-A, we determined the crystal structure of the VEGF-A HBD in complex with the core ligand binding domain of Nrp1, domain b1. There were two molecules in the asymmetric unit, with intermolecular interactions between the VEGF-A HBD and Nrp-b1 (Figure 6.1, A,

Table 6.1). The VEGF-A HBD of Chain A fully engages the Nrp1-b1 moiety of

Chain B and reveals an extended intermolecular interaction interface between

VEGF-A and Nrp1 (gold) (Figure 6.1, B). The intermolecular interface is formed by both exon seven (blue) and eight (green) encoded residues of the VEGF-A HBD

(Figure 6.1, B). Analysis of the HBD/Nrp1-b1 complex reveals that the interface is predominantly hydrophilic in nature and is stabilized by a network of hydrogen bonds and salt bridges. The HBD retains the overall structural architecture previously determined (Fairbrother et al., 1998), with an R.M.S.D. of 2.0Å

(overlaying residues 115-164). The orientation of the C-terminal peptide-like exon eight encoded residues is significantly different from that observed in solution. This unique orientation is due to the presence of an intramolecular salt-bridge formed between D142 and R163 as well as its direct association with Nrp1 (Figure 6.1, C

& D). The Nrp1 b1 domain shows no significant differences from previously

102 determined structures with a R.M.S.D. of 0.8Å (PDB=2QQI, overlaying residues

274-429).

103 A. Chain A: B. C. VEGF-A Chain B: Nrp1-b1 linker Exon 8 C R163 D142 D320 R164

N Exon 7

Chain A: N D. Nrp1-b1

linker

Chain B: Nrp1-b1 VEGF-A C Figure 6.1: The crystal structure of the VEGF-A HBD in complex with Nrp1.

(A) Chain A (green) and Chain B (blue) crystallized in an anti-parallel fashion with the Chain A VEGF-A HBD fully engaging the Nrp1-b1 domain of Chain B, and that of Chain B engaged by the symmetry related Nrp1-b1 domain of Chain A. The intermolecular complex enclosed in the dashed box is shown in (B) A space-filling model revealing the specific interface with VEGF-A164 exon seven and eight (Chain

A) encoded residues with Nrp1 (Chain B). (C) A stereo view of the 2FO-FC electron density map contoured at 1.0 sigma of Nrp1 (gold) and exon seven (blue) and eight

(green) of VEGF-A164. An intramolecular salt bridge between D142 and R163 of

VEGF-A164 and intermolecular salt-bridge between D320 of Nrp1 and R164 of

VEGF-A164 are observable. (D) A stereo view of the Fo-Fc omit electron density map for the HBD residues contoured at 3.0 sigma

104

Isoform specific binding of VEGF-A to Nrp1

Differences in potency, functionality, and receptor binding have been reported for VEGF-A164 and VEGF-A120, which differ only in the inclusion of exon seven in VEGF-A164. To delineate the different roles of these exons in Nrp binding, the dose dependent binding of AP-VEGF-A164 and AP-VEGF-A120 to Nrp1 was analyzed (Figure 6.2). VEGF-A164 (black line) bound Nrp1 with high affinity,

Kd=3.0nM±0.2nM. As expected for high density coupling of Nrp1, the observed binding is consistent with the reported tight cell surface binding (Soker et al., 1998).

VEGF-A120 (gray dashed line) also bound Nrp1, but with lower affinity

(Kd=22nM±1nM). These data indicate that both exon eight containing isoforms of

VEGF-A are able to bind to Nrp1, but differ in their affinity.

105 300 VEGF-A164

VEGF-A120

200

100 Retained AP-VEGF-A Activity AP-VEGF-A Retained

hydrolyzed/min/µL) (µmole p-NPP

0 0 1 2 3 4 5 6 7 8 9 10 11 [VEGF-A] (nM)

Figure 6.2: VEGF-A164 binds to Nrp1 with high-affinity.

VEGF-A164 (black line) binds Nrp1 with a Kd=3.0nM±0.2nM. VEGF-A120 (gray dashed line) binds Nrp1 with a Kd=22nM±1nM.

106 Exon eight encoded residues are essential for high-affinity VEGF-A binding to Nrp1

Close examination of the residues of exon eight that interact with Nrp1 reveals that VEGF-A164 utilizes a C-terminal arginine (R164) to bind Nrp1 (Figure

6.3, A). Analysis of the intermolecular interface in this region indicates that R164 contributes a majority of the interaction with Nrp1. While the electron density associated with the side-chain indicates some disorder (Figure 6.1, C & D), the binding mode is seen in both molecules in the asymmetric unit, and is very similar to the previously determined structure with the inhibitory peptide Tuftsin (TKPR), a peptide mimic of VEGF-A exon eight (Figure 6.3, B). R164 is fully engaged by b1, burying 247 Å2 in the binding pocket. A salt-bridge is formed with D320 and the free carboxy-terminus forms hydrogen bonds with S346, T349, and Y353 of the b1 domain (Figure 6.3, A).

To characterize the contribution of the exon eight encoded C-terminal arginine to binding, we analyzed site-directed mutants of both VEGF-A and Nrp1.

To determine the role of the salt-bridge, the C-terminal arginine was first mutated to alanine (Figure 6.3, C). Retention of R164A by Nrp1 (black bar) was reduced by 97% relative to WT VEGF-A164. To determine the role of the carboxy-terminal hydrogen bond network, a VEGF-A164 construct with a C-terminal di-alanine addition was generated. Retention of R164R+AA by Nrp1 (gray bar) was reduced by 87% relative to WT VEGF-A164. The observed contributions from both side- chain and carboxy-terminus emphasize the unique requirement for a carboxy- terminal arginine in high affinity Nrp ligands. Lastly, a charge reversal was

107 produced. R164E (red bar) showed no significant binding to Nrp1. These data demonstrate the essential role of a carboxy-terminal arginine in VEGF.

To complement these results we examined the amount of VEGF-A164 binding retained when the Nrp1 binding pocket was occluded. T316, which sits at the base of the Nrp1 binding pocket adjacent to D320 (Figure 6.3, D), was mutated to arginine to generate a binding-deficient Nrp1 mutant. The binding of AP-Nrp1 or AP-T316R to VEGF-A164 was determined (Figure 6.3, E). Strikingly, occlusion of the Nrp1-b1 binding pocket in AP-T316R completely abolished binding with

VEGF-A164. These data demonstrate the essential role of the Nrp1-b1 C-terminal arginine binding pocket in mediating high-affinity VEGF-A binding.

108 A. B.

T349 R164

S346

D320 Y353

C. 100 D. E. VEGF-A 164 100 R164R+AA R164A WT Nrp1

) 15 R164E T316R Activity Activity 164 75 164 164

10 50 (% of WT Nrp1) WT of (% (% of WT VEGF-A WT (%of 5 25 Retained AP-Nrp1 Activity Retained Retained AP-VEGF-A Retained

0 0 Figure 6.3: The exon eight encoded C-terminal arginine of VEGF-A mediates high-affinity Nrp1 binding.

(A) R164 forms specific contacts with the b1 binding pocket of Nrp1. The guanidinium moiety forms a salt-bridge with D320 carboxylate oxygen’s (dashed red lines, 3.08Å and 3.32Å). The carboxy-terminus forms hydrogen bonds with three Nrp1-b1 residues (dashed gray lines, 3.08Å, 2.95Å, and 3.13Å to S346,

T349, and Y353, respectively). (B) Tuftsin binds to the Nrp1-b1 domain

(PDB=2ORZ) utilizing the same C-terminal arginine binding mode. (C)

Mutagenesis demonstrates a specific role for the side chain (R164A, black bar) and carboxy-terminus (R164R+AA, gray bar) in Nrp1 binding. Charge reversal

(R164E, red bar) completely abolishes binding to Nrp1. Statistical comparison of mean wild-type and mutant binding demonstrates significant differences,

109 p<0.0002, between all constructs. AP tagged wild-type and mutant VEGF was used at an activity of 25 μmole p-NPP hydrolyzed/min/μL. (D) A surface representation of VEGF-A bound to Nrp1 reveals the critical location of T316 (red) and illustrates the mechanism by which mutation to arginine would occlude binding. (E) Occlusion of the Nrp1 binding pocket in the Thr-Arg Nrp1 mutant

(T316R, red bar) completely abolishes binding to VEGF-A164. AP-tagged wild-type and mutant Nrp1 was used at an activity of 1 μmole p-NPP hydrolyzed/min/μL.

110 Exon seven residues directly physically engage the L1 loop of Nrp1

The enhanced affinity of VEGF-A164 for Nrp1 versus that of VEGF-A120 suggests a role for the exon seven encoded residues in the interaction with Nrp1.

The reported structure reveals that specific exon seven encoded residues also directly engage Nrp1. The interface with exon seven encoded residues is more extended and involves K146, E151, and E154. The residue with the largest interface contribution is E154, with 67 Å2 buried surface area at the interface. The side chain of E154 forms a hydrogen bond with both the backbone amide and side- chain hydroxyl of T299 in the Nrp1 L1 loop (Figure 6.4, A). The role of E154 in Nrp binding was examined (Figure 6.4, B). An E154A (purple bar) mutant showed reduced Nrp1 binding, but the reduction was not as pronounced as that observed for R164 mutants.

111 A. B. VEGF-A 100 164 E154A

y t i v

t 75 c ) A

164 164 A T299 - F

G 50 E V - E154 P A

d e (% of WT VEGF-A (%WT of n

i 25 a t e R

Figure 6.4: E154 of VEGF-A164 exon seven contributes to Nrp1 binding.

(A) E154 interacts with the side chain hydroxyl (bond distance=2.73Å) and backbone amide (bond distance=3.16Å) of T299 of the Nrp1 L1 loop. (B) Mutation of E154 to alanine (E154A, purple bar) reduces binding to Nrp1 (p<0.0004). AP tagged wild-type and mutant VEGF was used at an activity of 25 μmole p-NPP hydrolyzed/min/μL.

112

Selective VEGF-A binding to Nrp1

Nrp1 and Nrp2 b1 domains are structurally highly homologous (Figure 6.5,

A). However, there are regions that are divergent. The L1 loops of Nrp1 and Nrp2 have distinct amino acid composition: the Nrp1 L1 loop T299 is replaced by D301 in the Nrp2 L1 loop. This replacement would be expected to result in electrostatic repulsion of E154 of VEGF-A164. We generated a Nrp1 L1 loop chimeric mutant, replacing the L1 loop of Nrp1 (299-TN-300) with the Nrp2 loop (301-DGR-303), and assessed its ability to bind VEGF-A164 (Figure 6.5, B). This mutant shows a

75% reduction in its ability to bind VEGF-A164 (gold/cyan bar) relative to WT Nrp1

(gold bar). These data directly demonstrate that the VEGF-A164 E154 interaction with the L1 loop of Nrp1 contributes to binding.

As observed, substitution of the L1 loop of Nrp2 into Nrp1 dramatically reduces Nrp1 binding to VEGF-A164. This led us to consider whether the physical interaction we describe may be directly reflected in decreased affinity of VEGF-

A164 for Nrp2. To test this, we assayed the binding of VEGF-A164 to Nrp2. In fact,

VEGF-A164 shows dramatically weaker binding to Nrp2, with approximately fifty- fold lower affinity (Figure 6.5, C, black line, Kd=150nM±4nM) relative to Nrp1

(Figure 6.2, black line, Kd=3nM). Our data suggests that the observed binding selectivity may be due to the exon seven encoded residues. To test this, we compared the binding of VEGF-A120 to the two Nrp receptors. Indeed, binding of

VEGF-A120 to Nrp2 is unchanged (Figure 6.5, C, gray dashed line, Kd=23nM±1nM) from that of Nrp1 (Figure 6.2, gray dashed line, Kd=22nM).

113 To confirm that the observed binding selectivity involves electrostatic repulsion between E154 and Nrp2, we tested the binding of E154A to Nrp2 (Figure

6.5, D). Indeed, E154A (purple bar) shows three-fold higher retention by Nrp2 relative to WT VEGF-A164 (blue).

114 A. L1 B. Nrp1: YST-NWS 100 Cons: YS W+ Nrp2: YSDGRWT Nrp1 75 TN-DGR

50 (%of WT Nrp1)

25 Retained AP-Nrp1 Activity AP-Nrp1 Retained

C. 25 D. 400 VEGF-A 164 VEGF-A164 20 VEGF-A120 E154A 300 ) Activity 164 15 164

200 10

(% of WT VEGF-A WT (% of 100 5 Retained AP-VEGF-A Activity AP-VEGF-A Retained (µmole p-NPP hydrolyzed/min/µL)(µmolep-NPP Retained AP-VEGF-A Retained

0 0 0 12 345 67 891011 [VEGF-A] (nM)

Figure 6.5: Exon seven encoded residues of VEGF-A164 are responsible for

Nrp1 binding selectivity.

(A) Superimposition of Nrp1 (PDB=1KEX) and Nrp2 b1 domains (PDB=2QQJ, residues 276-427) reveals a similar overall architecture, R.M.S.D.=0.7Å, but unique amino acid composition of the L1 loop, with Nrp1-T299 and Nrp2-D301 highlighted in gold and cyan, respectively. (B) Chimeric Nrp1, containing the L1 loop of Nrp2 (gold/cyan), loses over 75% binding to VEGF-A164 relative to WT Nrp1

(gold) (p<0.000004). AP-tagged wild-type and mutant Nrp1 was used at an activity of 1 μmole p-NPP hydrolyzed/min/μL. (C) VEGF-A120, which lacks exon seven, has essentially the same affinity for Nrp2 (gray dashed line, Kd=23nM±1nM) as it does for Nrp1. VEGF-A164 retains exon seven and has dramatically reduced affinity for

Nrp2 (black line, Kd=150nM±4nM) compared to Nrp1. (D) Nrp2 shows three-fold

115 higher retention of E154A (purple bar) relative to WT VEGF-A164 (blue bar)

(p<0.0003). AP tagged wild-type and mutant VEGF was used at an activity of 25

μmole p-NPP hydrolyzed/min/μL.

116 Discussion

Together, these data demonstrate a mechanism for selective VEGF binding to Nrp. We report the first detailed picture of the structural basis for the binding of

Nrp1 and VEGF-A. The interface is shown to involve regions encoded by both exons seven and eight and is found to determine splice-form specific receptor binding and selectivity (Figure 6). Together with mutagenesis of key interfacial residues, our data define the specific contribution of different regions of VEGF-A.

The exon eight encoded C-terminus of VEGF-A is confirmed to be necessary for high-affinity Nrp binding. The C-terminal arginine of VEGF-A is shown to engage the Nrp1-b1 domain binding pocket utilized by all known ligands (Parker et al.,

2010; Starzec et al., 2007; Teesalu et al., 2009; Vander Kooi et al., 2007). A number of mutations to residues in the C-terminal arginine binding pocket have been reported (Herzog et al., 2011). These mutations modulate binding to Nrp1 and show different signaling properties. We report here that T316R represents a true binding-deficient Nrp1 mutant that will be useful for future studies.

117

Figure 6.6: The HBD of VEGF-A is responsible for selective binding to the

Nrp1 b1 domain.

Exon eight encoded residues mediate high-affinity binding whereas exon seven encoded residues primarily govern selectivity.

118 Exon seven encoded residues are also found to directly physically interact with Nrp1. In particular, we show that E154 directly physically engages T299 in the L1 loop of Nrp1, resulting in enhanced and selective Nrp1 binding. The importance of this interaction interface is further underlined by the recently reported VEGF blocking antibody, Anti-Nrp1B (Liang et al., 2007). Surprisingly, this antibody was shown not to block the expected C-terminal arginine binding site of the b1 domain. Instead it binds to residues in the nearby loops including T299 of the Nrp1 L1 loop (Appleton et al., 2007). These data reveal that Anti-Nrp1B engages a binding site on Nrp1 that is shared with VEGF-A164, thus explaining its ability to potently inhibit VEGF-A binding to Nrp1.

It is interesting to note that the other VEGF family members that signal via

Nrp1 also possess conserved electronegative residues at positions analogous to

E154: D158 in VEGF-B167 and E194 in placental growth factor 3 (PlGF) isoform-3.

This suggests that electrostatic repulsion between VEGF family members and

Nrp2 may be a general mechanism governing ligand binding selectivity.

While the Nrp2 L1 loop chimera resulted in increased VEGF-A164 binding

(data not shown), the observed binding is still significantly lower than that observed for wild-type Nrp1. Consistent with this observation, mutating R287 and N290 of

Nrp2, which are immediately N-terminal to the L1 loop, has been reported to enhance the binding of VEGF-A to Nrp2 (Geretti et al., 2007). Since the L1 loop of Nrp2 is highly conserved, it will also be interesting to explore if there are distinct mechanisms that promote selective Nrp2 binding by its in vivo ligands VEGF-C and VEGF-D. Indeed, there may be Nrp-dependent physical mechanisms

119 differentiating Nrp1 dependent angiogenesis and Nrp2 dependent lymphangiogenesis. This also suggests that design of specific Nrp inhibitors that exploit the difference in the L1 loop may be attainable.

The current Chapter and Chapter 6 have focused on understanding the mechanism of Nrp binding to the VEGF family of ligands. Earlier Chapters

(Chapters 3-4) have provided the molecular basis for Sema/Nrp binding. We have found both distinct and overlapping mechanisms that facilitate differential ligand binding to the Nrp family. These data led us to consider how inhibitory modalities could be designed to selectively inhibit one family of ligands over another.

120 Table 6.1: Data collection and refinement statistics Data Collection Beamline APS 22-ID Wavelength 1.0000 Space group P43 Unit cell parameters (a, b, c) 114.97, 114.97, 50.94 Unique reflections 18014 Completeness (%) 91.5(55.4) Resolution (Å) 2.65(2.74-2.65) Rmerge (%) 12.2(53.9) Redundancy 4.3(1.7) I/σ(I) 11.4(2.02) Refinement Resolution limits (Å) 20.0(2.65) No. reflections/no. to compute Rfree 17074/916 R(Rfree) 21.2(26.7) No. protein residues 432 No. solvent/ion molecules 19 No. phosphate molecules 1 RMSD Bond, Å 0.005 RMSD Angle, ° 1.01 Ramachandran outlier/favored (%) 0/96 Residues with bad bonds/angles (%) 0/0 Rotamer outliers (%) 0.26

Table 6.1: Data collection and refinement statistics for Chapter 6

Reprinted in full or part with permission from: Structural Basis for Selective Vascular Endothelial Growth Factor-A (VEGF-A) Binding to Neuropilin-1. Matthew W. Parker, Ping Xu, Xiaobo Li, and Craig W. Vander Kooi, Journal of Biological Chemistry 2012 287, 11082-11089. Copyright 2013 ASBMB.

121 CHAPTER 7: MECHANISM OF SELECTIVE VEGF-A BINDING BY NRP1

REVEALS A BASIS FOR SPECIFIC LIGAND INHIBITION

Introduction

The Nrp family of receptors coordinate ligand-binding events that mediate endothelial cell migration and proliferation and neuronal chemorepulsion (reviewed in (Zachary et al., 2009)). There are two Nrp homologues, Nrp1 and Nrp2, which share the same overall domain architecture with 44% identity in their primary sequence. Nrp ligands include the VEGF family of pro-angiogenic cytokines

(Soker et al., 1998) and the Sema3 family (He and Tessier-Lavigne, 1997;

Kolodkin et al., 1997) of axon guidance molecules (Nakamura et al., 1998). Both

VEGF and Sema3 family of ligands are composed of multiple genes, splice forms, and proteolytic products with different receptor binding specificity and physiological function.

Soluble receptors capable of sequestering specific ligands are an attractive modality for blocking ligand-dependent signaling pathways. VEGF-Trap, a soluble chimera of VEGFR, containing the domains necessary for ligand binding (Holash et al., 2002), has been approved for use as a clinical agent blocking VEGF-A dependent angiogenesis (Stewart et al., 2012). The identification of endogenously expressed soluble Nrp receptors (sNrp) with anti-tumor activity (Gagnon et al.,

2000) has prompted interest in the use of engineered Nrp molecules as specific pathway inhibitors. It was recently reported that mutation to the Nrp2-b1 domain enhances its ability to bind VEGF-A (Geretti et al., 2007) and that administration of this truncated receptor effectively antagonizes VEGF-A dependent angiogenic

122 signaling (Geretti et al., 2010). A Sema3 specific inhibitor would be of significant utility. While Sema3 mediated repulsive cues are essential during development, they pose a significant barrier to axonal regrowth following injury (de Wit and

Verhaagen, 2003). This is particularly the case in repair following spinal cord injury. In response to spinal cord injury a glial scar forms that serves as a barrier to regenerating axons. Sema3 family members are produced by meningeal cells located in the glial scar and are a major component of the repulsive cues that prevent axonal regeneration (De Winter et al., 2002; Niclou et al., 2003). Blocking inhibitory cues represents one fundamental mode of regenerative therapy for partial cord injuries (Fawcett, 2006). However, due to limited understanding of the determinants of Nrp binding specificity, no soluble Nrp-trap exists that is specific for Sema3.

The conserved Nrp architecture provides Nrp homologues with the ability to bind ligands using a common binding mode. As demonstrated in Chapters 5 and

6, the b1 coagulation factor domain of Nrp1 and Nrp2 contains a conserved cleft optimally suited for binding a C-terminal arginine that is necessary for ligand binding (Parker et al., 2012c; Starzec et al., 2007; Vander Kooi et al., 2007). All

VEGF family members contain a C-terminal arginine. Additionally, in Chapters 3 and 4 we demonstrated that all Sema3 family members contain at least one conserved furin recognition sequence that is endogenously cleaved to liberate a

C-terminal arginine (Adams et al., 1997; Parker et al., 2010). Indeed, binding to this shared site underlies the observed competition between VEGF and Sema3

(Geretti et al., 2007; Parker et al., 2010).

123 In physiological context, physical and functional specificity is observed between receptor-ligand pairs. Nrp1 acts as the functional receptor for VEGF-A

(Soker et al., 1998) and Sema3A (He and Tessier-Lavigne, 1997; Kolodkin et al.,

1997) and Nrp2 acts as the functional receptor for VEGF-C (Karkkainen et al.,

2001) and Sema3F (Giger et al., 1998). The mechanism underlying specific

Sema3 family member binding by Nrp1 and Nrp2 has been shown to involve dual- site binding. The Nrp b1 domain binding to the Sema3 C-terminal domain is necessary for high-affinity binding but does not display specificity for Nrp1 or Nrp2

(Giger et al., 1998; Parker et al., 2010). Secondarily, the N-terminal a1 domain of

Nrp1 and Nrp2 selectively binds the sema domain of different Sema3 family members (Chen et al., 1998; Koppel et al., 1997; Merte et al., 2010). Indeed, a

Sema3 binding-deficient Nrp has been reported which disrupts the unique a1/sema interaction (Gu et al., 2002). However, since there is no known secondary binding site, the basis for specificity in VEGF binding remains unclear.

It was recently demonstrated that the essential VEGF-A164/165 (VEGF-A) isoform binds preferentially to Nrp1 (Parker et al., 2012c). The C-terminal arginine binding cleft of the Nrp1-b1 domain is formed by three loops which are a common feature among coagulation factor domains. While the cleft is conserved, a significant number of residues surrounding this binding pocket differ between Nrp1 and Nrp2 and may contribute to the observed ligand binding specificity. Indeed, the L1 loop of the Nrp b1 domain has been shown to contribute to the observed preferential binding of VEGF-A to Nrp1 (Parker et al., 2012c). However, this interaction alone is insufficient to explain the marked difference in potency of

124 VEGF-A binding to Nrp1 and Nrp2, therefore other molecular determinants of this preferential binding must exist. The recently reported Nrp2 mutation, when Nrp2

R287 is replaced with the corresponding Nrp1 E285, shows enhanced VEGF-A binding and, importantly, unchanged binding to Sema3F (Geretti et al., 2007).

Together, these data suggest that the Nrp b1 domain contains distinct features that govern specific ligand binding and could be exploited to produce specific inhibitors of Nrp ligands. By understanding how selective VEGF binding is achieved, and whether these regions also affect Sema3 binding, a Nrp molecule that specifically bound Sema3 and blocked Sema3 repulsive cues could be designed as a potentially important therapeutic tool.

In the present study we determined the basis for preferential Nrp1 binding to VEGF-A. Mutagenesis of residues surrounding the shared C-terminal arginine binding pocket identifies residues that differ between Nrp1 and Nrp2 and underlie the observed VEGF-A specificity. A chimeric Nrp2, which combines the identified mutations, is capable of binding VEGF-A similarly to Nrp1 whereas a chimeric Nrp1 shows significant loss of VEGF-A binding. We further show that Nrp1 and Nrp2 both bind Sema3F with similar affinity and that both Nrp2 and the chimeric Nrp1 can selectively sequester Sema3. These data establish that unique mechanisms are used by Sema3 and VEGF-A to mediate specific Nrp binding, revealing a basis for the use of engineered Nrp molecules as selective inhibitors of Sema3.

125 Results

Soluble Nrp1 selectively inhibits VEGF-A

To assay the inhibitory potency of Nrp1 and Nrp2, we tested the ability of

Nrp1 and Nrp2 to inhibit VEGF-A binding to Nrp1 affinity plates in a dose dependent manner (Figure 7.1). Nrp1 was able to potently inhibit the binding of

VEGF-A to Nrp1 affinity plates with an IC50 = 1.8 μM (log IC50 = -5.7 ± 0.2) (black line, Figure 7.1). In contrast, Nrp2 was able to inhibit binding only at the highest concentrations with an IC50≈ 310 μM (log IC50 = -3.5 ± 0.4) (grey line, Figure 7.1).

Each experiment was performed in triplicate with unique protein preparations to provide a direct measurement of inter-assay variability, with 11% inter-assay variability observed for Nrp1 compared to 3.5% intra-assay variability. These data demonstrate, both qualitatively and quantitatively, the robust quality and reproducibility of the reported data. The greater than 100-fold difference in IC50 between Nrp1 and Nrp2 is consistent with the previously reported VEGF binding selectivity (Parker et al., 2012c).

126

Figure 7.1: Nrp1 selectively inhibits VEGF-A binding.

The ability of Nrp1 and Nrp2 to selectively sequester AP-VEGF-A from Nrp1 adsorbed on affinity plates was assessed. The amount of retained AP-VEGF-A was measured and the Nrp concentration recorded where half-inhibition was achieved. Nrp1 inhibited the binding of VEGF-A with an IC50 = 1.8 μM (black line).

Inhibition by Nrp2 was seen only at the highest concentration of protein attainable with an estimated IC50 ≈ 310 μM (grey line). Experiments were performed in triplicate and reported as the mean ± 1 S.D.

127 Identification of the regions of Nrp1 that confer selective VEGF-A binding

The selective ability of Nrp1 to inhibit VEGF-A binding led us to consider the mechanism of this specificity. The predominant Nrp structural determinants mediating VEGF-A binding have been localized to domain b1 (Geretti et al., 2007;

Gu et al., 2002; Parker et al., 2012c). The b1 domains of Nrp1 and Nrp2 are 45% identical. Conserved between both are the residues which form the C-terminal arginine binding pocket demonstrated to be essential for VEGF-A/Nrp binding

(Vander Kooi et al., 2007) (Figure 7.2, A, asterisk). Alignment of Nrp1 and Nrp2 domain b1 reveals diversity in the primary sequence conservation of regions surrounding the binding pocket. There are three different categories of residues: those that are well conserved in both Nrp1 and Nrp2, those that are not well conserved, and those which are distinct between Nrp1 and Nrp2 but well conserved within each ortholog. We hypothesized that residues in this last category likely underlie the observed specific Nrp1/VEGF-A binding. Further, residues nearby the binding pocket, especially those in the coagulation-factor loops (L1-L3) often utilized in ligand binding in coagulation-factor domain proteins

(Fuentes-Prior et al., 2002), most likely underlie specific binding.

Based on these two criteria, i.e., that residues are separately conserved within Nrp1 and Nrp2 and that they are nearby the binding site, four regions were selected: Nrp1:E285/Nrp2:R287, Nrp1:299-TN-300/Nrp2:301-DGR-303,

Nrp1:304-ER-305/Nrp2:307-QQ-308, and Nrp1:350-KKK-352/Nrp2:353-QNG-355

(Figure 7.2, B). Mutant proteins were produced by swapping the selected sequences between the two Nrp homologues. Nrp residues that contribute to the

128 observed specificity would be expected to reduce Nrp1 binding and enhance Nrp2 binding when reversed. These chimeric Nrp molecules were expressed with an N- terminal AP tag. The binding of AP-Nrp constructs to VEGF-A affinity plates was measured and normalized relative to wild-type Nrp1 or Nrp2 binding. Reduced binding to VEGF-A was observed for all Nrp1 chimeras relative to wild-type Nrp1

(Figure 7.2, C), suggesting a role of these sequences in VEGF-A binding. Three of the four Nrp2 chimeras showed enhanced affinity for VEGF-A relative to wild- type Nrp2 (Figure 7.2, D). The reduction in VEGF-A binding for both Nrp1 ER/QQ and Nrp2 QQ/ER suggests that these mutations may destabilize the protein and this mutation was therefore excluded from further study. These results suggest that

E285, 299-TN-300, and 350-KKK-352 of the Nrp1-b1 domain contribute to selective VEGF-A binding.

129

Figure 7.2: Nrp1 residues mediate specific VEGF-A binding.

(A) Alignment of orthologous Nrp1 and Nrp2 b1 domains shows conservation of residues critical for C-terminal arginine binding (marked with a *) but variability within regions surrounding the interloop cleft (orange: Nrp1:E285/Nrp2:R287

(Geretti et al., 2007); green: Nrp1:299-TN-300/Nrp2:301-DGR-303; blue:

Nrp1:304-ER-305/Nrp2:307-QQ-308; purple: Nrp1:350-KKK-352/Nrp2:353-QNG-

355). Below the alignment is a conservation histogram illustrating identity across the displayed sequences. (B) Surface representation of the Nrp b1 domain reveals that the direct VEGF-A binding region (gold) (Parker et al., 2012c) is closely associated with the selected regions (colored according to 2A) in three- dimensional space. (C) VEGF-A binding of Nrp1 mutants reveals loss of binding for each mutant protein compared to wild-type. Retained AP-Nrp1 binding is reported as the percent of retained wild-type AP-Nrp1. (D) Determination of the

VEGF-A binding capacity of Nrp2 mutants reveals that three of the four Nrp2 chimeras show enhanced VEGF-A binding compared to wild-type. Retained AP-

130 Nrp2 binding is reported as the percent of retained wild-type AP-Nrp2. Experiments were performed in triplicate and reported as the mean ± 1 S.D.

131 Nrp2Chimera inhibits VEGF-A binding

To assess whether the identified loops compose the predominant features of the Nrp1-b1 domain that confer specific VEGF-A binding, we generated a Nrp1 and Nrp2 molecule which incorporated all three different mutations in a single construct, termed Nrp1Chimera and Nrp2Chimera. To ensure that the mutant proteins were well folded, CD was utilized to assess wild-type and chimeric constructs

(Figure 7.3, A). All proteins produce spectra consistent with the expected β- sandwich architecture of the protein. Further, wild-type and mutant proteins show superimposable spectra demonstrating that the mutations are not structurally deleterious. As a quantitative measure of the selectivity of NrpChimera proteins for

VEGF-A, we assayed the potency of Nrp1Chimera and Nrp2Chimera in inhibiting AP-

VEGF-A binding to Nrp1 affinity plates (Figure 7.3, B). The dose-dependent ability of these constructs to inhibit VEGF-A binding was compared to wild-type Nrp1 and

Nrp2 (Figure 7.1). The Nrp1Chimera showed a significant loss in potency, with an

IC50 = 62 μM (log IC50= -4.2 ± 0.1) (blue line, Figure 7.3, B), an over 30-fold loss in potency compared to Nrp1 (black line, Figure 7.1). Strikingly, Nrp2Chimera gained fifty-fold potency relative to Nrp2, with an IC50 = 3.9 μM (log IC50 = -5.4 ± 0.2)

(green line, Figure 7.3, B), nearly to the level observed for wild-type Nrp1.

132

Figure 7.3: NrpChimera molecules exhibit reversed VEGF-A specificity.

(A) The secondary structure of WT Nrp and NrpChimera was assessed by CD. The overlapping spectra of NrpChimera with wild-type Nrp demonstrate that the incorporated mutations are not structurally deleterious. (B) Nrp1Chimera (blue line) and Nrp2Chimera (green line) were tested for their ability to selectively sequester AP-

VEGF-A from Nrp1 adsorbed on affinity plates. The NrpChimera molecules show reversed VEGF-A specificity with Nrp1Chimera having a marked reduction in

Chimera inhibitory potency (IC50 = 62 μM) and Nrp2 exhibiting a significant gain in potency (IC50 = 3.9 μM). Wild-type Nrp1 (black dotted line, IC50 = 1.8 μM) and Nrp2

(grey dotted line, IC50 ≈ 310 μM) are shown for comparison (data from Figure 7.1).

Experiments were performed in triplicate and reported as the mean ± 1 S.D.

133 Nrp1 and Nrp2 equivalently inhibit Sema3F binding

VEGF and Sema3 are the two major ligand families of Nrp receptors. The

Sema3 family of ligands has been demonstrated to bind Nrp receptors via a dual- site binding mechanism. The Sema3 semaphorin domain mediates specific ligand binding via the Nrp a1 domain (Koppel et al., 1997) and the Sema3 basic C- terminus mediates common high-affinity binding to the Nrp b1b2 domains (Gu et al., 2002). To confirm that the Nrp1 and Nrp2 b1b2 domains bind to Sema3, and that they do not display specificity, we measured their ability to inhibit the binding of AP-Sema3F to Nrp1. Both Nrp1 and Nrp2 showed a dose-dependent inhibition of AP-Sema3F binding with equivalent observed potency for Nrp1, IC50 = 2.0 μM

(log IC50 = -5.7 ± 0.1), and Nrp2, IC50 = 2.7 μM (log IC50 = -5.6 ± 0.3) (Figure 7.4,

A). To confirm that this interaction is mediated by the C-terminal domains of

Sema3, we assayed the ability of Nrp1 and Nrp2 b1b2 domains to inhibit the binding of AP-Sema3F-Ig-basic to Nrp1. Consistent with the potency against full- length Sema3F, Nrp1 inhibited AP-Sema3F-Ig-basic binding with an IC50 = 1.2 μM

(log IC50 = -5.9 ± 0.1) and Nrp2 inhibited AP-Sema3F-Ig-basic binding with an IC50

= 6.2 μM (log IC50 = -5.2 ± 0.2) (Figure 7.4, B).

These data demonstrate that the b1b2 domain of Nrp1 and Nrp2 contain structural determinants capable of C-terminal Sema3F binding and that this binding does not show specific binding to the two Nrp homologues. To confirm that the interaction between the basic domain of Sema3F and Nrp1-b1b2 is conserved across the Sema3 family, and that this interaction site overlaps with that for VEGF-A, we measured the ability of a peptide corresponding to the C-terminus

134 of both Sema3F and Sema3A to inhibit VEGF-A binding (Figure 7.4, C). C- furSema3F and C-furSema3A, peptides corresponding to the furin-activated forms of Sema3F and Sema3A, respectively, were assayed for their ability to competitively inhibit the binding of VEGF-A. Both peptides showed potent, dose- dependent inhibition of AP-VEGF-A binding with IC50 = 22 nM (log IC50 = -7.7 ±

0.04) and 67 nM (log IC50 = -7.2 ± 0.04) for C-furSema3F and C-furSema3A, respectively. These data confirm that the Sema3 family of ligands utilize their basic

C-terminus for equivalent high-affinity binding to the Nrp b1b2 domains, and that this binding is competitive with that of VEGF-A.

135 Figure 7.4: Nrp inhibits Sema3F binding through interaction with the C- terminal basic domain.

(A) Nrp1 and Nrp2 dependent inhibition of AP-Sema3F binding to Nrp1 affinity plates was measured. Both Nrp homologues showed similar ability to compete for

AP-Sema3F, with Nrp1 inhibiting with an IC50 = 2.0 μM and Nrp2 with an IC50 =

2.7. μM. (B) The ability of Nrp1 and Nrp2 to selectively sequester AP-Sema3F-Ig- basic from Nrp1 adsorbed on affinity plates was assessed. The amount of retained

AP-Sema3F-Ig-basic was measured and the Nrp concentration recorded where half-inhibition was achieved. Nrp1 (black line) and Nrp2 (grey line) had similar ability for inhibiting Sema3F binding with IC50 = 1.2 μM and IC50 = 6.2 μM, respectively. (C) Two peptides corresponding to the C-terminal basic domain of

Sema3A (C-furSema3A) and Sema3F (C-furSema3F) were analyzed for their ability to inhibit AP-VEGF-A binding to Nrp1 affinity plates. Both peptides potently inhibited binding, with an IC50 = 22nM and 67nM for C-furSema3F and C- furSema3A, respectively. Experiments were performed in triplicate and reported as the mean ± 1 S.D.

136 Nrp2 and Nrp1Chimera relieve C-furSema mediated inhibition of AP-VEGF-A binding to Nrp1

Nrp1 and Nrp2 display similar affinity for Sema3F (Figure 7.4, A & B) but

Nrp2 has markedly reduced binding to VEGF-A relative to Nrp1 (Figure 7.1). This data suggests that Nrp2 may be a potent and specific inhibitor of Sema3 binding to Nrp receptors. To demonstrate the use of Nrp2 as a selective Sema3 inhibitor, we assayed the ability of Nrp2 to selectively relieve Sema3-dependent inhibition of VEGF-A binding. Selectivity would be demonstrated by a reduction of C- furSema mediated-inhibition and resultant gain in AP-VEGF-A binding to Nrp1 affinity plates. Nrp1 showed no ability to relieve the inhibition of Sema3F-mediated inhibition. In fact, Nrp1 directly sequestered VEGF-A at high Nrp concentrations resulting in complete loss of binding as expected (black line, Figure 7.5).

Remarkably, Nrp2 significantly enhanced the amount of VEGF-A retained on Nrp1 plates to 63% the level of retention seen in the absence of inhibition (grey line,

Figure 7.5). This provides direct evidence that Nrp2 is able to directly and specifically sequester Sema3. Similarly, 37% recovery was seen with Nrp1Chimera

(blue line, Figure 7.5), consistent with the reversal of specificity seen for VEGF-A inhibition (Figure 7.3, B). These data demonstrates that the unique mechanisms utilized by Nrp to preferentially bind different members of the Sema3 and VEGF family ligands can be exploited to create Nrp inhibitors specific for different Nrp ligand families.

137

Figure 7.5: Nrp2 and Nrp1Chimera preferentially sequester Sema3F.

The ability of Nrp1, Nrp2, and Nrp1Chimera to selectively sequester C-furSema was determined through combined incubation of Nrp, C-furSema3F, and AP-VEGF-A in Nrp1 adsorbed affinity plates. Nrp1 was unable to relieve C-furSema-dependent inhibition and completely abolished VEGF-A binding (black line). Conversely, titration with Nrp2 promoted 63% recovery of VEGF-A binding demonstrating selective sequestration of Sema3F (grey line). Nrp1Chimera also relieved C- furSema-dependent inhibition, promoting recovery of VEGF-A binding to 37% the level of uninhibited VEGF-A (blue line). Experiments were performed in triplicate and reported as the mean ± 1 S.D.

138 Discussion

These data establish the basis for selective Nrp ligand binding to the b1 coagulation factor domain of Nrp. While both VEGF and Sema3 ligand families share a partially overlapping ligand binding site in the Nrp b1 domain (Parker et al., 2010), we demonstrate a series of residues that differ between Nrp1 and Nrp2 and contribute to the observed specific VEGF-A binding.

Nrp2Chimera possesses nearly full reversal of selectivity to the level of Nrp1, indicating that the three altered regions represent the predominant regions mediating specific VEGF-A binding. The TN/DGR L1-loop is highly divergent between Nrp1 and Nrp2 and possesses the only insertion/deletion in the b1 domain. Mutation of this region results in the largest net loss of VEGF-A binding to Nrp1, consistent with a direct interaction involving a hydrogen bond between

Nrp1-T299 and VEGF-A-E154 (Parker et al., 2012c). The KKK/QNG L3-loop produces the largest net gain in Nrp2 binding to VEGF-A. Conserved residues in the L3 loop form one wall of the C-terminal arginine binding pocket. The non- conserved KKK/QNG residues do not appear to directly engage VEGF-A in the bound form. The residues are, however, very distinct in physical properties with the Nrp1 loop being significantly more electropositive. VEGF-A contains a series of conserved acidic residues in the C-terminus of its heparin-binding domain, including E154 discussed above, which contribute to Nrp binding. This suggests that the L3-loop may function by electrostatic steering of VEGF-A. This may represent a general mechanism allowing specific binding of other Nrp1 specific ligands such as VEGF-B and placental growth factor (PlGF) that also possess

139 acidic residues in their heparin-binding domains. In contrast, Sema3 family members and Nrp2-specific VEGF-C and VEGF-D do not possess these acidic residues. A mutant Nrp2 containing R287E has previously been shown to possess enhanced binding to VEGF-A (Geretti et al., 2007). We show that charge reversal produces a decrease in Nrp1 binding to VEGF-A indicating a direct contribution to ligand binding selectivity. Geretti and colleagues proposed that this may be due to enhancing the electronegative potential of the b1 domain of Nrp2 favoring binding to the electropositive VEGF-A. R287 is located in a helical region directly adjacent to the L1 loop, suggesting that these regions interact to correctly position these two structural elements, and thus the critical ligand binding coagulation factor loops. While the NrpChimera molecules substantially reverse the observed specificity for VEGF-A, the reversal does not reach the corresponding wild-type level and one or more additional regions may also contribute to selective binding.

140 CHAPTER 8: MICROPLATE-BASED SCREENING FOR SMALL MOLECULE

INHIBITORS OF NRP2/VEGF-C INTERACTIONS

Collectively, this dissertation has provided molecular details explaining the mechanism of Nrp binding to multiple ligands, including Sema3, VEGF-A, and

VEGF-C. Using structural approaches we have demonstrated the physical basis for Nrp1 and Nrp2 binding to the VEGF-A and VEGF-C family of ligands. Thus, we are uniquely poised to develop inhibitors of Nrp ligand binding. Indeed, we have already demonstrated the use of peptides (Chapters 3, 4, and 6) and soluble receptor fragments (Chapters 6 and 7) for VEGF and Sema3 inhibition. However, we are interested in expanding our investigation of Nrp inhibitory modalities through the development of small molecule inhibitors that would be of significant therapeutic interest.

Nrp2 inhibition is of particular interest due to its important role in tumorigenesis. Nrp2 is highly expressed in lymphatic endothelial cells where it is important for physiological lymphangiogenesis and can contribute to pathological tumor lymphangiogenesis (Ellis, 2006; Parker et al., 2012a; Stacker et al., 2002).

Additionally, expression of Nrp2 and its ligand, VEGF-C, are induced in a variety of cancer types and their expression is directly correlated with metastasis to regional lymph nodes and advanced stage disease. Thus, blocking Nrp2 activation by VEGF-C represents a promising anti-lymphangiogenesis therapeutic strategy.

Mature VEGF-C, produced through proteolytic removal of N- and C-terminal propeptides, binds domains b1b2 of Nrp2 (Karpanen et al., 2006). A number of C- terminal arginine containing molecules have been identified as inhibitors of

141 Nrp/ligand binding (Parker et al., 2013). However, these compounds have limited potency and thus identification of inhibitors that target secondary binding sites are of significant interest (Guo et al., 2013). To identify novel small molecule inhibitors of VEGF-C binding to Nrp2-b1b2, we developed and optimized an assay that utilizes Nrp2-affinity plates and an alkaline phosphatase (AP) fusion of VEGF-C

(AP-VEGF-C) for detection of competitive inhibitors of ligand binding. We used this assay to screen the National Institutes of Health (NIH) Clinical Collection, provided through the NIH Molecular Libraries Roadmap Initiative, and we report the identification of three inhibitors of VEGF-C binding to Nrp2.

To make Nrp2-b1b2-affinity plates, we expressed domains b1b2 of human

Nrp2 (Nrp2-b1b2, residues 276-595) in Escherichia coli (E. coli) strain Rosetta

Gami-2(DE3) (Novagen/EMD Millipore, Billerica, MA) as a hexa-histidine fusion protein from pET28b (Novagen/EMD Millipore). Nrp2-b1b2 was purified in a two- step process via immobilized metal affinity chromatography (IMAC) (HIS-Select

HF Nickel Affinity Gel, Sigma-Aldrich, St. Louis, MO) followed by heparin affinity chromatography (HiTrap Heparin HP, GE Healthcare Life Sciences, Pittsburgh,

PA) (Vander Kooi et al., 2007). Purified Nrp2-b1b2 was then diluted to 50 μg/mL with 50 mM Na2CO3 pH 10.4 and immediately added to 96-well protein high-bind microplates (Plate #9018, Corning, Corning, NY) and incubated for 1 h at 37°C.

Bovine serum albumin (BSA) (Simga-Aldrich) was used to coat control wells. The plate was then aspirated, washed 5x 100 μL with PBS-T (phosphate buffered saline, 0.1% Tween 20), and stored with 100 μL PBS-T at 4°C. To detect VEGF-C binding to Nrp2-affinity plates, we produced mature mouse VEGF-C (residues 108-

142 223) fused at its N-terminus to the reporter gene alkaline phosphatase (AP) (AP-

VEGF-C). AP-VEGF-C was expressed via PEI mediated transient transfection of

Chinese hamster ovary suspension cells (CHO-S) (Aricescu et al., 2006; Longo et al., 2013) from the pAPtag-5 vector (GenHunter Corporation, Nashville, TN). AP-

VEGF-C conditioned media was then buffer exchanged into assay buffer (50 mM

NaCl, 200 mM Tris pH 7.4) and concentrated to an activity of 7 x 10-1 u/mL. AP-

VEGF-C binding to Nrp2-b1b2-affinity plates was detected by adding 100 μL of 1X

AP substrate and the evolution of para-nitrophenol phosphate was monitored by measuring 405 nm absorption on a microplate reader. Prior to analysis the absorbance of the control wells (BSA-coated) was used to background correct the data.

The assay conditions for screening, including pH and buffer, salt, and

DMSO concentration, were optimized. First, we analyzed AP-VEGF-C binding to

Nrp2-b1b2-affinity plates over a pH range of 3 to 9 (Figure 8.1, A). 100 mM citrate was used to buffer samples with a pH less than 6.5 and 100 mM Tris was used to buffer samples with a pH greater than 6.5. The highest AP-VEGF-C binding was observed as a stable plateau between pH = 5.5 and pH = 8. Reduction of pH below

5 eventually resulted in a complete loss of binding. The observed pH sensitivity of

AP-VEGF-C/Nrp2-b1b2 binding is consistent with the known requirement for a C- terminal arginine in Nrp ligands (Parker et al., 2012b; Vander Kooi et al., 2007). In light of the observed sensitivity of AP-VEGF-C/Nrp2-b1b2 binding to extremes in pH, we determined the effect of buffer concentration on binding (Figure 8.1, B).

AP-VEGF-C binding to Nrp2-b1b2-affinity plates was analyzed at physiological pH

143 = 7.4 with Tris concentrations between 0 mM and 500 mM. The assay tolerated a range of Tris concentrations, with the highest binding observed at 150 mM.

Additionally, we determined the effect of ionic strength on AP-VEGF-C/Nrp2-b1b2 binding (Figure 8.1, C). Binding was measured in 200 mM Tris pH 7.4 over a range of NaCl concentrations, from 0 mM to 300 mM. The highest AP-VEGF-C binding was observed at 50 mM NaCl and greater than 75% binding was observed for all other NaCl concentrations tested. Collectively, we used these data to define the assay buffer as 50 mM NaCl, 200 mM Tris pH 7.4 for producing the highest signal- to-noise while maintaining tight control over pH.

As a final optimization step, we determined the effect of dimethylsulfoxide

(DMSO), a common solvent used in small molecule libraries, on AP-VEGF-C binding to Nrp2-b1b2-affinity plates (Figure 8.1, D). AP-VEGF-C was prepared in assay buffer and binding to Nrp2-b1b2 was analyzed in the presence of increasing concentrations of DMSO, from 0% to 10% (v/v). DMSO markedly inhibited AP-

VEGF-C/Nrp2-b1b2 binding with 60% of AP-VEGF-C binding lost at 2% DMSO.

Therefore, in order to maintain robust assay signal, a maximum of 2% DMSO was not exceeded in subsequent experiments.

144

Figure 8.1: Assay optimization.

(A) The pH sensitivity of AP-VEGF-C/Nrp2-b1b2 binding was analyzed over a pH range of 3 to 9. (B) The effect of increasing Tris concentrations on AP-VEGF-

C/Nrp2-b1b2 was determined. (C) AP-VEGF-C/Nrp2-b1b2 binding was analyzed for a range of NaCl concentrations. (D) DMSO tolerance was determined by titrating DMSO with AP-VEGF-C and analyzing binding to Nrp2-b1b2-affinity plates.

145

With the optimized assay conditions, we then validated our approach by testing the ability of the Nrp-binding inhibitory peptide, ATWLPPR (Parker et al.,

2010; Starzec et al., 2007; Starzec et al., 2006), which was initially identified as a competitive inhibitor of Nrp1 ligand binding, to inhibit AP-VEGF-C binding to Nrp2- b1b2 (Figure 8.2, A). Serial dilutions of ATWLPPR from 300 μM to 3 μM were prepared in assay buffer containing AP-VEGF-C and 2% DMSO. ATWLPPR fully inhibited AP-VEGF-C/Nrp2-b1b2 binding with an IC50 = 43 μM. This is consistent with the previously published potency of this peptide against Nrp1 (IC50 = 60 μM) and confirms that the core ligand binding pocket engaged by ATWLPPR is conserved between Nrp1 and Nrp2 (Starzec et al., 2006). These data confirm the functionality of our assay in characterizing inhibitors of VEGF-C/Nrp2-b1b2 binding.

C-terminal arginine-like compounds have established efficacy for inhibition of VEGF/Nrp binding (Jarvis et al., 2010). However, these compounds suffer from limited potency and thus we aimed to identify non-C-terminal arginine-like inhibitors. The NIH Clinical Collection is a diverse small molecule library of approximately 450 drug-like compounds that have a history of clinical testing with known safety profiles. We screened the NIH Clinical Collection for inhibitors of

VEGF-C/Nrp2-b1b2 binding (Figure 8.2, B, PubChem BioAssay AID: 16941), testing each compound at 200 μM (2% DMSO). The quality of the results were evaluated by two commonly used parameters in high-throughput screens: signal- to-noise (S/N) and Z-factor (Z), the latter takes into account both the assay

146 dynamic range and signal variability. By these two parameters our assay was defined as high quality with an average across-plate S/N = 32 and Z = 0.55 (Zhang et al., 1999). The data were Z-score normalized (rescaled to set the mean signal

(Σ) = 0 with a standard deviation (σ) = 1) (Malo et al., 2006), thus allowing direct comparison of each compound’s activity across different plates. The Z-score for each compound was plotted and a stringent “hit” criteria was defined as differing by more than 3.5 standard deviations from the mean. This selection criteria restricted hits to <1% of the screened compounds and included only inhibitors

(Figure 8.2, B, dashed line). Importantly, all hit compounds were tested for the ability to directly alter the activity of AP, resulting in a false positive, and no effect was observed.

Three compounds exceeded the criteria for a hit: Zafirlukast (σ = -5.3), dihydrexidine HCl (σ = -4.3), and Actinomycin D (σ = -4.0). Each hit was confirmed by preparing a concentrated compound stock (50 mM) in 100% DMSO and measuring the inhibitory potency against AP-VEGF-C/Nrp2-b1b2 binding in triplicate (Figure 8.2, C). All hit compounds showed greater than 90% inhibition of

AP-VEGF-C/Nrp2-b1b2 binding and their relative potencies were consistent with the initial screen. Zafirlukast (Santa Cruz Biotechnology, Dallas, TX) was the most potent inhibitor with an IC50 = 66 μM, dihydrexidine hydrochloride (Tocris

Bioscience, Bristol, UK) was the weakest inhibitor (IC50 = 113 μM), and

Actinomycin D (Tocris Bioscience) fell in between these two compounds (IC50 = 92

μM). Zileuton (Santa Cruz Biotechnology), a drug structurally distinct from

Zafirlukast but also a leukotriene receptor antagonist, was tested alongside the

147 compounds and showed no ability to inhibit AP-VEGF-C/Nrp2-b1b2, indicating the specific activity of these compounds.

148

Figure 8.2: Screen for small molecule inhibitors.

(A) The ability of the peptide ATWLPPR to inhibit AP-VEGF-C/Nrp2-b1b2 binding was determined. ATWLPPR inhibited binding with an IC50 = 43 μM. (B) The NIH

Clinical Collection small molecule library was screened for inhibitors of AP-VEGF-

C/Nrp2-b1b2 binding. Three compounds exceeded the hit criteria (dashed line), reducing binding by more than 3.5 standard deviations from the mean. (C) Hits were validated by titrating each compound with AP-VEGF-C and measuring Nrp2- b1b2 binding. Zafirlukast, Actinomycin D, and Dihydrexidine inhibited AP-VEGF-

C/Nrp2-b1b2 binding with IC50 = 66, 92, and 113 μM, respectively. Zileuton was unable to inhibit binding. (D) Chemical structure of the identified hit compounds.

149

Reprinted in full or part with permission from: Microplate-based screening for small molecular inhibitors of neuropilin-2/vascular endothelial growth factor-C interactions. Matthew W. Parker and Craig W. Vander Kooi, Analytical Biochemistry 2014 15 (453), 4-6. Copyright 2013 Elsevier.

150 CHAPTER 9: DISCUSSION

Future Directions of Nrp Research

Cumulatively, the work described in this dissertation defines the molecular basis for binding of the Sema3 and VEGF family of ligands to the Nrp family of essential cell surface receptors. Importantly, we have demonstrated that both Nrp1 and Nrp2 contain a C-terminal arginine binding pocket used for the high-affinity binding of both ligand families along with accessory sites to tune potency and specificity. The utilization of a common binding site by both VEGF and Sema3 underlies their observed competitive binding and provides an explanation for their functional competition observed in vivo.

Because of the complex interplay of Nrp with other molecules, a number of critical areas regarding the mechanism of Nrp function remain to be explored. Of particular interest is the specific role for Nrp in distinct signaling cascades and tissue types. Connected to this is the extent to which Nrp functions by the same general mechanism in highly diverse pathways or if there are distinct mechanisms employed in different signaling cascades.

Nrp Function as an Essential Co-receptor

The nature of Nrp coupling to signaling receptors to form a functional signaling holocomplex is a major area of future research. In particular, unique heterophilic receptor/receptor contacts are likely critical to the formation and stability of the complex. Although necessary for signaling, the role of the Nrp MAM domain is largely unknown. Thus, the MAM domain might physically couple to cognate receptors. Indeed, the membrane proximal seventh Ig-like domain of

151 VEGFR-2 has been shown to form a dimer required for activation (Ruch et al.,

2007; Yang et al., 2010), which could be coupled to MAM domain function in the holo-complex. Further, the specific function of the transmembrane domain of Nrp may be to directly couple to signaling receptors or to allow precise physical arrangement of the intracellular domain. The transmembrane domain of VEGFR-

2 has been demonstrated to facilitate correct orientation of homo-dimers to couple ligand binding to receptor activation (Dosch and Ballmer-Hofer). Nrp may require similar specific alignment of the transmembrane domain and may, in fact, directly couple to the transmembrane domain of cognate signaling receptors. Indeed, it is possible that the extracellular juxtamembrane, transmembrane, and intracellular juxtamembrane domains may coordinately function to physically couple ligand binding to receptor activation.

The function of the intracellular domain of Nrp and the extent to which Nrp can function independently of a canonical signaling receptor, for example VEGFR or Plexin, remains to be determined. Of particular interest are the molecules that can form direct interactions or indirectly bridge with Nrp. While GIPC is clearly a critical Nrp adaptor protein, other PDZ-domain proteins have also been suggested to function with Nrp. In particular, Nrp was recently identified as a positive regulator of hedgehog signaling (Hillman et al., 2011) and the PDZ-domain protein required for this activity is currently being pursued. Connected to this is the extent to which autocrine signaling represents a fundamental mode of Nrp activation in disease, with particular interest in the connection to cancer stem cell maintenance.

152 Physical basis for Sema3 function via Nrp receptors

We have investigated the molecular basis for Sema3 binding to the Nrp family of receptors. In Chapter 3 we demonstrated that the Sema3 family is processed at a conserved furin recognition site in the C-terminal domain. This processing directly regulates Nrp binding and VEGF competition. Additionally,

Chapter 4 describes specific differences in the furin recognition site across the

Sema3 family and how the strength of the consensus site and Nrp binding is inversely correlated. While it is clear that only the processed form of Sema3 is able to function as a VEGF pathway inhibitor, the role of C-terminal processing of

Sema3 in axon guidance is an intriguing area that remains to be explored. The C- terminal domain of Sema3 is necessary, but not sufficient, for its function in axon guidance. Sema3 additionally requires an interaction between its sema domain and the a1 domain of Nrp (Gu et al., 2002; Klostermann et al., 1998; Mamluk et al., 2002). It is interesting to note that the C-terminal Fc fusion of Sema3, which represents the unprocessed form of Sema3, is able to cause axon repulsion in situ

(Appleton et al., 2007; Cheng et al., 2004). Thus, both processed and unprocessed forms of Sema3 are able to function in situ in axon guidance. In vivo, furin processing may be solely utilized as a mechanism regulating the anti- angiogenic activity of Sema3 or it may alter the potency and range of activity of

Sema3 in axon guidance. Alternatively, VEGF has well characterized neurotrophic and neuroprotective effects and it is possible that furin processing of Sema3 could affect the ability of VEGF to compete for Nrp binding on the surface of neuronal and glial cells.

153 While the basis for ligand binding specificity has come into focus through our work, the contribution of physical mechanisms governing ligand binding selectivity versus tissue specific expression remains to be determined. The physical basis for divalent Sema3 binding to Nrp remains an important unanswered question. Specifically, the contribution and coupling between the a1 and b1 interaction sites and the effect of post-translational modification by furin to binding and signaling remain outstanding questions.

Intriguingly, these data also have direct implications for the in vivo function of differential furin processing in Kallmann’s syndrome, a genetic disease characterized by incomplete development of olfactory nerve fibers and defective migration of neuroendocrine cells. A recent analysis of patients revealed mutations in the Sema3A gene underlying the disease, including R733H (Hanchate et al.,

2012). R733 is the C-1 residue of the Sema3A.1 furin site (Adams et al., 1997).

Although R733H was efficiently expressed and secreted, it exhibited a marked and significant reduction in its ability to signal (Hanchate et al., 2012). The results presented within this dissertation suggest that this mutation would reduce furin processing at the Sema3A.1 site, one of three furin consensus sites within the basic domain. Thus, differential processing and Nrp engagement of the different furin sites has important function in both normal physiology and pathology. Our data provide a mechanistic basis for understanding the effect of mutations on both furin processing and Nrp binding.

Semaphorin-like proteins are produced by a variety of viruses that utilize molecular mimicry. Various poxviruses encode SemaV family members, which

154 are sema domain proteins homologous to the N-terminus of semaphorin (Yazdani and Terman, 2006). SemaV family members have been shown to induce changes in host cytoskeletal dynamics, thereby altering the adherence and spreading of infected cells (Walzer et al., 2005). It has recently been shown that Nrp is essential for HTLV-1 viral entry (Ghez et al., 2006). The HTLV-1 coat protein is a heparin binding protein that directly interacts with Nrp (Lambert et al., 2009). Further, the interaction and infectivity of HTLV-1 can be attenuated by both VEGF-A and peptide inhibitors of Nrp (Lambert et al., 2009). Intriguingly, the HTLV coat protein that interacts with Nrp requires furin processing for maturation and infectivity

(Hasegawa et al., 2002). From our studies, we suggest that HTLV-1 utilizes molecular mimicry of the mature furin processed form of Sema3 to target the shared Sema3/VEGF binding site in the b1 domain of Nrp. This insight provides a novel avenue for potential therapeutic intervention in HTLV-1 infected individuals.

Nrp Inhibition in Spinal Cord Injury

Nrp function in Sema3 signaling is important not only for physiological axon guidance, but also for signaling in spinal cord injury (Pasterkamp and Giger, 2009;

Zhou et al., 2008). Following spinal cord injury, a glial scar forms to stabilize and seal the wound. However, the glial scar also serves as a barrier to regenerating axons due to production of axon repulsion molecules. Sema3 expression significantly contributes to the inhibitory nature of the glial scar and so inhibitors of

Nrp-Sema3 signaling hold promise as therapeutics for treatment of spinal cord injury (De Winter et al., 2002; Fawcett, 2006; Pasterkamp et al., 1999).

155 The use of soluble Nrp as a modulator of VEGF-mediated angiogenesis has been the target of numerous studies. However, soluble Nrp as a modality for blocking the chemorepulsive function of Sema3 is less established even though clear clinical applications exist for neutralizing molecules in spinal cord injury

(Niclou et al., 2006). The efficacy of targeting this signaling axis is demonstrated by a small molecule inhibitor of the Sema3A/Nrp1 interaction (Kikuchi et al., 2003) that, when given to rats following spinal cord transection, shows enhanced regeneration of axons across the glial scar (Kaneko et al., 2006). Additionally, initial work has demonstrated that Nrp2 inhibition allows penetration of axons into a model glial scar (Shearer et al., 2003). Two outstanding issues remain to be solved. First, many members of the Sema3 family can mediate this deleterious axonal repulsion, so a potent pan-Sema3 inhibitory modality is desired (De Winter et al., 2002). Second, selective inhibition of Sema3 signaling without effecting

VEGF-A signaling is desired to maximally promote recovery following injury. Nrp2 molecules engineered for increased potency, by oligomerization with an Fc or related strategy, have the potential to be used as selective Sema3-traps. One

B promising approach is to repurpose s9Nrp2 for use against Sema3.

Through understanding the molecular basis for Nrp’s specific binding of its ligands, engineered soluble Nrp receptors can be designed with specificity for a particular ligand. The binding of Sema3F shows little selectivity between Nrp1 and

Nrp2. Further, the observed inhibitory potency of Nrp1 for VEGF-A and Sema3F is virtually identical in absolute terms. Thus, Nrp1 was found to be equally capable of sequestering both VEGF-A and Sema3F and suggests that Nrp1 has utility as

156 a broad-spectrum Nrp ligand inhibitor. In contrast, Nrp2 is found to selectively sequester Sema3F. The reported chimeric Nrps have utility for discriminating between the contribution of VEGF-A and Sema3F function in a particular system.

Previous work has reported Nrp1 mutations in the a1 domain that allow production of a VEGF-A selective Nrp1 (Gu et al., 2002). The Nrp1Chimera reported here represents a complimentary molecule that is selective for Sema3F, with utility in differentiating between specific effects mediated by the different Nrp1 ligands.

Additionally, molecules specific for certain members of the Sema3 family could potentially be produced by combinatorial approaches using the a1 domain of

Nrp1/Nrp2 for specificity in combination with the b1 domain of Nrp2. Our studies demonstrate the mechanism underlying Nrp coagulation factor domain-mediated ligand binding selectivity and advances the search for potent and selective inhibitors of Nrp signaling.

Physical basis for VEGF function via Nrp receptors

We have determined the structure of VEGF-A and VEGF-C in complex with their cognate Nrp receptor, Nrp1 and Nrp2, respectively. The loss of Nrp binding by VEGF-C and VEGF-A C-terminal arginine mutants, or to Nrp mutants with an occluded binding pocket, demonstrates the use of a C-terminal arginine for ligand engagement. Indeed, both the VEGF-A and VEGF-C free carboxylate forms extensive interactions with the Nrp-b1 binding pocket. Interestingly, the C-terminal domain is the most variable region within the VEGF family and VEGF-C is not the only VEGF family member that, in the absence of post-translational modification, lacks a C-terminal arginine. Of the five VEGF family members, three family

157 members contain Nrp-binding isoforms that lack this structural motif (VEGF-C,

VEGF-D, and VEGF-B186). VEGF-D, a close structural and functional homologue of VEGF-C, is processed at an equivalent site in its C-terminus to produce a C- terminal arginine (Stacker et al., 1999) and thus likely utilizes a similar binding mode to Nrp2. This observation provides additional functional insight, as loss of

VEGF-D C-terminal processing also ablates function in vivo (Harris et al., 2013).

The collective results from studies of both VEGF-C (Chapter 5) and VEGF-

A (Chapter 6) raise important questions regarding the function and receptor binding properties of the VEGF-B family of pro-angiogenic cytokines. Interestingly, there are three VEGF-B isoforms, VEGF-B167, VEGF-B127, and VEGF-B186, all of which differ in their C-terminal domain (Olofsson et al., 1996a; Olofsson et al.,

1996b). Characterization of VEGF-B186 demonstrated that it exhibited proteolytic- dependent binding to Nrp1 and identified the site of proteolysis as R227 (Makinen et al., 1999). These data suggest that VEGF-B186 may exhibit a similar mechanism of activation as that observed for VEGF-C. Alternatively, VEGF-B167 contains a heparin-binding domain similar to VEGF-A165 and VEGF-B127 appears homologous to VEGF-A121. Thus, the VEGF-B family contains multiple members that share similarities with divergent members of the VEGF family and, as such, represents an important area for future research.

Our data also suggest a physical basis for the observed functional differences of VEGF family members and isoforms within specific VEGF families.

We have demonstrated that VEGF-A164 selectively physically engages Nrp1 and exhibits dramatically reduced binding to Nrp2, thus explaining the functional

158 specificity of VEGF-A164 in angiogenesis. Further, VEGF-A164 is well documented as the most potent VEGF-A isoform in stimulating angiogenesis. It is well recognized that VEGFR dimerization, while necessary, is insufficient for activation.

Indeed, a specific dimeric organization of the juxtamembrane domain of VEGFR-

2 has been shown to be critical to couple ligand binding to intracellular receptor activation (Dosch and Ballmer-Hofer; Yang et al., 2010). It is possible that the observed binding imposes specific steric constraints, allowing a stable organization of the heterohexameric VEGF-A/Nrp1/VEGFR-2 signaling complex.

Indeed, the heparin binding residues of the HBD (Krilleke et al., 2007) and Nrp1- b1 domain (Vander Kooi et al., 2007) are positioned spatially close together in the complex. This spatial orientation would allow binding of a single glycosaminoglycan (GAG)/heparin chain by the protein complex and further reinforce the specific orientation of the signaling complex. Our data establishes the unique physical engagement of VEGF-A164 by Nrp1 and opens up new avenues to explore the specific physical mechanism of Nrp in angiogenesis.

Nrp Inhibition in Tumor Angiogenesis

Inhibitory modalities targeting Nrp1-dependent VEGF-A induced angiogenesis have been extensively explored. A soluble Nrp splice form, containing only the ligand binding region of the extracellular domain of Nrp has been employed and found to inhibit tumorigenesis (Gagnon et al., 2000). A number of peptides and a synthetic peptidomimetic inhibitor of Nrp have been described (Jarvis et al., 2010; Parker et al., 2010; Starzec et al., 2006; von Wronski et al., 2006). Methods to overexpress inhibitory molecules, for example Sema3A,

159 have been reported with promising activity in tumor angiogenesis (Casazza et al.,

2011). Finally, antibodies targeting both Nrp1 and Nrp2 have been developed that show promising activity in animal models with observed devascularization of solid tumors and decreases in metastasis (Caunt et al., 2008; Liang et al., 2007). These anti-Nrp antibodies are currently being tested in clinical trials. Surprisingly, one of the observed side effects of administration of the Nrp1 antibody, MNRP1685A, was platelet depletion (Weekes et al., 2014). This was shown to be due to platelet activation, aggregation, and clearance.

While significant effort has been devoted to blocking aberrant VEGF-A signaling, only recently has VEGF-C been identified as a promising target for cancer therapeutics. Aberrant activation of VEGF-C signaling via Nrp2 is associated with cancer initiation, survival, and progression (Ellis, 2006; Stacker et al., 2002). The Nrp2/VEGF-C signaling axis contributes to tumorigenesis via multiple mechanisms. Mimicking its physiological function, VEGF-C signaling via

Nrp2 stimulates lymphatic vessel recruitment to tumors and directly contributes to cancer metastasis (Caunt et al., 2008). Importantly, the role of VEGF-C and Nrp2 in tumorigenesis is not exclusively associated with aberrant lymphangiogenesis.

Indeed, in situ studies have demonstrated that autocrine VEGF-C signaling in breast cancer cells stimulates cellular motility (Timoshenko et al., 2007). Further, recent reports indicate that cancer cell survival is enhanced through VEGF-

C/Nrp2-dependent autophagy (Stanton et al., 2012) and that autocrine Nrp2 signaling maintains the population of cancer stem cells (Goel et al., 2013). VEGF-

C also functions to protect prostate cancer cells from oxidative stress in a Nrp2-

160 dependent fashion (Muders et al., 2009). Thus, selective inhibition of Nrp2 represents a promising, multipronged anti-cancer therapeutic strategy.

Secreted splice forms of angiogenic receptors have essential roles in vivo

(Albuquerque et al., 2009; Ambati et al., 2006; Kendall and Thomas, 1993) and have been engineered to serve as therapeutic inhibitors that block aberrant pathway activation by ligand sequestration (Stewart, 2012). In Chapter 5 we

B demonstrate that the alternative Nrp2 splice form, s9Nrp2 , potently sequesters

VEGF-C and inhibits binding to Nrp2. The biological function and localized tissue- specific expression of s9Nrp2 is of significant interest. Indeed, s9Nrp2 may be analogous or complement the secreted splice form of VEGFR-2, sVEGFR-2, which functions as an endogenous lymphangiogenesis inhibitor in the eye (Albuquerque et al., 2009). VEGF-D also functions in lymphatic angiogenesis and has been shown to have partially overlapping biological function with VEGF-C and important pathological functions (Haiko et al., 2008; Harris et al., 2013; Karpanen et al.,

2006). The conservation of Nrp2-interacting residues between VEGF-C and

B VEGF-D strongly suggest that s9Nrp2 will equivalently sequester both VEGF-C and VEGF-D. In contrast, Nrp2 is physically and functionally separate from VEGF-

B A (Parker et al., 2012b; Parker et al., 2012c). Thus, s9Nrp2 is likely to selectively sequester the lymphangiogenic-specific VEGF family members, VEGF-C and

VEGF-D.

Development of Nrp Inhibitors

This dissertation has demonstrated parallel strategies for inhibiting Nrp ligand binding, including peptide inhibitors, soluble receptor fragments, and small

161 molecules. These unique inhibitory modalities provide an important basis for future studies aimed at optimizing the selectivity and potency of Nrp inhibitors for use therapeutically in cancer and spinal cord injury.

Two major classes of peptide-based Nrp inhibitors have been described

(Starzec et al., 2006; von Wronski et al., 2006). Both are relatively small monomeric peptides (5-7 residues) with modest inhibitory potency (mid-mM). It has been unclear if this modest potency is due to specific features of the peptides or if it represents a general problem with this mode of inhibiting angiogenesis. Our results from Chapter 2 reveal that C-furSema is able to inhibit binding of VEGF-A to Nrp with an increase in potency of two to three orders of magnitude relative to previous inhibitors. It will be interesting to determine the physical basis for this enhanced potency. It is notable that while dimeric VEGF was found to directly antagonize Sema3 mediated growth cone collapse, a monomeric peptide inhibitor derived from the C-terminus of VEGF reversed this effect (Cheng et al., 2004). C- furSema contains the strictly conserved intermolecular disulfide, and the multimeric state of the peptide may well contribute to its enhanced potency.

Additionally, the C-terminal region of all known anti-angiogenic Sema3s shows conservation beyond the terminal 5-7 residues. This suggests that additional binding pockets on Nrp may be employed which are not exploited by current generation peptides. Together, these results strongly suggest that potent peptide inhibitors of Nrp can be produced, opening exciting avenues to design novel inhibitors based on Sema3F and other endogenous angiogenesis inhibitors.

162 A second inhibitory modality is soluble receptor fragments that function by sequestering ligand from the active signaling complex. In Chapters 7 we describe a monomeric fragment of Nrp capable of sequestering VEGF and Sema3 with modest potency. The practice of engineering inhibitor multimerization to increase potency is well established for soluble receptor fragments. Most commonly, soluble receptors are dimerized by expression as an Fc fusion protein (e.g. VEGF-trap, discussed earlier, Fc-sVEGFR3 (Lin et al., 2005), and Fc-sVEGFR1/2 (Holash et

B al., 2002)). The dimeric soluble fragment of Nrp2 presented in Chapter 7, s9Nrp2 , represents a unique mechanism for generation of a multimeric protein that maintains the benefits of avidity but does not require introduction of an exogenous

B polypeptide sequence. s9Nrp2 showed significant gains in potency relative to the monomeric inhibitor against VEGF-C and remains to be tested against the Sema3 family. Importantly, this is likely to function as a selective VEGF-C inhibition as

VEG-A is physically and functionally separate from Nrp2. Additional optimization

B of s9Nrp2 potency, selectivity, and stability is an important future direction for the development of a therapeutically useful inhibitor.

In parallel with the strategies described above, we report in Chapter 8 the development of a VEGF-C/Nrp2 binding assay that we used to screen a small molecule compound library for inhibitors of Nrp2. We identified three novel inhibitors of VEGF-C/Nrp2 binding. Intriguingly, both Actinomycin D and

Zafirlukast are used to treat disease states, tumor angiogenesis and asthma, respectively, where anti-angiogenesis has been suggested to be an important beneficial secondary effect of treatment (Lee et al., 2011) (Blumberg, 1974). In

163 both cases, a direct contribution of Nrp inhibition in the biological activities of these compounds is an important area for future studies. Additionally, the development of this assay opens the door for screening other large chemically diverse small molecule libraries to identify more potent inhibitors of VEGF-C function. Notably, none of the three molecules identified contain a C-terminal arginine-like moiety, thus providing novel lead compounds for optimization and combinatorial approaches towards the production of a potent and selective small molecules Nrp2 inhibitor. Continued development of novel inhibitory modalities targeting the different fundamental mechanisms of Nrp action will be both mechanistically informative and have direct relevance to human health.

164 REFERENCES

Adams, R.H., and Eichmann, A. (2010). Axon guidance molecules in vascular patterning. Cold Spring Harbor perspectives in biology 2, a001875.

Adams, R.H., Lohrum, M., Klostermann, A., Betz, H., and Puschel, A.W. (1997). The chemorepulsive activity of secreted semaphorins is regulated by furin- dependent proteolytic processing. The EMBO journal 16, 6077-6086.

Albuquerque, R.J., Hayashi, T., Cho, W.G., Kleinman, M.E., Dridi, S., Takeda, A., Baffi, J.Z., Yamada, K., Kaneko, H., Green, M.G., et al. (2009). Alternatively spliced vascular endothelial growth factor receptor-2 is an essential endogenous inhibitor of lymphatic vessel growth. Nature medicine 15, 1023-1030.

Algire, G.H., and H.W., C. (1945). VASCULAR REACTIONS OF NORMAL AND MALIGNANT TISSUES INVIVO .1. VASCULAR REACTIONS OF MICE TO WOUNDS AND TO NORMAL AND NEOPLASTIC TRANSPLANTS. Journal of the National Cancer Institute 6, 73-85.

Ambati, B.K., Nozaki, M., Singh, N., Takeda, A., Jani, P.D., Suthar, T., Albuquerque, R.J., Richter, E., Sakurai, E., Newcomb, M.T., et al. (2006). Corneal avascularity is due to soluble VEGF receptor-1. Nature 443, 993-997.

Appleton, B.A., Wu, P., Maloney, J., Yin, J., Liang, W.C., Stawicki, S., Mortara, K., Bowman, K.K., Elliott, J.M., Desmarais, W., et al. (2007). Structural studies of neuropilin/antibody complexes provide insights into semaphorin and VEGF binding. The EMBO journal 26, 4902-4912.

Arese, M., Serini, G., and Bussolino, F. (2011). Nervous vascular parallels: axon guidance and beyond. The International journal of developmental biology 55, 439- 445.

Aricescu, A.R., Lu, W., and Jones, E.Y. (2006). A time- and cost-efficient system for high-level protein production in mammalian cells. Acta crystallographica Section D, Biological crystallography 62, 1243-1250.

Bachelder, R.E., Crago, A., Chung, J., Wendt, M.A., Shaw, L.M., Robinson, G., and Mercurio, A.M. (2001). Vascular endothelial growth factor is an autocrine survival factor for neuropilin-expressing breast carcinoma cells. Cancer research 61, 5736-5740.

Bachelder, R.E., Lipscomb, E.A., Lin, X., Wendt, M.A., Chadborn, N.H., Eickholt, B.J., and Mercurio, A.M. (2003). Competing autocrine pathways involving alternative neuropilin-1 ligands regulate chemotaxis of carcinoma cells. Cancer research 63, 5230-5233.

165 Bagci, T., Wu, J.K., Pfannl, R., Ilag, L.L., and Jay, D.G. (2009). Autocrine semaphorin 3A signaling promotes glioblastoma dispersal. Oncogene 28, 3537- 3550.

Bagri, A., Tessier-Lavigne, M., and Watts, R.J. (2009). Neuropilins in tumor biology. Clinical cancer research : an official journal of the American Association for Cancer Research 15, 1860-1864.

Baldwin, M.E., Halford, M.M., Roufail, S., Williams, R.A., Hibbs, M.L., Grail, D., Kubo, H., Stacker, S.A., and Achen, M.G. (2005). Vascular endothelial growth factor D is dispensable for development of the lymphatic system. Molecular and cellular biology 25, 2441-2449.

Ballmer-Hofer, K., Andersson, A.E., Ratcliffe, L.E., and Berger, P. (2011). Neuropilin-1 promotes VEGFR-2 trafficking through Rab11 vesicles thereby specifying signal output. Blood 118, 816-826.

Banerjee, S., Sengupta, K., Dhar, K., Mehta, S., D'Amore, P.A., Dhar, G., and Banerjee, S.K. (2006). Breast cancer cells secreted platelet-derived growth factor- induced motility of vascular smooth muscle cells is mediated through neuropilin-1. Molecular carcinogenesis 45, 871-880.

Banerjee, S.K., Zoubine, M.N., Tran, T.M., Weston, A.P., and Campbell, D.R. (2000). Overexpression of vascular endothelial growth factor164 and its co- receptor neuropilin-1 in estrogen-induced rat pituitary tumors and GH3 rat pituitary tumor cells. Int J Oncol 16, 253-260.

Bates, D.O., Cui, T.G., Doughty, J.M., Winkler, M., Sugiono, M., Shields, J.D., Peat, D., Gillatt, D., and Harper, S.J. (2002). VEGF165b, an inhibitory splice variant of vascular endothelial growth factor, is down-regulated in renal cell carcinoma. Cancer research 62, 4123-4131.

Beck, B., Driessens, G., Goossens, S., Youssef, K.K., Kuchnio, A., Caauwe, A., Sotiropoulou, P.A., Loges, S., Lapouge, G., Candi, A., et al. (2011). A vascular niche and a VEGF-Nrp1 loop regulate the initiation and stemness of skin tumours. Nature 478, 399-403.

Becker, P.M., Waltenberger, J., Yachechko, R., Mirzapoiazova, T., Sham, J.S., Lee, C.G., Elias, J.A., and Verin, A.D. (2005). Neuropilin-1 regulates vascular endothelial growth factor-mediated endothelial permeability. Circulation research 96, 1257-1265.

Benachour, H., Seve, A., Bastogne, T., Frochot, C., Vanderesse, R., Jasniewski, J., Miladi, I., Billotey, C., Tillement, O., Lux, F., et al. (2012). Multifunctional Peptide-conjugated hybrid silica nanoparticles for photodynamic therapy and MRI. Theranostics 2, 889-904.

166 Bender, R.J., and Mac Gabhann, F. (2013). Expression of VEGF and semaphorin genes define subgroups of triple negative breast cancer. PloS one 8, e61788.

Bielenberg, D.R., Pettaway, C.A., Takashima, S., and Klagsbrun, M. (2006). Neuropilins in neoplasms: expression, regulation, and function. Exp Cell Res 312, 584-593.

Blumberg, N. (1974). Tumor angiogenesis factor. Speculations on an approach to cancer chemotherapy. The Yale journal of biology and medicine 47, 71-81.

Brozzo, M.S., Bjelic, S., Kisko, K., Schleier, T., Leppanen, V.M., Alitalo, K., Winkler, F.K., and Ballmer-Hofer, K. (2012). Thermodynamic and structural description of allosterically regulated VEGFR-2 dimerization. Blood 119, 1781- 1788.

Cai, H., and Reed, R.R. (1999). Cloning and characterization of neuropilin-1- interacting protein: a PSD-95/Dlg/ZO-1 domain-containing protein that interacts with the cytoplasmic domain of neuropilin-1. The Journal of neuroscience : the official journal of the Society for Neuroscience 19, 6519-6527.

Cao, S., Yaqoob, U., Das, A., Shergill, U., Jagavelu, K., Huebert, R.C., Routray, C., Abdelmoneim, S., Vasdev, M., Leof, E., et al. (2010a). Neuropilin-1 promotes cirrhosis of the rodent and human liver by enhancing PDGF/TGF-beta signaling in hepatic stellate cells. The Journal of clinical investigation 120, 2379-2394.

Cao, Y., E, G., Wang, E., Pal, K., Dutta, S.K., Bar-Sagi, D., and Mukhopadhyay, D. (2012). VEGF exerts an angiogenesis-independent function in cancer cells to promote their malignant progression. Cancer research 72, 3912-3918.

Cao, Y., Szabolcs, A., Dutta, S.K., Yaqoob, U., Jagavelu, K., Wang, L., Leof, E.B., Urrutia, R.A., Shah, V.H., and Mukhopadhyay, D. (2010b). Neuropilin-1 mediates divergent R-Smad signaling and the myofibroblast phenotype. The Journal of biological chemistry 285, 31840-31848.

Cao, Y., Wang, L., Nandy, D., Zhang, Y., Basu, A., Radisky, D., and Mukhopadhyay, D. (2008). Neuropilin-1 upholds dedifferentiation and propagation phenotypes of renal cell carcinoma cells by activating Akt and sonic hedgehog axes. Cancer research 68, 8667-8672.

Carmeliet, P. (2003). Blood vessels and nerves: common signals, pathways and diseases. Nature reviews Genetics 4, 710-720.

Carmeliet, P. (2005). VEGF as a key mediator of angiogenesis in cancer. Oncology 69 Suppl 3, 4-10.

Casazza, A., Fu, X., Johansson, I., Capparuccia, L., Andersson, F., Giustacchini, A., Squadrito, M.L., Venneri, M.A., Mazzone, M., Larsson, E., et al. (2011). Systemic and targeted delivery of semaphorin 3A inhibits tumor angiogenesis and

167 progression in mouse tumor models. Arteriosclerosis, thrombosis, and vascular biology 31, 741-749.

Caunt, M., Mak, J., Liang, W.C., Stawicki, S., Pan, Q., Tong, R.K., Kowalski, J., Ho, C., Reslan, H.B., Ross, J., et al. (2008). Blocking neuropilin-2 function inhibits tumor cell metastasis. Cancer cell 13, 331-342.

Chen, H., Chedotal, A., He, Z., Goodman, C.S., and Tessier-Lavigne, M. (1997). Neuropilin-2, a novel member of the neuropilin family, is a high affinity receptor for the semaphorins Sema E and Sema IV but not Sema III. Neuron 19, 547-559.

Chen, H., He, Z., Bagri, A., and Tessier-Lavigne, M. (1998). Semaphorin-neuropilin interactions underlying sympathetic axon responses to class III semaphorins. Neuron 21, 1283-1290.

Chen, V.B., Arendall, W.B., 3rd, Headd, J.J., Keedy, D.A., Immormino, R.M., Kapral, G.J., Murray, L.W., Richardson, J.S., and Richardson, D.C. (2010). MolProbity: all-atom structure validation for macromolecular crystallography. Acta crystallographica Section D, Biological crystallography 66, 12-21.

Cheng, L., Jia, H., Lohr, M., Bagherzadeh, A., Holmes, D.I., Selwood, D., and Zachary, I. (2004). Anti-chemorepulsive effects of vascular endothelial growth factor and placental growth factor-2 in dorsal root ganglion neurons are mediated via neuropilin-1 and cyclooxygenase-derived prostanoid production. The Journal of biological chemistry 279, 30654-30661.

Christensen, C., Ambartsumian, N., Gilestro, G., Thomsen, B., Comoglio, P., Tamagnone, L., Guldberg, P., and Lukanidin, E. (2005). Proteolytic processing converts the repelling signal Sema3E into an inducer of invasive growth and lung metastasis. Cancer research 65, 6167-6177.

Cui, J.Z., Hinz, B.J., Greve, M.D., Potter, M.J., Hornan, D., Samad, A., To, E., and Matsubara, J.A. (2003). Expression of neuropilin-1 in choroidal neovascular membranes. Canadian journal of ophthalmology Journal canadien d'ophtalmologie 38, 41-45.

Curreli, S., Arany, Z., Gerardy-Schahn, R., Mann, D., and Stamatos, N.M. (2007). Polysialylated neuropilin-2 is expressed on the surface of human dendritic cells and modulates dendritic cell-T lymphocyte interactions. The Journal of biological chemistry 282, 30346-30356.

De Winter, F., Oudega, M., Lankhorst, A.J., Hamers, F.P., Blits, B., Ruitenberg, M.J., Pasterkamp, R.J., Gispen, W.H., and Verhaagen, J. (2002). Injury-induced class 3 semaphorin expression in the rat spinal cord. Exp Neurol 175, 61-75. de Wit, J., and Verhaagen, J. (2003). Role of semaphorins in the adult nervous system. Progress in neurobiology 71, 249-267.

168 Delcombel, R., Janssen, L., Vassy, R., Gammons, M., Haddad, O., Richard, B., Letourneur, D., Bates, D., Hendricks, C., Waltenberger, J., et al. (2013). New prospects in the roles of the C-terminal domains of VEGF-A and their cooperation for ligand binding, cellular signaling and vessels formation. Angiogenesis 16, 353- 371.

Dick, G., Tan, C.L., Alves, J.N., Ehlert, E.M., Miller, G.M., Hsieh-Wilson, L.C., Sugahara, K., Oosterhof, A., van Kuppevelt, T.H., Verhaagen, J., et al. (2013). Semaphorin 3A binds to the perineuronal nets via chondroitin sulfate type E motifs in rodent brains. The Journal of biological chemistry 288, 27384-27395.

Dosch, D.D., and Ballmer-Hofer, K. Transmembrane domain-mediated orientation of receptor monomers in active VEGFR-2 dimers. FASEB journal : official publication of the Federation of American Societies for Experimental Biology 24, 32-38.

Ebos, J.M., Lee, C.R., Cruz-Munoz, W., Bjarnason, G.A., Christensen, J.G., and Kerbel, R.S. (2009). Accelerated metastasis after short-term treatment with a potent inhibitor of tumor angiogenesis. Cancer cell 15, 232-239.

El Mourabit, H., Poinat, P., Koster, J., Sondermann, H., Wixler, V., Wegener, E., Laplantine, E., Geerts, D., Georges-Labouesse, E., Sonnenberg, A., et al. (2002). The PDZ domain of TIP-2/GIPC interacts with the C-terminus of the integrin alpha5 and alpha6 subunits. Matrix biology : journal of the International Society for Matrix Biology 21, 207-214.

Ellis, L.M. (2006). The role of neuropilins in cancer. Mol Cancer Ther 5, 1099-1107.

Emsley, P., Lohkamp, B., Scott, W.G., and Cowtan, K. (2010). Features and development of Coot. Acta crystallographica Section D, Biological crystallography 66, 486-501.

Erskine, L., Reijntjes, S., Pratt, T., Denti, L., Schwarz, Q., Vieira, J.M., Alakakone, B., Shewan, D., and Ruhrberg, C. (2011). VEGF signaling through neuropilin 1 guides commissural axon crossing at the optic chiasm. Neuron 70, 951-965.

Fairbrother, W.J., Champe, M.A., Christinger, H.W., Keyt, B.A., and Starovasnik, M.A. (1998). Solution structure of the heparin-binding domain of vascular endothelial growth factor. Structure 6, 637-648.

Fantin, A., Schwarz, Q., Davidson, K., Normando, E.M., Denti, L., and Ruhrberg, C. (2011). The cytoplasmic domain of neuropilin 1 is dispensable for angiogenesis, but promotes the spatial separation of retinal arteries and veins. Development 138, 4185-4191.

Fasman, G.D. (1989). Prediction of protein structure and the principles of protein conformation (New York: Plenum Press).

169 Favier, B., Alam, A., Barron, P., Bonnin, J., Laboudie, P., Fons, P., Mandron, M., Herault, J.P., Neufeld, G., Savi, P., et al. (2006). Neuropilin-2 interacts with VEGFR-2 and VEGFR-3 and promotes human endothelial cell survival and migration. Blood 108, 1243-1250.

Fawcett, J.W. (2006). Overcoming inhibition in the damaged spinal cord. Journal of neurotrauma 23, 371-383.

Ferrara, N., Gerber, H.P., and LeCouter, J. (2003). The biology of VEGF and its receptors. Nature medicine 9, 669-676.

Ferrara, N., and Henzel, W.J. (1989). Pituitary follicular cells secrete a novel heparin-binding growth factor specific for vascular endothelial cells. Biochemical and biophysical research communications 161, 851-858.

Ferrara, N., Hillan, K.J., and Novotny, W. (2005). Bevacizumab (Avastin), a humanized anti-VEGF monoclonal antibody for cancer therapy. Biochemical and biophysical research communications 333, 328-335.

Ferrell, R.E., Kimak, M.A., Lawrence, E.C., and Finegold, D.N. (2008). Candidate gene analysis in primary lymphedema. Lymphatic research and biology 6, 69-76.

Folkman, J. (1971). Tumor angiogenesis: therapeutic implications. The New England journal of medicine 285, 1182-1186.

Fuentes-Prior, P., Fujikawa, K., and Pratt, K.P. (2002). New insights into binding interfaces of coagulation factors V and VIII and their homologues lessons from high resolution crystal structures. Curr Protein Pept Sci 3, 313-339.

Fuh, G., Garcia, K.C., and de Vos, A.M. (2000). The interaction of neuropilin-1 with vascular endothelial growth factor and its receptor flt-1. The Journal of biological chemistry 275, 26690-26695.

Fujisawa, H. (2004). Discovery of semaphorin receptors, neuropilin and plexin, and their functions in neural development. Journal of neurobiology 59, 24-33.

Fukasawa, M., Matsushita, A., and Korc, M. (2007). Neuropilin-1 interacts with integrin beta1 and modulates pancreatic cancer cell growth, survival and invasion. Cancer biology & therapy 6, 1173-1180.

Gaboriaud, C., Gregory-Pauron, L., Teillet, F., Thielens, N.M., Bally, I., and Arlaud, G.J. (2011). Structure and properties of the Ca(2+)-binding CUB domain, a widespread ligand-recognition unit involved in major biological functions. The Biochemical journal 439, 185-193.

Gagnon, M.L., Bielenberg, D.R., Gechtman, Z., Miao, H.Q., Takashima, S., Soker, S., and Klagsbrun, M. (2000). Identification of a natural soluble neuropilin-1 that binds vascular endothelial growth factor: In vivo expression and antitumor activity.

170 Proceedings of the National Academy of Sciences of the United States of America 97, 2573-2578.

Geretti, E., Shimizu, A., Kurschat, P., and Klagsbrun, M. (2007). Site-directed mutagenesis in the B-neuropilin-2 domain selectively enhances its affinity to VEGF165, but not to semaphorin 3F. The Journal of biological chemistry 282, 25698-25707.

Geretti, E., van Meeteren, L.A., Shimizu, A., Dudley, A.C., Claesson-Welsh, L., and Klagsbrun, M. (2010). A mutated soluble neuropilin-2 B domain antagonizes vascular endothelial growth factor bioactivity and inhibits tumor progression. Molecular cancer research : MCR 8, 1063-1073.

Ghez, D., Lepelletier, Y., Lambert, S., Fourneau, J.M., Blot, V., Janvier, S., Arnulf, B., van Endert, P.M., Heveker, N., Pique, C., et al. (2006). Neuropilin-1 is involved in human T-cell lymphotropic virus type 1 entry. J Virol 80, 6844-6854.

Giger, R.J., Urquhart, E.R., Gillespie, S.K., Levengood, D.V., Ginty, D.D., and Kolodkin, A.L. (1998). Neuropilin-2 is a receptor for semaphorin IV: insight into the structural basis of receptor function and specificity. Neuron 21, 1079-1092.

Glinka, Y., and Prud'homme, G.J. (2008). Neuropilin-1 is a receptor for transforming growth factor beta-1, activates its latent form, and promotes regulatory T cell activity. Journal of leukocyte biology 84, 302-310.

Goel, H.L., Pursell, B., Chang, C., Shaw, L.M., Mao, J., Simin, K., Kumar, P., Vander Kooi, C.W., Shultz, L.D., Greiner, D.L., et al. (2013). GLI1 regulates a novel neuropilin-2/alpha6beta1 integrin based autocrine pathway that contributes to breast cancer initiation. EMBO molecular medicine 5, 488-508.

Goel, H.L., Pursell, B., Standley, C., Fogarty, K., and Mercurio, A.M. (2012). Neuropilin-2 regulates alpha6beta1 integrin in the formation of focal adhesions and signaling. Journal of cell science 125, 497-506.

Gordon, K., Schulte, D., Brice, G., Simpson, M.A., Roukens, M.G., van Impel, A., Connell, F., Kalidas, K., Jeffery, S., Mortimer, P.S., et al. (2013). Mutation in vascular endothelial growth factor-C, a ligand for vascular endothelial growth factor receptor-3, is associated with autosomal dominant milroy-like primary lymphedema. Circulation research 112, 956-960.

Gordon, V.M., Klimpel, K.R., Arora, N., Henderson, M.A., and Leppla, S.H. (1995). Proteolytic activation of bacterial toxins by eukaryotic cells is performed by furin and by additional cellular proteases. Infect Immun 63, 82-87.

Greenblatt, M., and Shubi, P. (1968). Tumor angiogenesis: transfilter diffusion studies in the hamster by the transparent chamber technique. Journal of the National Cancer Institute 41, 111-124.

171 Grunewald, F.S., Prota, A.E., Giese, A., and Ballmer-Hofer, K. (2010). Structure- function analysis of VEGF receptor activation and the role of coreceptors in angiogenic signaling. Biochimica et biophysica acta 1804, 567-580.

Gu, C., Limberg, B.J., Whitaker, G.B., Perman, B., Leahy, D.J., Rosenbaum, J.S., Ginty, D.D., and Kolodkin, A.L. (2002). Characterization of neuropilin-1 structural features that confer binding to semaphorin 3A and vascular endothelial growth factor 165. The Journal of biological chemistry 277, 18069-18076.

Gualandris, A., Noghero, A., Geuna, M., Arese, M., Valdembri, D., Serini, G., and Bussolino, F. (2009). Microenvironment drives the endothelial or neural fate of differentiating embryonic stem cells coexpressing neuropilin-1 and Flk-1. FASEB journal : official publication of the Federation of American Societies for Experimental Biology 23, 68-78.

Guo, H.F., Li, X., Parker, M.W., Waltenberger, J., Becker, P.M., and Vander Kooi, C.W. (2013). Mechanistic Basis for the Potent Anti-Angiogenic Activity of Semaphorin 3F. Biochemistry.

Guttmann-Raviv, N., Kessler, O., Shraga-Heled, N., Lange, T., Herzog, Y., and Neufeld, G. (2006). The neuropilins and their role in tumorigenesis and tumor progression. Cancer Lett 231, 1-11.

Hagberg, C.E., Falkevall, A., Wang, X., Larsson, E., Huusko, J., Nilsson, I., van Meeteren, L.A., Samen, E., Lu, L., Vanwildemeersch, M., et al. (2010). Vascular endothelial growth factor B controls endothelial fatty acid uptake. Nature 464, 917- 921.

Haiko, P., Makinen, T., Keskitalo, S., Taipale, J., Karkkainen, M.J., Baldwin, M.E., Stacker, S.A., Achen, M.G., and Alitalo, K. (2008). Deletion of vascular endothelial growth factor C (VEGF-C) and VEGF-D is not equivalent to VEGF receptor 3 deletion in mouse embryos. Molecular and cellular biology 28, 4843-4850.

Hamerlik, P., Lathia, J.D., Rasmussen, R., Wu, Q., Bartkova, J., Lee, M., Moudry, P., Bartek, J., Jr., Fischer, W., Lukas, J., et al. (2012). Autocrine VEGF-VEGFR2- Neuropilin-1 signaling promotes glioma stem-like cell viability and tumor growth. The Journal of experimental medicine 209, 507-520.

Hanchate, N.K., Giacobini, P., Lhuillier, P., Parkash, J., Espy, C., Fouveaut, C., Leroy, C., Baron, S., Campagne, C., Vanacker, C., et al. (2012). SEMA3A, a gene involved in axonal pathfinding, is mutated in patients with Kallmann syndrome. PLoS genetics 8, e1002896.

Harper, S.J., and Bates, D.O. (2008). VEGF-A splicing: the key to anti-angiogenic therapeutics? Nature reviews Cancer 8, 880-887.

Harris, N.C., Davydova, N., Roufail, S., Paquet-Fifield, S., Paavonen, K., Karnezis, T., Zhang, Y.F., Sato, T., Rothacker, J., Nice, E.C., et al. (2013). The propeptides

172 of VEGF-D determine heparin binding, receptor heterodimerization, and effects on tumor biology. The Journal of biological chemistry 288, 8176-8186.

Hasegawa, H., Tatsumi, M., Ogawa-Goto, K., Takahashi, H., Kojima, A., Iwasaki, T., Kurata, T., Sata, T., Takeuchi, T., Sheehy, N., et al. (2002). Processing of the HTLV-II envelope precursor glycoprotein gp63 by furin is essential for cell fusion activity. AIDS Res Hum Retroviruses 18, 1253-1260.

Haspel, N., Zanuy, D., Nussinov, R., Teesalu, T., Ruoslahti, E., and Aleman, C. (2011). Binding of a C-end rule peptide to the neuropilin-1 receptor: a molecular modeling approach. Biochemistry 50, 1755-1762.

He, H., Yang, T., Terman, J.R., and Zhang, X. (2009). Crystal structure of the plexin A3 intracellular region reveals an autoinhibited conformation through active site sequestration. Proceedings of the National Academy of Sciences of the United States of America 106, 15610-15615.

He, Z., and Tessier-Lavigne, M. (1997). Neuropilin is a receptor for the axonal chemorepellent Semaphorin III. Cell 90, 739-751.

Herzog, B., Pellet-Many, C., Britton, G., Hartzoulakis, B., and Zachary, I.C. (2011). VEGF binding to NRP1 is essential for VEGF stimulation of endothelial cell migration, complex formation between NRP1 and VEGFR2, and signaling via FAK Tyr407 phosphorylation. Molecular biology of the cell 22, 2766-2776.

Hillman, R.T., Feng, B.Y., Ni, J., Woo, W.M., Milenkovic, L., Hayden Gephart, M.G., Teruel, M.N., Oro, A.E., Chen, J.K., and Scott, M.P. (2011). Neuropilins are positive regulators of Hedgehog signal transduction. Genes & development 25, 2333-2346.

Holash, J., Davis, S., Papadopoulos, N., Croll, S.D., Ho, L., Russell, M., Boland, P., Leidich, R., Hylton, D., Burova, E., et al. (2002). VEGF-Trap: a VEGF blocker with potent antitumor effects. Proceedings of the National Academy of Sciences of the United States of America 99, 11393-11398.

Holmes, D.I., and Zachary, I. (2005). The vascular endothelial growth factor (VEGF) family: angiogenic factors in health and disease. Genome biology 6, 209.

Holmes, D.I., and Zachary, I.C. (2008). Vascular endothelial growth factor regulates stanniocalcin-1 expression via neuropilin-1-dependent regulation of KDR and synergism with fibroblast growth factor-2. Cellular signalling 20, 569-579.

Hosaka, M., Nagahama, M., Kim, W.S., Watanabe, T., Hatsuzawa, K., Ikemizu, J., Murakami, K., and Nakayama, K. (1991). Arg-X-Lys/Arg-Arg motif as a signal for precursor cleavage catalyzed by furin within the constitutive secretory pathway. The Journal of biological chemistry 266, 12127-12130.

173 Hota, P.K., and Buck, M. (2012). Plexin structures are coming: opportunities for multilevel investigations of semaphorin guidance receptors, their cell signaling mechanisms, and functions. Cellular and molecular life sciences : CMLS 69, 3765- 3805.

Houck, K.A., Leung, D.W., Rowland, A.M., Winer, J., and Ferrara, N. (1992). Dual regulation of vascular endothelial growth factor bioavailability by genetic and proteolytic mechanisms. The Journal of biological chemistry 267, 26031-26037.

Hu, B., Guo, P., Bar-Joseph, I., Imanishi, Y., Jarzynka, M.J., Bogler, O., Mikkelsen, T., Hirose, T., Nishikawa, R., and Cheng, S.Y. (2007). Neuropilin-1 promotes human glioma progression through potentiating the activity of the HGF/SF autocrine pathway. Oncogene 26, 5577-5586.

Janin, J. (1999). Wet and dry interfaces: the role of solvent in protein–protein and protein–DNA recognition. Structure 7, R277-R279.

Janssen, B.J., Malinauskas, T., Weir, G.A., Cader, M.Z., Siebold, C., and Jones, E.Y. (2012). Neuropilins lock secreted semaphorins onto plexins in a ternary signaling complex. Nature structural & molecular biology 19, 1293-1299.

Janssen, B.J., Robinson, R.A., Perez-Branguli, F., Bell, C.H., Mitchell, K.J., Siebold, C., and Jones, E.Y. (2010). Structural basis of semaphorin-plexin signalling. Nature 467, 1118-1122.

Jardin, B.A., Zhao, Y., Selvaraj, M., Montes, J., Tran, R., Prakash, S., and Elias, C.B. (2008). Expression of SEAP (secreted alkaline phosphatase) by baculovirus mediated transduction of HEK 293 cells in a hollow fiber bioreactor system. Journal of biotechnology 135, 272-280.

Jarvis, A., Allerston, C.K., Jia, H., Herzog, B., Garza-Garcia, A., Winfield, N., Ellard, K., Aqil, R., Lynch, R., Chapman, C., et al. (2010). Small molecule inhibitors of the neuropilin-1 vascular endothelial growth factor A (VEGF-A) interaction. Journal of medicinal chemistry 53, 2215-2226.

Jeltsch, M., Kaipainen, A., Joukov, V., Meng, X., Lakso, M., Rauvala, H., Swartz, M., Fukumura, D., Jain, R.K., and Alitalo, K. (1997). Hyperplasia of lymphatic vessels in VEGF-C transgenic mice. Science 276, 1423-1425.

Jia, H., Bagherzadeh, A., Hartzoulakis, B., Jarvis, A., Lohr, M., Shaikh, S., Aqil, R., Cheng, L., Tickner, M., Esposito, D., et al. (2006). Characterization of a bicyclic peptide neuropilin-1 (NP-1) antagonist (EG3287) reveals importance of vascular endothelial growth factor exon 8 for NP-1 binding and role of NP-1 in KDR signaling. The Journal of biological chemistry 281, 13493-13502.

Jones, E.A., Yuan, L., Breant, C., Watts, R.J., and Eichmann, A. (2008). Separating genetic and hemodynamic defects in neuropilin 1 knockout embryos. Development 135, 2479-2488.

174 Joukov, V., Pajusola, K., Kaipainen, A., Chilov, D., Lahtinen, I., Kukk, E., Saksela, O., Kalkkinen, N., and Alitalo, K. (1996). A novel vascular endothelial growth factor, VEGF-C, is a ligand for the Flt4 (VEGFR-3) and KDR (VEGFR-2) receptor tyrosine kinases. The EMBO journal 15, 1751.

Joukov, V., Sorsa, T., Kumar, V., Jeltsch, M., Claesson-Welsh, L., Cao, Y., Saksela, O., Kalkkinen, N., and Alitalo, K. (1997). Proteolytic processing regulates receptor specificity and activity of VEGF-C. The EMBO journal 16, 3898-3911.

Joyal, J.S., Sitaras, N., Binet, F., Rivera, J.C., Stahl, A., Zaniolo, K., Shao, Z., Polosa, A., Zhu, T., Hamel, D., et al. (2011). Ischemic neurons prevent vascular regeneration of neural tissue by secreting semaphorin 3A. Blood 117, 6024-6035.

Kaneko, S., Iwanami, A., Nakamura, M., Kishino, A., Kikuchi, K., Shibata, S., Okano, H.J., Ikegami, T., Moriya, A., Konishi, O., et al. (2006). A selective Sema3A inhibitor enhances regenerative responses and functional recovery of the injured spinal cord. Nature medicine 12, 1380-1389.

Karkkainen, M.J., Ferrell, R.E., Lawrence, E.C., Kimak, M.A., Levinson, K.L., McTigue, M.A., Alitalo, K., and Finegold, D.N. (2000). Missense mutations interfere with VEGFR-3 signalling in primary lymphoedema. Nature genetics 25, 153-159.

Karkkainen, M.J., Haiko, P., Sainio, K., Partanen, J., Taipale, J., Petrova, T.V., Jeltsch, M., Jackson, D.G., Talikka, M., Rauvala, H., et al. (2004). Vascular endothelial growth factor C is required for sprouting of the first lymphatic vessels from embryonic veins. Nature immunology 5, 74-80.

Karkkainen, M.J., Saaristo, A., Jussila, L., Karila, K.A., Lawrence, E.C., Pajusola, K., Bueler, H., Eichmann, A., Kauppinen, R., Kettunen, M.I., et al. (2001). A model for gene therapy of human hereditary lymphedema. Proceedings of the National Academy of Sciences of the United States of America 98, 12677-12682.

Karpanen, T., Heckman, C.A., Keskitalo, S., Jeltsch, M., Ollila, H., Neufeld, G., Tamagnone, L., and Alitalo, K. (2006). Functional interaction of VEGF-C and VEGF-D with neuropilin receptors. FASEB journal : official publication of the Federation of American Societies for Experimental Biology 20, 1462-1472.

Karplus, P.A., and Faerman, C. (1994). Ordered water in macromolecular structure. Current opinion in structural biology 4, 770-776.

Kawasaki, T., Kitsukawa, T., Bekku, Y., Matsuda, Y., Sanbo, M., Yagi, T., and Fujisawa, H. (1999). A requirement for neuropilin-1 in embryonic vessel formation. Development 126, 4895-4902.

Kendall, R.L., and Thomas, K.A. (1993). Inhibition of vascular endothelial cell growth factor activity by an endogenously encoded soluble receptor. Proceedings of the National Academy of Sciences of the United States of America 90, 10705- 10709.

175 Kessler, O., Shraga-Heled, N., Lange, T., Gutmann-Raviv, N., Sabo, E., Baruch, L., Machluf, M., and Neufeld, G. (2004). Semaphorin-3F is an inhibitor of tumor angiogenesis. Cancer research 64, 1008-1015.

Keyt, B.A., Berleau, L.T., Nguyen, H.V., Chen, H., Heinsohn, H., Vandlen, R., and Ferrara, N. (1996). The carboxyl-terminal domain (111-165) of vascular endothelial growth factor is critical for its mitogenic potency. The Journal of biological chemistry 271, 7788-7795.

Khatib, A.M., Lahlil, R., Scamuffa, N., Akimenko, M.A., Ernest, S., Lomri, A., Lalou, C., Seidah, N.G., Villoutreix, B.O., Calvo, F., et al. (2010). Zebrafish ProVEGF-C expression, proteolytic processing and inhibitory effect of unprocessed ProVEGF- C during fin regeneration. PloS one 5, e11438.

Kiedzierska, A., Smietana, K., Czepczynska, H., and Otlewski, J. (2007). Structural similarities and functional diversity of eukaryotic discoidin-like domains. Biochimica et biophysica acta 1774, 1069-1078.

Kikuchi, K., Kishino, A., Konishi, O., Kumagai, K., Hosotani, N., Saji, I., Nakayama, C., and Kimura, T. (2003). In vitro and in vivo characterization of a novel semaphorin 3A inhibitor, SM-216289 or xanthofulvin. The Journal of biological chemistry 278, 42985-42991.

Kitsukawa, T., Shimizu, M., Sanbo, M., Hirata, T., Taniguchi, M., Bekku, Y., Yagi, T., and Fujisawa, H. (1997). Neuropilin-semaphorin III/D-mediated chemorepulsive signals play a crucial role in peripheral nerve projection in mice. Neuron 19, 995-1005.

Klagsbrun, M., Takashima, S., and Mamluk, R. (2002). The role of neuropilin in vascular and tumor biology. Advances in experimental medicine and biology 515, 33-48.

Klostermann, A., Lohrum, M., Adams, R.H., and Puschel, A.W. (1998). The chemorepulsive activity of the axonal guidance signal semaphorin D requires dimerization. The Journal of biological chemistry 273, 7326-7331.

Kolodkin, A.L., Levengood, D.V., Rowe, E.G., Tai, Y.T., Giger, R.J., and Ginty, D.D. (1997). Neuropilin is a semaphorin III receptor. Cell 90, 753-762.

Kolodkin, A.L., Matthes, D.J., and Goodman, C.S. (1993). The semaphorin genes encode a family of transmembrane and secreted growth cone guidance molecules. Cell 75, 1389-1399.

Koncina, E., Roth, L., Gonthier, B., and Bagnard, D. (2007). Role of semaphorins during axon growth and guidance. Advances in experimental medicine and biology 621, 50-64.

176 Koppel, A.M., Feiner, L., Kobayashi, H., and Raper, J.A. (1997). A 70 amino acid region within the semaphorin domain activates specific cellular response of semaphorin family members. Neuron 19, 531-537.

Kreuter, M., Woelke, K., Bieker, R., Schliemann, C., Steins, M., Buechner, T., Berdel, W.E., and Mesters, R.M. (2006). Correlation of neuropilin-1 overexpression to survival in acute myeloid leukemia. Leukemia.

Krilleke, D., DeErkenez, A., Schubert, W., Giri, I., Robinson, G.S., Ng, Y.S., and Shima, D.T. (2007). Molecular mapping and functional characterization of the VEGF164 heparin-binding domain. The Journal of biological chemistry 282, 28045-28056.

Kukk, E., Lymboussaki, A., Taira, S., Kaipainen, A., Jeltsch, M., Joukov, V., and Alitalo, K. (1996). VEGF-C receptor binding and pattern of expression with VEGFR-3 suggests a role in lymphatic vascular development. Development 122, 3829-3837.

Lambert, S., Bouttier, M., Vassy, R., Seigneuret, M., Petrow-Sadowski, C., Janvier, S., Heveker, N., Ruscetti, F.W., Perret, G., Jones, K.S., et al. (2009). HTLV-1 uses HSPG and neuropilin-1 for entry by molecular mimicry of VEGF165. Blood 113, 5176-5185.

Laskowski, R.A., and Swindells, M.B. (2011). LigPlot+: multiple ligand-protein interaction diagrams for drug discovery. Journal of chemical information and modeling 51, 2778-2786.

Latil, A., Bieche, I., Pesche, S., Valeri, A., Fournier, G., Cussenot, O., and Lidereau, R. (2000). VEGF overexpression in clinically localized prostate tumors and neuropilin-1 overexpression in metastatic forms. International journal of cancer Journal international du cancer 89, 167-171.

Leahy, D.J., Dann, C.E., 3rd, Longo, P., Perman, B., and Ramyar, K.X. (2000). A mammalian expression vector for expression and purification of secreted proteins for structural studies. Protein Expr Purif 20, 500-506.

Lee, C.C., Kreusch, A., McMullan, D., Ng, K., and Spraggon, G. (2003). Crystal structure of the human neuropilin-1 b1 domain. Structure 11, 99-108.

Lee, C.G., Ma, B., Takyar, S., Ahangari, F., Delacruz, C., He, C.H., and Elias, J.A. (2011). Studies of vascular endothelial growth factor in asthma and chronic obstructive pulmonary disease. Proceedings of the American Thoracic Society 8, 512-515.

Leppanen, V.M., Prota, A.E., Jeltsch, M., Anisimov, A., Kalkkinen, N., Strandin, T., Lankinen, H., Goldman, A., Ballmer-Hofer, K., and Alitalo, K. (2010). Structural determinants of growth factor binding and specificity by VEGF receptor 2.

177 Proceedings of the National Academy of Sciences of the United States of America 107, 2425-2430.

Leppanen, V.M., Tvorogov, D., Kisko, K., Prota, A.E., Jeltsch, M., Anisimov, A., Markovic-Mueller, S., Stuttfeld, E., Goldie, K.N., Ballmer-Hofer, K., et al. (2013). Structural and mechanistic insights into VEGF receptor 3 ligand binding and activation. Proceedings of the National Academy of Sciences of the United States of America 110, 12960-12965.

Leung, D.W., Cachianes, G., Kuang, W.J., Goeddel, D.V., and Ferrara, N. (1989). Vascular endothelial growth factor is a secreted angiogenic mitogen. Science 246, 1306-1309.

Liang, W.C., Dennis, M.S., Stawicki, S., Chanthery, Y., Pan, Q., Chen, Y., Eigenbrot, C., Yin, J., Koch, A.W., Wu, X., et al. (2007). Function blocking antibodies to neuropilin-1 generated from a designed human synthetic antibody phage library. Journal of molecular biology 366, 815-829.

Lin, J., Lalani, A.S., Harding, T.C., Gonzalez, M., Wu, W.W., Luan, B., Tu, G.H., Koprivnikar, K., VanRoey, M.J., He, Y., et al. (2005). Inhibition of lymphogenous metastasis using adeno-associated virus-mediated gene transfer of a soluble VEGFR-3 decoy receptor. Cancer research 65, 6901-6909.

Lohela, M., Bry, M., Tammela, T., and Alitalo, K. (2009). VEGFs and receptors involved in angiogenesis versus lymphangiogenesis. Current opinion in cell biology 21, 154-165.

Longo, P.A., Kavran, J.M., Kim, M.S., and Leahy, D.J. (2013). Transient mammalian cell transfection with polyethylenimine (PEI). Methods in enzymology 529, 227-240.

Love, C.A., Harlos, K., Mavaddat, N., Davis, S.J., Stuart, D.I., Jones, E.Y., and Esnouf, R.M. (2003). The ligand-binding face of the semaphorins revealed by the high-resolution crystal structure of SEMA4D. Nature structural biology 10, 843- 848.

Lymboussaki, A., Olofsson, B., Eriksson, U., and Alitalo, K. (1999). Vascular endothelial growth factor (VEGF) and VEGF-C show overlapping binding sites in embryonic endothelia and distinct sites in differentiated adult endothelia. Circulation research 85, 992-999.

Maiti, R., Van Domselaar, G.H., Zhang, H., and Wishart, D.S. (2004). SuperPose: a simple server for sophisticated structural superposition. Nucleic acids research 32, W590-594.

Makinen, T., Olofsson, B., Karpanen, T., Hellman, U., Soker, S., Klagsbrun, M., Eriksson, U., and Alitalo, K. (1999). Differential binding of vascular endothelial

178 growth factor B splice and proteolytic isoforms to neuropilin-1. The Journal of biological chemistry 274, 21217-21222.

Malo, N., Hanley, J.A., Cerquozzi, S., Pelletier, J., and Nadon, R. (2006). Statistical practice in high-throughput screening data analysis. Nature biotechnology 24, 167- 175.

Mamluk, R., Gechtman, Z., Kutcher, M.E., Gasiunas, N., Gallagher, J., and Klagsbrun, M. (2002). Neuropilin-1 binds vascular endothelial growth factor 165, placenta growth factor-2, and heparin via its b1b2 domain. The Journal of biological chemistry 277, 24818-24825.

Matsushita, A., Gotze, T., and Korc, M. (2007). Hepatocyte growth factor-mediated cell invasion in pancreatic cancer cells is dependent on neuropilin-1. Cancer research 67, 10309-10316.

Mattila, M.M., Ruohola, J.K., Karpanen, T., Jackson, D.G., Alitalo, K., and Harkonen, P.L. (2002). VEGF-C induced lymphangiogenesis is associated with lymph node metastasis in orthotopic MCF-7 tumors. International journal of cancer Journal international du cancer 98, 946-951.

McColl, B.K., Baldwin, M.E., Roufail, S., Freeman, C., Moritz, R.L., Simpson, R.J., Alitalo, K., Stacker, S.A., and Achen, M.G. (2003). Plasmin activates the lymphangiogenic growth factors VEGF-C and VEGF-D. The Journal of experimental medicine 198, 863-868.

McCoy, A.J., Grosse-Kunstleve, R.W., Adams, P.D., Winn, M.D., Storoni, L.C., and Read, R.J. (2007). Phaser crystallographic software. J Appl Cryst 40, 658-674.

McTigue, M.A., Wickersham, J.A., Pinko, C., Showalter, R.E., Parast, C.V., Tempczyk-Russell, A., Gehring, M.R., Mroczkowski, B., Kan, C.C., Villafranca, J.E., et al. (1999). Crystal structure of the kinase domain of human vascular endothelial growth factor receptor 2: a key enzyme in angiogenesis. Structure 7, 319-330.

Merte, J., Wang, Q., Vander Kooi, C.W., Sarsfield, S., Leahy, D.J., Kolodkin, A.L., and Ginty, D.D. (2010). A forward genetic screen in mice identifies Sema3A(K108N), which binds to neuropilin-1 but cannot signal. The Journal of neuroscience : the official journal of the Society for Neuroscience 30, 5767-5775.

Miao, H.Q., Lee, P., Lin, H., Soker, S., and Klagsbrun, M. (2000). Neuropilin-1 expression by tumor cells promotes tumor angiogenesis and progression. FASEB journal : official publication of the Federation of American Societies for Experimental Biology 14, 2532-2539.

Miao, H.Q., Soker, S., Feiner, L., Alonso, J.L., Raper, J.A., and Klagsbrun, M. (1999). Neuropilin-1 mediates collapsin-1/semaphorin III inhibition of endothelial

179 cell motility: functional competition of collapsin-1 and vascular endothelial growth factor-165. The Journal of cell biology 146, 233-242.

Miaskowski, C., Dodd, M., Paul, S.M., West, C., Hamolsky, D., Abrams, G., Cooper, B.A., Elboim, C., Neuhaus, J., Schmidt, B.L., et al. (2013). Lymphatic and angiogenic candidate genes predict the development of secondary lymphedema following breast cancer surgery. PloS one 8, e60164.

Migdal, M., Huppertz, B., Tessler, S., Comforti, A., Shibuya, M., Reich, R., Baumann, H., and Neufeld, G. (1998). Neuropilin-1 is a placenta growth factor-2 receptor. The Journal of biological chemistry 273, 22272-22278.

Molloy, S.S., Bresnahan, P.A., Leppla, S.H., Klimpel, K.R., and Thomas, G. (1992). Human furin is a calcium-dependent serine endoprotease that recognizes the sequence Arg-X-X-Arg and efficiently cleaves anthrax toxin protective antigen. The Journal of biological chemistry 267, 16396-16402.

Muders, M.H., Zhang, H., Wang, E., Tindall, D.J., and Datta, K. (2009). Vascular endothelial growth factor-C protects prostate cancer cells from oxidative stress by the activation of mammalian target of rapamycin complex-2 and AKT-1. Cancer research 69, 6042-6048.

Muller, Y.A., Li, B., Christinger, H.W., Wells, J.A., Cunningham, B.C., and de Vos, A.M. (1997). Vascular endothelial growth factor: crystal structure and functional mapping of the kinase domain receptor binding site. Proceedings of the National Academy of Sciences of the United States of America 94, 7192-7197.

Murga, M., Fernandez-Capetillo, O., and Tosato, G. (2005). Neuropilin-1 regulates attachment in human endothelial cells independently of vascular endothelial growth factor receptor-2. Blood 105, 1992-1999.

Murshudov, G.N. (1997). Refinement of macromolecular structures by the maximum-likelihood method. Acta crystallographica Section D, Biological crystallography 53, 240-255.

Nakamura, F., Tanaka, M., Takahashi, T., Kalb, R.G., and Strittmatter, S.M. (1998). Neuropilin-1 extracellular domains mediate semaphorin D/III-induced growth cone collapse. Neuron 21, 1093-1100.

Narazaki, M., and Tosato, G. (2006). Ligand-induced internalization selects use of common receptor neuropilin-1 by VEGF165 and semaphorin3A. Blood 107, 3892- 3901.

Neufeld, G., Shraga-Heled, N., Lange, T., Guttmann-Raviv, N., Herzog, Y., and Kessler, O. (2005). Semaphorins in cancer. Frontiers in bioscience : a journal and virtual library 10, 751-760.

180 Niclou, S.P., Ehlert, E.M., and Verhaagen, J. (2006). Chemorepellent axon guidance molecules in spinal cord injury. Journal of neurotrauma 23, 409-421.

Niclou, S.P., Franssen, E.H., Ehlert, E.M., Taniguchi, M., and Verhaagen, J. (2003). Meningeal cell-derived semaphorin 3A inhibits neurite outgrowth. Molecular and cellular neurosciences 24, 902-912.

Nowak, D.G., Woolard, J., Amin, E.M., Konopatskaya, O., Saleem, M.A., Churchill, A.J., Ladomery, M.R., Harper, S.J., and Bates, D.O. (2008). Expression of pro- and anti-angiogenic isoforms of VEGF is differentially regulated by splicing and growth factors. Journal of cell science 121, 3487-3495.

Ochiumi, T., Kitadai, Y., Tanaka, S., Akagi, M., Yoshihara, M., and Chayama, K. (2006). Neuropilin-1 is involved in regulation of apoptosis and migration of human colon cancer. Int J Oncol 29, 105-116.

Olofsson, B., Pajusola, K., Kaipainen, A., von Euler, G., Joukov, V., Saksela, O., Orpana, A., Pettersson, R.F., Alitalo, K., and Eriksson, U. (1996a). Vascular endothelial growth factor B, a novel growth factor for endothelial cells. Proceedings of the National Academy of Sciences of the United States of America 93, 2576- 2581.

Olofsson, B., Pajusola, K., von Euler, G., Chilov, D., Alitalo, K., and Eriksson, U. (1996b). Genomic organization of the mouse and human genes for vascular endothelial growth factor B (VEGF-B) and characterization of a second splice isoform. The Journal of biological chemistry 271, 19310-19317.

Otwinowski, Z., and Minor, W. (1997). Processing of X-ray diffraction data collected in oscillation mode. Macromolecular Crystallography, Pt A 276, 307-326.

Paez-Ribes, M., Allen, E., Hudock, J., Takeda, T., Okuyama, H., Vinals, F., Inoue, M., Bergers, G., Hanahan, D., and Casanovas, O. (2009). Antiangiogenic therapy elicits malignant progression of tumors to increased local invasion and distant metastasis. Cancer cell 15, 220-231.

Pan, Q., Chanthery, Y., Liang, W.C., Stawicki, S., Mak, J., Rathore, N., Tong, R.K., Kowalski, J., Yee, S.F., Pacheco, G., et al. (2007a). Blocking neuropilin-1 function has an additive effect with anti-VEGF to inhibit tumor growth. Cancer cell 11, 53- 67.

Pan, Q., Chathery, Y., Wu, Y., Rathore, N., Tong, R.K., Peale, F., Bagri, A., Tessier-Lavigne, M., Koch, A.W., and Watts, R.J. (2007b). Neuropilin-1 binds to VEGF121 and regulates endothelial cell migration and sprouting. The Journal of biological chemistry 282, 24049-24056.

Parker, M.W., Guo, H.F., Li, X., Linkugel, A.D., and Vander Kooi, C.W. (2012a). Function of members of the neuropilin family as essential pleiotropic cell surface receptors. Biochemistry 51, 9437-9446.

181 Parker, M.W., Hellman, L.M., Xu, P., Fried, M.G., and Vander Kooi, C.W. (2010). Furin processing of semaphorin 3F determines its anti-angiogenic activity by regulating direct binding and competition for neuropilin. Biochemistry 49, 4068- 4075.

Parker, M.W., Linkugel, A.D., and Vander Kooi, C.W. (2013). Effect of C-Terminal Sequence on Competitive Semaphorin Binding to Neuropilin-1. Journal of molecular biology.

Parker, M.W., and Vander Kooi, C.W. (2014). Microplate-based screening for small molecule inhibitors of Neuropilin-2/VEGF-C interactions. Analytical biochemistry.

Parker, M.W., Xu, P., Guo, H.F., and Vander Kooi, C.W. (2012b). Mechanism of Selective VEGF-A Binding by Neuropilin-1 Reveals a Basis for Specific Ligand Inhibition. PloS one 7, e49177.

Parker, M.W., Xu, P., Li, X., and Vander Kooi, C.W. (2012c). Structural basis for selective vascular endothelial growth factor-A (VEGF-A) binding to neuropilin-1. The Journal of biological chemistry 287, 11082-11089.

Pasterkamp, R.J., and Giger, R.J. (2009). Semaphorin function in neural plasticity and disease. Curr Opin Neurobiol 19, 263-274.

Pasterkamp, R.J., Giger, R.J., Ruitenberg, M.J., Holtmaat, A.J., De Wit, J., De Winter, F., and Verhaagen, J. (1999). Expression of the gene encoding the chemorepellent semaphorin III is induced in the fibroblast component of neural scar tissue formed following injuries of adult but not neonatal CNS. Molecular and cellular neurosciences 13, 143-166.

Pellet-Many, C., Frankel, P., Jia, H., and Zachary, I. (2008). Neuropilins: structure, function and role in disease. The Biochemical journal 411, 211-226.

Perret, G.Y., Starzec, A., Hauet, N., Vergote, J., Le Pecheur, M., Vassy, R., Leger, G., Verbeke, K.A., Bormans, G., Nicolas, P., et al. (2004). In vitro evaluation and biodistribution of a 99mTc-labeled anti-VEGF peptide targeting neuropilin-1. Nuclear medicine and biology 31, 575-581.

Potiron, V.A., Roche, J., and Drabkin, H.A. (2009). Semaphorins and their receptors in lung cancer. Cancer Lett 273, 1-14.

Prahst, C., Heroult, M., Lanahan, A.A., Uziel, N., Kessler, O., Shraga-Heled, N., Simons, M., Neufeld, G., and Augustin, H.G. (2008). Neuropilin-1-VEGFR-2 complexing requires the PDZ-binding domain of neuropilin-1. The Journal of biological chemistry 283, 25110-25114.

Puschel, A.W. (2002). The function of neuropilin/plexin complexes. Advances in experimental medicine and biology 515, 71-80.

182 Rey-Gallardo, A., Escribano, C., Delgado-Martin, C., Rodriguez-Fernandez, J.L., Gerardy-Schahn, R., Rutishauser, U., Corbi, A.L., and Vega, M.A. (2010). Polysialylated neuropilin-2 enhances human dendritic cell migration through the basic C-terminal region of CCL21. Glycobiology 20, 1139-1146.

Robinson, S.D., Reynolds, L.E., Kostourou, V., Reynolds, A.R., da Silva, R.G., Tavora, B., Baker, M., Marshall, J.F., and Hodivala-Dilke, K.M. (2009). Alphav beta3 integrin limits the contribution of neuropilin-1 to vascular endothelial growth factor-induced angiogenesis. The Journal of biological chemistry 284, 33966- 33981.

Rose-John, S., and Heinrich, P.C. (1994). Soluble receptors for cytokines and growth factors: generation and biological function. The Biochemical journal 300 ( Pt 2), 281-290.

Rosenfeld, P.J., Brown, D.M., Heier, J.S., Boyer, D.S., Kaiser, P.K., Chung, C.Y., and Kim, R.Y. (2006). Ranibizumab for neovascular age-related macular degeneration. The New England journal of medicine 355, 1419-1431.

Rossignol, M., Gagnon, M.L., and Klagsbrun, M. (2000). Genomic organization of human neuropilin-1 and neuropilin-2 genes: identification and distribution of splice variants and soluble isoforms. Genomics 70, 211-222.

Roth, L., Agemy, L., Kotamraju, V.R., Braun, G., Teesalu, T., Sugahara, K.N., Hamzah, J., and Ruoslahti, E. (2012). Transtumoral targeting enabled by a novel neuropilin-binding peptide. Oncogene 31, 3754-3763.

Roth, L., Nasarre, C., Dirrig-Grosch, S., Aunis, D., Cremel, G., Hubert, P., and Bagnard, D. (2008). Transmembrane domain interactions control biological functions of neuropilin-1. Molecular biology of the cell 19, 646-654.

Ruch, C., Skiniotis, G., Steinmetz, M.O., Walz, T., and Ballmer-Hofer, K. (2007). Structure of a VEGF-VEGF receptor complex determined by electron microscopy. Nature structural & molecular biology 14, 249-250.

Saaristo, A., Karkkainen, M.J., and Alitalo, K. (2002). Insights into the molecular pathogenesis and targeted treatment of lymphedema. Annals of the New York Academy of Sciences 979, 94-110.

Sahay, A., Molliver, M.E., Ginty, D.D., and Kolodkin, A.L. (2003). Semaphorin 3F is critical for development of limbic system circuitry and is required in neurons for selective CNS axon guidance events. The Journal of neuroscience : the official journal of the Society for Neuroscience 23, 6671-6680.

Sakurai, A., Doci, C.L., and Gutkind, J.S. (2012). Semaphorin signaling in angiogenesis, lymphangiogenesis and cancer. Cell research 22, 23-32.

183 Salikhova, A., Wang, L., Lanahan, A.A., Liu, M., Simons, M., Leenders, W.P., Mukhopadhyay, D., and Horowitz, A. (2008). Vascular endothelial growth factor and semaphorin induce neuropilin-1 endocytosis via separate pathways. Circulation research 103, e71-79.

Schwarz, Q., Gu, C., Fujisawa, H., Sabelko, K., Gertsenstein, M., Nagy, A., Taniguchi, M., Kolodkin, A.L., Ginty, D.D., Shima, D.T., et al. (2004). Vascular endothelial growth factor controls neuronal migration and cooperates with Sema3A to pattern distinct compartments of the facial nerve. Genes & development 18, 2822-2834.

Senger, D.R., Galli, S.J., Dvorak, A.M., Perruzzi, C.A., Harvey, V.S., and Dvorak, H.F. (1983). Tumor cells secrete a vascular permeability factor that promotes accumulation of ascites fluid. Science 219, 983-985.

Serini, G., Bussolino, F., Maione, F., and Giraudo, E. (2013). Class 3 semaphorins: physiological vascular normalizing agents for anti-cancer therapy. Journal of internal medicine 273, 138-155.

Serini, G., Valdembri, D., Zanivan, S., Morterra, G., Burkhardt, C., Caccavari, F., Zammataro, L., Primo, L., Tamagnone, L., Logan, M., et al. (2003). Class 3 semaphorins control vascular morphogenesis by inhibiting integrin function. Nature 424, 391-397.

Shearer, M.C., Niclou, S.P., Brown, D., Asher, R.A., Holtmaat, A.J., Levine, J.M., Verhaagen, J., and Fawcett, J.W. (2003). The astrocyte/meningeal cell interface is a barrier to neurite outgrowth which can be overcome by manipulation of inhibitory molecules or axonal signalling pathways. Molecular and cellular neurosciences 24, 913-925.

Shintani, Y., Takashima, S., Asano, Y., Kato, H., Liao, Y., Yamazaki, S., Tsukamoto, O., Seguchi, O., Yamamoto, H., Fukushima, T., et al. (2006). Glycosaminoglycan modification of neuropilin-1 modulates VEGFR2 signaling. The EMBO journal 25, 3045-3055.

Shraga-Heled, N., Kessler, O., Prahst, C., Kroll, J., Augustin, H., and Neufeld, G. (2007). Neuropilin-1 and neuropilin-2 enhance VEGF121 stimulated signal transduction by the VEGFR-2 receptor. FASEB journal : official publication of the Federation of American Societies for Experimental Biology 21, 915-926.

Soker, S., Fidder, H., Neufeld, G., and Klagsbrun, M. (1996). Characterization of novel vascular endothelial growth factor (VEGF) receptors on tumor cells that bind VEGF165 via its exon 7-encoded domain. The Journal of biological chemistry 271, 5761-5767.

Soker, S., Gollamudi-Payne, S., Fidder, H., Charmahelli, H., and Klagsbrun, M. (1997). Inhibition of vascular endothelial growth factor (VEGF)-induced endothelial

184 cell proliferation by a peptide corresponding to the exon 7-encoded domain of VEGF165. The Journal of biological chemistry 272, 31582-31588.

Soker, S., Miao, H.Q., Nomi, M., Takashima, S., and Klagsbrun, M. (2002). VEGF165 mediates formation of complexes containing VEGFR-2 and neuropilin- 1 that enhance VEGF165-receptor binding. J Cell Biochem 85, 357-368.

Soker, S., Takashima, S., Miao, H.Q., Neufeld, G., and Klagsbrun, M. (1998). Neuropilin-1 is expressed by endothelial and tumor cells as an isoform-specific receptor for vascular endothelial growth factor. Cell 92, 735-745.

Song, H., Ming, G., He, Z., Lehmann, M., McKerracher, L., Tessier-Lavigne, M., and Poo, M. (1998). Conversion of neuronal growth cone responses from repulsion to attraction by cyclic nucleotides. Science 281, 1515-1518.

Stacker, S.A., Achen, M.G., Jussila, L., Baldwin, M.E., and Alitalo, K. (2002). Lymphangiogenesis and cancer metastasis. Nature reviews Cancer 2, 573-583.

Stacker, S.A., Stenvers, K., Caesar, C., Vitali, A., Domagala, T., Nice, E., Roufail, S., Simpson, R.J., Moritz, R., Karpanen, T., et al. (1999). Biosynthesis of vascular endothelial growth factor-D involves proteolytic processing which generates non- covalent homodimers. The Journal of biological chemistry 274, 32127-32136.

Stanton, M.J., Dutta, S., Zhang, H., Polavaram, N.S., Leontovich, A.A., Honscheid, P., Sinicrope, F.A., Tindall, D.J., Muders, M.H., and Datta, K. (2012). Autophagy control by VEGF-C/NRP-2 axis in cancer and its implication for treatment resistance. Cancer research.

Starzec, A., Ladam, P., Vassy, R., Badache, S., Bouchemal, N., Navaza, A., du Penhoat, C.H., and Perret, G.Y. (2007). Structure-function analysis of the antiangiogenic ATWLPPR peptide inhibiting VEGF(165) binding to neuropilin-1 and molecular dynamics simulations of the ATWLPPR/neuropilin-1 complex. Peptides 28, 2397-2402.

Starzec, A., Vassy, R., Martin, A., Lecouvey, M., Di Benedetto, M., Crepin, M., and Perret, G.Y. (2006). Antiangiogenic and antitumor activities of peptide inhibiting the vascular endothelial growth factor binding to neuropilin-1. Life sciences 79, 2370-2381.

Staton, C.A., Kumar, I., Reed, M.W., and Brown, N.J. (2007). Neuropilins in physiological and pathological angiogenesis. The Journal of pathology 212, 237- 248.

Stephenson, J.M., Banerjee, S., Saxena, N.K., Cherian, R., and Banerjee, S.K. (2002). Neuropilin-1 is differentially expressed in myoepithelial cells and vascular smooth muscle cells in preneoplastic and neoplastic human breast: a possible marker for the progression of breast cancer. International journal of cancer Journal international du cancer 101, 409-414.

185 Stewart, M.W. (2012). Aflibercept (VEGF Trap-eye): the newest anti-VEGF drug. The British journal of ophthalmology 96, 1157-1158.

Stewart, M.W., Grippon, S., and Kirkpatrick, P. (2012). Aflibercept. Nature reviews Drug discovery 11, 269-270.

Stringer, S.E. (2006). The role of heparan sulphate proteoglycans in angiogenesis. Biochemical Society transactions 34, 451-453.

Stuttfeld, E., and Ballmer-Hofer, K. (2009). Structure and function of VEGF receptors. IUBMB life 61, 915-922.

Takagi, S., Hirata, T., Agata, K., Mochii, M., Eguchi, G., and Fujisawa, H. (1991). The A5 antigen, a candidate for the neuronal recognition molecule, has homologies to complement components and coagulation factors. Neuron 7, 295- 307.

Takagi, S., Kasuya, Y., Shimizu, M., Matsuura, T., Tsuboi, M., Kawakami, A., and Fujisawa, H. (1995). Expression of a cell adhesion molecule, neuropilin, in the developing chick nervous system. Developmental biology 170, 207-222.

Takagi, S., Tsuji, T., Amagai, T., Takamatsu, T., and Fujisawa, H. (1987). Specific cell surface labels in the visual centers of Xenopus laevis tadpole identified using monoclonal antibodies. Developmental biology 122, 90-100.

Takahashi, T., Fournier, A., Nakamura, F., Wang, L.H., Murakami, Y., Kalb, R.G., Fujisawa, H., and Strittmatter, S.M. (1999). Plexin-neuropilin-1 complexes form functional semaphorin-3A receptors. Cell 99, 59-69.

Takahashi, T., and Strittmatter, S.M. (2001). Plexina1 autoinhibition by the plexin sema domain. Neuron 29, 429-439.

Tani, T.T., and Mercurio, A.M. (2001). PDZ interaction sites in integrin alpha subunits. T14853, TIP/GIPC binds to a type I recognition sequence in alpha 6A/alpha 5 and a novel sequence in alpha 6B. The Journal of biological chemistry 276, 36535-36542.

Teesalu, T., Sugahara, K.N., Kotamraju, V.R., and Ruoslahti, E. (2009). C-end rule peptides mediate neuropilin-1-dependent cell, vascular, and tissue penetration. Proceedings of the National Academy of Sciences of the United States of America 106, 16157-16162.

Tennessen, J.A., Bigham, A.W., O'Connor, T.D., Fu, W., Kenny, E.E., Gravel, S., McGee, S., Do, R., Liu, X., Jun, G., et al. (2012). Evolution and functional impact of rare coding variation from deep sequencing of human exomes. Science 337, 64-69.

186 Timoshenko, A.V., Rastogi, S., and Lala, P.K. (2007). Migration-promoting role of VEGF-C and VEGF-C binding receptors in human breast cancer cells. British journal of cancer 97, 1090-1098.

Tordjman, R., Lepelletier, Y., Lemarchandel, V., Cambot, M., Gaulard, P., Hermine, O., and Romeo, P.H. (2002). A neuronal receptor, neuropilin-1, is essential for the initiation of the primary immune response. Nature immunology 3, 477-482.

Vacca, A., Scavelli, C., Serini, G., Di Pietro, G., Cirulli, T., Merchionne, F., Ribatti, D., Bussolino, F., Guidolin, D., Piaggio, G., et al. (2006). Loss of inhibitory semaphorin 3A (SEMA3A) autocrine loops in bone marrow endothelial cells of patients with multiple myeloma. Blood 108, 1661-1667.

Valdembri, D., Caswell, P.T., Anderson, K.I., Schwarz, J.P., Konig, I., Astanina, E., Caccavari, F., Norman, J.C., Humphries, M.J., Bussolino, F., et al. (2009). Neuropilin-1/GIPC1 signaling regulates alpha5beta1 integrin traffic and function in endothelial cells. PLoS biology 7, e25.

Vander Kooi, C.W. (2013). Chapter Twenty One - Megaprimer Method for Mutagenesis of DNA. In Methods in enzymology, L. Jon, ed. (Academic Press), pp. 259-269.

Vander Kooi, C.W., Jusino, M.A., Perman, B., Neau, D.B., Bellamy, H.D., and Leahy, D.J. (2007). Structural basis for ligand and heparin binding to neuropilin B domains. Proceedings of the National Academy of Sciences of the United States of America 104, 6152-6157.

Varshavsky, A., Kessler, O., Abramovitch, S., Kigel, B., Zaffryar, S., Akiri, G., and Neufeld, G. (2008). Semaphorin-3B is an angiogenesis inhibitor that is inactivated by furin-like pro-protein convertases. Cancer research 68, 6922-6931.

Vieira, J.M., Schwarz, Q., and Ruhrberg, C. (2007). Selective requirements for NRP1 ligands during neurovascular patterning. Development 134, 1833-1843. von Wronski, M.A., Raju, N., Pillai, R., Bogdan, N.J., Marinelli, E.R., Nanjappan, P., Ramalingam, K., Arunachalam, T., Eaton, S., Linder, K.E., et al. (2006). Tuftsin binds neuropilin-1 through a sequence similar to that encoded by exon 8 of vascular endothelial growth factor. The Journal of biological chemistry 281, 5702- 5710. von Wronski, M.A., Tweedle, M.F., and Nunn, A.D. (2007). Binding of the C- terminal amino acids of VEGF121 directly with neuropilin-1 should be considered. FASEB journal : official publication of the Federation of American Societies for Experimental Biology 21, 1292; author reply 1293.

Waltenberger, J., Claesson-Welsh, L., Siegbahn, A., Shibuya, M., and Heldin, C.H. (1994). Different signal transduction properties of KDR and Flt1, two receptors for

187 vascular endothelial growth factor. The Journal of biological chemistry 269, 26988- 26995.

Walzer, T., Galibert, L., Comeau, M.R., and De Smedt, T. (2005). Plexin C1 engagement on mouse dendritic cells by viral semaphorin A39R induces actin cytoskeleton rearrangement and inhibits integrin-mediated adhesion and chemokine-induced migration. J Immunol 174, 51-59.

Wang, L., Mukhopadhyay, D., and Xu, X. (2006). C terminus of RGS-GAIP- interacting protein conveys neuropilin-1-mediated signaling during angiogenesis. FASEB journal : official publication of the Federation of American Societies for Experimental Biology 20, 1513-1515.

Wang, L., Zeng, H., Wang, P., Soker, S., and Mukhopadhyay, D. (2003). Neuropilin-1-mediated vascular permeability factor/vascular endothelial growth factor-dependent endothelial cell migration. The Journal of biological chemistry 278, 48848-48860.

Watanabe, T., Murakami, K., and Nakayama, K. (1993). Positional and additive effects of basic amino acids on processing of precursor proteins within the constitutive secretory pathway. FEBS letters 320, 215-218.

Weekes, C.D., Beeram, M., Tolcher, A.W., Papadopoulos, K.P., Gore, L., Hegde, P., Xin, Y., Yu, R., Shih, L.M., Xiang, H., et al. (2014). A phase I study of the human monoclonal anti-NRP1 antibody MNRP1685A in patients with advanced solid tumors. Investigational new drugs.

West, D.C., Rees, C.G., Duchesne, L., Patey, S.J., Terry, C.J., Turnbull, J.E., Delehedde, M., Heegaard, C.W., Allain, F., Vanpouille, C., et al. (2005). Interactions of multiple heparin binding growth factors with neuropilin-1 and potentiation of the activity of fibroblast growth factor-2. The Journal of biological chemistry 280, 13457-13464.

Xu, Y., Yuan, L., Mak, J., Pardanaud, L., Caunt, M., Kasman, I., Larrivee, B., Del Toro, R., Suchting, S., Medvinsky, A., et al. (2010). Neuropilin-2 mediates VEGF- C-induced lymphatic sprouting together with VEGFR3. The Journal of cell biology 188, 115-130.

Yamada, Y., Oike, Y., Ogawa, H., Ito, Y., Fujisawa, H., Suda, T., and Takakura, N. (2003). Neuropilin-1 on hematopoietic cells as a source of vascular development. Blood 101, 1801-1809.

Yang, Y., Xie, P., Opatowsky, Y., and Schlessinger, J. (2010). Direct contacts between extracellular membrane-proximal domains are required for VEGF receptor activation and cell signaling. Proceedings of the National Academy of Sciences of the United States of America 107, 1906-1911.

Yazdani, U., and Terman, J.R. (2006). The semaphorins. Genome biology 7, 211.

188 Yuan, L., Moyon, D., Pardanaud, L., Breant, C., Karkkainen, M.J., Alitalo, K., and Eichmann, A. (2002). Abnormal lymphatic vessel development in neuropilin 2 mutant mice. Development 129, 4797-4806.

Zachary, I.C., Frankel, P., Evans, I.M., and Pellet-Many, C. (2009). The role of neuropilins in cell signalling. Biochemical Society transactions 37, 1171-1178.

Zanuy, D., Kotla, R., Nussinov, R., Teesalu, T., Sugahara, K.N., Aleman, C., and Haspel, N. (2013). Sequence dependence of C-End Rule peptides in binding and activation of Neuropilin -1 Receptor. Journal of structural biology.

Zhang, F., Tang, Z., Hou, X., Lennartsson, J., Li, Y., Koch, A.W., Scotney, P., Lee, C., Arjunan, P., Dong, L., et al. (2009). VEGF-B is dispensable for blood vessel growth but critical for their survival, and VEGF-B targeting inhibits pathological angiogenesis. Proceedings of the National Academy of Sciences of the United States of America 106, 6152-6157.

Zhang, J.H., Chung, T.D., and Oldenburg, K.R. (1999). A Simple Statistical Parameter for Use in Evaluation and Validation of High Throughput Screening Assays. Journal of biomolecular screening 4, 67-73.

Zheng, M., Streck, R.D., Scott, R.E., Seidah, N.G., and Pintar, J.E. (1994). The developmental expression in rat of proteases furin, PC1, PC2, and carboxypeptidase E: implications for early maturation of proteolytic processing capacity. The Journal of neuroscience : the official journal of the Society for Neuroscience 14, 4656-4673.

Zhou, Y., Gunput, R.A., and Pasterkamp, R.J. (2008). Semaphorin signaling: progress made and promises ahead. Trends Biochem Sci 33, 161-170.

189 VITA

1. Place of birth:

Beaufort, South Carolina

2. Educational institutions attended and degrees already awarded:

2010: Western State College of Colorado (WSC), Gunnison, CO Bachelor of Arts, Department of Biology Advisor: Dr. Cassandra Osborne

3. Scholastic and professional honors:

Travel Grant: ACA Annual Meeting (UK) College of Medicine Research Incentive Award (UK) Alumni Award for Excellence (WSC) Noddin-McKenny Award (WSC)

4. Professional publications:

1. Li X., Parker M.W., Vander Kooi C.W. (2014). Control of cellular motility by neuropilin-mediated physical interactions. Biomolecular Concepts (In Press).

2. Parker M.W., Vander Kooi C.W. (2014) Plate-based assay for measuring direct Neuropilin-Semaphorin interactions. Methods in Molecular Biology: Semaphorins and Semaphorin Signaling (In Press).

3. Parker M.W., Vander Kooi C.W. (2014). Microplate-based screening for small molecule inhibitors of Neuropilin-2/VEGF-C interactions. Analytical Biochemistry (453): 4-6.

4. Guo H.F., Li X., Parker M.W., Waltenberger J., Becker P.M., Vander Kooi C.W. (2013). Mechanistic basis for the potent anti-angiogenic activity of semaphorin 3F. Biochemistry 52(43): 7551-7558.

5. Parker M.W., Linkugel A.D., Vander Kooi, C.W. (2013) Effect of C-terminal sequence on competitive semaphorin binding to neuropilin-1. J Mol Biol. (425)22: 4405-4414.

6. Parker M.W., Xu P., Guo H.F., Vander Kooi C.W. (2012) Mechanism of selective VEGF-A binding by neuropilin-1 reveals a basis for specific ligand inhibition. PLoS One 7(11): e49177 .

7. Parker M.W., Guo H.F., Li X., Linkugel A.D., Vander Kooi C.W. (2012) Function of members of the neuropilin family as essential pleiotropic cell surface receptors. Biochemistry 51(47): 9437-9446.

190

8. Parker M.W., Xu P., Li X., Vander Kooi C.W. (2012) Structural basis for selective vascular endothelial growth factor-A (VEGF-A) binding to neuropilin-1. J Biol Chem. 287(14):11082-11089.

9. Parker M.W., Hellman L.M., Xu P., Fried M.G., Vander Kooi C.W. (2010) Furin processing of semaphorin 3F determines its anti-angiogenic activity by regulating direct binding and competition for neuropilin. Biochemistry 49(19): 4068-4075.

______Matthew William Parker

191