The Role of Proprotein Convertases in Cancer

MCGILL UNIVERSITY

The role of Proprotein Convertases in Cancer

By Xiaowei Sun

Department of Medicine Division of Experimental medicine McGill University Montreal, Canada December 2012

A thesis submitted to McGill University in fulfillment of the requirements for the degree of Doctor of Philosophy

Acknowledgements

I wish to express my most sincere gratitude and appreciation to Dr Nabil Seidah for his supervision, guidance, encouragement and stimulating discussions throughout the development of this thesis.

I would also like to give my deepest thanks to Dr Annik Prat, who was closely implicated in this thesis. I was grateful to all her guidance, patience and support.

I would also express my appreciation to my thesis committee members: Dr Hugh Bennett, Dr Jean-François Côté, Dr Marc Prentki and Dr Jean-Bernard Denault. I am very much grateful to my academic advisor Claude Lazure who offered tremendous help through this thesis and to my future career.

I would like to thank all the current and former members of Seidah’s laboratory, Ann Chamberland, Rachid Essalmani, Delia Susan-Resiga, Josée Hamelin; Anna Roubtsova; Jadwiga Marcinkiewicz; Brigitte Mary, Marie-Claude Asselin, Suzanne Benjannet and Zuhier Awan. I especially want to express my sincerest gratitude to my colleagues and friends Grisel Luna, Maryssa Canuel, Estelle Rousselet, Johann Guillemot, WooJin Kim, Maxim Denis and Lorelei Durand for their great support in the lab as well as outside the lab.

Finally, I dedicate this work to my loving parents, without their unconditional love and encouragement this work would not have been possible. I would also like to thank my friends Jing, Victor, Cecilia and Ling for their friendship and support in the most difficult time.

ABSTRACT

Cancer is a leading cause of death worldwide and accounts for about one fifth of all death in the Western world. In 2008, nearly 12.7 million new cancer cases and 7.6 million cancer deaths occurred worldwide. The development of cancer is a multistage process, during which cells acquire a series of mutations that eventually lead to unrestrained cell growth, evasion of cell death, angiogenesis, invasion of the surrounding tissue and finally spreading to other parts of the body. The mammalian proprotein convertases (PCs) constitute a family of nine secretory serine proteases that are related to bacterial subtilisin and yeast kexin. They have been associated with cancer since the early 1990s. By processing cancer-associated factors, PCs are believed to play key roles in almost every step of cancer development. Seven of these PCs (PC1, PC2, furin, PC4, PC5/6, PACE4 and PC7) activate, or less frequently inactivate, a wide variety of substrates, including hormones, growth factors, receptors, adhesion molecules, angiogenic factors, metalloproteases. Among these substrates, some of them are key factors controlling cancer progression and metastasis. The last member of this family proprotein convertase subtilisin kexin 9 (PCSK9) only cleaves itself and participates in maintaining the levels of cholesterol, which was shown to have impacts on cancer incidence.

In this thesis, I focused on the role of two PCs, PC5/6 and PCSK9, in cancer development. I first showed that PC5/6 is systematically down-regulated in human and mice intestinal tumors. In ApcMin/+ mice which are a colonic cancer model and develop numerous adenocarcinomas along the intestinal tract, the specific knockout of PC5/6 in the intestine and colon leads to higher number of tumors, particularly in duodenum. This suggests that PC5/6 plays a protective role against tumorigenesis in the intestine. Although PC5/6 is protective in intestinal cancer, it has been shown to promote tumor progression in other cancer types e.g., brain and skin. Interestingly, PC5/6 is inhibited by

some natural inhibitors, the latent TGFβ binding proteins 2 and 3 (LTBP-2, -3). These two proteins reduce the enzymatic activity of PC5/6A and reduce the bio-availability of PC5/6A by sequestering the zymogen proPC5/6 in the extracellular matrix. Finally, I demonstrated that the lack of PCSK9 leads to a significantly lower level of liver metastasis of melanoma cells. This cancer protective effect is due to low plasma cholesterol levels as well as high apoptosis in liver stroma and metastasized tumors that are associated with PCSK9 deficiency.

In summary, the present cumulative data define some of the in vivo roles of PC5/6 and PCSK9 in cancer and should enhance our appreciation of the physiological impact of PC inhibition.

iii

Résumé

Le cancer est à l’origine d’environ 20% des décès dans le monde occidental. En 2008, près de 12,7 millions de nouveaux cas et 7,6 millions de décès sont survenus dans le monde. Le développement du cancer est un processus multi-étapes, au cours duquel les cellules acquièrent des mutations qui mènent à une croissance cellulaire incontrôlée, l'échappement à la mort cellulaire, l'angiogenèse, l'invasion des tissus environnants et, finalement, la propagation des tumeurs à d'autres parties du corps. Les proprotéines convertases (PC) constituent une famille de neuf sérine-protéases sécrétoires similaires à la subtilisine bactérienne et à la kexine de levure. Dès le début des années 1990, il a été montré que les PC contrôlent l’activation de facteurs clés dans presque toutes les étapes du développent du cancer. Les sept premiers membres de la famille des PC (PC1, PC2, furin, PC4, PC5/6, PACE4 et PC7) activent, et moins fréquemment inactivent, des hormones, des facteurs de croissance, des récepteurs, des molécules d’adhésion, des facteurs angiogéniques, ou encore des métalloproteases, qui peuvent être impliqués dans la progression du cancer. Le dernier membre de cette famille, la proprotéine convertase de type subtilisine/kexin 9 (PCSK9), participe au maintien de l’homéostasie du cholestérol. Ce dernier a été lié à la fois positivement et négativement au dévelopement de cancers. Au cours de mon travail de thèse, je me suis concentrée sur le rôle de deux PC, PC5/6 et PCSK9, dans le développement du cancer. J’ai d'abord montré que l’expression de PC5/6 est systématiquement diminuée dans les tumeurs intestinales chez l'homme et chez la souris. Chez les souris ApcMin/+, modèle du cancer du côlon, qui développent de nombreux adénocarcinomes le long de l’intestin, l’absence de PC5/6 spécifiquement dans l'intestin et le côlon conduit à un plus grand nombre de tumeurs, en particulier dans le duodénum. Ceci suggère que PC5/6 ait un rôle protecteur contre la tumorigenèse dans l'intestin. Cependant, PC5/6 a été montré favoriser la progression de tumeurs dans d’autres types de cancers, par exemple le cancer du cerveau et de la peau. J’ai ensuite mis

en évidence l’existence de deux inhibiteurs naturels de PC5/6 : les «latent TGFβ binding proteins -2 et -3» (LTBP-2 et -3). Ils réduient l'activité enzymatique de PC5/6A ainsi que sa disponibilité en séquestrant le zymogène dans la matrice extracellulaire. Enfin, j’ai également démontré qu’une déficience en PCSK9 conduit à plus de métastases hépatiques, induites par injection de cellules mélanocytaires. Cet effet protecteur est dû aux faibles taux de cholestérol associé à la perte de PCSK9 et à une mort cellulaire programmée accrue dans le foie injecté ainsi que dans les métastases chez les souris n’exprimant pas PCSK9. En résumé, ces travaux définissent le rôle des PC5/6 et PCSK9 dans le cancer in vivo et nous aident à évaluer l'impact physiologique potentiel de l’inhibition des PC.

Table of Contents

Acknowledgement ...... i Abstract ...... ii Résumé ...... iv Table of contents ...... vi List of figures ...... ix List of tables...... x List of abbreviations ...... xi Chapter I: Introduction ...... 1 [A] Proprotein convertases...... 1 A.1 Discovery ...... 1 A.2 Protein structure ...... 3 A.3 Tissue distribution and cellular localization ...... 4 A.4 Zymogen activation...... 6 A.5 PC5/6: an essential convertases ...... 7 A.5.1 A unique proprotein convertase with two isoforms ...... 7 A.5.2 Cellular trafficking and zymogen activation of PC5/6 ...... 8 A.5.3 Transforming growth factor β-like factors: PC-substrates ...... 9 A.6 PCSK9: third gene associated with hypercholesterolemia ...... 10 A.6.1 Cholesterol biosynthesis and regulation ...... 11 A.6.2 PCSK9: a key player in cholesterol homeostasis ...... 13 A.6.3 Structure and function of PCSK9...... 14 A.6.4 PCSK9 functions beyond LDLR regulation ...... 15 [B] Cancer and the Proprotein Convertases ...... 18 B.1 Causes and progression of human cancer ...... 18 B.2 Hallmarks of cancer ...... 20

B.3 Treatment of human cancer ...... 22 B.4 The proprotein convertases and cancer ...... 23 B.4.1 The proprotein convertases expression in cancer ...... 23 B.4.2 The proprotein convertases and tumor growth ...... 25 B.4.3 The proprotein convertases and angiogenesis ...... 26 B.4.4 The proprotein convertases and cell invasion ...... 27 B.4.5 Cross-talk between the substrates of the proprotein convertases ...... 29 B.4.6 The proprotein convertases modify the incidence of cancer by controlling cholesterol homeostasis ...... 31 B.5 Targeting the proprotein convertases in cancer therapy ...... 32 B.5.1 Inhibitors of the proprotein convertases ...... 32 B.5.2 Inhibition of the proprotein convertase as a potential cancer therapy...... 33 Chapter II: Manuscript 1: Molecular Cancer 2009 ...... 36 2.1 Preface...... 36 2.2 The proprotein convertase PC5/6 is protective against intestinal tumorigenesis: in vivo mouse model...... 38 2.3 Discussion and conclusion ...... 66 Chapter III: Manuscript 2: Journal of Biological Chemistry 2011 ...... 70 3.1 Preface...... 70 3.2 Latent transforming growth factor beta-binding proteins-2 and -3 inhibit the proprotein convertase 5/6A...... 72 3.3 Discussion and conclusions ...... 98 Chapter IV: Manuscript 3: Neoplasia 2012 ...... 103 4.1 Preface...... 103 4.2 Proprotein convertase subtilisin/kexin type 9 deficiency reduces melanoma metastasis in liver ...... 105 4.3 Discussion and conclusion ...... 140 4.3.1 PCSK9 inhibition and risk of cancer ...... 140 4.3.2 PCSK9 regulate cancer cell adhesion ...... 141 4.3.3 PCSK9 regulates cancer cell apoptosis in liver ...... 144 4.3.4 PCSK9-deficiency enhances liver metastasis independently to the LDLR ...... 148

vii

4.3.5 Diet-induced hypercholesterolemia, but not familial hypercholesterolemia, increases liver metastasis ...... 150 Chapter VI: Conclusions and future perspectives ...... 152 Chapter VII: Personal contributions to the subjects of the thesis...... 154 Reference List ...... 155

viii

List of figures

Figure 1 Classification of proteases based on the degree of identity of their catalytic domain primary sequences...... 2

Figure 2 Schematic representation of the primary structure of mammalian PCs...... 4

Figure 3 Cellular localization and trafficking of the proprotein convertases...... 6

Figure 4 Cholesterol biosynthesis and its regulation by the SREBPs pathways...... 12

Figure 5 PCSK9 induces LDLR degradation via extracellular and intracellular pathways ...... 14

Figure 6 PCSK9 affects gene expression pathways beyond cholesterol metabolism in HepG2 cells ...... 17

Figure 7 Cancer progression from benign tumor to metastasis...... 19

Figure 8: The proprotein convertases in cancer...... 24

Figure 9: The opposite role of furin and PC5/6A in brain tumor progression...... 73

Figure 10: Schematic diagram depicting how LTBP-2 and -3 reduce mature PC5/6A bioavailability...... 79

Figure 11: Sequence homology among the prosegments of human PC1/3, PC2, furin, PACE4 and PC5/6...... 80

Figure 12: LRP1 is a novel target of PCSK9...... 159

Figure 13: The loss of PCSK9 enhances apoptosis in liver stroma and tumors in an LDLR-independent manner...... 162

Figure 14: The loss of PCSK9 increases hepatic metastasis in an LDLR-independent manner...... 166

List of tables

Table 1 Classifications and examples of human carcinogens ...... 18

List of abbreviations

α1-PDX: α1-antitrypsin Portland GOF: gain-of-function

aa: amino acid HMG-CoA: 3-hydroxy-3-methylglutaryl CoA ADH: autosomal dominant hypercholesterolemia HMGCR: 3-hydroxy-3-methylglutaryl CoA reductase ADAM: a disintegrin and metalloproteinase HSPG: heparan sulfate proteogylcan

APC: adenomatous polyposis coli IGF: Insulin-like growth factor

APOB: apolipoprotein B KO: knockout

ApoER2:apolipoprotein E receptor-2 LDLR: low-density lipoprotein receptor

Bcl-2: B cell lymphoma-2 LOF: loss-of-function

BMP11: bone morphogenic protein 11, LRP1: low-density lipoprotein receptor- related protein 1 CAM: cell adhesion molecule Min: multiple intestinal neoplasia CHRD: cysteine-histidine rich domain MMP: matrix metalloprotease CMK: chloromethylketone MT1-MMP: membrane type 1-matrix CRD: cysteine-rich domain metalloprotease

CVDs cardiovascular diseases NARC-1: neural apoptosis-regulated convertase-1 dKO: double knockout NF-κB: nuclear factor-κB ECM: extracellular matrix NSCLC: non-small cells lung carcinoma ER: endoplasmic reticulum SCAP: sterol regulatory element binding EGF-A: epidermal growth factor-A protein cleavage activating protein

EMT: epithelial-mesenchymal transition SKI-1subtilisin kexin isozyme-1

Gdf11: growth and differentiation factor SREBP: sterol regulatory element binding 11 protein

xii

SRE sterol response element TIMP: tissue inhibitors of metalloprotease

PC: proprotein convertase siRNA: small interference RNA

PDGF: platelet-derived growth factor VEGF: vascular endothelial growth factor

PTM: post-translational modification VLDLR: very-low-density lipoprotein receptor TGFβ: transforming growth factorβ

TGN: trans-Golgi network

xiii

Chapter I: Introduction

The diversity of protein/peptide products generated by a given genome depends not only on the exact number of protein-coding genes but also on multiple other processes. Prominent among these are the post-translational modifications (PTMs) of the primary protein product that influence its structure and/or function. Most PTMs are introduced into target proteins/peptides by specific enzymes. Some modification are reversible, such as serine/threonine/tyrosine-phosphorylation and S-nitrosylation; but the most common ones are irreversible or covalent, including asparagine- and serine/theronine- glycosylation, tyrosine-sulfation, serine-octanoylation, disulfide bridge formation, N- terminal acetylation, C-terminal amidation, protein lipidation and cholesterol attachments. Another most common irreversible modification involves the cleavage of the peptide bonds of proteins/peptides at specific sites. Analysis of human and mouse genomes revealed ~600 proteases, which vary in substrate specificity, mechanism of cleavage, cellular location and stability. These proteases can be classified based on their active site involved in proteolysis. Accordingly, the proteases are classified in five major families: serine, cysteine, aspartic acid, threonine and metallo-proteases, based on the critical nucleophilic amino acid (aa) in the active site (1) (Figure 1). Asparatic acid and metallo-protease use H2O as nucleophile to break peptide bonds. While serine, cysteine and threonine protease use their own respective aa to hydrolyze peptide bonds. Serine proteases are the most abundant proteases and are divided into two classes: those closer to trypsin/chymotrpsin and those related to bacterial subtilisin and yeast kexin (2).

[A] Proprotein convertases

A.1 Discovery

A sub-class of serine-proteases of the subtilisn/kexin type, named proprotein convertases (PC) family, was discovered and characterized in the last 22 years. Furin is the first PC gene partially discovered in 1986, and its complete genetic sequence was

finally reported in 1988 (3) Its discovery came from its high degree of homology shared with yeast kexin (4), which is responsible for the cleavage of pro-α mating factor and pro-killer toxin (5). The characterization of furin led to the discovery of six other kexin- like PCs, PC1/3, PC2, PC4, PC5/6, PACE4 and PC7 from 1991 to 1996. The 8th member of PC family, subtilisin kexin isozyme-1 (SKI-1), also known as site-1 protease (S1P), is discovered in 1999. It is a pyrolysin-like homologue of subtilisin. The last member of PC family is PCSK9, which was discovered in 2003 in our laboratory. This PC family thus comprises 9 members.

The first seven convertases are basic aa-specific proteases that cleave substrates at the general consensus motif (K/R)-(X)-(K/R)↓, where X = 0, 1, 2 or 3 aa (6). SKI-1 cleaves substrates at the consensus motif R-X-(hydrophobic)-X↓ (7). The last member PCSK9 has only one substrate. It autocatalytically cleaves itself at the VFAQ152↓ sequence within its prosegment (8).

Figure 1 Classification of proteases based on the degree of identity of their catalytic domain primary sequences. (adapted from Seidah & Chretien 1999 (9))

A.2 Protein structure

PCs are multi-domain proteases. Some domains are well conserved, while others are specific to each PC (Figure 2). All PCs contain a similar N-terminal structure, consisting of a signal peptide that is required for entry of the PC into the endoplasmic reticulum (ER) preceding its intracellular trafficking within the secretory pathway. The signal peptide is followed by a prosegment, acting as an intramolecular chaperone assisting the folding the zymogen, as well as an inhibitor controlling enzymatic activity of the mature protease. Following the prosegment, PCs possess a subtilisin/kexin-related catalytic domain where the catalytic triad aspartic acid, histidine and serine, and the oxyanion hole asparagine are located (10). All 7 basic aa-specific PCs have a conserved β-barrel P-domain following the catalytic domain. This domain apparently stabilizes the enzyme and appears to be necessary for the folding of the convertases in the ER (11, 12). PCs mostly differ in their C-terminal domains, which regulate uniquely the subcellular localization and trafficking of each PC (6). Some PCs including furin, PACE4, PC5/6A and PC5/6B have a cysteine-rich domain (CRD), which contains multiple repeats of the consensus motif: Cys-X2-3-Cys-X3-4-Cys-X2-7-Cys-X5-10-Cys-X2-Cys-X9-13-Cys-X3-5-Cys-

X7-16. In furin and PC5/6B, the CRD is followed by a transmembrane domain and a cytosolic tail. PC7 and SKI-1 are also type-I transmembrane bound proteases, but do not contain a CRD. Although PC5/6A and PACE4 do not have a transmembrane domain, they can be retained at the cell surface via their CRD, which anchors them to the cell surface via binding to heparan sulfate proteoglycans (HSPGs) and tissue inhibitors of metalloproteases (TIMPs) (13, 14). Lastly, PCSK9 exhibits a cysteine-histidine rich domain (CHRD) (15), which seems to be required for the internalization and lysosomal trafficking of the complex of PCSK9 and its target, the low-density lipoprotein receptor (LDLR) (See Chapter A.6.3).

Figure 2 Schematic representation of the primary structures of mammalian PCs. (reprinted from Seidah & Prat 2012 (16)) The seven kexin-like basic-aa-specific PCs, pyrolysin-like SKI-1 and proteinase K-like PCSK9 are individually boxed to emphasize their distinct homology. Notice that PC5/6 has two alternatively spliced forms, namely PC5/6A and PC5/6B. The various domains, N-glycosylation positions and the catalytic triad residues Asp, His and Ser, and the oxyanion hole Asn (Asp for PC2) are shown. PCs differ by their unique C-terminal domains, furin, PACE4, PC5/6A and PC5/6B have a CRD, whereas PCSK9 has a CHRD. In addition, furin, PC5/6B, PC7 and SKI-1 have a transmembrane domain following by a cytosolic tail.

A.3 Tissue distribution and cellular localization

While some PCs, furin, PACE4, PC5/6, PC7 and SKI-1 are ubiquitously or widely expressed, the expression of other PCs is more restricted. PC1/3 and PC2 are mostly localized in immature and dense-core secretory granule of neural cells (e.g., in brain and hypophyse) and endocrine cells (e.g., in pancreas and pituitary gland) (17, 18) (Figure 3). Their localizations are consistent with their implications in activation of prohormones

within the acidic regulated secretory pathway e.g., β-endorphin, insulin (19). PC4 expression is specific to germ cells within pachytene spermatocytes and round spermatids (20). Furin, PACE4, PC5/6A, PC5/6B are PC7 reach the cell surface via the constitutive pathway. The ubiquitously expressed membrane-bound PCs, furin, PC5/6B and PC7, can be recycled from the cell surface back to the trans-Golgi network (TGN) through endosomes (Figure 3). The endocytosis sorting motif YXXØ (Ø is a hydrophobic residue) locating in their cytosolic tail directs the internalization of the enzymes (21). Although furin and PC5/6B can be shed into soluble forms, PC7 is never shed and remains entirely membrane-bound. By contrast, the widely expressed PACE4 and PC5/6A are constitutively secreted into the extracellular matrix (ECM), however they can also be retained at the cell surface via their CRD. Unlike other membrane-bound PCs, SKI-1 does not reach the cell surface. It is mostly localized in the cis/medial-Golgi may transit to the endosomes/lysosomes directly from the TGN (22). However, some SKI-1 can be shed into the medium via an autocatalytic cleavage event (23). Accordingly, furin, PC4, PC5/6A and PC5/6B, PACE4, PC7 and SKI-1 process precursors while trafficking through the constitutive secretory pathway (6). The last member of the family PCSK9 is mainly expressed in the liver, small intestine and kidney (24, 25). It is also expressed transiently in the central nervous system during development, e.g., at embryonic day (E)12-E13 in telencephalon and E17-E20 in cerebellar neurons. In adult brain, it is only found in the olfactory peduncle and cerebellum (24). PCSK9 is secreted from the TGN into the medium as an soluble enzymatically inactive complex of the protease and its prosegment (8, 24). PCSK9 can also be internalized into endosomes by binding to the LDLR at the cell surface (26). The internalized PCSK9•LDLR complex in the clathrin- coated vesicles is then sent to lysosomes for degradation (27, 28) (See chapter A.6.3).

Figure 3 Cellular localization and trafficking of the proprotein convertases. (reprinted from Seidah et al, 2008 (6)) After exiting the ER, PC1/3 and PC2 are sorted to immature secretory granules where they are activated. They will then traffic to dense core granules where they process many secretory precursor proteins awaiting signals for regulated secretion. The other basic PCs are constitutively secreted. Furin, PC5/6, and PACE4 reach the cell surface from the TGN, whereas PC7 can reach the cell surface from both the TGN and the ER directly. PC5/6A and PACE4 can be retained at the cell surface and/or ECM by binding to HSPGs and/or to TIMPs. The membrane-bound furin, PC5/6B and PC7 can be recycled through endosomes back to the TGN. SKI-1 is mostly concentrated in the cis/medial Golgi. PCSK9 is the only PC secreted as enzymatically inactive prosegment•PCSK9 heterodimer. It can be endocytosed by binding the LDLR at the cell surface. The PCSK9•LDLR complex is targeted to lysosomes for degradation. SG: secretory granules; endo: endosomes; TGN: trans-Golgi network; TIMP: tissue inhibitor of metalloprotease; HSPG: heparan sulphate proteoglycans.

A.4 Zymogen activation

PCs are initially synthesized as inactive zymogens, which undergo a first autocatalytic processing at the C-terminus of the prosegment in the ER, except for SKI-1 where the first autocatalytic cleavage occurs in the middle of the prosegment. The processing will generate a complex of the inhibitory prosegment and the rest of the

molecule (prosegment•PC) (6). PC2 is an exception. ProPC2 exits the ER as a complex with is chaperone pro7B2, which first undergoes a furin cleavage into mature 7B2 in the TGN. The complex proPC2•7B2 then undergoes an autocatalytic cleavage upon transit into immature secretory granules (29). For the other PCs, the complex prosegment•PC exits the ER and sorts to specific subcellular compartments where the catalytic activity is acquired. However, to be fully activated, most PCs require a second autocatalytic cleavage within the prosegment to liberate the active enzyme from the inhibitory prosegment (6). PC4 and PC7do not require a second cleavage and get rid of their prosegment along the secretory route, likely as a result of lower pHs. PCSK9 does not undergo a second cleavage, it is secreted as an inactive prosegment•PC heterodimer. Therefore, the exquisite auto-activation mechanism of PC represents the first line of regulation that ensures the convertase to be only active at the correct intracellular/extracellular site(s) for the in trans processing of specific substrates. Thus, PC1/3 and PC2 are active only in secretory granules (30). Furin (31) and PC7 (32) are active in the TGN and endosomes, as well as at the cell surface. The secreted PC5/6A and PACE4 are activated by a second auto-cleavage event of the prosegment at the cell surface, thus are active at the cell surface, in the endosomes and ECM (14).

Since most of my thesis work focused on PC5/6 and PCSK9, I will elaborate on these two specific convertases in more details in the next two sections.

A.5 PC5/6: an essential convertases

A.5.1 A unique proprotein convertase with two isoforms

In 1993, PC5/6 was simultaneously discovered in our laboratory named PC5 (33) and in Nakayama’s group who called it PC6 (34). PC5/6 is the only PC that has two isoforms, soluble PC5/6A (915aa, 21 exons) (33) and membrane-bound PC5/6B (1877aa, 38 exons) (34), generated by differential splicing of its exons. Both transcripts share the first 20 exons encoding the signal peptide, prosegment, catalytic domain, P-domain, and the CRD (Figure 2). The 21st exon of PC5/6A that encodes the last 38 residues is replaced in PC5/6B with by 18 additional exons encoding an extended CRD, transmembrane domain and a cytoplasmic tail (34). Thus, PC5/6A possesses five tandem

repeats of the cysteine rich motif Cys-X2-3-Cys-X3-4-Cys-X2-7-Cys-X5-10-Cys-X2-Cys-X9-

13-Cys-X3-5-Cys-X7-16, while PC5/6B contains 22 such repeats.

PC5/6 is widely expressed. In situ hybridization and quantitative PCR analysis of the PC5/6 mRNA tissue distribution revealed that the small intestine and adrenal are the richest sources of PC5/6 (35). PC5/6 expression can be also detected very early during mouse embryonic development, appearing first in extra-embryonic tissues at the implantation stage (E4.5–E6.5) (35). It is also specifically expressed in cells of the maternal embryonic junction by E6.5. In addition, PC5/6A is the major isoform in most tissues except in the small intestine and kidney, where PC5/6B is the dominant form (35).

A.5.2 Cellular trafficking and zymogen activation of PC5/6

Soluble PC5/6A is sorted to the cell surface via the constitutive secretory pathway as well as partially to secretory granules of regulated secretory pathway (13, 36). Mature PC5/6A can be found both in the medium and at the surface via the binding of its CRD to HSPGs (13, 14) (Figure 3). The cell surface binding favors a close encounter with cell surface substrates such as HSPG-bound endothelial lipoprotein lipases (14, 37) and possibly transforming growth factor β (TGFβ)-like factors (see chapter A.5.3). In contrast, PC5/6B contains a transmembrane domain and a cytosolic tail. Its trafficking is controlled by sorting signals in the cytosolic tail and their interaction with specific sorting adaptors (36, 38). Endocytosis of PC5/6B is directed by the YEKL (YXXL) motif within its cytosolic tail. An acidic cluster in the proximity of the transmembrane domain directs the TGN localization of PC5/6B. A second distal-membrane acidic cluster is sufficient to recapitulates the cytosolic distribution, which is similar to that of the full PC5/6B cytosolic tail (36, 38).

PC5/6 is synthesized as an inactive zymogen. It undergoes a first autocatalytic processing in the ER at RTKR116↓ (mouse sequence), resulting in a tight binding complex of the inhibitory prosegment with the protease allowing the protein to exit the ER (36, 39). It is then fully activated after a second autocatalytic cleavage within the prosegment at RTIKR84↓. In the case of PC5/6A, the second cleavage mostly occurs at the cell surface, where PC5/6A is tethered via its CRD (13, 14).

A.5.3 Transforming growth factor β-like factors: PC-substrates

In 2008, our laboratory identified a major in vivo substrate of PC5/6 by a phenotypical analysis of PC5/6 knockout (KO) mice. PC5/6 KO newborn mice die at birth, lack of tail and exhibit some major defects in antero-posterior axis with extra- thoracic and -lumbar vertebrates, (18 and 8 instead of 13 and 6, respectively) and kidney agenesis (40). Some of these phenotypes were reported in growth and differentiation factor 11 (Gdf11, also known as bone morphogenic protein 11, BMP11) KO mice, suggesting that Gdf11 is a potential in vivo substrate of PC5/6. In vitro analysis further confirmed that Gdf11 is preferentially processed by both PC5/6A and B. And the cleavage occur at RSRR296↓NL (40).

Gdf11 belongs to the TGFβ superfamily, which comprises activins, nodals, bone morphogenic proteins, growth differentiating factors and canonical TGFβs. Upon synthesis, TGFβ factors form homodimers in the ER which are then cleaved at site 1 intracellularly or extracellularly into an N-terminal inhibitory prosegment and mature C- terminal domain that remain non-covalently associated. These complexes are named latent TGFβ complex. To be efficiently secreted, these latent complexes have to bind to latent TGFβ binding proteins (LTBPs) (41). Upon secretion, the latent TGFβ complex accumulates in the ECM (42) still through the binding to LTBPs (43) which allows the complex to interacte with HSPGs (44). The inactive latent TGFβ complex awaits a local activation signal that results in a second cleavage at site 2 within the prosegment to release the mature, active TGFβ.

Various TGFβ superfamily members, including TGFβ1 (45), lefty (14) and BMP4 (46) were reported to be cleaved by basic aa-specific PCs. However, these substrates can be cleaved by more than one PC. And more often, furin is the most efficient PC to cleave TGFβ-like protiens. In contrast, proGdf11 is the first substrate that was reported to be best processed by PC5/6. This PC5/6-proGdf11 enzyme-substrate specificity may reside in the primary sequence of the cleavage site. PC5/6 prefers an asparagine residue at the P1’ position, the first position after the cleavage site RSRR296↓NL (40). The specificity can also be due to the more advantageous cellular localization of PC5/6 compared to that

of furin, PC5/6A is anchored at the surface by binding HSPGs, which puts it in a perfect geographic position to process proGdf11.

A.6 PCSK9: third gene associated with hypercholesterolemia

PCSK9, the last member of the PC family, originally called NARC-1 for “neural apoptosis-regulated convertase 1,” was first discovered and characterized in our laboratory in 2003 (24). PCSK9 is the third gene, following LDLR (47) and the gene of apolipoprotein B (APOB) (48), associated with familial hypercholesterolemia (49). Familial hypercholesterolemia is characterized by high plasma cholesterol levels, specifically very high levels of low-density lipoprotein cholesterol (LDLc). It is inherited in an autosomal dominant pattern, with a prevalence of 1:500 for heterozygotes and 1:1,000,000 for homozygotes (50). Loss-of-function (LOF) mutations in LDLR and APOB, both involved in the internalization of LDL particles into the cells, are associated with autosomal dominant hypercholesterolemia (ADH), whereas it is gain-of-function (GOF) mutations in PCSK9 that associate with ADH. The most strong GOF mutations D347Y and R469W cause more than 3-fold increase in LDLc, 350 mg/dL and 357 mg/dL, respectively (51, 52). Non-sense mutations resulting in a LOF of PCSK9 are more common and found to associate with hypocholesterolemia in ~2% of black familial hypercholesterolemia subjects (53). Two individuals were found to totally lack PCSK9 (∆R97/Y142X and C679X/C679X) (54, 55). They exhibit 85% reduction of LDLc (15 mg/dL) (54).

High cholesterol levels is one of the most potent risk factors that contributes to the development of cardiovascular diseases (CVDs) and stroke (56). The world health organization estimated 17.3 million deaths from CVDs in 2008, a trend that will continue to increase in the next two decades. The prevention of CVDs can be achieved by lowering total cholesterol levels. Therefore, inhibition of PCSK9 emerged as a promising therapeutic approach to combat hypercholesterolemia and prevent CVDs (57).

A.6.1 Cholesterol biosynthesis and its regulation

Cholesterol plays important roles in animal life. It is essential to build and maintain cellular membranes. It is also an important precursor molecule for the synthesis of bile acids, vitamin D and steroid hormones. The structure of cholesterol (Figure 4A) was discovered in early 1920’ by German chemists Heinrich Wieland and Adolf Windaus who won the Nobel Prize in chemistry in 1928 for this work. Subsequent work by Konrad Bloch and Feodor Lynen established cholesterol biosynthesis pathway and they were also rewarded the Nobel Prize in physiology and medicine in 1964. Cholesterol is synthesized from two-carbon block acetyl-CoA, which are condensed by 3-hydroxy-3-methylglutaryl CoA (HMG-CoA) synthase to form HMG-CoA (Figure 4A). This molecule is then reduced to mevalonate by the enzyme HMG-CoA reductase (HMGCR), the rate-limiting step in cholesterol synthesis. HMGCR is inhibited by statins, the most prescribed cholesterol lowering drug. It contains an HMG-like moiety preventing the binding of HMG-CoA to HMGCR (58, 59). Mevalonate is then converted and decarboxylated to isopentenyl pyrophosphate (60). Six units of this molecule are condensed to form squalene, which is then cyclized to form lanosterol. Finally, lanosterol is converted to cholesterol by a series of oxidation, reduction and demethylation reactions. ~70% cholesterol is synthesized in various tissues such as liver, intestine and adrenal glands and other ~30% cholesterol comes from diet (61). Cells obtain exogenous cholesterol via the LDLR-mediated uptake of plasma lipoproteins, such as LDL. LDL particles bind the LDLR via their single copy of APOB. The complex is internalized into clathrin-coated vesicles and then delivered to lysosomes, where hydrolysis of the core cholesteryl esters occurs (62). The free cholesterol is then released, to be used immediately by cells or to be stored after being re-esterified by acyl-CoA cholesterol acyltransferase (63). High cholesterol levels are associated with various pathogenic conditions such as CVD and atherosclerosis. Thus body cholesterol homeostasis need to be strictly maintained. In 1985, Michael Brown and Joseph Goldstein were awarded the Nobel Prize in physiology and medicine for their work on the mechanism feedback regulation of cholesterol metabolism. A large amount of genes participating in the biosynthesis and metabolism of cholesterol is directly regulated at the transcription level by the cholesterol contents.

Figure 4 Cholesterol biosynthesis and its regulation by the SREBPs pathways. (A) Cholesterol biosynthetic pathway. (adapted from Goldstein et al, (60)) (B) Model for the sterol regulation of SREBP proteolysis. (reprinted from Horton et al,. 2002 (64)) SREBP: sterol regulatory element binding protein; SCAP: SREBP cleavage activating protein; SRE: sterol response element; S1P: site-1 protease (S1P/SKI-1); S2P: site-2 protease; bHLH: basic helix-loop-helix-leucine.

A family of transcription factors, sterol regulatory element binding proteins (SREBPs) comprising three isoforms SREBP-1a, SREBP-1c and SREBP-2, stands in the center of cholesterol and fatty acid homeostasis regulation. SREBPs belong to the large family of basic helix-loop-helix-leucine zipper (bHLH-Zip) transcription factors (65, 66). Newly synthesized SREBPs are inactive and embedded in the ER membrane where they form a complex with SREBP cleavage activating protein (SCAP), which contains a sterol sensory domain (67) (Figure 4B). With high cellular sterol concentration, SCAP confines SREBPs to the ER (68, 69). Under low cellular cholesterol concentrations, SCAP escorts

SREBPs to the cis/medial Golgi, where SREBPs undergo two consecutive proteolytic cleavages by SKI-1/S1P and site-2 protease (S2P) to release the N-terminal bHLH-Zip domain, designated nuclear SREBP (nSREBP) (70-73). nSREBP then translocates to the nucleus and binds to the sterol response element (SRE) in the promoter/enhancer regions of target genes, e.g., for nSREBP-2 it binds to and activates SRE elements necessary for the synthesis of 30 different mRNAs, including those coding for HMGCR and LDLR(64).

A.6.2 PCSK9: key player in cholesterol homeostasis

In adult animals, PCSK9 is mainly expressed in hepatocytes and enterocytes which are the major site of cholesterol metabolism (8). PCSK9 transcription is regulated by cholesterol concentration (74, 75). In mice, PCSK9 transcription is downregulated when cholesterol is in excess caused by a high cholesterol diet feeding (74). Indeed, the promoter of human, mouse and rat PCSK9 genes all contain SREs, and PCSK9 was further identified as one of the direct targets of SREBP-2 (74, 76). On the other hand, PCSK9 directly regulates cholesterol level. The mice lacking PCSK9 exhibit ~40% lower total cholesterol and ~80% lower LDLc compared to WT mice (25, 77). PCSK9 regulates cholesterol metabolism by controlling the protein levels of the LDLR, the major controller of blood LDLc levels by clearing LDLc from blood. Total loss of PCSK9 in mouse results in an increased protein level of the hepatic LDLR, thereby increasing the clearance of circulating LDLc and leading to low plasma LDLc levels (25, 77). PCSK9 decreases cell surface LDLR by enhancing its degradation in endosomes/lysosomes. Extracellular and intracellular PCSK9 both participate in this process, which are named the extracellular and intracellular pathways (Figure 5). The extracellular pathway requires binding of PCSK9 to cell surface LDLR, which results in the internalization of the complex through clathrin coated vesicles (78), whereby the complex is targeted by an as yet unknown mechanism to the endosomal/lysosomal degradative pathway. It has been shown that even blocking cell surface LDLR endocytosis, PCSK9 still enhances LDLR degradation (79). This is possibly due to a direct targeting of LDLR from the ER or the TGN by intracellular PCSK9 to lysosomes via the intracellular pathway (Figure 5).

Figure 5 PCSK9 induces LDLR degradation through the extracellular and intracellular pathway. (adapted from Poirier et al,. 2009 (79)) In the absence of PCSK9, LDL particles bind to the LDLR via APOB and the complex is endocytosed. The LDL particles are released in the endosomes and the LDLR is recycled back to the cell surface. In the presence of PCSK9, PCSK9 binds the LDLR either on the cell surface (extracellular pathway) or directly in TGN when both proteins transit through secretory pathway (intracellular pathway). Upon binding, the PCSK9•LDLR complex is then sent to lysosomes for degradation.

A.6.3 Structure and function of PCSK9

Human PCSK9 (692 aa) comprises a signal peptide (aa 1–30), a prosegment (aa 31– 152), a catalytic domain (aa 153-407), a hinge domain (aa 408-452) and C-terminal CHRD domain (aa 453-692) (Figure 3). Following synthesis, PCSK9 is autocatalytically cleaved at the VFAQ152↓SIP after the prosegment in the ER (8, 24). Different from most of the other PCs that undergo a second autocatalytic cleavage in order to remove the inhibitory prosegment and become enzymatically active, PCSK9 is the only PC secreted in complex with its inhibitory prosegment, thereby remaining enzymatically inactive (8).

Indeed, PCSK9 enhanced LDLR degradation was reported to be independent of its protease activity (80). PCSK9 and soluble LDLR co-crystal structure studies revealed that it is indeed the catalytic domain of PCSK9 (aa 153–156 and 367–381) that is in direct contact with the epidermal growth factor-A (EGF-A) domain of the LDLR (81). The most severe GOF mutation associated with hypercholesterolemia, D374Y, is within this interaction domain. PCSK9D347Y displays ~10-25-fold higher affinity of PCSK9 toward the LDLR (15). The prosegment of PCSK9 negatively regulates the interaction between this convertases and LDLR, since the removal of its N-terminal acidic stretch (aa 31–53) led to 7-fold stronger binding of PCSK9 to LDLR (81). As for the C-terminal CHRD, whose function is poorly understood, several mutations (GOF: R469W, E482G, R496W, F515L and H553R; LOF: Q554E and R434W) were identified in this region, but their underlying mechanisms remain to be determined. CHRD is composed by three modules (M1, 2 and 3). We recently showed that the M2 domain (aa 530-603) is essential for the extracellular (but not intracellular) activity of PCSK9 on the LDLR and that the increased positive charge of the double mutation HQ553,554RR enhances the activity of PCSK9 on LDLR and hence results in a GOF (Luna et al., 2012 in press).

A.6.4 PCSK9 functions beyond LDLR regulation

To better delineate the function(s) of PCSK9, some in vitro and in vivo studies were conducted in order to dissect role(s) of PCSK9 independent from its function in regulating the LDLR. Two microarray studies revealed that a large number of gene expressions were altered in HepG2 cells overexpressing WT PCSK9 or its GOF mutant D374Y (82, 83). Among these genes, the ones implicated in “cholesterol biosynthesis” and “sterol metabolism” are upregulated. This is probably a secondary effect of lower extracellular cholesterol uptake due to PCSK9 mediated LDLR degradation. Interestingly, PCSK9 regulates a number of pathways independent of its effects on cholesterol (83). These pathways include “protein ubiquitination”, “xenobiotic metabolism”, “cell cycle” and “inflammation and stress response” (83) (Figure 6). These two genome-profiling studies shed some light on the function of PCSK9 beyond cholesterol metabolism in hepatocytes.

A lot of efforts have been invested to identify targets of PCSK9 other than the LDLR. In 2008, our laboratory showed that in vitro PCSK9 reduces the protein levels of two receptors belonging to LDLR-like protein family, apolipoprotein E receptor-2 (ApoER2) and very-low-density lipoprotein receptor (VLDLR) (84). In addition, PCSK9 was also found to decrease the level of a major receptor of hepatitis C virus (HCV), the tetraspanin protein CD81, in the hepatic cell line HuH7 and in mouse liver, leading to reduced HCV infection (85). Recently, our laboratory was able to identify VLDLR as an in vivo target of PCSK9. VLDLR binds chylomicrons and VLDLs, which are rich in triglycerides. VLDLR delivers fatty acids deriving from triglycerides to peripheral tissues, especially into adipose tissue. PCSK9 KO mice exhibits ~4 (male) to 40-fold (female) more cell surface VLDLR in adipose tissue (86). Hence female PCSK9 KO mice exhibit significantly higher circulating triglyceride level (86).

In addition, in vivo studies also reveal some PCSK9 physiological roles that are independent to its function in regulating cholesterol homeostasis. Particularly after PCSK9 total and tissue-specific KO mice being generated (25, 77), some studies clearly demonstrated that PCSK9 also regulates cellular apoptosis, proliferation and differentiation. Apoptosis: PCSK9 was initially called neural apoptosis-regulated convertase-1 (NARC-1), since its expression was upregulated during serum-deprivation induced apoptosis in primary cerebellar neurons (24, 87, 88). PCSK9 was subsequently shown to act as a pro-apoptotic factor in cultured cerebellar granule neurons (87, 89). However, in some other tissue or cells, PCSK9 rather has an anti-apoptotic effect. In HepG2 cells, the overexpression of PCSK9 results in a repression of cellular apoptotic pathway genes (82). In pancreatic islets, the loss of PCSK9 results in signs of apoptosis and necrosis, as well as malformation, which probably explain the hypoinsulinemic, hyperglycemic and glucose intolerant phenotype in PCSK9 KO mice by the age of 4 months (90). Proliferation: PCSK9 also has been shown to regulate hepatocyte proliferation. PCSK9 KO mice exhibit impaired liver regeneration after partial hepatectomy due to a significant delay of cellular proliferation (77). This delay is probably due to severely low cholesterol levels in PCSK9 KO mice, since a high cholesterol diet, which restores plasma cholesterol levels, is able to correct such proliferative delay. Differentiation: PCSK9 was also believed to be implicated in

neurogenesis since it is transiently expressed in mouse central nervous system during brain development and its overexpression in primary neuronal cultures enhances the differentiation of cortical neurons (24). PCSK9 was also reported to downregulate LDLR levels during brain development (91). However a role of PCSK9 in brain development and neuronal differentiation is not yet determined.

Taken together, the above in vitro and in vivo studies support the idea that PCSK9 is implicated in some physiological and/or pathological processes independently of LDLR. As inhibitors or silencers of PCSK9 which are developed to reduce plasma cholesterol levels, will be available on the market within a few years, a thorough understanding of PCSK9 function is urgently needed to assess consequences and long term safety of PCSK9 inhibition (16).

Figure 6 PCSK9 affects gene expression pathways beyond cholesterol metabolism in HepG2 cells. (reprinted from Lan et al, 2010 (83)) The biological process category for the 427 genes regulated by PCSK9 but not by cholesterol starvation. Only the categories with ≥10 genes and P < 0.05 are shown.

[B] Cancer and the Proprotein Convertases

Cancer comprises more than 200 different diseases. It is a leading cause of death worldwide and accounts for about one fifth of all deaths in the developed countries of the Western world. In 2008, nearly 12.7 million new cancer cases and 7.6 million cancer deaths occurred worldwide. Deaths from cancer worldwide are projected to continue to rise, with an estimated 13.1 million deaths in 2030 (GLOBOCAN database. http://globocan.iarc.fr). PCs that have been associated with cancer shortly after their discovery play some central roles in the development of cancer.

B.1 Causes and progression of human cancer

Cancer can be caused by exogenous chemicals, physical or biological carcinogens and endogenous processes (Table 1), which induce genomic instabilities including single mutation in specific genes, or alteration, amplification and loss of large regions of the genome. GOF mutations in oncogenes and LOF mutations in tumor suppressor genes have been identified in virtually every form of cancer and are believed to be at the origin of the development of cancer.

Table 1 Classifications and examples of human carcinogens (reprinted from W.A.Schulz, 2007 (92))

Figure 7 Cancer progression from benign tumor to metastasized tumor (reprinted from Valastyan and Weinberg Cell 2011 (93)) Cancer cells carrying genetic changes enabling local proliferative and invasive properties can invade surrounding stromal tissues and degrade the basement membrane and then exit their primary sites of growth and intravaste. The survival cancer cells in the circulation will arrest at a distant organ site, extravasate, adapt to survive and thrive in the microenvironments of foreign tissues.

The transformation of a normal cell into a cancerous one and finally into malignant tumors is a multistage process (93). A cancer cell arises from a normal cell whose genome has been mutated. At the early stage of cancer, these cancer cells undergo local expansion and are not invasive. As the result of further accumulation of genetic alterations, a clone of cancer cells will acquire the invasive ability and become malignant. These malignant cancer cells will invade the surrounding ECM and stromal cell layers, then intravasate into the lumina of blood vessels. The cancer cells that overcome damage incurred by hemodynamic shear forces and escape the immune system in the circulation will arrest at various organ sites and extravasate into the parenchyma of distant tissues. These cells have to survive in these foreign microenvironments in order to form micrometastases. The cancer cells can enter in dormancy and subsequently reinitiate their

proliferative programs at metastatic sites, thereby generating macroscopic detectable neoplastic growths. Figure 7 summarizes the each steps in cancer progression.

B.2 Hallmarks of cancer

In spite of their diversity, different types of cancer share some common properties, which were proposed as “hallmarks of cancer” (94). These hallmarks corresponds to six biological capabilities acquired during the multistep development of cancer, including: (1) self-sufficiency in growth signals, (2) insensitivity to antigrowth signals, (3) evasion of apoptosis, (4) replicative immortalization, (5) sustained angiogenesis and (6) tissue invasion and metastasis (94). These hallmarks enable a normal cell to become tumorigenic and ultimately malignant, thereby providing a foundation for mechanism- based targeted therapies to treat human cancer.

Self-sufficiency in growth signals: cancer cells acquire the capability to induce and sustain proliferative signaling by producing specific growth factors and/or by stimulating the tumor-associated stroma to secrete growth factors (95). Additionally, most cancer cells express high levels of cell surface growth factor receptors rendering them hyper- responsive to growth factors. Cancer cells can also acquire growth autonomy through the constitutive activation of intracellular components of proliferative signaling pathways. For instance, the inactivation of intracellular APC (adenomatous polyposis coli) leads to the constitutive activation of proliferative Wnt signaling pathway, thereby giving rise to familial adenomatous polyposis (96, 97). In carriers of this mutated gene, the risk of colorectal cancer by the age of 40 years is almost 100% (96, 98).

Insensitivity to anti-growth signals and evasion of apoptosis: In addition to maintaining positively cell proliferation, cancer cells can evade growth suppressive signals and cell death generally through inactivation of tumor suppressors. Tumor suppressor proteins act as the central nodes in cellular regulatory pathway (99). They control the decision of cells to proliferate, to enter in the senescence, a viable but non- proliferative state, or to apoptose. For example, when tumor suppressor p53 receives inputs from cellular stresses e.g., genomic damage and oxidative stress, it initiates cell- cycle arrest and a cell-death checkpoint until the condition is normalized. Inactivating

somatic mutations in p53 are frequently found in human cancer cells. The heterozygous inactivation of the p53 gene renders mice highly prone to develop cancer early in life (100). .

Replicative immortalization: To form macroscopic tumors, cancer cells need to acquire limitless replicative potentials. A normal cell lineage undergoes only a limited number of cell division before they enter into senescence or apoptosis. In contrast, cancer cells are capable of infinite number of division, called immortality, which is normally specific to stem cells. Telomeres, which protect the ends of chromosomes, are key to enable cells with unlimited replicative capacity (101). In cancer cells, immortalization relies on the over-expression of telomerase that adds telomere repeats to the ends of telomeric DNA, thereby reducing telomere erosion after each cell division (102).

Sustained angiogenesis: the nutrients and oxygen supplied by vasculature are crucial for normal cell function and survival. Like the normal tissue, the growth of a tumor beyond a certain size requires neovasculature, which is induced and sustained by angiogenesis, a process of sprouting new vessels from existing ones. After organogenesis, vasculatures remain largely quiescent, except during wound healing and the female reproductive cycling process, in which angiogenesis is turned on transiently. In contrast, tumors can induce and sustain angiogenic processes by disequilibrating the balance of angiogenic inducers and inhibitors (103). For instance, the expression of angiogenesis initiators, such as vascular endothelial growth factors (VEGFs), is up-regulated by oncogenic signaling (104), whereas the expression of angiogenic inhibitor thrombospondin-1 is down-regulated (105).

Tissue invasion and metastasis: at later stages of most types of cancer, tumors grow beyond their primary site, invade the neighboring tissues and eventually spread to distant organs where they form new tumors. Invasion and metastasis, which distinguish benign from malignant tumors, account for 90% of human cancer death (106). To acquire invasive/metastatic properties, cancer cells hijack the normal embryonic development process “epithelial-mesenchymal transition” (EMT), which is characterized by a loss of cell-to-cell attachment and an increase in cell mobility as a result of repression of

adhesion protein such as cadherins and intergrins (107). The initiation and maintenance of oncogenic EMT occurs via the activation of a set of oncogenes, e.g., Ras, β-catenin and TGFβs (107, 108). Additionally, the invasive/metastatic ability of cancer cells is enhanced i) by upregulating the expression or by increasing the activity of ECM- degrading proteases, e.g. matrix metalloprotease-9 (MMP-9) and ii) by downregulating the expression of protease inhibitors, e.g., tissue inhibitor of TIMPs, in tumor and/or stromal cells (109). Metastasis is considered as the last and the most lethal step in human cancer progression, however the growth of disseminated cancer cells to from a macroscopic tumor in distal organs is the most limiting step. The disseminated cells need to develop their own solutions to adapt the new microenvironment. These adaptations require the acquisition of one or several of the five capacities described above.

B.3 Treatment of human cancer

Surgery, irradiation and medications can be used to treat cancers. The choice of therapy depends strongly on the stage of the cancer. Surgery and irradiation are usually used to treat localized tumors. In contrast, drug chemotherapy treats leukemia, lymphomas, soft tissue tumors and metastatic or locally advanced tumors. More often, a combination of these therapies is employed. The removal of localized tumors by surgery is followed by chemotherapy or irradiation to attack residual local tumors and metastases. This is called “adjuvant” therapy. Conversely, chemotherapy can be applied before the surgery to shrink the tumor mass and facilitate the complete removal of the tumor. This kind of treatment is called “neo-adjuvant” therapy (92). The efficacy of chemotherapy or radiotherapy largely depends on cancer type. Some cancers are highly sensitive to these therapies, e.g., some lymphomas and testicular carcinomas (110), whereas some are not, e.g., renal cell carcinomas (111).

In spite of the availability of abundant cancer therapies that alleviate cancer symptoms and prolong survival, many cancers remain incurable today. Thus novel cancer therapies are urgently needed. The advances in molecular biology of cancer have a significant impact on cancer therapies. For instance, immunotherapy and gene therapy has been already introduced into clinic and prove their efficacy in some cancers.

B.4 The proprotein convertases and cancers

B.4.1 The proprotein convertase expression in cancer

PCs have been associated to cancer since the early 1990s. By controlling activation or inactivation of cancer-associated proteins, PCs are believed to play key roles in almost every step of cancer development. Although the mechanistic link between PCs activities and acquisition of malignant properties are sometimes poorly understood, the altered expression of PCs has been reported in various human cancers and/or cancer cell lines.

PC1/3 and PC2 expressions were associated with the cancers of neuroendocrine origin. Both PCs are expressed in pheochromocytomas but not in normal adrenal tissues (112), pituitary adenomas (113) and carcinoids (114). PC1/3 and PC2 are also highly expressed in small cells lung carcinomas (115). Furin, which is the most extensively studied PC in cancer, was found to be overexpressed in cancer of breast (116), colon, head and neck (117) and non-small cells lung carcinomas (NSCLCs) (115). In ovarian cancer, increased expression of furin is correlated with decreased survival, suggesting the possible use of furin as a marker of tumor progression (118). PACE4 is highly expressed in NSCLCs (115) and breast cancer (116). Overexpressing PACE4 was able to convert non-invasive keratinocytes and squamous cell carcinoma cell line into a more invasive variant (119, 120). Recently, PACE4 is shown to be highly expressed in all different clinical stages of human prostate tumor tissues. And PACE4 silencing reduces prostate cancer cell line proliferation, clonogenic activity, and ability to grow as xenografts in nude mice, suggesting that PACE4 is a therapeutic target of prostate cancer (121)

Few studies addressed the role of PC5/6 and PC7 in cancer. PC5/6 was shown to be expressed in human colon cancer (122). PC7 is detected in human breast cancer, but not PC5/6 (116). Moreover, the level of PC5/6 expression correlates with the aggressiveness of NSCLC cell line and head and neck derived squamous cell carcinoma cell line, while PC7 expression is consistently high in all these cell lines (123).

These expression profiling studies provided solid evidences for the association between PCs and cancer. Although the precise mechanisms remain sometimes unclear,

many studies were able to reveal that PCs regulate tumor progression and metastasis through activating/inactivating specific factors involved in human cancer (Figure 8).

Figure 8: The proprotein convertases in cancer. PCs activate or inactivate a variety of precusors of growth factors, receptors, angiogenic factors, adhesion molecules and matrix metalloproteases that were shown to be implicated in crucial steps of tumor progression, from tumorigenesis to metastasis. IGF- 1: insulin-like growth factor; IGF-1R: IGF-1 receptor; TGFβ: transformation growth factor b; PDGF: platelet-derived growth factors; VEGF: vascular endothelial growth factors; MMP: matrix metalloprotease; MT1-MMP: membrane type 1-MMP; ADAM: a disintegrin and metalloproteinase; ADAM-TS: ADAM with thrombospondin motifs. Solid line: activation; dotted line: inactivation.

B.4.2 The proprotein convertases and tumor growth

As described in chapter B.2, in order to maintain proliferation and growth, cancer cells secrete a large amount of growth factors and upregulate cell surface growth factor receptors. The majority of growth factors and their receptors are synthesized as inactive precursors that must be converted into active forms by proteolytic processing. One or more PCs were reported to activate some of these factors, including insulin-like growth factors (IGFs), platelet-derived growth factors (PDGFs) and TGFβs.

IGF-1 and IGF-1R: high circulating levels of IGF-1 and overexpression of its receptor IGF-1R were directly link to high risk of cancer (124) and to accelerated tumor progression (125), through their ability to increase proliferation and pro-survival signals. IGF-1 is synthesized as two precursor isoforms, pro-IGF-1A and B. Both are cleaved by furin, or less efficiently by PC5/6A, to generate active IGF-1 (126). Its receptor IGF-1R also requires furin cleavage to generate its α and β chain which are indispensible for IGF- 1 binding and intracellular signaling (127). Furin is the primary IGF-1R convertase, although other PCs can process pro-IGF-1R in the furin-deficient LoVo cell line (128). In agreement, inhibition of furin activity resulted in reduced level of active IGF-1R, resistance to IGF-1 and decreased cell growth (127).

PDGFs: The binding of the potent mitogenic factors PDGFs on their tyrosine kinase receptors (PDGFR) stimulates various cellular responses including cell growth, proliferation and differentiation. The upregulation of PDGFs and their receptors were reported in various human solid tumors, such as glioblastomas and prostate carcinomas (129). All PDGFs (A, B, C and D) form disulfide-linked homodimers, except PDGF-A and -B that can also form functional heterodimers. Both pro-PDGF-A and -B processing can be mediated by furin, PACE4, PC5/6 and PC7 (130, 131). But furin remains the most potent PDGF convertase (130, 131). Mutation of the PC-cleavage site RGRR81↓ led to an inhibition (partial processing) or even a complete abolition of the processing, which subsequently reduced downstream tyrosine phosphorylation signaling and cell growth (130, 131).

TGFβs: three isoforms, TGFβ1 to 3, form the TGFβ family which is also a part of the TGFβ superfamily comprising additionally activins, inhibins, bone morphogenetic proteins, growth differentiation factors and Müllerian inhibiting substance (132). TGFβs play a dual role in the progression of human cancers. They act as tumor suppressors early in carcinogenesis by inhibiting tumor cell growth and promoting differentiation and apoptosis, but switch to tumor promoters later in carcinogenesis by enhancing tumor growth, invasion and metastasis (133). The levels of TGFβ signaling pathway components are frequently altered in human cancer (134).

TGFβs are produced as inactive precursors and require two sequential proteolytic cleavages to be active as homodimers. The first cleavage separates the C-terminal prosegment from the N-terminal mature TGFβ. Furin is mainly responsible for this cleavage in TGFβ1 at RHRR283↓ in humans, although PACE4, PC5/6B and PC7 can partially substitute for furin activity (45). After the first cleavage, mature TGFβ1 remains non-covalently associated with its prosegment prior to secretion and requires a second cleavage within the prosegment to release active TGFβ1. Numerous cell surface or ECM located enzymes were reported to be responsible for this second cleavage, such as thrombospondin-1 (135), plasmin (136) as well as furin (137, 138). In contrast, although TGFβ2 contains the PC consensus cleavage site, it is insensitive to furin cleavage (139). This differential sensitivity of TGFβ1 and β2 to PCs is probably determined by the tertiary structure of their prosegment. The processing of TGFβ3 is not clearly determined. On the other hand, the expression of furin is upregulated by mature TGFβs via a positive feedback mechanism which further amplifies the levels of mature TGFβs and its downstream signaling (140).

B.4.3 The proprotein convertases and angiogenesis

Angiogenesis, the fifth hallmark of cancer, is necessary for expansion of tumor mass. Without angiogenesis, continued cell proliferation alone can only give rise to dormant, microscopic tumors of ~1 mm3 or less (141). VEGFs, a group of secreted glycoproteins including VEGF-A to -D and placental growth factor, are major stimulators

of angiogenesis and lymphangiogenesis during carcinogenesis (104). Among these factors, VEGF-C and -D were reported to be activated by PCs.

VEGF-C and -D bind both to vascular endothelial receptor VEGFR-2 and to lymphatic endothelial receptor VEGFR-3. VEGF-C expression was detected in a wide range of human cancers and was associated with lymphatic propagation of carcinomas (142, 143). VEGF-D enhances tumor growth and metastatic spread via the lymphatic vessels in animal models (144). It is associated with metastasis and decreased survival in human cancer (145). VEGFs consist of a central VEGF homology domain (VHD), which contains the binding sites for VEGF receptors, and N- and C-terminal propeptides flanking the VHD. They are initially secreted as homodimers of the precursors that can be proteolytically processed to remove the N- and C-terminal propeptides (146, 147). Furin, PC5/6 and PC7 are involved in the proteolytic processing of proVEGF-C at HSIIRR227↓ that releases the C-terminal propeptide (148). As to proVEGF-D, furin and PC5/6 mediate the cleavage of both C- and N-terminal propeptides, whereas PC7 promotes the cleavage of the C-terminal propeptide only (149).

B.4.4 The proprotein convertases and cellular invasion

In more advanced stages of cancer progression, cancer cells become invasive and invade surrounding tissues, basement membrane and eventually leave their primary site. Controlling cell motility and adhesiveness by cell adhesion molecules (CAMs) and promoting degradation of the ECM by matrix metalloproteases (MMPs) are the crucial elements in the process of cell invasion.

Cadherins: these type-1 transmembrane proteins are key cell adhesion molecules (CAMs) that mediate Ca2+-dependent homophilic intercellular interaction. The functions of cadherins in cancer pathogenesis have been extensively investigated and the invasive/metastatic potential of tumor cells inversely correlates with cadherin expression (150, 151). Epithelial-cadherin (E-cadherin), the prototype of cadherins, has been characterized as a potent suppressor of cell invasion and metastasis (151). E-cadherin is synthesized as an inactive precursor that requires the post-translational cleavage of the prosegment at the

N-terminal to be active. Pro-E-cadherin was first shown to be cleaved by furin at RQKR154↓, when they were coexpressed (152). The cleavage also occurs in the furin- deficient LoVo cell line, suggesting that PC5/6 and/or PACE4 also mediate the processing (152, 153). Neural-cadherin (N-cadherin) is also produced as a precursor. It requires the removal of its prosegment to exert its adhesive property (154). It was recently shown to be activated by furin via the cleavage at RQKR159↓DW, as the lack of furin expression resulted in non-adhesive pro-N-cadherin accumulation at the glioma cell surface, thereby rendering tumor cell less invasive (155). In contrast, PC5/6 was shown to inactivate N- cadherin by processing it at a more C-terminal site to that of furin at RIRSDR187↓DK , thereby eliminating the critical tryptophan161 for homotypic interactions of N-cadherin (156).

Integrins: a large family of heterodimeric transmembrane glycoproteins contain two distinct chains, called α and β subunits. Integrins attach cells either to ligands present on other cells, or to the ECM of the basement membrane, which provide the traction required for tumor cell invasion. Integrins can also enhance tumor cell survival by preventing pro-apoptotic signaling cascades initiated by anoikis and increasing survival signaling. The expression of some integrins correlates with tumor progression and patient survival (157). αvβ3, α5β1 and αvβ6 that are usually expressed at low or undetectable levels in most adult epithelia are highly upregulated in some tumors. Whereas α2β1 expression decreases in tumor cells which promotes their dissemination (158). Some α subunits possess a potential dibasic aa PC-cleavage site. The endoproteolytic processing has been shown to be required for the proper conformation and adhesion properties of integrins (159). The subunits α3, α4, α5, α6 and αv were identified in vitro as furin and/or PC5/6 substrates (160-162)

MMPs: these metalloproteases participate in the ECM remodeling and allow the disseminated cancer cells to invade adjacent tissues. Located at the cancer cell and stroma interface, the ECM is the barrier that limits the spreading of tumor cells. Degradation of the ECM is therefore one of the first steps in tumor invasion. MMPs, with more than 20 members identified so far, are the principal matrix-degrading proteases. All the MMPs

are synthesized as inactive zymogens that must be processed in order to express enzymatic activities. The activation requires two sequential cleavages. The first cleavage attacks the proteinase-susceptible region located in the middle of the propeptide, and induces conformational changes in the propeptide that renders the second activation site to be readily auto-cleaved (163). Stromelysin-3 (MMP-11) was the first MMP that has been demonstrated to be directly processed into its enzymatic active form. Furin and PACE4 are responsible for this cleavage at RNRQKR98↓ in the TGN that separates the catalytic domain from the prosegment prior to the secretion of the mature form into the extracellular space (164, 165). MMP-2 is one of the most studied MMPs. It is implicated in the digestion of type IV collagen and gelatin and has been frequently found elevated in human cancers. It correlates with advanced tumor stages, increased metastasis and poor prognosis (166). Co-expression of furin and pro-MMP-2 in the COS-1 cell line results in a direct cleavage of the N-terminal portion of the propeptide at RKPR72↓ in the TGN (167). Interestingly, unlike stromelysin-3, MMP-2 is inactivated by this cleavage. Indeed, the activation of MMP-2 follows the conventional two-step proteolytic processing. Pro-MMP-2 is first cleaved at the cell surface by membrane type 1-MMP (MT1-MMP) (168), which belongs to a sub-family of six MMPs containing a transmembrane domain. This cleavage is followed by an intermolecular autocatalytic cleavage that results in a complete removal of the prosegment and generates the active MMP-2 (169). MT1-MMP therefore enhances the invasiveness of cancer cells by controlling MMP-2 activation (170). It also participates in a direct digestion of the ECM components such as collagen types I (171), laminin-5 (172) and intergrins (173). To make things more complicated, furin and PC5/6 cleave pro-MT1-MMP at RRKR111↓ leading to the activation of MT1-MMP (174, 175). Furin thus acts as a double-edged sword: directly incapacitating MMP-2 and indirectly activating MMP-2 following activation of MT1-MMP.

B.4.5 Cross-talk between the substrates of the proprotein convertases

To make matters even more complicated, PC substrates interact actively with each other to regulate cancer progression. For instance, the expression of MT1-MMP and

MMP-2 can be stimulated by IGF-1 (176) and TGFβs (177), respectively. As the latter two proteins are activated by PCs, this reinforces the regulation of PC/MT1-MMP/MMP- 2 activation cascade (Figure 8).

Some PC-activated substrates can be further regulated by a disintegrin and metalloproteinases (ADAMs) which are a family of membrane-associated metalloproteinases, and related ADAMs with thrombospondin type-1 modules (ADAM- TSs). ADAMs act as ‘sheddases’ that cleave transmembrane protein ectodomains at sites close to the cell membrane. The upregulation of ADAMs has been documented for a number of cancers, with correlations to the degree of progression. The first and most studied member of this family, ADAM17, is known as tumor necrosis factor-α converting enzyme (TACE). It is processed and activated by PC1/3, PC2, furin, PACE4 and PC5/6 (178, 179). ADAM10 (180), ADAM12 (181) and ADAM15 (182) were next shown to be activated by furin as well. ADAM10 has also been shown to be activated by PC7 (183).

Activated ADAM17 sheds and solubilizes membrane bound growth factors, such as pro-TNFα, pro-TGFα and heparin-binding epidermal growth factor, which initiating a paracrine signaling thereby enhancing tumor proliferation (184). ADAM17 membrane activity is increased by the IGF-1/IGF-1R pathway (185), which is also initially activated by PCs. Thus, PCs control both upstream and downstream events of ADAM17 activation cascade. ADAM10 mediates proteolysis of cell surface E-cadherin (186) and ADAM15 release the ectodomain of N-cadherin, which leads to inactivation of N-cadherin (187). Hence, ADAM10 and ADAM15 increase cancer cell migration and metastatic properties. In contrast, both N- and E-cadherin are initially activated by furin, but N-cadherin and possibly E-cadherin are also inactivated by PC5/6, suggesting that PCs can tightly control the activity of adhesion molecules either by a direct activation/inactivation, or by an indirect inactivation via ADAMs.

PCs play a complex and diverse role in tumor progression (Figure 8), they can activate oncoproteins (e.g., IGF-1), as well as inactivate it (e.g., MMP-2). PCs can exert their functions at very upstream positions of the proteolytic cascade to activate/inactivate

oncoproteins (e.g., MT1-MMP-MMP-2, ADAM-Cadherin) or at downstream sites of the cascade to directly process oncoproteins (e.g., MMP-2, cadherin). Thus, it is impossible to draw a general conclusion regarding the roles PCs in cancer/metastasis. They can act either as pro-oncogenic or as anti-carcinogenic enzymes.

B.4.6 The proprotein convertases modify cancer incidence by controlling cholesterol homeostasis

Unlike the well-established association between coronary heart disease and cholesterol (188), the relation between cancer risk and cholesterol remains controversial. However, there are convincing human studies showing that cholesterol is an important cancer risk modifier (189-193). Two major concerns limit us to demonstrate a clear association between cancer and cholesterol. i) In most cancer patients, plasma cholesterol levels decline prior to the diagnosis of cancer (194). It is probably due to this preclinical phenomenon that some earlier cohort studies suggested an association between low cholesterol levels and increased risk of cancer (195, 196). ii) Cancer is a highly heterogeneous disease. The correlation between cholesterol and cancer differs markedly by cancer types. Inverse associations were demonstrated for breast, cervix (189, 190), liver (192, 193), stomach and lung (only in men) (190) cancers, while positive associations were found for brain (189), colon (190) and prostate cancers (191). For those cancers which are positively associated with high cholesterol levels, lowering cholesterol can be considered as an approach to reduce incidence and control tumor progression. Some clinical studies have been explored the effectiveness of lowering cholesterol to treat these cancers. Statins, the most commonly prescribed cholesterol lowering drugs, have been shown to reduce the risk of prostate (197) and colorectal cancer (198). In addition, a statin clinical trial for the prevention and/or treatment of prostate cancer has been proposed and will be conducted in Canada (199). Both SKI-1 and PCSK9 are key regulators of cholesterol homeostasis. SKI-1 proteolytically activates SREBPs (200), which control the expression of the key enzymes in cholesterol and fatty acid biosynthesis (64). PCSK9 targets the LDLR for lysosomal degradation, thereby decreasing cholesterol clearance. Thus, it is reasonable to presume that SKI-1 and PCSK9

increase the risk of the cancers that are positively associated with cholesterol. Inhibition of SKI-1 and PCSK9 will be beneficial to control these cancer progressions.

Taken together, above evidences strongly support the important roles of PCs at different stages of tumor progression, from tumorigenesis to metastasis, from tumor cells to stromal cells. Since PCs control the activation/inactivation states of key cancer- associated proteins, they merited the name “master switches” of cancer development (123).

B.5 Targeting the proprotein convertases in cancer therapy

As described in the previous section, there are convincing evidences and still incoming studies that support the important implication of PCs in accelerating cancer progression. Thus, the development of PC inhibitors will have a potential clinical implication. The approaches used to develop PC specific inhibitors include active-site- directed chloromethylketone (CMK), reversible peptide-based inhibitors and various engineered protein-based inhibitors (16)

B.5.1 Inhibitors of the proprotein convertases

Natural inhibitors: proSAAS and 7B2 are specific for PC1/3 and PC2, respectively. The C-terminal part of the neuroendocrine protein 7B2 inhibits PC2 with high potency (Ki=57 nM) (201). More importantly, 7B2 is essential for PC2 activation (202). The binding of pro7B2 to proPC2 in the ER allows the productive folding of proPC2 and its exit from this compartment. Pro7B2 is first cleaved in the TGN by furin (203) and the C- terminal domain acts as an inhibitor of mature PC2 until the complex reaches immature secretory granules, where the C-terminal domain of 7B2 is further cleaved by PC2, thereby liberating the active enzyme (204). The granin-like protein proSAAS was identified as an endogenous inhibitor of PC1/3 (Ki in micromolar range) (205). Unlike 7B2, it is not necessary for PC1/3 stability, folding and activation. Recently, plasminogen activator inhibitor-1 was shown to form an SDS-stable complex with furin, which leads

to the inhibition of the intra-Golgi activity of furin (IC50 = 480 nM) and thereby inhibiting furin-dependent insulin receptor and ADAM17 maturation (206).

Active-site-directed CMK: the first synthesized inhibitor consists of peptidyl sequences encompassing the PC recognition motif coupled to a CMK moiety. These molecules enter the active site of enzyme resulting in the formation of a covalent enzyme-inhibitor complex (207, 208). Among this type of inhibitors, decanoyl-RVKR- CMK is the best known inhibitor that inhibits all PCs with a Ki of 0.1-2 nM (209). Unfortunately, the lack of selectivity, plus the cytotoxicity and the unstable nature of the chloromethane group preclude its use in human.

Peptide-based inhibitors: are potent competitive inhibitors. These molecules are 4- to 20-aa long that contain the minimal PC cleavage motif. The most promising and best characterized inhibitors of this kind are the PC prosegments. As discussed in Chapter A.4, to be activated, all the PCs, except PCSK9, undergo autocleavage events to release the active enzyme from their own inhibitory prosegment. Although the prosegments have been found to be potent with Ki in the low nanomolar range, they show modest specificity toward their cognate PCs (39, 210, 211). For example, in trans the prosegment

of furin has higher inhibitory potency towards PC5/6 (IC50=0.4 nM) than furin (IC50=4 nM) (211).

Protein-based inhibitors: are developed based on the expression of the proteins containing either natural or bioengineered PC consensus motif. This class of inhibitors includes α2-macroglobulin (212), proteinase inhibitor 8 (213), turkey ovomucoid third domain (214) and α1-antitrypsin (209, 215). Among these, the most relevant one is an α1-antitrypsin variant: α1-antitrypsin Portland (α1-PDX). α1-antitrypsin is a physiological inhibitor of neutrophil elastase. A point mutation at its active site AIPM358 to AIPR358 changed the specificity from an inhibitor of elastase into an inhibitor of thrombin, named α1-antitrypsin Pittsburg. Then, a second engineered mutation changed the alanine at P4 to an arginine in order to obtain a PC consensus site RIPR358, which gave rise to a potent PC competitive inhibitor α1-PDX (216). Although this inhibitor shows higher selectivity towards furin with a Ki of 0.6 nM, it can inhibit, to some extent, all other PCs, especially PC5/6 with a Ki of 2.3 nM (209, 215).

B.5.2 Inhibition of the proprotein convertase as a potential cancer therapy

Many studies reported the successful use of PC inhibitors in vitro and in vivo to inhibit tumorigenesis, tumor invasiveness and metastasis. The group of Klein-Szanto demonstrated that the inhibition of PC activities by the overexpression of α1-PDX or furin prosegment in head and neck squamous cell carcinoma (217, 218) and in the astrocytoma cell line (219) diminished their proliferation, tumorigenicity and invasiveness in vitro and in vivo. Recently, this group also demonstrated that dec-RVKR- CMK blocked skin squamous cell carcinoma proliferation and reduced skin tumor incidence and metastasis in a chemical carcinogenesis mouse model (220). The group of Khatib obtained similar results in a human colon adenocarcinoma HT-29 cells. The overexpression of α1-PDX in this cell line results in lower liver metastases after being injected into nude mice (T cell immunodeficient mice) (127).

Recently, an American group developed an autologous tumor cell vaccine, named Fang vaccine (221), which provides the individual patient’s tumor antigen array and in addition a plasmid encoding granulocyte-macrophage colony-stimulating factor (GMCSF) to stimulate potent antitumor immunity. The vaccine also contains a bifunctional short hairpin RNA interference (bi-shRNAi) targeting furin in order to reduce the expression of furin. Furin can activate both TGFβ1 and TGFβ2, which promote tumor malignancy and suppress immune responses (222). Knockdown of furin mRNA will simultaneously block TGFβ1 and TGFβ2 activities, thereby preventing endogenous immunosuppression. The outcome of phase I trial is promising. Fang vaccine elicited immune response in 50% patients which correlated with prolonged survival. In contrast, some studies showed PC inhibition rather enhances cancer development. Luis group in France showed that inhibition of PC activity either by overexpressing α1-PDX or with dec-RVKR-CMK resulted in an elevated HT-29 cell migration and invasion as well as increased lung metastasis after HT-29 cells inoculation into immunosuppressed newborn rats (223). Moreover, a recent study found that the risk of developing aggressive hepatocellular carcinoma is enhanced by low levels of furin (224).

In spite of the seemingly bright future of the clinical applications of PC inhibitors in cancer therapy, there remains a lot of unresolved issues and potential problems. Further efforts should be made to clarify the function of specific PCs in cancer development and

to develop selective PC inhibitors that are capable of blocking specifically one PC activity, without affecting the other members of the family.

My thesis work focused on characterizing function of PC5/6 and PCSK9 in cancer progression by using tissues specific or total knockout mice (manuscript 1 and 3). My work also explored some natural inhibitory mechanism which is specific to one PC, named PC5/6 (manuscript 2).

Chapter II: Manuscript 1

Title: The proprotein convertase PC5/6 is protective against intestinal tumorigenesis: in vivo mouse model. Author: Xiaowei Sun, Rachid Essalmani, Nabil G Seidah and Annik Prat Journal: Molecular Cancer. 2009, volume 8: p73-81.

2.1 Preface

Although for a long time, PCs have been associated with cancer, the specific role of PC5/6 in cancer progression has not yet been vigorously investigated. Some gene profiling studies demonstrated the overexpression of PC5/6 in the aggressive NSCLC cell line (123). In contrast, PC5/6 mRNA is only detected in 2 out of 30 human breast cancer tumors (116). PC5/6 was also shown to cleave some key players in cancer development, such as TGFβ, PDGFs, E-cadherin, which largely overlap with furin substrates (See Chapter A.2.4). Taken together, none of these studies could reveal a specific physiological function of PC5/6 in either promoting tumorigenesis or protecting from tumor progression. This chapter describes the first attempt to assess the in vivo role of PC5/6 in cancer development.

Because the small intestine is the second highest site of PC5/6 expression (35) and the PC5/6 KO mice die at birth, we generated PC5/6 intestine-specific knockout (iKO) mice using a Villin-Cre recombinase system. QPCR analysis confirmed that the efficiency of KO in small intestines is more than 90%. Furthermore, to induce intestinal tumors, both flox/flox and iKO mice were crossed with ApcMin/+ mice carrying a heterozygote Min (multiple intestinal neoplasia) mutation in the tumor suppressor gene Apc (96, 97). We first examined the mRNA level of PC5/6 and found a systematical

downregulation of PC5/6 in human colorectal carcinomas as well as in ApcMin/+ mice intestinal tumors. The lack of PC5/6 in intestine led to a higher tumor number, especially in the duodenum, and a premature mortality. This study suggested that PC5/6 plays a protective role in intestinal tumorigenesis.

This was the first study exploring the in vivo role of PC5/6 in tumor development and successfully demonstrated a protective role of PC5/6 in intestinal tumorigenesis. These results were unexpected, as most of the PC5/6 substrates, e.g. IGF-1, TGFβ, promote tumorigenesis. Our study suggested that PC5/6 activates some tumor suppressors or inactivates some oncoproteins in intestine. This study will serve as a stepping-stone to identify cancer-related substrates of PC5/6 and to discover its functions in other types of cancer and tissues.

37 The proprotein convertase PC5/6 is protective against intestinal

tumorigenesis: in vivo mouse model

Xiaowei Sun, Rachid Essalmani, Nabil G. Seidah and Annik Prat*

Laboratory of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, affiliated to the University of Montreal, 110 Pine Avenue West, Montreal, Quebec,

Canada

*Corresponding author: Dr. Annik Prat, Clinical Research Institute of Montreal, 110 Pine Avenue West, Montreal, Quebec, Canada, H2W1R7 Telephone: 1-514-987-5738 E-mail: [email protected]

Emails: Xiaowei Sun – [email protected] Rachid Essalmani – [email protected] Nabil G. Seidah – [email protected]

Abstract

Background: The secretory basic amino acid-specific proprotein convertases (PCs) have often been associated with cancer/metastasis. By controlling the cleavage of cancer- associated proteins, PCs play key roles in multiple steps of cancer development. Most analyses of the implication of PCs in cancer/metastasis relied on the use of in vitro overexpression systems or inhibitors that can affect more than one PC. Aside from the role of furin in salivary gland tumorigenesis, no other in vivo genetic model of PC- knockout was reported in relation to cancer development. Results: Since PC5/6 is highly expressed in the small intestine, the present study examined its in vivo role in intestinal tumorigenesis. Analysis of human intestinal tumors at various stages showed a systematic down-regulation of PC5/6 expression. Since gene inactivation of PC5/6 leads to lethality at birth, we generated mice lacking PC5/6 in enterocytes and analyzed the impact of the presence or absence of this PC in the mouse ApcMin/+ model that develops numerous adenocarcinomas along the intestinal tract. This resulted in viable mice with almost no expression of PC5/6 in small intestine, but with no overt phenotype. The data showed that by themselves ApcMin/+ tumors express lower levels of PC5/6 mRNA, and that the lack of PC5/6 in enterocytes results in a significantly higher tumor number in the duodenum, with a similar trend in other intestinal segments. Finally, the absence of PC5/6 is also associated with a premature mortality of ApcMin/+ mice. Conclusion: Overall, these data suggest that intestinal PC5/6 is protective towards tumorigenesis, especially in mouse duodenum, and possibly in human colon.

Background Nine secretory proprotein convertases (PCs) of the subtilisin/kexin type (genes PCSK1 to PCSK9) were identified in mammals and are known as: PC1/3, PC2, furin, PC4, PC5/6, PACE4, PC7, SKI-1/S1P and PCSK9 [1,2]. The first 7 convertases cleave secretory precursor proteins at single or paired basic residues [2], whereas SKI-1/S1P [3] and PCSK9 [4] do not require a basic residue at the cleavage site. The basic amino acid (aa)- specific convertases process precursors of growth factors, receptors, polypeptide hormones, adhesion molecules, proteases, as well as cell surface proteins of infectious viruses and bacteria [2]. In some cases, furin and/or PC5/6 inactivate proteins such as endothelial and lipoprotein lipases [5], PCSK9 [6] and N-cadherin (Maret D. et al., submitted).

Overexpression of PC5/6, PACE4 and furin revealed that these proteinases can often cleave the same precursors, indicating a functional redundancy [6-12]. Evidence for in vivo redundancy was provided by furin inactivation in the liver, which revealed that most of the precursors analyzed were still processed, although to a lesser extent, in the absence of this ubiquitous convertase [13]. In contrast, in vivo studies demonstrated that in a spatio-temporal manner furin can uniquely process the Ac45 subunit of the vacuolar type H+-ATPase in pancreatic β-cells [14] and PC5/6 the TGFβ-like growth and differentiation factor Gdf11 in the developing embryo [15,16].

Various precursors cleaved by overexpressed furin, PC5/6, PACE4 and PC7 have been previously implicated in cancer and associated metastatic processes [17-19]. A correlation between the mRNA levels of some of these PCs and the degree of tumorigenicity has been reported [9,18-27]. Furthermore, injection/implantation of various cell lines expressing PC inhibitors, such as the antitrypsin derivative α1-PDX [9,12,20,24,27,28] or the inhibitory prodomain of PCs [26] suggested a critical role of the PCs in tumor growth and/or metastasis.

The convertase PC5/6 (previously known as PC5 or PC6) was characterized in 1993 and shown to be composed of two differentially spliced isoforms, a short 915 aa soluble

PC5/6A [29], and a long membrane-bound 1877 aa PC5/6B [30]. In adult rodents, PC5/6 exhibits a wide tissue distribution [29], which in mice when analyzed by quantitative PCR (QPCR) revealed that the adrenal cortex and small intestine are the richest sources of PC5/6A and PC5/6B, respectively [31]. However, the function of PC5/6 in these tissues has not been addressed. PC5/6 can bind cell surface heparan sulfate proteoglycans and tissue inhibitors of metalloproteases via its C-terminal Cys-rich domain [32]. It also seems to differ from the other convertases in that it can get activated at the cell surface [1,33]. Knockout of the PC5/6 gene (Pcsk5) revealed that Pcsk5-/- animals die at birth due to multiple malformations, including defects in antero-posterior patterning and heart formation [15,16]. Defective specification of segment identity, which leads to an increased number of thoracic and lumbar vertebrae and lack of tail, is likely due to the absence of processing of Gdf11 [15,16,34]. No obvious malformations were seen in the small intestine of Pcsk5-/- embryos [15].

The specific role of PC5/6 in tumorigenesis/metastasis has not yet been investigated. PC5/6 expression was not detected in human breast, and generally not induced in breast cancer since it was present in only 2/30 tumors [35]. In contrast, its mRNA levels seem to correlate with tumor aggressiveness of head and neck- and lung tumor-derived cell lines [18], suggesting that PC5/6 may play a different role in metastasis compared to tumor growth. Whether this is related to its ability to process adhesion molecules [36], including the α-chain of various integrins [7,37] and N-cadherin (Maret D. et al., submitted) is not yet clear.

Colorectal cancer is the third most common form of cancer in the Western world. As a mouse model for this pathology, we used the ApcMin/+ strain that harbors a heterozygote Min (multiple intestinal neoplasia) mutation in the Apc (adenomatous polyposis coli) gene. These mice spontaneously develop polyps all along the small intestine [38,39]. In order to assess the role of PC5/6 in intestinal tumorigenesis, we generated PC5/6 intestine-specific knockout mice (iKO) and crossed them with ApcMin/+ mice. Our data show that mice carrying the Min mutation but lacking PC5/6 tend to exhibit a higher tumor number than ApcMin/+ mice, especially in duodenum, and die significantly earlier.

Methods

Animals Tg(Vil-cre) mice (stock number 004586) [40] and ApcMin/+ mice (stock number 002020) [39] were from The Jackson Laboratory. Conditional knockout mice, in which the proximal promoter and exon 1 of Pcsk5 were flanked with loxP sites (Pcsk5flox/flox) [15], were crossed with Tg(Vil-cre) mice that express Cre under the control of the villin promoter. After two generations, Pcsk5flox/flox mice carrying (intestinal KO; iKO) or not (wild type; WT) one copy of the transgene were obtained and further intercrossed, yielding the F4 progeny used in this study, which exhibits a mixed background consisting of ~70% C57BL/6; 25% 129Sv and less than 5% SJL. When expressed, Cre leads to the recombination of the two loxP sites present in Pcsk5, resulting in the excision of ~3kb of DNA including exon 1 (1 alleles) and thereby gene inactivation.

Tumor scoring in mouse intestine

Four month old mice were sacrificed by CO2 asphyxiation, and the whole intestine was immediately removed and rinsed with ice-cold PBS. The intestine was divided into duodenum, jejunum, ileum and colon. All sections were carefully split longitudinally, fixed in 8% paraformaldehyde, stained with 8% methylene blue and the tumors were counted under a binocular microscope.

Quantitative RT-PCR Tissue samples were dissected from PBS-rinsed intestine. Total RNA was extracted using Trizol reagent (Invitrogen), as recommended by the manufacturer. Typically, 250 ng of total RNA were used for cDNA synthesis in a total volume of 20 µL using

SuperScript II reverse transcriptase, 25 µg/mL oligo(dT)12–18, 0.5 mM 2'-deoxynucleoside 5'-triphosphates, and 40 U of RNaseOUT, all products from Life Technologies, and used according to the recommendations of the manufacturer. cDNAs of human adenocarcinomas were purchased from Origene. The quantitative PCR (QPCR) was performed as previously described [41]. Specific primers (Table 1) were used for the

42 simultaneous amplification of the normalizing cDNA for ribosomal protein S14 (human) or S16 (mouse), and the gene of interest.

In situ hybridization Mouse cRNA probes corresponding to the coding region for aa 20 to 348 of PC5/6 were synthesized using 35S-UTP and 35S-CTP (>1,000 Ci/mmol; Amersham Bioscience, Piscataway, NJ). Cryosections (8-10 m) were fixed for 1 hour in 4% formaldehyde and hybridized overnight at 55°C as previously described [42]. For autoradiography, the sections were dipped in photographic emulsion (NTB-2, Kodak, Rochester, NY), exposed for 6–12 days, and developed in D19 solution (Kodak).

PCNA immunohistochemistry Tissues were fixed overnight in 4% paraformaldehyde at 4ºC and embedded in paraffin. Proliferation cell nuclear antigen (PCNA) was visualized in sections of 6 m thickness by incubation with a mouse antibody (1:50; Vector laboratories, Burlingame, CA) and a biotin-labeled secondary antibody (PerkinElmer, Boston, MA), and revelation with the Vectastain kit (Vector laboratories). Sections were also counterstained with hematoxylin and eosin.

Table 1: Sequences of primers used for QPCR

Assessed Forward Primer Reverse Primer mRNA human ACTCTTCAGAGGGTGGCTA GCTGGAACAGTTCTTGAATC

PC5AB mouse TGACCACTCTTCAGAGAATGGATAC GAGATACCCACTAGGGCAGC

PC5AB mouse AGGATTCAAGAACTGTTCCA AGCATACAGAAGCCTCCTT

PC5A mouse GCAATGCCTCCCACTCCC TGCTCGTAAAACTCAGCCTCC

PC5B mouse CATGACTACTCTGCTGATGG GAACGAGAGTGAACTTGGTC

Furin

Cre ATGATCCGAATAACTACCTG ACAATATTTACATTGGTCCAG human GGCAGACCGAGATGAATCCTCA CAGGTCCAGGGGTCTTGGTCC

S14 mouse GCTACCAGGGCCTTTGAGATG AGGAGCGATTTGCTGGTGTGG

S16

Results

Expression of PC5/6 is lower in intestinal tumors versus adjacent normal tissues Mining cancer gene expression database (www.oncomine.org) revealed that PC5/6 expression was significantly reduced in 7 out of 10 tumor types (P < 0.0001); [see figure S1; Additional file 1]. Since PC5/6 expression is highest in the adult small intestine [29,31], and as no data were available for intestinal cancers, PC5/6 mRNA levels were analyzed by QPCR in 22 human colon tumors at stages I, II, III or IV and compared to those of their match-paired normal adjacent tissue (Figure 1A). PC5/6 expression was on average ~7.6-fold lower in these human tumors. To assess whether PC5/6 was similarly regulated in mouse, we used the ApcMin/+ mice, which spontaneously develop numerous tumors in the small intestine due to the heterozygote mutation Min in the Apc gene. This mutation was originally discovered in patients suffering from familial adenomatous polyposis and frequently found in sporadic colorectal cancers [38,39]. ApcMin/+-induced tumors in the mouse small intestine constitute a good model for colonic tumorigenesis in human. We first quantified the expression levels of furin, PC5/6, PACE4 and PC7, which transit through the constitutive secretory pathway and cleave their substrates after basic residues [2]. While PACE4 and PC7 did not show any significant change, furin and PC5/6 mRNA levels were on average ~1.5-fold higher (P = 0.003) and lower (P = 0.0008), respectively (Figure 1B). Closer analysis of the duodenum-, jejunum- and ileum- associated tumors versus their adjacent normal tissues revealed a 1.9-, 1.2- and 1.4-fold higher furin levels, respectively, and a 2-, 1.7- and 1.1-fold lower PC5/6 expression, respectively (Figure 1C). Using specific primers, we showed that this lower level primarily affected PC5/6B transcripts [see figure S2; Additional file 2], which dominate in intestine [31]. The above data thus indicated that PC5/6 is down-regulated in many tumor types, including intestinal ones, and that in the latter furin undergoes an opposite up-regulation. Both PC5/6 and furin exhibited the greatest changes in the duodenum. These data prompted us to verify if intestinal tumorigenesis was favored in absence of PC5/6.

Figure S1: Down-regulation of PC5/6 expression in various cancers. Description: Datasets were retrieved from ONCOMINE (a cancer microarray database and integrated data-mining platform) with a threshold of P < 0.0001. PC5/6 expression value in tumors was log2 transformed and normalized by that in the adjacent normal tissue.

Figure 1: Decreased expression of PC5/6 in intestinal tumors versus adjacent normal tissues. (A) RNA samples from human colonic adenocarcinomas (stage I, II, III or IV) and their adjacent normal tissues were submitted to QPCR analysis (n = 6, 7, 7 and 2 for stages I, II, III and IV, respectively). (B) In each small intestine section (duodenum, jejunum and ileum) from 3 ApcMin/+ mice, 2 tumors and their adjacent normal tissue (6 couples/section) were dissected and assessed for the expression levels of furin, PC5/6, PACE4 and PC7 by QPCR. Normalized expression values are shown for the 18 samples of normal tissues and 18 samples of tumors. (C) Expression of PC5/6 and furin in tumors was also analyzed by intestinal section. All mRNA levels in tumors were normalized to their respective normal tissue expression and have been log2 transformed, with the median of the total 18 samples set to 0. *, P < 0.05; **, P < 0.005; ***, P < 5.10-11 (Student’s t test).

Figure 2: Decreased expression of PC5/6B, but not PC5/6A, in intestinal tumors versus adjacent normal tissues. Description: Specific primers were used for QPCR analysis of the two PC5/6 isoforms. Normal (N) and tumoral (T) expression of PC5A and PC5B was assessed by using

48 isoform-specific primers. Error bars represent SEM and n = 6 for each intestine section. *, P < 0.05 for PC5/6B (Student’s t test)

Conditional inactivation of Pcsk5 in enterocytes To explore the in vivo role of PC5/6 in intestinal tumor formation, we specifically inactivated its gene in enterocytes using a loxP/Cre system. Pcsk5flox/flox mice were bred to Tg(Vil-cre) mice that expressed the Cre recombinase under the direction of the villin promoter, specifically expressed in enterocytes [40]. Pcsk5flox/flox mice carrying one copy of the transgene (iKO; Tg+/0) or none (WT; Tg0/0) were generated. To verify that the presence of the transgene resulted in an efficient inactivation of Pcsk5 in enterocytes, we analyzed PC5/6 mRNA levels using QPCR and in situ hybridization in 3 mice of each genotype. Duodenum, jejunum, ileum and colon sections were dissected for further RNA extraction and tissue sectioning. Cre expression under the villin promoter in iKO mice was highest in duodenum and progressively diminished along the intestinal tract to reach ~25% of the duodenum level in the distal colon (Figure 2A). In WT mice, PC5/6 expression is elevated in the small intestine, especially in the duodenum, as compared to colon (Figure 2B). Indicative of the Cre efficiency all along the intestine, the absolute numbers of PC5/6 mRNA remaining in all sections of iKO intestine were very similar : 1.6 to 3.1 PC5/6 mRNA / 1000 S16 mRNA. Furthermore, in situ hybridization with a PC5/6 cRNA probe confirmed that PC5/6 transcripts were strongly reduced in iKO intestinal enterocytes (Figure 3). The low residual expression observed by QPCR (Figure 2B) and in situ hybridization labeling suggest that in the small intestine PC5/6 is mainly expressed in enterocytes, but to a much less extent expressed in other cell types all along the intestine. Finally, the morphology and proliferation of enterocytes was assessed by immunohistochemistry. No gross malformation was observed and labeling with PCNA, a marker for proliferation, was not significantly different between the two genotypes [see figure S3; Additional file 3].

Figure 2: Efficient inactivation of Pcsk5 in iKO mice. (A) Cre expression was assessed in intestinal segments from 3 iKO mice. Expression values were normalized to that of S16 mRNA. (B) PC5/6 expression was quantified in each intestinal segment from 3 WT and 3 iKO mice and normalized to that of S16. Error bars represent SEM.

Figure 3: Detection of PC5/6 transcripts in WT and iKO intestine by in situ hybridization. Cryosections were hybridized with a PC5/6-specific probe, stained with cresyl violet and dipped in an autoradiography emulsion. The extent of 35S labeling was visualized on dark field.

Figure S3: Unaffected enterocyte proliferation in iKO mice. Description: Representative PCNA immunohistochemistry of WT and iKO jejunum sections is shown. Quantitative analysis was achieved by counting PCNA-positive nuclei in 3 random fields in duodenum, jejunun and ileum in 3 mice per genotype. Error bars represent SEM.

PC5/6 deficiency has a significant impact on Min mutation-induced tumorigenesis in the duodenum Intercrossing of [Pcsk5flox/flox Tg(Vil-cre)+/0] with [Pcsk5flox/flox ApcMin/+] generates 25% mice that carry only the Min mutation (WTMin), and exhibit normal levels of PC5/6 in intestine. Another 25% of these mice carry both the Min mutation and the Cre transgene (iKOMin), and lack PC5/6 expression in enterocytes. Duodenum, jejunum and ileum from 11 WTMin mice and 17 iKOMin mice were dissected out, opened longitudinally and stained with methylene blue (Figure 4A). All the tumors, including those exceeding 2 mm in diameter, were counted along the entire section of each tissue. The average tumor density

(tumors/cm) in the duodenum of iKOMin mice was significantly higher than that in WTMin mice (P = 0.01; Figure 4B). In iKO mice, the duodenum is the tissue in which the PC5/6 drop was the most drastic (Figure 2). However, although this trend was observed in other intestinal sections, it did not reach statistical significance, and the total number of tumors in iKOMin mice, 58 versus 46 in WT mice, was not significantly higher (Figure 4C). In addition, the numbers of large tumors (>2 mm; Figure 4C) were very similar in both cases. Overall, this analysis indicates that only in duodenum does the loss of PC5/6 significantly enhance intestinal tumorigenesis.

PC5/6 deficiency shortens the half-life of ApcMin/+ mice Apc Min/ + mice having a pure C57BL/6 background were reported to die by 120 days of age [38,39], likely due to severe chronic anemia [38]. In this study, WTMin mice exhibited a longer half-life of 180 days, possibly due to their mixed background (see Methods). However, in the absence of intestinal PC5/6, this half-life was significantly shortened to 140 days (P = 0.03; Figure 5), suggesting that PC5/6 exerts a protective effect on these mice. ApcMin/+ mice develop anemia with a severity that seems to depend on the density of intestinal adenomas [38]. Considering that iKOMin mice had a trend for higher numbers of tumors, especially in the duodenum, premature death of iKOMin mice could be the result of more severe chronic anemia [38], which could be exacerbated by multiple hemorrhages, as observed in the liver and subcutaneously in PC5/6 knockout mice [15]. In the future, it may be valuable to examine whether PC5/6 levels correlate with the survival rate, or intestinal bleeding/anemia of patients that suffer from colorectal carcinomas.

Figure 4: Intestinal tumor formation in WTMin and iKOMin mice. (A) Representative sections of WTMin and iKOMin ileum stained with methylene blue. Arrows point at visualized tumors. (B) Total tumor numbers and large tumor (> 2 mm) numbers in WTMin and iKOMin intestine of 4 month-old WTMin (n = 11) and iKOMin mice (n =17). (C) Numbers of tumors per cm of duodenum, jejunum, ileum or colon in the above mice are shown. *, P < 0.05 (Student’s t test)

Figure 5. Decreased survival of ApcMin/+ mice in the absence of PC5/6. Survival rates of WTMin (n=21) and iKOMin (n=22) mice were compared. P = 0.03 (Log- rank test)

Figure S4: Relative expression of PC5/6 and furin in WT intestine. Description: The PC5/6 and furin expression was assessed on each intestinal segment from 3 WT mice. The expression value was normalized to that of S16 mRNA. Error bars represent SEM.

Discussion

The use of general PC-inhibitors such as α1-PDX or pro-furin revealed that PC-inhibition decrease tumorigenesis and metastasis in nude mice [9,12,20,26], but enhance metastasis in immunosuppressed newborn rats [43]. This is probably due to the ability of overexpressed PC-inhibitors to block the activity of more than one convertase [44], which may exert opposite regulating effects and modulate multiple processes. Thus, mice lacking a specific convertase should represent a more powerful tool to assess the specific function of a single convertase. Of all the PC knockout mice, those lacking furin [45] and PC5/6 [15,16] exhibit a fully penetrant embryonic lethal phenotype, precluding their use in adult mouse studies. Tissue-specific knockouts thus provide a potential approach to test their effect in cancer/metastasis. So far, the in vivo role of a specific PC in tumorigenesis was only investigated in mice lacking furin in salivary glands among other tissues [46]. In these mice, the simultaneous inactivation of furin and overexpression of the PLAG1 transcription factor, which induced the formation of adenomas in salivary glands, showed that the absence of furin delayed tumorigenesis [46], suggesting a pro- tumorigenic effect of furin. The present study is the first attempt to assess the role of PC5/6 in cancer development using knockout mice. The impact of PC5/6 has been analyzed here exclusively in vivo, using the ApcMin/+ intestinal tumorigenesis model. We first evaluated PC5/6 mRNA levels in intestinal tumors versus normal tissue obtained from colon cancer patients (Figure 1A) or ApcMin/+ mice (Figure 1B and C), and showed that PC5/6 is systematically downregulated in intestinal tumors. To probe the role of PC5/6 in tumorigenesis, we compared the number and size of intestinal tumors in ApcMin/+ mice lacking or not PC5/6 (Figure 4). The data showed a trend for an enhanced tumorigenesis in PC5/6-deficient mice, reaching significance only in the duodenum (Figure 4B) where PC5/6 is primarily expressed (Figure 2A), suggesting that it may exert specific functions therein. This result was unexpected in view of the reported reduced tumorigenesis by general PC-inhibitors [18,20-22].

Could PC5/6 specifically process a tumor-suppressor or inactivate a tumorigenic factor, and hence act in an opposite fashion to other basic aa-specific PCs? Opposing functions can occur by cleavage of the same substrate at different sites, as illustrated by the ability of furin to activate the cell adhesion molecule N-cadherin and PC5/6 to inactivate it (Maret D. et al., submitted). In the duodenum, PC5/6 was only 1.7-fold less abundant than furin, while its ratio to furin was 3- to 10-fold lower in other segments of the intestine [see figure S4; Additional file 4]. Thus, tumorigenesis in the duodenum may depend on the balance between activation and/or inactivation of proteins by resident furin and PC5/6, respectively. In tumors of the duodenum, PC5/6 mRNA levels are ~7-fold lower than those of furin (Figure 1C). Thus, the pro-tumorigenic properties of furin [46] may in some cases overshadow the protective effect of PC5/6. We surmise that within the duodenum, furin may activate precursors implicated in epithelial to mesenchymal transition, involved in early tumorigenesis and invasion/metastasis [47], such as E- cadherin [48] and TGF-β [49], while PC5/6 may inhibit tumorigenesis, e.g., via inactivation of adhesion proteins such as N-cadherin (Maret D. et al., submitted), resulting in a lower number of tumors.

Conclusions

Future studies aimed to identify the implicated substrates will require an extensive comparative analysis of ApcMin/+-induced tumors isolated from mice lacking PC5/6, furin or both in enterocytes. Whether the mechanism behind the shortened survival of ApcMin/+ mice lacking PC5/6 (Figure 5) is due to more severe hemorrhages resulting from a greater vessel fragility induced by the loss of PC5/6 [15] would require a more detailed examination. Furthermore, the importance of specific PCs in the invasion/metastasis process, which is heavily regulated by adhesion molecules processed by PCs [17,27] is yet to be fully investigated in an appropriate in vivo model. Finally, this is the first report that emphasizes the opposite roles of furin and PC5/6 in tumorigenesis. Thus, recently proposed treatments aimed to reduce furin activity [9,18-27] should include careful monitoring of their effects on PC5/6 levels and/or activity.

Abbreviations aa, amino acid; α1-PDX, α1-antitrypsin Portland; Apc, adenomatous polyposis coli; iKO, intestinal knockout of the Pcsk5 gene; Min, multiple intestinal neoplasia; PC, Proprotein convertase; PCNA, proliferation cell nuclear antigen; Pcsk5, Proprotein convertase subtilisin/kexin type 5; QPCR, quantitative RT-PCR; Tg, transgene; WT, wild type.

Competing interests

This work was supported by Canadian Institutes of Health Research grant # 44363, a Canada Chair # 201652, and a Strauss foundation grant. The authors declare that they have no competing interests.

Authors contributions

All authors read and approved the final manuscript.

XS carried out all the mouse analyses, tumor measurements and other experiments as well as the genotyping. RE generated the PC5/6 conditional knockout mice and helped in the analyses of their phenotypes, NGS participated in the design of the experiments, analysis of the data and writing of the manuscript, and AP was the major driver of the project implicated in all aspects of the research.

Acknowlegments

We thank Edwige and Martin Marcinkiewicz for their help for in the in situ hybridization analysis and Claudia Toulouse for the excellent animal care. We also acknowledge the editorial assistance of Brigitte Mary.

References

1. Seidah NG, Mayer G, Zaid A, Rousselet E, Nassoury N, Poirier S, Essalmani R, Prat A: The activation and physiological functions of the proprotein convertases. Int J Biochem Cell Biol 2008, 40:1111-1125. 2. Seidah NG, Chretien M: Proprotein and prohormone convertases: a family of subtilases generating diverse bioactive polypeptides. Brain Res 1999, 848:45- 62. 3. Pasquato A, Pullikotil P, Asselin MC, Vacatello M, Paolillo L, Ghezzo F, Basso F, Di Bello C, Dettin M, Seidah NG: The Proprotein Convertase SKI-1/S1P: in vitro analysis of lassa virus glycoprotein-derived substrates and ex vivo validation of irreversible peptide inhibitors. J Biol Chem 2006, 281:23471- 23481. 4. Seidah NG, Prat A: The proprotein convertases are potential targets in the treatment of dyslipidemia. J Mol Med 2007, 85:685-696. 5. Jin W, Fuki IV, Seidah NG, Benjannet S, Glick JM, Rader DJ: Proprotein convertases are responsible for proteolysis and inactivation of endothelial lipase. J Biol Chem 2005, 280:36551-36559. 6. Benjannet S, Rhainds D, Hamelin J, Nassoury N, Seidah NG: The proprotein convertase PCSK9 is inactivated by furin and/or PC5/6A: Functional consequences of natural mutations and post-translational modifications. J Biol Chem 2006, 281:30561-30572. 7. Lissitzky JC, Luis J, Munzer JS, Benjannet S, Parat F, Chretien M, Marvaldi J, Seidah NG: Endoproteolytic processing of integrin pro-alpha subunits involves the redundant function of furin and proprotein convertase (PC) 5A, but not paired basic amino acid converting enzyme (PACE) 4, PC5B or PC7. Biochem J 2000, 346 :133-138. 8. Benjannet S, Elagoz A, Wickham L, Mamarbachi M, Munzer JS, Basak A, Lazure C, Cromlish JA, Sisodia S, Checler F, Chrétien M, Seidah NG.: Post- translational processing of beta-secretase (beta-amyloid- converting enzyme)

and its ectodomain shedding. The pro- and transmembrane/cytosolic domains affect its cellular activity and amyloid-beta production. J Biol Chem 2001, 276:10879-10887. 9. Scamuffa N, Siegfried G, Bontemps Y, Ma L, Basak A, Cherel G, Calvo F, Seidah NG, Khatib AM: Selective inhibition of proprotein convertases represses the metastatic potential of human colorectal tumor cells. J Clin Invest 2008, 118:352-363. 10. Nour N, Basak A, Chretien M, Seidah NG: Structure-Function Analysis of the Prosegment of the Proprotein Convertase PC5A. J Biol Chem 2003, 278:2886- 2895. 11. Fujii Y, Sakaguchi T, Kiyotani K, Yoshida T: Comparison of substrate specificities against the fusion glycoprotein of virulent Newcastle disease virus between a chick embryo fibroblast processing protease and mammalian subtilisin-like proteases. Microbiol Immunol 1999, 43:133-140. 12. Khatib AM, Siegfried G, Prat A, Luis J, Chretien M, Metrakos P, Seidah NG: Inhibition of proprotein convertases is associated with loss of growth and tumorigenicity of HT-29 human colon carcinoma cells: importance of insulin-like growth factor-1 (IGF-1) receptor processing in IGF-1- mediated functions. J Biol Chem 2001, 276:30686-30693. 13. Roebroek AJ, Taylor NA, Louagie E, Pauli I, Smeijers L, Snellinx A, Lauwers A, Van de Ven WJ, Hartmann D, Creemers JW: Limited redundancy of the proprotein convertase furin in mouse liver. J Biol Chem 2004, 279:53442- 53450. 14. Louagie E, Taylor NA, Flamez D, Roebroek AJ, Bright NA, Meulemans S, Quintens R, Herrera PL, Schuit F, Van de Ven WJ, Creemers JW.: Role of furin in granular acidification in the endocrine pancreas: identification of the V- ATPase subunit Ac45 as a candidate substrate. Proc Natl Acad Sci U S A 2008, 105:12319-12324. 15. Essalmani R, Zaid A, Marcinkiewicz J, Chamberland A, Pasquato A, Seidah NG, Prat A: In vivo functions of the proprotein convertase PC5/6 during mouse

development: Gdf11 is a likely substrate. Proc Natl Acad Sci U S A 2008, 105:5750-5755. 16. Szumska D, Pieles G, Essalmani R, Bilski M, Mesnard D, Kaur K, Franklyn A, El Omari K, Jefferis J, Bentham J, Taylor JM, Schneider JE, Arnold SJ, Johnson P, Tymowska-Lalanne Z, Stammers D, Clarke K, Neubauer S, Morris A, Brown SD, Shaw-Smith C, Cama A, Capra V, Ragoussis J, Constam D, Seidah NG, Prat A, Bhattacharya S.: VACTERL/caudal regression/Currarino syndrome-like malformations in mice with mutation in the proprotein convertase Pcsk5. Genes Dev 2008, 22:1465-1477. 17. Khatib AM, Siegfried G, Chretien M, Metrakos P, Seidah NG: Proprotein convertases in tumor progression and malignancy : novel targets in cancer therapy. Am J Pathol 2002, 160:1921-1935. 18. Bassi DE, Fu J, Lopez DC, Klein-Szanto AJ: Proprotein convertases: "master switches" in the regulation of tumor growth and progression. Mol Carcinog 2005, 44:151-161. 19. Coppola JM, Bhojani MS, Ross BD, Rehemtulla A: A small-molecule furin inhibitor inhibits cancer cell motility and invasiveness. Neoplasia 2008, 10:363-370. 20. Bassi DE, Lopez DC, Mahloogi H, Zucker S, Thomas G, Klein-Szanto AJ: Furin inhibition results in absent or decreased invasiveness and tumorigenicity of human cancer cells. Proc Natl Acad Sci U S A 2001, 98:10326-10331. 21. Bassi DE, Mahloogi H, Klein-Szanto AJ: The proprotein convertases furin and PACE4 play a significant role in tumor progression. Mol Carcinog 2000, 28:63-69. 22. Khatib AM, Bassi D, Siegfried G, Klein-Szanto AJ, Ouafik L: Endo/exo- proteolysis in neoplastic progression and metastasis. J Mol Med 2005, 83:856- 864. 23. Cheng M, Xu N, Iwasiow B, Seidah N, Chretien M, Shiu RP: Elevated expression of proprotein convertases alters breast cancer cell growth in response to estrogen and tamoxifen. J Mol Endocrinol 2001, 26:95-105.

24. Siegfried G, Basak A, Cromlish JA, Benjannet S, Marcinkiewicz J, Chretien M, Seidah NG, Khatib AM: The secretory proprotein convertases furin, PC5, and PC7 activate VEGF-C to induce tumorigenesis. J Clin Invest 2003, 111:1723- 1732. 25. Mercapide J, Lopez DC, Bassi DE, Castresana JS, Thomas G, Klein-Szanto AJ: Inhibition of Furin-mediated Processing Results in Suppression of Astrocytoma Cell Growth and Invasiveness. Clin Cancer Res 2002, 8:1740- 1746. 26. Lopez DC, Bassi DE, Zucker S, Seidah NG, Klein-Szanto AJ: Human carcinoma cell growth and invasiveness is impaired by the propeptide of the ubiquitous proprotein convertase furin. Cancer Res 2005, 65:4162-4171. 27. Lapierre M, Siegfried G, Scamuffa N, Bontemps Y, Calvo F, Seidah NG, Khatib AM: Opposing function of the proprotein convertases furin and PACE4 on breast cancer cells' malignant phenotypes: role of tissue inhibitors of metalloproteinase-1. Cancer Res 2007, 67:9030-9034. 28. Siegfried G, Khatib AM, Benjannet S, Chretien M, Seidah NG: The proteolytic processing of pro-platelet-derived growth factor-A at RRKR(86) by members of the proprotein convertase family is functionally correlated to platelet- derived growth factor-A-induced functions and tumorigenicity. Cancer Res 2003, 63:1458-1463. 29. Lusson J, Vieau D, Hamelin J, Day R, Chretien M, Seidah NG: cDNA structure of the mouse and rat subtilisin/kexin-like PC5: a candidate proprotein convertase expressed in endocrine and nonendocrine cells. Proc Natl Acad Sci U S A 1993, 90:6691-6695. 30. Nakagawa T, Murakami K, Nakayama K: Identification of an isoform with an extremely large Cys-rich region of PC6, a Kex2-like processing endoprotease. FEBS Lett 1993, 327:165-171. 31. Essalmani R, Hamelin J, Marcinkiewicz J, Chamberland A, Mbikay M, Chretien M, Seidah NG, Prat A: Deletion of the gene encoding proprotein convertase 5/6 causes early embryonic lethality in the mouse. Mol Cell Biol 2006, 26:354- 361.

32. Nour N, Mayer G, Mort JS, Salvas A, Mbikay M, Morrison CJ, Overall CM, Seidah NG: The Cysteine-rich Domain of the Secreted Proprotein Convertases PC5A and PACE4 Functions as a Cell Surface Anchor and Interacts with Tissue Inhibitors of Metalloproteinases. Mol Biol Cell 2005, 16:5215-5226. 33. Mayer G, Hamelin J, Asselin MC, Pasquato A, Marcinkiewicz E, Tang M, Tabibzadeh S, Seidah NG: The regulated cell surface zymogen activation of the proprotein convertase PC5A directs the processing of its secretory substrates. J Biol Chem 2008, 283:2373-2384. 34. McPherron AC, Lawler AM, Lee SJ: Regulation of anterior/posterior patterning of the axial skeleton by growth/differentiation factor 11. Nat Genet 1999, 22:260-264. 35. Cheng M, Watson PH, Paterson JA, Seidah N, Chretien M, Shiu RP: Pro-protein convertase gene expression in human breast cancer. Int J Cancer 1997, 71:966-971. 36. Kalus I, Schnegelsberg B, Seidah NG, Kleene R, Schachner M: The proprotein convertase PC5A and a metalloprotease are involved in the proteolytic processing of the neural adhesion molecule L1. J Biol Chem 2003, 278:10381- 10388. 37. Bergeron E, Basak A, Decroly E, Seidah NG: Processing of alpha4 integrin by the proprotein convertases: histidine at position P6 regulates cleavage. Biochem J 2003, 373:475-484. 38. Moser AR, Pitot HC, Dove WF: A dominant mutation that predisposes to multiple intestinal neoplasia in the mouse. Science 1990, 247:322-324. 39. Su LK, Kinzler KW, Vogelstein B, Preisinger AC, Moser AR, Luongo C, Gould KA, Dove WF: Multiple intestinal neoplasia caused by a mutation in the murine homolog of the APC gene. Science 1992, 256:668-670. 40. Madison BB, Dunbar L, Qiao XT, Braunstein K, Braunstein E, Gumucio DL: Cis elements of the villin gene control expression in restricted domains of the vertical (crypt) and horizontal (duodenum, cecum) axes of the intestine. J Biol Chem 2002, 277:33275-33283.

41. Dubuc G, Chamberland A, Wassef H, Davignon J, Seidah NG, Bernier L, Prat A: Statins upregulate PCSK9, the gene encoding the proprotein convertase neural apoptosis-regulated convertase-1 implicated in familial hypercholesterolemia. Arterioscler Thromb Vasc Biol 2004, 24:1454-1459. 42. Marcinkiewicz M, Marcinkiewicz J, Chen A, Leclaire F, Chretien M, Richardson P: Nerve growth factor and proprotein convertases furin and PC7 in transected sciatic nerves and in nerve segments cultured in conditioned media: their presence in Schwann cells, macrophages, and smooth muscle cells. J Comp Neurol 1999, 403:471-485. 43. Nejjari M, Berthet V, Rigot V, Laforest S, Jacquier MF, Seidah NG, Remy L, Bruyneel E, Scoazec JY, Marvaldi J, Luis J.: Inhibition of proprotein convertases enhances cell migration and metastases development of human colon carcinoma cells in a rat model. Am J Pathol 2004, 164:1925-1933. 44. Benjannet S, Savaria D, Laslop A, Munzer JS, Chretien M, Marcinkiewicz M, Seidah NG: Alpha1-antitrypsin Portland inhibits processing of precursors mediated by proprotein convertases primarily within the constitutive secretory pathway. J Biol Chem 1997, 272:26210-26218. 45. Roebroek AJ, Umans L, Pauli IG, Robertson EJ, van Leuven F, Van de Ven WJ, Constam DB: Failure of ventral closure and axial rotation in embryos lacking the proprotein convertase Furin. Development 1998, 125:4863-4876. 46. De Vos L, Declercq J, Rosas GG, Van Damme B, Roebroek A, Vermorken F, Ceuppens J, Van D, V, Creemers J: MMTV-cre-mediated fur inactivation concomitant with PLAG1 proto-oncogene activation delays salivary gland tumorigenesis in mice. Int J Oncol 2008, 32:1073-1083. 47. Chen X, Halberg RB, Burch RP, Dove WF: Intestinal adenomagenesis involves core molecular signatures of the epithelial-mesenchymal transition. J Mol Histol 2008, 39:283-294. 48. Posthaus H, Dubois CM, Muller E: Novel insights into cadherin processing by subtilisin-like convertases. FEBS Lett 2003, 536:203-208.

49. Dubois CM, Blanchette F, Laprise MH, Leduc R, Grondin F, Seidah NG: Evidence that furin is an authentic transforming growth factor-beta1- converting enzyme. Am J Pathol 2001, 158:305-316.

2.3 Discussions and Conclusions

PCs exert their functions through activation or inactivation of their substrates. However, different PCs share some functional redundancy due to an overlapping substrates preference. This has been extensively demonstrated in in vitro studies, whereby a single precursor can be processed by several PCs. In contrast, each PC fulfills specific processing events in vivo, as revealed by the distinct phenotypes resulting from their respective gene inactivation in mice (225, 226). Furthermore, due to the lack of specificity of the PC inhibitors, knocking-out individual PC in an animal model, either completely or in a specific tissue seems to be the only way to investigate physiological function(s) of a given PC. Therefore, in order to study the specific function of PC5/6 in cancer development, we generated PC5/6 intestine-specific KO mice, rather than using the conventional inhibitors α1-PDX or PC prosegment overexpressing systems. The residual expression of PC5/6 in small intestine was less than 10%, which confirms the efficient inactivation of PC5/6.

The inactivation of PC5/6 in intestine of ApcMin/+ mice led to higher tumorigenesis and premature mortality. This result was quite unexpected, as the majority of previous reports revealed that general PC-inhibition decreased tumorigenesis and metastasis in mice (127, 217, 218, 220, 227). This work thus revealed a unique function of PC5/6 in cancer development, which was probably hidden by the inhibition of other PCs exerting opposite functions when using general PC-inhibitors, such as α1-PDX. It also emphasized the importance of targeting PC individually to discover its unique in vivo functions.

In this manuscript, we also demonstrated the differential expression of furin and PC5/6 in intestinal cancer. While PC5/6 is systematically downregulated in intestinal tumors, furin expression is upregulated. A very recent publication extended the study to human. QPCR analysis of the biopsy from endometrial cancer patients demonstrated that furin expression is consistently upregulated in tumors, whereas PC5/6, PACE4 and PC7 expressions are reduced in higher grade cancers (228).

The differential regulation of furin and PC5/6 expression implies that these two convertases play opposite roles in cancer development. The group of Colman in collaboration with our lab confirmed this hypothesis in malignant gliomas (155, 156). Semi-quantitative reverse-transcriptional PCR revealed that the expression of furin is lower in highly invasive U251 glioma cells relative to less invasive U343 cells (Figure 9A). In contrast, PC5/6A expression is high in U251 cells compared to that in U343 cells. As control, similar levels of PACE4 and PC7 were detected in U343 and U251 cells. Although this different expression of furin and PC5/6 is more heterogeneous in vivo, it is in general consistent with in vitro findings (Figure 9B). The biopsies from human brain tumors, such as glioblastomas, showed that higher grade tumors (CT-001, OP-132, OP- 122) express lower levels of furin compared to low grade ones (OP-71, CT-005). Whereas PC5/6A levels were elevated in higher grade brain tumors, except for OP-71 whose PC5/6A level was similar to OP-132 (Figure 9B). Further studies demonstrated that the aggressiveness of tumors or cancer cell lines can be indeed modulated by the different expression of furin and PC5/6A. This is in part due to a differential cleavage of N-Cadherin by furin and PC5/6A. In HeLa cells which express furin but not PC5/6, the coexpression of PC5/6A and proN-cadherin resulted in a ~20 kDa prosegment species detected by anti-proN-cadherin antibody (Figure 9C). This 20 kDa fragment is larger than the prosegment of N-cadherin (~17kDa) and corresponds to an N-terminal cleavage product that would result from the inactivation of N-cadherin. In fact, furin processes N- cadherin at the consensus activating site RQKR159↓DW, while PC5/6A processes the protein at a second site RIRSDR187↓DK (Figure 9D). The cleavage at the second site removes the downstream Trp161 residue which was shown to be crucial for cadherin mediated adhesion (154), therefore resulting in an inactivation of the adhesive properties of N-cadherin. In addition, the subcellular localization of PC5/6A is important to properly exert its proteolytic activities. PC5/6A lacking its C-terminal CRD (PC5∆CRD), which mediates cell surface anchoring of the enzyme via binding to HSPGs and the TIMPs, cannot generate the inactivating form of N-Cadherin (Figure 9C, right panel). Thus, PC5/6A probably cleaves N-Cadherin at the plasma membrane.

N-Cadherin mediates mesenchymal adhesion, while E-cadherin mediates epithelial adhesion. A switch from E-cadherin to N-cadherin, which induces the invasion of epithelial cell into stromal cells, was observed in a variety of carcinomas (229, 230). However, primary brain tumors do not undergo E- to N-cadherin transition and express intrinsically high level of N-cadherin. Thus, the aggressiveness of brain tumors can be modulated by the post-translation processing of proN-cadherin. In highly invasive gliomal cells, a significant amount of non-adhesive unprocessed pro-N-cadherin was present on the cell surface, which promotes cancer cell migration and invasion (155). A model was thus proposed in Figure 9E. In high grade gliomas, the endogenous low levels of furin and high levels of PC5/6 result in the presence at the cell surface of a mixture of the inactive unprocessed proN-cadherin and PC5/6-inactivated form of N-Cadherin lacking aa 160-187. These cells are more invasive and have the ability to form secondary metastases. Conversely, gliomas expressing high furin and low PC5/6A levels will have high levels of mature N-cadherin at the cell surface which enhances adhesiveness of tumor cells, thereby rendering them less invasive (156).

The above studies and present manuscript both support the idea that furin and PC5/6 exert opposite functions in some cancers. These data also suggest that PC functions are highly tissue specific. PC5/6 plays a protective role in intestinal cancer, whereas it promotes brain tumor invasiveness. Furin seems to play a pro-tumorigenic role, as most of its substrates are key cancer promoting factors. However, furin activity is associated with low invasive brain glioma tumors. Although PCs are promising drug targets in human cancer, the careful analysis of their in vivo function emphasizes the necessity to target selectively one PC without affecting other PC activities, and ideally to deliver the PC-inhibitor/silencer in a tissue-specific fashion. However, since PCs share high structural similarities, the discovery/development of a PC-specific inhibitor is a major challenge. Chapter III, thus deals with our attempt to search PC5/6-specific inhibitors.

A C

D B high grade low grade E

Figure 9: The opposite role of furin and PC5/6A in brain tumor progression. (Adapted from Maret et al, in press Neoplasia) (A) Semi-quantitative reverse-transcriptional PCR was carried out to quantify the expression of furin, PC5A, PACE4 and PC7 in U343 and U251 cells, GAPDH expression was used as a normalizing control. (B) Quantitative PCR analysis of PC5/6A and furin in biopsy of human brain tumors. The number of mRNA is normalized to that of S14. (C) HeLa cells were transiently transfected with N-cadherin and full length PC5/6A or empty vector (pCXN2 and pIRES-EGFP). PC5/6A with a truncated cysteine-rich domain (PC5/6A∆CRD) was transfected instead of full length PC5/6A.The conditioned medium was concentrated and resolved on a 15% SDS-PAGE gel and N-cadherin cleavage were detected with the anti-proN-cadherin antibody. The cleavage products with molecular weight of 17 kDa and 20 kDa corresponded to processing of proN-cadherin at the consensus site, and at the second site, respectively. (D) Schematic illustration of two N- cadherin processing site by furin or PC5/6A. (E) Schematic scheme depicting surface cadherin expression during glioma progression. Glioma cells in the main tumor mass associate with each other via N-cadherin mediated adhesion (low PC5/6A and high furin expression). The accumulation of surface proN-cadherin and inactive N-cadherin due to low furin and high PC5/6 expression allows cell detachment. These cells with are more invasive, but in general do not metastasize.

Chapter III: Manuscript 2

Title: Latent transforming growth factor beta-binding proteins-2 and -3 inhibit the proprotein convertase 5/6A.

Author: Xiaowei Sun, Rachid Essalmani, Delia Susan-Resiga, Annik Prat, and Nabil G. Seidah Journal: Journal of Biological Chemistry. 2011, volume 286, No 33: p29063-73.

3.1 Preface

PC5/6 is an essential convertase, as PC5/6 KO mice die at birth (40, 231). Furthermore, non synonymous mutations in the human PCSK5 gene were found in VACTERL (vertebral, anorectal, cardiac tracheoesphageal, renal, limb malformation) patients that exhibit birth defects (231). PC5/6 is a protective protease, as specific intestinal KO mice are more susceptible to intestinal tumorigenesis (See Chapter II). PC5/6 endothelial KO mice exhibit cardiovascular hypothrophy (232). Thus, we dwelled on the question of: “what are the molecular mechanisms that regulate PC5/6 activity?”. So far, some studies only partially answered this query. First, PC5/6 enzymatic activity can be inhibited by its own prosegment (nM IC50), in spite of its lack of specificity towards furin (39). Second, the subcellular localization of PC5/6A is regulated by its CRD that localizes PC5/6A to HSPG-rich domains at the cell surface and/or in the extracellular matrix (ECM). This cellular localization is important for the enzyme to efficiently cleave cell surface proteins (13, 14). Lastly, adrenocorticotropic hormone or 8- bromo-cyclic AMP could stimulate PC5/6A transcription as well as its zymogen activation by enhancing the second autocatalytic cleavage of its prosegment at the cell surface, thereby increasing the releasing of the active form of PC5/6A (14).

In this manuscript, we reported a novel regulatory mechanism of PC5/6A activity through co-expression of its endogenous inhibitors, latent transforming growth factor β binding proteins 2 and 3 (LTBP-2 and -3). The co-expression of PC5/6A with LTBP-2 and -3 inhibits Gdf11 processing and its downstream Smad2 signaling. In fact, LTBP-2

and -3 form a stable complex with proPC5/6A, which leads to the specific targeting of the zymogen to the ECM, thereby reducing the availability of the active enzyme in media. Moreover, the extensive co-localization of PC5/6 and LTBP-3 mRNAs in mice both at embryonic and adult stages further emphasized the physiological relevance of this inhibition.

Most PCs do not exit the ER unless a prior autocatalytic cleavage of their prosegment at Site 1 occurs. The complex of prosegment•PC then exits the ER towards the Golgi before reaching its final destination in an active form in other subcellular organelles (cell surface, endosomes, dense core secretory granules) or to be secreted. PC2 was the first convertase that has been shown to exit the ER as a zymogen proPC2 in complex with its chaperone pro7B2 and only becomes autocleaved/activated in dense core secretory granules (6, 16) Our study provided the second example whereby proPC5/6A can exit the ER in complex with LTBP-2 or -3, providing a new mechanism for regulating the activity of PC5/6A.

LATENT TGF BINDING PROTEINS-2 AND -3 INHIBIT THE PROPROTEIN CONVERTASE 5/6A

Xiaowei Sun, Rachid Essalmani, Delia Susan-Resiga, Annik Prat and Nabil G. Seidah¶

Laboratory of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal (affiliated to the University of Montreal) 110 Pine Ave. West, Montreal, QC H2W 1R7, Canada

¶Address correspondence to: Nabil G. Seidah: Laboratory of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, 110 Pine Avenue West Montreal, QC H2W 1R7, Canada. Tel: (514) 987-5609; E-mail: [email protected]

ABSTRACT

The basic amino acid-specific proprotein convertase 5/6 (PC5/6) is an essential secretory protease, as knockout mice die at birth and exhibit multiple homeotic transformation defects, including impaired bone morphogenesis and lung structure. Some of the observed defects were attributed to impaired processing of the TGF--like growth differentiating factor 11 precursor (proGdf11). In this work we present evidence that the latent TGF- binding proteins 2 and 3 (LTBP-2, -3) inhibit the extracellular processing of proGdf11 by PC5/6A. This is partly due to the binding of LTBPs in the endoplasmic reticulum (ER) to the zymogen proPC5/6A, thus allowing the complex to exit the ER and be sequestered as an inactive zymogen in the extracellular matrix (ECM), but not at the cell surface. This results in lower levels of PC5/6A in the media, without affecting those of PACE4, Furin or a soluble form of PC7. The secreted soluble protease specific activity of PC5/6A or a variant lacking the C-terminal Cys-rich domain (PC5/6-CRD) are significantly decreased when co-expressed with LTBPs in cells. A similar enzymatic inhibition seems to apply to PACE4 and Furin. In situ hybridization analyses revealed extensive co-localization of PC5/6 and LTBP-3 mRNAs in mice at embryonic day 15.5 and post partum day 1. In conclusion, this is the first time that a zymogen of the proprotein convertases was shown to exit the ER in the presence of LTBPs, representing a potential novel mechanism for the regulation of PC5/6A activity, e.g., in tissues such as bone and lung where LTBP-3 and PC5/6 co-localize.

INTRODUCTION

The mammalian proprotein convertases (PCs) form a family of nine serine proteinases related to subtilisin that primarily modify the activation state of a wide range of bioactive proteins. Seven PCs, PC1 (also known as PC3), PC2, Furin, PC4, PC5/6 (also known as PC5 or PC6), PACE4, and PC7, cleave protein precursors at basic sites during their transit through the secretory pathway and/or at the cell surface. Among these basic amino acid-specific PCs, Furin, PC5/6, PACE4, and PC7 are ubiquitous or widely distributed, although they exhibit characteristic patterns of expression in specific tissues and cells. PC5/6 is the only member of the PC family that exists as two isoforms: soluble PC5/6A (1) and membrane-bound PC5/6B, which has an extended C-terminal Cys-rich domain (CRD) (2). Except in the small intestine and kidney, PC5/6A is the major isoform in all other tissues (3). PC5/6A is synthesized as an inactive zymogen (proPC5/6A). It undergoes a first autocatalytic processing in the endoplasmic reticulum (ER) at RTKR116↓ (supplemental Fig. S1), resulting in a tight binding complex of the inhibitory prosegment with the protease, allowing the protein to exit the ER. It is then activated by a second autocatalytic cleavage within the prosegment at RTIKR84↓ (Fig. S1), which mostly occurs on the cell surface, where PC5/6A is anchored through its CRD that binds to heparan sulfate proteoglycans (HSPGs) (4). PC5/6 knockout (KO) mice die at birth and the newborn pups recapitulate all the phenotypes observed in mice lacking growth and differentiation factor 11 (Gdf11, also known as bone morphogenic protein 11, BMP11) (5), including an altered antero-posterior patterning with extra thoracic and lumbar vertebrae, lack of tail and kidney agenesis (6,7). Gdf11 thus seems selectively cleaved by PC5/6 during development. In agreement, in vitro and ex vivo analyses showed a high selectivity of PC5/6 for Gdf11 compared to the other constitutively secreted PCs, PACE4, Furin and PC7 (6). Gdf11 belongs to the transforming growth factor  (TGF) superfamily, which includes activins, nodals, BMPs, growth differentiating factors and canonical TGFs. These TGF- like factors form homodimers that are cleaved at site 1 (S1) intracellularly or extracellularly into N-terminal inhibitory prodomains and mature C-terminal domains that remain non- covalently associated. The prodomain of canonical TGFs interacts with latent TGFbinding proteins (LTBPs) that facilitate the secretion of the ligand and target the latent complex to the extracellular matrix (ECM) (8). LTBPs belong to the LTBP/fibrillin superfamily, a group of high molecular weight ECM protein that contains several 8-cysteine repeats. LTBP-1,-3 and -

4 form a disulfide bond with TGF prodomains through a Cys in their third 8-cysteine repeat, resulting in a large complex (9). Upon secretion, this complex accumulates in the ECM (10), likely via binding of LTBPs to fibrillin-1 (11), and awaits a local activation that requires a second cleavage at site 2 (S2) within the prodomain to release the mature and active TGF ligand. Despite the identification of PC5/6 as the enzyme responsible for the first cleavage of proGdf11, little is known about how and where the cleavage occurs. The interaction between Gdf11 and LTBPs has never been investigated. Myostatin (also known as Gdf8) is the closest member to Gdf11, sharing 92% amino acid identity in their mature domain and 49% identity in their prodomain (12). It was recently demonstrated that LTBP-2 and -3 interact non- covalently with pro-myostatin early along the secretory pathway. LTBP-3 furthermore reduces pro-myostatin secretion by retaining it in the ECM and therefore inhibits its cleavage by membrane-bound Furin (13). As a consequence, LTBP-3 negatively regulates myostatin signaling. Whether this inhibitory effect also occurs with Gdf11 is not known. We thus analyzed the effect of LTBP-2 or LTBP-3 on Gdf11 trafficking, processing and activity. Herein, we demonstrated that PC5/6A cleaves proGdf11 extracellularly and that LTBP-2 and -3 inhibit this cleavage and consequently decrease the Gdf11-mediated Smad2 signaling. We also showed that inhibition of Gdf11 cleavage is partly due to proPC5/6A sequestration in the ECM.

EXPERIMENTAL PROCEDURES

Expression Constructs − Mouse 7B2 (14,15), PC1/3, PC2 (16), PC5/6A (17), PC5/6B (18), PC5/6-CRD, PC5/6AR116A (17), human PACE4 (17), human Furin (19) and soluble rat PC7 (20), with or without a C-terminal V5 tag, were expressed using pIRES2-EGFP vectors (Clontech). Mouse Gdf11 (21) and HA epitope-tagged human LTBP-2 and mouse LTBP-3 (13) were expressed using pcDNA3 vectors.

Cell culture and transfection − COS-1 and HEK293 cell lines were grown in Dulbecco’s modified Eagle’s medium with 10% fetal bovine serum (Invitrogen). All cells were maintained at 37°C under 5% CO2. At about 80–90% confluence, COS-1 and HEK293 cells were transiently transfected using Lipofectamine 2000 (Invitrogen) and Effectene (Qiagen), respectively. Stable transfectants of proGdf11 were obtained in HEK293 and COS-1 cells upon hygromycin B selection.

Cell surface biotinylation, immunoprecipitation, and Western blotting − Cultured cells were washed with serum-free medium 24h post-transfection and incubated with serum-free medium for the following 24h. Media were then collected, and cells were lysed in 50 mM Tris-HCl, pH 7.8, 150 mM NaCl, 1% Nonidet P-40, 0.5% sodium deoxycholate and 0.1% SDS, in the presence of a mixture of protease inhibitors (Roche Applied Science). Lysates were centrifuged and supernatants were collected. For media swap experiments, serum-free media were collected at 48h post-transfection and swapped. Following 24h incubations, media and cells were collected. For cell surface biotinylation, HEK293 cells were transiently transfected. 48h post- transfection, the cells were washed with cold PBS adjusted to pH8.0 and biotinylated with 0.2 mg/ml of sulfo-NHS-LC-Biotin (Pierce) for 30 min at 4°C. The cells were then incubated with 100mM glycine for 5 min to quench the reaction. As a control, the cells are treated with 0.05% trypsin for 15 min on the ice to remove all cell surface protein. Trypsin was then inactivated with 10% fetal bovine serum. The cells were finally washed with PBS and harvested as described above. The cell lysates were then incubated with streptavidin coupling with agarose (Fluka) for 12h. The beads were washed six times by lysis buffer and then resolved by Western blotting. For immunoprecipitation, cell lysates and media were incubated overnight at 4°C with a PC5/6 prosegment rabbit polyclonal antibody (Ab) pPC5 (18) (1:100;), anti-V5 (1:500; Sigma) or anti-HA (1:500; Covance). Protein A-agarose beads were then added to the antigen-antibody complexes, incubated for 3h, and washed six times with the above lysis buffer and one time with cold PBS buffer. Protein samples were heated in reducing Laemmli buffer, resolved on SDS-polyacrylamide gels, electrotransfered onto PVDF membranes, incubated with specific primary and secondary antibodies, and revealed by chemiluminescence (Amersham). The following Ab were used: anti-pPC5 (1:2000), anti-PC7 (1:10,000; (22)), anti-mPC1-NT (1:2000, (16)), anti-mPC2 (1:2000, (16)), anti-Furin (1:1000, Alexis), -actin (1:5000; Sigma), anti-pSmad2 (1:1000; Cell signaling), anti-protein C (1:3000; Roche), anti-HA (1:3000; Covance) and horseradish peroxidase (HRP)-conjugated mouse anti-V5 (1:10,000; Sigma). Bound primary Ab were detected using anti-mouse IgG-HRP or anti-rabbit-IgG-HRP secondary Ab (both at 1:10,000; Amersham).

Immunocytochemistry − COS-1 cells were platted on glass bottom culture dishes (MatTek) and transfected the following day. After 24h, cells were washed three times with PBS and fixed in 3.7% paraformaldehyde for 10 min at room temperature. For intracellular labeling, cells were permeabilized in methanol for 3 min at -20°C, then washed in PBS and incubated for 5 min in 150 mM glycine. Cells, either permeabilized or not, were washed once in PBS,

incubated for 30 min in 1% BSA in PBS (blocking solution) and further incubated overnight at 4°C with monoclonal mouse anti-V5 (1:200), anti-pPC5/6 (1:100), anti-protein C (1:200) or anti-HA (1:500) in blocking solution. The next day, cells were washed four times with PBS and incubated for 45 min with secondary Ab: anti-rabbit IgGs or anti-mouse IgGs coupled to either Alexa-fluor-488 (green), 555 (red), or 647 (blue) (Molecular Probes, Eugene, OR). Cells were then washed four times with PBS and mounted in glycerol 1,4- diazabicyclo[2.2.2]octane (DABCO; Sigma-Aldrich). Immunofluorescence analyses were performed with a Zeiss LSM-710 confocal microscope.

In situ hybridization– To generate mouse LTBP-3 cRNA, a fragment of cDNA corresponding aa 995-1257 were amplified by PCR using the primers 5’CATATTGTTTGGGGCAGAGATCT and 5’GCCAGCTTTGCAGACACAG, and then subcloned into pDrive vector (Qiagen). Mouse PC5/6A cRNA probes corresponding to the coding region for aa 20 to 348 were described previously (23). Both probe were synthesized using 35S-UTP and 35S-CTP (>1,000 Ci/mmol; Amersham Bioscience). Cryosections (8-10 μm) were fixed for 1h in 4% formaldehyde and hybridized overnight at 55°C. For autoradiography, the sections were dipped in photographic emulsion (NTB-2, Kodak), exposed for 5 days, and developed in D19 solution (Kodak).

In vitro activity assay − Enzymatic in vitro assays were performed in 100 l of buffer (2 mM CaCl2, 25 mM Tris-HCl, pH 7.0) at 37°C in the presence of 100 M of the PC-substrate pyroglutamic acid-RTKR-7-amido-4-methyl-coumarin and 60 l of medium from HEK293 cells. The release of free 7-amino-4-methylcoumarin was detected with a Spectra MAX GEMINI EM microplate spectrofluorimeter (Molecular Devices; excitation, 360 nm; emission, 460 nm) (24). To calculate the relative fluorescence units (RFU), the background fluorescence was subtracted from fluorescent readings of each well. The enzymatic activity was determined by the slope of the linear portion of the data plot of RFU versus time and expressed in RFU/min.

RESULTS

PC5/6A is not responsible for proGdf11 processing at S2 – It was previously reported that PC5/6 can cleave TGF-like factors and BMP4 at both S1 and S2 (25). To examine whether proGdf11 could be cleaved by PC5/6A and/or PC5/6B at S2, we co-expressed proGdf11 carrying an N-terminal protein C (pC) and C-terminal FLAG tags (Fig. 1A) with PC5/6A or PC5/6B in HEK293 cells. Analysis of the media 48h post-transfection by Western blotting using a pC-specific Ab revealed the efficient cleavage by both PC5/6 isoforms of the Gdf11 precursor (proGdf11, 53 kDa) into a 37 kDa species corresponding to the N-terminal prodomain (Fig. 1B). We did not observe any lower molecular mass product (expected ~25 kDa), indicating that PC5/6A and PC5/6B do not cleave proGdf11 at another site (S2) within the prodomain (Fig. 1B). PC5/6 seems to be the unique PC responsible for Gdf11 processing in vivo during embryonic development (6,7). However, the ability of the three other constitutively secreted PCs, Furin, PC7 and PACE4, to cleave proGdf11 at S1, and possibly at another site, was also assessed. While Furin and PACE4 both cleaved proGdf11 at S1, albeit to a lesser extent than PC5/6, PC7 did not. Here also as for PC5/6, no cleavage at another site by these PCs was observed (Fig. 1B), suggesting that none of the PCs are responsible for cleavage at S2. However, a second cleavage of the prodomain of Gdf11 at S2 has been reported to be performed by the metalloprotease BMP-1/Tolloid (21,26).

ProGdf11 is cleaved by PC5/6A extracellularly – Previous results showed that PC5/6A is activated at the cell surface upon a second autocatalytic cleavage of its prosegment (4). We thus hypothesized that proGdf11 could be cleaved extracellularly either at the cell surface or in the media. We first separately expressed in HEK293 cells proGdf11, PC5/6A, or an unrelated protein 7B2 as control. 7B2 is a neuroendocrine-specific protein (14,27). Its primary function is to specifically bind the zymogen proPC2 in the ER and consequently to allow the productive folding of proPC2 and its exit from this compartment (28,29). 7B2 is first cleaved in the Golgi by Furin (30) and the C-terminal domain acts as an inhibitor of mature PC2 until the complex reaches immature secretory granules, where the C-terminal domain of 7B2 is further cleaved by PC2, thereby liberating the active enzyme allowing it to act in trans on other substrates (27,31). After 24h, the media were collected and incubated for 12h with HEK293 cells transiently transfected with cDNAs coding for either proGdf11 or PC5/6A, and then analyzed the media by immunoblotting. On one hand, addition of a medium containing PC5/6A to cells expressing proGdf11 led to a ~80% cleavage versus the basal control ~40% cleavage observed with 7B2 (Fig. 2). On the other hand, addition of a medium containing proGdf11 to

cells expressing PC5/6A led to its complete cleavage into Gdf11 (Fig. 2). These data demonstrate that PC5/6A can cleave proGdf11 at the cell surface and/or in the medium.

Fig. 1 PC5/6A does not cleave proGdf11 at site S2. (A) Schematic representation of the proGdf11 structure showing the position of the N- terminal protein C (pC) and C-terminal FLAG (FG) tags, the amino acid sequence surrounding the cleavage site at Arg296↓ (site S1). And the speculated site 2. (B) Expression of proGdf11 in HEK293 cells alone (-) or with different PCs, including PC5/6A, PC5/6B, Furin, PC7 and PACE4. At 48h post transfection the media were analyzed by Western blotting using anti-pC Ab. The migration positions of proGdf11 (53 kDa) and its prodomain (37 kDa) are shown.

Fig. S1 Alignment of the prosegments of mouse PC5/6 and human PACE4 and Furin. Identical residues are emphasized as well as the two autocatalytic cleavage sites at

Arg116↓ and Arg84↓ for mPC5/6, Arg149↓ and Arg117↓ for hPACE4 and Arg107↓ and

Arg86↓ for hFurin.

Fig. 2 ProGdf11 is cleaved extracellularly by PC5/6A. The media from HEK293 cells transfected individually with V5-tagged PC5/6A, pC-tagged proGdf11 or an unrelated protein 7B2 were collected 24h after transfection and incubated with HEK293 cells overexpressing either proGdf11 or PC5/6A for 12h. Western blot analysis of the media used anti-pC Ab to reveal proGdf11 and its prodomain, and anti-V5 Ab to reveal PC5/6A.

Fig. 3 LTBP-2 and -3 interact with proGdf11 without affecting its secretion. ProGdf11, HA-tagged LTBP-2 (L2) or LTBP -3 (L3) and/or 7B2, a negative control protein, were co-expressed in HEK293 cells. Cell lysates were immunoprecipitated with anti-HA Ab and the immunoprecipitated proteins, or lysates and media were analyzed by Western blotting using anti-HA to reveal LTBP-2 or -3 and anti-pC to reveal Gdf11. Note that, in cell extracts, proGdf11 migrates as two bands with the upper one only seen in the media, likely due to terminal N-glycosylation trimming of secreted proGdf11 in the Golgi apparatus.

LTBP-2 and -3 interact with proGdf11 without affecting its secretion – LTBP-2 and -3 were shown to interact intracellularly with pro-myostatin, the closest TGF-like member to Gdf11, and to sequester it in the ECM (13). To examine whether LTBPs play a similar role on proGdf11, we individually co-expressed in HEK293 cells their HA-tagged forms with proGdf11 or 7B2 as a control (Fig. 3). The migration positions of proGdf11 (~53 kDa), LTBP-2 (~250 kDa) and LTBP-3 (~150 kDa) and the levels of their co-expression were first detected by Western blotting of the input lysates (Fig. 3, left panel). Evidence for binding of proGdf11 with LTBP-2 or -3 was obtained by immunoprecipitation of cell lysates with the HA Ab and revelation by the pC Ab (Fig. 3, middle panel). The small amount of proGdf11 seen in cells only expressing Gdf11 and 7B2, is likely due to a non-specific pull-down of proGdf11 by the HA Ab. In contrast to myostatin (13), proGdf11 in the media of cells co- expressing either 7B2, LTBP-2 or LTBP-3 was found at similar levels, indicating that it is not retained in cells or ECM by these LTBPs (Fig. 3, right panel). To further investigate the cellular interaction between Gdf11 and LTBPs, they were transiently expressed either alone or together in COS-1 cells, and immunocytochemistry was

performed using pC and HA Ab. Under permeabilizing conditions, we observed an almost complete co-localization of LTBP-2 or -3 with proGdf11 in a perinuclear ER-like compartment (Fig. S2, left panel). When the cells were not permeabilized, proGdf11 and LTBP-2 or -3 co-localized at the cell surface (Fig. S2, right panel). However, Gdf11 can bind the cell surface even in absence of either LTBPs (Fig. S2, top panels), suggesting that the Gdf11-LTBP interaction is not critical for the intracellular and cell surface localization of Gdf11.

Fig. S2 Co-localization of Gdf11 and LTBP-2 and -3 in COS-1 cells. Gdf11 was co-expressed transiently alone or with LTBP-2 or -3 in COS-1 cells. The intracellular or cell surface co-localization of LTBP-2 or -3 (anti-HA Ab; green) and proGdf11 (anti-pC Ab; red) was analyzed in permeabilized or non-permeabilized COS-1 cells, respectively. Arrows point to sites where co-localizations were evident. Bar = 10 μm. These data are representative of three independent experiments.

LTBP-2 and -3 retain proPC5/6A in the ECM and prevent its maturation – Although LTBP-2 and -3 do not affect proGdf11 levels in the media, their co-expression with PC5/6A in HEK293 or COS-1 cells reduced the levels of soluble PC5/6A in the media by 30-50%, as assessed by a C-terminal V5-Ab (Fig. 4A, upper panels). It is important to note that soluble PC5/6A in the media is only immunoreactive to the V5 and not to the prosegment (pPC5) Ab (see later in Fig. 7, bottom panel), suggesting an activated form that lost its prosegment (4). We also noted that no intracellular PC5/6A accumulation was observed in either cell line. Interestingly, removal of the HSPG-binding C-terminal CRD domain (PC5/6-CRD) (17) abrogated the reduction of the level of PC5/6A in the media by either LTBP-2 or -3 (Fig. 4A, middle panel). The effect seems to be specific to PC5/6A, since in HEK293 media the levels of either PACE4 (the closest member to PC5/6A) or a soluble form of PC7 (sPC7) (32) are not affected by the co-expression of LTBP-2 or -3 (Fig. 4A, lower panels). Since both LTBP- 2 (33) and -3 (34) were shown to assemble onto fibrillar structures in the ECM, we investigated whether the lower soluble levels of PC5/6A were due to its sequestration within the ECM. For this, PC5/6A was co-expressed with the unrelated protein 7B2 or LTBP-2 or -3 in COS-1 and HEK293 cells. After removing the cells in lysis buffer, plastic-bound proteins were extracted in boiling Laemmli buffer with a cell scraper. Western blotting revealed that upon co-expression with LTBP-2 or -3, ~1.5-3.6 fold more proPC5/6A was trapped within the ECM, based on the molecular mass (~110 kDa) and the immunoreactivity to both pPC5 and the C-terminal V5-tag Ab (Fig. 4B). Since the increased proPC5/6A retention in the ECM was not associated with an increased cell lysis in the presence of LTBP-2 or 3 (absence of - actin in the ECM or media, not shown), this suggests that LTBP-2 or -3 facilitates the exit of non-cleaved proPC5/6A zymogen from the ER, which normally does not exit this compartment (4,17). More LTBP-3 than LTBP-2 was associated with the ECM in COS-1 cells, but not in HEK293 cells (Fig. 4B), which may explain the lower levels of media LTBP- 3 (Fig. 3). However, the amount of sequestered proPC5/6A was similar with both LTBPs (Fig. 4B). Thus, it seems that the level of LTBPs is not limiting for the ability of proPC5/6A to exit the ER and to be sequestered in the ECM. We presume that in a similar fashion both LTBPs primarily allow exit of the zymogen proPC5/6A from the ER and the complex with LTBPs is sequestered very efficiently in the ECM. The retention of proPC5/6A in the ECM probably occurs via both the C-terminal CRD of PC5/6A (4) and LTBP-2 (35), or -3, all of which likely bind HSPGs and/or fibrillin-containing microfibrils (8). This is consistent with the fact that the media level of PC5/6-CRD, which does not bind HSPGs, is not affected by LTBP-2 or -3. Finally, cell surface biotinylation revealed that LTBP-2 and -3 reduce the levels of cell surface associated PC5/6A by ~20-30% (Fig. 4C), as it did in the media (Fig. 4A).

To confirm that LTBPs facilitate the exit of proPC5/6A from the ER and its sequestration in the ECM, we co-expressed LTBP-2 with the non-cleavable proPC5/6A-R116A mutant (Fig. S1), which results in a proPC5/6A form that does not exit the ER (17). Indeed, we can only detect proPC5/6A in cells and not in media (Fig. 4D). However, the cellular co-expression with LTBP-2 clearly enhanced the levels of proPC5/6A-R116A retained in the ECM as compared to those of the 7B2 control (Fig. 4D). The small amount of proPC5/6A in the ECM in absence of co-expressed LTBP-2 may be due to endogenous expression of these proteins in HEK293 cells. Altogether, these data suggest that the decreased levels of PC5/6A in the media and cell surface are likely due to a pool of proPC5/6A that escaped autocatalytic cleavage at the primary RTKR116↓ site (Fig. S1) in the ER, and that is subsequently secreted and trapped in the ECM and cannot be readily auto-activated into PC5/6A under basal conditions.

LTBP-2 and -3 inhibit the cleavage of proGdf11 by PC5/6A – It was previously shown that the sequestration of pro-myostatin in the ECM by LTBP-3 can inhibit its processing by Furin (13). Based on the fact that the both cellular and media levels of proGdf11 are not affected by LTBP-2 or -3 (Fig. 3), we concluded that proGdf11 is not appreciably sequestered by LTBPs in the ECM, whereas proPC5/6A was. To examine whether proGdf11 processing was affected by LTBP-2 or -3, the latter were co-expressed with PC5/6A in HEK293 or COS-1 cells that stably express Gdf11. Because mature Gdf11 (17 kDa) was almost undetectable with an anti- FLAG Ab (21) despite the presence of 3 FLAG epitopes at its C-terminus, we rather analyzed proGdf11 or its remaining prodomain after cleavage (53 and 37 kDa, see Fig. 1), using protein C Ab. In COS-1 cells, proGdf11 cleavage was modulated by the quantity of transfected PC5/6A cDNA. It was almost complete with 50 ng (96%), and reduced to 56% and 19% with 10 ng and 1 ng, respectively (Fig. 5, left panel). When LTBP-2 or -3 were co-expressed with 50 ng of PC5/6A, cleavage of proGdf11 was reduced from 96% to 40% and 18%, respectively. Similar data were obtained in HEK293 cells (Fig. 5, right panel). Herein, the 100% cleavage obtained with 10 ng of PC5/6A-expressing vector was reduced to 64% and 51% upon LTBP- 2 or -3 co-expressions, respectively. In both cell lines, LTBP-2 and, to a larger extent, LTBP- 3 inhibited the PC5/6A-mediated proGdf11 cleavage into mature Gdf11.

Fig. 4 LTBP-2 and -3 allow the exit of proPC5/6A from the ER and the complex is then secreted and retained in the ECM. (A) V5-tagged PC5/6A and HA-tagged LTBP-2 (L2) or LTBP -3 (L3) or the negative control protein 7B2, were co-expressed in COS-1 or HEK293 cells. PC5/6A was assessed by Western blotting using anti-V5 and anti-pPC5 Ab. in (A) the media and cell lysates, as well as (B) ECM fractions. LTBP-2 or -3 were also detected in the ECM fractions using HA Ab. As controls, V5-tagged PACE4 and a soluble form of PC7 (sPC7) were quantified by Western blotting in media and cell lysates from HEK293 using V5 and PC7 Ab, respectively. (C) HEK293 cells co-transfected with PC5/6A with 7B2 or LTBP-2 or -3 were biotinylated and pulled down (PD) with streptavidin-agarose. The pulled down proteins were then analyzed by Western blotting using V5 and HA Ab. As negative control for cell surface proteins, the cells co-expressing PC5/6A and 7B2 were treated with trypsin, which removes all cell surface proteins. (D) PC5/6A mutant R116A was either co-transfected with 7B2 or LTBP2 in HEK293 cells. The cell lysates, media and ECM fraction were analyzed by Western blotting using pPC5 Ab. Immunoblots were submitted to quantitative analysis using the ImageQuant software. The intensity was calculated relative to that of the control, which was fixed at 1. These data are representative of at least three independent experiments giving consistent and reproducible results.

Fig. 5 LTBP-2 and -3 partially and dose dependently inhibit the cleavage of proGdf11 by PC5/6A. COS-1 cells (left) and HEK293 cells (right) stably expressing proGdf11 were transfected with PC5/6A and/or, LTBP-2 (L2) or -3 (L3). Conditioned media were analyzed by Western blotting using anti-pC Ab that recognizes the proGdf11 and its prodomain. The total quantity of cDNA used for transfection was kept constant by using a vector expressing an unrelated protein, 7B2. The quantity (ng) of transfected PC5/6A cDNA was indicated for both cell lines. The protein level of PC5/6A and -actin were revealed in total cell lysates by Western blotting using V5 Ab and anti--actin Ab. The immunoblots were submitted to quantitative analysis by using ImageQuant. The percentage cleavage was calculated from the ratio of prodomain/(proGdf11+prodomain). These data are representative of at least three independent experiments.

PC5/6A inhibition by LTBP-2 and -3 requires their intracellular interaction – To determine whether the LTBPs inhibited the overall PC5/6A activity or specifically the proGdf11 cleavage, we measured PC5/6A activity on a fluorogenic substrate pERTKR-7-amido-4- methylcoumarin (pERTKR-AMC) (24) in the media of HEK293 cells that co-expressed PC5/6A with a control 7B2 or with LTBP-2 or -3. Although PC5/6A protein levels in the media of HEK293 cells were reduced by ~30% upon co-expression with LTBP-2 and -3, PC5/6A activity was reduced by ~80% and ~70%, respectively (Fig. 6A). Normalization of PC5/6A activity to its V5-immunoreactive protein quantity revealed that LTBP-2 and -3 achieved a ~75% and ~65% inhibition of activity, respectively (Fig. 6A). However, no inhibition of pERTKR-AMC cleavage was obtained when media containing PC5/6A were co- incubated with those containing LTBP-2 or -3 (Fig. 6A), suggesting that the inhibitory complex formation occurs intracellularly before secretion. In agreement, proGdf11 cleavage was not affected when media containing proGdf11, PC5/6A, LTBP-2 or -3 were co-incubated (Fig. 6B). We therefore conclude that PC5/6A inhibition requires a prior intracellular interaction with LTBP-2 or -3. Furthermore, the in vitro activity of soluble PC5/6-CRD construct lacking the C-terminal CRD, and hence cannot bind HSPGs, is also inhibited by co- expression with LTBP-2 or -3 (Fig. 6C). Interestingly, while the level of PC5/6-CRD in the media is not affected by the co-expression of LTBP-2 or -3 (Fig. 4), the latter inhibit the in vitro activity of PC5/6-CRD by ~55-60% (Fig. 6C). Notably, since their media levels are unchanged, PACE4 (Fig. 4A) and shed Furin (Fig. S3A) are presumably not retained in the ECM in the presence of LTBP-2 or -3. In contrast, their activities in the medium are reduced by ~70% for PACE4 (Fig. 6D) and 50-60% for shed Furin (Fig. S3B). As negative controls, we have also compared the effects of LTBP-2 or -3 on the secreted proprotein convertases PC1/3 or PC2 (Fig. S3). The media levels and activities of these two convertases were not affected by either LTBP-2 or -3. Therefore, inhibition of PC5/6A by LTBPs is a consequence of two additive effects. One is through reducing active PC5/6A media levels, and another one occurs via reduction of its enzymatic activity (Fig. 6A). However, we did not observe a reduction in media levels of PC5/6-CRD (Fig 4A and 6C). This suggested that the reduction of PC5/6A levels by LTBPs in the media is CRD-dependent. It is consistent with the fact that PC5/6A-CRD is less inhibited by LTBPs (Fig. 6C). We conclude that LTBP-2 and -3 can selectively reduce the media levels of full length PC5/6A, which binds HSPGs, but not those of PC5/6-CRD. Furthermore, they do not affect the media levels of Furin, PACE4 or sPC7. In addition, LTBP-2 and -3 can inhibit the activity of PC5/6A (as well as PACE4 and Furin), independent of its ability to bind HSPGs,

suggesting that they can also inhibit the catalytic subunit of these convertases independently from their ability to be sequestered in the ECM.

Fig. 6 LTBP-2 and -3 reduce secreted PC5/6A and PACE4 activities when co-expressed in the same cells. The conditioned media from HEK293 cells co-expressing PC5/6A (A), PC5/6-CRD (C) or PACE4 (D) with LTBP-2 or -3 were first analyzed by Western blotting to quantify the level of these PCs (left bars). In vitro enzymatic assay using a fluorogenic substrate pERTKR-7- amido-4-methylcoumarin was used to measure the activity of either the above media, or those expressing individually PCs, LTBP-2 or -3 that were then mixed and incubated for 2h. For equal amount of PCs, the relative activity was normalized to relative fluorescence units (RFU) in media containing only PCs fixed to 100% (shaded area). The error bars indicate standard error of three independent experiments. *, P <0.05; **, P < 0.005 (Student’s t test). (B) Conditioned media from HEK293 cells expressing proGdf11 were incubated with those expressing individually PC5/6A or LTBP-2 or -3. Cleavage of proGdf11 was revealed by Western blotting using anti-pC Ab. The intensity was quantified by ImageQuant. The % cleavage was calculated from the ratio of prodomain/(proGdf11+prodomain).

Fig. S3 LTBP-2 and -3 reduce secreted Furin activities but not PC1/3 and PC2 activities when co-expressed in the same cells. (A) The conditioned media from HEK293 cells co-expressing furin, PC1/3 or PC2 with LTBP-2 or -3 were first analyzed by Western blotting to quantify the level of these PCs using ImageQuant. The intensity was calculated relative to that of the control, which was fixed at 1.(B) In vitro enzymatic assay using a fluorogenic substrate pERTKR-7-amido-4- methylcoumarin was used to measure the activity of either the above media, or those expressing individually PCs, LTBP-2 or -3 that were then mixed and incubated for 2h. The relative activity was normalized to relative fluorescence units (RFU) in media containing only PCs, fixed to 100%. The error bars indicate standard error of three independent experiments. **, P < 0.005 (Student’s t test).

ProPC5/6A forms an intracellular complex with proGdf11 and LTBP-2 or -3 – To understand the mechanism of PC5/6A inhibition by LTBPs, we analyzed the interactions between the enzyme, substrate and inhibitors by co-transfection in HEK293 cells. Immunoprecipitation of PC5/6A with a V5 Ab and Western blots using various Ab demonstrated the interaction of proPC5/6A with proGdf11 or LTBP-2 and -3 (Fig. 7). Furthermore, in triply-transfected cells, proPC5/6A can co-immunoprecipitate with proGdf11 and either LTBP-2 or -3. This suggests that these three proteins form an intracellular complex. Note that the levels of V5-immunoprecipiated proPC5/6A in cells are lower in the presence of either proGdf11, LTBPs or both (Fig. 7, left panel). Since direct Western blots did not show this difference (Fig. 4A), we presume that complex formation partially interferes with the recognition of the V5-epitope in the C-terminus of PC5/6A. In the media however, only PC5/6A was found, lacking its prosegment, and neither Gdf11 nor LTBP-2 or -3 were precipitated with PC5/6A (Fig. 7). Our previous results showed that, in the cell, the majority of PC5/6A immunoreactivity is associated with its zymogen proPC5/6A (4,17). PC5/6A is then fully activated at the cell surface and released into the medium free of its prosegment (4). This was confirmed by the analysis of V5-immunoprecipitated proteins with the prosegment pPC5 Ab, which generated a similar pattern to that obtained with V5 Ab (Fig. 7; lower panels). These data suggest that proPC5/6A, and not PC5/6A, stably interacts with proGdf11 and LTBP-2 or -3 in the cells. Further evidence for the existence of a ternary complex was obtained upon subcellular co- localizations by immunofluorescence. PC5/6A was co-expressed with proGdf11 or LTBP-2 or -3 in COS-1 cells. Under permeabilizing conditions, PC5/6A primarily co-localized with Gdf11 and LTBP-2 or -3 (Fig. 8) in an ER-like compartment (20). Under non-permeabilizing conditions, PC5/6A co-localized with Gdf11 or LTBP-2 or -3 at the cell surface (Fig. 8). Note that we also showed that Gdf11 co-localized with LTBP-2 or -3 in the cell and at the cell surface (Fig. S2). Taken together, these data demonstrate that proPC5/6A, Gdf11 and LTBP-2 or -3 form intracellular and cell surface complexes. This may enhance the efficiency of proGdf11 processing by PC5/6A at the cell surface and its regulation by LTBP-2 and -3.

Fig. 7 ProPC5/6A forms an intracellular complex with proGdf11, LTBP-2 or -3. In HEK293 cells, PC5/6A was expressed either with 7B2 or co-expressed with proGdf11 and 7B2 and/or LTBP-2 (L2) or -3 (L3), as indicated. The quantity of transfected PC5/6A cDNA was kept constant (10 ng) and the total level of cDNAs transfected was also kept constant (50 ng) by including a corresponding quantity of 7B2 cDNA, except for the triple transfections. PC5/6A from cell lysates or media was immunoprecipitated using anti-V5 Ab and the immunoprecipitated proteins detected by Western blotting using anti-HA, anti-pC, anti-V5 and anti-pPC5.

Fig. 8 PC5/6A co-localizes with ProGdf11 and LTBP-2 or -3 in cells and on the cell surface. PC5/6A was co-expressed transiently with either proGdf11 or LTBP-2/-3 in COS-1 cells. The intracellular and cell surface immunofluorescence staining was performed in permeabilized and non-permeabilized COS-1 cells respectively, using anti-HA Ab against LTBP-2 and -3 (green), anti-pC Ab against Gdf11 (green) and anti-V5 Ab against PC5/6A (red). Arrows point to sites where co-localizations were evident. Bar = 10 μm. These data are representative of three independent experiments.

LTBPs regulate the activity of Gdf11 – It has been demonstrated that Gdf11 was able to activate the downstream Smad signaling pathway by inducing Smad2 phosphorylation in Xenopus (36). Additionally, a second cleavage of the prodomain of Gdf11 at S2 by BMP- 1/Tolloid has been shown to be important for Gdf11-dependent activation of Smad2 phosphorylation (37). We decided to examine the effect on Gdf11 activity of: (i) the cleavage at S1 by PC5/6A, and (ii) PC5/6A inhibition by LTBP-2 and -3. The biological activity of Gdf11 can be measured by the extent of Smad2 phosphorylation, which was assessed by immunoblotting using an Ab to phosphorylated Smad2 (pSmad2). HEK293 cells transfected with proGdf11 exhibited a ~2.5 fold higher pSmad2 signal than those transfected with LTBP-2, -3 or PC5/6A alone (Fig. 9), which indicated the basal level

of pSmad2. Although proGdf11 is likely activated by endogenous PC5/6A in HEK293 cells (3), the co-expression of PC5/6A with proGdf11 increased pSmad2 levels by 8.6-and 3.4- fold as compared to that observed with the expression of either PC5/6 or proGdf11 alone (Fig. 9). In the absence or presence of exogenous PC5/6A, LTBP-2 or -3 reduced pSmad2 levels by 1.3-2.9-fold (Fig. 9), demonstrating that PC5/6A inhibition by LTBP-2 and -3 led to reduced signaling of Gdf11.

Co-localization of PC5/6 and LTBP-3 – In order to probe the physiological significance of the down regulation of PC5/6 by LTBPs, we compared by in situ hybridization histochemistry their expression in mouse both at embryonic day 15.5 (E15.5) and at postpartum day 1 (P1) (Fig. 10). The data show an extensive co-localization of PC5/6 mRNA with those of LTBP-3 at E15.5 in various tissues and organs, including vertebrae, ribs, blood vessels, and vertebral bodies of the tail. At P1, clear co-localizations were seen in the stomach, kidney, lung, femur and humerus bones, ciliary bodies, and alveolar bones. Thus, LTBP-3 and PC5/6A co-localize at various developmental stages in bones and other tissues, befitting the deduced function of PC5/6 from its knockout phenotype in bone morphogenesis, as well as lung and tail formation (6,7).

Fig. 9 ProGdf11 processing regulates the phosphorylation of Smad2. ProGdf11, LTBP-2, -3 or PC5/6A were expressed alone or co-expressed in HEK293 cells as indicated. The cell lysates were detected by Western blotting using anti-phosphoSmad2 Ab and anti--actin Ab. The immunoblots were submitted to quantitative analysis by using ImageQuant.

Fig. 10 Comparative in situ hybridization shows co-localization of PC5/6 and LTBP3 mRNA inE15.5 mouse embryo and newborn mouse. E15.5 mouse embryo and newborn (P1) mouse were cut at lateral plan into 6 M sections. The sections were stained with cresyl violet (top), or hybridized with 35S-labelled mPC5/6 (middle) and mLTBP-3 (bottom) specific probes. PC5/6 and LTBP mRNA labelling were seen as bright under dark field illumination in the X-ray film autoradiogram. The white arrows indicate the major tissues of co-localisation. Br, brain; H, heart; In, intestine; Ki, kidney; R, ribs; SpC, spinal cord; T, tail; V, vertebrae; VB, vertebral bodies; AB, alveolar bone; Ad, adipose tissue, intercapsular; E, epiphysis; F, femur; CB, ciliary body; Hu, humerus; Li, liver; Lu, lung; St, stomach.

DISCUSSION

Previous work (6,7) showed that PC5/6 is the proGdf11 convertase through cleavage at S1, and our present work further demonstrates that neither PC5/6A nor any of the other PCs cleave at S2 (Fig. 1). The metalloprotease BMP1/Tolloid was reported to cleave proGdf11 within the prodomain at the S2 site Gly119↓Asp120, leading to the dissociation of the prodomain from mature Gdf11 (21). Interestingly, it was suggested that PC5/6 activates BMP1/Tolloid (38), via cleavage at the PC-like site RSRR125↓AA (39). We can thus conclude that proGdf11 activation requires two sequential cleavages: the first one at S1 by PC5/6, and the second one by activated BMP1/Tolloid at S2. Both occur extracellularly (Fig. 2 and (21)). The zymogens of the PCs do not exit the ER until a first autocatalytic cleavage takes place (40). An exception is proPC2 that can exit the ER as a complex with the inhibitory pro7B2 (27,29,41). Our data revealed a second exception whereby proPC5/6A (aa 35-915) can exit the ER by forming a complex with LTBPs, without undergoing the canonical autocatalytic cleavage at Arg116↓ (Fig. S1). Cleavage in the ER leads to an enzymatically inactive complex of the prosegment (aa 35-116) and mature PC5/6A (aa 117-915) (1,42). Subsequently, a second autocatalytic cleavage at Arg84↓ (Fig. S1) occurring at the cell surface or in the ECM releases the active form of PC5/6A into the medium (4). Upon co-expression of full length PC5/6A and LTBP-2 or -3, proPC5/6A was found in the ECM, and ~30-50% lower levels of mature PC5/6A were detected in the medium (Fig. 4A). In contrast, the levels of mature PC5/6-CRD in the medium were not affected by either LTBP-2 or -3 (Fig. 4A), suggesting that binding to HSPGs is important for ECM sequestration. The co-localization (Fig. 8) and co-immunoprecipitation of cellular proPC5/6A and LTBP-2 or -3 (Fig. 7), likely originating from the ER where they are mostly localized, suggested that these proteins form a complex very early along the secretory pathway. This complex may allow the exit of proPC5/6A from the ER and its accumulation in the ECM, but not at the cell surface (Figs. 4B,C). This was confirmed with the PC5/6A-R116A mutant, which remains as a zymogen that accumulated in the ECM in the presence of LTBP-2 (Fig. 4D). Whether this ECM-bound pool of proPC5/6A can be subsequently activated locally is not clear. LTBP-2 and -3 inhibit the processing of proGdf11 by PC5/6A (Fig. 5) by reducing the levels of PC5/6A in the media, but also by lowering the specific enzymatic activity of PC5/6A (Fig. 6A). Indeed, when equal quantities of secreted PC5/6A obtained from cells either expressing it alone or co-expressing it with LTBP-2 or -3, were assessed in vitro PC5/6A enzymatic activity was reduced by ~65% in the presence of co-expressed LTBPs (Fig. 6A). However, mixing media containing LTBP-2 or -3 and PC5/6A did not affect the

protease activity (Fig. 6A). Thus, inhibition requires both PC5/6A and LTBP-2 or -3 to be co- expressed in the same cells. This suggests that LTBP-2 or -3 can inhibit PC5/6A activity, likely via a prior intracellular complex formation. The complex may remain stable and be directed to the ECM as an inactive zymogen, or it could dissociate and allow PC5/6A activation, albeit less efficiently than in the absence of LTBPs. The fact that inhibition of PC5/6A activity can only be observed when LTBPs are co-expressed with the convertase, suggested that the intracellular interaction between LTBPs and PC5/6A can induce some conformational change in the protease early in the secretory pathway, which would lead to reduced secreted enzymatic activity. Conversely, the levels of PACE4 and shed Furin in the media of HEK293 cells were not affected upon their co-expression with either LTBP-2 or -3 (Figs. 4A and S3A). This suggested that different from PC5/6A, LTBPs do not sequester proPACE4 or proFurin in the ECM, which may be due to unique sequences in their prosegment that need to be identified (Fig. S1). However, the protease activity of PC5/6A (-70-80%), PACE4 (-70%) and Furin (- 50 to 60%) are reduced in the presence of LTBPs (Figs. 6, S3), suggesting that the overexpression of the latter can also inhibit the catalytic activity of these enzymes in the media, but not those of PC1/3 or PC2 (Fig. S3). Under limiting amounts of endogenous levels of LTBP-2 or -3, it is possible that the co-expressed PC5/6A (Fig. 10) would be mostly affected in both its secretion level and protease activity. In the presence of LTBP-2 or -3, the media levels of proGdf11 are not affected (Fig. 3, right panel), suggesting that LTBPs do not sequester proGdf11 in the ECM. In contrast, LTBP-2 or -3 were shown to retain pro-myostatin in the ECM (13), thereby reducing the levels of mature myostatin in the medium. The abundance of the convertase PACE4 in skeletal muscle (43,44), suggests that it could cleave pro-myostatin. In contrast, the levels of secreted mature TGF in the media of HEL cells were reported to be enhanced by the presence of LTBPs (45). The difference may be due to the fact that proTGF is best processed by membrane- bound Furin (46), while proGdf11 and pro-myostatin are likely to be best processed by PC5/6A (6) and possibly PACE4, both of which bind HSPGs at the cell surface and ECM via their C-terminal CRD domain (40). Gdf11 globally regulates antero-posterior axial patterning by controlling the spatial- temporal expression of Hox genes. The loss of Gdf11 in mice causes anterior homeotic transformation of the axial skeleton and absence of caudal segments (5). During embryonic development, the concentration of active Gdf11 is critical for its patterning function. Here we provided evidence for a mechanism by which PC5/6 and LTBP-2 or 3 tightly regulate Gdf11 activation, thereby modulating the gradient of active Gdf11. First, the production of active Gdf11 depends on the level of PC5/6A (Fig. 5). Secondly, the quantity of extracellular active

PC5/6A is negatively regulated by LTBP-2 and -3. These regulations of mature Gdf11 production were reflected in its downstream signaling, as reflected by Smad2 phosphorylation, which is reduced in the presence of LTBPs (Fig. 9). We also noted that the reduction in pSmad2 signal caused by LTBP-3 is more significant than that by LTBP-2. This is consistent with the fact that inhibition of proGdf11 processing is more drastic with LTBP-3 than that with LTBP-2 (Fig. 5). Our work provided the first evidence that PC5/6A processing of proGdf11 into Gdf11 is important for its downstream activity, and thus inhibition of this cleavage reduces Gdf11 activity Very recently, the first natural inhibitor of Furin was identified as plasminogen activator inhibitor 1 (PAI-1) (47). Herein, we identified LTBP-2 and -3 as natural inhibitors of PC5/6A. What is the physiological significance of the observed inhibition of PC5/6A activity by LTBPs and in what tissues will this be operating? First, we noted that in some tissues Pcsk5 KO mice (6,7) exhibit opposite phenotypes to those observed in Ltbp-3 KO mice (48). For example, while lung alveoli are collapsed in Pcsk5 KO mice, Ltbp-3 KO mice exhibit a much larger volume of alveoli. In addition, Pcsk5 KO mice exhibit ossification defects (6,7) while Ltbp-3 KO mice exhibit an osteopetrotic-like phenotype with increased bone deposition and excessive trabecular mass (48). Thus, it is possible that LTBP-3 downregulates PC5/6 activity, specifically in lung and bones. In support of this hypothesis, in situ hybridization of PC5/6 and LTBP-3 mRNAs during development at E15.5 and at P1 revealed a large degree of co-localization in the lung, various bones including humerus, femur and alveolar bones, as well as kidney and tail (Fig. 10). Interestingly, microarray data on hypophosphatemic mice that exhibit bone mineralization defects, suggested that in bone osteocytes PC5/6 activates the bone forming BMP1/Tolloid (39), and inactivates the bone desorbing fibroblast growth factor FGF23 by cleavage at the RHTR179↓SA site (38). Further studies will be needed to confirm that PC5/6 is involved in bone mass upregulation and that LTBP-3 may downregulate this activity.

FOOTNOTES

We are grateful to Dr. Gaoxiang Ge (University of Wisconsin) for the mouse Gdf11 cDNA and Malcolm Whitman (Harvard Medical School) for the cDNAs of human LTBP-2 and mouse LTBP-3. We especially thank Marie-Claude Asselin for cell culture, Edwidge Marcinkiewicz for in situ hybridization, and Brigitte Mary for editorial assistance. This research was supported by CIHR grants MOP-44363, a Strauss Foundation grant, and a Canada Chair # 216684.

The abbreviations used are: Ab, antibody; ER, endoplasmic reticulum; ECM, extracellular matrix; HSPG: heparan sulfate proteoglycan; aa: amino acid; SA, streptavidin; PC5/6, proprotein convertase 5/6; PCs, proprotein convertases; TGF transforming growth factor LTBP, latent TGF binding protein; Gdf11: growth differentiating factor 11; BMP, bone morphogenic protein; pSmad2: phospho-Smad2; CRD, cysteine rich domain; KO, knock-out; S1, site 1; S2, site 2; IP, immunoprecipitation; WB, western blot; HRP, horseradish peroxidase; PBS, phosphate-buffered saline.

3.3 Conclusions and Discussions

In manuscript 2, we reported the discovery of two natural PC5/6A inhibitors, LTBP-2 and -3. These two proteins do not interact with the active site of PC5/6, rather with its prosegment. The removal of the PC5/6 prosegment after two autocatalytic cleavage reactions (Sites 1 and 2) abolishes the interaction between LTBPs and mature PC5/6A in the media. Although the molecular mechanism of this inhibition has not been clearly demonstrated, the engagement of PC5/6 prosegment in the interaction with LTBPs probably affects the ability of the prosegment to mediate correct folding of the convertase. The PC prosegments are generally believed to act as intra-molecular chaperones which induce the correct folding of their catalytic domains (245, 246). Binding of LTBPs to the prosegment of PC5/6 probably modifies the conformation of the enzyme, thereby preventing zymogen processing and/or resulting in decreased enzymatic activity.

LTBP-2 and -3 not only inhibit PC5/6A by affecting its conformation, but also specifically reduce the level of secreted mature PC5/6A by 20 to 30% without affecting the secretion of the other PCs (PC1/3, PC2, furin, PACE4 and soluble PC7). Indeed, LTBPs regulate the secretory trafficking of PC5/6A by bypassing its first autocleavage event at RTIKR84↓ in the ER and targeting the zymogen directly into the ECM (Figure 10). In addition, the zymogen proPC5/6A can be sequestered in the ECM only with an intact CRD, as the secretion of PC5/6-∆CRD remains unchanged when co-expressed with LTBP-2 or -3. Under normal condition, the enzymatic activity of PC5/6 is detected in the TGN (247) and at the cell surface (14). The detour and sequestration of proPC5/6A reduces the bioavailability of active enzyme to its substrates, and thereby inhibits PC5/6A activity.

Unfortunately, the inhibitory activities of LTBP-2 and -3 are not specific to PC5/6A. The co-expression of LTBPs with different PCs led to reduced in vitro cleavage of fluorogenic peptides, with varied inhibitory efficiency PC5/6A = PACE4 (-70%) > furin (50%) >> PC1/3 = PC2 (0%). Their prosegment sequence alignments demonstrated

that the homology is higher among PC5/6, furin and PACE4 than that among PC5/6, PC1/3 and PC2 (Figure 11). This suggests that LTBPs also bind to furin and PACE4 but not to PC1/3 and PC2. And further studies are required to confirm such interactions. If LTBP-2 and -3 truly bind to the prosegment of furin and PACE4, why only proPC5/6A is sequestered in the ECM by LTBP-2 and -3? One possibility is that the binding of LTBP-2 and -3 to proPC5/6 as well as other PCs (furin, PACE4 or PC5/6-∆CRD) takes place in the ER, very early in the secretory pathway, which affects their folding. However, the interaction between LTBPs and PC5/6A is strong enough so that the two proteins remain associated all along the secretory pathway. The complex is eventually secreted and retained in the ECM. In contrast, the interaction between LTBP-2 / -3 with furin, PACE4 or PC5/6-∆CRD is probably not strong enough. The complex can be dissociated along the secretory pathway when transits through the more acidic compartments of the TGN (pH=6.2).

autocleavage prodomain LTBPs

- ER Cell surface media mature PC5/6A

Activation? ER ECM media + LTBPs

Figure 10: Schematic diagram depicting LTBP-2 and -3 reduce mature PC5/6A bioavailability. In the absence of LTBPs, proPC5/6A undergoes two autocatalytical cleavages, first in the ER and second at the cell surface. Mature PC5/6A is secreted into the media. In the presence of LTBPs, proPC5/6A bypasses the autocatalytic cleavage and can be retained in the ECM, which leads to decreased levels of mature active PC5/6A in the media.

B Sequence aligment of prodomain of human PC5/6, Sequence aligment of prodomain of human PC5/6, furin and PACE4 PC1/3 and PC2 similarity

Relative residue position Relative residue position

Figure 11: Sequence homology among the prosegments of human PC1/3, PC2, furin, PACE4 and PC5/6. (A) Sequence alignment of prosegment of human PC5/6, furin, PACE4, PC1 and PC2. Identical residues were displayed in red, similar residues were displayed in green and other residues are displayed in black. The conserved regions are highlighted in gray. (B) Plots of sequence conservation over proPC5/6, profurin and proPACE4 alignment and proPC5, proPC1/3 and proPC2 alignment.

To design a specific inhibitor for a protease, pharmaceutical chemists usually primarily target the active site of the enzyme by mimicking the structure of its natural peptide substrate. These small molecules are known as active-site directed inhibitors. However, the high sequence homology of the PC catalytic domains (50%-70%) and the remarkably similar shape of the active-site cleft among kexin, PC1/3, PC2, PC4, furin, PACE5, PC5/6 and PC7 (233), make the active-site directed strategy very difficult to apply to the development of PC inhibitors. In agreement, dec-RVKR-CMK developed using this strategy, shows very low selectivity. The structural similarity among PC active

100

sites also explains the high overlapping substrate specificity among PCs. The redundancy among the members of an enzyme family sometimes constitutes an advantage because the presence of alternative processing capabilities attenuates the deleterious consequences caused by the deletion/inhibition of one member of the family. This redundancy concept has been extensively demonstrated in a wide range of enzyme families, such as MMPs (234) and caspases (235, 236). Therefore, in many instances in order to block completely the processing of one substrate; several PCs need to be inhibited at the same time.

Despite these huge challenges, a lot of efforts continue to be invested in order to understand the specificity generated by substrates. A recent study shed some light on the PC specificity that resides in the peptide sequences surrounding the consensus cleavage motif. This comparative cleavage study analyzed the efficiency of PC2, PC4, furin, PACE4, PC5/6 and PC7 in cleaving over 100 decapeptides, which comprised P6-P4’ peptide sequence with R-X-(K/R/X)-R↓ consensus PC-cleavage motif in the center (237). The P2 valine enhances the activity of PC4, PC5/6 and PACE4 over that of furin, PC2 and PC7. A P4’ methionine increases cleavage by PC4, PC5/6, PACE4 and PC7, but not by furin and PC2. This kind of comparative studies with a larger number of peptides with longer sequence will allow us to better determine the cleavage preference of one PC. This will also provide some hints for the identification of in vivo substrate-PC couple as well as for PC inhibitor design. Nevertheless, the 3-dimensional structure of the substrate will ultimately define whether other exosite interactions can exist between the substrate and its cognate convertase, which enhances the specificity and selectivity of the proteolysis by a given PC at any given time and place along the intracellular secretory pathway.

To selectively inhibit one PC, short interfering RNA (siRNA, ~22 nucleotides in length) presents the advantage of targeting the mRNA issued from the transcription of theoretically any gene with high specificity (238). However, some important issues, such as toxicity, have to be addressed before the clinical use of siRNAs as anti-PC drugs. A systematic administration of such drugs will require caution, since furin, PC5/6 and SKI- 1 total KO mice are lethal or die at birth (40, 239, 240). Although PACE4 KO mice are viable (75% of newborns), they exhibit severe craniofacial malformations (241). And PC1/3 and PC2 KO mice manifest hormone and neuroendocrine deficiency (242, 243).

101

Although PC7 KO mice do not have apparent symptoms, they exhibit anxiolytic and novelty seeking behaviors (Besnard et al., Nature in press). All these striking phenotypes imply that a chronic PC siRNA treatment may result in severe toxicity. One strategy to reduce this toxicity and side effect of is to specifically inhibit either intracellular or extracellular PC activity for those PCs including furin, PC5/6 and PACE4, whose activities can be found in both compartments. This concept has been proven by a furin inhibitor. In order to target furin activity in viral infection, a peptide TPRARRRKKRT inhibits furin cell-surface processing without interfering with its normal intracellular function (244). Unfortunately, due to the mechanism of siRNAs, they cannot distinguish between the intra- or extracellular activity of the target PC. In addtion, the poor intracellular uptake and short half-life sometimes limit the clinical use of siRNAs (238).

Alternative strategies to develop highly specific PC inhibitors which do not rely on active-site directed small molecules and gene silencing are more attractive. The development of allosteric inhibitors and the use of biological agents (natural inhibitors or monoclonal antibodies) represent the most clinically promising approaches to target PCs. As PC5/6 is a secreted enzyme and it is mainly activated at the cell surface (14), its targeting will not be limited by the problems of intracellular delivery. Hence, the use of natural inhibitors and monoclonal antibodies to inhibit PC5/6 could be highly relevant.

In manuscript 2, our results proposed some strategies to selectively inhibit PC activity: i) targeting folding of PCs; ii) targeting the trafficking of PCs. However, our work is still preliminary with lots of points need to be clarified : i) Do LTBPs bind to furin and PACE4 prosegments?; ii) As LTBP-2 and -3 are large ECM molecules comprising multi-domains, which domain interacts with PC5/6A?; iii) Can this pool of ECM-sequestered proPC5/6 be activated? If it is the case, what is the signal and molecular mechanism?

102

Chapter IV: Manuscript 3

Title: Proprotein convertase subtilisin/kexin type 9 deficiency reduces melanoma metastasis in liver Authors: Xiaowei Sun, Rachid Essalmani, Robert Day, Majid Khatib, Nabil G Seidah and Annik Prat Journal: Neoplasia. 2012 Dec;14 (12):1122-31

4.1 Preface

PCSK9 is predominantly expressed in liver and its primary function is to enhance LDLR degradation, thereby increasing the levels of circulating LDLc. PCSK9 inhibition emerged as a promising therapy to treat hypercholesterolemia, which attracted a lot of pharmaceutical companies (16). A monoclonal antibody (mAb) from Regeneron (REGN727) that blocks PCSK9-LDLR interaction is currently in a phase III clinical trials (248, 249). The mAbs from Amgen (AMG145) and from Pfizer (RN316) are also in phase II clinical trials (16). In order to prevent side effects associated with PCSK9 inhibition, it is very important to better delineate physiological functions of PCSK9, particularly beyond its function in cholesterol regulation (See Chapter A.6.4). On the other hand, cholesterol is a strong cancer risk modifier. Low cholesterol levels have been associated with high risk of primary liver cancer in human (192, 193), whereas an inhibitor of cholesterol synthesis was able to reduce liver metastasis in mice (250). Thus, it is of general interest to evaluate the safety of PCSK9 inhibition in the context of liver cancer/metastasis development.

In this manuscript, we reported the first study to our knowledge that examined the in vivo role of PCSK9 in cancer development. We showed that the absence of PCSK9 reduces liver metastasis, through its ability to strongly reduce circulating cholesterol

103

levels. More importantly, PCSK9 has an anti-apoptotic role in hepatic cell lines, primary hepatocytes and in the livers of mice that received a splenic injection of B16F1 melanoma cells. In vivo, PCSK9 also maintains low levels of TNFα, which is probably the trigger of cell apoptosis.

In summary, our study evaluated the effect of PCSK9-deficiency in cancer and proved that the total loss of PCSK9 does not promote cancer progression, but rather decreases the propensity of liver metastasis of invasive melanoma cells in liver. PCSK9 inhibition will be beneficial to control liver metastasis progression. These results support the safety of PCSK9 inhibition and further propose the use of such inhibitors to reduce cancer metastasis.

104

Proprotein convertase subtilisin/kexin type 9 deficiency reduces melanoma metastasis in liver1,2

Xiaowei Sun*, Rachid Essalmani*, Robert Day†, Abdel M. Khatib‡, Nabil G. Seidah* and Annik Prat* *Laboratory of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, University of Montreal, Montreal, QC, Canada; †Institut de pharmacologie de Sherbrooke, Université de Sherbrooke, Sherbrooke, QC, Canada; ‡INSERM U1029, Université Bordeaux 1, Talence, France

Address all correspondence to: Annik Prat, PhD, Laboratory of Biochemical Neuroendocrinology, Clinical Research Institute of Montreal, 110 Pine Ave W, Montreal, QC, Canada H2W 1R7. E-mail: [email protected] Tel: 1-514-987-5738

Key words: PCSK9, metastasis, cholesterol, apoptosis, TNF

Running title: The loss of PCSK9 reduces liver metastasis

1This work was supported by CIHR grants MOP 102741 (to NGS and AP) and CTP 82946 (to RD, NGS and AP), a Strauss Foundation grant (to NGS), and a Canada Chair # 216684 (to NGS). 2This article refers to supplementary materials, which are designated by Table W1 and Figures W1 to W4 are available online at www.neoplasia.com.

Abbreviations: PCSK9: proprotein convertase subtilisin/kexin type 9; LDLR, low-density lipoprotein receptor; LDLc: low-density lipoprotein cholesterol; HCD: high cholesterol diet; TC: total cholesterol; WT: wild type; qPCR: quantitative PCR, LPDS: lipoprotein-deficient serum; TNFtumor necrosis factor ; Bcl-2: B-cell lymphoma-2;

NF-B: nuclear factor -light-chain-enhancer of activated B cell; TRAF2: TNF receptor-associated factor 2.

105

Abstract High circulating cholesterol is associated with hypercholesterolemia, atherosclerosis and stroke. However, the relation between cholesterol and tumorigenesis/metastasis is controversial. The proprotein convertase subtilisin/kexin type 9 (PCSK9) regulates low-density lipoprotein cholesterol homeostasis by targeting the low-density lipoprotein receptor (LDLR) for degradation. PCSK9 is mostly expressed in liver, which is one of the most common sites for metastatic disease. To reveal the function of PCSK9 and also evaluate the impact of cholesterol in liver metastasis development, B16F1 melanoma cells were injected into wild-type and Pcsk9-/- mice to induce liver metastasis. On chow diet, Pcsk9-/- mice harbored 2-fold less liver metastases than wild-type mice. This decrease is related to low cholesterol levels in Pcsk9-/- mice, as the protection was lost after normalizing Pcsk9-/- cholesterol levels by a 2 week-high cholesterol diet. Furthermore, a prolongation of this diet strongly increased metastasis in both genotypes, suggesting that high cholesterol levels promote metastatic progression. The protective effect of the PCSK9 deficiency is also associated with increased apoptosis in liver stroma and metastases. TNF mRNA and protein were respectively higher in liver stroma and plasma of injected mice, likely increasing the apoptotic TNF signaling. Furthermore, the anti-apoptotic factor B-cell lymphoma 2 was downregulated. TNF regulation is LDLR-independent, as its mRNA level was similarly upregulated in mice lacking both PCSK9 and LDLR. Our findings show that PCSK9 deficiency reduces liver metastasis by its ability to lower cholesterol levels, and by possibly enhancing TNF-mediated apoptosis.

106

Introduction Proprotein convertase subtilisin/kexin type 9 (PCSK9) belongs to a family of secretory serine proteases, named proprotein convertases (PCs) [1]. Gain-of-function mutations in the PCSK9 gene are associated with familial hypercholesterolemia [2]. In contrast, loss-of-function mutations are associated with reduced low-density lipoprotein cholesterol (LDLc) and lower risk of coronary artery disease [3,4]. PCSK9 regulates cholesterol homeostasis by controlling the protein levels of the low-density lipoprotein receptor (LDLR), which uptakes LDL particles into cells. Binding of PCSK9 to cell surface LDLR promotes its internalization and degradation in endosomal/lysosomal compartments [5-7]. PCSK9 inhibition therefore emerged as a promising therapy to treat hypercholesterolemia [8]. Beyond its role in maintaining cholesterol homeostasis, PCSK9 is implicated in numerous biological processes. Microarray studies revealed that PCSK9 overexpression leads to the dysregulation of many pathways, including cell cycle, apoptosis, inflammatory and stress responses [9,10]. In vivo studies also suggest that PCSK9 is implicated in these processes [11,12]. Following partial hepatectomy, Pcsk9-/- mice exhibit a delay in hepatocyte proliferation and enhanced apoptosis, which can be rescued by feeding mice a high cholesterol diet (HCD) [11]. Moreover, Pcsk9-/- mice exhibit hypoinsulinemia, hyperglycemia, and glucose intolerance [12]. Because biological processes such as cell cycle and proliferation are modified in cancer, we suspected that PCSK9 could regulate tumorigenesis/metastasis. Furthermore, as PCSK9 inhibitors or silencers may soon be available on the market [8], it was of interest to evaluate the safety of PCSK9 deficiency in relation to cancer/metastasis. So far, only two human studies examined a possible link between PCSK9 polymorphisms known to lower circulating LDLc levels and the risk of cancer [13,14]. Although one of the studies established a statistically valid association between low circulating LDLc and increased risk of cancer [13], neither study could associate heterozygote PCSK9 loss-of-function

107

variants (R46L, -15% LDLc; Y142X or C679X; -29% LDLc) with a higher cancer incidence. Using our PCSK9 knockout mouse models [11,15], we directly addressed the function of PCSK9 in cancer development. In this study, upon injection of B16F1 mouse melanoma cells into spleen, which preferentially metastasize to the liver [16], Pcsk9-/- mice developed less hepatic metastatic melanomas. This protective effect of the PCSK9 deficiency was abrogated by increased total cholesterol (TC) levels following a HCD. We also showed that Pcsk9-/- mice exhibited increased liver stromal and tumoral apoptosis, likely due to the activation of the TNF pathway.

Materials and Methods

Cell culture, plasmid, transfection and Western blotting HEK293 cells and B16F1 cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Invitrogen) and HepG2 cells in Modified Eagle’s Medium (MEM; Invitrogen) supplemented with 10% fetal bovine serum (FBS; Invitrogen). All cells were maintained at 37°C under 5% CO2. The cDNAs of mouse or human wild type (WT) PCSK9 and its mutant form, PCSK9D374Y were cloned into the pIRES2-EGFP vector (Clontech), as described previously [1]. Non-target (NTsh) and PCSK9-targeted (PC9sh) shRNA constructs, generation of lentiviral particles and transductions of HepG2 cells to obtain stable pools were described previously [17]. Human HEK293 and HepG2 cells and mouse B16F1 cells were transfected using Effectene (Qiagen), Fugene HD (Roche) and Lipofectamine 2000 (Invitrogen) reagents, respectively. Stable B16F1 and HepG2 cells overexpressing either WT or PCSK9D374Y were obtained upon puromycin selection. HEK293 cells were transfected with the vector pIRES2-EGFP empty or encoding V5-tagged PCSK9 or PCSK9D374Y, washed at 24 hours post-transfection, then incubated

108

with serum-free medium for another 24 hours. Media were then collected or transferred onto B16F1 or HepG2 cells for 12 hours. Media and cell lysate proteins were separated by 8% SDS-polyacrylamide gel-electrophoresis, blotted onto PVDF membranes and incubated with a polyclonal human PCSK9 (1:2000) [7], human or mouse LDLR (1:1,000; R & D Systems), -actin (1:5000; Sigma), or horseradish peroxidase-conjugated V5 (1:10,000; Sigma) antibodies (Ab).

RNA isolation, quantitative PCR (qPCR) and qPCR array Liver samples were dissected out from the right medial lobe. For metastatic livers, the absence of tumor contamination in the adjacent normal tissue samples (stroma) was verified by qPCR analysis of transcripts for tyrosinase, an abundant enzyme implicated in melanin synthesis in B16-derived tumors (Figure W1). All cell and tissue samples were submitted to RNA extraction using Trizol reagent (Invitrogen). cDNA were then prepared as described previously [18]. All qPCRs were carried out in a final volume of 10 μL using the SYBER green supermaster (Quanta biosciences) following manufacturer’s protocol. Liver WT and Pcsk9-/- stromal cDNAs were also submitted to a mouse apoptosis qPCR array (SuperArray) according to the manufacturer’s protocol. Specific primers were used for the amplification of the following cDNAs: furin [18], PCSK9 [11], HMG-CoA reductase (forward 5’-GTACGGAGAAAGCACTGCTGAA; reverse 5’-

TGACTGCCAGAATCTGCATGTC), LDLR (forward 5’-GTATGAGGTTCCTGTCCATC; reverse

5’-CCTCTGTGGTCTTCTGGTAG), TNFforward 5’-CCAGAACTCCAGGCGGTGCC; reverse

5’-CTGATGAGAGGGAGGCCATTTGGGA), TNFR1 (forward 5’-CCACCCGCAACGTCCTGACA; reverse 5’-AGGCACGCCATCCACCACAG), Bcl-2 (forward 5’-TGAACCGGCATCTGCACACCTG; reverse 5’-AAACAGAGGTCGCATGCTGGGG) and TRAF2 (forward

5’-TGCCCGCAGAGAGGTGGAGAG; reverse 5’-CCTTCTCGCTGAGGCGGACC). Simultaneously, cDNAs for TBP were amplified for normalization (forward:

5’-CCTAGTGGAGGTGCCTTGGA; reverse 5’-GGTTGCCACCTGAAGTCACA).

109

Figure W1. Tyrosinase expression in liver tumor and stroma. The absence of contamination of the liver stroma samples by tumoral cells was verified by assessing tyrosinase expression by qPCR in these samples (n=12). Tyrosinase is implicated in the synthesis of the melanin pigment that is abundantly produced by tumoral cells. The primers used for the amplification of tyrosinase cDNAs were: forward

5’-CACAGGCACCTATGGCCAAATGAAC; reverse 5’-CCAGTATGGAACAGTGAAGTTCTCATC. Error bars indicate SEM.

Animals and experimental metastasis assay ApcMin/+ mice were obtained from the Jackson Laboratory [19]. Generation of Pcsk9-/- mice that lack the proximal promoter and first exon of the Pcsk9 gene was described previously [11], as well as that of mice lacking LDLR [20] or both LDLR and PCSK9 [15]. All mice were on the C57BL/6J background. The following procedures were approved by the animal care committee of the Institut de Recherches Cliniques de Montreal. Six to eight week-old female mice were anesthetized with 2% isoflurane

110

inhalation. B16F1 cells were suspended in PBS (105 cells/100 μL) and injected directly into spleen, and mice were sacrificed 12 days later. Mice were fed either a chow diet (2018 Teklad Global) or a HCD (TD.08464 Harlan Teklad) for 14 or 26 days prior to sacrifice day. Livers were then removed to be photographed. Tumor and liver areas were quantified by image software NorthernEclipse and tumor density was calculated as the ratio of tumor area and total liver area.

B16F1 cell adhesion assay Primary hepatocytes were prepared using 10-12 week-old mice and the two-step collagenase perfusion method, as described previously [21]. Hepatocytes were seeded in a 24 well-plate (105/well) coated with 0.5 mg/mL fibronectin (Sigma Aldrich). Only the wells in which a perfect cell monolayer was formed were used for adhesion assay. B16F1 cells (3.106) were labeled with 30 Ci of 3H-thymidine (PerkinElmer Life Sciences) for 24 hours. Cells were then trypsinized, washed with PBS and suspended in 10 mL of DMEM containing 10% FBS or 10% lipoprotein-deficient serum (LPDS; Biomedical Technologies). 3H-labeled B16F1 cells (104 cells in 100 L) were then added on hepatocyte monolayers, which were pre-incubated for 3 hours with FBS- or LPDS-containing media. After 1 hour adhesion, media were removed. Cells were washed and incubated with cold 10% trichloroacetic acid for 30 minutes, lysed in a 0.5 M NaOH/0.1% SDS solution and finally submitted to 3H radioactivity counting.

In vitro cell growth Cells (104/well) were distributed in 96-well plates and subjected to MTS [3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazo lium)] cell proliferation assay (Promega) at multiple time point. Optical absorption was read at 490 nm using a Vmax microplate reader (Molecular device).

111

Apoptosis assay B16F1 and HepG2 cells were grown either in the absence or presence of 1 M staurosporine (Sigma) for 12 hours. Cells were collected in reaction buffer (20 mM PIPES pH7.2, 100 mM NaCl, 1 mM EDTA, 0.1% CHAPS, 10% sucrose). Liver, hepatic tumors and adjacent stromal tissues were dissected and homogenized (Polytron; Brinkmann Instruments) in reaction buffer. Cell lysis was completed by two cycles of 10 minutes freezing and thawing. Caspase-3 activity was then measured at 37°C in 80 g of proteins using 30 g of the caspase-3 fluorogenic substrate, acetyl-Asp-Glu-Val-Asp-7-amido-4-methylcoumarin (Ac-DEVD-AMC; Calbiochem). The fluorescence generated by free AMC was detected with a SpectraMAX GEMINI EM microplate spectrofluorometer (Molecular Devices; excitation, 380 nm; emission, 460 nm), and caspase-3 activity was expressed as the released relative fluoresce units/minute. Apoptosis in liver paraffin sections was assessed using In Situ Cell Death Detection Kit, POD according to the manufacturer instructions (Roche).

Plasma measurement Plasma total cholesterol was measured using the Infinity kit (Thermo scientific). Specific ELISA kits were used to measure mouse plasma PCSK9 (CircuLex) and TNF (Cell Signaling), according to the manufacturer’s recommendations.

112

Results

PCSK9 expression is higher in intestinal tumors than in adjacent normal tissues To probe the possible involvement of PCSK9 in cancer development, we looked at its expression in various cancers reported in the Oncomine database (http://www.oncomine.org). PCSK9 expression was upregulated by 1.2- to 2.8-fold in all cancer types assessed (Figure W2), except in the steroid-dependent breast and prostate carcinomas, where PCSK9 was downregulated by 40-50%. We completed these data with the analysis of tumors in intestine, one of the major sites of PCSK9 expression [1]. At 4 months of age, the ApcMin/+ mice, which constitute a colorectal cancer model, spontaneously develop along their small intestine ~1 tumor/cm [18,19]. A representative hematoxylin and eosin-stained tumor section is shown in Figure 1. QPCR analysis of ApcMin/+ tumors and their adjacent normal tissues demonstrated that PCSK9 mRNA was 3.1-fold higher in tumors (Figure 1). This upregulation was likely due to the transcriptional activation of the cholesterogenic pathway by the sterol response element-binding protein 2, as LDLR and hydroxymethylglutaryl-CoA (HMG-CoA) reductase mRNA levels were also increased (2-fold). Thus, PCSK9 was upregulated in intestinal tumors as in most cancers.

113

Figure W2. Expression of PCSK9 in various cancers. Datasets were retrieved from ONCOMINE (a cancer microarray database and integrated data-mining platform) with a threshold of P < 0.0001. PCSK9 expression value in tumors was log2 transformed and normalized by that in the adjacent normal tissue. Cancers in which PCSK9 was upregulated or downregulated are listed in red or blue, respectively.

114

Figure 1. PCSK9 is upregulated in intestinal tumors. An Apcmin/+ mouse small intestine section stained with hematoxylin and eosin shows the structure of an intestinal tumor (T) and its adjacent normal tissue (N). Expression of PCSK9, LDLR and HMG-CoA reductase was assessed by qPCR in 18 couples of tumors and their adjacent normal tissue collected from 3 Apcmin/+ mice of 4 months of age. mRNA levels in tumors were normalized to that of their respective adjacent stromal tissue, which was set to 1. Error bars represent SEM. *, P<0.05; **, P<0.01 (Student's t test).

115

Figure 2. The loss of PCSK9 decreases hepatic metastasis in a cholesterol-dependent manner. (A) Mice were fed a chow diet or a high cholesterol diet (HCD) for 14 or 26 days before euthanasia, and were injected with B16F1 cells 12 days before euthanasia. Livers were dissected out and representative livers and liver sections are shown. Insets emphasize the accumulation of lipid droplets in the livers of mice fed a HCD for 26 days. (B) Tumor density (area of tumors/total area of liver) was evaluated in 12 to 18 mice per genotype and per condition. (C) Total cholesterol levels were measured in plasma collected on day of sacrifice. Error bars represent SEM. *, P<0.05, **, P<0.01, ***, P<0.001 (Student's t test).

116

Development of hepatic tumors is reduced in Pcsk9-/- mice The liver is the major site of PCSK9 expression with ~10-fold higher levels of mRNA than in colon and intestine [1]. To investigate the in vivo role of PCSK9 in hepatic cancer development, we used a metastasis model, in which B16F1 mouse melanoma cells are injected into the spleen of C57BL/6 mice. Metastases were formed on the surface of the liver and were easily identified after 12 days because of the black melanin pigment they synthesize (Figure 2A). When 200,000 cells were injected into WT and Pcsk9-/- mice fed a regular chow diet, Pcsk9-/- mice harbored 2-fold less hepatic melanoma foci than WT mice (Figure 2, A and B). To assess whether the protective effect of PCSK9 deficiency was due to its associated hypocholesterolemia (TC=77 mg/dL versus 112 mg/dL for WT mice; Figure 2C) [11,22], mice were fed a HCD (0.4% cholesterol) for 14 days before euthanasia. Although this diet did not affect TC levels in WT mice, it brought Pcsk9-/- mice TC to 102 mg/dL, similar to 114 mg/dL in WT mice (Figure 2C). In parallel to TC levels, tumor density rose in Pcsk9-/- mice from 13% on a chow diet to 20% on a HCD, but did not change in WT mice (27% versus 25%; Figure 2 A and B). When mice were fed a HCD for 26 days before euthanasia, TC levels were increased by 1.5- and 1.8-fold, and tumor density rose by ~2- and ~4-fold, in WT and Pcsk9-/- mice, respectively (Figure 2). In addition, mice fed a 26 day-HCD exhibited tumors with a very high degree of necrosis. The loss of statistical significance between the WT and Pcsk9-/- tumor densities when mice were fed a HCD indicates that the protective effect of PCSK9 deficiency mainly resides in its ability to lower circulating cholesterol levels. Furthermore, our data show that a high dietary cholesterol intake significantly enhances the development of liver metastases.

117

B16F1 cells do not express PCSK9 and their LDLR levels are not sensitive to exogenous PCSK9 To understand the underlying molecular mechanism(s) behind the protective role of PCSK9 deficiency on liver metastases development, we analyzed by qPCR the mRNA expression of PCSK9 and other members of the PC family in B16F1 cells (Figure 3A and W3). Importantly, although PCSK9 expression was abundant in the human hepatocellular carcinoma HepG2 cells, it was extremely low in B16F1 cells and undetectable in B16F1 cell-derived hepatic tumors collected 12 days post-injection (Figure 3A). This suggests that the higher metastasis observed in WT mice is due to liver PCSK9 that modulates the host microenvironment, rather than to PCSK9 from B16F1 cells. Thus, B16F1 tumor cells constitute a good model to study the exclusive impact of host PCSK9 on metastasis. To mimic the in vivo paracrine/endocrine influence of host PCSK9, we incubated B16F1 cells overnight with media from HEK293 cells expressing V5-tagged mouse PCSK9, human PCSK9 or its most potent gain-of-function mutant PCSK9D374Y [23,24], and analyzed PCSK9-enhanced LDLR degradation by Western blotting. Surprisingly, none of the three PCSK9 forms were able to degrade the LDLR in B16F1 cells. In contrast, the same media incubated with HepG2 cells can trigger LDLR degradation (Figure 3). Although exogenous PCSK9 did not affect LDLR levels, it likely bound the cell surface LDLR of B16F1 cells, as V5 immunoreactive species were detected in cell lysates, especially with PCSK9D374Y (Figure 3B), known to exhibit a >10-fold better binding to the LDLR [24]. Metastasis is a complex multi-step process including cell extravasation from the primary tumor, intravasation and adhesion in distant organs, and proliferation to form secondary tumors [25]. In our experimental model, the final metastatic volume was shown to depend on tumor cell implantation and focal growth of metastases [26]. We thus decided to focus on the function of PCSK9 in cancer cell adhesion and proliferation.

118

Figure 3. B16F1 cells do not express PCSK9 and their LDLR is insensitive to exogenous PCSK9. (A) PCSK9 expression was quantified by qPCR in HepG2 cells, B16F1 cells, and B16F1 cell-derived liver tumors (n=3 for each). As a control, expression of the ubiquitous furin was also quantified. Error bars represent SEM. (B) Media of HEK293 cells that express the empty pIRES2-EGFP vector, mouse (mWT) or human (hWT) WT PCSK9, or human gain-of-function PCSK9D374Y (hD347Y), all containing a C-terminal V5 tag, were collected and transferred onto B16F1 or HepG2 cells for 12 hours. HEK293 cell media and B16F1 and HepG2 cell lysates were analyzed by Western blotting using a V5 Ab to reveal PCSK9, and mouse or human LDLR and -actin Abs. Band intensities were quantified by ImageJ software and LDLR intensities were first normalized to that of -actin (LDLR/act). Ratios were then normalized to that of the empty vector that was set to 1. A representative experiment out of 3 is shown.

119

Figure W3. PC expression in B16F1 cells. Mouse furin, PC5/6, PACE4, PC7, PCSK9 and SKI-1 expression was quantified by qPCR in B16F1 cells (n=3) and B16F1 cell-derived liver tumors (n=3). The primers used to amplify these cDNAs were reported previously (Sun et al., 2009, Mol Cancer 8:73).

B16F1 cell adhesion to primary hepatocytes increases in the absence of PCSK9 The role of PCSK9 in cell adhesion was assessed by measuring the attachment of B16F1 cells to primary hepatocytes (Figure 4). B16F1 cells were radioactively labeled by 3H-thymidine incorporation and then added directly onto a monolayer of mouse primary hepatocytes. After 1 hour incubation in a cholesterol-rich (10% FBS) or cholesterol-deficient (10% lipoprotein-deficient serum, LPDS) medium, cells were washed, lysed and the radioactivity counted. In 10% FBS, 41% more B16F1 cells adhered onto Pcsk9-/- versus WT hepatocytes. This different capacity of adhesion is probably caused by a cell surface protein other than LDLR, as the same adhesion difference was observed between Ldlr-/- and Ldlr-/- Pcsk9-/- (dKO) hepatocytes. In contrast,

120

B16F1 adhesion was insensitive to the genotype of hepatocytes in LPDS medium (Figure 4). Because Pcsk9-/- mice developed less metastasis than WT mice, it was surprising to observe that Pcsk9-/- primary hepatocytes bound more B16F1 cells than WT hepatocytes in FBS media. This suggests that, in vivo, hypocholesterolemia blunts the adhesion potential of B16F1 cells to liver, and reduces metastasis via a different mechanism than a change in adhesion.

PCSK9 does not modulate the growth and apoptosis of B16F1 cells but affects those of liver cells We then examined the impact of PCSK9 on B16F1 cell proliferation. We first incubated B16F1 cells with or without 5 g/mL of purified human PCSK9 [27] and measured cell growth using an MTS assay. Similar growth curves were obtained (Figure 5A), indicating that host PCSK9 should not affect B16F1 cell proliferation in vivo. To further challenge B16F1 cell growth, cells stably expressing an empty vector (pIR), human PCSK9 or its gain-of-function mutant PCSK9D374Y were grown in 10% FBS or 10% LPDS medium (Figure 5B). Cell proliferation was drastically reduced in the lipoprotein-deficient medium, indicating that B16F1 cell growth strongly depends on cholesterol. However, growth curves were not affected by PCSK9 overexpression, suggesting that B16F1 cell proliferation is not affected in vivo by PCSK9. Finally, as a control, we examined the growth of HepG2 cells whose LDLR levels are sensitive to exogenous PCSK9 (Figure 3B). HepG2 cells that stably express human PCSK9, a non-targeting shRNA (NTsh) or a PCSK9 antisense shRNA (PC9sh) [17] were grown in 10% FBS media (Figure 5C). Although higher PCSK9 levels did not affect growth, as observed for B16F1 cells, knocking down PCSK9 significantly reduced HepG2 cell growth. The slower growth of HepG2 cells lacking PCSK9 may be due to increased cell death. Accordingly, cellular apoptosis was estimated by measuring caspase-3 enzymatic activity

121

at basal levels and in the presence of staurosporine, an apoptosis inducer. HepG2 cells lacking PCSK9 (PC9sh) exhibited 170% to 200% higher caspase-3 activity versus control HepG2 cells (NTsh; Figure 5D). Similarly, caspase-3 activity in Pcsk9-/- primary hepatocytes was 170% higher than in WT ones (Figure 5D). In contrast, B16F1 cells pre-incubated with or without 5 g/mL of purified PCSK9 for 24 hours exhibited similar caspase-3 activity, in the absence or presence ofstaurosporine (Figure 5D). Thus, in vitro, only the absence of PCSK9 had an effect on HepG2 cell proliferation, likely through increased apoptosis, whereas PCSK9 had no major effect on B16F1 cell proliferation or apoptosis. In vivo, basal caspase-3 activities in WT and Pcsk9-/- livers were similar. However, like in HepG2 cells, injected livers lacking PCSK9 exhibited 50% higher caspase-3 activities, both in B16F1 cell-derived liver tumors and their adjacent stromal tissues (Figure 5E). This was confirmed by TUNEL assay on corresponding liver sections. In tumoral regions, the labeling was highly concentrated in the center of the tumor, making any quantification of the number of cells in apoptosis difficult. However, Pcsk9-/- stromal regions exhibited a stronger labeling than WT ones, whereas non-injected livers sections only showed a background signal. In summary, these data suggest that the absence of PCSK9 causes higher apoptosis in hepatic stromal and tumoral cells. The increased apoptosis observed in tumors is probably indirect, as incubation of B16F1 cells with PCSK9 failed to change their caspase-3 activity.

122

Figure 4. The adhesion of B16F1 cells to the primary hepatocytes is regulated by PCSK9 and cholesterol. B16F1 cells were incubated with 3H-thymidine for 24 hours and then added onto WT, Pcsk9-/-, Ldlr-/- and Ldlr-/-Pcsk9-/- primary hepatocytes, which were pre-incubated either in 10% FBS or in 10% LPDS containing media for 3 hours. After 1 hour of incubation, unadhered B16F1 cells were washed out, the remaining cells were lysed and radioactivity in the total lysates was counted. The error bars indicate SEM of four independent experiments. *, P<0.05; (Student's t test).

123

Figure 5. PCSK9 affects cell proliferation and apoptosis of hepatic cells, but not those of B16F1 cells. (A) Proliferation of B16F1 cells incubated with or without 5g/mL of purified PCSK9, and grown in media containing 10% FBS, was measured using a MTS assay. (B) B16F1 cells stably expressing the empty pIRES2-EGFP vector (pIR), V5 tagged human PCSK9 (PCSK9) or its variant PCSK9D374Y (D347Y) were grown in media containing 10% FBS or 10% LPDS and their proliferation measured. PCSK9 expression in these cell lines was assessed by Western blot analysis of serum-free media using a V5 Ab (inset). (C) HepG2 cells stably expressing non-targeting shRNAs (NTsh), PCSK9-targeting shRNAs (PC9sh) or human PCSK9 were grown in media containing 10% FBS and their proliferation was measured. PCSK9 expression and efficiency of PCSK9 knock-down in these cell lines was assessed by Western blot analysis of serum-free media using a V5 Ab (inset). (D) Caspase-3 activity was measured in HepG2 cells stably expressing NTsh and PC9sh, WT and Pcsk9-/- primary hepatocytes, and B16F1 cells pre-incubated with or without 5g/mL of purified PCSK9 for 24 hours,

124

either in the absence or presence of 1 M staurosporine. (E) Caspase-3 activity was assessed in non-injected liver extracts, and in liver stroma and tumor extracts obtained from mice 12 days post-B16F1 cell injection. Liver sections from both non-injected and B16F1 cell injected WT and Pcsk9-/- mice were submitted to a TUNEL assay. Arrows point at labeled apoptotic nuclei and dashed lines separate tumoral (T) and stromal (S) areas. The error bars indicate SEM of three independent experiments (A-D) or data for 6 mice/genotype in E. *, P<0.05, **, P<0.01, ***, P<0.001 (Student's t test).

TNF pathway is affected in Pcsk9-/- liver To identify the underlying molecular pathway that promotes apoptosis in the absence of PCSK9, we performed a qPCR array analysis of 84 key genes involved in apoptosis on WT and Pcsk9-/- liver stroma tissues that were collected 12 days after B16F1 cell injection. A set of genes was dysregulated in Pcsk9-/- mice (Table W1), and the expression of selected genes was validated by qPCR in non-injected livers, hepatic tumors and stroma samples (Figure 6). The most upregulated gene encodes tumor necrosis factor  (TNF, an inflammatory cytokine. QPCR analysis confirmed that TNF expression was 50% higher in Pcsk9-/- stromal samples than in WT samples (Figure 6). Pcsk9-/- versus WT non-injected livers also showed a trend for higher TNF expression, but TNF expression was similar in WT and Pcsk9-/- tumors.

To better assess the role of PCSK9 in TNF gene regulation, we took advantage of our transgenic mice, Tg(Apoe-PCSK9), in which the apoE promoter drives the expression of PCSK9 essentially in the liver and macrophages [11]. First, Tg(Apoe-PCSK9) livers expressed 38% less TNF than Pcsk9-/- livers (Figure W4). Second, naïve peritoneal macrophages express >200-fold higher levels of TNF than liver. When isolated from Pcsk9-/- mice, their TNF expression was 70% and 130% higher than in macrophages isolated from WT and Tg(Apoe-PCSK9), respectively (Figure W4). Thus, these data

125

confirm the existence of an inverse correlation between PCSK9 and TNF expression levels. The expression of the TNF receptor, TNFR1 (gene Tnfsfr1a), was similar in WT and Pcsk9-/- intact livers. However, injection of B16F1 cells significantly induced TNFR1 expression by 50% and 110% in WT and Pcsk9-/- liver stroma tissue, respectively, resulting in a 40% higher expression of TNFR1 in the Pcsk9-/- stroma (Figure 6). Like TNFTNFR1 is similarly expressed in liver tumors developed in WT or Pcsk9-/- mice. Thus, the 40% higher expression of TNFR1 in the Pcsk9-/- stroma should potentiate TNF signaling in this tissue. The binding of TNF to TNFR1 can simultaneously elicit apoptotic and nuclear factor kappa(NF-B) mediated cell survival signals [28]. NF-B negatively regulates apoptosis by inducing the expression of anti-apoptotic factors [29], including B-cell lymphoma-2 (Bcl-2) [30] and the TNF receptor-associated factor 2 (TRAF2), whose deficiency was shown to increase cell sensitivity to TNF-induced apoptosis [31,32] (Figure 6; lower right panel). Expression of both TRAF2 and Bcl-2 was 30% lower in Pcsk9-/- liver stroma than in WT ones (Figure 6). In addition, Bcl-2 was downregulated by 20% in Pcsk9-/- tumors as compared to WT ones. This suggests that the absence of PCSK9 reduces the NF-B survival response, and thus further increases cell susceptibility to apoptosis. We then verified whether TNF upregulation depended on the LDLR. Mice lacking the LDLR (Ldlr-/-) or both the LDLR and PCSK9 (Ldlr-/-Pcsk9-/-; dKO) [11] and that exhibit similar high plasma TC levels were analyzed before or after B16F1 cell injection. The higher TNFexpression associated with the lack of PCSK9 was conserved in the LDLR-deficient background. TNF mRNA levels were 90% higher in stroma (Figure 6; lower middle panel), showing that TNFupregulation does not depend on the LDLR. In summary, these data indicate that, in an LDLR-independent manner, PCSK9 deficiency leads to a stronger TNF pro-apoptotic signaling in liver stroma. In addition,

126

Pcsk9-/- stroma and tumor exhibited a reduced NF-B pro-survival response that likely reinforces apoptosis in these tissues.

Table W1. Comparative expression of 84 key genes associated with apoptosis in WT and Pcsk9-/- liver stroma.

127

Figure 6. TNF pathway is altered in Pcsk9-/- mice. Expression of TNF TNFR1, TRAF2 and Bcl-2 was analyzed by qPCR in non-injected livers (n=9/genotype), and in liver stroma and tumors (n=5-7/genotype) collected from mice 12 days post-injection. TNF expression was also assessed in non-injected livers (n=12/genotype), and in liver stroma and tumors (n=5/genotype) collected from Ldlr-/- and Pcsk9-/- Ldlr-/- (dKO) mice (hatched bars). Error bars indicate SEM. *, P<0.05; **, P<0.01 (Student's t test). In the lower right panel, a scheme of the TNFsignaling pathway is shown. Binding of TNF to TNFR1 triggers intracellular recruitment of TRADD (TNF-receptor associated death domain) to TNFR1, which provides an assembly platform for TRAF2 and FADD (Fas-associated protein with a death domain). The binding of TRAF2 leads to NF-B activation that induces the transcription of survival and anti-apoptotic proteins, including Bcl-2 and TRAF2. The binding of FADD leads to caspase activation and apoptosis. The events regulated by the absence of PCSK9 are indicated by open arrows.

128

Figure W4. TNF expression in livers and naïve macrophages. TNFand PCSK9 expression was analyzed by qPCR in non-injected livers and naïve macrophages isolated from WT, Pcsk9-/- and PCSK9 transgenic (Tg; Zaid et al., 2008, Hepatology 48:646) mice (n=6-10/genotype). Naïve macrophages were isolated from the mouse peritoneal cavity by lavaging with 10 mL of PBS. After centrifugation at low speed, the cell pellet was washed twice with 10mL PBS. Cells were then allowed to adhere to a plastic culture dish (35mm) for 3 hours in RPMI 1640 containing 10% fetal bovine serum. Non-adherent cells were washed off and adherent cells (essentially macrophages) were harvested for RNA extraction. Error bars indicate SEM; *, P < 0.05 (Student's t test).

129

Figure 7. Plasma TNF and PCSK9 are upregulated along metastasis formation. (A) TNF was measured before injection, 2 days and 12 days post-injection in the plasma of WT and Pcsk9-/- mice (n=12/genotype). (B) Circulating PCSK9 and plasma total cholesterol were measured before injection, 2 days and 12 days post-injection in WT mice (n=12). (C) PCSK9 and LDLR expression levels were measured by qPCR in non-injected livers and in B16F1 cell-injected livers (stroma) from WT mice (n=5). The data are represented by the means ± SEM. *, P<0.05; **, P<0.01; ***, P<0.001 (Student's t test).

Circulating TNF and PCSK9 are co-regulated in the course of metastasis progression. To confirm that the upregulation of TNF expression led to a higher TNF secretion, we measured the plasma levels of TNF by ELISA before, 2 days and 12 days post-B16F1 cell injection. At all times, TNF levels were ~2-fold higher in Pcsk9-/- plasma than in WT plasma (Figure 7A). In addition, in both genotypes, TNF levels increased with time, reaching 4-fold higher levels at 12 days post-injection. In the same WT plasma samples, circulating PCSK9 increased by 1.8-fold at 12 days post-injection,

130

whereas cholesterol levels were unaffected (Figure 7B). In agreement, we confirmed by qPCR that PCSK9 transcripts were 2.3-fold higher in stroma than in non-injected livers, whereas LDLR transcripts were unchanged (Figure 7C). This suggests that the increases in TNFandPCSK9 were independent of cholesterol [33].

Discussion In this study, we investigated the role of PCSK9 in liver metastasis formation, following B16F1 cells injection in the spleen of WT and Pcsk9-/- mice. Because B16F1 cells and their derived tumors do not express PCSK9 and exhibit LDLR levels that are insensitive to exogenous PCSK9, we exclusively examined the role of the host PCSK9 in metastasis development. First, we showed that PCSK9 enhances liver metastasis development via its role in maintaining high circulating cholesterol levels, as (i) a HCD that brought back the low cholesterol levels of Pcsk9-/- mice to WT levels resulted in increased metastasis in Pcsk9-/- mice, and (ii) an extended HCD enhanced liver metastasis development in both genotypes. Second, we showed that PCSK9 protects HepG2 cells, primary hepatocytes and liver stroma and metastases against apoptosis, at least in part by maintaining low levels of TNF expression in an LDLR/cholesterol-independent manner. Tumor development depends on the reciprocal interaction of the tumor with its microenvironment. Locally activated stromal cells modify the proliferative and invasive behavior of cancer cells [34]. Diminished liver metastases in Pcsk9-/- mice revealed that the lack of PCSK9 generates a less permissive environment for tumor expansion. First, tumor cell proliferation may be indirectly limited in Pcsk9-/- mice, as the proliferation of B16F1 cells in vitro strongly depends on cholesterol. Second, increased apoptosis in liver stroma and metastases may contribute to lower metastasis in Pcsk9-/- mice. However, the relative contributions of lower proliferation versus higher apoptosis to the Pcsk9-/- protective phenotype remain to be determined. PCSK9 was originally named “neural apoptosis-regulated candidate-1” because of its

131

upregulation upon induction of apoptosis by serum withdrawal [35]. Since then, the implication of PCSK9 in apoptosis has been frequently evoked. In hepatic cells and tissues, PCSK9 seems to be anti-apoptotic. A microarray study showed that overexpression of the gain-of-function PCSK9D347Y downregulates some pro-apoptotic genes in HepG2 cells [10]. Additionally, the lack of PCSK9 enhances apoptosis during liver regeneration [11]. In contrast, an in vitro study revealed that PCSK9 has a pro-apoptotic effect in primary cultures of cerebellar granular neurons through its ability to degrade the apolipoprotein receptor 2 ([36]. However, unlike neuronal primary cultures, liver expresses very low levels of this receptor. Thus, the present study constitutes the first in vivo evidence that PCSK9 has an anti-apoptotic effect in the mouse liver. What are the underlying mechanisms of the anti-apoptotic properties of PCSK9? Surprisingly, even though PCSK9 had no effect on B16F1 cell apoptosis, B16F1 cell-derived tumors exhibited higher caspase-3 activities, suggesting apoptosis was initiated by a stromal apoptotic factor. TNFis a good candidate. It was upregulated in liver stroma, but not in tumors. It was also 2-fold higher in Pcsk9-/- plasma. Furthermore, TNF is a potent anti-cancer agent used to treat advanced soft-tissue sarcomas and melanoma, due to its ability to induce tumor cell apoptosis [37,38]. Although TNF can also activate NF-B-mediated cell survival responses, NF-B signaling was rather reduced in Pcsk9-/- stroma, as lower mRNA levels of its downstream targets Bcl-2 and TRAF2 were measured. Finally, the modest, but significant, TNF upregulation observed was independent of (i) circulating cholesterol levels, which remained unchanged during metastasis development, or (ii) the LDLR, as the same TNF increase was observed in mice lacking both PCSK9 and the LDLR, and exhibiting hypercholesterolemia (TC ~300 mg/dL). One of the objectives of this work was to assess the safety of PCSK9 inhibition which may widely used to control hypercholesterolemia. Unlike the clear association between hypercholesterolemia and elevated risk of cardiovascular diseases [4], the relation

132

between cholesterol and cancer remains unclear. This is mainly due to the high heterogeneity of this disease. Cholesterol levels have been positively correlated with cancers of brain [39], colon [40] and prostate [41], suggesting that cholesterol-lowering drugs could be beneficial to manage these cancers progression. For instance, statins that inhibit cholesterol synthesis have been proven to be effective to limit the growth and progression of prostate cancer [42,43]. In contrast, cholesterol levels have been negatively correlated with cancers of breast [39], lung [40] and liver [44,45]. However, the latter correlation is confounded by a preclinical effect, in which plasma cholesterol declines prior to the diagnosis of cancer [46]. Furthermore, although a negative correlation between the risk of primary liver cancer and cholesterol has been shown in human [44,45], a mouse study demonstrated that lower plasma cholesterol levels, achieved by a cholesterol synthesis inhibitor, significantly reduced liver metastasis [47]. In agreement with the latter study, the present work that monitored liver metastasis progression, but not primary liver cancer, strongly suggests an adverse role of cholesterol. Thus, future PCSK9 inhibitors that lower LDLc [8] may also be beneficial to control hepatic metastasis progression. The beneficial effect of PCSK9 deficiency resides (i) in its associated hypocholesterolemia that may decrease B16F1 cell growth, and (ii) in increased cell apoptosis, likely due to chronically elevated levels of TNF. Even though this work was focused on melanoma, it suggests that a PCSK9 inhibitor, initially developed to treat hypercholesterolemia [8], may be useful in therapies directed against melanoma and possibly other types of cancer metastasis.

Acknowledgements We are grateful to Rex Parker from Bristol Myers Squibb for purified human PCSK9. We thank Anna Roubtsova and Marie-Claude Asselin for excellent technical assistance.

133

References

[1] Seidah NG, Benjannet S, Wickham L, Marcinkiewicz J, Jasmin SB, Stifani S, Basak A, Prat A, and Chretien M (2003). The secretory proprotein convertase neural apoptosis-regulated convertase 1 (NARC-1): liver regeneration and neuronal differentiation. Proc. Natl. Acad. Sci. U. S. A. 100, 928-933. [2] Abifadel M, Varret M, Rabes JP, Allard D, Ouguerram K, Devillers M, Cruaud C, Benjannet S, Wickham L, Erlich D et al. (2003). Mutations in PCSK9 cause autosomal dominant hypercholesterolemia. Nat. Genet. 34, 154-156. [3] Cohen J, Pertsemlidis A, Kotowski IK, Graham R, Garcia CK, and Hobbs HH (2005). Low LDL cholesterol in individuals of African descent resulting from frequent nonsense mutations in PCSK9. Nat. Genet. 37, 161-165. [4] Cohen JC, Boerwinkle E, Mosley TH, Jr., and Hobbs HH (2006). Sequence variations in PCSK9, low LDL, and protection against coronary heart disease. New England Journal of Medicine 354, 1264-1272. [5] Maxwell KN, Fisher EA, and Breslow JL (2005). Overexpression of PCSK9 accelerates the degradation of the LDLR in a post-endoplasmic reticulum compartment. Proc. Natl. Acad. Sci. U. S. A. 102, 2069-2074. [6] Zhang DW, Garuti R, Tang WJ, Cohen JC, and Hobbs HH (2008). Structural requirements for PCSK9-mediated degradation of the low-density lipoprotein receptor. Proc. Natl. Acad. Sci. U. S. A. 105, 13045-13050. [7] Nassoury N, Blasiole DA, Tebon OA, Benjannet S, Hamelin J, Poupon V, McPherson PS, Attie AD, Prat A, and Seidah NG (2007). The Cellular Trafficking of the Secretory Proprotein Convertase PCSK9 and Its Dependence on the LDLR. Traffic. 8, 718-732. [8] Seidah NG, and Prat A (2012). The biology and therapeutic targeting of the proprotein

134

convertases. Nat. Rev. Drug Discov. 11, 367-383. [9] Lan H, Pang L, Smith MM, Levitan D, Ding W, Liu L, Shan L, Shah VV, Laverty M, Arreaza G et al. (2010). Proprotein convertase subtilisin/kexin type 9 (PCSK9) affects gene expression pathways beyond cholesterol metabolism in liver cells. J Cell Physiol 224, 273-281. [10] Ranheim T, Mattingsdal M, Lindvall JM, Holla OL, Berge KE, Kulseth MA, and Leren TP (2008). Genome-wide expression analysis of cells expressing gain of function mutant D374Y-PCSK9. J Cell Physiol 217, 459-467. [11] Zaid A, Roubtsova A, Essalmani R, Marcinkiewicz J, Chamberland A, Hamelin J, Tremblay M, Jacques H, Jin W, Davignon J et al. (2008). Proprotein convertase subtilisin/kexin type 9 (PCSK9): Hepatocyte-specific low-density lipoprotein receptor degradation and critical role in mouse liver regeneration. Hepatology 48, 646-654. [12] Mbikay M, Sirois F, Mayne J, Wang GS, Chen A, Dewpura T, Prat A, Seidah NG, Chretien M, and Scott FW (2010). PCSK9-deficient mice exhibit impaired glucose tolerance and pancreatic islet abnormalities. FEBS Lett. 584, 701-706. [13] Benn M, Tybjaerg-Hansen A, Stender S, Frikke-Schmidt R, and Nordestgaard BG (2011). Low-density lipoprotein cholesterol and the risk of cancer: a mendelian randomization study. J. Natl. Cancer Inst. 103, 508-519. [14] Folsom AR, Peacock JM, and Boerwinkle E (2007). Sequence variation in proprotein convertase subtilisin/kexin type 9 serine protease gene, low LDL cholesterol, and cancer incidence. Cancer Epidemiol. Biomarkers Prev. 16, 2455-2458. [15] Roubtsova A, Munkonda MN, Awan Z, Marcinkiewicz J, Chamberland A, Lazure C, Cianflone K, Seidah NG, and Prat A (2011). Circulating Proprotein Convertase Subtilisin/Kexin 9 (PCSK9) Regulates VLDLR Protein and Triglyceride Accumulation in Visceral Adipose Tissue. Arterioscler. Thromb. Vasc. Biol. 31, 785-791. [16] Barbera-Guillem E, Alonso-Varona A, and Vidal-Vanaclocha F (1989). Selective

135

implantation and growth in rats and mice of experimental liver metastasis in acinar zone one. Cancer Res. 49, 4003-4010. [17] Poirier S, Mayer G, Poupon V, McPherson PS, Desjardins R, Ly K, Asselin MC, Day R, Duclos FJ, Witmer M et al. (2009). Dissection of the endogenous cellular pathways of PCSK9-induced LDLR degradation: Evidence for an intracellular route. J. Biol. Chem. 284, 28856-28864. [18] Sun X, Essalmani R, Seidah NG, and Prat A (2009). The proprotein convertase PC5/6 is protective against intestinal tumorigenesis: in vivo mouse model. Mol. Cancer 8, 73-81. [19] Su LK, Kinzler KW, Vogelstein B, Preisinger AC, Moser AR, Luongo C, Gould KA, and Dove WF (1992). Multiple intestinal neoplasia caused by a mutation in the murine homolog of the APC gene. Science 256, 668-670. [20] Ishibashi S, Brown MS, Goldstein JL, Gerard RD, Hammer RE, and Herz J (1993). Hypercholesterolemia in low density lipoprotein receptor knockout mice and its reversal by adenovirus-mediated gene delivery. J. Clin. Invest. 92, 883-893. [21]Essalmani R, Susan-Resiga D, Chamberland A, Abifadel M, Creemers JW, Boileau C, Seidah NG, and Prat A (2011). In vivo evidence that furin from hepatocytes inactivates PCSK9. J Biol. Chem. 286, 4257-4263. [22] Rashid S, Curtis DE, Garuti R, Anderson NN, Bashmakov Y, Ho YK, Hammer RE, Moon YA, and Horton JD (2005). Decreased plasma cholesterol and hypersensitivity to statins in mice lacking Pcsk9. Proc. Natl. Acad. Sci. U. S. A. 102, 5374-5379. [23] Timms KM, Wagner S, Samuels ME, Forbey K, Goldfine H, Jammulapati S, Skolnick MH, Hopkins PN, Hunt SC, and Shattuck DM (2004). A mutation in PCSK9 causing autosomal-dominant hypercholesterolemia in a Utah pedigree. Hum. Genet. 114, 349-353. [24] Cunningham D, Danley DE, Geoghegan KF, Griffor MC, Hawkins JL, Subashi TA, Varghese AH, Ammirati MJ, Culp JS, Hoth LR et al. (2007). Structural and

136

biophysical studies of PCSK9 and its mutants linked to familial hypercholesterolemia. Nat. Struct. Mol. Biol. 14, 413-419. [25] Khatib AM, Siegfried G, Chretien M, Metrakos P, and Seidah NG (2002). Proprotein convertases in tumor progression and malignancy : novel targets in cancer therapy. Am. J Pathol. 160, 1921-1935. [26] Barbera-Guillem E, Barcelo JR, Urcelay B, Alonso-Varona AI, and Vidal-Vanaclocha F (1988). Noncorrelation between implantation and growth of tumor cells for their final metastatic efficiency. Invasion Metastasis 8, 266-284. [27] Benjannet S, Saavedra YG, Hamelin J, Asselin MC, Essalmani R, Pasquato A, Lemaire P, Duke G, Miao B, Duclos F et al. (2010). Effects of the prosegment and pH on the activity of PCSK9: evidence for additional processing events. J. Biol. Chem. 285, 40965-40978. [28] Heyninck K, and Beyaert R (2001). Crosstalk between NF-kappaB-activating and apoptosis-inducing proteins of the TNF-receptor complex. Mol. Cell Biol. Res. Commun. 4, 259-265. [29] Barkett M, and Gilmore TD (1999). Control of apoptosis by Rel/NF-kappaB transcription factors. Oncogene 18, 6910-6924. [30] Tamatani M, Che YH, Matsuzaki H, Ogawa S, Okado H, Miyake S, Mizuno T, and Tohyama M (1999). Tumor necrosis factor induces Bcl-2 and Bcl-x expression through NFkappaB activation in primary hippocampal neurons. J. Biol. Chem. 274, 8531-8538. [31] Tada K, Okazaki T, Sakon S, Kobarai T, Kurosawa K, Yamaoka S, Hashimoto H, Mak TW, Yagita H, Okumura K et al. (2001). Critical roles of TRAF2 and TRAF5 in tumor necrosis factor-induced NF-kappa B activation and protection from cell death. J. Biol. Chem. 276, 36530-36534. [32] Wang CY, Mayo MW, Korneluk RG, Goeddel DV, and Baldwin AS, Jr. (1998). NF-kappaB antiapoptosis: induction of TRAF1 and TRAF2 and c-IAP1 and c-IAP2

137

to suppress caspase-8 activation. Science 281, 1680-1683. [33] Dubuc G, Chamberland A, Wassef H, Davignon J, Seidah NG, Bernier L, and Prat A (2004). Statins upregulate PCSK9, the gene encoding the proprotein convertase neural apoptosis-regulated convertase-1 implicated in familial hypercholesterolemia. Arterioscler. Thromb. Vasc. Biol. 24, 1454-1459. [34] Tlsty TD, and Coussens LM (2006). Tumor stroma and regulation of cancer development. Annu. Rev. Pathol. 1, 119-150. [35] Chiang LW, Grenier JM, Ettwiller L, Jenkins LP, Ficenec D, Martin J, Jin F, DiStefano PS, and Wood A (2001). An orchestrated gene expression component of neuronal programmed cell death revealed by cDNA array analysis. Proc. Natl. Acad. Sci. U. S. A 98, 2814-2819. [36] Kysenius K, Muggalla P, Matlik K, Arumae U, and Huttunen HJ (2012). PCSK9 regulates neuronal apoptosis by adjusting ApoER2 levels and signaling. Cell Mol. Life Sci. [37] Eggermont AM, de Wilt JH, and ten Hagen TL (2003). Current uses of isolated limb perfusion in the clinic and a model system for new strategies. Lancet Oncol. 4, 429-437. [38] Alegre JF, Duarte M, Sureda GM, Bretcha BP, Dussan C, Ballester A, Crespo A, and Brugarolas MA (2012). [Tumor necrosis factor alpha and melfalan-based hyperthermic isolated limb perfusion in locally advanced extremity soft tissue sarcomas and melanomas]. Cir. Esp. 90, 114-120. [39] Kritchevsky SB, and Kritchevsky D (1992). Serum cholesterol and cancer risk: an epidemiologic perspective. Annu. Rev. Nutr. 12, 391-416. [40] Kitahara CM, Berrington de GA, Freedman ND, Huxley R, Mok Y, Jee SH, and Samet JM (2011). Total cholesterol and cancer risk in a large prospective study in Korea. J. Clin. Oncol. 29, 1592-1598. [41] Platz EA, Clinton SK, and Giovannucci E (2008). Association between plasma

138

cholesterol and prostate cancer in the PSA era. Int. J. Cancer 123, 1693-1698. [42] Gutt R, Tonlaar N, Kunnavakkam R, Karrison T, Weichselbaum RR, and Liauw SL (2010). Statin use and risk of prostate cancer recurrence in men treated with radiation therapy. J. Clin. Oncol. 28, 2653-2659. [43] Murtola TJ, Syvala H, Pennanen P, Blauer M, Solakivi T, Ylikomi T, and Tammela TL (2011). Comparative effects of high and low-dose simvastatin on prostate epithelial cells: the role of LDL. Eur. J. Pharmacol. 673, 96-100. [44] Iso H, Ikeda A, Inoue M, Sato S, and Tsugane S (2009). Serum cholesterol levels in relation to the incidence of cancer: the JPHC study cohorts. Int. J. Cancer 125, 2679-2686. [45] Borena W, Strohmaier S, Lukanova A, Bjorge T, Lindkvist B, Hallmans G, Edlinger M, Stocks T, Nagel G, Manjer J et al. (2012). Metabolic risk factors and primary liver cancer in a prospective study of 578,700 adults. Int. J. Cancer 131, 193-200. [46] Kritchevsky SB, Wilcosky TC, Morris DL, Truong KN, and Tyroler HA (1991). Changes in plasma lipid and lipoprotein cholesterol and weight prior to the diagnosis of cancer. Cancer Res. 51, 3198-3203. [47] Broitman SA, Wilkinson J, Cerda S, and Branch SK (1996). Effects of monoterpenes and mevinolin on murine colon tumor CT-26 in vitro and its hepatic "metastases" in vivo. Adv. Exp. Med. Biol. 401, 111-130.

139

4.3 Discussions and Conclusions

4.3.1 PCSK9 inhibition and risk of cancer

According to Canadian health measures survey, over 40% of Canadians aged 20 to 79 have an elevated level of total cholesterol. About 20% patients cannot reach optimal serum LDLc levels using the currently available medications. Statins, inhibitors of HMGCR which plays a central role in cholesterol biosynthesis, are the most effective and commonly used cholesterol-lowering drugs (251). However, statins paradoxically increase the expression of PCSK9 (75, 252), which degrades the LDLR and opposes the serum LDLc-lowering effect of statins. Thus, PCSK9 inhibition will enhance statin efficacy and can also be used as alternative therapy for statin-resistant patients. In addition, the total absence of circulating PCSK9 in two healthy women leading to a hypocholesterolemic phenotype, suggests that PCSK9 is a promising and relatively safe target to treat hypercholesterolemia (253). Cholesterol-lowering drugs represent a $40 billion global market. Therefore, PCSK9 draws enormous attention from large pharmaceutical companies since its discovery in 2003 by our group (24) and the confirmation of its role in LDLc regulation in humans (49). So far, several PCSK9 neutralizing agents are in clinical trials (16, 249) most advanced one is a PCSK9 monoclonal antibody (REGN727) which blocks the PCSK9≡LDLR interaction (248, 254). Its phase-II trial outcome demonstrated 30% to 65% serum LDLc reduction from baseline either in combination with statin therapy or not (249). Another promising approach under preclinical development to inhibit the extracellular interaction of PCSK9 with EGF-A domain of the LDLR is the use of adnectins (BMS-962476) which are variants of the 10th type III domain of human fibronectin comprising a PCSK9 binding loop that contains a small molecular structure that mimicks the EGF-A domain of the LDLR (SX-PCSK9) (16, 80). A 13-mer locked nucleic acid (LNA) compound designed to reduce intracellularly PCSK9 mRNA levels (ALN-PCS02) was in Phase I clinical trial (255, 256), but it seems its further use has been stopped for unknown reasons. PCSK9 antisense RNAi approaches by developed by Alnylam are however ongoing (16). The huge clinical and socioeconomic impact of PCSK9 makes the fundamental studies of this

140

protein particularly important. These studies will reveal the side-effects associated with the long-term administration of the PCSK9 inhibitors/silencers before they enter the drug market.

Liver, the major organ of cholesterol metabolism, is where PCSK9 is predominantly expressed (24) and from where all circulating PCSK9 originates (86). Liver is also one of the most common sites for cancer metastasis, accounting for 25% of all metastases to solid organs (257). Secondary liver tumors are far more common than primary hepatomas. Liver metastasis thus represents a good model to investigate the function of PCSK9 in cancer/metastasis development. Our study reported in manuscript 3 clearly demonstrated that the loss of PCSK9 significantly reduces liver metastasis formation in a B16F1 melanoma cell splenic injection mouse model. These results have high clinical relevance because they suggest that the use of PCSK9 inhibitor/silencer will not only be safe in the context of liver metastasis, but will also beneficial to control melanoma liver metastasis.

4.3.2 PCSK9 regulates cell adhesion

We further undertook some mechanistic studies to understand how PCSK9 regulates liver metastasis progression. Our data suggest that the protective effect of PCSK9-deficiency against liver metastasis largely resides in: i) hypocholesterolemia of Pcsk9-/- mice, which limits growth of B16F1 cells; and ii) high TNFα levels which result in enhanced tumor apoptosis. On the other hand, both cholesterol and PCSK9-deficiency increase B16F1 cell adhesion, thus in vivo hypocholesterolemia counterbalances PCSK9- deficiency effects, which makes cell adhesion not a contributing factor to low liver metastasis in Pcsk9-/- mice. In the next sections, I will discuss in more detail the PCSK9 functions related to cell adhesion and apoptosis, which seem to be independent of PCSK9 role in regulating the LDLR.

In manuscript 3, we demonstrated that Pcsk9-/- hepatocytes have greater ability to bind B16F1 cells, which is independent of the presence of the LDLR on hepatocytes, but dependent on the presence of cholesterol. We thus hypothesize that PCSK9 targets a protein regulating cell-cell interaction and its cell surface expression is regulated by

141

cholesterol. Our preliminary data strongly suggest that LDLR-related protein 1 (LRP1) is this protein.

Figure 12: LRP1 is a novel target of PCSK9. (A) The LDL receptor family (reprinted from Willnow et al,. 1999 (268)). Members of the LDLR family share common structural domains including: a single membrane anchor (transmembrane domain), complement-type repeats (which make up the ligand-binding domains) and epidermal growth factor (EGF) precursor homology domains (required for ligand binding and acid-dependent release of ligands in endosomes), NPxY designates the four-aa motif -Asn-Pro-X-Tyr- that mediates clustering of the receptors into coated pits. O-linked sugar domains are found in some, but not all, of the receptors. (B) LRP1 degradation is enhanced by PCSK9. B16F1 cells were stably transfected with empty vector (pIR-V5), human WT PCSK9 (hPCSK9) or GOF mutant PCSK9D374Y (hD347Y), both tagged with V5. The levels of the LDLR and the LRP1 were assessed in total cell lysates using antibodies recognized mouse LDLR and LRP1. The expression of PCSK9 and its mutant was also assessed in the media using a V5 antibody. The immunoblots were submitted to quantitative analysis using the ImageJ software. The intensity of the LDLR and LRP1 band were normalized by that of respective β-actin and then calculated relative to the value of the empty vector (pIR-V5), which was fixed at 1.

142

LRP1 (also known as α2 macroglobulin receptor or CD91) is a member of the LDLR family (258), sharing structural similarity with the LDLR (259) (Figure 12A). LRP1 is a ubiquitously expressed type-1 transmembrane receptor. The mature receptor protein is derived from a 600 kDa precursor processed by furin to generate an 85 kDa membrane-bound C-terminal fragment (intracellular and transmembrane domains) and a non covalently attached 515 kDa N-terminal fragment (extracellular domain) (259). LRP1 is a large endocytotic receptor that mediates endocytosis of more than 30 different ligands ranging from lipoproteins, extracellular matrix proteins, protease/protease inhibitor complexes and growth factors to viral surface glycoproteins (260, 261). This wide range of ligands suggests that LRP1 plays diverse biological roles in lipid metabolism, homeostasis of proteinases, cellular entry of viruses and toxins, cellular signal transduction and neurotransmission. The critical biological roles of the LRP1 was also confirmed by gene knockout experiments showing that Lrp1-/- mice die early during embryonic development (262).

Among the multifunctions of LRP1, it has been shown to coordinate cell adhesion

and macrophage migration by binding directly to intergrin αMβ2 (263-265). Our preliminary data showed that LRP1 degradation is enhanced by PCSK9 in HepG2 (Canuel et al, submitted) and B16F1 cells (Figure 12B). B16F1 cells stably transfected with human WT PCSK9 and its GOF mutant PCSK9D374Y exhibit significantly 40-50% lower protein levels of LRP1 when compared to those cells transfected with an empty vector pIR-V5, whereas the LDLR was not sensitive to PCSK9 (Figure 12B and see manuscript 3, Figure 3). Furthermore, LRP1 expression is highly regulated by cholesterol. It is induced by hypercholesterolemia and reduced upon statin cholesterol- lowering intervention (266, 267). Thus, these three arguments i) regulating cell-cell interaction, ii) targeted by PCSK9 and iii) the expression controlled by cholesterol make LRP1 a potential underlying factor that determine the higher adhesive properties of Pcsk9-/- hepatocytes than Pcsk9+/+ ones. We therefore propose the following scenarios: in a cholesterol-proficient environment, PCSK9 induces the degradation of LRP1, which results in decreased B16F1 cell adhesion to hepatocytes. In a cholesterol-deficient environment, since the expression of LRP1 is already repressed, PCSK9 mediated LRP1

143

degradation is less pronounced, so that we did not observe significant decrease in B16F1 cell adhesion (See manuscript 3, Figure 4). This also explains why higher adhesion properties of Pcsk9-/- hepatocytes were not reflected by lower metastases in Pcsk9-/- livers. Although PCSK9-deficiency increases the LRP1 levels, the severe hypocholesterolemia in Pcsk9-/- mice probably blunts this increase and leads to minor or no effect on cell adhesion.

4.3.3 PCSK9 regulates cell apoptosis in liver

In manuscript 3, we also demonstrated that the loss of PCSK9 directly increases apoptosis in hepatic cells. Indeed, this is not the first time that the PCSK9 has been related to apoptosis. PCSK9 was discovered and initially called neural apoptosis- regulated candidate-1 (NARC-1), because of its upregulation during serum deprivation induced apoptosis in primary cerebellar neurons (87, 88). It is clear now that this upregulation is due to the decrease of cholesterol level in the cerebellar media that activates the transcription of PCSK9 via SREBP pathway (74, 76), but not due to apoptosis itself. Since then the function of PCSK9 in apoptosis has been frequently evoked. However, the results are controversial and none of these studies were able to clearly demonstrate an in vivo function of PCSK9 in apoptosis.

PCSK9 has also been proposed to have anti-apoptotic effect in hepatocytes based on: i) microarray studies identified a set of apoptotic factors, such as death effectors domain containing 2 (DEDD2) and Brca1 associated ring domain 1 (BARD1) that were downregulated in HepG2 cells overexpressing PCSK9D374Y (82); ii) hepatocyte apoptosis is enhanced in Pcsk9-/- mice after partial hepatectomy (77). This increase in apoptosis can be compromised by a high cholesterol diet (HCD), suggesting that the anti-apoptotic role of PCSK9 during liver regeneration is cholesterol-dependent. In contrast, in neuronal cells in vitro studies suggest a pro-apoptotic role of PCSK9, for instance in primary cerebellar neurons (87, 89). Another study revealed that PCSK9 could potentiate apoptosis in neuronal cells through degrading the ApoER2, which promotes the anti- apoptotic pathway (89). The opposing apoptotic roles of PCSK9 in hepatic cells/liver and in neuronal cells may be due to the different expression of ApoER2, which is almost

144

exclusively found in neuronal cells. In order to resolve this tissue/cell-specific observation and to reveal the in vivo implication of PCSK9 in liver apoptosis, we undertook both in vitro and in vivo studies. Our results consistently demonstrated an enhancement of apoptosis in HepG2 cells, primary hepatocytes, hepatic stroma and metastases in the absence of PCSK9.

To further verify whether this anti-apoptotic role of PCSK9 in liver is mediated by the LDLR, we extended our studies to the mice lacking the LDLR. Caspase-3 activity measurement showed that the double knockout Ldlr-/-Pcsk9-/- (dKO) mice exhibited 1.8- and 1.5-fold higher apoptosis than Ldlr-/- mice in liver stroma and tumors, respectively (Figure 13). These increases are consistent with those in Pcsk9-/- versus WT mice, suggesting that the loss of PCSK9 increases apoptosis independent of the presence or absence of the LDLR. Thus, the anti-apoptotic function of PCSK9 is LDLR-independent.

3.0 * x1.5 2.5

2.0 x1.5

1.5 * * x1.8 1.0 x1.6

0.5 Caspase - 3 activity (rfu/min/ug) 0.0 WT Pcsk9-/-Ldlr-/- dKO WT Pcsk9-/- Ldlr-/- dKO WT Pcsk9-/- Ldlr-/- dKO liver stroma tumor

Figure 13: The loss of PCSK9 increases apoptosis in liver stroma and tumors in an LDLR-independent manner. Caspase-3 activity was assessed in non-injected liver, liver stroma and tumors obtained from mice 12 days post-B16F1 cell injection (6-9 mice/genotype) by measuring the fluorescent generated by the cleavage of a caspase-3 specific fluorogenic substrate, acetyl-Asp-Glu-Val-Asp-7-amido-4-methylcoumarin. The error bars indicate SEM. *, P<0.05 (Student's t test).

145

The apoptosis qPCR array analysis allowed us to identify TNFα as a potential underlying apoptosis mediator which is upregulated by the loss of PCSK9 (manuscript 3, Figure 5). Although at basal level (non-injected livers), the loss of PCSK9 only tends to upregulate TNFα expression when comparing Pcsk9-/- mice to WT ones or Ldlr-/- ones to Ldlr-/-Pcsk9-/- mice, the difference did not reach statistical significance. In the liver stroma of mice that received B16F1 cells injection 12 days before, the expression of TNFα increased significantly 1.6-fold in Pcsk9-/- versus WT mice. This increase was also observed in an LDLR-deficient background, since 1.9-fold higher TNFα was expressed by Ldlr-/-Pcsk9-/- liver stroma versus Ldlr-/- ones. Thus, the upregulation of TNFα by the loss of PCSK9 is also independent of the LDLR.

TNFα was initially discovered as a cytokine that causes tumor regression by triggering cancer cell death (269). The initial hope that TNFα could be useful as a potent anti-tumor therapy waned due to its high toxicity. Nevertherless, high doses of TNFα infused directly into tumors may yield the desired therapeutic outcome. TNFα was approved in Europe to treat locally advanced soft-tissues sarcomas and melanomas by isolated limb perfusion (270, 271). And it is also clear now that TNFα contributes to tumorigenic properties of cancer cells by regulating proliferation, invasion and metastasis. TNFα acts as a growth factor on a wide variety of cancer cells. It activates the nuclear factor-κB (NF-κB), which induces the expression of a large number of genes involved in cell growth, invasion and metastasis, such as cell-cycle genes cyclin D1 (272), growth factor IL-6 (273), matrix metalloprotease MMP9 (274) and angiogenic factor VEGFs (275). In addition, the activation of NF-κB antagonizes apoptosis by induce anti-apoptotic genes such as the caspase inhibitors, cellular FLICE-inhibitory protein (c-FLIP) (276) and B cell lymphoma-2 (Bcl-2) (277). Given the opposite cell fate (death and survival) that TNFα can trigger, it will be difficult to predict the consequences of TNFα activation in tumorigenesis and metastasis. Indeed, animal studies reported contradictory influences of TNFα on tumor growth. On the one hand, TNFα-deficient mice are resistant to development of benign and malignant skin tumors (278), and intraperitoneal administration of recombinant TNFα enhanced liver metastasis induced

146

by the intravenous and subcutaneous injections of lymphoma cells (279). On the other hand, intraperitoneal administration of TNFα resulted in more than 60% inhibition of liver metastasis induced by M109 lung carcinomas cells inoculation (280). Furthermore, a study using an intravital microscopy system that allows visualizing in vivo the interaction between host and EGFP-transfected colon adenocarcinoma cell CT-26 (281), demonstrated that TNFα plays a bidirectional role in metastasis development (282). At the early stages, TNFα facilitates CT-26 cell adhesion by increasing expression of adhesion molecules such as Intercellular Adhesion Molecule-1, Vascular Adhesion Molecule-1 and E-Selectin, whereas at later stages, TNFα triggers tumor apoptosis resulting in an overall lower liver metastasis. Interestingly, the loss of PCSK9 causes similar effects, higher B16F1 cell adhesion to host hepatocytes and enhanced tumor apoptosis, which lead to low liver metastasis outcome. Whether these phenotypes in Pcsk9-/- mice reside in the higher local TNFα levels (~1.5-fold higher in liver stroma) and circulating TNFα (~2-fold higher in plasma) need further in vivo investigations.

The primary role of TNFα is in immune responses. It is a pro-inflammatory cytokine that is mainly produced by macrophages in the acute phase of infection. TNFα initiates a cascade of cytokines and increases vascular permeability, thereby leading to the recruitment of macrophage and neutrophils to the site of infection. A high TNFα secretion is implicated in pathologies of several autoimmune diseases including inflammatory bowel disease, rheumatoid arthritis and Crohn’s disease (283), which explain in part the general toxicity of TNFα treatment. TNFα is also one of the key cytokines expressed in the acute phase of trauma. Similar to PCSK9, TNFα was shown to be upregulated after partial hepatectomy (77, 284). Although TNFα can cause potentially dangerous effects to regenerating hepatocytes, such as enhanced apoptosis and increased release of mitochondrial reactive oxygen species (285), TNFα also activates proliferative NF-κB signaling, which allows hepatocytes to overcome apoptotic and oxidative stress. In fact, the liver tolerates TNFα-initiated stress well. It was shown that a systemic administration of TNFα up to 25µg/200g in healthy rats potentiated hepatocytes to proliferate rather than die (286, 287). Nevertheless, TNFα can cause liver injury if the

147

liver has been pre-exposed to ethanol (288), infection (289) and obesity (290), which can interrupt the induction of survival factors. It also has been shown that ethanol-ingestion delays liver regeneration (288, 291). TNFα hepatic mRNA and plasma levels are both higher in ethanol-fed rats than in control ones before and 1 day after partial hepatectomy. However the induction of hepatic NF-κB signaling is blocked, which potentiates the apoptotic pathway of hepatocytes (292). Indeed, these ethanol-fed rats are quite similar with Pcsk9-/- mice. They both exhibit: i) liver regeneration delays after partial hepatectomy; ii) higher levels of TNFα mRNA in liver and higher protein levels in plasma before and after the trauma (hepatectomy or B16F1 injection); and iii) decreased NF-κB signaling response. We can speculate that like ethanol, the loss of PCSK9 enhances hepatocytes vulnerability to stress/trauma. In addition, both partial hepatectomy and B16F1 cell injection induce TNFα and PCSK9 expression, suggesting that these two proteins are co-regulated when cells adapt to stress/trauma. PCSK9 may protect the liver by modulating TNFα levels in order to prevent excessive cell death.

4.3.4 PCSK9-deficiency enhances liver metastasis independently to the LDLR

So far, our data has shown that PCSK9 regulates cell adhesion and apoptosis independent of the LDLR. So, is the impact of PCSK9 on liver metastasis progression independent of the LDLR as well? In order to dissect the function(s) of PCSK9 from its role in LDLR regulation in vivo, we needed to use mice in a genetic background lacking the LDLR. Our laboratory therefore generated LDLR-deficient mice either expressing PCSK9 (Ldlr-/-) or not (Ldlr-/-Pcsk9-/-). 200,000 B16F1 melanoma cells were injected into the spleen of Ldlr-/- and Ldlr-/-Pcsk9-/- mice fed a regular diet. Twelve-day post-injection, in contrast to Pcsk9-/- mice which formed 2-fold less liver metastases than WT mice, Ldlr- /-Pcsk9-/- mice harbored ~1.5-fold more metastases than WT and Ldlr-/-mice (Figure 14A, B). These data revealed that in the absence of the LDLR, the presence of PCSK9 rather than its absence limits liver metastasis. In the LDLR-deficient genetic background, Ldlr-/- and Ldlr-/-Pcsk9-/- mice exhibited similar plasma cholesterol levels which are ~2.5-fold higher than WT mice (77) (Figure 14C). Cholesterol is thus no longer a contributing factor to the different metastatic phenotype between Ldlr-/- and Ldlr-/-Pcsk9-/- mice, while cell adhesion and apoptosis become determinant. We can hypothesize that PCSK9-

148

deficiency mediated increase in B16F1 cell adhesion weigh probably more than the apoptosis enhancement, which results in an overall high metastasis outcome in Ldlr-/- Pcsk9-/- mice. Furthermore, this intriguing finding also suggests that PCSK9-dificiency associated hypocholesterolemia has a greater ability to reduce metastatic progression, masking the LDLR-independent pro-metastatic function of PCSK9-deficiency.

A B

* 50 * * 40 * 30

10 Liver metastases density (%) 0 WT Pcsk9-/- Ldlr-/- Pcsk9-/- Ldlr-/- WT Pcsk9-/- Ldlr-/- Pcsk9-/-Ldlr-/-

C Total cholesterol (mg/dL) Student t Test WT 112 ± 3 Pcsk9-/- 77 ± 2 P = 6x10-6 WT (4wk HCD) 174 ± 2 Pcsk9-/- (4wk HCD) 142 ± 7 P = 0.01 Ldlr-/- 325 ± 36 -/- -/- Ldlr Pcsk9 322 ± 54 P = 0.4

Figure 14: The loss of PCSK9 increases hepatic metastasis in a LDLR-independent manner. (A) Representative experiment in which livers were dissected from mice 12 days post-B16F1 cell injection. (B) Tumor density (area of tumors/total area of liver) was evaluated in 12 to 18 mice per genotype. The error bars represent SEM. *, P<0.05, Student t test. (C) Total plasma cholesterol of various mice strains.

149

4.3.5 Diet-induced hypercholesterolemia but not familial hypercholesterolemia increases liver metastasis

In manuscript 3, we concluded that hypercholesterolemia promotes liver metastasis progression. Surprisingly, Ldlr-/- mice developed as many metastases as WT ones, despite the hypercholesterolemic phenotype (total cholesterol: 2.5-fold higher than WT mice, 1.8-fold higher than WT mice fed 4-week HCD (Figure 14C). However, these results are in agreement with the literature which suggests that the LDLR has not been so far associated with any types of cancer. This also suggests that diet-induced hypercholesterolemia, but not familial hypercholesterolemia, promotes liver metastases development. In agreement, a U.S.A population based study showed that high cholesterol consumption is significantly associated with higher risk of cirrhosis or liver cancer, whereas serum cholesterol level is not (293). In fact, diet- and LDLR deficient-induced high cholesterol levels have different impact on body and cellular physiology. In WT mice, the LDLR accounts for 88% cholesterol clearance. When the LDLR is deleted, the clearance of LDLc will thus be reduced to ~10% of normal conditions, which results in a ~14-fold increase in plasma LDLc levels, from 7 to 101 mg/dl (294). Upon internalization of LDLc particles by the LDLR, SREBP signaling pathways will be inhibited, leading to the suppression of HMGCR expression in order to reduce cholesterol biosynthesis and the LDLR expression to decrease LDL particles uptake. In contrast, the activity of acyl-CoA cholesteryl acyltransferase will be enhanced in order to convert internalized cholesterol to cholesteryl esters (62, 64). Despite the high circulating cholesterol levels, due to the absence of the LDLR, the intracellular loading of cholesterol in Ldlr-/- mice is very low. This increases the rate of hepatic cholesterol synthesis by 1.7-fold and further exacerbates hypercholesterolemic phenotype (294). On the other hand, although diet-induced hypercholesterolemia can decrease LDLR expression to 25-50% in order to reduce LDLc uptake (295), a long-term HCD causes an accumulation of cholesterol in the cell due to the continuous internalization of cholesterol from the extracellular space (296). In a more severe scenario, the intracellular cholesterol over-loading leads to a toxic built up of

150

unesterified cholesterol (free cholesterol), which causes ER stress and even cell death (297). It has been shown that it is the accumulation of intracellular cholesterol, rather than the levels of extracellular cholesterol, which is central to the development of atherosclerosis (298). In early atherosclerotic lesions, an overloading of esterified cholesterol transforms the macrophage into a foam cell, which is the hallmark of atherosclerosis. In a later stage, an accumulation of free cholesterol will induce macrophage apoptosis, thereby contributing to plaque necrotic core formation (298, 299). Regarding cancer, our work also suggests that the intracellular cholesterol content is the determining factor that promotes liver metastases formation. Even though Ldlr-/- mice exhibit higher extracellular circulating cholesterol levels as compared to mice fed a HCD for 4-weeks, their intracellular cholesterol levels are very different: Ldlr-/- mice having low cholesterol content, while HCD-treated mice have high cholesterol content. Thus, the reason that HCD treated mice developed higher liver metastases versus WT or Ldlr-/- mice resides in their high intracellular cholesterol levels.

151

Chapter VI: Conclusions and Future perspectives

Twenty two years have elapsed since the discovery of the first PCs. During this period an enormous knowledge has been accumulated from the in vitro and in vivo characterization of the biological functions of the PCs (300). In various pathologies, PCs activities have been shown to be dysregulated and their inhibition has been proven to be beneficial (16). However, due to the redundancy shared among different PCs, the challenge of the design of safe and potent inhibitors of some PCs depends on a better understanding of their specific in vivo functions. My thesis work represents two different ways to better delineate the in vivo functions of PCs: i) From function to inhibition: Since PC5/6 has not been clearly associated with any diseases so far and there are no PC5/6 selective inhibitors, my work began with demonstration of a protective in vivo role of PC5/6 in intestinal tumorigenesis and continued towards the identification of two natural inhibitors LTBP-2 and -3. ii) From inhibition to function: PCSK9 is promising target to treat hypercholesterolemia and several inhibitors are currently in an advanced stage of clinical development. Thus, it is very important to evaluate the safety of PCSK9 inhibition. My work demonstrated that PCSK9-deficiency reduces melanomas liver metastasis, suggesting that PCSK9 inhibitors will not only have no adverse effect in cancer patients having liver metastasis, but also will reduce the progression of metastasis. It remains to be determined how general is this observation to other types of metastatic cancers. In summary, my work allows us to better understand the in vivo roles of PCs in cancer. Given the high therapeutic potential of PC- based drugs, I hope these findings will help to limit side effects of these therapies in cancer and associated metastasis. One of the biggest difficulties in studying specific PC functions is the redundancy shared by the constitutive PCs (furin, PACE4, PC5/6 and PC7). Thus, in order to elucidate the physiological roles of these PCs, KO mouse models were generated to establish direct links between the absence of a given PC and a phenotype. Because PC5

152

and furin KOs are lethal, tissue-specific KOs of furin and PC5 were also generated in order to bypass early lethality and to study altered physiological pathways. On the other hand, PCSK9 inhibition so far seems to be a safe strategy to treat hypercholesterolemia. However, PCSK9 KO mice exhibit some physiological defects including delayed liver regeneration (77), hyperglycemia, hypoinsulinemia and glucose intolerance (90), and could be subjected to higher hepatitis C infection (85), implying some adverse effects can be associated with PCSK9 inhibition. The use of PCSK9 inhibitors should be limited in patients having liver diseases, diabetes or hepatitis C infection. The phenotypes of PCSK9 KO mice need to be more thoroughly examined to ensure an appropriate application of PCSK9 inhibitors. Moreover, efforts still need to be invested into identification of in vivo PCSK9 targets and the fundamental molecular mechanisms by which PCSK9 enhances the degradation of its targets.

In perspectives, future PC studies will focus on the elucidation of the in vivo substrates by using total or tissue-specific knockout mice. The identification of the specific substrates/targets will lead to the revelation of novel physiological functions of these PCs as well as innovative therapeutic applications.

153

Chapter VII: Personal contributions to the subjects of the thesis

The results presented in Chapter II and III have been published and Chapter IV is in press. My direct contributions are as follows.

Manuscript 1: This work was published in Molecular Cancer volume 8: page 73-81, 2009. I performed all the experiments, data analysis and interpretations. I participated in the preparation of the manuscript with Annik Prat and Nabil G. Seidah, who also participated in the conceptualization and supervision of this work.

Manuscript 2 This work was published in Journal of Biological Chemistry Volume 286: page 29063- 29073, 2011. I performed all the experiments, except Figure 1 describing the cleavage of proGdf11 by PCs was done by Rashid Essalmani. I analyzed and interpreted all the data. I drafted the manuscript which was then corrected by Nabil G Seidah, and then finally reviewed by Annik Prat. Nabil G Seidah and Annik Prat also participated in the conceptualization and supervision of this work.

Manuscript 3 31This work is now published in Neoplasia. Volume 14: page 1122-1131, 2012. I performed all the experiments, data analysis and interpretations. I also drafted the manuscript, which was then corrected and edited by Annik Prat, and finally reviewed by Nabil G. Seidah. Annik Prat and Nabil G. Seidah also participated in the conceptualization and supervision of this work.

154

Reference List

(1) Puente XS, Sanchez LM, Overall CM, Lopez-Otin C. Human and mouse proteases: a comparative genomic approach. Nat Rev Genet 2003;4:544-58.

(2) Siezen RJ, Leunissen JA. Subtilases: the superfamily of subtilisin-like serine proteases. Protein Sci 1997;6:501-23.

(3) Mizuno K, Nakamura T, Ohshima T, Tanaka S, Matsuo H. Yeast KEX2 genes encodes an endopeptidase homologous to subtilisin-like serine proteases. Biochem Biophys Res Commun 1988;156:246-54.

(4) Roebroek AJ, Schalken JA, Bussemakers MJ, van HH, Onnekink C, Debruyne FM, et al. Characterization of human c-fes/fps reveals a new transcription unit (fur) in the immediately upstream region of the proto-oncogene. Mol Biol Rep 1986;11:117-25.

(5) Fuller RS, Sterne RE, Thorner J. Enzymes required for yeast prohormone processing. Annu Rev Physiol 1988;50:345-62.

(6) Seidah NG, Mayer G, Zaid A, Rousselet E, Nassoury N, Poirier S, et al. The activation and physiological functions of the proprotein convertases. Int J Biochem Cell Biol 2008;40:1111-25.

(7) Pasquato A, Pullikotil P, Asselin MC, Vacatello M, Paolillo L, Ghezzo F, et al. The proprotein convertase SKI-1/S1P. In vitro analysis of Lassa virus glycoprotein-derived substrates and ex vivo validation of irreversible peptide inhibitors. J Biol Chem 2006;281:23471-81.

(8) Benjannet S, Rhainds D, Essalmani R, Mayne J, Wickham L, Jin W, et al. NARC- 1/PCSK9 and its natural mutants: zymogen cleavage and effects on the low density lipoprotein (LDL) receptor and LDL cholesterol. J Biol Chem 2004;279:48865-75.

155

(9) Seidah NG, Chretien M. Proprotein and prohormone convertases: a family of subtilases generating diverse bioactive polypeptides. Brain Res 1999;848:45-62.

(10) Carter P, Wells JA. Dissecting the catalytic triad of a serine protease. Nature 1988;332:564-8.

(11) Gluschankof P, Fuller RS. A C-terminal domain conserved in precursor processing proteases is required for intramolecular N-terminal maturation of pro-Kex2 proteaseEMBO J 1994;13:2280-8.

(12) Zhou A, Martin S, Lipkind G, LaMendola J, Steiner DF. Regulatory roles of the P domain of the subtilisin-like prohormone convertases1. J Biol Chem 1998;273:11107-14.

(13) Nour N, Mayer G, Mort JS, Salvas A, Mbikay M, Morrison CJ, et al. The cysteine- rich domain of the secreted proprotein convertases PC5A and PACE4 functions as a cell surface anchor and interacts with tissue inhibitors of metalloproteinases. Mol Biol Cell 2005;16:5215-26.

(14) Mayer G, Hamelin J, Asselin MC, Pasquato A, Marcinkiewicz E, Tang M, et al. The regulated cell surface zymogen activation of the proprotein convertase PC5A directs the processing of its secretory substrates. J Biol Chem 2008;283:2373-84.

(15) Cunningham D, Danley DE, Geoghegan KF, Griffor MC, Hawkins JL, Subashi TA, et al. Structural and biophysical studies of PCSK9 and its mutants linked to familial hypercholesterolemia. Nat Struct Mol Biol 2007;14:413-9.

(16) Seidah NG, Prat A. The biology and therapeutic targeting of the proprotein convertases. Nat Rev Drug Discov 2012.

(17) Day R, Schafer MK, Watson SJ, Chretien M, Seidah NG. Distribution and regulation of the prohormone convertases PC1 and PC2 in the rat pituitary. Mol Endocrinol 1992;6:485-97.

156

(18) Marcinkiewicz M, Day R, Seidah NG, Chretien M. Ontogeny of the prohormone convertases PC1 and PC2 in the mouse hypophysis and their colocalization with corticotropin and alpha-melanotropin. Proc Natl Acad Sci U S A 1993;90:4922-6.

(19) Smeekens SP, Montag AG, Thomas G, biges-Rizo C, Carroll R, Benig M, et al. Proinsulin processing by the subtilisin-related proprotein convertases furin, PC2, and PC3. Proc Natl Acad Sci U S A 1992;89:8822-6.

(20) Seidah NG, Day R, Hamelin J, Gaspar A, Collard MW, Chretien M. Testicular expression of PC4 in the rat: molecular diversity of a novel germ cell-specific Kex2/subtilisin-like proprotein convertase. Mol Endocrinol 1992;6:1559-70.

(21) Trowbridge IS. Endocytosis and signals for internalization. Curr Opin Cell Biol 1991;3:634-41.

(22) Pullikotil P, Benjannet S, Mayne J, Seidah NG. The proprotein convertase SKI- 1/S1P: alternate translation and subcellular localization. J Biol Chem 2007;282:27402-13.

(23) Elagoz A, Benjannet S, Mammarbassi A, Wickham L, Seidah NG. Biosynthesis and cellular trafficking of the convertase SKI-1/S1P: ectodomain shedding requires SKI-1 activity. J Biol Chem 2002;277:11265-75.

(24) Seidah NG, Benjannet S, Wickham L, Marcinkiewicz J, Jasmin SB, Stifani S, et al. The secretory proprotein convertase neural apoptosis-regulated convertase 1 (NARC-1): liver regeneration and neuronal differentiation. Proc Natl Acad Sci U S A 2003;100:928- 33.

(25) Rashid S, Curtis DE, Garuti R, Anderson NN, Bashmakov Y, Ho YK, et al. Decreased plasma cholesterol and hypersensitivity to statins in mice lacking Pcsk9. Proc Natl Acad Sci U S A 2005;102:5374-9.

(26) Zhang DW, Lagace TA, Garuti R, Zhao Z, McDonald M, Horton JD, et al. Binding of proprotein convertase subtilisin/kexin type 9 to epidermal growth factor-like repeat A

157

of low density lipoprotein receptor decreases receptor recycling and increases degradation J Biol Chem 2007;282:18602-12. (27) Park SW, Moon YA, Horton JD. Post-transcriptional regulation of low density lipoprotein receptor protein by proprotein convertase subtilisin/kexin type 9a in mouse liver. J Biol Chem 2004;279:50630-8.

(28) Maxwell KN, Breslow JL. Adenoviral-mediated expression of Pcsk9 in mice results in a low-density lipoprotein receptor knockout phenotype. Proc Natl Acad Sci U S A 2004;101:7100-5.

(29) Mbikay M, Seidah NG, Chretien M. Neuroendocrine secretory protein 7B2: structure, expression and functions. Biochem J 2001;357:329-42.

(30) Benjannet S, Rondeau N, Paquet L, Boudreault A, Lazure C, Chretien M, et al. Comparative biosynthesis, covalent post-translational modifications and efficiency of prosegment cleavage of the prohormone convertases PC1 and PC2: glycosylation, sulphation and identification of the intracellular site of prosegment cleavage of PC1 and PC2. Biochem J 1993;294 ( Pt 3):735-43.

(31) Creemers JW, Vey M, Schafer W, Ayoubi TA, Roebroek AJ, Klenk HD, et al. Endoproteolytic cleavage of its propeptide is a prerequisite for efficient transport of furin out of the endoplasmic reticulum. J Biol Chem 1995;270:2695-702.

(32) Seidah NG, Hamelin J, Mamarbachi M, Dong W, Tardos H, Mbikay M, et al. cDNA structure, tissue distribution, and chromosomal localization of rat PC7, a novel mammalian proprotein convertase closest to yeast kexin-like proteinases. Proc Natl Acad Sci U S A 1996;93:3388-93.

(33) Lusson J, Vieau D, Hamelin J, Day R, Chretien M, Seidah NG. cDNA structure of the mouse and rat subtilisin/kexin-like PC5: a candidate proprotein convertase expressed in endocrine and nonendocrine cells. Proc Natl Acad Sci U S A 1993;90:6691-5.

158

(34) Nakagawa T, Murakami K, Nakayama K. Identification of an isoform with an extremely large Cys-rich region of PC6, a Kex2-like processing endoprotease. FEBS Lett 1993;327:165-71.

(35) Essalmani R, Hamelin J, Marcinkiewicz J, Chamberland A, Mbikay M, Chretien M, et al. Deletion of the gene encoding proprotein convertase 5/6 causes early embryonic lethality in the mouse. Mol Cell Biol 2006;26:354-61.

(36) De B, I, Marcinkiewicz M, Malide D, Lazure C, Nakayama K, Bendayan M, et al. The isoforms of proprotein convertase PC5 are sorted to different subcellular compartments. J Cell Biol 1996;135:1261-75.

(37) Jin W, Fuki IV, Seidah NG, Benjannet S, Glick JM, Rader DJ. Proprotein convertases [corrected] are responsible for proteolysis and inactivation of endothelial lipase. J Biol Chem 2005;280:36551-9.

(38) Xiang Y, Molloy SS, Thomas L, Thomas G. The PC6B cytoplasmic domain contains two acidic clusters that direct sorting to distinct trans-Golgi network/endosomal compartments. Mol Biol Cell 2000;11:1257-73.

(39) Nour N, Basak A, Chretien M, Seidah NG. Structure-function analysis of the prosegment of the proprotein convertase PC5A. J Biol Chem 2003;278:2886-95.

(40) Essalmani R, Zaid A, Marcinkiewicz J, Chamberland A, Pasquato A, Seidah NG, et al. In vivo functions of the proprotein convertase PC5/6 during mouse development: Gdf11 is a likely substrate. Proc Natl Acad Sci U S A 2008;105:5750-5.

(41) Rifkin DB. Latent transforming growth factor-beta (TGF-beta) binding proteins: orchestrators of TGF-beta availability. J Biol Chem 2005;280:7409-12.

(42) Dallas SL, Miyazono K, Skerry TM, Mundy GR, Bonewald LF. Dual role for the latent transforming growth factor-beta binding protein in storage of latent TGF-beta in the extracellular matrix and as a structural matrix protein. J Cell Biol 1995;131:539-49.

159

(43) Isogai Z, Ono RN, Ushiro S, Keene DR, Chen Y, Mazzieri R, et al. Latent transforming growth factor beta-binding protein 1 interacts with fibrillin and is a microfibril-associated protein. J Biol Chem 2003;278:2750-7.

(44) Parsi MK, Adams JR, Whitelock J, Gibson MA. LTBP-2 has multiple heparin/heparan sulfate binding sites. Matrix Biol 2010;29:393-401.

(45) Dubois CM, Blanchette F, Laprise MH, Leduc R, Grondin F, Seidah NG. Evidence that furin is an authentic transforming growth factor-beta1-converting enzyme. Am J Pathol 2001;158:305-16.

(46) Cui Y, Jean F, Thomas G, Christian JL. BMP-4 is proteolytically activated by furin and/or PC6 during vertebrate embryonic development. EMBO J 1998;17:4735-43.

(47) Brown MS, Goldstein JL. Familial hypercholesterolemia: A genetic defect in the low-density lipoprotein receptor. N Engl J Med 1976;294:1386-90.

(48) Vega GL, Grundy SM. In vivo evidence for reduced binding of low density lipoproteins to receptors as a cause of primary moderate hypercholesterolemia. J Clin Invest 1986;78:1410-4.

(49) Abifadel M, Varret M, Rabes JP, Allard D, Ouguerram K, Devillers M, et al. Mutations in PCSK9 cause autosomal dominant hypercholesterolemia. Nat Genet 2003;34:154-6.

(50) Davignon J, Dufour R. An Atlas of Investigation and Diagnosis. Oxford: Clinical Publishing; 2007.

(51) Allard D, Amsellem S, Abifadel M, Trillard M, Devillers M, Luc G, et al. Novel mutations of the PCSK9 gene cause variable phenotype of autosomal dominant hypercholesterolemia. Hum Mutat 2005;26:497.

(52) Leren TP. Mutations in the PCSK9 gene in Norwegian subjects with autosomal dominant hypercholesterolemia. Clin Genet 2004;65:419-22.

160

(53) Cohen J, Pertsemlidis A, Kotowski IK, Graham R, Garcia CK, Hobbs HH. Low LDL cholesterol in individuals of African descent resulting from frequent nonsense mutations in PCSK9. Nat Genet 2005;37:161-5.

(54) Zhao Z, Tuakli-Wosornu Y, Lagace TA, Kinch L, Grishin NV, Horton JD, et al. Molecular characterization of loss-of-function mutations in PCSK9 and identification of a compound heterozygote. Am J Hum Genet 2006;79:514-23.

(55) Hooper AJ, Marais AD, Tanyanyiwa DM, Burnett JR. The C679X mutation in PCSK9 is present and lowers blood cholesterol in a Southern African population 1. Atherosclerosis 2007;193:445-8.

(56) Brown MS, Goldstein JL. Biomedicine. Lowering LDL--not only how low, but how long? Science 2006;311:1721-3.

(57) Seidah NG. PCSK9 as a therapeutic target of dyslipidemia. Expert Opin Ther Targets 2009;13:19-28.

(58) Endo A. A gift from nature: the birth of the statins. Nat Med 2008;14:1050-2.

(59) Endo A, Tsujita Y, Kuroda M, Tanzawa K. Inhibition of cholesterol synthesis in vitro and in vivo by ML-236A and ML-236B, competitive inhibitors of 3-hydroxy-3- methylglutaryl-coenzyme A reductase. Eur J Biochem 1977;77:31-6.

(60) Goldstein JL, Brown MS. Regulation of the mevalonate pathway. Nature 1990;343:425-30.

(61) Dietschy JM. Regulation of cholesterol metabolism in man and in other species. Klin Wochenschr 1984;62:338-45.

(62) Brown MS, Goldstein JL. Receptor-mediated control of cholesterol metabolism. Science 1976;191:150-4.

(63) Buhman KF, Accad M, Farese RV. Mammalian acyl-CoA:cholesterol acyltransferases. Biochim Biophys Acta 2000;1529:142-54.

161

(64) Horton JD, Goldstein JL, Brown MS. SREBPs: activators of the complete program of cholesterol and fatty acid synthesis in the liver. J Clin Invest 2002;109:1125-31.

(65) Hua X, Yokoyama C, Wu J, Briggs MR, Brown MS, Goldstein JL, et al. SREBP-2, a second basic-helix-loop-helix-leucine zipper protein that stimulates transcription by binding to a sterol regulatory element. Proc Natl Acad Sci U S A 1993;90:11603-7.

(66) Yokoyama C, Wang X, Briggs MR, Admon A, Wu J, Hua X, et al. SREBP-1, a basic-helix-loop-helix-leucine zipper protein that controls transcription of the low density lipoprotein receptor gene. Cell 1993;75:187-97.

(67) Hua X, Nohturfft A, Goldstein JL, Brown MS. Sterol resistance in CHO cells traced to point mutation in SREBP cleavage-activating protein. Cell 1996;87:415-26.

(68) Matsuda M, Korn BS, Hammer RE, Moon YA, Komuro R, Horton JD, et al. SREBP cleavage-activating protein (SCAP) is required for increased lipid synthesis in liver induced by cholesterol deprivation and insulin elevation. Genes Dev 2001;15:1206-16.

(69) Korn BS, Shimomura I, Bashmakov Y, Hammer RE, Horton JD, Goldstein JL, et al. Blunted feedback suppression of SREBP processing by dietary cholesterol in transgenic mice expressing sterol-resistant SCAP(D443N). J Clin Invest 1998;102:2050-60.

(70) Wang X, Sato R, Brown MS, Hua X, Goldstein JL. SREBP-1, a membrane-bound transcription factor released by sterol-regulated proteolysis. Cell 1994;77:53-62.

(71) Sakai J, Duncan EA, Rawson RB, Hua X, Brown MS, Goldstein JL. Sterol-regulated release of SREBP-2 from cell membranes requires two sequential cleavages, one within a transmembrane segment. Cell 1996;85:1037-46.

(72) Rawson RB, Zelenski NG, Nijhawan D, Ye J, Sakai J, Hasan MT, et al. Complementation cloning of S2P, a gene encoding a putative metalloprotease required for intramembrane cleavage of SREBPs. Mol Cell 1997;1:47-57.

(73) Seidah NG, Mowla SJ, Hamelin J, Mamarbachi AM, Benjannet S, Toure BB, et al. Mammalian subtilisin/kexin isozyme SKI-1: A widely expressed proprotein convertase

162

with a unique cleavage specificity and cellular localization. Proc Natl Acad Sci U S A 1999;96:1321-6.

(74) Maxwell KN, Soccio RE, Duncan EM, Sehayek E, Breslow JL. Novel putative SREBP and LXR target genes identified by microarray analysis in liver of cholesterol-fed mice. J Lipid Res 2003;44:2109-19.

(75) Dubuc G, Chamberland A, Wassef H, Davignon J, Seidah NG, Bernier L, et al. Statins upregulate PCSK9, the gene encoding the proprotein convertase neural apoptosis- regulated convertase-1 implicated in familial hypercholesterolemia. Arterioscler Thromb Vasc Biol 2004;24:1454-9.

(76) Horton JD, Shah NA, Warrington JA, Anderson NN, Park SW, Brown MS, et al. Combined analysis of oligonucleotide microarray data from transgenic and knockout mice identifies direct SREBP target genes. Proc Natl Acad Sci U S A 2003;100:12027- 32.

(77) Zaid A, Roubtsova A, Essalmani R, Marcinkiewicz J, Chamberland A, Hamelin J, et al. Proprotein convertase subtilisin/kexin type 9 (PCSK9): hepatocyte-specific low- density lipoprotein receptor degradation and critical role in mouse liver regeneration. Hepatology 2008;48:646-54.

(78) Nassoury N, Blasiole DA, Tebon OA, Benjannet S, Hamelin J, Poupon V, et al. The cellular trafficking of the secretory proprotein convertase PCSK9 and its dependence on the LDLR. Traffic 2007;8:718-32.

(79) Poirier S, Mayer G, Poupon V, McPherson PS, Desjardins R, Ly K, et al. Dissection of the endogenous cellular pathways of PCSK9-induced low density lipoprotein receptor degradation: evidence for an intracellular route. J Biol Chem 2009;284:28856-64.

(80) McNutt MC, Kwon HJ, Chen C, Chen JR, Horton JD, Lagace TA. Antagonism of secreted PCSK9 increases low-density lipoprotein receptor expression in HEPG2 cells. J Biol Chem 2009;284:10551-70.

163

(81) Kwon HJ, Lagace TA, McNutt MC, Horton JD, Deisenhofer J. Molecular basis for LDL receptor recognition by PCSK9. Proc Natl Acad Sci U S A 2008;105:1820-5.

(82) Ranheim T, Mattingsdal M, Lindvall JM, Holla OL, Berge KE, Kulseth MA, et al. Genome-wide expression analysis of cells expressing gain of function mutant D374Y- PCSK9. J Cell Physiol 2008;217:459-67.

(83) Lan H, Pang L, Smith MM, Levitan D, Ding W, Liu L, et al. Proprotein convertase subtilisin/kexin type 9 (PCSK9) affects gene expression pathways beyond cholesterol metabolism in liver cells. J Cell Physiol 2010;224:273-81.

(84) Poirier S, Mayer G, Benjannet S, Bergeron E, Marcinkiewicz J, Nassoury N, et al. The proprotein convertase PCSK9 induces the degradation of low density lipoprotein receptor (LDLR) and its closest family members VLDLR and ApoER2. J Biol Chem 2008;283:2363-72.

(85) Labonte P, Begley S, Guevin C, Asselin MC, Nassoury N, Mayer G, et al. PCSK9 impedes hepatitis C virus infection in vitro and modulates liver CD81 expression. Hepatology 2009;50:17-24.

(86) Roubtsova A, Munkonda MN, Awan Z, Marcinkiewicz J, Chamberland A, Lazure C, et al. Circulating proprotein convertase subtilisin/kexin 9 (PCSK9) regulates VLDLR protein and triglyceride accumulation in visceral adipose tissue. Arterioscler Thromb Vasc Biol 2011;31:785-91.

(87) Bingham B, Shen R, Kotnis S, Lo CF, Ozenberger BA, Ghosh N, et al. Proapoptotic effects of NARC 1 (= PCSK9), the gene encoding a novel serine proteinase. Cytometry A 2006;69:1123-31.

(88) Chiang LW, Grenier JM, Ettwiller L, Jenkins LP, Ficenec D, Martin J, et al. An orchestrated gene expression component of neuronal programmed cell death revealed by cDNA array analysis. Proc Natl Acad Sci U S A 2001;98:2814-9.

(89) Kysenius K, Muggalla P, Matlik K, Arumae U, Huttunen HJ. PCSK9 regulates neuronal apoptosis by adjusting ApoER2 levels and signaling. Cell Mol Life Sci 2012.

164

(90) Mbikay M, Sirois F, Mayne J, Wang GS, Chen A, Dewpura T, et al. PCSK9- deficient mice exhibit impaired glucose tolerance and pancreatic islet abnormalities. FEBS Lett 2010;584:701-6.

(91) Rousselet E, Marcinkiewicz J, Kriz J, Zhou A, Hatten ME, Prat A, et al. PCSK9 reduces the protein levels of the LDL receptor in mouse brain during development and after ischemic stroke. J Lipid Res 2011;52:1383-91.

(92) Schulz WA. Molecular Biology of Human Cancers: An Advanced Student's Textbook. Springer; 2007.

(93) Valastyan S, Weinberg RA. Tumor metastasis: molecular insights and evolving paradigms. Cell 2011;147:275-92.

(94) Hanahan D, Weinberg RA. The hallmarks of cancer. Cell 2000;100:57-70.

(95) Cheng N, Chytil A, Shyr Y, Joly A, Moses HL. Transforming growth factor-beta signaling-deficient fibroblasts enhance hepatocyte growth factor signaling in mammary carcinoma cells to promote scattering and invasion. Mol Cancer Res 2008;6:1521-33.

(96) Su LK, Kinzler KW, Vogelstein B, Preisinger AC, Moser AR, Luongo C, et al. Multiple intestinal neoplasia caused by a mutation in the murine homolog of the APC gene. Science 1992;256:668-70.

(97) Moser AR, Pitot HC, Dove WF. A dominant mutation that predisposes to multiple intestinal neoplasia in the mouse. Science 1990;247:322-4.

(98) Korinek V, Barker N, Morin PJ, van WD, de WR, Kinzler KW, et al. Constitutive transcriptional activation by a beta-catenin-Tcf complex in APC-/- colon carcinoma. Science 1997;275:1784-7.

(99) Sherr CJ, McCormick F. The RB and p53 pathways in cancer. Cancer Cell 2002;2:103-12.

165

(100) Vazquez A, Bond EE, Levine AJ, Bond GL. The genetics of the p53 pathway, apoptosis and cancer therapy. Nat Rev Drug Discov 2008;7:979-87.

(101) Blasco MA. Telomeres and human disease: ageing, cancer and beyond. Nat Rev Genet 2005;6:611-22.

(102) Sherr CJ, DePinho RA. Cellular senescence: mitotic clock or culture shock? Cell 2000;102:407-10.

(103) Hanahan D, Folkman J. Patterns and emerging mechanisms of the angiogenic switch during tumorigenesis. Cell 1996;86:353-64.

(104) Carmeliet P. VEGF as a key mediator of angiogenesis in cancer. Oncology 2005;69 Suppl 3:4-10.

(105) Kazerounian S, Yee KO, Lawler J. Thrombospondins in cancer. Cell Mol Life Sci 2008;65:700-12.

(106) Sporn MB. The war on cancer. Lancet 1996;347:1377-81.

(107) Yang J, Weinberg RA. Epithelial-mesenchymal transition: at the crossroads of development and tumor metastasis. Dev Cell 2008;14:818-29.

(108) Oft M, Peli J, Rudaz C, Schwarz H, Beug H, Reichmann E. TGF-beta1 and Ha-Ras collaborate in modulating the phenotypic plasticity and invasiveness of epithelial tumor cells. Genes Dev 1996;10:2462-77.

(109) Chambers AF, Matrisian LM. Changing views of the role of matrix metalloproteinases in metastasis. J Natl Cancer Inst 1997;89:1260-70.

(110) Feldman WE. Effect of prior antibiotic therapy on concentrations of bacteria in CSF4. Am J Dis Child 1978;132:672-4.

(111) Motzer RJ, Bander NH, Nanus DM. Renal-cell carcinoma. N Engl J Med 1996;335:865-75.

166

(112) Konoshita T, Gasc JM, Villard E, Takeda R, Seidah NG, Corvol P, et al. Expression of PC2 and PC1/PC3 in human pheochromocytomas. Mol Cell Endocrinol 1994;99:307-14.

(113) Takumi I, Steiner DF, Sanno N, Teramoto A, Osamura RY. Localization of prohormone convertases 1/3 and 2 in the human pituitary gland and pituitary adenomas: analysis by immunohistochemistry, immunoelectron microscopy, and laser scanning microscopy. Mod Pathol 1998;11:232-8.

(114) Kajiwara H, Itoh Y, Itoh J, Yasuda M, Osamura RY. Immunohistochemical expressions of prohormone convertase (PC)1/3 and PC2 in carcinoids of various organs. Tokai J Exp Clin Med 1999;24:13-20.

(115) Mbikay M, Sirois F, Yao J, Seidah NG, Chretien M. Comparative analysis of expression of the proprotein convertases furin, PACE4, PC1 and PC2 in human lung tumours. Br J Cancer 1997;75:1509-14.

(116) Cheng M, Watson PH, Paterson JA, Seidah N, Chretien M, Shiu RP. Pro-protein convertase gene expression in human breast cancer. Int J Cancer 1997;71:966-71.

(117) Bassi DE, Mahloogi H, Al-Saleem L, Lopez De CR, Ridge JA, Klein-Szanto AJ. Elevated furin expression in aggressive human head and neck tumors and tumor cell lines. Mol Carcinog 2001;31:224-32.

(118) Page RE, Klein-Szanto AJ, Litwin S, Nicolas E, Al-Jumaily R, Alexander P, et al. Increased expression of the pro-protein convertase furin predicts decreased survival in ovarian cancer. Cell Oncol 2007;29:289-99.

(119) Hubbard FC, Goodrow TL, Liu SC, Brilliant MH, Basset P, Mains RE, et al. Expression of PACE4 in chemically induced carcinomas is associated with spindle cell tumor conversion and increased invasive ability. Cancer Res 1997;57:5226-31.

(120) Mahloogi H, Bassi DE, Klein-Szanto AJ. Malignant conversion of non-tumorigenic murine skin keratinocytes overexpressing PACE4. Carcinogenesis 2002;23:565-72.

167

(121) D'Anjou F, Routhier S, Perreault JP, Latil A, Bonnel D, Fournier I, et al. Molecular Validation of PACE4 as a Target in Prostate Cancer. Transl Oncol 2011;4:157-72.

(122) Rovere C, Barbero P, Maoret JJ, Laburthe M, Kitabgi P. Pro- neurotensin/neuromedin N expression and processing in human colon cancer cell lines. Biochem Biophys Res Commun 1998;246:155-9.

(123) Bassi DE, Fu J, Lopez de CR, Klein-Szanto AJ. Proprotein convertases: "master switches" in the regulation of tumor growth and progression. Mol Carcinog 2005;44:151- 61.

(124) Chan JM, Stampfer MJ, Giovannucci E, Gann PH, Ma J, Wilkinson P, et al. Plasma insulin-like growth factor-I and prostate cancer risk: a prospective study. Science 1998;279:563-6.

(125) Kaleko M, Rutter WJ, Miller AD. Overexpression of the human insulinlike growth factor I receptor promotes ligand-dependent neoplastic transformation. Mol Cell Biol 1990;10:464-73.

(126) Duguay SJ, Milewski WM, Young BD, Nakayama K, Steiner DF. Processing of wild-type and mutant proinsulin-like growth factor-IA by subtilisin-related proprotein convertases. J Biol Chem 1997;272:6663-70.

(127) Scamuffa N, Siegfried G, Bontemps Y, Ma L, Basak A, Cherel G, et al. Selective inhibition of proprotein convertases represses the metastatic potential of human colorectal tumor cells. J Clin Invest 2008;118:352-63.

(128) Lehmann M, Andre F, Bellan C, Remacle-Bonnet M, Garrouste F, Parat F, et al. Deficient processing and activity of type I insulin-like growth factor receptor in the furin- deficient LoVo-C5 cells. Endocrinology 1998;139:3763-71.

(129) George D. Platelet-derived growth factor receptors: a therapeutic target in solid tumors. Semin Oncol 2001;28:27-33.

168

(130) Siegfried G, Khatib AM, Benjannet S, Chretien M, Seidah NG. The proteolytic processing of pro-platelet-derived growth factor-A at RRKR(86) by members of the proprotein convertase family is functionally correlated to platelet-derived growth factor- A-induced functions and tumorigenicity. Cancer Res 2003;63:1458-63.

(131) Siegfried G, Basak A, Prichett-Pejic W, Scamuffa N, Ma L, Benjannet S, et al. Regulation of the stepwise proteolytic cleavage and secretion of PDGF-B by the proprotein convertases. Oncogene 2005;24:6925-35.

(132) de CM. The transforming growth factor-beta superfamily of receptors. Cytokine Growth Factor Rev 2004;15:1-11.

(133) Akhurst RJ, Derynck R. TGF-beta signaling in cancer--a double-edged sword. Trends Cell Biol 2001;11:S44-S51.

(134) Levy L, Hill CS. Alterations in components of the TGF-beta superfamily signaling pathways in human cancer. Cytokine Growth Factor Rev 2006;17:41-58.

(135) Schultz-Cherry S, Chen H, Mosher DF, Misenheimer TM, Krutzsch HC, Roberts DD, et al. Regulation of transforming growth factor-beta activation by discrete sequences of thrombospondin 1. J Biol Chem 1995;270:7304-10.

(136) Khalil N, Corne S, Whitman C, Yacyshyn H. Plasmin regulates the activation of cell-associated latent TGF-beta 1 secreted by rat alveolar macrophages after in vivo bleomycin injury. Am J Respir Cell Mol Biol 1996;15:252-9.

(137) Moldovan F, Pelletier JP, Mineau F, Dupuis M, Cloutier JM, Martel-Pelletier J. Modulation of collagenase 3 in human osteoarthritic cartilage by activation of extracellular transforming growth factor beta: role of furin convertase. Arthritis Rheum 2000;43:2100-9.

(138) Blakytny R, Ludlow A, Martin GE, Ireland G, Lund LR, Ferguson MW, et al. Latent TGF-beta1 activation by platelets. J Cell Physiol 2004;199:67-76.

169

(139) Kusakabe M, Cheong PL, Nikfar R, McLennan IS, Koishi K. The structure of the TGF-beta latency associated peptide region determines the ability of the proprotein convertase furin to cleave TGF-betas. J Cell Biochem 2008;103:311-20.

(140) Blanchette F, Rivard N, Rudd P, Grondin F, Attisano L, Dubois CM. Cross-talk between the p42/p44 MAP kinase and Smad pathways in transforming growth factor beta 1-induced furin gene transactivation. J Biol Chem 2001;276:33986-94.

(141) Udagawa T, Fernandez A, Achilles EG, Folkman J, D'Amato RJ. Persistence of microscopic human cancers in mice: alterations in the angiogenic balance accompanies loss of tumor dormancy. FASEB J 2002;16:1361-70.

(142) Salven P, Lymboussaki A, Heikkila P, Jaaskela-Saari H, Enholm B, Aase K, et al. Vascular endothelial growth factors VEGF-B and VEGF-C are expressed in human tumors. Am J Pathol 1998;153:103-8.

(143) Kishimoto K, Sasaki A, Yoshihama Y, Mese H, Tsukamoto G, Matsumura T. Expression of vascular endothelial growth factor-C predicts regional lymph node metastasis in early oral squamous cell carcinoma. Oral Oncol 2003;39:391-6.

(144) Stacker SA, Caesar C, Baldwin ME, Thornton GE, Williams RA, Prevo R, et al. VEGF-D promotes the metastatic spread of tumor cells via the lymphatics. Nat Med 2001;7:186-91.

(145) Stacker SA, Achen MG, Jussila L, Baldwin ME, Alitalo K. Lymphangiogenesis and cancer metastasis. Nat Rev Cancer 2002;2:573-83.

(146) Joukov V, Sorsa T, Kumar V, Jeltsch M, Claesson-Welsh L, Cao Y, et al. Proteolytic processing regulates receptor specificity and activity of VEGF-C. EMBO J 1997;16:3898-911.

(147) Stacker SA, Stenvers K, Caesar C, Vitali A, Domagala T, Nice E, et al. Biosynthesis of vascular endothelial growth factor-D involves proteolytic processing which generates non-covalent homodimers. J Biol Chem 1999;274:32127-36.

170

(148) Siegfried G, Basak A, Cromlish JA, Benjannet S, Marcinkiewicz J, Chretien M, et al. The secretory proprotein convertases furin, PC5, and PC7 activate VEGF-C to induce tumorigenesis. J Clin Invest 2003;111:1723-32.

(149) McColl BK, Paavonen K, Karnezis T, Harris NC, Davydova N, Rothacker J, et al. Proprotein convertases promote processing of VEGF-D, a critical step for binding the angiogenic receptor VEGFR-2. FASEB J 2007;21:1088-98.

(150) Vleminckx K, Vakaet L, Jr., Mareel M, Fiers W, van RF. Genetic manipulation of E-cadherin expression by epithelial tumor cells reveals an invasion suppressor role. Cell 1991;66:107-19.

(151) Van RF, Berx G. The cell-cell adhesion molecule E-cadherin. Cell Mol Life Sci 2008;65:3756-88.

(152) Posthaus H, Dubois CM, Laprise MH, Grondin F, Suter MM, Muller E. Proprotein cleavage of E-cadherin by furin in baculovirus over-expression system: potential role of other convertases in mammalian cells. FEBS Lett 1998;438:306-10.

(153) Tsuji A, Ikoma T, Hashimoto E, Matsuda Y. Development of selectivity of alpha1- antitrypsin variant by mutagenesis in its reactive site loop against proprotein convertase. A crucial role of the P4 arginine in PACE4 inhibition. Protein Eng 2002;15:123-30.

(154) Tamura K, Shan WS, Hendrickson WA, Colman DR, Shapiro L. Structure-function analysis of cell adhesion by neural (N-) cadherin. Neuron 1998;20:1153-63.

(155) Maret D, Gruzglin E, Sadr MS, Siu V, Shan W, Koch AW, et al. Surface expression of precursor N-cadherin promotes tumor cell invasion. Neoplasia 2010;12:1066-80.

(156) Maret D, Sadr MS, Sadr ES, Colman DR, Del Maestro RF, Seidah NG. Opposite Roles of Furin and PC5A in N-Cadherin Processing. Neoplasia 2012;14:880-92.

(157) Desgrosellier JS, Cheresh DA. Integrins in cancer: biological implications and therapeutic opportunities. Nat Rev Cancer 2010;10:9-22.

171

(158) Kren A, Baeriswyl V, Lehembre F, Wunderlin C, Strittmatter K, Antoniadis H, et al. Increased tumor cell dissemination and cellular senescence in the absence of beta1- integrin function. EMBO J 2007;26:2832-42.

(159) Delwel GO, Hogervorst F, Sonnenberg A. Cleavage of the alpha6A subunit is essential for activation of the alpha6Abeta1 integrin by phorbol 12-myristate 13-acetate. J Biol Chem 1996;271:7293-6.

(160) Lehmann M, Rigot V, Seidah NG, Marvaldi J, Lissitzky JC. Lack of integrin alpha- chain endoproteolytic cleavage in furin-deficient human colon adenocarcinoma cells LoVo. Biochem J 1996;317 ( Pt 3):803-9.

(161) Lissitzky JC, Luis J, Munzer JS, Benjannet S, Parat F, Chretien M, et al. Endoproteolytic processing of integrin pro-alpha subunits involves the redundant function of furin and proprotein convertase (PC) 5A, but not paired basic amino acid converting enzyme (PACE) 4, PC5B or PC7. Biochem J 2000;346 Pt 1:133-8.

(162) Bergeron E, Basak A, Decroly E, Seidah NG. Processing of alpha4 integrin by the proprotein convertases: histidine at position P6 regulates cleavage. Biochem J 2003;373:475-84.

(163) Nagase H. Activation mechanisms of matrix metalloproteinases. Biol Chem 1997;378:151-60.

(164) Pei D, Weiss SJ. Furin-dependent intracellular activation of the human stromelysin- 3 zymogen. Nature 1995;375:244-7.

(165) Santavicca M, Noel A, Angliker H, Stoll I, Segain JP, Anglard P, et al. Characterization of structural determinants and molecular mechanisms involved in pro- stromelysin-3 activation by 4-aminophenylmercuric acetate and furin-type convertases. Biochem J 1996;315 ( Pt 3):953-8.

(166) Stetler-Stevenson WG, Aznavoorian S, Liotta LA. Tumor cell interactions with the extracellular matrix during invasion and metastasis. Annu Rev Cell Biol 1993;9:541-73.

172

(167) Cao J, Rehemtulla A, Pavlaki M, Kozarekar P, Chiarelli C. Furin directly cleaves proMMP-2 in the trans-Golgi network resulting in a nonfunctioning proteinase. J Biol Chem 2005;280:10974-80.

(168) Sato H, Takino T, Okada Y, Cao J, Shinagawa A, Yamamoto E, et al. A matrix metalloproteinase expressed on the surface of invasive tumour cells. Nature 1994;370:61- 5.

(169) Atkinson SJ, Crabbe T, Cowell S, Ward RV, Butler MJ, Sato H, et al. Intermolecular autolytic cleavage can contribute to the activation of progelatinase A by cell membranes. J Biol Chem 1995;270:30479-85.

(170) Sato H, Seiki M. Membrane-type matrix metalloproteinases (MT-MMPs) in tumor metastasis. J Biochem 1996;119:209-15.

(171) Ohuchi E, Imai K, Fujii Y, Sato H, Seiki M, Okada Y. Membrane type 1 matrix metalloproteinase digests interstitial collagens and other extracellular matrix macromolecules. J Biol Chem 1997;272:2446-51.

(172) Koshikawa N, Giannelli G, Cirulli V, Miyazaki K, Quaranta V. Role of cell surface metalloprotease MT1-MMP in epithelial cell migration over laminin-5. J Cell Biol 2000;148:615-24.

(173) Ratnikov BI, Rozanov DV, Postnova TI, Baciu PG, Zhang H, DiScipio RG, et al. An alternative processing of integrin alpha(v) subunit in tumor cells by membrane type-1 matrix metalloproteinase. J Biol Chem 2002;277:7377-85.

(174) Pei D, Weiss SJ. Transmembrane-deletion mutants of the membrane-type matrix metalloproteinase-1 process progelatinase A and express intrinsic matrix-degrading activity. J Biol Chem 1996;271:9135-40.

(175) Yana I, Weiss SJ. Regulation of membrane type-1 matrix metalloproteinase activation by proprotein convertases. Mol Biol Cell 2000;11:2387-401.

173

(176) Zhang D, Brodt P. Type 1 insulin-like growth factor regulates MT1-MMP synthesis and tumor invasion via PI 3-kinase/Akt signaling. Oncogene 2003;22:974-82.

(177) McMahon S, Laprise MH, Dubois CM. Alternative pathway for the role of furin in tumor cell invasion process. Enhanced MMP-2 levels through bioactive TGFbeta. Exp Cell Res 2003;291:326-39.

(178) Peiretti F, Canault M, prez-Beauclair P, Berthet V, Bonardo B, Juhan-Vague I, et al. Intracellular maturation and transport of tumor necrosis factor alpha converting enzyme. Exp Cell Res 2003;285:278-85.

(179) Srour N, Lebel A, McMahon S, Fournier I, Fugere M, Day R, et al. TACE/ADAM- 17 maturation and activation of sheddase activity require proprotein convertase activity. FEBS Lett 2003;554:275-83.

(180) Anders A, Gilbert S, Garten W, Postina R, Fahrenholz F. Regulation of the alpha- secretase ADAM10 by its prodomain and proprotein convertases. FASEB J 2001;15:1837-9.

(181) Loechel F, Gilpin BJ, Engvall E, Albrechtsen R, Wewer UM. Human ADAM 12 (meltrin alpha) is an active metalloprotease. J Biol Chem 1998;273:16993-7.

(182) Lum L, Reid MS, Blobel CP. Intracellular maturation of the mouse metalloprotease disintegrin MDC15. J Biol Chem 1998;273:26236-47.

(183) Lopez-Perez E, Zhang Y, Frank SJ, Creemers J, Seidah N, Checler F. Constitutive alpha-secretase cleavage of the beta-amyloid precursor protein in the furin-deficient LoVo cell line: involvement of the pro-hormone convertase 7 and the disintegrin metalloprotease ADAM10. J Neurochem 2001;76:1532-9.

(184) Murphy G. The ADAMs: signalling scissors in the tumour microenvironment. Nat Rev Cancer 2008;8:929-41.

174

(185) Jacobsen KT, Adlerz L, Multhaup G, Iverfeldt K. Insulin-like growth factor-1 (IGF-1)-induced processing of amyloid-beta precursor protein (APP) and APP-like protein 2 is mediated by different metalloproteinases. J Biol Chem 2010;285:10223-31.

(186) Reiss K, Maretzky T, Ludwig A, Tousseyn T, de SB, Hartmann D, et al. ADAM10 cleavage of N-cadherin and regulation of cell-cell adhesion and beta-catenin nuclear signalling. EMBO J 2005;24:742-52.

(187) Najy AJ, Day KC, Day ML. ADAM15 supports prostate cancer metastasis by modulating tumor cell-endothelial cell interaction. Cancer Res 2008;68:1092-9.

(188) Cohen JC, Boerwinkle E, Mosley TH, Jr., Hobbs HH. Sequence variations in PCSK9, low LDL, and protection against coronary heart disease. N Engl J Med 2006;354:1264-72.

(189) Kritchevsky SB, Kritchevsky D. Serum cholesterol and cancer risk: an epidemiologic perspective. Annu Rev Nutr 1992;12:391-416.

(190) Kitahara CM, Berrington de GA, Freedman ND, Huxley R, Mok Y, Jee SH, et al. Total cholesterol and cancer risk in a large prospective study in Korea. J Clin Oncol 2011;29:1592-8.

(191) Platz EA, Clinton SK, Giovannucci E. Association between plasma cholesterol and prostate cancer in the PSA era. Int J Cancer 2008;123:1693-8.

(192) Iso H, Ikeda A, Inoue M, Sato S, Tsugane S. Serum cholesterol levels in relation to the incidence of cancer: the JPHC study cohorts. Int J Cancer 2009;125:2679-86.

(193) Borena W, Strohmaier S, Lukanova A, Bjorge T, Lindkvist B, Hallmans G, et al. Metabolic risk factors and primary liver cancer in a prospective study of 578,700 adults. Int J Cancer 2012;131:193-200.

(194) Kritchevsky SB, Wilcosky TC, Morris DL, Truong KN, Tyroler HA. Changes in plasma lipid and lipoprotein cholesterol and weight prior to the diagnosis of cancer. Cancer Res 1991;51:3198-203.

175

(195) Strasak AM, Pfeiffer RM, Brant LJ, Rapp K, Hilbe W, Oberaigner W, et al. Time- dependent association of total serum cholesterol and cancer incidence in a cohort of 172,210 men and women: a prospective 19-year follow-up study. Ann Oncol 2009;20:1113-20.

(196) Knekt P, Reunanen A, Aromaa A, Heliovaara M, Hakulinen T, Hakama M. Serum cholesterol and risk of cancer in a cohort of 39,000 men and women. J Clin Epidemiol 1988;41:519-30.

(197) Shannon J, Tewoderos S, Garzotto M, Beer TM, Derenick R, Palma A, et al. Statins and prostate cancer risk: a case-control study. Am J Epidemiol 2005;162:318-25.

(198) Poynter JN, Gruber SB, Higgins PD, Almog R, Bonner JD, Rennert HS, et al. Statins and the risk of colorectal cancer. N Engl J Med 2005;352:2184-92.

(199) Moyad MA, Klotz LH. Statin Clinical Trial (REALITY) for prostate cancer: an over 15-year wait is finally over thanks to a dietary supplement. Urol Clin North Am 2011;38:325-31.

(200) Cheng D, Espenshade PJ, Slaughter CA, Jaen JC, Brown MS, Goldstein JL. Secreted site-1 protease cleaves peptides corresponding to luminal loop of sterol regulatory element-binding proteins. J Biol Chem 1999;274:22805-12.

(201) Lindberg I, van den Hurk WH, Bui C, Batie CJ. Enzymatic characterization of immunopurified prohormone convertase 2: potent inhibition by a 7B2 peptide fragment. Biochemistry 1995;34:5486-93.

(202) Westphal CH, Muller L, Zhou A, Zhu X, Bonner-Weir S, Schambelan M, et al. The neuroendocrine protein 7B2 is required for peptide hormone processing in vivo and provides a novel mechanism for pituitary Cushing's disease. Cell 1999;96:689-700.

(203) Paquet L, Bergeron F, Boudreault A, Seidah NG, Chretien M, Mbikay M, et al. The neuroendocrine precursor 7B2 is a sulfated protein proteolytically processed by a ubiquitous furin-like convertase. J Biol Chem 1994;269:19279-85.

176

(204) Zhu X, Rouille Y, Lamango NS, Steiner DF, Lindberg I. Internal cleavage of the inhibitory 7B2 carboxyl-terminal peptide by PC2: a potential mechanism for its inactivation. Proc Natl Acad Sci U S A 1996;93:4919-24.

(205) Fricker LD, McKinzie AA, Sun J, Curran E, Qian Y, Yan L, et al. Identification and characterization of proSAAS, a granin-like neuroendocrine peptide precursor that inhibits prohormone processing. J Neurosci 2000;20:639-48.

(206) Bernot D, Stalin J, Stocker P, Bonardo B, Scroyen I, Alessi MC, et al. Plasminogen activator inhibitor 1 is an intracellular inhibitor of furin proprotein convertase. J Cell Sci 2011;124:1224-30.

(207) Garten W, Hallenberger S, Ortmann D, Schafer W, Vey M, Angliker H, et al. Processing of viral glycoproteins by the subtilisin-like endoprotease furin and its inhibition by specific peptidylchloroalkylketones. Biochimie 1994;76:217-25.

(208) Jean F, Boudreault A, Basak A, Seidah NG, Lazure C. Fluorescent peptidyl substrates as an aid in studying the substrate specificity of human prohormone convertase PC1 and human furin and designing a potent irreversible inhibitor. J Biol Chem 1995;270:19225-31.

(209) Jean F, Stella K, Thomas L, Liu G, Xiang Y, Reason AJ, et al. alpha1-Antitrypsin Portland, a bioengineered serpin highly selective for furin: application as an antipathogenic agent. Proc Natl Acad Sci U S A 1998;95:7293-8.

(210) Boudreault A, Gauthier D, Lazure C. Proprotein convertase PC1/3-related peptides are potent slow tight-binding inhibitors of murine PC1/3 and Hfurin. J Biol Chem 1998;273:31574-80.

(211) Zhong M, Munzer JS, Basak A, Benjannet S, Mowla SJ, Decroly E, et al. The prosegments of furin and PC7 as potent inhibitors of proprotein convertases. In vitro and ex vivo assessment of their efficacy and selectivity. J Biol Chem 1999;274:33913-20.

177

(212) Van RL, Ayoubi T, Van D, V, Marynen P. Inhibition of intracellular proteolytic processing of soluble proproteins by an engineered alpha 2-macroglobulin containing a furin recognition sequence in the bait region. Biochem J 1997;326 ( Pt 2):507-14.

(213) Dahlen JR, Jean F, Thomas G, Foster DC, Kisiel W. Inhibition of soluble recombinant furin by human proteinase inhibitor 8. J Biol Chem 1998;273:1851-4.

(214) Lu W, Zhang W, Molloy SS, Thomas G, Ryan K, Chiang Y, et al. Arg15-Lys17- Arg18 turkey ovomucoid third domain inhibits human furin. J Biol Chem 1993;268:14583-5.

(215) Benjannet S, Savaria D, Laslop A, Munzer JS, Chretien M, Marcinkiewicz M, et al. Alpha1-antitrypsin Portland inhibits processing of precursors mediated by proprotein convertases primarily within the constitutive secretory pathway. J Biol Chem 1997;272:26210-8.

(216) Anderson ED, Thomas L, Hayflick JS, Thomas G. Inhibition of HIV-1 gp160- dependent membrane fusion by a furin-directed alpha 1-antitrypsin variant. J Biol Chem 1993;268:24887-91.

(217) Bassi DE, Lopez De CR, Mahloogi H, Zucker S, Thomas G, Klein-Szanto AJ. Furin inhibition results in absent or decreased invasiveness and tumorigenicity of human cancer cells. Proc Natl Acad Sci U S A 2001;98:10326-31.

(218) Lopez de CR, Bassi DE, Zucker S, Seidah NG, Klein-Szanto AJ. Human carcinoma cell growth and invasiveness is impaired by the propeptide of the ubiquitous proprotein convertase furin. Cancer Res 2005;65:4162-71.

(219) Mercapide J, Lopez De CR, Bassi DE, Castresana JS, Thomas G, Klein-Szanto AJ. Inhibition of furin-mediated processing results in suppression of astrocytoma cell growth and invasiveness. Clin Cancer Res 2002;8:1740-6.

(220) Bassi DE, Zhang J, Cenna J, Litwin S, Cukierman E, Klein-Szanto AJ. Proprotein convertase inhibition results in decreased skin cell proliferation, tumorigenesis, and metastasis. Neoplasia 2010;12:516-26.

178

(221) Senzer N, Barve M, Kuhn J, Melnyk A, Beitsch P, Lazar M, et al. Phase I Trial of "bi-shRNAi(furin)/GMCSF DNA/Autologous Tumor Cell" Vaccine (FANG) in Advanced Cancer. Mol Ther 2011.

(222) Li MO, Wan YY, Sanjabi S, Robertson AK, Flavell RA. Transforming growth factor-beta regulation of immune responses. Annu Rev Immunol 2006;24:99-146.

(223) Nejjari M, Berthet V, Rigot V, Laforest S, Jacquier MF, Seidah NG, et al. Inhibition of proprotein convertases enhances cell migration and metastases development of human colon carcinoma cells in a rat model. Am J Pathol 2004;164:1925-33.

(224) Huang YH, Lin KH, Liao CH, Lai MW, Tseng YH, Yeh CT. Furin overexpression suppresses tumor growth and predicts a better postoperative disease-free survival in hepatocellular carcinoma PLoS One 2012;7:e40738.

(225) Scamuffa N, Calvo F, Chretien M, Seidah NG, Khatib AM. Proprotein convertases: lessons from knockouts. FASEB J 2006;20:1954-63.

(226) Creemers JW, Khatib AM. Knock-out mouse models of proprotein convertases: unique functions or redundancy? Front Biosci 2008;13:4960-71.

(227) Khatib AM, Siegfried G, Prat A, Luis J, Chretien M, Metrakos P, et al. Inhibition of proprotein convertases is associated with loss of growth and tumorigenicity of HT-29 human colon carcinoma cells: importance of insulin-like growth factor-1 (IGF-1) receptor processing in IGF-1-mediated functions. J Biol Chem 2001;276:30686-93.

(228) Singh H, Heng S, Nicholls PK, Li Y, Tai LT, Jobling T, et al. Proprotein convertases in post-menopausal endometrial cancer: Distinctive regulation and non- invasive diagnosis. Biochem Biophys Res Commun 2012.

(229) Tran NL, Nagle RB, Cress AE, Heimark RL. N-Cadherin expression in human prostate carcinoma cell lines. An epithelial-mesenchymal transformation mediating adhesion withStromal cells. Am J Pathol 1999;155:787-98.

179

(230) Kang Y, Massague J. Epithelial-mesenchymal transitions: twist in development and metastasis. Cell 2004;118:277-9.

(231) Szumska D, Pieles G, Essalmani R, Bilski M, Mesnard D, Kaur K, et al. VACTERL/caudal regression/Currarino syndrome-like malformations in mice with mutation in the proprotein convertase Pcsk5. Genes Dev 2008;22:1465-77.

(232) Marchesi C, Essalmani R, Lemarie CA, Leibovitz E, Ebrahimian T, Paradis P, et al. Inactivation of endothelial proprotein convertase 5/6 decreases collagen deposition in the cardiovascular system: role of fibroblast autophagy. J Mol Med (Berl) 2011;89:1103-11.

(233) Henrich S, Lindberg I, Bode W, Than ME. Proprotein convertase models based on the crystal structures of furin and kexin: explanation of their specificity. J Mol Biol 2005;345:211-27.

(234) Vihinen P, la-aho R, Kahari VM. Matrix metalloproteinases as therapeutic targets in cancer. Curr Cancer Drug Targets 2005;5:203-20.

(235) Thornberry NA, Rano TA, Peterson EP, Rasper DM, Timkey T, Garcia-Calvo M, et al. A combinatorial approach defines specificities of members of the caspase family and granzyme B. Functional relationships established for key mediators of apoptosis. J Biol Chem 1997;272:17907-11.

(236) Stennicke HR, Renatus M, Meldal M, Salvesen GS. Internally quenched fluorescent peptide substrates disclose the subsite preferences of human caspases 1, 3, 6, 7 and 8. Biochem J 2000;350 Pt 2:563-8.

(237) Remacle AG, Shiryaev SA, Oh ES, Cieplak P, Srinivasan A, Wei G, et al. Substrate cleavage analysis of furin and related proprotein convertases. A comparative study. J Biol Chem 2008;283:20897-906.

(238) Dykxhoorn DM, Palliser D, Lieberman J. The silent treatment: siRNAs as small molecule drugs. Gene Ther 2006;13:541-52.

180

(239) Roebroek AJ, Umans L, Pauli IG, Robertson EJ, van LF, Van d, V, et al. Failure of ventral closure and axial rotation in embryos lacking the proprotein convertase Furin. Development 1998;125:4863-76.

(240) Yang J, Goldstein JL, Hammer RE, Moon YA, Brown MS, Horton JD. Decreased lipid synthesis in livers of mice with disrupted Site-1 protease gene. Proc Natl Acad Sci U S A 2001;98:13607-12.

(241) Constam DB, Robertson EJ. SPC4/PACE4 regulates a TGFbeta signaling network during axis formation. Genes Dev 2000;14:1146-55.

(242) Furuta M, Yano H, Zhou A, Rouille Y, Holst JJ, Carroll R, et al. Defective prohormone processing and altered pancreatic islet morphology in mice lacking active SPC2. Proc Natl Acad Sci U S A 1997;94:6646-51.

(243) Zhu X, Orci L, Carroll R, Norrbom C, Ravazzola M, Steiner DF. Severe block in processing of proinsulin to insulin accompanied by elevation of des-64,65 proinsulin intermediates in islets of mice lacking prohormone convertase 1/3. Proc Natl Acad Sci U S A 2002;99:10299-304.

(244) Remacle AG, Gawlik K, Golubkov VS, Cadwell GW, Liddington RC, Cieplak P, et al. Selective and potent furin inhibitors protect cells from anthrax without significant toxicity. Int J Biochem Cell Biol 2010;42:987-95.

(245) Muller L, Cameron A, Fortenberry Y, Apletalina EV, Lindberg I. Processing and sorting of the prohormone convertase 2 propeptide. J Biol Chem 2000;275:39213-22.

(246) Anderson ED, Molloy SS, Jean F, Fei H, Shimamura S, Thomas G. The ordered and compartment-specfific autoproteolytic removal of the furin intramolecular chaperone is required for enzyme activation. J Biol Chem 2002;277:12879-90.

(247) Stawowy P, Graf K, Goetze S, Roser M, Chretien M, Seidah NG, et al. Coordinated regulation and colocalization of alphav integrin and its activating enzyme proprotein convertase PC5 in vivo. Histochem Cell Biol 2003;119:239-45.

181

(248) Chan JC, Piper DE, Cao Q, Liu D, King C, Wang W, et al. A proprotein convertase subtilisin/kexin type 9 neutralizing antibody reduces serum cholesterol in mice and nonhuman primates. Proc Natl Acad Sci U S A 2009;106:9820-5.

(249) Stein EA, Gipe D, Bergeron J, Gaudet D, Weiss R, Dufour R, et al. Effect of a monoclonal antibody to PCSK9, REGN727/SAR236553, to reduce low-density lipoprotein cholesterol in patients with heterozygous familial hypercholesterolaemia on stable statin dose with or without ezetimibe therapy: a phase 2 randomised controlled trial. Lancet 2012;380:29-36.

(250) Broitman SA, Wilkinson J, Cerda S, Branch SK. Effects of monoterpenes and mevinolin on murine colon tumor CT-26 in vitro and its hepatic "metastases" in vivo. Adv Exp Med Biol 1996;401:111-30.

(251) Briel M, Nordmann AJ, Bucher HC. Statin therapy for prevention and treatment of acute and chronic cardiovascular disease: update on recent trials and metaanalyses. Curr Opin Lipidol 2005;16:601-5.

(252) Awan Z, Seidah NG, MacFadyen JG, Benjannet S, Chasman DI, Ridker PM, et al. Rosuvastatin, Proprotein Convertase Subtilisin/Kexin Type 9 Concentrations, and LDL Cholesterol Response: the JUPITER Trial. Clin Chem 2012;58:183-9.

(253) Seidah NG, Prat A. The proprotein convertases are potential targets in the treatment of dyslipidemia. J Mol Med 2007;85:685-96.

(254) Ni YG, Di MS, Condra JH, Peterson LB, Wang W, Wang F, et al. A PCSK9- binding antibody that structurally mimics the EGF(A) domain of LDL-receptor reduces LDL cholesterol in vivo. J Lipid Res 2011;52:78-86.

(255) Gupta N, Fisker N, Asselin MC, Lindholm M, Rosenbohm C, Orum H, et al. A locked nucleic acid antisense oligonucleotide (LNA) silences PCSK9 and enhances LDLR expression in vitro and in vivo. PLoS One 2010;5:e10682.

182

(256) Lindholm MW, Elmen J, Fisker N, Hansen HF, Persson R, Moller MR, et al. PCSK9 LNA Antisense Oligonucleotides Induce Sustained Reduction of LDL Cholesterol in Nonhuman Primates. Mol Ther 2011.

(257) Abbruzzese JL, Abbruzzese MC, Lenzi R, Hess KR, Raber MN. Analysis of a diagnostic strategy for patients with suspected tumors of unknown origin. J Clin Oncol 1995;13:2094-103.

(258) Willnow TE, Nykjaer A, Herz J. Lipoprotein receptors: new roles for ancient proteins. Nat Cell Biol 1999;1:E157-E162.

(259) Herz J, Hamann U, Rogne S, Myklebost O, Gausepohl H, Stanley KK. Surface location and high affinity for calcium of a 500-kd liver membrane protein closely related to the LDL-receptor suggest a physiological role as lipoprotein receptor. EMBO J 1988;7:4119-27.

(260) Krieger M, Herz J. Structures and functions of multiligand lipoprotein receptors: macrophage scavenger receptors and LDL receptor-related protein (LRP). Annu Rev Biochem 1994;63:601-37.

(261) Herz J, Strickland DK. LRP: a multifunctional scavenger and signaling receptor. J Clin Invest 2001;108:779-84.

(262) Herz J, Clouthier DE, Hammer RE. LDL receptor-related protein internalizes and degrades uPA-PAI-1 complexes and is essential for embryo implantation. Cell 1992;71:411-21.

(263) Spijkers PP, da Costa MP, Westein E, Gahmberg CG, Zwaginga JJ, Lenting PJ. LDL-receptor-related protein regulates beta2-integrin-mediated leukocyte adhesion. Blood 2005;105:170-7.

(264) Ranganathan S, Cao C, Catania J, Migliorini M, Zhang L, Strickland DK. Molecular basis for the interaction of low density lipoprotein receptor-related protein 1 (LRP1) with integrin alphaMbeta2: identification of binding sites within alphaMbeta2 for LRP1. J Biol Chem 2011;286:30535-41.

183

(265) Cao C, Lawrence DA, Li Y, Von Arnim CA, Herz J, Su EJ, et al. Endocytic receptor LRP together with tPA and PAI-1 coordinates Mac-1-dependent macrophage migration. EMBO J 2006;25:1860-70.

(266) Llorente-Cortes V, Casani L, Cal R, Llenas A, Juan-Babot O, Camino-Lopez S, et al. Cholesterol-lowering strategies reduce vascular LRP1 overexpression induced by hypercholesterolaemia. Eur J Clin Invest 2011;41:1087-97.

(267) Kim S, Kim CH, Vaziri ND. Upregulation of hepatic LDL receptor-related protein in nephrotic syndrome: response to statin therapy. Am J Physiol Endocrinol Metab 2005;288:E813-E817.

(268) Willnow TE, Nykjaer A, Herz J. Lipoprotein receptors: new roles for ancient proteins. Nat Cell Biol 1999;1:E157-E162.

(269) Carswell EA, Old LJ, Kassel RL, Green S, Fiore N, Williamson B. An endotoxin- induced serum factor that causes necrosis of tumors. Proc Natl Acad Sci U S A 1975;72:3666-70.

(270) Eggermont AM, de Wilt JH, ten Hagen TL. Current uses of isolated limb perfusion in the clinic and a model system for new strategies. Lancet Oncol 2003;4:429-37.

(271) Alegre JF, Duarte M, Sureda GM, Bretcha BP, Dussan C, Ballester A, et al. Tumor necrosis factor alpha and melfalan-based hyperthermic isolated limb perfusion in locally advanced extremity soft tissue sarcomas and melanomas. Cir Esp 2012.

(272) Joyce D, Albanese C, Steer J, Fu M, Bouzahzah B, Pestell RG. NF-kappaB and cell-cycle regulation: the cyclin connection. Cytokine Growth Factor Rev 2001;12:73-90.

(273) Libermann TA, Baltimore D. Activation of interleukin-6 gene expression through the NF-kappa B transcription factor. Mol Cell Biol 1990;10:2327-34.

(274) Bond M, Fabunmi RP, Baker AH, Newby AC. Synergistic upregulation of metalloproteinase-9 by growth factors and inflammatory cytokines: an absolute requirement for transcription factor NF-kappa B. FEBS Lett 1998;435:29-34.

184

(275) Yoshida S, Ono M, Shono T, Izumi H, Ishibashi T, Suzuki H, et al. Involvement of interleukin-8, vascular endothelial growth factor, and basic fibroblast growth factor in tumor necrosis factor alpha-dependent angiogenesis. Mol Cell Biol 1997;17:4015-23.

(276) Micheau O, Lens S, Gaide O, Alevizopoulos K, Tschopp J. NF-kappaB signals induce the expression of c-FLIP. Mol Cell Biol 2001;21:5299-305.

(277) Tamatani M, Che YH, Matsuzaki H, Ogawa S, Okado H, Miyake S, et al. Tumor necrosis factor induces Bcl-2 and Bcl-x expression through NFkappaB activation in primary hippocampal neurons. J Biol Chem 1999;274:8531-8.

(278) Moore RJ, Owens DM, Stamp G, Arnott C, Burke F, East N, et al. Mice deficient in tumor necrosis factor-alpha are resistant to skin carcinogenesis. Nat Med 1999;5:828- 31.

(279) Orosz P, Kruger A, Hubbe M, Ruschoff J, Von HP, Mannel DN. Promotion of experimental liver metastasis by tumor necrosis factor. Int J Cancer 1995;60:867-71.

(280) Schultz RM, Altom MG. Protective activity of recombinant murine tumor necrosis factor-alpha and interferon-gamma against experimental murine lung carcinoma metastases. J Interferon Res 1990;10:229-36.

(281) Chambers AF, MacDonald IC, Schmidt EE, Koop S, Morris VL, Khokha R, et al. Steps in tumor metastasis: new concepts from intravital videomicroscopy. Cancer Metastasis Rev 1995;14:279-301.

(282) Sturm JW, Magdeburg R, Berger K, Petruch B, Samel S, Bonninghoff R, et al. Influence of TNFA on the formation of liver metastases in a syngenic mouse model. Int J Cancer 2003;107:11-21.

(283) Armaka M, Apostolaki M, Jacques P, Kontoyiannis DL, Elewaut D, Kollias G. Mesenchymal cell targeting by TNF as a common pathogenic principle in chronic inflammatory joint and intestinal diseases. J Exp Med 2008;205:331-7.

185

(284) Diehl AM, Rai RM. Liver regeneration 3: Regulation of signal transduction during liver regeneration. FASEB J 1996;10:215-27.

(285) Schulze-Osthoff K, Beyaert R, Vandevoorde V, Haegeman G, Fiers W. Depletion of the mitochondrial electron transport abrogates the cytotoxic and gene-inductive effects of TNF. EMBO J 1993;12:3095-104.

(286) Feingold KR, Soued M, Grunfeld C. Tumor necrosis factor stimulates DNA synthesis in the liver of intact rats. Biochem Biophys Res Commun 1988;153:576-82.

(287) Beyer HS, Stanley M. Tumor necrosis factor-alpha increases hepatic DNA and RNA and hepatocyte mitosis. Biochem Int 1990;22:405-10.

(288) Orrego H, Crossley IR, Saldivia V, Medline A, Varghese G, Israel Y. Long-term ethanol administration and short- and long-term liver regeneration after partial hepatectomy. J Lab Clin Med 1981;97:221-30.

(289) Akita K, Okuno M, Enya M, Imai S, Moriwaki H, Kawada N, et al. Impaired liver regeneration in mice by lipopolysaccharide via TNF-alpha/kallikrein-mediated activation of latent TGF-beta. Gastroenterology 2002;123:352-64.

(290) Yang SQ, Lin HZ, Lane MD, Clemens M, Diehl AM. Obesity increases sensitivity to endotoxin liver injury: implications for the pathogenesis of steatohepatitis. Proc Natl Acad Sci U S A 1997;94:2557-62.

(291) Wands JR, Carter EA, Bucher NL, Isselbacher KJ. Inhibition of hepatic regeneration in rats by acute and chronic ethanol intoxication. Gastroenterology 1979;77:528-31.

(292) Yang SQ, Lin HZ, Yin M, Albrecht JH, Diehl AM. Effects of chronic ethanol consumption on cytokine regulation of liver regeneration. Am J Physiol 1998;275:G696- G704.

186

(293) Ioannou GN, Morrow OB, Connole ML, Lee SP. Association between dietary nutrient composition and the incidence of cirrhosis or liver cancer in the United States population. Hepatology 2009;50:175-84.

(294) Osono Y, Woollett LA, Herz J, Dietschy JM. Role of the low density lipoprotein receptor in the flux of cholesterol through the plasma and across the tissues of the mouse. J Clin Invest 1995;95:1124-32.

(295) Rudling M. Hepatic mRNA levels for the LDL receptor and HMG-CoA reductase show coordinate regulation in vivo. J Lipid Res 1992;33:493-501.

(296) Henkel AS, Anderson KA, Dewey AM, Kavesh MH, Green RM. A chronic high- cholesterol diet paradoxically suppresses hepatic CYP7A1 expression in FVB/NJ mice. J Lipid Res 2011;52:289-98.

(297) Tabas I. The role of endoplasmic reticulum stress in the progression of atherosclerosis. Circ Res 2010;107:839-50.

(298) Tabas I. Consequences of cellular cholesterol accumulation: basic concepts and physiological implications. J Clin Invest 2002;110:905-11.

(299) Devries-Seimon T, Li Y, Yao PM, Stone E, Wang Y, Davis RJ, et al. Cholesterol- induced macrophage apoptosis requires ER stress pathways and engagement of the type A scavenger receptor. J Cell Biol 2005;171:61-73.

(300) Seidah NG. The proprotein convertases, 20 years later. Methods Mol Biol 2011;768:23-57.

187