Evaluation of Chloride Intracellular Channels 4 and 1 Functions in Developmental and Pathological Angiogenesis

Jennifer Jean Tung

Submitted in partial fulfillment of the requirements for the degree of Doctor of Philosophy under the Executive Committee of the Graduate School of Arts and Sciences

COLUMBIA UNIVERSITY

2012

Evaluation of Chloride Intracellular Channels 4 and 1 Functions in Developmental and

Pathological Angiogenesis

Jennifer Jean Tung

Members of the chloride intracellular channel (CLIC) protein family have been

implicated as regulators of tubulogenesis, a critical step in the formation of new blood

vessels during angiogenesis. We sought to determine CLIC4 and CLIC1 function in

angiogenesis. We hypothesized that CLIC4 and CLIC1 act in endothelial lumen

formation and promote both developmental and pathological angiogenesis.

Using in vitro studies, we found that CLIC4 promotes endothelial proliferation,

network formation, capillary-like sprouting, and lumen formation. In vivo, Clic4

knockout mice display a mild defect in retinal vascular development and an apparent

decrease in retinal macrophage content. By implanting murine tumor cells in Clic4

knockout mice, we discovered that Clic4 affects the establishment of lung metastases.

Endothelial and smooth muscle cell content of tumors are comparable to wild type, but

overall vessel architecture is altered. In studying CLIC1, we found that CLIC1

knockdown results in reduced endothelial proliferation, directed migration, network

formation, and capillary-like sprouting in vitro . In vivo analysis revealed no apparent

angiogenic phenotype in the developing retinas of Clic1 knockout mice.

We developed Clic4 -/-;Clic1 -/- double mutant embryos, which were unable to develop beyond 9.5 dpc. Whole mount staining of Clic4 -/-;Clic1 -/- 9.5 dpc embryos for vasculature revealed an angiogenic defect, most notable along the intersomitic vessels and in the brain. Endothelial content is reduced in Clic4;Clic1 double knockout embryos, and double nullizygous embryos were growth retarded. Preliminary histological analysis of Clic4 -/-;Clic1 -/- 9.5 dpc embryos suggests altered aortic development, reduced proliferation, and increased apoptosis.

I conclude that CLIC4 and CLIC1 function in endothelial proliferation and morphogenesis and that Clic4 and Clic1 are required for embryonic development.

Together, our findings indicate that CLIC4 and CLIC1 are important in developmental angiogenesis and should be considered in elucidating the molecular mechanisms of tubulogenesis.

Contents

List of Figures ……………………………………………………………………... iii

List of Tables ……………………………………………………………………... vi

Acknowledgements ……………………………………………………………... vii

Chapter 1. Introduction …………………………………………………….... 1

1.1 Overview of angiogenesis in development and pathology ..…….… 1 1.2 Mechanisms of tubulogenesis ….………………………………...... 2 1.3 Chloride intracellular channels ....……………………….………… 4 1.4 Evidence for CLIC function in angiogenesis ..….…………………. 6 1.5 Figures ...…………………………………………………………... 11

Chapter 2. CLIC4 is involved in endothelial cell proliferation and morphogenesis in vitro …………………………………………... 13

2.1 Abstract .…………………………………………………………… 13 2.2 Introduction .……………………………………………………….. 14 2.3 Altered CLIC4 expression does not affect endothelial cell migration .………………………………………………………….. 17 2.4 CLIC4 expression promotes endothelial cell proliferation ..…...…. 18 2.5 CLIC4 promotes endothelial capillary-like network formation …... 19 2.6 Altered CLIC4 expression affects capillary-like sprouting and lumen formation .……………………………………………….….. 21 2.7 Discussion .………………………………………………………… 22 2.8 Figures ………………………………………………………...... 26 2.9 Additional discussion ..…………………………………………….. 40 2.10 Supplementary figures ...…………………………………………... 41

Chapter 3. Generation and characterization of Clic4 knockout mice ….….. 47

3.1 Abstract .…………………………………………………………… 47 3.2 Introduction .……………………………………………………….. 48 3.3 Clic4 targeting scheme …………………………………………….. 49 3.4 Clic4 knockout mice are underrepresented, but viable and fertile .…………………………………………………………. 51 3.5 Clic4 knockout mice display subtle defects in postnatal angiogenesis and may have reduced macrophage content in the developing retina ….………………………………………… 52 3.6 Clic4 knockout may enhance tumor growth and metastasis ..…..... 53 3.7 Discussion …………..…………………………………………...… 55 3.8 Figures ………………………………………………..…………… 59 3.9 Tables ……………………………………………………………… 69

Chapter 4. Chloride intracellular channel 1 functions in endothelial cell growth and migration ………………………………….…… 70

4.1 Abstract .…………………………………………………………… 70 4.2 Introduction .……………………………………………………….. 71 4.3 Generation of endothelial cell lines with stable CLIC1 knockdown .………………………………………………………... 73 4.4 CLIC1 knockdown inhibits endothelial cell growth .……………… 74 4.5 CLIC1 knockdown inhibits endothelial migration .……………….. 75 4.6 CLIC1 influences expression of select integrins on endothelial cells .……………………………………………...…… 76 4.7 CLIC1 plays a role in capillary-like network formation .……..….. 77 4.8 Reduced CLIC1 expression affects capillary-like sprouting and branching .………………………………………………….….. 78 4.9 Discussion .………………………………………………………… 79 4.10 Figures ……………....…………………………………………….. 83 4.11 Additional discussion .…………………………………………...… 95 4.12 Supplementary figures …………………………………………….. 97 4.13 Supplementary tables ……………………………………………… 103

Chapter 5. Clic4 and Clic1 are required for embryonic development …….. 104

5.1 Abstract …………………………………………………………… 104 5.2 Introduction ……………………………………………………….. 105 5.3 Generation of Clic4 and Clic1 double mutants …………………… 107 5.4 Clic4 -/-;Clic1 -/- double knockout mice are embryonic lethal at 9.5 dpc ………………………………………………..…... 108 5.5 Clic4 -/-;Clic1 -/- double knockout mice display aberrant vascular development ………………………………………………109 5.6 Aortic development and proliferation may be altered in Clic4 -/-;Clic1 -/- embryos ………………………………………….. 110 5.7 Discussion …………………………………………………………. 111 5.8 Figures …………………………………………………………….. 116 5.9 Tables ……………………………………………………………… 134

Chapter 6. Conclusions and future directions ………………………………. 136

Chapter 7. Materials and methods …………………………………………... 147

Bibliography ………………………………………………………………………. 164

Appendix Regulation of cardiovascular development and integrity by the heart of glass–cerebral cavernous malformation protein pathway …………………………………….. 176

List of Figures

Chapter 1.

Fig 1.1 Cell hollowing and cord hollowing models for tubulogenic mechanisms …………………………………………... 11

Chapter 2

Fig 2.1 Establishing CLIC4 knockdown and overexpression in endothelial cells ……………………………………………….... 26

Fig 2.2 CLIC4 expression has no effect on endothelial cell migration ….... 28

Fig 2.3 CLIC4 expression promotes endothelial cell proliferation ……...... 30

Fig 2.4 Altering CLIC4 expression affects endothelial cell survival ……… 32

Fig 2.5 CLIC4 expression promotes in vitro endothelial network formation …...... 34

Fig 2.6 Reduced CLIC4 expression inhibits endothelial network formation independent of its proliferative defect ………………….. 36

Fig 2.7 CLIC4 expression affects capillary-like sprouting in vitro ………...38

Supp Fig 2.1 Clic1 and Clic4 are most highly expressed in cultured endothelial cells …………………………………………… 41

Supp Fig 2.2 CLIC4 expression affects endothelial morphogenesis in a three-dimensional extracellular matrix ……………..… 43

Supp Fig 2.3 CLIC4 expression does not affect ion conductance at the plasma membrane in HEK 293 cells …………………... 45

Chapter 3

Fig 3.1 Diagram of Clic4 targeting strategy and ES cell targeting confirmation …………………………………………...... 59

Fig 3.2 Confirmation of Clic4 knockout …………………………………... 61

Fig 3.3 Clic4 homozygous knockout mice are phenotypically normal at 9.5 dpc ………………………………………………...…63

iii

Fig 3.4 Clic4 knockout mice show reduced retinal vasculature and macrophage content …………………………………………... 65

Fig 3.5 Tumorigenesis is enhanced by Clic4 knockout ………………...... 67

Chapter 4

Fig 4.1 Establishing human endothelial cell lines with CLIC1 knockdown ………………………………………………… 83

Fig 4.2 CLIC1 knockdown has mild effects on viability and significantly inhibits endothelial cell growth ……………………… 85

Fig 4.3 Reduced CLIC1 expression inhibits endothelial migration ……….. 87

Fig 4.4 CLIC1 expression plays a role in the expression of various integrins …………………………………………………… 89

Fig 4.5 Knocked down CLIC1 expression inhibits capillary-like network formation and branching morphogenesis ………………… 91

Fig 4.6 Reduced CLIC1 expression affects capillary-like sprouting and branching …………………………………………… 93

Supp Fig 4.1 Confirmation of CLIC4 overexpression and CLIC1 knockdown ………………………………………… 97

Supp Fig 4.2 Partial rescue of the CLIC1 knockdown network formation defect by CLIC4 ………………………………... 99

Supp Fig 4.3 Loss of Clic1 does not affect postnatal retinal angiogenesis ……………………………………………….. 101

Chapter 5

Fig 5.1 Clic4;Clic1 double knockout embryos present cardiac edema at 10.5 dpc …………………………………………………. 116

Fig 5.2 Embryos missing three Clic alleles are represented at 10.5 dpc ….. 118

Fig 5.3 Clic4 +/+ ;Clic1 -/- 9.5 dpc embryos are indistinguishable from Clic4 +/+ ;Clic1 +/+ embryos …………………………………... 120

Fig 5.4 In utero death for Clic4 -/-;Clic1 -/- embryos occurs at 9.5 dpc with growth retardation and an angiogenic defect …...……………. 122

Fig 5.5 Endothelial content is reduced in Clic4 -/-;Clic1 -/- developing embryos ……………………………………………….. 124

Fig 5.6 Aortic development may be altered in Clic4 -/-;Clic1 -/- embryos ..... 126

Fig 5.7 Jag1 signaling may be altered in Clic4 -/-;Clic1 -/- embryos …..……. 128

Fig 5.8 Overall proliferation may be reduced in developing Clic4 -/-;Clic1 -/- embryos ………………………………………….... 130

Fig 5.9 Overall apoptosis may be increased in developing Clic4 -/-;Clic1 -/- embryos …………………………………………… 132

Chapter 6

Model 6.1 ……………………………………………………………………. 141

Model 6.2 ……………………………………………………………………. 143

Model 6.3 ……………………………………………………………………. 144

List of Tables

Chapter 3

Table 3.1 Clic4 -/- mice are underrepresented at birth ……………….... 69

Chapter 4

Supp Table 4.1 Clic1 -/- mice confer with expected Mendelian ratios ……... 103

Chapter 5

Table 5.1 Inheritance from Clic4 +/-;Clic1 +/- intercrosses deviates from predicted Mendelian ratios …………………. 134

Table 5.2 Mendelian inheritance is restored at 9.5 dpc …………….... 135

Acknowledgements

Thank you to Sean West, Glenn Chris Tan, Roy Bisnar-Maute, Vikram Ranade,

Shahrnaz Kemal, Qing Wang, Danny Concepcion, Kally Pan, Katie Lelli, Tiffany Zee,

Jeana Bisnar-Maute, John Seeley, Zach Ortiz, Grant Osborn, Mary Morales, Shera

Williams, Mallorie Lenn, Gilda Hernandez, Krissy Jackson, Sarah French, Steve Canham,

James Ji, Lauren Palermo, Lindsay Castrillo, Laura Schindler, Birget Moe, Ian Tattersall, and Emerson Lim. Without these people, I would probably still be wearing pajamas in public and I would generally be less sane. Thanks also to all of my other friends, Carrie

Shawber and lab, and members of the Kitajewski lab, especially Claire Vech Reeves,

Thaned Kangsamaksan, and Anshula Sharma.Special thanks to Dave Franco and

Michelle Whaley, both of who are directly responsible for putting me on this career track.

Heartfelt thanks to my mentor Jan Kitajewski for teaching me proficiency, efficiency, consistency, and confidence in facing challenges. Jan’s attention, concern, and guidance have been indispensable to both my professional and personal maturation.

Thank you to my committee members Oliver Hobert, who constantly challenged me to be a better scientist; Gil Di Paolo, for being both incredibly nice and encouraging; and

Thomas Ludwig, whose support and expertise greatly aided this study and my mental well-being. To all four of you, your patience has also been deeply appreciated.

Finally, special gratitude goes to my parents SaSa and Dick Tung whose support, constant availability, pride, encouragement, and love have been essential to my functioning.

vii

Chapter One: Introduction

This chapter contains excerpts from a review that I co-authored with Ian Tattersall

[1].

1.1 Overview of angiogenesis in development and pathology

The processes responsible for blood vasculature development are termed vasculogenesis for primary vessel structure growth, and angiogenesis. Angiogenesis is the formation of new blood vessels from the sprouting, branching, and pruning of preexisting vessels. It is an essential process for mammalian embryonic maturation, proper organ development, and tissue growth and repair. Deviations in angiogenic regulation are important factors in pathological processes such as malignant growth and transformation among many others [2, 3].

The basic cellular mechanisms of angiogenesis begin with pro-angiogenic factors stimulating endothelial cells to degrade the local basement membrane surrounding an existing vessel [4]. Endothelial cells then rearrange, proliferate, and migrate into the surrounding stroma. New sprouts form lumens in a process termed tubulogenesis. After lumen formation, tubes branch and reconnect to form vascular networks in a process termed anastomosis. Pruning of newly formed capillaries and elementary sprouts follows anastomosis, and the basement membrane is reformed. Finally, accessory cells such as pericytes or smooth muscle cells are recruited to help stabilize the vessel, produce extracellular matrix components, and provide control of luminal diameter [5]. 2

When angiogenic regulation is derailed, a number of pathologies such as macular degeneration, diabetic retinopathy, and ischemia can occur. With respect to tumor development, Judah Folkman first hypothesized in 1971 that without tumor-recruited angiogenesis to provide nourishment, waste removal, and a means of escape for tumor cells, the tumor would remain localized or become necrotic [6]. Since its inception, a number of studies have gathered evidence in support of this tumor angiogenesis hypothesis and are well-reviewed [7]. The mechanism driving tumor angiogenesis has been termed the “angiogenic switch” and stipulates a pathological shift to a pro- angiogenic state from the carefully regulated balance of inhibitory and activating angiogenic signals in a normal cell environment [8]. With these early studies, the concept of targeting angiogenesis in treating pathologies was discovered, and targeting tumor angiogenesis is now a well-established goal in anti-cancer drug development [3]. A better understanding of the basic mechanisms that drive angiogenesis could provide new avenues for drug development and tumor targeting.

1.2 Mechanisms of tubulogenesis

Without tubulogenesis, endothelial cords would fail to transport fluid and key components of blood. While it is accepted that endothelial tubulogenesis involves the establishment of apicobasal polarity, vesicular trafficking, and cytoskeletal reorganization, the underlying mechanisms of tubulogenesis remain unclear [9-15]. Recent studies have implicated a broad array of cell surface and signaling proteins in tubulogenesis and several models have been proposed to explain the mechanics of tubulogenesis. 3

An initial model for endothelial tubulogenesis, referred to as “cell hollowing,” involves the coalescence of pinocytic vesicles to generate lumens (Fig 1.1) [16-20]. This model emphasizes the importance of vesicular and vacuolar development and movement within the endothelial cell. By cell hollowing, a single endothelial cell creates a de novo lumen through the coalescence of vesicles/vacuoles intracellularly while cell invasion into the surrounding stroma occurs. These vesicles carry apical markers to distinguish the future luminal membrane from the basal membrane of the endothelial cell, and this process occurs without cell loss [17]. Once the cell is “hollowed,” the lumen contained within the cell opens on either end through intercellular vacuolar fusion by exocytosis to form a “seamless vessel” without junctions [20]. Evidence for this model is found most notably in vitro and by analysis of zebrafish intersomitic vessel sprouting [16, 18, 20].

An alternative model for endothelial tubulogenesis, known as “cord hollowing,” emphasizes the importance of endothelial cell-cell junctions and the establishment of apicobasal polarity (Fig 1.1) [21-24]. In this model, a polarized multi-cellular endothelial cord migrates into the surrounding stroma and loses its apicobasal polarity. During this step, cell junctions in the cord break and the cord thickens to roughly two or three cells.

A de novo lumen is formed extracellularly without cell loss when an external cue signals for repolarization, and surfaces in contact with the extracellular matrix accrue basal markers while apical markers target endothelial cell-cell junctions at the central axis via vesicles. Apical vesicles then fuse at the central axis, separating the cell surfaces with fluid-filled pockets of what will ultimately be a lumen continuous within itself and the parent vessel. Studies using C. elegans and zebrafish provide notable data supporting this model [22-25]. 4

In addition to the cell hollowing and cord hollowing models, other models of tubulogenesis have been proposed. Each model has supporting evidence and utility in explaining the mechanics behind endothelial tubulogenesis. One recent study of the developing mouse aorta proposes a mechanism by which lumens are formed extracellularly when VEGF-A activates a chain of events leading to F-actin rearrangements, ultimately separating two contacting endothelial cells [26]. This process involves Rho-dependent kinase and is not known to invoke vesicular trafficking. Another model, termed “selective sprouting”, is supported by studies in the developing zebrafish cardinal vein [27]. The selective sprouting model stipulates that venous endothelial cells migrate from the dorsal aorta to surround erythrocytes and are retained by VEGF-

A/VEGFR-2 signaling, a process analogous to vasculogenesis during embryogenesis.

The fact that several models of endothelial tubulogenesis are supported by a variety of observations and experiments suggests that the mechanism utilized for lumen formation could be context-dependent. With further research and advancing means for studying tubulogenesis, we may find that lumen formation results from a combination of proposed models or that the method for tubulogenesis depends on vessel type and caliber.

1.3 Chloride intracellular channels

Chloride intracellular channels (CLICs), so named for their homology to p64, are putative ion channels belonging to the p64 gene family and are distantly related to glutathione S-transferases (GSTs) [28, 29]. CLICs were first discovered by the Al-

Awqati group when they identified a protein while searching for novel chloride ion channels [28, 30]. In their studies, they found indanyloxyacetic acid 94 (IAA94), a small 5 molecule inhibitor of ion channel modulator ethacrynic acid, moderately inhibited chloride channel activity in bovine kidney cortex membrane vesicles [28]. From this, they purified what is now known as p64 and identified it as a bovine chloride channel localized to apical and intracellular membranes [29, 31]. CLIC1 and CLIC4 were then discovered as homologues of bovine p64, however they differed structurally from p64 in that they did not possess an N-terminal extension like p64 [32-34].

There are 6 known CLIC paralogues in mammals, named CLIC1-6, and the family is defined by a roughly 230 conserved amino acid core sequence at the C-terminus and three conserved cysteine residues [35]. CLIC-like proteins are present in invertebrates and CLICs are highly conserved in vertebrates [36]. CLICs are not known to have a signal peptide, however they do possess a putative transmembrane domain

(PTM) near the N-terminus responsible for forming the pore of the membrane-bound channel [33, 37]. Sequence analysis, crystal structures, and electrophysiological studies on CLICs indicate that they form single-pass transmembrane proteins by polymerizing

[36]. CLIC4 has been observed as a homotrimer [38] while CLIC1 has been observed as a homodimer [39]. In addition to being redox-regulated, studies suggest that the dramatic transition from cytoplasmic globular state to membrane-bound channel structure of

CLICs may be influenced by pH [40-42].

CLICs have previously been shown to be involved in a variety of processes that include secretion, cell division, apoptosis, and cell mobility [43-47]. The CLIC family of proteins is widely expressed in multicellular organisms and auto-assemble to form intracellular ion channels that auto-insert into specific cellular membranes [36]. Unlike other mammalian ion channels, CLICs are present in cells as both soluble cytoplasmic 6 proteins and integral membrane proteins, and this transition appears to be redox-regulated

[37, 39, 48]. CLIC localization varies according to cell type and its localization appears to be specific to its function. In the case of CLIC4, its localization also appears to be dependent on its expression level [49].

The CLIC family of proteins has recently been implicated as having a role in this tubulogenic process. One study found that CLIC-like protein EXC-4 in C. elegans plays a critical role in both the proper development and maintenance of the C. elegans excretory canal [50]. Exc-4 mutants presented luminal cysts and other morphological defects at the surface of the canal. EXC-4 was also found to be continuously needed for normal tube size maintenance.

1.4 Evidence for CLIC function in angiogenesis

In both the cell hollowing and cord hollowing models for tubulogenesis, vesicular trafficking plays an integral role [13, 16, 18]. The origin and purpose of these vesicles remains under contention. Since the C. elegans excretory canal is thought to form a lumen by much the same mechanism as human capillaries [20], and CLICs are conserved between vertebrates and invertebrates, this suggests a role for CLICs in endothelial tubulogenesis.

Of the six known mammalian CLICs, only CLIC4 and CLIC1 are reported to be highly expressed in endothelial cells [51-57]. CLIC5 levels are found to be low in placental endothelial cells [52] and high in glomerular endothelial cells [57]. CLIC1,

CLIC4, and CLIC5 are also regulated by F-actin when reconstituted in planar lipid bilayers [43], suggesting a potential role in cytoskeletal reorganization. Both CLIC4 and 7

CLIC1 have been shown to form single-pass, dimeric channels that have poorly selective, redox-regulated chloride channel activity [39, 58]. Crystal structure and functional analyses indicate that CLIC4 and CLIC1 possess a putative transmembrane domain

(PTM) near the N-terminus [59].

Potential regulation of CLICs by redox signaling provides an interesting avenue by which CLICs may function in angiogenesis. Structural studies on CLIC1 have revealed two stable crystal structures: soluble and integral membrane forms [39]. The N- domain undergoes the most drastic structural transition, lending further credence to the

N-domain being the transmembrane domain. The two crystal structure states of CLIC1 observed were a reduced, soluble monomeric form, and an oxidized, non-covalent dimer.

As hypoxic or hyperoxic conditions produce oxidative stress, it would seem intuitive for redox signaling to be a regulator of angiogenesis. Research on the role of reactive oxygen species (ROS) in angiogenesis signaling has been an emerging field, but

ROS presence has been implicated in a number of vascular and cardiovascular diseases such as ischemia and hypertension [60, 61]. Studies have now shown that an increase in

ROS is often accompanied by induction of angiogenic regulators such as VEGF and Ang-

1 in endothelial cells [62, 63] while antioxidants inhibit angiogenic responses in endothelial cells [64, 65]. With the implication that CLIC translocation from the cytoplasm to membranes is regulated by redox signaling, and given that redox signaling mediates the angiogenic response, one could postulate that the oxidized, membrane- bound form of CLIC plays a role in hypoxia-induced angiogenesis.

Along these lines, studies have noted a correlation between pH and CLIC chloride conductivity [40, 42, 59, 66]. Specifically, channel activity increases as pH is reduced, 8 consistent with the proposed redox regulation of stable CLIC tertiary structures. Having sensitivity to pH could suggest another means by which CLICs may be regulated in functioning during tubulogenesis as cytoplasmic vesicles are often acidic. Alternatively,

CLICs may play a role in regulating vesicular acidification coupled with a proton pump

[36].

Little attention has been given to CLIC5 and CLIC1 as tubulogenic regulators thus far, but there is increasing evidence for a role for CLIC4 in lumen formation. CLIC4 has been found to be localized to vesicles in human umbilical vein endothelial cells [49] and large vacuoles in mouse heart endothelial cells [67]. Analyses of retinal vasculature in developing mice and adult mice challenged by an oxygen toxicity assay, reveals stunted vascular development in Clic4 -/- mice [67]. CLIC4 may also influence vacuolar formation in endothelial cells by regulating pH [67]. Furthermore, CLIC4 localizes to the midbody and centrosome of cultured bovine aortic endothelial cells and is enriched at cell-cell junctions in choriocarcinoma cells [68], suggesting a role in establishing or maintaining cell polarization. CLIC4 is also RhoA-regulated in a variety of cells [46].

Upon G13-coupled Rho-A activating receptor stimulation, CLIC4 translocates to the plasma membrane but does not modulate chloride currents, implying a hitherto unknown function of CLIC4 in the cell [46].

While not necessarily a regulator of tubulogenesis, CLIC1 may also play a role in regulating angiogenesis. The same C. elegans studies on exc-4 found that human Clic1 fused to the putative transmembrane domain of exc4 was able to rescue the cystic disruption phenotype of the exc4 mutant excretory canal [37]. CLIC1 may also affect apical membrane recycling as suggested by its localization to the apical domain in several 9 columnar epithelia [54]. In addition, both CLIC1 and CLIC4 have been suggested to interact with the actin cytoskeleton [43], implicating them in organelle trafficking.

Together, the data described above suggest that CLIC4 may function in physiological and pathological angiogenesis, specifically during tubulogenesis. As

CLIC4 has been more extensively characterized, we have chosen to focus mainly on defining a role for CLIC4 in angiogenesis. As we acknowledge the possibility that CLIC1 may functionally overlap with CLIC4 and note that only CLIC4 and CLIC1 are strongly expressed in endothelial cells, we also focus on CLIC1 as a potential angiogenic regulator.

We tested our hypothesis that CLIC4 and CLIC1 contribute singly and in concert to tubulogenesis and angiogenesis. First, we analyzed endothelial cell behavior in a series of in vitro assays that mimic individual steps of angiogenesis when CLIC4 or CLIC1 expression levels are altered. Second, we generated Clic4 knockout mice and acquired

Clic1 knockout mice for analyses. Third, we analyzed tumor growth on Clic4 mutant mice. Finally, we generated Clic4;Clic1 double knockout mice.

In chapters 2 and 4, we describe how CLIC4 and CLIC1 were determined to be effectors of endothelial cell growth, survival, network formation, and capillary sprouting in vitro . We also determined that only CLIC4 affects lumen formation in vitro while only

CLIC1 affects endothelial cell migration. Chapter 3 describes our analysis of the Clic4 single knockout mice. We show that loss of Clic4 results in postnatal underrepresentation and subtly delayed vascular development in the retina. Surviving Clic4 knockout mice are viable and fertile. In chapter 3, we describe our finding that tumors implanted on

Clic4 knockout mice tend to be larger by weight than controls with more metastases to the lungs. Chapter 5 describes our finding that Clic4 and Clic1 are concurrently essential 10 for embryonic development as evidenced by embryonic lethality of Clic4;Clic1 double mutants, which is not presented by Clic1 single mutants while Clic4 single mutants present embryonic lethality with incomplete penetrance. In addition, we show that

Clic4;Clic1 double mutants exhibit an angiogenic phenotype at embryonic time of death, suggesting that Clic4 and Clic1 overlap functionally in regulating angiogenesis. Finally in chapter 6, we discuss possible mechanisms of CLIC4 and CLIC1 function. 11

1.5 Figures

Figure 1.1

Figure 1.1 | Cell hollowing and cord hollowing models for tubulogenic mechanisms.

In this depiction, green shading represents the basal domain of the cell while red shading represents the apical domain. (Center panel) Representation of the process of pinocytosis.

(Left panel) Cell hollowing begins with a cord of endothelial cells undergoing

pinocytosis. The resultant pinocytic vesicles then fuse intracellularly to form vacuoles,

which further fuse. Apical markers on the vacuolar membrane target the vacuoles to the

site of the future lumen while the vacuoles continue to fuse, creating a membrane bound

lumen. The lumen of this “hollowed cell” then fuses intercellularly with surrounding

hollowed cells by exocytosis of vacuoles to form a continuous lumen. A cross-section of

a vessel formed in such a manner would reveal a vessel without cell junctions, or a

“seamless vessel.” (Right panel) During invasion by an endothelial cord into the

surrounding extracellular matrix, the endothelial cells lose their apicobasal polarity in the

cord-hollowing model. An external cue then triggers repolarization and membranes

contacting the extracellular matrix accumulate basal markers while pinocytic vesicles

carrying apical markers fuse at the site of the future lumen. Vesicles continue to fuse at

the apical domain creating “pockets of lumen” that eventually merge to become a

continuous lumen.

Chapter Two: CLIC4 is involved in endothelial cell proliferation and

morphogenesis in vitro

Based on previously reported data implicating CLIC4 in tubulogenesis (section

1.4), we profiled human umbilical vein endothelial cells (HUVEC) for CLIC expression by semi-quantitative PCR [69]. We found CLIC4 and 1 to be the most strongly expressed of the 6 human CLICs analyzed (Supp Fig 2.1). We initially sought to determine which steps of angiogenesis CLIC4 expression affected by using a series of in vitro assays that mimic individual steps of angiogenesis. This chapter contains the manuscript I authored as published in Angiogenesis (Feb 2009) as well as additional results and discussion. I

performed all data gathering and analysis for the material in this chapter with technical

support and guidance from Martin Nakatsu, Christopher Hughes, and Christine Yoon.

Claire Vech Reeves and Joseph Dufraine contributed ideas and guidance early in the

project. Dazhi Xiong and Robert Kass performed patch-clamp assessment for chloride

conductance. Oliver Hobert, Mark Berryman, and Jan Kitajewski participated in

analyzing results, authoring, and revising the manuscript. In addition, Mark Berryman

provided an antibody used in analyses of this chapter.

2.1 Abstract

New capillaries are formed through angiogenesis and an integral step in this

process is endothelial tubulogenesis. The molecular mechanisms driving tube formation

during angiogenesis are not yet delineated. Recently, the chloride intracellular channel 4

(CLIC4)-orthologue EXC-4 was found to be necessary for proper development and 14 maintenance of the Caenorhabditis elegans excretory canal, implicating CLIC4 as a regulator of tubulogenesis. Here, we studied the role of CLIC4 in angiogenesis and endothelial tubulogenesis. We report the effects of inhibiting or inducing CLIC4 expression on distinct aspects of endothelial cell behavior in vitro . Our experiments utilized RNA interference to establish cultured human endothelial cell lines with significant reduction of CLIC4 expression, and a CLIC4-expressing lentiviral plasmid was used to establish CLIC4 overexpression in endothelial cells. We observed no effect on cell migration and a modest effect on cell survival. Reduced CLIC4 expression decreased cell proliferation, capillary network formation, capillary-like sprouting, and lumen formation. This suggests that normal endogenous CLIC4 expression is required for angiogenesis and tubulogenesis. Accordingly, increased CLIC4 expression promoted proliferation, network formation, capillary-like sprouting, and lumen formation. We conclude that CLIC4 functions to promote endothelial cell proliferation and to regulate endothelial morphogenesis and is thus involved in multiple steps of in vitro angiogenesis.

2.2 Introduction

Angiogenesis entails the formation of new blood vessels from the sprouting, branching, and pruning of pre-existing vessels and is essential for embryonic development as well as tissue growth and repair. Deviations in the regulation of angiogenesis are important factors in pathological processes such as atherosclerosis, diabetic retinopathy, and malignant tumor growth. The basic cellular mechanisms of angiogenesis begin with pro-angiogenic factors stimulating endothelial cells to degrade the local basement membrane surrounding an existing vessel followed by endothelial cell 15 rearrangement, proliferation, and migration into the surrounding stroma [4]. A neo-vessel is generated when new sprouts form lumens in a process termed tubulogenesis.

Anastomosis of tubes occurs after the formation of lumens and involves the branching and reconnecting of vessels to form networks. Pruning of elementary sprouts follows and the basement membrane is reformed. Finally, accessory cells such as pericytes or smooth muscle cells are recruited to help stabilize the vessel and provide control of luminal diameter.

In endothelial biology, the critical underlying process of endothelial tube formation and tube maintenance is still poorly understood. There are several current models for mammalian vascular tubulogenesis. One model involves a chain of endothelial cells forming vesicles that enlarge, merge, and fuse intra- and intercellularly to form contiguous lumens [20]. This process involves the acquisition of cell apicobasal polarity, cytoskeletal reorganization, assembly of intercellular junctions, and membrane domain specification [9]. The formation of the Caenorhabditis elegans excretory canal provides an example of this process [50].

Recently, members of the glutathione S-transferase-related chloride intracellular channel (CLIC) protein family have been implicated in tubulogenesis. The CLIC family, composed of six family members in mammals, is defined by a roughly 230 conserved amino acid core sequence at the C-terminus [54]. CLIC family proteins are widely expressed in multicellular organisms and are involved in a variety of processes including tube formation, secretion, cell division, apoptosis, and cell motility [35]. Several CLIC family proteins have been shown to form intracellular ion channels that auto-assemble and auto-insert into specific cellular membranes [35]. In addition to their association with 16 membranes, CLIC proteins can also exist as soluble cytosolic and nuclear proteins, suggesting that CLICs may have alternate, non-channel functions [42]. Currently, only

CLIC1 and CLIC4 are reported to be expressed in endothelial cells [49, 52, 54].

Among CLIC proteins, CLIC4 has been studied most extensively. CLIC4 has

been shown to be associated with cytoskeletal proteins such as dynamin I, actin, and

tubulin suggesting a role in regulating cellular morphology, membrane trafficking, and

cell signaling [68, 70]. A recent study found that CLIC4-orthologue EXC-4 in C. elegans

plays a critical role in both the proper development and maintenance of the C. elegans

excretory canal [50]. Luminal cysts and other morphological defects were observed at the

surface of the canal in exc-4 mutants, providing the first evidence that CLICs function in tubulogenesis. It was also demonstrated that EXC-4 was continuously needed for normal tube size maintenance [50].

This study investigates human CLIC4 as a regulator of angiogenic processes by using in vitro techniques that mimic multiple distinct steps of angiogenesis. Based on the

EXC-4 precedent, we hypothesized that reducing CLIC4 expression in endothelial cells

would negatively affect angiogenesis, while increasing CLIC4 expression would promote

endothelial angiogenesis. Human primary endothelial cells were used to represent their in

vivo counterpart and are suggested to be better suited for in vitro studies than

immortalized endothelial lines based upon assessment of gene expression patterns [71].

We indeed found that reduced CLIC4 expression levels in endothelial cells inhibited

endothelial network formation and cell proliferation. CLIC4 knockdown affected

capillary-like network formation and lumen formation. Reduced CLIC4 expression had

no effect on cell migration but led to a modest effect on cell survival. CLIC4 17 overexpression promoted proliferation, network formation, capillary-like sprouting, and lumen formation in vitro . Together, these results indicate that CLIC4 expression is required for multiple steps of angiogenesis.

2.3 Altered CLIC4 expression does not affect endothelial cell migration

To assess the function of CLIC4 during individual steps of angiogenesis in vitro , we generated HUVEC lines stably infected with lentivirus-based vectors to reduce or increase expression of CLIC4. CLIC4 knockdown and overexpression were confirmed at the protein level by immunoblotting (Fig 2.1a, b). Immunoblotting for CLIC1 showed that Clic4 shRNA did not target Clic1 , which is also expressed in endothelial cells, indicating that the shRNA treatments were specific for Clic4 (Fig 2.1a).

Established knockdown cell lines were referred to as shRNA3 and shRNA5 with plko sc (pLKO.1-puro vector containing scrambled shRNA insert) as the control. The

CLIC4 overexpression cell line was named CLIC4 with pccl (pCCL.pkg.wpre) containing GFP as its respective control. The cellular morphology of knockdown and overexpression HUVEC lines was qualitatively assessed. For this analysis, equivalent numbers of cells from each of the lines were seeded and cell cultures were photographed

48 h later, when cells were still subconfluent (Fig 2.1c, d). We detected no major changes in cellular morphology with the exception of HUVEC shRNA5, which appeared more rounded relative to control cells (Fig 2.1c). We noted that both HUVEC shRNA3 and

HUVEC shRNA5 lines appeared sparser than control or overexpressing cells, indicating that CLIC4 knockdown may reduce cell growth. 18

Previous studies have documented that CLIC4 is associated with cytoskeletal proteins in a variety of cells [43, 68, 70] and that CLIC4 expression levels affect cell motility in some cell types [72]. To determine whether endothelial cell migration is affected by reduced or increased CLIC4 expression, we utilized a migration assay to assess effects of CLIC4 on directed cell migration. For this assay, we scraped a line through confluent monolayers of the HUVEC CLIC4 knockdown and overexpression cultures grown on type I collagen. Endothelial cell migration into the cell-free area was then monitored at various time points over a 12 h period. CLIC4 knockdown or overexpression resulted in no observable effect on the migration rates of HUVECs when compared with control lines. Representative data are shown in Figure 2.2.

2.4 CLIC4 expression promotes endothelial cell proliferation

Chloride intracellular channel 4 has been documented to play a role in cell proliferation of squamous cancer cell lines and to affect survival of keratinocytes, squamous cancer cell lines, and human osteosarcoma lines [47, 51]. To explore the role of CLIC4 in endothelial cell proliferation and survival, we utilized WST-8-based survival and proliferation assays on our CLIC4 knockdown and overexpressing cell lines.

Proliferation assays were performed by scoring cells after seeding equal numbers of

HUVECs on type I collagen-coated plates and culturing in serum free medium (SFM) supplemented with epidermal growth factor (EGF) and vascular endothelial growth factor-A (VEGF-A) for 96 h. We found that reduction of CLIC4 expression led to significant inhibition of endothelial cell proliferation (Fig 2.3a), while overexpression of

CLIC4 resulted in a significant promotion of endothelial cell proliferation (Fig 2.3b).

These results indicate that expression of CLIC4 promotes endothelial cell proliferation. 19

For survival assays, equivalent numbers of knockdown, overexpression, or control

HUVEC were seeded on type I collagen-coated plates and cultured for 24 or 48 h in SFM alone or SFM supplemented with the survival signal EGF. We found that in the absence of EGF, CLIC4 knockdown had no significant effect on endothelial cell survival at 24 h when compared with control lines (Fig 2.4). At 48 h, knockdown lines display a modest but significant increase in cell survival (Fig 2.4a). CLIC4 overexpression significantly decreased endothelial cell survival at 48 h, but not at 24 h (Fig 2.4b). Thus, in contrast to the pronounced results for proliferation, CLIC4 expression plays a limited role in decreasing endothelial cell survival.

2.5 CLIC4 promotes endothelial capillary-like network formation

Since CLIC4 has been suggested to function in tubulogenesis, we assessed the effect of CLIC4 expression on endothelial cell morphogenesis by determining the ability of the various CLIC4 knockdown or overexpressing cell lines to organize into capillary- like networks. To assess capillary-like network formation, HUVEC were seeded between two porcine collagen gel layers and cultured in SFM containing VEGF-A and EGF for 96 h. As shown in Figure 2.5a, CLIC4 knockdown resulted in a dramatic reduction in cord formation, network formation, and branching, whereas overexpression resulted in a modest but significant increase in capillary-like network formation. Quantification of surface area occupied by network structures confirmed that CLIC4 knockdown results in a significant decrease in surface area occupation by endothelial cells: HUVEC shRNA3 and HUVEC shRNA5 covered only 72.7 and 70.2% of the surface area relative to

HUVEC plko sc set at 100% (Fig 2.5b). By comparison, HUVEC CLIC4 overexpression 20 covered 112.4% of the surface area relative to HUVEC pccl set at 100%, indicating a significant increase in endothelial surface area occupation by endothelial networks.

Results from quantifying the number of branch points per field of view also indicate that CLIC4 expression affects endothelial network formation. One branch point was considered any focal intersection of two or more cords. Knockdown cell lines exhibited a significant reduction in the number of branch points per field of vision when compared with plko sc control (Fig 2.5c). Consistently, a significant increase in branch point numbers was found in CLIC4-overexpressing lines compared with control pccl (Fig

2.5d). These results indicate that CLIC4 expression promotes endothelial network formation and branching.

It has been reported that the network formation assay we utilize here is dependent to some extent on cell proliferation [73]. Thus, we included an additional control group to help distinguish potential effects of CLIC4 knockdown on proliferation versus network formation. This control group consisted of plko sc cells treated with mitomycin C (mmc;

25 µg/ml for 45 min), which blocks cell division. We found that CLIC4 knockdown

results in significantly more severe inhibition of network formation when compared with

mitomycin C-treated control cells (Fig 2.6a). The inhibitory effect of mitomycin C on cell

proliferation was confirmed using the WST-8 proliferation assay done with endothelial

cells grown as a monolayer for 120 h, the duration of the network formation assay (Fig

2.6b). In fact, mitomycin C-treated control HUVEC showed significantly greater

inhibition of proliferation than HUVEC shRNA3.

Quantification of network surface area and number of branch points (Fig 2.6c, d)

confirmed that CLIC4 knockdown cells formed a less extensive capillary-like network 21 than mitomycin C-treated control cells. HUVEC network surface area occupation analysis indicates that HUVEC shRNA3 and shRNA5 covered only 74.1 and 63.9% of the surface area occupied by mitomycin C-treated control, respectively. Branch point number analysis at 5× objective also indicates that CLIC4 knockdown lines significantly inhibit network branching when compared with mitomycin C-treated control. We conclude that reducing CLIC4 expression has effects on in vitro endothelial network formation independent from the effects of reduced CLIC4 expression on proliferation.

2.6 Altered CLIC4 expression affects capillary-like sprouting and lumen

formation

Since the network structures in the previous assay are formed from cords without lumens, we assessed the effect of CLIC4 expression on in vitro capillary-like tube formation in a fibrin bead assay. HUVEC were adhered to dextran-coated Cytodex-3 beads and embedded in a fibrin clot. D551 fibroblasts were then seeded and cultured in

EGM-2 medium for 10 days as a monolayer on top of the fibrin clot to provide secreted factors [74]. Photographic documentation occurred on or between days 3 and 11, with sprout formation assessment beginning on day 3; sprout extension, branching, and lumen formation assessed from days 4 to 11; and anastomosis assessed from days 7 to 11. The sprouts resulting from this assay were multicellular, lumen-containing processes, consistent with previously published results [75].

Chloride intracellular channel 4 knockdown HUVEC lines exhibited stunted sprouting and formed fewer sprouts (Fig 2.7a, b). Sprouts that formed from CLIC4 knockdown HUVECs showed undersized sprouting, indicating a defect in sprout 22 elongation into the surrounding environment. Although HUVEC shRNA3 cells formed only shortened sprouts, these sprouts still contained lumen. In contrast, HUVEC shRNA5 cells, which have less CLIC4 expression than HUVEC shRNA3, showed a reduction in lumen-containing structures. This was consistent with results from a separate but similar assay in which HUVEC were embedded in a thick collagen layer and allowed to undergo morphogenesis (Supp Fig 2.2) [76].

Quantification of HUVEC shRNA5 results revealed that this lumen-forming defect was significant when compared with the number of lumen-containing sprouts control cells formed (Fig 2.7c). Disjointed lumen formation was also observed within

CLIC4 knockdown cell line shRNA5 (Fig 2.7a, b). CLIC4 overexpression resulted in increased and more rapid sprouting when compared with pccl control cells (Fig 2.7d).

Quantification of these results indicated that CLIC4 overexpression led to an increase in lumen-containing sprouts (Fig 2.7e). Together, these data demonstrate the necessity for appropriate CLIC4 expression to allow for in vitro capillary sprouting. Disrupting CLIC4 function was found to impair HUVEC morphogenesis and lumen formation.

2.7 Discussion

Chloride channels are known to function in several cellular processes such as maintaining membrane potentials, controlling cell volume, and regulating pH. To date, at least five independent chloride channel families have been documented in mammals [51].

To emphasize the importance of intracellular chloride regulation further, defects of intracellular chloride channels are associated with numerous diseases including those that affect the human neuromuscular, renal, skeletal, and respiratory systems [77]. As the 23 most recently discovered intracellular chloride channel family, our knowledge of CLIC functions is still evolving, and this study sought to determine if intracellular chloride channels also play a role in angiogenic processes and pathologies.

Currently, there are six identified mammalian CLICs defined by a C-terminal sequence of roughly 230 amino acids that display GST-like structure and function [54].

CLICs are highly homologous to each other and are conserved between vertebrate and invertebrate species [37]. Unlike other mammalian ion channels, CLICs are present in cells as both soluble cytoplasmic proteins and integral membrane proteins [54]. CLIC proteins have been shown to localize to intracellular membranes such as those of the

Golgi, endoplasmic reticulum, mitochondria, and other membrane types [37, 50, 51].

CLIC4 localization can vary in different cell types and this seems to be specific to its function [51]. For example, in pro-apoptotic keratinocytes, CLIC4 translocates from the mitochondria to the nucleus, with the level of translocation being a factor in stress response regulation [45, 78]. Overexpression of CLIC4 in cells was also shown to promote membrane localization where CLIC4 may exhibit anion channel activity [79].

In general, the activities of CLIC proteins are not well-documented and their precise functions in cellular processes are poorly understood. Even less is known about

CLIC4 function in endothelial cells. To date, CLIC1 and CLIC4 are the only members of the CLIC family confirmed to be expressed in endothelial cells [49, 52, 54]. However, recent research identified a C. elegans homologue of CLIC4, EXC-4, as having tubulogenic functions, indicating a potential role for CLIC4 in endothelial angiogenesis

[50]. Another study found CLIC4 to be among 120 proteins affected during VEGF-A driven endothelial tubulogenesis [49]. To support the notion of a CLIC4 role in 24 tubulogenesis, localization of CLIC4 on tubulogenic vesicles should be tracked at endogenous, knockdown, and overexpressing levels during angiogenesis either in vitro or ex vivo in a developing mouse retina.

We hypothesized that reduced CLIC4 expression in primary endothelial cells would negatively affect angiogenic processes, while increasing CLIC4 expression would lead to promoted endothelial angiogenesis. To test this hypothesis, RNAi were used to generate stable cultured primary HUVEC lines exhibiting CLIC4 knockdown expression.

CLIC4-overexpressing HUVEC lines were also created using a lentiviral construct expressing full-length CLIC4. To analyze the effects of CLIC4 expression on various steps of in vitro angiogenesis, we utilized in vitro assays assessing endothelial cell survival, proliferation, migration, network formation, tube formation, and capillary sprouting. We present here a comprehensive analysis of the potential angiogenic functions of CLIC4 in vitro .

We found that CLIC4 is involved in several steps of in vitro angiogenesis after observing aberrant network formation, capillary sprouting, and lumen formation due to altered CLIC4 expression (Figs 2.5, 2.6 and 2.7). We also found that reduced CLIC4 expression has a modest effect on endothelial cell survival (Fig 2.4) and a significant effect on cell proliferation (Fig 2.3). With these results, we conclude that CLIC4 expression is required for proper endothelial cell proliferation and morphogenesis in vitro .

Thus, we have defined a novel angiogenic signaling regulator and present CLIC4 as a potential target for anti-angiogenic therapies.

Using transient transfection of human telomerase-immortalized microvascular endothelial (TIME) cells with CLIC4 antisense constructs, a previous study [49] 25 documented decreased in vitro angiogenesis with an indication of reduced tube-like structures. In contrast to this report, our study documented several steps in angiogenesis dependent on proper CLIC4 expression, including cellular proliferation, endothelial network formation, and lumen formation, providing a deeper understanding of the role of

CLIC4 in angiogenesis. This is the first study on CLIC4 function in primary endothelial cells (HUVEC), demonstrating a role for CLIC4 in endothelial proliferation and branching morphogenesis.

The mechanism by which CLIC4 functions to regulate in vitro angiogenesis is still unclear. A current model for endothelial tubulogenesis is based on vesicular fusion and cord hollowing [9, 20]. This process is known to involve cell apicobasal polarity [9,

20], but the factors and molecular mechanisms that regulate this process are still not well understood. CLIC4 has previously been localized to intracellular vesicles in proliferating endothelial cells [49]. In C. elegans , EXC-4 is reported to localize to the apical luminal membrane of the excretory cell [50]. It is possible that the chloride ion channel function of CLIC4 plays a role in regulating vacuole formation, enlargement, and fusion since chloride channels are known to regulate water transport between a cell and its environment. As an ion channel, another plausible mechanism for CLIC4 involvement in angiogenesis would involve regulating electric potentials generated by the acidification required for vesicular fusion [80]. The resolution of the precise mechanisms of CLIC4 function in angiogenesis is an important future goal to understand this novel angiogenic regulator further.

2.8 Figures

Figure 2.1

Figure 2.1 | Establishing CLIC4 knockdown and overexpression in endothelial cells.

(a) Immunoblotting for CLIC4 confirms knockdown of CLIC4 protein expression in shRNA3 and shRNA5 endothelial cell lines compared with plko sc control line. CLIC1 immunoblotting confirms that knockdown is specific to CLIC4 and CLIC1 expression is not affected. (b) Immunoblotting for CLIC4 confirms CLIC4 overexpression in the

CLIC4 endothelial cell line compared with pccl control. Alpha-tubulin serves as a loading control. (c) HUVEC shRNA3 knockdown cells exhibit no major change in endothelial cell morphology while HUVEC shRNA5 appear to be more rounded when compared with plko sc. HUVEC refer to endothelial cells that have not been infected with any constructs. (d) No major change was observed in endothelial cell morphology between pccl control cells and CLIC4-overexpressing cells

Figure 2.2

Figure 2.2 | CLIC4 expression has no effect on endothelial cell migration. (a) CLIC4 knockdown cell lines shRNA3 and shRNA5 exhibited no noticeable change in directed endothelial cell migration 6 h after scraping in a migration assay when compared with control plko sc. (b) Likewise, CLIC4-overexpressing cells exhibited no noticeable change in endothelial cell migration 6 h after scraping when compared with control pccl.

Figure 2.3

Figure 2.3 | CLIC4 expression promotes endothelial cell proliferation. (a) Inhibition of CLIC4 expression by shRNA resulted in decreased endothelial cell proliferation when compared with plko sc control. (b) CLIC4 overexpression resulted in a significant increase in endothelial cell proliferation when compared with control pccl. (*p < 0.05;

**p < 0.01).

Figure 2.4

Figure 2.4 | Altering CLIC4 expression affects endothelial cell survival. (a) CLIC4 knockdown lines significantly increased cell survival at 48 h, but not 24 h post-seeding.

(b) CLIC4 overexpression resulted in significantly decreased endothelial cell survival at

48 h, but not 24 h post-seeding. (*p < 0.05; *p < 0.01)

Figure 2.5

Figure 2.5 | CLIC4 expression promotes in vitro endothelial network formation. (a)

Microscopy of CLIC4 knockdown or overexpressing cells in a collagen network formation assay shows decreased network formation with CLIC4 knockdowns, while

CLIC4 overexpression results in increased network formation when compared with respective controls. (b) Quantification confirms that CLIC4 knockdown results in decreased network formation by showing that reduced CLIC4 expression leads to a significant reduction in surface area coverage by endothelial cells. CLIC4 overexpression leads to increased network formation with a significant increase in surface area occupied by endothelial cells. Control standards were set at 100% surface area occupation. (c)

Quantification of branch points for each knockdown cell line further confirms CLIC4 knockdown inhibition on endothelial network formation. Knockdown cell lines form significantly fewer branch points than control plko sc. Quantification was done by counting visible branch points per field of view at 10× objective. (d) CLIC4- overexpressing cells form significantly more branch points than control pccl.

Quantification was done by counting visible branch points per field of view at 5× objective. One branch point was considered any focal intersection of two or more cords.

(*p < 0.05; **p < 0.01)

Figure 2.6

Figure 2.6 | Reduced CLIC4 expression inhibits endothelial network formation independent of its proliferative defect. (a) Comparison of network formation by knockdown cell lines and control cells treated with mitomycin C indicates that there is inhibition of network formation independent from reduced CLIC4 inhibition of proliferation. (b) A proliferation assay confirms mitomycin C growth arrest of plko sc control cells. (c) Quantification of surface area occupied by endothelial cells confirms an additional defect associated with reduced CLIC4 expression independent of reduced

CLIC4 proliferation inhibition. Mitomycin C-treated control standards were set at 100% surface area occupation. (d) Quantification of branch points further confirms an additional defect associated with reduced CLIC4 expression separate from reduced

CLIC4 proliferation inhibition. (*p < 0.05 compared with plko sc; **p < 0.01 compared with plko sc; + p < 0.05 compared with plko sc + mmc; ++ p < 0.01 compared with plko sc

+ mmc).

Figure 2.7

Figure 2.7 | CLIC4 expression affects capillary-like sprouting in vitro . (a) Microscopy indicates that HUVEC shRNA3 have altered tube morphogenesis in that they tend to form shorter, more stunted tube structures as indicated by a red arrow. HUVEC shRNA5 show a decrease in the number of branches per sprout. HUVEC shRNA5 also exhibited stunted sprouts indicated by a red arrow. Unlike HUVEC shRNA3 cells, HUVEC shRNA5 cells did not form lumens. Black arrows indicate robust, lumen-containing sprouts. (b) Microscopy from a separate trial confirms the phenotypes of HUVEC shRNA3 stunted tube structures and HUVEC shRNA5 stunted sprouts without lumens. (c)

Quantification of the number of lumen-containing sprouts shows that HUVEC shRNA5 exhibits a significant decrease in the number of lumen-containing sprouts formed when compared with plko sc control. (d) Microscopy of CLIC4-overexpressing endothelial cells show increased sprouting and lumen formation when compared with plko sc control.

Lumen-containing sprouts are indicated by black arrows. (e) Quantification for CLIC4- overexpressing endothelial cells indicates a significant increase in the number of lumen- containing sprouts. (*p < 0.05; **p < 0.01)

2.9 Additional discussion

Two issues not addressed in our manuscript [56] are how CLIC4’s putative chloride channel activity and other possible functions of CLIC4 could affect angiogenesis.

To date, CLIC4 has only shown poorly selective anion channel activity in artificial bilayers [58] and has not shown chloride conductance in cultured endothelial cells [46].

Consistent with these observations, Dazhi Xiong, from our collaborator Robert Kass’s laboratory, found no channel conductance in HEK 293 cells transfected with CLIC4-GFP overexpression vector (Supp Fig 2.3).

Meanwhile, the GST-domain of CLIC4 is implicated in regulating translocation to the plasma membrane upon sphingosine-1-phosphate (S1P) or lysophosphatidic acid

(LPA) induction [46], suggesting a previously unstudied role for the GST-domain of

CLIC4. CLIC4 has also been implicated in immune responses to bacterial lipopolysaccharide induction, suggesting a role for CLIC4 in inflammation and macrophage function [81]. As S1P, LPA, and macrophages are well-documented mediators of angiogenesis [82-84], the role of CLIC4 in angiogenesis could certainly depend on one of these pathways instead of, or in addition to, its putative chloride channel activity. There are currently no reports addressing the potential roles of CLIC4 when it is resident in the cytoplasm. The varying roles of CLIC4 in differing cell types support the idea of multiple CLIC4 domains regulating its different functions.

2.10 Supplementary figures

Supplementary Figure 2.1

Supplementary Figure 2.1 | Clic1 and Clic4 are most highly expressed in cultured endothelial cells. HUVEC were screened for the expression of CLIC1-6 using specific primers through 30 PCR cycles in a semi-quantitative PCR screen. CLIC1 and CLIC4 show highest expression while CLIC2 and CLIC5 are expressed at extremely low levels.

Gene fragments cloned into a TA-cloning vector using the same screening primers were used as positive controls. 43

Supplementary Figure 2.2

Supplementary Figure 2.2 | CLIC4 expression affects endothelial morphogenesis in a three-dimensional extracellular matrix. HUVEC with CLIC4 knockdown exhibit reduced ability to form endothelial tube-like structures. Structures formed are tortuous and contain disjointed or absent lumens, indicated by red arrows in comparison to black arrows of control cells. HUVEC overexpressing CLIC4 are qualitatively comparable to control cells in terms of lumen formation (black arrows) and vessel morphology. 45

Supplementary Figure 2.3

Supplementary Figure 2.3 | CLIC4 expression does not affect ion conductance at the plasma membrane in HEK 293 cells. Patch clamping was used to assess ion channel conductance at the plasma membrane of single HEK 293 cells transfected to express either GFP (control, left panel) or CLIC4-GFP (CLI4 overexpressing, right panel).

Representative results from single cells are shown here. Currents shown here represent potassium currents known to be endogenous to HEK cells. No difference in currents were found between GFP-expressing control cells and CLIC4-overexpressing cells, suggesting that CLIC4 does not have any ion channel activity in HEK 293 cells at the plasma membrane in single cell suspension.

Chapter Three: Generation and characterization of Clic4 knockout mice

To assess the physiological relevance of our in vitro findings, we designed and generated Clic4 knockout mice for in vivo analyses. This chapter details the generation and resultant examination of our Clic4 knockout mice. I performed all data gathering and analysis for the material in this chapter. Thomas Ludwig provided assistance in the initial design of the knockout mouse targeting construct and Southern blot genotyping strategy.

Victor Lin performed electroporation of the targeting vector into ES cells and microinjection into blastocysts. Thaned Kangsamaskin and Ian Tattersall provided technical assistance with tumor implantation studies. Valeriya Borisenko performed sectioning on embedded tumors. Irina Jilishitz assisted in performing histology on tumor sections. Both Thomas Ludwig and Jan Kitajewski participated in analyzing results.

3.1 Abstract

CLIC4 belongs to the CLIC family of proteins, which exist as both cytoplasm- soluble and membrane-bound proteins in a wide variety of tissues. To examine the biological role of CLIC4 during vascular development, we engineered conditional Clic4 knockout mice. Germline deletion of Clic4 was achieved by Cre-mediated recombination of the conditional allele. Mice heterozygous for Clic4 are viable, fertile, and appear normal. Clic4 homozygous knockout mice are fertile and viable, however they are underrepresented and have a subtle angiogenic defect in the developing retina characterized by less dense retinal vasculature. Preliminary results also reveal a 48 macrophage recruitment defect and enhanced tumorigenesis in Clic4 homozygous mutant mice.

3.2 Introduction

The chloride intracellular channel (CLIC) family of proteins comprises CLIC1-6 in mammals. CLICs are highly conserved in vertebrates and invertebrates [42], and their presence as both soluble and membrane-bound channels makes them unique as channel proteins [54]. Thus far, chloride channel activity has been shown for CLIC1, 2, 4, and 5

[32, 33, 43, 66].

The subcellular localization of CLIC4 seems to be context-dependent, suggesting that CLIC4 may play roles in a variety of biological processes. CLIC4 has been localized to the cytoplasm, nucleus, mitochondria, at cell junctions, centrosomes, on intracellular vesicles, the plasma membrane, the nuclear membrane, and the mitochondrial membrane in various cell types [45, 49, 85, 86]. CLIC4 is reported to bind brain dynamin I directly and interact with several cytoskeletal elements such as actin, ezrin, and tubulin [70, 87].

CLIC4 is also shown to be VEGF- [49], F-actin- [43], S1P- [46], and p53-regulated [45,

88], suggesting roles for CLIC4 in such diverse processes as angiogenesis, maintaining cell polarity, migration, and cell cycle regulation. In addition, CLIC4 is found in all tissues that have been analyzed, indicating ubiquitous expression [70, 78]. Furthermore, the putative channel activity of CLIC4 points to a function in maintaining membrane polarity or regulating cell/vesicle volume [80].

The structure of CLIC4 suggests that it is a single-pass transmembrane protein with the transmembrane domain near the N-terminus [33, 58]. A conformational change 49 from its globular cytoplasmic state to its transmembrane state is thought to be induced with redox reactions [59]. The channel activity of CLIC4 is still being assessed, though it is thought to form poorly selective anion channels with activity dependent upon pH [58].

As a soluble monomer, CLIC4 is structurally similar to CLIC1 [59], but CLIC4 has been found in its soluble state as a homotrimer, which imparts an organization distinct from

CLIC1 [38].

CLIC4 is a 253 amino acid putative ion channel that is highly conserved across a range of species. Only two amino acid substitutions separate mouse CLIC4 from human

CLIC4 [89]. In humans, Clic4 is located at 1p36.11 while it is at chromosome 4 D3; 4 in mice. Mouse Clic4 contains five introns and six exons with the coding sequence beginning 225 base pairs into exon 1.

The functional roles of CLIC4 remain unclear. To investigate the role of CLIC4 in vivo , we disrupted Clic4 in mice through the generation of conditional Clic4 knockout mice using homologous recombination in embryonic stem cells. Preliminary analyses indicate that Clic4 is not required for developmental angiogenesis or normal development, however loss of Clic4 results in decreased angiogenic activity and enhanced tumor development.

3.3 Clic4 targeting scheme

As the putative transmembrane domain (PTM) of Clic4 is thought to be invaluable to its proper localization and function [37, 50], we targeted exon 2 for conditional deletion. Exon 2 encodes the pore-lining α-helix and second β-sheet of the

PTM and is 110 base pairs [37, 89]. Upon Cre-mediated recombination, the elimination 50 of exon 2 would result in deletion of a majority of the PTM and a frameshift mutation. If a resulting splice variant were generated and stable, it would encode a roughly 10 kDa protein comprising 24 amino acids, encoded by exon 1 and 28 amino acids encoded by the disrupted exon 3 before the introduction of a stop codon. This protein would have no similarity to endogenously produced CLIC4.

The targeting vector utilizes loxP sites, one each placed in introns upstream and downstream of exon 2 and in the same orientation (Fig 3.1a). A flrted (flanked by FRT) neomycin cassette was placed within intron 1 for positive selection of ES cell clones. The herpes simplex virus thymidine kinase (HSV-tk) gene cassette was used for negative selection in the targeting vector. This conditional targeting vector was termed pClic4floxed and constructed from a BAC RP23 genomic DNA clone.

pClic4floxed was electroporated into CSL3 ES cells derived from

129S6/SvEvTac-Car3 (designated 129) background mice, and 192 neomycin-resistant clones were isolated. Southern blotting and PCR were used to identify clones that had undergone homologous recombination and carried the correctly targeted Clic4 , termed the Clic4 fl allele. Southern blot analysis of PstI-digested genomic DNA with a 5’ 32 P- labeled probe external to the targeting vector was used to confirm proper integration of the upstream loxP site (Fig 3.1b) while PCR confirmed proper integration of the downstream loxP site (Fig 3.1c). Of the 192 isolated clones, 11 ES cell clones were found to be properly targeted with both loxP elements. Two of these ES cell clones were microinjected into C57BL/6 (designated B6) background host blastocysts for generation of chimeric mice. Germline transmission was established first by observing agouti coat color and then through genotyping. Subsequent genotyping for Clic4 fl was performed by 51

PCR using the same genotyping scheme to confirm proper integration of the downstream loxP site.

Once Clic4 fl chimeric mice were generated, mouse embryonic fibroblasts (MEFs)

were isolated from mice heterozygous for the Clic4 fl allele and transfected with

adenovirus expressing either GFP or Cre recombinase to ensure loxP recombination

events were occurring. Screening for recombination events was done by Southern blot

analysis using Pst1-digested genomic DNA isolated from the MEFs (Fig 3.1d). After

confirming Cre recombination, Clic4 fl heterozygotes were intercrossed to produce

Clic4 f/fll homozygous mice in a mixed B6;129 hybrid background.

3.4 Clic4 knockout mice are underrepresented, but viable and fertile

For analysis of the Clic4 null allele, we crossed F1 generation B6;129 Clic4 fl/fl

homozygous male mice with female mice expressing EIIa-Cre transgenic mice, allowing

for global removal of Clic4 . The Ella adenoviral promoter drives Cre-recombinase

expression in oocytes and preimplantation embryos [90]. All subsequent genotyping for

Clic4 knockout mice occurred by PCR analysis as schematized in Figure 3.1a. Clic4 +/-

were crossed with pure B6 wild-type mice for 8 generations to generate Clic4 +/- mice in a pure B6 background. The data reported in this chapter were obtained from studies on these Clic4 knockout mice considered to be in a pure B6 genetic background.

Clic4 +/- mice were viable, healthy, and fertile; however crosses between two

Clic4 +/- mice produced Clic4 -/- in skewed Mendelian ratios (Table 3.1a). Of the 25% mice

from a pool of 78 resultant offspring expected to be Clic4 -/-, only 14 were Clic4 -/-

indicating that Clic4 -/- mice are born 28.2% less than Mendelian distribution predicts as 52 analyzed by a χ2 test (p < 0.03). Postnatal morbidity is unlikely to account for this

discrepancy as all pups born were genotyped. Furthermore, genotyping embryos at 9.5

days post coitus (dpc) indicates that Mendelian distribution is restored (p > 0.05) (Table

3.1b). PCR analysis and Western blotting of lung tissue isolated from Clic4 homozygous

knockout mice confirmed corresponding loss of CLIC4 (Fig 3.2).

Whole mount staining of 9.5 dpc embryos with endothelial marker endomucin

showed no overt differences between vasculature of Clic4 -/- mice and wild-type littermates suggesting that vasculogenesis and angiogenesis are not affected by deletion of Clic4 (Fig 3.3). Surviving Clic4 -/- mice are viable, fertile, and did not exhibit gross physical phenotypes apart from a mild runting, consistent with reports of adult Clic4 -/- mice being smaller in size by weight [67]. Nonetheless, this underrepresentation of Clic4 -

/- mice suggests that CLIC4 is required for embryonic viability in a small percentage of embryos. Thus, CLIC4 plays a role in development, but only minimally or rarely.

3.5 Clic4 knockout mice display subtle defects in postnatal angiogenesis and may

have reduced macrophage content in the developing retina

Although surviving Clic4 -/- mice appeared to develop normally, we wanted to

evaluate the effect of Clic4 loss on postnatal angiogenesis, given its established role in in

vitro angiogenesis as described in Chapter 2. To study postnatal angiogenesis in our

Clic4 -/- mice, we dissected retinas at specific stages of vascular development and

performed FITC-conjugated isolectin staining on retina whole mounts to view the

vasculature. The mouse retina is avascular at birth, and angiogenesis in the mouse retina

is well-characterized in defined stages [91, 92]. Briefly, from birth to postnatal day 7 (P7), 53 angiogenesis from the central retinal artery forms a capillary network that extends through the retina to the periphery, termed the primary plexus or superficial layer. From

P7 to P21, angiogenesis occurs first perpendicular to the primary plexus and then parallel to create the intermediate and deep vascular plexuses. While vasculature, marked by endothelium-specific marker IB4, appeared well-organized in P5 and P9 retinas isolated from Clic4 -/- mice, vascular density appeared reduced at all levels analyzed (Fig 3.4a), consistent with published data [67]. Vasculature in the deep vascular plexus of Clic4 -/- mice also presents robust vascular structures that are not notable in wild type mice, suggesting an increased amount of vessel regression or pruning.

Since macrophages are known to mediate anastomosis during angiogenesis through their responses to and secretion of angiogenic factors [82, 93-95], we also assessed macrophage content using macrophage-specific marker F4/80. Preliminary results suggest reduced macrophage content in retinas from Clic4 -/- mice as indicated by

qualitatively lighter F4/80 staining, suggesting one possible mechanism by which Clic4

knockout results in less dense retinal vasculature (Fig 3.4b). Lack of properly localized

macrophages may be responsible for the increased pruning or regression seen in Clic4 -/-

mice retinas at the deep vascular plexus.

3.6 Clic4 knockout may enhance tumor growth and metastasis

While we found normal physiological angiogenesis to be only subtly affected, other reports suggest that this phenotype is enhanced when pathological angiogenesis is induced [67]. Thus, we sought to determine if Clic4 knockout affects tumor-induced angiogenesis or tumorigenesis. To examine the effects of Clic4 knockout on 54 tumorigenesis, we used the murine Lewis lung carcinoma model, which has a well- characterized metastatic pathway to the lungs [96]. The Lewis lung carcinoma line is syngeneic to the B6 genetic background, and these Lewis lung carcinoma cells express luciferase, enabling non-invasive visualization of tumor progression. Briefly, Lewis lung carcinoma cells were injected subcutaneously into the upper right flank of Clic4 -/- mice or

wild type littermates. 21 days after implantation of Lewis lung carcinoma cells, tumors

and lungs were harvested and analyzed. As the Lewis lung carcinoma cells are Clic4 +/+ , this presents an environment in which only the invading host cells are Clic4-/-, providing a means by which we are able to study the effects of Clic4 knockout only on invading

vasculature.

In preliminary experiments, we found that tumors isolated from Clic4 -/- mice

tended to be larger by weight than tumors isolated from heterozygous littermates (p <

0.05) (Fig 3.5a). Although not statistically significant, average tumor size by weight from

Clic4 -/- mice tended to be greater than tumors from wild type littermates. There was no

statistically significant difference between tumors isolated from Clic4 +/- and Clic4 +/+ mice based on average size by weight. We also noted larger and more numerous metastases in lungs isolated from Clic4 -/- mice (Fig 3.5b). Metastases were visualized

using the IVIS molecular imaging system to detect fluorescent signals in luciferin-soaked

lungs.

Immunofluorescent histology with macrophage marker F4/80 performed on tumor

sections suggests a reduction in macrophage content in tumors isolated from Clic4 -/- mice

(Fig 3.5c); however F4/80 does not differentiate between macrophage subtypes. This finding could indicate a shift in M1:M2 macrophage phenotype ratios and further analysis 55 is needed to determine macrophage profile. Qualitatively, endothelial cell (CD31) and smooth muscle cell ( αSMA) content is comparable in tumors isolated from Clic4 -/- or

Clic4 +/+ mice, but overall vessel architecture appears altered (Fig 3.5d). Altered macrophage content and vessel architecture suggest two mechanisms by which Clic4 knockout may be enhancing tumorigenesis.

3.7 Discussion

Based on previous reports of CLIC4 being involved in in vitro angiogenesis [49,

56], we sought to study the role of CLIC4 in a physiological setting by generating a Clic4 knockout mouse line. The ubiquitous expression and evolutionary conservation of CLIC4 across vertebrates and invertebrates suggests an important role for the normal function of

CLIC4 [42, 70]. Because of this implication, we had anticipated embryonic lethality upon

Clic4 knockout and designed a conditional Clic4 targeting strategy (Fig 3.1). Surprisingly, we found Clic4 -/- mice to be viable and fertile, albeit underrepresented at birth (Table 3.1).

This difference in penetrance may be due to an underlying defect in the hypoxic response

during prenatal development. Alternatively, Western blotting confirmed that Clic4 -/- mice

did not express CLIC4 (Fig 3.2).

Analysis of Clic4 -/- embryos at 9.5 dpc, a critical time for embryonic angiogenesis, suggests that postnatal underrepresentation is not due to an angiogenic defect, although angiogenic defects may only occur in a small subset of Clic4 -/- embryos (Fig 3.3).

Development progresses normally up to 9.5 dpc as embryo genotypes coincide with

Mendelian ratios at this time point (Table 3.1). Nevertheless, surviving Clic4 -/- mice present subtle defects in postnatal angiogenesis in the retina, characterized by reduced 56 vascular density and the presence of less robust vascular structures (Fig 3.4a). These findings are consistent with in vitro analyses that concluded knocking down CLIC4 reduces endothelial network formation and capillary-like sprouting [56]. Also consistent with previous findings, the reduced vascular density in the retina could be an effect of reduced endothelial cell growth since knocking down CLIC4 has been shown to reduce endothelial proliferation [56]. Increased vessel regression could be indicative of reduced vessel patency, which would be consistent with the hypothesis that CLIC4 plays a role in mediating tubulogenesis as capillaries without patent lumens would experience less sheer stress and be pruned [97, 98].

Preliminary results suggest that developing retinas from Clic4 -/- mice have reduced macrophage content (Fig 3.4b). This defect could be due to either a macrophage recruitment defect due to CLIC4 deficiency or a migratory defect in macrophages incurred by Clic4 deletion since the mice were global Clic4 knockouts. Regardless, the macrophage deficiency could account for the lower vascular density in Clic4 -/- mouse

retinas as macrophages have been shown to mediate anastomosis during angiogenesis

[82]. More in-depth analysis will be required to define the effects of Clic4 knockout on

angiogenesis in either the endothelium or macrophages. Our conditional targeting design

makes the generation of endothelial or macrophage tissue-specific Clic4 knockout

possible for such studies. A macrophage-specific Clic4 knockout line would further

address the question of whether or not the macrophage defect is cell autonomous.

Based on our findings of reduced retinal vasculature and macrophage content in

Clic4 -/- retinas, we initially hypothesized tumorigenesis in Clic4 -/- mice to be inhibited by reduced tumor-induced angiogenesis and decreased inflammation due to lower 57 macrophage content [8, 99, 100]. Interestingly, we instead found tumorigenesis to be enhanced in preliminary experiments, evidenced by larger tumors and more lung metastases from Clic4 -/- mice compared to their wild type littermates (Fig 3.5a & b).

Upon histological analysis of tumor sections with endothelial, smooth muscle, and macrophage specific markers (CD31, αSMA, and F4/80, respectively), we found macrophage content to be reduced in tumors isolated from Clic4 -/- mice (Fig 3.5c). We

also found endothelial and smooth muscle content to be relatively normal, however

overall vessel architecture may be modified (Fig 3.5d).

One possibility to explain the unexpected findings from the tumor studies could

be that an altered macrophage profile occurs due to loss of Clic4 . Loss of Clic4 has been

shown to be protective from lipopolysaccharide (LPS)-induced death characterized by a

reduced inflammatory response [81]. Consistent with these findings, the same study

found that CLIC4 overexpression enhanced the inflammatory response of macrophages

after LPS exposure. These data indicate that CLIC4 plays a role in the innate immune

response to microbial exposure functioning through macrophages. These reports together

with my findings of reduced macrophage content in both Clic4 -/- developing retinas and tumors harvested from Clic4 -/- mice suggest that Clic4 -/- mice may have reduced abilities to clear cancer cells upon tumor implantation and during the early stages of tumorigenesis.

While macrophage content may be reduced in Clic4 -/- isolated tumors, the macrophage profile could be shifted from an M1 inflammatory macrophage phenotype to a more M2 pro-tumor macrophage phenotype. Tumor-associated macrophages (TAM) are now accepted to be of the M2 phenotype [101, 102]. The M1/M2 phenotypes of 58 macrophages have been well-reviewed [103, 104], but briefly, the M1 macrophage phenotype is generally associated with inflammation, tumor resistance, and tissue destruction while the M2 macrophage phenotype imparts a more pro-angiogenic, tissue remodeling, tumor promoting character. Cytokines and microbes are responsible for activating the two polarized macrophage phenotypes. It has been suggested that inhibiting

M2 activity and inducing M1 activity could restore the anti-tumor functions of TAM

[105].

Overall, this study suggests that the functional roles of Clic4 in murine developmental and postnatal angiogenesis can be compensated upon loss by Clic4- independent pathways, or conversely, Clic4 does not participate in physiological angiogenesis. Such partial phenotypes as postnatal underrepresentation, subtle angiogenesis defects, and reduced macrophage content could also be due to functional compensation by other CLICs, specifically CLIC1, which is also widely expressed and evolutionarily well-conserved. This study also constitutes the first assessment of Clic4 deletion effects on tumorigenesis and the discovery of a novel Clic4 deletion effect on macrophage content. Identification of critical Clic4 roles may depend on deletion of additional Clic family members, the deletion of Clic4 in a tissue-specific manner, and additional disease models.

3.8 Figures

Figure 3.1

Figure 3.1 | Diagram of Clic4 targeting strategy and ES cell targeting confirmation.

(a) Schematic of the wild type Clic4 allele, targeting vector p Clic4floxed , targeted Clic4 fl allele, and the recombined Clic4 - allele. Clic4 exons, loxP sites, neomycin resistance cassette, and FRT sites are represented by black boxes, gray triangles, a white box (NEO), and light gray ovals, respectively. P1 refers to the probe used for Southern blot analysis, and Pst1 diagnostic restriction sites are depicted. Positive bands from Southern blot analysis are represented by dotted lines. The start site is depicted by a black arrow off exon 1, and primer sets used for PCR screening are depicted as dark gray and light gray arrows. (b) Presence of a lower band at 7.1 kb from Southern blotting of Pst1 digested genomic DNA isolated from representative, putatively targeted ES cell line ES1 indicates proper integration of the flrted neomycin cassette carrying the upstream loxP site. (c)

Presence of a higher band at 239 bp from PCR of genomic DNA isolated from putatively targeted ES cell line ES1 confirms proper integration of the downstream loxP site. (d)

Genomic DNA isolated from MEFs of mice heterozygous for Clic4 fl and transfected with

GFP or Cre recombinase was digested with Pst1 and subject to Southern blotting.

Absence of a band at 7.1 kb in the presence of Cre recombinase and subsequent presence of an 8.6 kb band signifies the occurrence of Cre-mediated loxP recombination.

Figure 3.2

a. b.

Figure 3.2 | Confirmation of Clic4 knockout. (a) Presence of a 420 bp band confirms deletion of Clic4 exon 2 in a genotyping PCR of wild type (WT), Clic4 homozygous knockout ( -/-), and Clic4 heterozygous ( +/-) mice genomic DNA. A 1.2 kb band is

indicative of wild type Clic4 allele. (b) Absence of a band at 29 kD in a Western blot for

CLIC4 of protein isolated from Clic4 homozygous knockout mice lung cells confirms

that no CLIC4 is being expressed after Cre recombination. α-tubulin serves as a loading

control.

Figure 3.3

Figure 3.3 | Clic4 homozygous knockout mice are phenotypically normal at 9.5 dpc .

Clic4 -/- mice exhibit mild runting at e9.5 dpc, but are otherwise phenotypically normal.

Presence of vasculature highlighted by staining with endothelial marker endomucin

suggests developmental angiogenesis is progressing normally.

Figure 3.4 a.

Figure 3.4 | Clic4 knockout mice show reduced retinal vasculature and macrophage content. (a) Retinal whole mount staining with endothelial marker IB4 (green) shows reduced vasculature density in the primary vascular plexus at P5 (above) and the developing deep vascular plexus at P9 (below). White arrows indicate less robust vasculature. (b) Retinal whole mount staining with macrophage-specific F4/80 (red) indicates reduced macrophage content in Clic4 -/- mice retinas.

Figure 3.5

b. a.

Figure 3.5 | Tumorigenesis is enhanced by Clic4 knockout. (a) Tumors harvested from

Clic4 -/- mice are larger by weight than tumors harvested from Clic4 heterozygous littermates (p < 0.05) and tend to be larger than tumors from wild type littermates. No statistically significant difference in weight was found between tumors from Clic4 heterozygous mice and wild type littermates. (b) The average number of metastases found on lungs dissected from Clic4 knockout mice was greater than lungs from either

Clic4 heterozygous or wild type littermates. In addition, lungs dissected from Clic4 knockout mice presented more large metastases as indicated by dark gray shading. Larger and smaller metastases were differentiated by fluorescent signal intensity. (c)

Macrophage staining using F4/80 suggests reduced macrophage content in tumors from

Clic4 -/- mice. (d) Endothelial (green) and smooth muscle (red) content appear comparable

between tumors from Clic4 knockout and wild type mice. Vessel architecture may be

altered as suggested by a lack of differentiation between the endothelial and smooth

muscle layers. Endothelia are marked by CD31 while αSMA marks smooth muscle cells.

Nuclei are marked in blue with DAPI staining.

3.9 Tables

Table 3.1

a. b. Clic4 +/ - x Clic4 +/ - pups at birth Clic4 +/ - x Clic4 +/ - embryos at 9.5 dpc n = 78, p = 0.03 n = 48, p = 0.72 genotype expected actual genotype expected actual Clic4 -/- 19.5 14 Clic4 -/- 12 14 Clic4 +/ - 39 35 Clic4 +/ - 24 24 Clic4 +/+ 19.5 29 Clic4 +/+ 12 10

Table 3.1 | Clic4 -/- mice are underrepresented at birth. (a) Genotyping of 78 progeny from Clic4 +/- intercrosses reveals that Clic4 -/- mice are underrepresented at birth and that

Clic4 -/-:Clic4 +/-:Clic4 +/+ ratios deviate significantly from normal Mendelian ratio (p <

0.05). (b) Genotyping of 48 embryos collected at 9.5 dpc from Clic4 +/- intercrosses

concurs with normal Mendelian ratio (p > 0.05). Blue coloring indicates expected

numbers for that genotype according to Mendelian predictions while red coloring

indicates actual number of mice of that genotype collected.

Chapter Four: Chloride intracellular channel 1 functions in endothelial cell growth

and migration

Once we determined that CLIC4 was an angiogenic effector, we sought to characterize CLIC1 in much the same manner. Our goals were to determine which steps of angiogenesis are affected when inhibiting CLIC1 expression. Using in vitro analysis, we found that knocking down CLIC1 most drastically affects endothelial growth and network branch formation. We noted some overlapping phenotypes between CLIC1 knockdown and CLIC4 knockdown in endothelial cells, namely the effects on cell growth, network formation, and sprouting. Altering CLIC1 expression levels did not affect lumen formation like altering CLIC4 expression did, whereas only CLIC1 knockdown resulted in a migratory defect. This chapter contains the manuscript I authored on this subject as published in Journal of Angiogenesis Research (Nov 2010) as well as additional results and discussion. I performed all data gathering and analysis for the material in this chapter.

Jan Kitajewski participated in analyzing results. Jan Kitajewski participated in revising the manuscript, with assistance from Minji Kim and Sonia Hernandez during the editing process. Samuel N Breit provided the C lic1 -/- mice that are characterized in section 4.11.

4.1 Abstract

Little is known about the role of CLIC1 in endothelium. These studies investigate

CLIC1 as a regulator of angiogenesis by in vitro techniques that mimic individual steps

of the angiogenic process. Using shRNA against Clic1 , we determined the role of CLIC1 in primary human endothelial cell behavior. Here, we report that reducing CLIC1 71 expression causes a reduction in endothelial migration, cell growth, branching morphogenesis, capillary-like network formation, and capillary-like sprouting. FACS analysis showed that CLIC1 plays a role in regulating the cell surface expression of various integrins that function in angiogenesis including β1 and α3 subunits, as well as

αVβ3 and αVβ5. Together, these results indicate that CLIC1 is required for multiple steps

of in vitro angiogenesis and plays a role in regulating integrin cell surface expression.

4.2 Introduction

The chloride intracellular channel (CLIC) gene superfamily consists of seven

distinct paralogues (p64 and CLIC1-6) and constitutes a unique class of mammalian

channel proteins that exist as both cytoplasm-soluble proteins and membrane-bound

channels [106]. CLICs are structurally related to the glutathione S-transferase (GST)

superfamily and are defined by an approximately 240 conserved amino acid sequence at

the C-terminus [107]. Most of the distinct CLIC proteins are shown to form channels in

artificial bilayers [39, 59, 66, 108, 109], but their selectivity for chloride as channels is

still under contention [43, 58]. CLICs and their homologues are highly conserved among

both vertebrates and invertebrates [36, 37].

Since their discovery, members of the CLIC family have been implicated in such

diverse biological processes as apoptosis [45], differentiation [45, 110], cell cycle

regulation [106], and cell migration [43] in a variety of different cell types. In separate

studies, CLIC4 was found to promote endothelial proliferation and morphogenesis [56]

and to function in mouse retinal angiogenesis [67]. The current model for the angiogenic

function of CLIC4 involves CLIC4 channel activity in the acidification of vesicles [67], a 72 process that may be linked to lumen formation or tubulogenesis [20]. The Hobert group also demonstrated the requirement of C. elegans CLIC4 orthologue EXC-4 expression in preventing cystic disruption of an expanding C. elegans excretory canal and defined a role for EXC-4 in maintaining proper excretory canal lumen size[49]. A chimeric construct expressing human CLIC1 with the putative transmembrane domain (PTM) of exc4 is able to rescue the cystic disruption phenotype of the excretory canal in exc4 null mutants, suggesting that CLIC4 and CLIC1 may have overlapping functions [37].

To date, six CLIC genes (CLIC 1-6) are identified in mice and humans, and

CLIC1 and CLIC4 are reported to be strongly expressed in endothelial cells [49, 52, 54].

As CLIC4 is linked to the process of angiogenesis and lumen formation within endothelial cells [50, 67], interest in the possibility that other CLICs are involved in angiogenesis has grown. Structural studies indicate that oxidized CLIC1 forms dimers in artificial bilayers and vesicles with the PTM located near the N-terminus [39, 111]. It is also suggested that CLIC1 activity is dependent on pH [40]. Studies localize CLIC1 to the nuclear membrane and it is suggested that CLIC1 can regulate the cell cycle of CHO-

K1 cells [106]. CLIC1 is almost ubiquitously expressed in human and mouse adult and fetal tissue [106] and is shown to be F-actin regulated, suggesting that it could function in solute transport, during any number of stages in the cell cycle, or during cell migration

[43]. In several columnar epithelia tissue samples, including but not limited to the renal proximal tubes, small intestine, colon, and airways, CLIC1 is found to be expressed in the apical domains suggesting a role in apical membrane recycling [54]. The same study also finds that CLIC1 subcellular distribution is polarized in an apical fashion in human colon cancer cells while another study finds it localized to intracellular vesicles in renal 73 proximal tubule cells [112]. Since the process of angiogenesis is known to involve endothelial cytoskeletal reorganization, apical-basal polarization, and endothelial proliferation [9, 113], these studies suggest CLIC1 may function in endothelial morphogenesis by influencing some or all of these cellular and subcellular processes.

Most recently, the Breit group generated a CLIC1 knockout mouse and reported platelet dysfunction as well as inhibited clotting in CLIC1 nullizygous mice [114]. There are no other gross phenotypes reported in the CLIC1 nullizygous mice. Given the previously defined roles of CLIC4 in angiogenesis, the suggestion of functional redundancies between CLIC4 and CLIC1, and the implications of CLIC1 involvement in cytoskeletal organization and apical membrane recycling, we now seek to define the role of CLIC1 in endothelial cell behavior and angiogenesis.

Here, we demonstrate the importance of CLIC1 expression in multiple steps of in vitro angiogenesis as well as elucidate a role for CLIC1 in regulating integrin cell surface expression. We show that with reduced CLIC1 expression there is reduced endothelial migration, cell growth, branching morphogenesis, capillary-like network formation, and capillary-like sprouting. CLIC1 also plays a role in regulating the cell surface expression of various integrins important in angiogenesis, including αVβ3 and αVβ5 and subunits β1

and α3.

4.3 Generation of endothelial cell lines with stable CLIC1 knockdown

Human umbilical venous endothelial cell (HUVEC) lines with stable CLIC1

knockdown expression were used to determine the function of CLIC1 in various steps of

angiogenesis, modeled with in vitro assays. Endothelial cell lines were generated using 74 lentiviral-based vectors carrying Clic1 -targeting shRNA. Immunoblotting for CLIC1

confirmed knockdown at the protein level with a reduced band around 31 kDa (Fig 4. 1a)

[115]. Immunoblotting for CLIC4 indicated that the Clic1 shRNA did not target Clic4

and that CLIC4 expression levels were not affected by CLIC1 knockdown. CLIC1

knockdown HUVEC phenotypes were compared to control endothelial cells stably

expressing the lentivirus-based vector backbone carrying a scrambled shRNA insert that

is not known to target any human genes (hereafter referred to as "control").

The cellular morphology of CLIC1 knockdown HUVEC was qualitatively

assessed and we observed no major changes in morphology with respect to control (Fig

4.1b). For this analysis, the CLIC1 knockdown and control cell lines were seeded in

equal numbers as a subconfluent monolayer on collagen-coated plates and photographed

48 h later. While no gross morphological changes were present, we noted that CLIC1

knockdown cells appeared less dense than control cells, indicating that CLIC1

knockdown may affect endothelial cell growth.

4.4 CLIC1 knockdown inhibits endothelial cell growth

To determine if CLIC1 is involved in regulating endothelial cell viability or growth, we utilized WST-8 colorimetric cell counting assays to score cells under different conditions. Cell viability assays were performed on CLIC1 knockdown or control cells by seeding equal numbers of HUVEC on collagen-coated plates and scoring cells after being cultured for 48 h in serum-free medium (SFM) alone or SFM supplemented with the survival signal EGF. Cell counts in medium without survival factor EGF were then scored relative to cell counts of the same cell line in medium with 75

EGF to produce a percentage of cells that remain viable in medium without EGF. In the absence of EGF, we found that CLIC1 knockdown cells displayed a modest but significant increase in HUVEC viability (p < 0.05), indicating that CLIC1 knockdown may play a limited role in regulating cell viability (Fig 4.2a).

To determine the effects of knocking down CLIC1 on endothelial cell growth,

CLIC1 knockdown or control HUVEC were seeded in equal numbers on collagen coated plates and cultured in SFM supplemented with survival signal EGF and VEGF to induce endothelial cell growth. Cells were allowed to grow for 96 h, and cells were then scored using WST-8 colorimetric detection. In contrast to the effect on cell viability, we found that reduction of CLIC1 expression led to a pronounced and significant reduction of

HUVEC cell growth (p < 0.001), indicating that CLIC1 is involved in regulating endothelial cell growth (Fig 4.2b).

4.5 CLIC1 knockdown inhibits endothelial migration

Given the previously described associations of CLIC1 with cytoskeletal elements

[43], we utilized a directed cell migration scratch assay to determine if CLIC1 knockdown affects directed endothelial cell motility. This assay involved creating an

"open wound" across a confluent monolayer of HUVEC growing on collagen and documenting cell migration into the wounded area at various time points over a 12 h period. We found that CLIC1 knockdown resulted in reduced endothelial migration into the wounded area when assessed on a qualitative level (Fig 4.3a). Given that the CLIC1 knockdown migratory defects are qualitatively noticeable as early as 3 h post-wounding and the migration assay terminated only 12 h post-wounding, it was unlikely that any 76

CLIC1 knockdown effects on endothelial survival or proliferation affected migration assay outcomes.

To quantify the extent to which reduced CLIC1 expression was inhibiting directed endothelial cell migration, TScratch software was used for quantifying open surface area

[116]. Quantification of data collected from six separate experiments confirmed that

CLIC1 knockdown cells occupy significantly less surface area at 6 (p < 0.02), 9 (p <

0.001), and 12 h (p < 0.001) post-wounding (Fig 4.3b), indicating that CLIC1 knockdown significantly reduced directed endothelial cell migration as early as 6 h post-wounding.

Thus, inhibiting CLIC1 expression reduces directed endothelial cell migration.

4.6 CLIC1 influences expression of select integrins on endothelial cells

To explore the regulatory role of CLIC1 in endothelial migration further, we analyzed the effects of CLIC1 knockdown on various integrins by flow cytometry. For this assessment, CLIC1 knockdown and control HUVEC were cultured as subconfluent monolayers on either type I collagen-coated or fibronectin-coated plates and incubated with primary antibodies for various integrins. A secondary APC-conjugated antibody was then used to enable flow cytometry. Integrins examined include α2, β1, α3, αVβ3, and

αVβ5. Endothelial marker CD31 served as a positive control while the absence of

primary antibody served as a negative control. In addition to being important for cell

attachment and migration on specific extracellular matrices, each of the integrins

examined have been reported to affect the angiogenic process [117-121].

Expression analysis revealed the same general expression patterns for HUVEC

grown either on collagen-coated plates (Fig 4.4, left panels) or HUVEC grown on 77 fibronectin-coated plates (Fig 4.4, right panels). The expression assays were done three times and results from a typical flow experiment are depicted in Figure 4.4. A slight downward shift in CD31 was found to be typical of CLIC knockdown cells. Analysis from both plate types indicated that HUVEC with CLIC1 knockdown had increased cell surface expression of integrin subunits β1 and α3 while α2 remained mostly unaffected

(Fig 4.4). Integrin αVβ3 surface expression was also increased in CLIC1 knockdown

endothelial cells while αVβ5 expression was decreased on both extracellular matrix plate types. These results suggest that CLIC1 may be moderating endothelial migration through regulation of integrin expression at the cell surface.

4.7 CLIC1 plays a role in capillary-like network formation

We next determined the effect of CLIC1 knockdown on endothelial cell morphogenesis by assessing the ability of HUVEC to organize into capillary-like networks. To do this, CLIC1 knockdown or control HUVEC were seeded in equal numbers between two layers of porcine collagen gel and cultured in SFM supplemented with VEGF and EGF for 96 h. Qualitative assessment of the results indicated a reduction in cord, network, and branch formations in CLIC1 knockdown relative to control (Fig

4.5a). Network formation from cells with reduced CLIC1 expression was noticeably less dense when compared to control cells as indicated by more prevalent open surface area.

In addition to these defects, we found dramatic cell aggregations at branch points when

CLIC1 was knocked down, where one branch point was considered any intersection of two or more cords. This phenotype could have been due to the migratory defects 78 established earlier. As indicated in a previous study assessing CLIC4 function, these cell aggregations are not due to proliferative defects [56].

Quantification of these results revealed that there is a significant reduction in surface area coverage by CLIC1 knockdown cells (p < 0.01) (Fig 4.5b). This was accompanied by a significant decrease in branch points formed by CLIC1 knockdown cells (p < 0.001) (Fig 4.5c). The decrease in surface area coverage in CLIC1 knockdown could have been due to the previously established proliferative defect in CLIC1 knockdown HUVEC, however the reduction in branch point formation was novel, indicating that reduced CLIC1 expression inhibits endothelial network formation and branching.

4.8 Reduced CLIC1 expression affects capillary-like sprouting and branching

Next, we assessed the effect of CLIC1 knockdown on capillary-like tube formation in a fibrin bead assay, which allowed for assessment of endothelial growth, sprouting, branching, and lumen formation. For this assay, CLIC1 knockdown and control HUVEC were attached to dextran-coated beads and embedded in a fibrin clot with fibroblasts seeded as a monolayer on top of the clot. The clot was cultured in EGM-

2 medium, and HUVEC morphogenesis was monitored and photographically documented for 11 days. As previously reported, resultant sprouts from this assay are multicellular, lumen-containing processes [74, 75]. In this assay, sprouting is noticeable as early as three days post-embedding. Sprouts are reported to extend, anastomose, and undergo tubulogenesis from day 4 until the termination of the assay at day 11. 79

Our results indicated that knocking down CLIC1 expression results in stunted capillary sprouting and branching with little apparent effect on lumen formation (Fig 4.6a, b). Upon quantification of results from day 11, we found that the number of sprouts per bead and the average number of branches per sprout were significantly decreased in the

CLIC1 knockdown group (p < 0.001) while the number of lumen-containing sprouts was not different (Fig 4.6c-e). The branching defect found in this assay was consistent with the branching defect found in the previous capillary-like network formation assay.

Together, these data demonstrate that CLIC1 expression is required for capillary sprouting and branching morphogenesis, when assessed in vitro.

4.9 Discussion

We have previously shown that CLIC4 knockdown results in lower endothelial cell growth and inhibited network formation, similar to the CLIC1 knockdown results shown here [56]. However, in contrast to the failure of CLIC4 knockdown cells to form networks, CLIC1 knockdown cells formed rudimentary networks but had a branching morphogenesis defect whereby cell aggregates formed at branch points. We suspect these cell aggregates are a manifestation of the migratory defect and a failure of knockdown cells to undergo appropriate branching morphogenesis. This would be consistent with the observation that CLIC1 knockdown reduces endothelial cell migration, whereas CLIC4 knockdown has no effect on migration. Another difference our comparison highlights is the lack of a lumen formation defect in CLIC1 knockdown endothelial cells in contrast to the lumen formation defect previously found with CLIC4 knockdown [56]. To explore the possibility of functional redundancy between CLIC1 and CLIC4, we made several 80 attempts to generate HUVEC lines with both CLIC1 and CLIC4 knockdown. In contrast to HUVEC introduced with two different control vectors, all attempts to create double knockdown cell lines did not result in viable HUVEC. We found this observation consistent with the hypothesis that CLIC1 and CLIC4 may possess functional redundancies and that both are required for endothelial cell viability.

CLIC1 is found to be significantly up-regulated in highly metastatic gallbladder carcinoma cell lines [122]. More generally, chloride transport is reported to be integral in generating electrical signals that guide cell migration to wounds in corneal epithelium

[123], and chloride channels are implicated directly in enabling glioma migration as well as regulating breast cancer invasiveness [124, 125]. One may thus hypothesize that

CLIC1 is required for endothelial cell motility based upon its potential function as an ion channel.

Extensive work has been done to validate CLIC1 as an anion channel that can auto-insert into artificial bilayers and produce conductance [40, 108], however the channel selectivity is found to be poor with conductance being based on anion concentration [39-41, 106, 111]. As an anion channel, CLIC1 is redox-regulated and its sequence contains the putative transmembrane domain (PTM) purported to be essential for its proper integral membrane channel characteristics [32, 111]. CLIC1 ion channel activity is also shown to be pH-dependent with activity lowest around neutral pH [40].

Structural experiments show that low pH may stimulate the PTM for insertion [36, 126,

127]. It will be important to assess whether these mechanistic features of CLIC1 are important for endothelial cell motility. 81

One of the most interesting structural characteristics of CLIC1 is the fact that it can exist as both an integral membrane protein and a soluble cytoplasmic protein. In contrast to the well-documented anion channel activity of CLIC1, the roles of CLIC1 independent of its channel activity are largely unknown. Studies demonstrate that a variety of CLIC proteins interact with the actin cytoskeleton either directly [43, 57, 70] or indirectly, mediated by scaffolding proteins [68, 115, 128]. With this in mind, we postulate that CLIC1 may be regulating endothelial cell migration by regulation of cytoskeletal elements.

Endothelial cell migration and adhesion also depend on appropriate integrin expression [129]. Studies show that adhesion to the extracellular matrix through integrin heterodimers is essential for proper endothelial cell motility, and endothelial migration is at its greatest with intermediate levels of adhesion [130, 131]. Of interest to our study are integrins αVβ3, which can bind with fibronectin; αVβ5, which binds only vitronectin;

integrin subunits α2 and β1, which are known for binding collagens, laminins, and possibly fibronectin [132]; and the integrin α3 subunit, which binds fibronectin [133]. We

found that reducing CLIC1 expression increased β1, α3, and αVβ3 expression while

decreasing αVβ5 expression (Fig 4.3), suggesting a role for CLIC1 in mediating integrin presentation and a means by which CLIC1 may be affecting endothelial migration. By increasing the surface expression of β1, α3, and αVβ3, CLIC1 may be increasing endothelial cell adhesion to the extracellular matrix, inhibiting motility by preventing the cell from breaking its contact with the extracellular matrix. These shifts in integrin expression also provide a possible explanation for the cell growth and viability defects

[134]. In addition, it is possible that CLIC1 alters integrin binding affinity for their 82 ligands through inside-out signaling resulting in increased integrin expression but less efficient ligand binding [135]. To determine a mechanism by which CLIC1 may be influencing integrin cell surface expression, we conducted Western blotting for β1 and β5 integrin subunits and found that the protein levels of these integrins were unchanged by

CLIC1 knockdown (data not shown). Based on these preliminary results, we hypothesize that changes to integrin cell surface expression are a result of altered cell trafficking as integrins cycle through the endosomal pathway and as a chloride channel, CLIC1 contributes to endosomal acidification [136].

In summary, we demonstrated here that CLIC1 is involved in several steps of angiogenesis in vitro and concluded that CLIC1 plays a role in mediating endothelial cell growth, branching morphogenesis, and migration, possibly via regulation of integrin expression. We found that CLIC1 and CLIC4 knockdown phenotypes are similar in that both result in reduced cell growth, modestly increased cell viability, and inhibited network formation, but are different in that only CLIC1 knockdown inhibits migration and only CLIC4 knockdown affects lumen formation. It will be important to understand the molecular mechanisms by which CLIC1 functions in these diverse endothelial cell behaviors.

4.10 Figures

Figure 4.1

Figure 4.1 | Establishing human endothelial cell lines with CLIC1 knockdown . (a)

Immunoblotting with polyclonal rabbit anti-CLIC1 antibody confirmed reduced CLIC1 protein expression in CLIC1 knockdown (C1 kd) endothelial cell lines. Immunoblotting with rabbit polyclonal anti-CLIC4 antibody verified that CLIC4 expression was not altered by Clic1 -targetting shRNA. α-tubulin served as a loading control. (b) Established

HUVEC control and CLIC1 knockdown lines exhibited no gross morphological

differences when cultured as a subconfluent monolayer on collagen-coated dishes. Scale

bar, 30 µm.

Figure 4.2

a. b.

Figure 4.2 | CLIC1 knockdown has mild effects on viability and significantly inhibits endothelial cell growth . (a) CLIC1 knockdown mildly but significantly increased

viability in endothelial cells 48 h post-seeding. Quantification of viability is shown as

mean percentages relative to the respective starting number of cells with standard

deviation. (*p < 0.05). (b) Reduced CLIC1 expression drastically and significantly

inhibited endothelial cell growth. The line at 10000 cells on the y-axis indicates the

number of cells at the start of the assay. Quantification of cell growth is shown as the

mean number of cells for each group with standard deviation. (**p < 0.001).

Figure 4.3

Figure 4.3 | Reduced CLIC1 expression inhibits endothelial migration. (a) CLIC1

knockdown endothelial lines exhibited inhibited directed migration in a scratch assay,

noticeable as early as 3 h post-wounding. Photographs shown here have been analyzed

using the TScratch software for quantifying open surface area. Lighter shaded areas

indicate area occupied by endothelial cells while darker shaded areas indicate surface

area not occupied by endothelial cells. Scale bar, 50 µm. (b) Quantification using the

TScratch software showed that there was significantly more open surface area in the

CLIC1 knockdown group at and after 6 h post-wounding. Mean open surface area for

each group is shown with standard deviation. (*p < 0.02; **p < 0.001). 89

Figure 4.4

Figure 4.4 | CLIC1 expression plays a role in the expression of various integrins .

Histograms from a typical flow experiment indicate that HUVEC with CLIC1 knockdown (green) had increased β1, α3, and αVβ3 expression and decreased αVβ5 expression while α2 expression remained generally unchanged. Trends persisted despite being cultured on collagen or fibronectin extracellular matrix components. A downward shift in positive control endothelial marker CD31 was typical with CLIC1 knockdown cells.

Figure 4.5

b. c.

Figure 4.5 | Knocked down CLIC1 expression inhibits capillary-like network formation and branching morphogenesis . (a) Microscopy of control or CLIC1 knockdown lines in a collagen network formation assay indicated decreased network formation and defective branching morphogenesis in cells with CLIC1 knockdown. In addition to having less dense network patterning (exemplified by blue arrowheads), cell aggregations were found where normal branch points should have been (exemplified by red arrows). Black arrowheads indicate regularly spaced network patterning, and black arrows indicate normal branch points. Scale bar, 50 µm. (b) Quantification of surface area

occupied by endothelial cells showed a significant decrease in surface area occupation by

cells with CLIC1 knockdown. Mean surface area occupied by endothelial cells for each

cell line is shown with standard deviation. (*p < 0.01). (c) Quantification of branch point

numbers revealed a significant decrease in the number of branch point formations in cells

with CLIC1 knockdown. One branch point was considered any focal intersection of two

or more cords. The mean number of branch points for each cell line is shown with

standard deviation. (**p < 0.001).

Figure 4.6 a.

c. d. e.

Figure 4.6 | Reduced CLIC1 expression affects capillary-like sprouting and branching . (a) Microscopy on day 7 of the fibrin bead assay showed that endothelial cells with CLIC1 knockdown had altered sprouting morphology. CLIC1 knockdown sprouts tended to be shorter, less branched, and less robust. Black arrows indicate long, robust, lumen-containing sprouts while red arrows indicate stunted tube structures. Scale bar, 10 µm. (b) Microscopy on day 11 of the fibrin bead assay showed that endothelial

cells with reduced CLIC1 expression formed tube-like structures; however these

structures were stunted and possessed a branching defect when compared to structures

formed by control cells. Black arrows indicate anastomosis or long tube-like structures.

Red arrows indicate lack of branching or stunted tube-like structures. Scale bar, 30 µm.

(c) The average number of sprouts formed per bead was tabulated and quantified, shown here with standard deviation. Reduced CLIC1 expression was found to be accompanied by a significant reduction in the average number of sprouts per bead. (d) Quantification of the average number of branches per sprout indicated that CLIC1 knockdown endothelial cells formed significantly fewer branches per sprout. The average number of branches per sprout is shown with standard deviation. (e) Quantification of the number of lumen-containing sprouts revealed no significant difference between control cells and cells with CLIC1 knockdown. The average percentage of lumen-containing sprouts relative to total number of sprouts per cell line is shown with standard deviation. (**p <

0.001).

4.11 Additional discussion

Our main interest in CLIC1 laid in the fact that it was one of two highly expressed

CLIC proteins expressed in endothelial cells, along with CLIC4 (Section 2.9). We hypothesized that CLIC1 could functionally compensate for CLIC4 during angiogenesis as suggested by the mostly normal vascular development seen in Clic4 -/- mice (Chapter 3),

despite the drastic phenotypes resultant from CLIC4 knockdown in endothelial cells

(Chapter 2). To determine if CLIC4 and CLIC1 had overlapping functions in the

endothelial response to angiogenesis, we sought to rescue the CLIC1 knockdown network

formation phenotype with CLIC4 overexpression using our overexpression vector

described in Chapter 3. Western blotting confirmed CLIC1 knockdown and CLIC4

overexpression (Supp Fig 4.1). Preliminary results suggest that CLIC4 and CLIC1 do

overlap in function to some extent as there was a partial rescue by CLIC4 overexpression

of the network formation defect imparted by CLIC1 knockdown (Supp Fig 4.2).

In addition, we sought to determine the physiological relevance of our CLIC1 in

vitro data by analyzing the Clic1 -/- mouse for angiogenic defect. We also sought to establish any phenotypic similarities between the Clic1 -/- mice and our Clic4 -/- mice lines.

The Clic1 -/- mice were generated and initially characterized by members of Sam Breit’s group at St. Vincent’s Centre for Applied Medical Research in New South Wales,

Australia, and the findings are reported in Genesis [114]. Briefly, Clic1 -/- mice are viable

and fertile with no obvious developmental abnormalities. Further analysis revealed a

platelet dysfunction phenotype characterized by prolonged bleeding and decreased

platelet activation upon stimulation. After acquiring these mice, we determined that

Clic1 -/- mice indeed breed consistent with Mendelian ratios (Supp Table 4.1). Clic -/- 96 retinal vasculature is also indistinguishable from vasculature of wild type littermates unlike Clic4 -/- mice, which had a subtle vascular defect in the developing retina. This

indicates normal postnatal angiogenic development in Clic1 -/- mice (Supp Fig 4.3).

These additional data are consistent with the hypothesis that CLIC4 and CLIC1

possess overlapping functions. Partial rescue of the CLIC1 knockdown network

formation defect by CLIC4 is indicative of overlapping functions or crosstalk between

the two proteins. Furthermore, the absence of a drastic angiogenic phenotype in either

Clic4 or Clic1 mutant mice despite strong endothelial phenotypes in vitro suggests compensation by other pathways. To assess the possibility of functional redundancy further, deletion of both Clic4 and Clic1 in mice may be necessary.

4.12 Supplementary figures

Supplementary Figure 4.1

Supplementary Figure 4.1 | Confirmation of CLIC4 overexpression and CLIC1 knockdown. Western blotting for CLIC4 or CLIC1 confirms overexpression or knockdown of HUVEC lines, respectively. Csh2 denotes CLIC1 knockdown, CLIC4 denotes CLIC4 overexpression, and plko sc and gfp are control vectors. α-tubulin served

as a loading control.

Supplementary Figure 4.2

100

Supplementary Figure 4.2 | Partial rescue of the CLIC1 knockdown network formation defect by CLIC4. Overexpression of CLIC4 in CLIC1 knockdown cell lines partially rescues the network formation defect seen by CLIC1 knockdown alone

(csh2/gfp). Networks maintain cell aggregates at branch points compared with control

(plko sc/gfp and CLIC4/plko sc), but aggregations appear to be reduced in the rescue line

(csh2/CLIC4). 101

Supplementary Figure 4.3

102

Supplementary Figure 4.3 | Loss of Clic1 does not affect postnatal retinal angiogenesis. Staining with endothelium-specific marker IB4 (green) shows primary plexus retinal vasculature in P5 mice that are either homozygous null or wild type for

Clic1 to be indistinguishable.

103

4.13 Supplementary Tables

Supplementary Table 4.1

n = 27, p = 0.95 genotype expected actual Clic1-/- 6.75 7 Clic1+/- 13.5 14 Clic1+/+ 6.75 6

Supplementary Table 4.1 | Clic1 -/- mice confer with expected Mendelian ratios.

Genotyping of 27 mice from Clic1 +/- intercrosses reveal no deviations from expected

Mendelian ratios, according to a χ2 test for goodness of fit.

104

Chapter Five: Clic4 and Clic1 are required for embryonic development

To determine whether CLIC4 and CLIC1 act in concert during developmental angiogenesis, we generated mice mutant for both Clic4 and Clic1 , designated Clic4 -/-

;Clic1 -/-. This chapter details the generation and subsequent analysis of Clic4 -/-;Clic1 -/- double knockout mice. I performed all data gathering and analysis for the material in this chapter. Carrie Shawber provided advice on experimental design and analysis. Valeriya

Borisenko performed embryo sectioning and Irina Jilishitz provided technical assistance with histology. Jan Kitajewski and Carrie Shawber assisted in analyzing results.

5.1 Abstract

Both CLIC4 and CLIC1 have been implicated in mediating angiogenesis in vitro , but initial characterizations of Clic4 and Clic1 knockout mice have revealed only a mild angiogenic defect in Clic4 -/- mice and no angiogenic defect in Clic1 -/- mice. Neither

Clic4 -/- nor Clic1 -/- mice are known to exhibit angiogenic defects in utero . This study finds transgenic mice deficient for both Clic4 and Clic1 to be embryonic lethal with in

utero time of death at 9.5 dpc. Lethality is accompanied by a defect in angiogenesis

characterized by diminished or absent vasculature in the head and along the intersomitic

vessels. Clic4;Clic1 double knockouts are also found to be growth retarded, characterized

by smaller size. These results indicate that both CLIC4 and CLIC1 function concurrently

during embryonic angiogenesis and both proteins are required for proper embryonic

development.

105

5.2 Introduction

Of the 6 known mammalian CLICs, only CLIC4 and CLIC1 are shown to be highly expressed in endothelial cells [49, 52, 54]. CLIC4 and CLIC1 share similar protein structures and 60-66% amino acid homology in humans and mice [36, 59, 89]. Both proteins have been shown to elicit chloride channel conductance when reconstituted in artificial membranes. CLIC4 and CLIC1 are also shown to be widely expressed and evolutionarily well-conserved, suggesting an important role for both proteins in development. Indeed, both CLIC4 and CLIC1 have been found individually to have roles in diverse biological processes including angiogenesis and endothelial cell proliferation

[55, 56], and both are F-actin regulated [43]. In both the human and mouse genomes,

Clic4 and Clic1 localize to two different chromosomes.

CLIC4 was first implicated in angiogenesis as a regulator of tubulogenesis when

CLIC4 C. elegans orthologue EXC-4 was found to be essential for proper formation and maintenance of the C. elegans excretory canal, which is thought to form lumens much the

same way as mammalian capillaries [10, 37, 50]. CLIC4 was later found to be vascular

endothelial growth factor-A (VEGF-A)-regulated [49]. As VEGF-A is a well-

characterized effector of angiogenesis [137-140], this further implicated CLIC4 in the

angiogenic pathway. In addition, the cystic disruption of the C. elegans excretory canal that typifies the mutant exc-4 tubulogenic defect can be rescued with human Clic1 fused to the transmembrane domain of exc-4, suggesting functional overlap between CLIC4 and CLIC1 in tubulogenesis. Evidence implicating CLIC1 in angiogenesis stems from its localization to the apical domain of several columnar epithelia [54], linking it to apicobasal specification, a process required for proper tube formation. 106

Despite CLIC4 or CLIC1 knockdowns having drastic effects on the endothelial response in various in vitro angiogenesis assays (Chapter 3), neither Clic4 nor Clic1 knockout mice exhibit lethal defects. Reports of Clic1 -/- mice indicate that it is wholly

indistinguishable from wild type mice aside from a platelet activation defect, even under

ischemic conditions [114, 141] while Clic4 -/- mice exhibit reduced native collateral density [67, 141]. Interestingly, under ischemic conditions, both CLIC4 and CLIC1 expression are increased, further suggesting functional redundancy between the two proteins. Consistent with these data, Clic1 -/- mice do not exhibit any angiogenic defects in

the developing retina (Section 4.11) while Clic4 -/- mice exhibit only mild angiogenic defects in the developing retina (Section 3.5). Taken together with in vitro results from

CLIC4 or CLIC1 knockdown lines, these data suggest that an alternate pathway is functionally compensating for loss of either Clic4 or Clic1.

In primary cultured endothelial cells, CLIC4 and CLIC1 have some overlapping properties such as their roles in proliferation, branch formation, anastomosis, and capillary sprouting; however they also differ in that only CLIC1 affects endothelial directed migration while only CLIC4 has been shown to affect endothelial lumen formation. From their disparate roles in endothelial cells, we inferred that targeting both

Clic4 and Clic1 simultaneously would result in a more severe phenotype than targeting

either gene alone. Therefore, we generated transgenic mice in which both Clic4 and Clic1

are deleted. We found the Clic4 -/-;Clic1 -/- genotype to be embryonic lethal at 9.5 days

post coitus (dpc), exhibiting growth retardation and vascular development defects at time

of death. Clic4 -/-;Clic1 -/- embryos isolated at 9.5 dpc also show reduced endothelial content. The Clic4;Clic1 double knockout shows a more severe angiogenic defect than 107 found in either Clic4 or Clic1 single knockouts. Mice deficient for either Clic4 or Clic1 and heterozygous for the other Clic (Clic4 -/-;Clic1 +/- or Clic4 +/-;Clic1 -/-) are viable, but

mice deficient for Clic4 and heterozygous for Clic1 (Clic4 -/-;Clic1 +/-) appear runted. Our results indicate that both Clic4 and Clic1 are required for normal angiogenesis during early embryonic development.

5.3 Generation of Clic4 and Clic1 double mutants

Both CLIC4 and CLIC1 have been shown individually to affect angiogenesis in vitro , and endothelial cells express both proteins in high amounts [55, 56]. To determine the extent to which CLIC4 and CLIC1 affect angiogenesis during development, it was deemed necessary to target both genes to account for possible functional redundancies.

We obtained Clic1 -/- mice from Dr. Samuel N Breit, and we generated Clic4 -/- mice, described in chapter 4. As previously reported and summarized in section 4.11, Clic1 -/-

mice are viable, fertile, and possess no gross angiogenic phenotypes in the retina [114].

As detailed in chapter 3, our Clic4 -/- mice are underrepresented, have subtle defects in

developing retinal vasculature, and may have reduced macrophage content in the

developing retina.

Clic4 -/- knockout mice were mated with Clic1 -/- knockout mice to produce Clic4 +/-

;Clic1 +/- double heterozygotes. Clic4 +/-;Clic1 +/- females were then mated with Clic4 +/-

;Clic1 +/- males to generate double Clic4 -/-;Clic1 -/- mice, however no mice of this double knockout genotype were found in the F1 generation, deviating from predicted Mendelian ratios (p < 0.01) (Table 5.1), suggesting embryonic lethality. Mice representing all other expected genotypes including those missing three Clic alleles survived birth. Mice were 108 genotyped using PCR genotyping strategies outlined in chapter 4 and previous reports

[114].

5.4 Clic4 -/-;Clic1 -/- double knockout mice are embryonic lethal at 9.5 dpc

To increase chances of obtaining Clic4 -/-;Clic1 -/- embryos, we performed Clic4 -/-

;Clic1 +/- intercrosses alongside Clic4 +/-;Clic1 -/- intercrosses. The resulting F1 generation did not contain Clic4 -/-;Clic1 -/- mice. This deviation from Mendelian ratio again indicates

embryonic lethality for the Clic4 -/-;Clic1 -/- genotype (p < 0.01) (Table 5.2a). To

determine time of death, we performed embryo dissections at various gestational time

points.

At 11.5 dpc, Clic4 -/-;Clic1 -/- embryos were not present. Clic4 -/-;Clic1 -/- embryos collected at 10.5 dpc had expired presenting growth retardation and cardiac edema (Fig

5.1), indicative of vascular blockade [142, 143]. Pooled blood is also evident in double knockout embryos isolated at 10.5 dpc, suggestive of vessel hemorrhaging. Interestingly, embryos analyzed by whole mount endomucin staining at 10.5 dpc comprised embryos representative of all other genotypes including those missing three Clic alleles. At 10.5

dpc, embryos heterozygous for Clic4 and Clic1 appeared comparable to wild type 10.5

dpc embryos. Clic4 +/-;Clic1 -/- embryos were indistinguishable from Clic4 +/-;Clic1 +/-

littermates while Clic4 -/-;Clic1 +/- embryos exhibited growth retardation, but vasculature

was wholly present (Fig 5.2). Qualitatively, Clic4 -/-;Clic1 +/- embryos may also have

reduced endothelial content as suggested by lighter endomucin staining when compared

with Clic4 +/-;Clic1 +/- littermates. 109

Clic4 -/-;Clic1 -/- embryos dissected at 9.5 dpc were alive as evidenced by blood flow and beating hearts. Furthermore, Mendelian ratio was restored at 9.5 dpc (Table

5.2b) but not 10.5 or 11.5 dpc. These data establish time of death for Clic4 -/-;Clic1 -/-

embryos to be between 9.5 and 10.5 dpc.

5.5 Clic4 -/-;Clic1 -/- double knockout mice display aberrant vascular development

To determine cause of death, we intercrossed Clic4 +/-;Clic1 -/- mice to obtain

Clic4 -/-;Clic1 -/- embryos for comparison with Clic4 +/+ ;Clic1 -/- littermates at 9.5 dpc.

Clic4 +/+ ;Clic1 -/- were determined to be phenotypically indistinguishable from

Clic4 +/+ ;Clic1 +/+ embryos at 9.5 (Fig 5.3). Dissected 9.5 dpc embryos were analyzed by

hematoxylin and eosin (HE) staining and whole mount staining with endothelial marker

endomucin [144-146].

We found endomucin staining in Clic4 -/-;Clic1 -/- 9.5 dpc embryos to be consistently lighter, indicating reduced endothelial content (Fig 5.4a). Severely reduced vasculature along the intersomitic vessels and in the head indicate an angiogenic defect in

Clic4 -/-;Clic1 -/- mice. As major vessels are still present, we concluded that the double

knockout phenotype is unlikely due to vasculogenic defects. We also concluded that the

double knockout phenotype is unlikely due to arteriovenous specification defects because

the dorsal aorta and cardinal vein are discernible and present in the double knockout

embryos. The observed vascular phenotype of the Clic4 -/-;Clic1 -/- embryo is qualitatively

more severe than the effects of either Clic4 or Clic1 knockout alone on vascular

development at 9.5 dpc. Growth retardation indicated by smaller size is consistent with

previously noted proliferation defects found in vitro (Section 2.4). Runting could also be 110 consistent with disrupted vascular development because tissues that are not well perfused do not grow as well [147].

Clic4 -/-;Clic1 -/- embryos collected at 9.0 dpc and earlier presented no gross

morphological phenotypes aside from qualitatively lighter endothelial staining suggestive

of mildly reduced endothelial content (Fig 5.4b). Mild runting is noted in Clic4 -/-;Clic1-/-

9.0 dpc embryos. These data suggest that vessel regression may be occurring between 9.0 and 9.5 dpc to yield the 9.5 dpc angiogenic phenotype. Vessel regression may occur due to a lack of vessel patency [97, 98]. Qualitative analysis of embryos at 7.5 and 8.5 dpc suggest that vascular development occurs normally until 9.0 dpc.

5.6 Aortic development and proliferation may be altered in Clic4 -/-;Clic1 -/-

embryos

To assess vessel morphology more in depth, Clic4 -/-;Clic1 -/- embryos were collected at 9.5 dpc for fixed-frozen sectioning and histological analysis. Preliminary staining with endothelial marker CD31 is fainter in Clic4 -/-;Clic1 -/- sections indicating reduced endothelial content (Fig 5.5). Analysis with additional endothelial marker vascular endothelial growth factor receptor-2 (VEGFR-2) also suggests reduced endothelial content consistent with endomucin-stained whole mounts (Fig 5.6).

Preliminary VEGFR-2 staining of Clic4 -/-;Clic1 -/- embryo sections appears lighter in both

the heart and rump. In the Clic4 -/-;Clic1 -/- embryo heart, VEGFR-2 staining of the

endocardium is present, but the endothelium along the descending aorta appears thinner

while the aorta itself seems dilated (Fig 5.6a). This suggests impaired aortic development

in double knockout embryos. Analysis of VEGFR-2 and αSMA should be performed on 111 transverse embryo sections to assess aortic dilation further. While VEGFR-2 staining of the intersomitic vessels (ISVs) is present in Clic4 -/-;Clic1 -/-, VEGFR-2 stained ISVs

appear disjointed, suggesting disruption of ISV development (Fig 5.6b).

Jagged1 (Jag1) staining also appears fainter in Clic4 -/-;Clic1 -/- embryo sections,

most notably in the heart and at the otic vesicle (Fig 5.7a). This suggests that Clic4 may regulate Notch1 signaling through Jag1 [148, 149]. αSMA staining revealed no major

changes in smooth muscle cell content by qualitative assessment (Fig 5.7b). Based on

αSMA staining, trabeculation appears normal and the myocardium appears comparable in both samples.

At 9.5 dpc, all cell types are actively proliferating in a normal mouse embryo.

Preliminary analysis with proliferation marker phospho-histone H3 (Ser10) suggests reduced proliferation throughout the embryo along two different planes (Fig 5.8). While quantification of preliminary results does not reach statistical significance due to variance and a small sample size, the trend of reduced proliferation in Clic4 -/-;Clic1 -/- embryos is evident (Fig 5.8b). This finding is consistent with in vitro data (Section 2.4) as well as the

runted phenotype of Clic4 -/-;Clic1 -/- embryos. Clic4 -/-;Clic1 -/- 9.5 dpc embryos also tend

to exhibit moderately increased apoptosis, revealed by caspase 3 (CASP3) staining (Fig

5.9). The trend of increased apoptosis in Clic4 -/-;Clic -/- embryos is not necessarily

associated with endothelial cells.

5.7 Discussion

The effects of knocking down CLIC4 or CLIC1 on endothelial cell morphogenesis in vitro suggested that both CLIC4 and CLIC1 may play an important 112 role in vasculogenesis or angiogenesis during development [55, 56]. Conversely, knockout of either Clic4 or Clic1 does not result in major functional or developmental defects (Chapter 3 and Section 4.11). Although subtle angiogenic defects may have gone unnoticed during initial characterization, both Clic4 and Clic1 null mice were viable, fertile, and presented normal appearing vasculature by whole mount analysis. The similarities in expression patterns, structure, and effects on endothelial morphogenesis in vitro suggest that CLIC4 and CLIC1 may at least partially compensate for each other during critical roles. Here, we provide the first study supporting the concept of functional redundancy between CLIC4 and CLIC1 during development.

Intercrossing animals heterozygous for both Clic4 and Clic1 failed to produce

Clic4 -/-;Clic1 -/- double knockout mice (Table 5.1). Further analysis revealed that embryonic development is arrested in double knockouts at 9.5 dpc as evidenced by the restoration of Mendelian inheritance (Table 5.2). At 10.5 dpc, Clic4 -/-;Clic1 -/- embryos

had already expired and presented pooled blood near the heart as well as cardiac edema

(Fig 5.1). Embryos of all other genotypes aside from the Clic4 -/-;Clic1 -/- genotype were present, and deletion of three Clic alleles did not appear to affect vascular development

(Fig 5.2). Whole mount analysis of 9.5 dpc showed that Clic4 +/+ ;Clic1 -/- embryos are indistinguishable from Clic4 +/+ ;Clic1 +/+ embryos (Fig 5.3). At 9.0 dpc, Clic4 -/-;Clic1 -/-

embryos presented qualitatively normal vasculature along with mild runting, which

suggests vessel regression or deterioration between 9.0 dpc and 9.5 dpc to yield the 9.5

dpc phenotype of lacking vasculature (Fig 5.4). These data would be consistent with a

vessel occlusion or leakage phenotype [97, 98], however further analysis is required to

determine the mechanism by which this phenotype is manifested. 113

Histological analysis of 9.5 dpc Clic4 -/-;Clic1 -/- embryo sections were consistent with whole mount phenotypes (Fig 5.5, 5.6, & 5.7). Stainings with an additional endothelial marker VEGFR-2 and Jag1 appeared fainter in Clic4 -/-;Clic1 -/- embryo

sections suggestive of reduced endothelial content and reduced Notch1 signaling,

respectively. VEGFR-2 staining also suggested thinner endothelium along the descending

aorta. Interestingly, Notch1 is an established regulator of angiogenesis [150], and Notch1

ligand Jag1 is expressed in the descending aorta [151]. This presents a possible pathway

in which CLIC4 is functioning to exhibit its effects on angiogenesis and aortic

development.

Smooth muscle content does not appear to be altered, and the myocardium

appears normal. Overall proliferation tends to be reduced in double knockout embryos,

consistent with in vitro data and embryo runting (Fig 5.8). Meanwhile, apoptosis tends to

be increased in double knockout embryos (Fig 5.9), possibly due to malnourished tissue

as a result of defective vascular development.

Several knockout lines exhibit lethal vascular defects around 9.5 dpc such as

knockouts of VEGF pathway members, Notch1 , Jag1 , Ang1/Tie2 , and ERK5 . These

knockouts and others that are similar have been well reviewed [152]. Indeed an

analogous example to our Clic4;Clic1 double knockout is provided by the Hey1;Hey2 double knockout, whereby knockout of both Hey1 and Hey2 result in embryonic lethality after 9.5 dpc with major vascular deficiencies while either Hey1 or Hey2 single

knockouts result in no major vascular phenotypes [153].

Other possibilities that may be responsible for Clic4 -/-;Clic1 -/- embryonic lethality are placental insufficiency or defects in chorioallantoic fusion [154, 155]. A defect in 114 either placental development or chorioallantoic fusion would result in decreased nourishment for the developing embryo, causing delayed or stunted development. To determine if placental and chorioallantoic development are affected by Clic4;Clic1

knockout, the placenta should be assessed for proper trophoblastic differentiation and

vascular development. The placental labyrinth where fetal and maternal blood supply are

exchanged should be closely examined. In addition, 8.5 dpc embryos should be assessed

for proper chorioallantoic fusion.

Although embryonic lethality of the Clic4;Clic1 double knockout precludes

analysis of global Clic4 and Clic1 function in adult mice, we had initially generated the

Clic4 knockout to be conditional. With a conditional Clic4 knockout in a Clic1 null background, we are able to conduct studies on how Clic4 may contribute to vessel

maintenance and remodeling in adult mice through crosses with inducible, endothelium-

specific Cre drivers, such as flk-1-Cre-ER mice that are Tamoxifen-inducible. Such a

system would also allow studies on how Clic4;Clic1 may contribute to pathological

angiogenesis, such as angiogenesis during tumor development. Furthermore, crosses with

an early endothelia-specific Cre driver such as VE-Cadherin-Cre mice would enable

analysis of Clic4 function specifically in vascular development without interference from

the effect of Clic4 knockout in other cell types or potential compensation by Clic1 [156].

Our identification of the requirement for Clic4 and Clic1 in embryonic

development extends our knowledge of CLIC4 and CLIC1 biological functions and

identifies a novel pathway in angiogenic development. While our study provides direct

evidence that Clic4 and Clic1 are both required for proper embryonic development, the

challenge now is to determine if Clic4 and Clic1 function in a complex, in the same 115 pathway, or merely concurrently. A co-immunoprecipitation using either CLIC4 or

CLIC1 to precipitate would determine if the two proteins function together in a complex.

Downstream effectors of CLIC4 and CLIC1 should be determined by comparing microarray or RNA-Seq results from wild type, Clic4 -/-, and Clic1 -/- cells [157].

Clustering by using gene ontology and bioinformatic approaches would reveal relevant targets [158]. Common targets of CLIC4 and CLIC1 could then be assessed and used to participate in rescue experiments. To determine functional redundancy between CLIC4 and CLIC1, rescue experiments could be performed in zebrafish, whereby Clic4 or Clic1 mRNA is used to rescue any Clic1 or Clic4 morpholino phenotype, respectively [159]. 116

5.8 Figures

Figure 5.1

Clic4 -/-;Clic1 +/+ Clic4 -/-;Clic1 -/- Clic4 -/-;Clic1 -/-

117

Figure 5.1 | Clic4;Clic1 double knockout embryos present cardiac edema at 10.5 dpc.

Clic4 -/-;Clic1 -/- (middle and right panels) embryos dissected at 10.5 dpc were expired as indicated by a lack of blood flow or beating heart. Compared with Clic4 single null control embryos (left panel), double knockouts are also appear growth retarded, have accumulated fluid in the pericardial sac (outlined by red boxes), and present blood pooling near the heart (highlighted by yellow arrows).

118

Figure 5.2

119

Figure 5.2 | Embryos missing three Clic alleles are represented at 10.5 dpc. With the exception of deleting all four Clic4 and Clic1 alleles, all other genotypes are represented at 10.5 dpc. Whole mount staining of 10.5 dpc embryos with endomucin indicate that vasculature is developing properly at this stage even in genotypic backgrounds that are missing three Clic alleles (middle and right panel). Clic4;Clic1 double heterozygotes (left panel) and Clic4 +/-;Clic1 -/- (middle) embryos are comparable both in size and developed vasculature. When compared with Clic4 +/-;Clic1 +/- littermates, 10.5 dpc Clic4 -/-;Clic1 +/- embryos suggest runting.

120

Figure 5.3

121

Figure 5.3 | Clic4 +/+ ;Clic1 -/- 9.5 dpc embryos are indistinguishable from

Clic4 +/+ ;Clic1 +/+ embryos. Embryos dissected at 9.5 dpc underwent endomucin whole

mount staining to view the vasculature. Vasculature in embryos of both genotypes

appears to be wholly present and comparable in architecture.

122

Figure 5.4 a. 9.5 dpc

b. 9.0 dpc

123

Figure 5.4 | In utero death for Clic4 -/-;Clic1 -/- embryos occurs at 9.5 dpc with growth retardation and an angiogenic defect. Endomucin whole mount staining was used to analyze the vasculature of developing embryos. Overall lighter endomucin staining of

Clic4 -/-;Clic1 -/- embryos indicate reduced endothelial content. (a) Clic4;Clic1 double

knockout embryos isolated at 9.5 dpc are runted, indicating general growth retardation.

Double knockout embryos also show lack of vasculature in the head as indicated by the

red arrows and along the intersomitic vessels as indicated by the red arrowheads. (b)

Double knockout embryos isolated at 9.0 dpc are also runted, but present intact

vasculature even in the head and along the intersomitic vessels.

124

Figure 5.5

125

Figure 5.5 | Endothelial content is reduced in Clic4 -/-;Clic1 -/- developing embryos.

DAPI (blue) was used to counterstain nuclei. Endothelium-specific marker CD31 (green) was used to analyze vasculature in the developing heart by histology. Lighter staining indicates reduced endothelial content, as highlighted by white arrows.

126

Figure 5.6

127

Figure 5.6 | Aortic development may be altered in Clic4 -/-;Clic1 -/- embryos. Serial sections of 9.5 dpc embryos were analyzed using endothelial marker VEGFR-2 (red) for immunofluorescence histology. Nuclei are marked in blue with DAPI counterstaining.

VEGFR-2 (red) staining appears fainter in Clic4 -/-;Clic1 -/- embryo sections, consistent

with lighter endothelial staining seen in whole mounts. (a) VEGFR-2 staining in the heart

is noticeably fainter in Clic4 -/-;Clic1 -/- sections. Magnification of the boxed area

highlights the descending aorta, which appears dilated in addition to being more weakly

stained in double knockouts. The white line is the same size in both photographs and

emphasizes the differing diameters of the two aortas. Double knockout embryos also

seem to possess thinner endothelium lining the aorta. (b) VEGFR-2 staining of the

intersomitic vessels (ISVs) is fainter in Clic4 -/-;Clic1 -/- embryos, consistent with the

fainter staining seen in the heart. Yellow arrows highlight ISV structures in the Clic4 -/-

;Clic1 -/- embryo, which appear disjointed compared with ISVs in control embryos that are highlighted by white arrows. 128

Figure 5.7

129

Figure 5.7 | Jag1 signaling may be altered in Clic4 -/-;Clic1 -/- embryos. Nuclei are

marked in blue with DAPI counterstaining. (a) Jag1 staining appears fainter in Clic4 -/-

;Clic1 -/- embryos, particularly in the heart and the otic vesicle. The otic vesicle is highlighted by a yellow arrow in the double knockout compared to the white arrow signifying the otic vesicle of control. (b) α-SMA labeling of smooth muscle cells in red suggest that smooth muscle content is comparable between Clic4 +/+ ;Clic1 -/- and Clic4 -/-

;Clic1 -/- embryos.

130

Figure 5.8

131

Figure 5.8 | Overall proliferation may be reduced in developing Clic4 -/-;Clic1 -/- embryos. Phospho-histone H3 (Ser10) staining (red) was used to identify proliferating cells. Sections are counterstained with DAPI to mark nuclei. (a) Clic4;Clic1 double knockout embryos show overall reduced proliferation as suggested by reduced staining at two different section depths. (b) Quantification of several sections stained with proliferation marker phosphor-histone H3 is consistent with qualitative assessment based on trends, however results do not reach significance (p > 0.05). Clic4 -/-;Clic1 -/- embryo sections tend to present a lower percentage of proliferating cells.

132

Figure 5.9

133

Figure 5.9 | Overall apoptosis may be increased in developing Clic4 -/-;Clic1 -/- embryos. CASP3 staining (red) was used to identify cells undergoing apoptosis in 9.5

dpc embryo sections. Sections are also stained with VEGFR-2 (green) to indicate

endothelial cells and counterstained with DAPI to mark nuclei (blue). (a) Clic4 -/-;Clic1 -/- embryo sections show increased apoptosis by qualitative assessment. Cells positive for both CASP3 and VEGFR-2 do not appear more numerous in Clic4 -/-;Clic1 -/- embryo

sections, indicating cells undergoing apoptosis are not necessarily endothelial. (b)

Quantification of cells undergoing apoptosis is consistent with observations. Clic4-/-

;Clic1 -/- embryo sections tend to have an increased percentage of apoptotic cells, however

results are not statistically significant (p > 0/05). 134

5.9 Tables

Table 5.1

n = 125, p = 0.56E-05 genotype expected actual Clic4-/-;Clic1-/- 8 0 Clic4+/+;Clic1-/- 8 8 Clic4+/+;Clic1+/+ 8 3 Clic4+/-;Clic1-/- 16 31 Clic4-/-;Clic1+/- 16 11 Clic4-/-;Clic1+/+ 8 6 Clic4+/-;Clic1+/- 32 42 Clic4+/+;Clic1+/- 16 16 Clic4+/-;Clic1+/+ 16 11

Table 5.1 | Inheritance from Clic4 +/-;Clic1 +/- intercrosses deviates from predicted

Mendelian ratios. A χ2 test for goodness of fit indicates that the genotypic ratio of

resultant progeny from Clic4 +/-;Clic1 +/- intercrosses deviated significantly from ratios predicted by Mendelian genetics. Expected numbers are noted in blue while actual numbers are in red. Notably, of the 125 resultant mice from Clic4 +/-;Clic1 +/- intercrosses,

8 were expected to be of the Clic4 -/-;Clic1 -/- genotype, however none were and this is

highlighted in green.

135

Table 5.2

9.5 dpc Clic4 +/ -;Clic1 -/- intercrosses

n = 48, p = 0.17

a. genotype expected actual

Clic4 -/-;Clic1 +/ - intercrosses Clic4 -/-;Clic1 -/- 12 17

n = 41, p = 0.72E-04 Clic4 +/ -;Clic1 -/- 24 23

genotype expected actual Clic4 +/+ ;Clic1 -/- 12 8

Clic4 -/-;Clic1 -/- 10.25 0

Clic4 -/-;Clic1 +/ - 20.5 25 9.5 dpc Clic4 -/-;Clic1 +/ - intercrosses

Clic4 -/-;Clic1 +/+ 10.25 16 n = 34, p = 0.84

genotype expected actual

Clic4 -/-;Clic1 -/- 8.5 8

Clic4 -/-;Clic1 +/ - 17 16

Clic4 -/-;Clic1 +/+ 8.5 10

Table 5.2 | Mendelian inheritance is restored at 9.5 dpc. (a) At birth, genotypes of pups resultant from Clic4 -/-;Clic1 -/- intercrosses deviate significantly from Mendelian

ratio with no double nulls present. (b) Genotypic ratios of embryos at 9.5 dpc are

consistent with predicted Mendelian ratios from the indicated matings. Double nulls are

present at 9.5 dpc. Expected numbers based on Mendelian predictions are in blue while

red represents actual numbers. Significance was established using χ2 statistical analysis. 136

Chapter Six: Conclusions and future directions

We have examined the role of CLIC4 and CLIC1 in developmental and

pathological angiogenesis. During the conception of this study, we determined that

CLIC4 and CLIC1 were the two most highly expressed CLIC protein members in human

umbilical vein endothelial cells, although CLIC2 and CLIC5 were also present. Due to

their evolutionary conservation across invertebrates and vertebrates as well as their high

expression in endothelial cells, we anticipated noteworthy roles for both CLIC4 and

CLIC1 in mediating angiogenesis. This study identifies CLIC4 and CLIC1 as players in

angiogenesis and represents the first documentation of the requirement for both Clic4 and

Clic1 during embryonic development.

Analysis of cultured primary endothelial cells with CLIC4 knockdown or

overexpression revealed that CLIC4 promotes endothelial cell growth, network and

branch formation, capillary sprouting, and lumen formation in vitro . Likewise, cultured

primary endothelial cells with CLIC1 knockdown exhibited reduced endothelial cell

growth, network and branch formation, and capillary sprouting in vitro . Unlike cultured

endothelial cells with CLIC4 knockdown, CLIC1 knockdown cells exhibited no effect on

lumen formation but did show reduced directed migration. These results are described in

chapters 2 and 4.

In analyzing the roles of CLIC4 and CLIC1 in angiogenesis further, we generated

a Clic4 knockout mouse line and acquired the Clic1 knockout mouse line [114]. First, we

characterized the Clic4 knockout mice as described in chapter 3. We found a significant difference between predicted inheritance and the resultant genotypes of mice from Clic4 137 heterozygote intercrosses at birth, suggesting partial prenatal lethality for Clic4 nulls.

Dissecting embryos at 9.5 dpc revealed that inheritance deviation is occurring after 9.5 dpc as embryos of all genotypes appeared largely normal, even upon endothelial whole mount staining. As Mendelian ratios are restored at this time point, it is unlikely that embryos are suffering developmental arrest prior to 9.5 dpc [160, 161]. To determine the point at which embryos are experiencing death, embryos should be genetically evaluated at various time points after 9.5 dpc during gestation and phenotypically evaluated at the time when genotypes deviate from Mendelian ratios.

Based on reported information and our own studies, we postulate that partial prenatal lethality of Clic4 knockouts may be due to a subtle defect in the hypoxic response or hematopoiesis. Mice deficient for the hypoxia-response element in the Vegf promoter ( Vegf / mice) exhibit partial in utero lethality during late gestation caused by hitherto unknown reasons [162]. Speculation implicates vascular defects as the cause of death, and a recent study has shown inferior hematopoietic stem cell survival in Vegf /

mice [163]. In addition, VEFG-A, hypoxia-inducible factor 1 α (HIF-1α), and angiopoietin-2 expression levels were found to be reduced in Clic4 -/- mice in ischemic

conditions, indicating CLIC4 regulation of these elements [141].

Postnatally, we noted that Clic4 null mice had reduced vascular density in both

the developing primary and deep retinal vascular plexuses. The increased presence of

delicate sprouts in Clic4 knockout retinas suggests that the vascular defect is due to a

delay in development or increased vascular regression due to lack of vessel patency [164,

165]. Since whole mount analysis for macrophage content suggested that Clic4 -/- retinas

have reduced macrophage content, and macrophages are known to be involved in 138 mediating anastomosis, the latter must be considered [82]. To determine if vessel regression is the culprit, whole mount staining for type IV collagen, which is a main component of vascular basement membranes, should be used to mark vessel sleeves in the event of vessel regression [166]. Sectioning retinas and staining for endothelial and mural cell markers will also reveal any differences in vessel patency and stability. In addition to analyzing the retina further, the vascular beds in the kidneys as well as the ovaries, endometrium, and mammary glands of the female reproductive system should be analyzed for similar phenotypes [167, 168]. Since angiogenesis during estrus or pregnancy is characterized as intense, studying vascular development here may be useful in studying the role of Clic4 in the induction of physiological angiogenesis during adulthood [169-171]. This would also provide insight into the mechanisms driving the surprising phenotype we noted regarding tumorigenesis in Clic4 knockout mice.

Using the Lewis lung carcinoma model for tumorigenesis, we found that Clic4 -/-

mice tended to present larger tumors than wild-type and Clic4 +/- and more lung metastases than their wild type counterparts in preliminary experiments. Histological analysis of tumor sections did not reveal any gross differences in endothelial or smooth muscle cell content, however this was not quantified. While endothelial and smooth muscle cell content is not significantly altered by loss of Clic4 , the architecture of the

tumor vessels may be altered (Section 3.6). Furthermore, histology for macrophages in

tumors isolated from Clic4 -/- mice recapitulated the reduced macrophage content

phenotype we noted in Clic4 -/- retinas. We suggest that the altered macrophage content in

Clic4 knockout-isolated tumors may be indicative of a shift in macrophage phenotype toward tumor-associated macrophages. 139

One alternative mechanism we did not address is the possibility that Clic4 may be altering the innate immune response to tumorigenesis. If loss of Clic4 indeed results in reduced macrophage content, having fewer activated macrophages at the site of initial transformation may inhibit the hosts immune response to combat transformed cells [172].

As loss of Clic4 has already been reported to result in a reduced inflammatory response to bacterial infection, there is merit to investigating the possibility of this latter mechanism. Moreover, CLIC4 is suggested to function in a positive feedback loop leading to enhanced TGF-β signaling [173, 174], lending further credence to this latter

hypothesis of a reduced innate immune system. Without CLIC4, TGF-β signaling was

shown to be abrogated. Because TGF-β signaling can enhance hematopoietic differentiation factor M-CSF production in mesenchymal cells [175], loss of Clic4 could result in an increase of undifferentiated hematopoietic stem cells [176, 177]. Since we designed our Clic4 knockout line to be conditional, one could assess the effects of Clic4 knockout on the macrophage or hematopoietic lineage [178, 179]. In addition, the role of

Clic4 in regulating TGF-β function could implicate it in a vast milieu of important biological activities such as cell cycle regulation [180-182] and extracellular matrix deposition, which would also affect angiogenesis [183].

Clic1 knockout mice did not exhibit gross angiogenic phenotypes despite the

drastic effects of CLIC1 knockdown on in vitro angiogenesis as described in chapter 4.

During the initial characterization of Clic1 -/- mice, a clotting defect was noted, typified by

prolonged bleeding times as well as lower rates of platelet activation and aggregation

despite having higher platelet counts [114]. Since platelets are known to mediate the

angiogenic response during wound healing and under inflammatory circumstances [184], 140 it is possible that Clic1 -/- mice may have an angiogenic defect in a wound healing model.

Platelets are also reported to sequester angiogenesis regulators [185], so the increase in

platelet count in Clic1 -/- mice may also have an effect on induced or pathological angiogenesis. In addition, CLIC1 is suggested to play a role in macrophage activation as

CLIC1 expression is upregulated upon phorbol-12-myristate-13 acetate (PMA) exposure in a monocyte cell line and A β activation of microglia [186, 187]. Inhibition of CLIC1 has also been shown to reduce the inflammatory response in microglia [187]. Thus, like

CLIC4, CLIC1 may also affect the initiation of the host innate immune response.

A central idea in our analyses is that CLIC4 and CLIC1 may be functionally redundant, as suggested by their similar structures, sequence, tissue expression, and their roles in endothelial cell morphogenesis in vitro , which we had determined [55, 56]. As

such, our work focused mainly on the roles of CLIC4 and CLIC1 as well as their

potential overlapping functions; however CLIC2 and CLIC5 were also found to be

expressed in endothelial cells albeit at low levels, and we have not examined their

potential roles in regulating angiogenesis.

To determine if CLIC4 and CLIC1 overlap functionally, we attempted to generate

Clic4 -/-;Clic1 -/- mice, but we were unable to obtain pups of that genotype. Embryonic

analysis revealed that the Clic4 -/-;Clic1 -/- genotype is lethal resulting in death at 9.5 dpc

with evidence of an angiogenic defect as described in chapter 5. Specifically, Clic4 -/-

;Clic1 -/- embryos analyzed at 9.5 dpc exhibit drastically reduced vasculature in the head and at the intersomitic vessels while Clic4 -/-;Clic1 -/- embryos at 9.0 dpc exhibit mostly

normal vascular patterning. These data imply that while vasculogenesis occurs normally, 141 angiogenic vessels are deficient and regress by 9.5 dpc causing death. As such, we present this model to explain our observations:

Model 6.1

In this model, we suggest that when vessels fail to form lumens, the lack of shear stress from flowing blood and the lack of mechanotransduction signals for the pruning of occluded vessels [97, 165, 188, 189]. Cords unable to undergo tubulogenesis due to a 142 defect in vesicular fusion should appear multivacuolated. We postulated that endothelial cells of occluded vessels then undergo apoptosis or migrate from the site of vessel formation along with other cell types that were initially recruited during pruning [98]. A vessel sleeve composed of newly synthesized basement membrane components remains as a shadow of the regressed vessel [190]. We postulated that Clic4;Clic1 double knockout may lead to such vessel occlusions resulting in the regression of small vessels and capillaries. Further analysis for vessel sleeves by staining for basement membrane components would reveal if regression is the cause of our 9.5 dpc Clic4 -/-;Clic1 -/- vascular phenotype. An increased number in apoptotic endothelial cells or migratory mural cells near the site of vessel sleeves would further indicate vessel regression.

Our whole mount histological analyses with various endothelial markers suggested reduced endothelial content in double knockout embryos. Histology also revealed a dilated descending aorta of Clic4 -/-;Clic1 -/- 9.5 dpc embryos with thinner lining

endothelium. Double knockout embryos at both 9.0 and 9.5 dpc present growth

retardation, consistent with histological analysis of 9.5 dpc embryos for proliferation.

This is also consistent with in vitro data showing that CLIC4 and CLIC1 knockdown

reduces endothelial cell growth. Two causes of death we did not examine are the

possibilities that chorioallantoic fusion is not occurring properly or there is placental

insufficiency [191-193]. Defects in either of these processes could lead to nutritional

insufficiency, which could lead to regressing vessels and cause death [98, 194-196].

While the Clic1 knockout is a straight knockout, knockout of Clic4 is conditional.

To study the contribution of Clic4 to developmental angiogenesis without the possibility

of functional compensation from Clic1 , Clic4 conditional knockouts can be generated in 143 a Clic1 -/- background. Crossing the conditional Clic4 allele to endothelia-specific and

macrophage-specific Cre drivers in a Clic1 null background would be advantageous in

evaluating the contributions of Clic4 to angiogenesis and macrophage function. Indeed, it

would also help reveal specific processes during which Clic4 and Clic1 concurrently

function. Generating a conditional Clic4 knockout in a Clic1 null background will also allow angiogenic studies in postnatal Clic4 -/-;Clic1 -/- mice by using inducible Cre drivers.

Such a model would be beneficial in studying pathological angiogenesis as well as

determining if Clic4 and Clic1 are required for the maintenance of existing vasculature.

All of our discoveries and the data presented here either directly or indirectly implicate both CLIC4 and CLIC1 in angiogenesis. With our current knowledge, we suggest this model for how CLIC4 and CLIC1 may be mediating angiogenesis:

Model 6.2

We suggest that CLIC4 and CLIC1 both promote endothelial proliferation, networking,

branching, and sprouting either additively or synergistically, and all these processes lead

to angiogenesis. CLIC4 plays additional roles in regulating tubulogenesis and recruiting 144 or activating macrophages to mediate anastomosis. Meanwhile, CLIC1 plays the additional roles promoting endothelial cell migration as well as activating platelets during the induction of angiogenesis. Furthermore, we present this model with postulations as to how CLIC4 and CLIC1 may be contributing to additional functions:

Model 6.2

We suggest that CLIC4 mediates tubulogenesis through its functions in vesicular trafficking or acidification as indicated by dotted lines, and we also suggest that CLIC1 is mediating endothelial migration through its interactions with the cell cytoskeleton.

CLIC1 may also play a role in macrophage activation during induction of angiogenesis.

Since the groundbreaking C. elegans experiments linking CLICs to tubulogenesis

[50], much research has been done in an effort to link CLICs with vertebrate vessel formation. Results from our in vitro studies on CLIC4 are consistent with the original notion that CLIC4 regulates tubulogenesis. CLIC4 knockdown resulted in reduced lumen formation and a significant increase in lumen formation was seen with CLIC4 overexpression. A future study could determine if CLIC4 localize to specific intracellular 145 vesicles and determine the mechanism by which it may be modulating tubulogenesis. To determine if vertebrate CLIC4 possesses any functions independent of its putative channel properties, functional studies on CLIC4 excluding its PTM or with a mutated

PTM should also be done as structural and gene disruption studies have suggested an enzymatic role for CLIC4 [46].

Our studies also attempt to address the possible functional redundancy between

CLIC4 and CLIC1 suggested by the initial studies done in C. elegans [50] in the creation of a Clic4;Clic1 double knockout mouse line. The absence of living postnatal Clic4;Clic1 double knockout mice during our studies is consistent with the notion that CLIC4 and

CLIC1 are functionally redundant to some extent since knockout of either Clic4 or Clic1 alone results in little to no phenotype. Double knockout embryos collected around prenatal time of death also present an angiogenic phenotype not present in either single knockout, further supporting the role of both CLIC4 and CLIC1acting either concurrently or in concert during angiogenesis. Again, CLIC4 and CLIC1 intracellular localization during angiogenic processes should be determined as a next step. The co-localization of

CLIC4 and CLIC1 during tube formation would further support the hypothesis of functional redundancy between the two proteins.

In summary, we have established that CLIC4 and CLIC1 are involved in regulating angiogenesis through their roles in promoting endothelial proliferation, branching, network formation and capillary sprouting; CLIC4 plays a role in promoting lumen formation; loss of CLIC4 may enhance tumorigenesis while altering macrophage function; CLIC1 plays a role in mediating endothelial migration; CLIC4 and CLIC1 function concurrently during developmental angiogenesis; and both CLIC4 and CLIC1 146 are required for embryonic development. Judah Folkman essentially pioneered angiogenesis research and since then, targeting angiogenesis has become an accepted method of combating tumorigenesis and vascular disease. Much remains to be elucidated regarding the mechanisms that drive angiogenesis, however our work has identified a novel pathway leading to angiogenesis. By continuing to discover the programs that govern angiogenesis both physiologically and pathologically, we will be able to better understand this important biological process as well as uncover new potential targets for generating angiogenic therapies. 147

Chapter Seven: Materials and methods

Ethics statement

No human subjects were used in this study. The collection of human umbilical venous endothelial cells (HUVEC) was approved by the Columbia University IRB

(AAAE4646.). Transgenic mice were generated and maintained under IACUC protocol

(AAAB0592). Tumor studies were conducted under IACUC protocol (AAAA9302).

Statistics

Unless otherwise noted, independent two-tailed Student’s t-tests were performed

on all quantified data to determine significant differences between two means. Pearson’s

χ2 testing was used to determine goodness of fit between observed genotype distributions and theoretical Mendelian distributions during in vivo analyses. P values less than 0.05

were considered statistically significant, and equal variances were assumed. All error bars

shown represent standard deviation.

Semi-quantitative PCR analysis

Total RNA was isolated from HUVEC using the RNeasy kit (Qiagen, Valencia,

CA), and first strand cDNA synthesis was done using the Thermo Scientific Verso cDNA

kit (Waltham, MA). PCR was then performed for human Clic1-6 for 30 cycles and an

annealing occurred at 56°C for Clic5 , 64°C for Clic1-4, and 66°C for Clic6 for 20 sec.

Elongation time was 40 sec at 72°C. Primers used are as follows: Clic1 forward 5’-

CCAAAAGGCGGACCGAGACA-3’, reverse 5’-AGGGTGAGCTCGTTGCCATCCA- 148

3’; Clic2 forward 5’-TGTCCCTTTTGCCAACGCCTTT-3’, reverse 5’-

CCTTGGAGGAGCCAGGGTTTGT-3’; Clic3 forward 5’-

TGCCAGCGGCTCTTCATGGT-3’, reverse 5’-CGCGGAGAACTTGTGGAAAACG-

3’; Clic4 forward 5’-GCCCCTCATCGAGCTCTTCGTCAA-3’, reverse 5’-

CAGGAGACCCCTCTCCAGTGCTTCA-3’; Clic5 forward 5’-

CAACACAGCGGGCATCGACA-3’, reverse 5’-CCACATGGAGCTTGGGCAACA-3’;

Clic6 forward 5’-GAAGAGGAATCCCCCGACAGCA-3’, reverse 5’-

TGTTCCGGGAGCCAGGTTCT-3’.

Cells and culture

HUVEC were isolated from human umbilical veins as previously described [197] and cultured on dishes coated with type I rat tail collagen (VWR, West Chester, PA) in

EGM-2 BulletKit medium (Lonza, Basel, Switzerland). Detroit 551 fibroblasts and 293T fibroblasts were purchased from ATCC (Manassas, VA) and cultured in Eagle’s

Minimum Essential Medium (EMEM) (ATCC, Manassas, VA) and Iscove’s Modified

Dulbecco’s Medium (Invitrogen, Carlsbad, CA), respectively. Both media were supplemented with 10% FBS (Invitrogen, Carlsbad, CA), and 0.01% Pen-Strep

(Invitrogen, Carlsbad, CA). Unless otherwise noted, cells were cultured under standard conditions in a humidified incubator at 37°C, 5% CO2.

Gene silencing and overexpression

Human CLIC4 shRNA-containing constructs were purchased from Sigma–

Aldrich (St. Louis, MO) and screened for significant Clic4 knockdown in HUVEC by 149 immunoblotting as described below. Oligonucleotides were provided in lentiviral vector pLKO.1-puro, which carries puromycin resistance allowing for selection of shRNA- expressing cells. Two shRNA were selected with the target sequences of 5 ′-

GCATATAGTGATGTAGCCAAA-3′ and 5 ′-GCCGTAATGTTGAACAGAATT-3′ denoted as constructs shRNA3 and shRNA5, respectively. pLKO.1-puro, denoted as plko sc, expressing a scrambled shRNA insert that does not target any known genes, served as a control.

A human Clic1 shRNA-containing construct in lentiviral vector pLKO.1-puro

(Sigma-Aldrich, St. Louis, MO) was used to provide Clic1 knockdown in HUVEC, which was confirmed by immunoblotting. The Clic1 -targetting shRNA possessed the target sequence of 5'-CCTGTTGCCAAAGTTACACAT-3'. Again, lentiviral vector pLKO.1-puro expressing scrambled shRNA was used as the control (Sigma-Aldrich, St,

Louis, MO).

Full-length human CLIC4 cDNA was prepared by RT-PCR using mRNA from

HMVEC (human dermal microvascular endothelial cells) with the following primers: 5 ′-

ATGGATCCGCCACCATGGCGTTGTCGATGCCGCTGAATG-3′ forward and 5 ′-

ATGTCGACTTACTTGGTGAGTCTTTTGG-3′ reverse. The sequence of the cloned

CLIC4 was determined and found to be identical with that of human CLIC4 (AF097330) except for a G to A nucleotide change in position 291, which did not affect the coding sequence. The full-length CLIC4 cDNA was cloned into lentiviral pCCL.pkg.wpre

(referred to as pCCL-CLIC4) and screened for significant Clic4 overexpression by immunoblotting. pCCL.pkg.wpre-expressing GFP, denoted as pccl, served as a control for this overexpressing line. 150

Lentivirus-mediated stable expression of shRNA and full-length constructs in HUVEC

For lentiviral gene transfer, the lentiviral vector pLKO.1 was used for shRNA knockdown lines while pCCL was used for the overexpressing HUVEC line. About

2.5 × 10 6 293T packaging cells were seeded and transfected with 3 µg pVSVG, 5 µg pMDLg/pRRE, 2.5 µg pRSV-Rev, and either 10 µg pLKO.1-Clic4 shRNA (CLIC4

knockdown), pLKO.1-Clic1 shRNA (CLIC1 knockdown), pLKO.1-scrambled shRNA

(knockdown control), pCCL-CLIC4 (CLIC4 overexpression), or pCCL-GFP

(overexpression control). Lentivirus-containing supernatants were collected 48 and 56 h

after transfection, passed through a 0.45 µm filter, and added to 1 × 10 6 low-passage

HUVECs. About 48 h after infection, pLKO.1-expressing HUVECs were selected with

puromycin (3 µg/ml) for 72 h and maintained with puromycin at 1.5 µg/ml.

Immunoblotting

HUVEC protein lysates were prepared in TENT lysis buffer (50 mM Tris pH 8.0,

2 mM EDTA, 150 mM NaCl, and 1% Triton X-100) containing Protease Inhibitor

Cocktail Set IV (EMD Chemicals, Inc., Gibbstown, NJ) prepared according to the manufacturer's protocol. Lysates were boiled for 5 min after addition of SDS and β- mercaptoethanol-containing sample buffer. Protein concentrations were determined using the Bradford protein assay (Bio-Rad Laboratories, Hercules, CA) according to the manufacturer's protocol, and sample volumes were adjusted to equivalent concentrations for equal protein loading into SDS-PAGE. Protein was then electroblotted onto nitrocellulose membrane and blocking occurred in 5% milk dissolved in PBST (1× PBS with 0.2× Tween20). Incubation of primary antibody (1:250 for CLIC1; 1:250 for 151

CLIC4; or 1:5000 for α-tubulin) was done in 2.5% milk dissolved in PBST, and

incubation of secondary antibody (1:5000 for both HRP-conjugated goat anti-rabbit and

goat anti-mouse) occurred in 2.5% milk in PBST. Protein bands were visualized using

Enhanced Chemiluminescence (GE Healthcare Bio-Sciences Corp., Piscataway, NJ)

according to the manufacturer's protocol.

Primary polyclonal rabbit anti-human CLIC1 (B121) antibody was a gift from

Mark Berryman at Ohio University College of Osteopathic Medicine (Athens, OH) [115].

Primary polyclonal rabbit anti-human CLIC4 antibody was purchased from Abcam Inc.

(Cambridge, MA) while primary monoclonal mouse anti-α-tubulin antibody was purchased from Sigma-Aldrich (St. Louis, MO). Secondary goat anti-rabbit and goat anti- mouse horseradish peroxidase (HRP)-conjugated antibodies were purchased from Sigma-

Aldrich (St. Louis, MO).

Cell viability and cell growth analysis

HUVEC were seeded at 3 × 10 4 cells/well of a 24-well plate and cultured in either

serum free medium (SFM) alone or SFM with 20 ng/mL epidermal growth factor (EGF)

for 48 h for cell viability assays (Invitrogen, Carlsbad, CA). Cell numbers after 48 h were

quantified using Cell Counting Kit-8 WST-8 (Dojindo Molecular Technologies,

Gaithersburg, MD) according to the manufacturer's protocol. Similarly for cell growth

assays, HUVEC were seeded at 1 × 10 4 cells/well of a 24-well plate and cultured in SFM

with 20 ng/mL EGF and 20 ng/mL recombinant human vascular endothelial growth

factor-A (rhVEGF) for 96 h (Invitrogen, Carlsbad, CA). Again, cell numbers were scored

using Cell Counting Kit-8 WST-8, and a calibration curve was generated following 152

Dojindo's protocol (Gaithersburg, MD). Assays were performed in triplicate and replicated at least five times.

Migration “scratch” analysis

To determine changes in directed cell migration, cells were seeded to confluence on collagen-coated 12-well plates at 1 × 10 6 cells/well in EGM-2 medium (Lonza, Basel,

Switzerland). 24 h post-seeding, cell monolayers were bisected along the diameter of each well with a 200 µL pipette tip, creating an open "scratch" or "wound" that was clear

of cells. The dislodged cells were removed by three washes with 1× PBS, EGM-2

medium was replaced, and cells were incubated under standard conditions. Migration into

the open area was documented at 0, 3, 6, 9, and 12 h post-scratching. Quantification was

done using TScratch software and performed according to the creator's protocol [116].

Experiments were done in triplicate and repeated at least five times.

Flow cytometry

1.5 × 10 5 HUVEC were incubated at 4°C on a rotator for 1 h with primary antibody (1:50

for each of α2, β1, α3, αVβ3, αVβ5, CD31, and without primary for negative controls) in

PCN (0.1% NaN 3, 0.001 M MgCl 2, 0.5% FBS in 1× PBS). Cells were incubated with secondary antibody (1:100 for APC-conjugated goat anti-mouse in PCN) on a rotator for

45 min at 4°C and kept on ice overnight. The BD FACSCalibur was used to perform flow cytometry according to the manufacturer's protocol and data analysis was performed using CD CellQuest Pro software (San Jose, CA). Experiments were repeated at least three times. Primary monoclonal mouse anti-human antibodies for integrin subunit chains 153

α2, β1, and α3 were purchased from BD Biosciences (San Jose, CA) and primary monoclonal mouse anti-human antibodies for integrins αVβ3 and αVβ5 were purchased from Millipore (Billerica, MA). Primary monoclonal mouse anti-human CD31 antibody was purchased from Dako (Carpinteria, CA). Secondary goat anti-mouse allophycocyanin (APC)-conjugated AffiniPure IgG antibody was obtained from Jackson

ImmunoResearch Laboratories, Inc. (West Grove, PA).

Network formation and branching analysis

HUVEC were cultured at standard conditions as a monolayer between two layers of porcine collagen gel (Wako USA, Richmond, VA) as previously described for the network formation assay [73]. Briefly, HUVEC were seeded between two layers of porcine collagen gel at 1 × 10 5 cells/well of a 24-well plate. Gels were cultured in SFM supplemented with 20 ng/mL EGF and 20 ng/mL rhVEGF for 96 h and photographically documented. Experiments were performed in triplicate and repeated at least five times.

Quantification for surface area was done by assessing collagen gel sandwiches

96 h post-seeding by applying MTT (Dojindo Molecular Technologies, Gaithersburg,

MD) for 3 h, photographing the networks, and using Image-Pro Plus (Media Cybernetics,

Bethesda, MD) software to calculate the surface area sum occupied by colored objects, which in this case would be the MTT-treated HUVECs. Branch points were quantified by counting the number of branch points per field of vision at 10× objective for knockdown cell lines and 5× objective for overexpression lines. One branch point was considered any focal intersection of two or more cords. For experiments involving mitomycin C treatment, cells were exposed to mitomycin C (Sigma–Aldrich, St. Louis, MO) at 154

25 µg/ml for 45 min prior to seeding. After 45 min of exposure to mitomycin C, cells

were used immediately for downstream applications. Mitomycin C was dissolved in

dH 2O and kept at 4°C as a 1 mg/ml stock solution. Accordingly, cells without mitomycin

C treatment received treatment with dH 2O.

Capillary-like sprouting fibrin bead assay and quantification

The capillary sprouting assay was performed as previously described [75] with two modifications: thrombin concentration and Detroit 551 fibroblast seeding. Briefly,

HUVEC and Detroit 551 fibroblasts (ATCC, Manassas, VA) were cultured in M199 with

10% FBS and 0.01% Pen-Strep (Invitrogen, Carlsbad, CA). HUVEC were attached to dextran-coated Cytodex 3 microcarrier beads (GE Healthcare Bio-Sciences Corp.,

Piscataway, NJ) at 400 HUVEC/bead. HUVEC-coated beads were then embedded at

~250 beads/well of a 24-well plate in a fibrin clot consisting of 2 mg/ml fibrinogen

(Sigma–Aldrich, St. Louis, MO), 0.15 U/ml aprotinin (Sigma–Aldrich, St. Louis, MO), and 0.0625 U/ml thrombin (Sigma–Aldrich, St. Louis, MO). After 24 h, Detroit 551 fibroblasts were seeded on top of the fibrin gel at 1.5 × 10 5 cells/well. Clots were cultured at standard conditions in EGM-2 medium and the assay was allowed to run for 11 days.

Experiments were performed in triplicate and replicated at least three times.

Quantification for sprout number/bead, branch point/sprout, and the ratio of lumen-containing sprouts to non-lumen containing cords was done by counting the number of sprouts, branch points, or lumen-containing sprouts for 300 beads for each cell line in one experiment using normal optical microscopy. One branch point was considered any focal point where at least two vessels intersect or where one vessel 155 divides. A lumen-containing sprout was tabulated as any sprout with a continuous lumen that occupied at least 50% of the sprout length.

Three-dimensional endothelial morphogenesis assay

Three-dimensional analysis of endothelial morphogenesis in forming tube-like structures was performed as previously reported [76]. HUVEC were suspended at 1 x 10 6 cells/mL of porcine collagen gel as prepared according to the manufacturer’s instructions

(Wako USA, Richmond, VA). 100 mL aliquots of the collagen/cell mixture were plated into one well of a 24-well plate and allowed to polymerize for 30 min at 37°C, 5% CO 2, and cultured in SFM supplemented with 40 ng/mL EGF (Invitrogen, Carlsbad, CA), 40 ng/mL bFGF (Invitrogen, Carlsbad, CA), and 40 ng/mL rhVEGF-A165 (Research

Diagnostics, Inc, Concord, MA). Photographs using light microscopy documented progress at 48 h and 72 h. Experiments were performed in triplicate and repeated at least three times.

Patch clamp assay for channel activity

HEK 293 cells were transfected with either pCCL-CLIC4 and a positive selection marker (CD8) or pCCL-GFP using Lipofectamine transfection according to the manufacturer’s protocol (Invitrogen, Carlsbad, CA). Cells are suspended as single whole- cell suspensions in a bath solution of 132 mM NaCl, 4.8 mM KCl, 10 mM HEPES, 5 mM glucose, 1.2 mM MgCl 2, and 2 mM CaCl 2 at pH 7.4. A pipette solution of 110 mM K-

Asp, 5 mM ATP-K2, 11 mM EGTA, 10 mM HEPES, 5.5 mM (100 nM free) CaCl 2, and 1

mM MgCl 2 at pH 7.3 was applied. The holding potential was kept at -100 mV, pulsing 156 from -100 mV to +100 mV at 20 mV increments for pulse durations of 1 sec. Ion channel blockers were not used in our analysis to detect all ion channel activity resultant from the transfections.

Targeting vector construction

Multiple cloning steps were applied in the construction of the Clic4 targeting vector. Subcloning of Clic4 exon 2 from bacterial artificial chromosome (BAC) RP23-

57P18 into pDrive vector (QiagenValencia, CA) was performed using PCR with forward primer 5’-AGCCCGGGGGGGGCATCTTCATCAGACCA-3’ and reverse 5’-

ACTCTAGACCCTCAAAGCCCCCACAGAGATCCA-3’. An Xma1 site was attached to the 5’ end of the forward primer while an Xba1 site was attached to the 5’ end of the reverse primer for subsequent subcloning purposes. Likewise, 3.2 kb upstream and 3.2 kb downstream homology arms were subcloned into separate pDrive vectors using the following primers: upstream homology arm forward 5’-

ACGCGGCCGCCTGGATGGCAGTGGCTCATGGCAGA-3’ and reverse 5’-

ACCTCGAGCAGCTTGTGGCGCTGTTTGGA-3’; downstream homology arm forward

5’-ACGTCGACTAGAGCCCCCTCCCAGCTAGCACA-3’ and reverse 5’-

ACGGTACCTGCCATAAGTCTGCCACCAC-3’. A NotI site and an XhoI site was engineered to the 5’ ends of the upstream homology arm forward and reverse primers, respectively. To the 5’ ends of the downstream homology arm forward and reverse primers, a SalI site and an Acc651 site was added, respectively.

The majority of the targeting allele was assembled by directional cloning into pBluescript II SK (Fermentas, Ontario, Canada) using the three pDrive constructs 157 containing Clic4 exon2, the upstream homology arm, and the downstream homology arm described above as well as pBS containing a loxP -flrted neomycin cassette referred to as

FNFL (pBS-FNFL) and pBS-LoxP, which contained a single loxP site. NotI was used for ligation of the 5’ end of the upstream homology arm into pBS while the XhoI site of the

3’ end was ligated to the SalI-digested 3’ end of FNFL from pBS-FNFL; the 5’ end of

FNFL was ligated into pBS using XmaI to yield pBS containing the upstream homology arm ligated to the neomycin cassette (pBS-5F). In a separate pBS, XmaI was used for insertion of the 5’ end of Clic4 exon 2, which was digested with XbaI on its 3’ end and ligated to the NheI-digested 3’ end of the single loxP site from pBS-LoxP; the 5’ end of the lone loxP was ligated into pBS by SalI digestion (pBS-EL). In a third pBS, the 5F fragment of pBS-5F and the EL fragment of pBS-EL were ligated into pBS cut with NotI and SalI after lifting them from their vectors using NotI and XmaI, and XmaI and SalI, respectively (pBS-5FEL). Finally, pBS-5FEL was cut with SalI and Acc651 for ligation of the downstream homology arm, which had been digested from its vector using SalI and Acc651. The final product was pBS containing the upstream homology arm, followed by the FNFL cassette, Clic4 exon 2, the single loxP site, and the downstream homology arm (pBS-5FEL3). The entire 5FEL3 product was then lifted from pBS using

NotI and Acc651 and ligated into pPNT1, which had also been digested with NotI and

Acc651. pPNT1 contained the negative selection marker HSV-tk, which would have been downstream of the 5FEL3 insert (pPNT1-5FEL3). NotI was used to linearize the final pPNT1-5FEL3 targeting construct for downstream applications.

In all cases, PCR was performed using AccuPrime Taq DNA Polymerase High

Fidelity (Invitrogen, Carlsbad, CA). Ligations were done using T4 ligase, bacterial 158 transformations occurred in DH10B competent cells, and minipreparation of DNA was done according to standard protocols (Life Technologies Corporation, Carlsbad, CA). All gel extractions of PCR products or restriction enzyme-digested products were done using the QIAquick Gel Extraction kit (Qiagen, Valencia, CA). Sequencing (Genewiz, South

Plainfield, NJ) and restriction mapping were used at each cloning step to confirm successful cloning. pPNT1, pBS-FNFL, and pBS-LoxP were gifts from Thomas Ludwig

(Columbia University Medical Center, New York, NY). The final product was subjected to restriction mapping and sequencing to confirm proper cloning of all essential elements using primers that include those used for cloning as well as the following: to sequence and screen for the FNFL cassette, forward 5’- CCATGACCACGGCAACTCCT-3’ and reverse 5’-CGAGATCAGCAGCCTCTGT-3’ (referred to hereafter as the FNFL primer set); to sequence and screen for the single loxP site, forward 5’-

ACCCGTGGCTGCATGTGGGTGAAT-3’ and reverse 5’-

TGTCTTTGGGGACGATGGGAACAGA-3’ (referred to hereafter as the LoxS primer set).

Generation of Clic4 fl and Clic4 - transgenic mice

The pPNT1-5FEL3 targeting construct was given to Victor Lin at Columbia

University Medical Center’s Comprehensive Cancer Transgenic Facility (New York,

NY) for generation of chimeric mice. Briefly, the targeting construct was linearized using

NotI and electroporated into CSL3 ES cells derived from 129S6/SvEvTac-Car3 mice.

After negative selection against HSV-tk and neomycin (G418) selection, 192 ES cell clones were screened for homologous recombination using Southern blot analysis to 159 confirm integration of the FNFL cassette. Correctly targeted ES cell clones were further screened for integration of the single loxP site by PCR using the LoxS primer set. From the eleven correctly targeted clones, two were chosen for microinjection into C56BL/6 host blastocysts. Resultant chimeras were crossed with C57BL/6 mice and germline transmission was detected by agouti coat color. Germline transmission was confirmed by

Southern blotting and PCR using LoxS primers. Chimeras were crossed with C57BL/6 mice to establish the Clic4 fl mutation in a hybrid 129/B6 background.

To generate Clic4 - mice, female Clic4l f/fl mice were crossed with male EllaCre

homozygous mice [198], resulting in mice heterozygous for Clic4 knockout. Clic4 +/-

mice were then backcrossed with C57BL/6 mice for eight generations to produce Clic4

heterozygotes in a pure B6 background. Clic4 +/- mice were subsequently intercrossed to

produce Clic4 global knockout mice in B6 background. Clic4- mice were genotyped by

PCR using the forward primer from the FNFL primer set and the reverse primer from the

LoxS primer set.

Southern blotting analysis

To perform Southern blotting, genomic DNA (gDNA) was isolated from ES cell clones and digested with PstI. The digested DNA was then gel electrophoresed in a

0.75% agarose gel containing ethidium bromide in 1X TAE buffer at 35 V. gDNA in the agarose gel was depurinated in 0.25 M HCl for 20 min at room temperature, washed in water, and denatured in 0.5 M Na OH and 1.5 M NaCl for 20 min at room temperature.

The gDNA was then transferred to a nitrocellulose membrane by capillary action, which was then allowed to air dry for 10 min and baked at 80°C for 20 min. Prehybridization of 160 the membrane in rapid hybridization buffer with denatured salmon sperm for 30 min at

65°C then occurred, followed by hybridization with the 32P-labeled probe for 2 hr at

65°C. To generate the Southern probe, which is located upstream of the upstream homology arm, forward 5’-TCTACACTAAAGGTAAATGG-3’ and reverse 5’-

GGAATGTATCAGCGGGTTTTTG-3’ primers were used for PCR on wild type gDNA.

The PCR product was gel purified, placed in pDrive, sequenced, and removed again by

EcoRI digestion prior to 32 P-labeling. The Prime-It II Primer Labeling Kit (Stratagene, La

Jolla, CA) was used to radiolabel the probe. Following hybridization, the membrane was washed three times in three different solutions all at 65°C for 10 min: (1) 2X SSC, 0.5%

SDS (2) 1X SSC, 0.5% SDS, and (3) 0.1X SSC, 0.1% SDS. The radiolabeled membrane was then exposed to film.

Isolation of mouse embryonic fibroblasts

13.5 dpc embryos were minced and trypsinized after dissection and removal of the head and liver. Cells were plated on gelatin-coated plates. Resulting fibroblasts were cultured in DMEM (Invitrogen, Carlsbad, CA) supplemented with 10% FBS (Invitrogen,

Carlsbad, CA) and 0.01% Pen-Strep (Invitrogen, Carlsbad, CA) at 37°C in 5% CO 2. Yolk sacs were used for genotyping.

Whole mount immunohistochemistry

Yolk sacs and extraembryonic membranes were removed from embryos during

dissection, and embryos were fixed overnight at 4°C in 4% PFA. Embryos were then

washed three times for 10 min each in PBS, once in 50% methanol (diluted in PBS) for 161

10 min, and three times in 100% methanol for 5 min each. Dehydrated embryos were then bleached for 5 hr in methanol:DMSO:H 2O2 (4:1:1) at room temperature for 5 hr.

Subsequent to rehydration by a 20-min wash in 50% methanol, three 15-min washes in

PBT (0.2% BSA, 0.5% Triton X-100 in PBS), two 5-min washes in PBSMT (2% milk,

0.5% Triton X-100 in PBS), three 15-min washes in PBSMT, and one 1-hr wash in

PBSMT, embryos were incubated with rat anti-mouse endomucin (1:200, Santa Cruz

Biotechnology, Inc., Santa Cruz, CA) in PBSMT overnight at 4°C with rocking. Embryos

were then washed in 4°C PBSMT twice for 5 min, thrice for 15 min, and then seven

times for 1 hr before incubation with HRP-conjugated goat anti-rat secondary antibody

(1:2000, Sigma-Aldrich, St. Louis, MO) at 4°C overnight with rocking.

To visualize staining, embryos were first washed with PBSMT at 4°C twice for 5

min, thrice for 15 min, and five times for 1 hr before two 5 min and three 15min washes

with PBT at room temperature. Embryos were then incubated for 15 min in DAB solution

(250 µg/mL DAB, 0.08% NiCl 2 in PBT) before the addition of H 2O2 at 1:1000. All

embryos within a matched set were allowed to incubate in the H 2O2 solution for the same amount of time, and the color reaction was stopped by rinsing embryos twice in PBT.

After postfixing overnight at 4°C in 4%PFA, embryos were washed three times with PBT and cryopreserved with 80% glycerol in a series (10 min room temperature wash with

25% glycerol, 10 min in 50% glycerol, and 10 min in 80% glycerol), embryos were analyzed using NIS-Elements software on a Nikon SMZ1000 (Nikon, Inc., Melville,

NY).

162

Retina whole mount analysis

After removing whole eyes from sacrificed animals, eyes were fixed for 1 hr at

4°C on a rotator in 4% paraformaldehyde (PFA). Eyes were then washed three times with

PBS and retinas were dissected. Retinas were permeabilized overnight at 4°C in 1% BSA and 0.5% Triton X-100 in 1X PBS and subsequently washed three times for 30 min each in PBLEC (1X PBS, 1% Triton X-100, 0.1 mM CaCl 2, 0.1 mM MgCl 2, 0.1 mM McCl 2,

pH 6.8) before being incubated overnight at 4°C with rabbit anti-mouse F4/80 primary

antibody (1:200, eBioscience, San Diego, CA) in PBLEC. After three 15 min washes in

PBLEC and three 15 min washes in PB/2 (PBLEC diluted 1:1 in PBS), retinas were

incubated overnight at 4°C with donkey anti-rat Alexa Fluor 594 secondary antibody

(1:750, Invitrogen, Carlsbad, CA) and FITC-conjugated isolectin-B4 (1:250, Sigma-

Aldrich, St. Louis, MO) in PBLEC. Retinas were washed four times for 15 min each in

PB/4 (PBLEC diluted 1:3 in PBS), once in PBS for 15 min, postfixed for 20 min in 4%

PFA at room temperature, washed twice in PBS, flat-mounted with VectaShield

mounting medium (Vector Labs, Burlingame, CA), and visualized using ACT-1 software

on a Nikon ECLIPSE E800 microscope (Nikon, Inc., Melville, NY).

Histology and immunofluorescence

Yolk sacs and extraembryonic membranes were removed from dissected embryos,

which were then fixed overnight at 4°C in 4% PFA with rocking. Embryos were then

washed three times with PBS, cryopreserved in 30% sucrose (in PBS) overnight at 4°C,

and embedded in Tissue-Tek OCT Compound (Sakura Finetek USA, Inc., Torrance, CA) 163 for sectioning. For tumor analysis, tumors were dissected and fresh-frozen in Tissue-Tek

OCT Compound for sectioning. Resultant slides were stored at -20°C until ready for use.

To analyze, slides were postfixed in -20°C acetone for 3 min and washed twice for 5 min each in 1X PBS. For hematoxylin and eosin staining, the standard protocol was followed [199]. For immunofluorescence, tissue sections were blocked in IHC diluent

(3% BSA and 0.02% donkey serum in PBS) for 1 hr at room temperature, and sections were incubated with primary antibody in IHC diluent overnight at 4°C. Primary antibodies used include rat anti-mouse CD31 (1:250, BD Pharmingen, San Diego, CA),

Cy3-conjugated mouse anti-mouse αSMA (1:1000, Sigma-Aldrich, St. Louis, MO), rabbit anti-mouse phosphor-histone H3 (Ser 10) (1:200, Millipore, Billerica, MA), goat anti-rat Jagged 1 (1:75, R&D Systems, Minneapolis, MN), goat anti-mouse VEGFR-2

(1:100, R&D Systems, Minneapolis, MN), rabbit anti-human caspase 3 (1:100, Cell

Signaling Technology, Danvers, MA), and rabbit anti-mouse F4/80 (1:200. eBioscience,

San Diego, CA). Slides were then washed twice for 5 min each in PBS and incubated with secondary antibody in IHC diluent for 30 min at room temperature. Secondary antibodies used include Alexa Fluor 594 and Alexa Fluor 488 ( both at 1:1000,

Invitrogen, Carlsbad, CA). After two 5 min washes in PBS, VectaShield with DAPI

(Vector Labs, Burlingame, CA) was used to mount coverslips and counterstain nuclei.

Tissue sections were visualized using ACT-1 software on a Nikon ECLIPSE E800 microscope (Nikon, Inc., Melville, NY). Quantification was done using Image J software

(NIH, Bethesda, MD) on five sections for each embryo of two matched sets. Positively stained cells were quantified relative to DAPI-stained nuclei and calculated as average percentages. 164

Bibliography

1. Tung, J.J., I.W. Tattersall, and J. Kitajewski, Tips, Stalks, Tubes: Notch-Mediated Cell Fate Determination and Mechanisms of Tubulogenesis during Angiogenesis. Cold Spring Harb Perspect Med, 2012. 2(2): p. a006601. 2. Folkman, J., Role of angiogenesis in tumor growth and metastasis. Semin Oncol, 2002. 29 (6 Suppl 16): p. 15-8. 3. Hanahan, D. and R.A. Weinberg, Hallmarks of cancer: the next generation. Cell, 2011. 144 (5): p. 646-74. 4. Risau, W., Mechanisms of angiogenesis. Nature, 1997. 386 (6626): p. 671-4. 5. Stratman, A.N., et al., Endothelial cell lumen and vascular guidance tunnel formation requires MT1-MMP-dependent proteolysis in 3-dimensional collagen matrices. Blood, 2009. 114 (2): p. 237-47. 6. Folkman, J., Tumor angiogenesis: therapeutic implications. N Engl J Med, 1971. 285 (21): p. 1182-6. 7. Folkman, J., Angiogenesis and apoptosis. Semin Cancer Biol, 2003. 13 (2): p. 159- 67. 8. Bergers, G. and L.E. Benjamin, Tumorigenesis and the angiogenic switch. Nat Rev Cancer, 2003. 3(6): p. 401-10. 9. Hogan, B.L. and P.A. Kolodziej, Organogenesis: molecular mechanisms of tubulogenesis. Nat Rev Genet, 2002. 3(7): p. 513-23. 10. Lubarsky, B. and M.A. Krasnow, Tube morphogenesis: making and shaping biological tubes. Cell, 2003. 112 (1): p. 19-28. 11. Dejana, E., Endothelial cell-cell junctions: happy together. Nat Rev Mol Cell Biol, 2004. 5(4): p. 261-70. 12. Bryant, D.M. and K.E. Mostov, From cells to organs: building polarized tissue. Nat Rev Mol Cell Biol, 2008. 9(11): p. 887-901. 13. Iruela-Arispe, M.L. and G.E. Davis, Cellular and molecular mechanisms of vascular lumen formation. Dev Cell, 2009. 16 (2): p. 222-31. 14. Zeeb, M., B. Strilic, and E. Lammert, Resolving cell-cell junctions: lumen formation in blood vessels. Curr Opin Cell Biol, 2010. 22 (5): p. 626-32. 15. Datta, A., D.M. Bryant, and K.E. Mostov, Molecular regulation of lumen morphogenesis. Curr Biol, 2011. 21 (3): p. R126-36. 16. Folkman, J. and C. Haudenschild, Angiogenesis in vitro. Nature, 1980. 288 (5791): p. 551-6. 17. Wolff, J.R. and T. Bar, 'Seamless' endothelia in brain capillaries during development of the rat's cerebral cortex. Brain Res, 1972. 41 (1): p. 17-24. 18. Davis, G.E. and C.W. Camarillo, An alpha 2 beta 1 integrin-dependent pinocytic mechanism involving intracellular vacuole formation and coalescence regulates capillary lumen and tube formation in three-dimensional collagen matrix. Exp Cell Res, 1996. 224 (1): p. 39-51. 19. Davis, G.E., W. Koh, and A.N. Stratman, Mechanisms controlling human endothelial lumen formation and tube assembly in three-dimensional extracellular matrices. Birth Defects Res C Embryo Today, 2007. 81 (4): p. 270-85. 20. Kamei, M., et al., Endothelial tubes assemble from intracellular vacuoles in vivo. Nature, 2006. 442 (7101): p. 453-6. 165

21. Pollack, A.L., R.B. Runyan, and K.E. Mostov, Morphogenetic mechanisms of epithelial tubulogenesis: MDCK cell polarity is transiently rearranged without loss of cell-cell contact during scatter factor/hepatocyte growth factor-induced tubulogenesis. Dev Biol, 1998. 204 (1): p. 64-79. 22. Parker, L.H., et al., The endothelial-cell-derived secreted factor Egfl7 regulates vascular tube formation. Nature, 2004. 428 (6984): p. 754-8. 23. Jin, S.W., et al., Cellular and molecular analyses of vascular tube and lumen formation in zebrafish. Development, 2005. 132 (23): p. 5199-209. 24. Wang, Y., et al., Moesin1 and Ve-cadherin are required in endothelial cells during in vivo tubulogenesis. Development, 2010. 137 (18): p. 3119-28. 25. Leung, B., G.J. Hermann, and J.R. Priess, Organogenesis of the Caenorhabditis elegans intestine. Dev Biol, 1999. 216 (1): p. 114-34. 26. Strilic, B., et al., The molecular basis of vascular lumen formation in the developing mouse aorta. Dev Cell, 2009. 17 (4): p. 505-15. 27. Herbert, S.P., et al., Arterial-venous segregation by selective cell sprouting: an alternative mode of blood vessel formation. Science, 2009. 326 (5950): p. 294-8. 28. Landry, D.W., et al., Purification and reconstitution of chloride channels from kidney and trachea. Science, 1989. 244 (4911): p. 1469-72. 29. Landry, D., et al., Molecular cloning and characterization of p64, a chloride channel protein from kidney microsomes. J Biol Chem, 1993. 268 (20): p. 14948- 55. 30. Landry, D.W., et al., Epithelial chloride channel. Development of inhibitory ligands. J Gen Physiol, 1987. 90 (6): p. 779-98. 31. Redhead, C.R., et al., A ubiquitous 64-kDa protein is a component of a chloride channel of plasma and intracellular membranes. Proc Natl Acad Sci U S A, 1992. 89 (9): p. 3716-20. 32. Valenzuela, S.M., et al., Molecular cloning and expression of a chloride ion channel of cell nuclei. J Biol Chem, 1997. 272 (19): p. 12575-82. 33. Duncan, R.R., et al., Rat brain p64H1, expression of a new member of the p64 chloride channel protein family in endoplasmic reticulum. J Biol Chem, 1997. 272 (38): p. 23880-6. 34. Howell, S., R.R. Duncan, and R.H. Ashley, Identification and characterisation of a homologue of p64 in rat tissues. FEBS Lett, 1996. 390 (2): p. 207-10. 35. Ashley, R.H., Challenging accepted ion channel biology: p64 and the CLIC family of putative intracellular anion channel proteins (Review). Mol Membr Biol, 2003. 20 (1): p. 1-11. 36. Littler, D.R., et al., The enigma of the CLIC proteins: Ion channels, redox proteins, enzymes, scaffolding proteins? FEBS Lett, 2010. 584 (10): p. 2093-101. 37. Berry, K.L. and O. Hobert, Mapping functional domains of chloride intracellular channel (CLIC) proteins in vivo. J Mol Biol, 2006. 359 (5): p. 1316-33. 38. Li, Y., et al., Trimeric structure of the wild soluble chloride intracellular ion channel CLIC4 observed in crystals. Biochem Biophys Res Commun, 2006. 343 (4): p. 1272-8. 39. Littler, D.R., et al., The intracellular chloride ion channel protein CLIC1 undergoes a redox-controlled structural transition. J Biol Chem, 2004. 279 (10): p. 9298-305. 166

40. Tulk, B.M., S. Kapadia, and J.C. Edwards, CLIC1 inserts from the aqueous phase into phospholipid membranes, where it functions as an anion channel. Am J Physiol Cell Physiol, 2002. 282 (5): p. C1103-12. 41. Warton, K., et al., Recombinant CLIC1 (NCC27) assembles in lipid bilayers via a pH-dependent two-state process to form chloride ion channels with identical characteristics to those observed in Chinese hamster ovary cells expressing CLIC1. J Biol Chem, 2002. 277 (29): p. 26003-11. 42. Littler, D.R., et al., Comparison of vertebrate and invertebrate CLIC proteins: the crystal structures of Caenorhabditis elegans EXC-4 and Drosophila melanogaster DmCLIC. Proteins, 2008. 71 (1): p. 364-78. 43. Singh, H., M.A. Cousin, and R.H. Ashley, Functional reconstitution of mammalian 'chloride intracellular channels' CLIC1, CLIC4 and CLIC5 reveals differential regulation by cytoskeletal actin. FEBS J, 2007. 274 (24): p. 6306-16. 44. Shiio, Y., et al., Quantitative proteomic analysis of myc-induced apoptosis: a direct role for Myc induction of the mitochondrial chloride ion channel, mtCLIC/CLIC4. J Biol Chem, 2006. 281 (5): p. 2750-6. 45. Suh, K.S., et al., The organellular chloride channel protein CLIC4/mtCLIC translocates to the nucleus in response to cellular stress and accelerates apoptosis. J Biol Chem, 2004. 279 (6): p. 4632-41. 46. Ponsioen, B., et al., Spatiotemporal regulation of chloride intracellular channel protein CLIC4 by RhoA. Mol Biol Cell, 2009. 20 (22): p. 4664-72. 47. Suh, K.S., et al., Antisense suppression of the chloride intracellular channel family induces apoptosis, enhances tumor necrosis factor {alpha}-induced apoptosis, and inhibits tumor growth. Cancer Res, 2005. 65 (2): p. 562-71. 48. Redhead, C., et al., Subcellular distribution and targeting of the intracellular chloride channel p64. Mol Biol Cell, 1997. 8(4): p. 691-704. 49. Bohman, S., et al., Proteomic analysis of vascular endothelial growth factor- induced endothelial cell differentiation reveals a role for chloride intracellular channel 4 (CLIC4) in tubular morphogenesis. J Biol Chem, 2005. 280 (51): p. 42397-404. 50. Berry, K.L., et al., A C. elegans CLIC-like protein required for intracellular tube formation and maintenance. Science, 2003. 302 (5653): p. 2134-7. 51. Suh, K.S., et al., CLIC4, an intracellular chloride channel protein, is a novel molecular target for cancer therapy. J Investig Dermatol Symp Proc, 2005. 10 (2): p. 105-9. 52. Money, T.T., et al., Expression and cellular localisation of chloride intracellular channel 3 in human placenta and fetal membranes. Placenta, 2007. 28 (5-6): p. 429-36. 53. Suh, K.S., et al., Reciprocal modifications of CLIC4 in tumor epithelium and stroma mark malignant progression of multiple human cancers. Clin Cancer Res, 2007. 13 (1): p. 121-31. 54. Ulmasov, B., et al., Tissue and subcellular distribution of CLIC1. BMC Cell Biol, 2007. 8: p. 8. 55. Tung, J.J. and J. Kitajewski, Chloride intracellular channel 1 functions in endothelial cell growth and migration. J Angiogenes Res, 2010. 2: p. 23. 167

56. Tung, J.J., et al., Chloride intracellular channel 4 is involved in endothelial proliferation and morphogenesis in vitro. Angiogenesis, 2009. 12 (3): p. 209-20. 57. Wegner, B., et al., CLIC5A, a component of the ezrin-podocalyxin complex in glomeruli, is a determinant of podocyte integrity. Am J Physiol Renal Physiol, 2010. 298 (6): p. F1492-503. 58. Singh, H. and R.H. Ashley, CLIC4 (p64H1) and its putative transmembrane domain form poorly selective, redox-regulated ion channels. Mol Membr Biol, 2007. 24 (1): p. 41-52. 59. Littler, D.R., et al., Crystal structure of the soluble form of the redox-regulated chloride ion channel protein CLIC4. FEBS J, 2005. 272 (19): p. 4996-5007. 60. Das, D.K. and N. Maulik, Antioxidant effectiveness in ischemia-reperfusion tissue injury. Methods Enzymol, 1994. 233 : p. 601-10. 61. Griendling, K.K., D. Sorescu, and M. Ushio-Fukai, NAD(P)H oxidase: role in cardiovascular biology and disease. Circ Res, 2000. 86 (5): p. 494-501. 62. Harfouche, R., et al., Roles of reactive oxygen species in angiopoietin-1/tie-2 receptor signaling. FASEB J, 2005. 19 (12): p. 1728-30. 63. Yamaoka-Tojo, M., et al., IQGAP1, a novel vascular endothelial growth factor receptor binding protein, is involved in reactive oxygen species--dependent endothelial migration and proliferation. Circ Res, 2004. 95 (3): p. 276-83. 64. Tang, F.Y., N. Nguyen, and M. Meydani, Green tea catechins inhibit VEGF- induced angiogenesis in vitro through suppression of VE-cadherin phosphorylation and inactivation of Akt molecule. Int J Cancer, 2003. 106 (6): p. 871-8. 65. Tang, F.Y. and M. Meydani, Green tea catechins and vitamin E inhibit angiogenesis of human microvascular endothelial cells through suppression of IL-8 production. Nutr Cancer, 2001. 41 (1-2): p. 119-25. 66. Cromer, B.A., et al., Structure of the Janus protein human CLIC2. J Mol Biol, 2007. 374 (3): p. 719-31. 67. Ulmasov, B., et al., Chloride intracellular channel protein-4 functions in angiogenesis by supporting acidification of vacuoles along the intracellular tubulogenic pathway. Am J Pathol, 2009. 174 (3): p. 1084-96. 68. Berryman, M.A. and J.R. Goldenring, CLIC4 is enriched at cell-cell junctions and colocalizes with AKAP350 at the centrosome and midbody of cultured mammalian cells. Cell Motil Cytoskeleton, 2003. 56 (3): p. 159-72. 69. Ferre, F., Quantitative or semi-quantitative PCR: reality versus myth. PCR Methods Appl, 1992. 2(1): p. 1-9. 70. Suginta, W., et al., Chloride intracellular channel protein CLIC4 (p64H1) binds directly to brain dynamin I in a complex containing actin, tubulin and 14-3-3 isoforms. Biochem J, 2001. 359 (Pt 1): p. 55-64. 71. Boerma, M., et al., Comparative expression profiling in primary and immortalized endothelial cells: changes in gene expression in response to hydroxy methylglutaryl-coenzyme A reductase inhibition. Blood Coagul Fibrinolysis, 2006. 17 (3): p. 173-80. 72. Ronnov-Jessen, L., et al., Differential expression of a chloride intracellular channel gene, CLIC4, in transforming growth factor-beta1-mediated conversion of fibroblasts to myofibroblasts. Am J Pathol, 2002. 161 (2): p. 471-80. 168

73. Funahashi, Y., et al., Sulfonamide derivative, E7820, is a unique angiogenesis inhibitor suppressing an expression of integrin alpha2 subunit on endothelium. Cancer Res, 2002. 62 (21): p. 6116-23. 74. Nakatsu, M.N. and C.C. Hughes, An optimized three-dimensional in vitro model for the analysis of angiogenesis. Methods Enzymol, 2008. 443 : p. 65-82. 75. Nakatsu, M.N., et al., Angiogenic sprouting and capillary lumen formation modeled by human umbilical vein endothelial cells (HUVEC) in fibrin gels: the role of fibroblasts and Angiopoietin-1. Microvasc Res, 2003. 66 (2): p. 102-12. 76. Davis, G.E., K.J. Bayless, and A. Mavila, Molecular basis of endothelial cell morphogenesis in three-dimensional extracellular matrices. Anat Rec, 2002. 268 (3): p. 252-75. 77. Jentsch, T.J., T. Maritzen, and A.A. Zdebik, Chloride channel diseases resulting from impaired transepithelial transport or vesicular function. J Clin Invest, 2005. 115 (8): p. 2039-46. 78. Fernandez-Salas, E., et al., mtCLIC/CLIC4, an organellular chloride channel protein, is increased by DNA damage and participates in the apoptotic response to p53. Mol Cell Biol, 2002. 22 (11): p. 3610-20. 79. Proutski, I., N. Karoulias, and R.H. Ashley, Overexpressed chloride intracellular channel protein CLIC4 (p64H1) is an essential component of novel plasma membrane anion channels. Biochem Biophys Res Commun, 2002. 297 (2): p. 317- 22. 80. Nishi, T. and M. Forgac, The vacuolar (H+)-ATPases--nature's most versatile proton pumps. Nat Rev Mol Cell Biol, 2002. 3(2): p. 94-103. 81. He, G., et al., Role of CLIC4 in the host innate responses to bacterial lipopolysaccharide. Eur J Immunol, 2011. 41 (5): p. 1221-30. 82. Fantin, A., et al., Tissue macrophages act as cellular chaperones for vascular anastomosis downstream of VEGF-mediated endothelial tip cell induction. Blood, 2010. 116 (5): p. 829-40. 83. Lee, O.H., et al., Sphingosine 1-phosphate induces angiogenesis: its angiogenic action and signaling mechanism in human umbilical vein endothelial cells. Biochem Biophys Res Commun, 1999. 264 (3): p. 743-50. 84. Rivera-Lopez, C.M., A.L. Tucker, and K.R. Lynch, Lysophosphatidic acid (LPA) and angiogenesis. Angiogenesis, 2008. 11 (3): p. 301-10. 85. Chuang, J.Z., et al., A 29 kDa intracellular chloride channel p64H1 is associated with large dense-core vesicles in rat hippocampal neurons. J Neurosci, 1999. 19 (8): p. 2919-28. 86. Edwards, J.C., B. Tulk, and P.H. Schlesinger, Functional expression of p64, an intracellular chloride channel protein. J Membr Biol, 1998. 163 (2): p. 119-27. 87. Chuang, J.Z., S.Y. Chou, and C.H. Sung, Chloride intracellular channel 4 is critical for the epithelial morphogenesis of RPE cells and retinal attachment. Mol Biol Cell, 2010. 21 (17): p. 3017-28. 88. Fernandez-Salas, E., et al., p53 and tumor necrosis factor alpha regulate the expression of a mitochondrial chloride channel protein. J Biol Chem, 1999. 274 (51): p. 36488-97. 89. Flicek, P., et al., Ensembl 2011. Nucleic Acids Res, 2011. 39 (Database issue): p. D800-6. 169

90. Lakso, M., et al., Efficient in vivo manipulation of mouse genomic sequences at the zygote stage. Proc Natl Acad Sci U S A, 1996. 93 (12): p. 5860-5. 91. Uemura, A., et al., Angiogenesis in the mouse retina: a model system for experimental manipulation. Exp Cell Res, 2006. 312 (5): p. 676-83. 92. Gerhardt, H., et al., VEGF guides angiogenic sprouting utilizing endothelial tip cell filopodia. J Cell Biol, 2003. 161 (6): p. 1163-77. 93. Checchin, D., et al., Potential role of microglia in retinal blood vessel formation. Invest Ophthalmol Vis Sci, 2006. 47 (8): p. 3595-602. 94. Outtz, H.H., et al., Notch1 controls macrophage recruitment and Notch signaling is activated at sites of endothelial cell anastomosis during retinal angiogenesis in mice. Blood, 2011. 118 (12): p. 3436-3439. 95. Sunderkotter, C., et al., Macrophages and angiogenesis. J Leukoc Biol, 1994. 55 (3): p. 410-22. 96. Savai, R., et al., Analysis of tumor vessel supply in Lewis lung carcinoma in mice by fluorescent microsphere distribution and imaging with micro- and flat-panel computed tomography. Am J Pathol, 2005. 167 (4): p. 937-46. 97. Lobov, I.B., et al., The Dll4/Notch pathway controls postangiogenic blood vessel remodeling and regression by modulating vasoconstriction and blood flow. Blood, 2011. 117 (24): p. 6728-37. 98. Baffert, F., et al., Cellular changes in normal blood capillaries undergoing regression after inhibition of VEGF signaling. Am J Physiol Heart Circ Physiol, 2006. 290 (2): p. H547-59. 99. Bach, J.P., et al., Role of MIF in inflammation and tumorigenesis. Oncology, 2008. 75 (3-4): p. 127-33. 100. Nabeyama, A., et al., xCT deficiency accelerates chemically induced tumorigenesis. Proc Natl Acad Sci U S A, 2010. 107 (14): p. 6436-41. 101. Sica, A., et al., Tumour-associated macrophages are a distinct M2 polarised population promoting tumour progression: potential targets of anti-cancer therapy. Eur J Cancer, 2006. 42 (6): p. 717-27. 102. Biswas, S.K., A. Sica, and C.E. Lewis, Plasticity of macrophage function during tumor progression: regulation by distinct molecular mechanisms. J Immunol, 2008. 180 (4): p. 2011-7. 103. Sica, A., Role of tumour-associated macrophages in cancer-related inflammation. Exp Oncol, 2010. 32 (3): p. 153-8. 104. Mantovani, A., C. Garlanda, and M. Locati, Macrophage diversity and polarization in atherosclerosis: a question of balance. Arterioscler Thromb Vasc Biol, 2009. 29 (10): p. 1419-23. 105. Sica, A. and V. Bronte, Altered macrophage differentiation and immune dysfunction in tumor development. J Clin Invest, 2007. 117 (5): p. 1155-66. 106. Valenzuela, S.M., et al., The nuclear chloride ion channel NCC27 is involved in regulation of the cell cycle. J Physiol, 2000. 529 Pt 3 : p. 541-52. 107. Dulhunty, A., et al., The glutathione transferase structural family includes a nuclear chloride channel and a ryanodine receptor calcium release channel modulator. J Biol Chem, 2001. 276 (5): p. 3319-23. 170

108. Harrop, S.J., et al., Crystal structure of a soluble form of the intracellular chloride ion channel CLIC1 (NCC27) at 1.4-A resolution. J Biol Chem, 2001. 276 (48): p. 44993-5000. 109. Cromer, B.A., et al., Expression, purification, crystallization and preliminary X- ray diffraction analysis of chloride intracellular channel 2 (CLIC2). Acta Crystallogr Sect F Struct Biol Cryst Commun, 2007. 63 (Pt 11): p. 961-3. 110. Suh, K.S., et al., CLIC4 mediates and is required for Ca2+-induced keratinocyte differentiation. J Cell Sci, 2007. 120 (Pt 15): p. 2631-40. 111. Singh, H. and R.H. Ashley, Redox regulation of CLIC1 by cysteine residues associated with the putative channel pore. Biophys J, 2006. 90 (5): p. 1628-38. 112. Tulk, B.M. and J.C. Edwards, NCC27, a homolog of intracellular Cl- channel p64, is expressed in brush border of renal proximal tubule. Am J Physiol, 1998. 274 (6 Pt 2): p. F1140-9. 113. Ausprunk, D.H. and J. Folkman, Migration and proliferation of endothelial cells in preformed and newly formed blood vessels during tumor angiogenesis. Microvasc Res, 1977. 14 (1): p. 53-65. 114. Qiu, M.R., et al., Generation and characterization of mice with null mutation of the chloride intracellular channel 1 gene. Genesis, 2010. 48 (2): p. 127-36. 115. Berryman, M. and A. Bretscher, Identification of a novel member of the chloride intracellular channel gene family (CLIC5) that associates with the actin cytoskeleton of placental microvilli. Mol Biol Cell, 2000. 11 (5): p. 1509-21. 116. Geback, T., et al., TScratch: a novel and simple software tool for automated analysis of monolayer wound healing assays. Biotechniques, 2009. 46 (4): p. 265- 74. 117. San Antonio, J.D., et al., A key role for the integrin alpha2beta1 in experimental and developmental angiogenesis. Am J Pathol, 2009. 175 (3): p. 1338-47. 118. Mitchell, K., et al., Alpha3beta1 integrin in epidermis promotes wound angiogenesis and keratinocyte-to-endothelial-cell crosstalk through the induction of MRP3. J Cell Sci, 2009. 122 (Pt 11): p. 1778-87. 119. Brooks, P.C., et al., Integrin alpha v beta 3 antagonists promote tumor regression by inducing apoptosis of angiogenic blood vessels. Cell, 1994. 79 (7): p. 1157-64. 120. Kumar, C.C., et al., Inhibition of angiogenesis and tumor growth by SCH221153, a dual alpha(v)beta3 and alpha(v)beta5 integrin receptor antagonist. Cancer Res, 2001. 61 (5): p. 2232-8. 121. Silva, R., et al., Integrins: the keys to unlocking angiogenesis. Arterioscler Thromb Vasc Biol, 2008. 28 (10): p. 1703-13. 122. Wang, J.W., et al., Identification of metastasis-associated proteins involved in gallbladder carcinoma metastasis by proteomic analysis and functional exploration of chloride intracellular channel 1. Cancer Lett, 2009. 281 (1): p. 71- 81. 123. Cao, L., et al., Chloride channels and transporters in human corneal epithelium. Exp Eye Res, 2010. 90 (6): p. 771-9. 124. Soroceanu, L., T.J. Manning, Jr., and H. Sontheimer, Modulation of glioma cell migration and invasion using Cl(-) and K(+) ion channel blockers. J Neurosci, 1999. 19 (14): p. 5942-54. 171

125. Gruber, A.D. and B.U. Pauli, Tumorigenicity of human breast cancer is associated with loss of the Ca2+-activated chloride channel CLCA2. Cancer Res, 1999. 59 (21): p. 5488-91. 126. Stoychev, S.H., et al., Structural dynamics of soluble chloride intracellular channel protein CLIC1 examined by amide hydrogen-deuterium exchange mass spectrometry. Biochemistry, 2009. 48 (35): p. 8413-21. 127. Fanucchi, S., R.J. Adamson, and H.W. Dirr, Formation of an unfolding intermediate state of soluble chloride intracellular channel protein CLIC1 at acidic pH. Biochemistry, 2008. 47 (44): p. 11674-81. 128. Shanks, R.A., et al., AKAP350 at the Golgi apparatus. II. Association of AKAP350 with a novel chloride intracellular channel (CLIC) family member. J Biol Chem, 2002. 277 (43): p. 40973-80. 129. Hood, J.D. and D.A. Cheresh, Role of integrins in cell invasion and migration. Nat Rev Cancer, 2002. 2(2): p. 91-100. 130. Duband, J.L., et al., Neural crest cell locomotion induced by antibodies to beta 1 integrins. A tool for studying the roles of substratum molecular avidity and density in migration. J Cell Sci, 1991. 98 ( Pt 4) : p. 517-32. 131. Palecek, S.P., et al., Integrin-ligand binding properties govern cell migration speed through cell-substratum adhesiveness. Nature, 1997. 385 (6616): p. 537-40. 132. Kirchhofer, D., et al., Alpha 2 beta 1 integrins from different cell types show different binding specificities. J Biol Chem, 1990. 265 (2): p. 615-8. 133. Pasco, S., et al., Control of melanoma cell invasion by type IV collagen. Cancer Detect Prev, 2005. 29 (3): p. 260-6. 134. Stupack, D.G., et al., Apoptosis of adherent cells by recruitment of caspase-8 to unligated integrins. J Cell Biol, 2001. 155 (3): p. 459-70. 135. Brown, E. and N. Hogg, Where the outside meets the inside: integrins as activators and targets of signal transduction cascades. Immunol Lett, 1996. 54 (2- 3): p. 189-93. 136. Caswell, P.T. and J.C. Norman, Integrin trafficking and the control of cell migration. Traffic, 2006. 7(1): p. 14-21. 137. Ferrara, N. and T. Davis-Smyth, The biology of vascular endothelial growth factor. Endocr Rev, 1997. 18 (1): p. 4-25. 138. Klagsbrun, M. and S. Soker, VEGF/VPF: the angiogenesis factor found? Curr Biol, 1993. 3(10): p. 699-702. 139. Carmeliet, P., Basic Concepts of (Myocardial) Angiogenesis: Role of Vascular Endothelial Growth Factor and Angiopoietin. Curr Interv Cardiol Rep, 1999. 1(4): p. 322-335. 140. Dvorak, H.F., et al., Distribution of vascular permeability factor (vascular endothelial growth factor) in tumors: concentration in tumor blood vessels. J Exp Med, 1991. 174 (5): p. 1275-8. 141. Chalothorn, D., et al., Chloride intracellular channel-4 is a determinant of native collateral formation in skeletal muscle and brain. Circ Res, 2009. 105 (1): p. 89- 98. 142. Sarnoff, S.J. and E. Berglund, Neurohemodynamics of pulmonary edema. IV. Effect of systemic vasoconstriction and subsequent vasodilation on flow and 172

pressures in systemic and pulmonary vascular beds. Am J Physiol, 1952. 170 (3): p. 588-600. 143. Sarnoff, S.J. and L.C. Sarnoff, Neurohemodynamics of pulmonary edema. II. The role of sympathetic pathways in the elevation of pulmonary and stemic vascular pressures following the intracisternal injection of fibrin. Circulation, 1952. 6(1): p. 51-62. 144. Morgan, S.M., et al., Biochemical characterization and molecular cloning of a novel endothelial-specific sialomucin. Blood, 1999. 93 (1): p. 165-75. 145. Kuhn, A., et al., Expression of endomucin, a novel endothelial sialomucin, in normal and diseased human skin. J Invest Dermatol, 2002. 119 (6): p. 1388-93. 146. Brachtendorf, G., et al., Early expression of endomucin on endothelium of the mouse embryo and on putative hematopoietic clusters in the dorsal aorta. Dev Dyn, 2001. 222 (3): p. 410-9. 147. Carmeliet, P., Angiogenesis in life, disease and medicine. Nature, 2005. 438 (7070): p. 932-6. 148. Lewis, A.K., et al., Distinct expression patterns of notch family receptors and ligands during development of the mammalian inner ear. Mech Dev, 1998. 78 (1- 2): p. 159-63. 149. Kiernan, A.E., et al., The Notch ligand Jagged1 is required for inner ear sensory development. Proc Natl Acad Sci U S A, 2001. 98 (7): p. 3873-8. 150. Gridley, T., Notch signaling in vascular development and physiology. Development, 2007. 134 (15): p. 2709-18. 151. Loomes, K.M., et al., The expression of Jagged1 in the developing mammalian heart correlates with cardiovascular disease in Alagille syndrome. Hum Mol Genet, 1999. 8(13): p. 2443-9. 152. Thurston, G., Role of Angiopoietins and Tie receptor tyrosine kinases in angiogenesis and lymphangiogenesis. Cell Tissue Res, 2003. 314 (1): p. 61-8. 153. Fischer, A., et al., The Notch target genes Hey1 and Hey2 are required for embryonic vascular development. Genes Dev, 2004. 18 (8): p. 901-11. 154. Rossant, J. and J.C. Cross, Placental development: lessons from mouse mutants. Nat Rev Genet, 2001. 2(7): p. 538-48. 155. Cross, J.C., Z. Werb, and S.J. Fisher, Implantation and the placenta: key pieces of the development puzzle. Science, 1994. 266 (5190): p. 1508-18. 156. Alva, J.A., et al., VE-Cadherin-Cre-recombinase transgenic mouse: a tool for lineage analysis and gene deletion in endothelial cells. Dev Dyn, 2006. 235 (3): p. 759-67. 157. Wang, Z., M. Gerstein, and M. Snyder, RNA-Seq: a revolutionary tool for transcriptomics. Nat Rev Genet, 2009. 10 (1): p. 57-63. 158. Ashburner, M., et al., Gene ontology: tool for the unification of biology. The Gene Ontology Consortium. Nat Genet, 2000. 25 (1): p. 25-9. 159. Bill, B.R., et al., A primer for morpholino use in zebrafish. Zebrafish, 2009. 6(1): p. 69-77. 160. Papaioannou, V.E. and R. Behringer, Mouse phenotypes : a handbook of mutation analysis . 2005, Cold Spring Harbor, N.Y.: Cold Spring Harbor Laboratory Press. x, 235 p. 173

161. Turgeon, B. and S. Meloche, Interpreting neonatal lethal phenotypes in mouse mutants: insights into gene function and human diseases. Physiol Rev, 2009. 89 (1): p. 1-26. 162. Oosthuyse, B., et al., Deletion of the hypoxia-response element in the vascular endothelial growth factor promoter causes motor neuron degeneration. Nat Genet, 2001. 28 (2): p. 131-8. 163. Rehn, M., et al., Hypoxic induction of vascular endothelial growth factor regulates murine hematopoietic stem cell function in the low-oxygenic niche. Blood, 2011. 118 (6): p. 1534-43. 164. Fruttiger, M., Development of the retinal vasculature. Angiogenesis, 2007. 10 (2): p. 77-88. 165. Benjamin, L.E., I. Hemo, and E. Keshet, A plasticity window for blood vessel remodelling is defined by pericyte coverage of the preformed endothelial network and is regulated by PDGF-B and VEGF. Development, 1998. 125 (9): p. 1591-8. 166. Kalluri, R., Basement membranes: structure, assembly and role in tumour angiogenesis. Nat Rev Cancer, 2003. 3(6): p. 422-33. 167. Kang, D.H., et al., Impaired angiogenesis in the remnant kidney model: I. Potential role of vascular endothelial growth factor and thrombospondin-1. J Am Soc Nephrol, 2001. 12 (7): p. 1434-47. 168. Augustin, H.G., Angiogenesis in the female reproductive system. EXS, 2005(94): p. 35-52. 169. Reynolds, L.P., A.T. Grazul-Bilska, and D.A. Redmer, Angiogenesis in the female reproductive organs: pathological implications. Int J Exp Pathol, 2002. 83 (4): p. 151-63. 170. Reynolds, L.P., S.D. Killilea, and D.A. Redmer, Angiogenesis in the female reproductive system. FASEB J, 1992. 6(3): p. 886-92. 171. Findlay, J.K., Angiogenesis in reproductive tissues. J Endocrinol, 1986. 111 (3): p. 357-66. 172. Lamagna, C., M. Aurrand-Lions, and B.A. Imhof, Dual role of macrophages in tumor growth and angiogenesis. J Leukoc Biol, 2006. 80 (4): p. 705-13. 173. Shukla, A., et al., TGF-beta signalling is regulated by Schnurri-2-dependent nuclear translocation of CLIC4 and consequent stabilization of phospho-Smad2 and 3. Nat Cell Biol, 2009. 11 (6): p. 777-84. 174. Shukla, A. and S.H. Yuspa, CLIC4 and Schnurri-2: A dynamic duo in TGF-beta signaling with broader implications in cellular homeostasis and disease. Nucleus, 2010. 1(2): p. 144-9. 175. Takaishi, T., et al., TGF-beta-induced macrophage colony-stimulating factor gene expression in various mesenchymal cell lines. Am J Physiol, 1994. 267 (1 Pt 1): p. C25-31. 176. Ogawa, M., Differentiation and proliferation of hematopoietic stem cells. Blood, 1993. 81 (11): p. 2844-53. 177. Keller, G., et al., Hematopoietic commitment during embryonic stem cell differentiation in culture. Mol Cell Biol, 1993. 13 (1): p. 473-86. 178. Clausen, B.E., et al., Conditional gene targeting in macrophages and granulocytes using LysMcre mice. Transgenic Res, 1999. 8(4): p. 265-77. 174

179. de Boer, J., et al., Transgenic mice with hematopoietic and lymphoid specific expression of Cre. Eur J Immunol, 2003. 33 (2): p. 314-25. 180. Gordon, K.J. and G.C. Blobe, Role of transforming growth factor-beta superfamily signaling pathways in human disease. Biochim Biophys Acta, 2008. 1782 (4): p. 197-228. 181. Elliott, R.L. and G.C. Blobe, Role of transforming growth factor Beta in human cancer. J Clin Oncol, 2005. 23 (9): p. 2078-93. 182. Blobe, G.C., W.P. Schiemann, and H.F. Lodish, Role of transforming growth factor beta in human disease. N Engl J Med, 2000. 342 (18): p. 1350-8. 183. Roberts, A.B., B.K. McCune, and M.B. Sporn, TGF-beta: regulation of extracellular matrix. Kidney Int, 1992. 41 (3): p. 557-9. 184. Kisucka, J., et al., Platelets and platelet adhesion support angiogenesis while preventing excessive hemorrhage. Proc Natl Acad Sci U S A, 2006. 103 (4): p. 855-60. 185. Klement, G.L., et al., Platelets actively sequester angiogenesis regulators. Blood, 2009. 113 (12): p. 2835-42. 186. Kanno, T. and T. Takishima, Chloride and potassium channels in U937 human monocytes. J Membr Biol, 1990. 116 (2): p. 149-61. 187. Novarino, G., et al., Involvement of the intracellular ion channel CLIC1 in microglia-mediated beta-amyloid-induced neurotoxicity. J Neurosci, 2004. 24 (23): p. 5322-30. 188. Bongrazio, M., et al., Shear stress modulates the expression of thrombospondin-1 and CD36 in endothelial cells in vitro and during shear stress-induced angiogenesis in vivo. Int J Immunopathol Pharmacol, 2006. 19 (1): p. 35-48. 189. Owen, M.R., et al., Angiogenesis and vascular remodelling in normal and cancerous tissues. J Math Biol, 2009. 58 (4-5): p. 689-721. 190. Jaffe, E.A., et al., Synthesis of basement membrane collagen by cultured human endothelial cells. J Exp Med, 1976. 144 (1): p. 209-25. 191. Gurtner, G.C., et al., Targeted disruption of the murine VCAM1 gene: essential role of VCAM-1 in chorioallantoic fusion and placentation. Genes Dev, 1995. 9(1): p. 1-14. 192. Morin-Kensicki, E.M., et al., Defects in yolk sac vasculogenesis, chorioallantoic fusion, and embryonic axis elongation in mice with targeted disruption of Yap65. Mol Cell Biol, 2006. 26 (1): p. 77-87. 193. Roberts, A.W., et al., Placental defects and embryonic lethality in mice lacking suppressor of cytokine signaling 3. Proc Natl Acad Sci U S A, 2001. 98 (16): p. 9324-9. 194. Shaut, C.A., et al., HOXA13 Is essential for placental vascular patterning and labyrinth endothelial specification. PLoS Genet, 2008. 4(5): p. e1000073. 195. Brown, W.R., A review of string vessels or collapsed, empty basement membrane tubes. J Alzheimers Dis, 2010. 21 (3): p. 725-39. 196. Im, E. and A. Kazlauskas, New insights regarding vessel regression. Cell Cycle, 2006. 5(18): p. 2057-9. 197. Jaffe, E.A., et al., Culture of human endothelial cells derived from umbilical veins. Identification by morphologic and immunologic criteria. J Clin Invest, 1973. 52 (11): p. 2745-56. 175

198. Holzenberger, M., et al., Cre-mediated germline mosaicism: a method allowing rapid generation of several alleles of a target gene. Nucleic Acids Res, 2000. 28 (21): p. E92. 199. Lillie, R.D., Advances in histopathologic technic. Am J Clin Pathol, 1948. 18 (11): p. 867-73.

176 ARTICLES

Appendix

Regulation of cardiovascular development and integrity by the heart of glass–cerebral cavernous malformation protein pathway

Benjamin Kleaveland 1, Xiangjian Zheng1, Jian J Liu2, Yannick Blum3, Jennifer J Tung4, Zhiying Zou 1, Mei Chen1, Lili Guo1, Min-min Lu1, Diane Zhou1, Jan Kitajewski4, Markus Affolter 3, Mark H Ginsberg 2 &

Mark L Kahn1 ved. r Cerebral cavernous malformations (CCMs) are human vascular malformations caused by mutations in three genes of unknown function: KRIT1, CCM2 and PDCD10. Here we show that the heart of glass (HEG1) receptor, which in zebrafish has been linked to ccm gene function, is selectively expressed in endothelial cell s. Heg1–/– mice showed defective integrity of the heart, blood vessels and lymphatic vessels. Heg1–/–; Ccm2 lacZ/+ and Ccm2 lacZ/lacZ mice had more severe cardiovascular defects and died early in development owing to a failure of nascent endothelial cells to associate into patent vessels. This endothelial cell phenotype was All rights rese

shared by zebrafish embryos deficient in heg, krit1 or ccm2 and reproduced in CCM2-deficient human endothelial cells in vitro. Defects in the hearts of zebrafish lacking heg or ccm2, in the aortas of early mouse embryos lacking CCM2 and in the lymphatic Inc. vessels of neonatal mice lacking HEG1 were associated with abnormal endothelial cell junctions like those observed in human CCMs. Biochemical and cellular imaging analyses identified a cell-autonomous pathway in which the HEG1 receptor couples to KRIT1 at these cell junctions. This study identifi es HEG1-CCM protein signaling as a crucial regulator of heart and vessel

America, formation and integrity.

11,12 Nature CCMs are a common vascular malformation, with a prevalence of myocardium, of zebrafish embryos , suggesting that these proteins

0.1%–0.5% in the human population1. CCMs arise primarily in the operate in an endothelial cell–autonomous signaling pathway. An

2009 brain as thin-walled, dilated blood vessels that cause seizures, head- endothelial role for this pathway is also supported by studies of

© aches and stroke in midlife, often in association with focal hemor- KRIT1-deficient mice, which show lethal vascular defects at embry- rhage1,2. Familial CCM shows an autosomal dominant pattern of onic day 9 (E9); however, these studies did not detect high-level Krit1 inheritance and is caused by loss-of-function mutations in three genes: or Ccm2 gene expression in the mouse cardiovascular system, leading KRIT1 (also known as CCM1)1,3,4, CCM2 (also known as malcalver- to the proposal of cell-nonautonomous mechanisms of CCM patho- nin and osm)5,6 and PDCD10 (also known as CCM3)7. The genesis such as a requirement for CCM protein signaling in adjacent CCM proteins are putative adaptor proteins that interact biochemi- neuronal cells13,14. cally8–10 and participate in a signaling pathway that is not yet Here we used mice and zebrafish lacking the HEG1 receptor and the fully characterized. The mechanism by which loss of CCM protein CCM2 adaptor to investigate the role of this pathway in the cardio- signaling results in the development of vascular malformations is vascular system. Our studies support an endothelial cell–autonomous not known. mechanism in which the HEG1 receptor couples to CCM proteins to A clue to the role of CCM protein signaling in the cardiovascular regulate endothelial cell-cell junctions required for the formation and system comes from genetic studies in zebrafish, which reveal that loss maintenance of the heart and vessels. Complete loss of function in this of krit1, ccm2 or heg (encoding the type I transmembrane receptor pathway resulted in an inability of emerging endothelial cells to heart of glass) results in a dilated heart phenotype early in develop- associate into a functional cardiovascular system in mouse and ment11,12. This phenotype is characterized by heart failure associated zebrafish embryos. Less-complete loss of function permitted heart with enlarged cardiac chambers in which the endocardium is covered and vessel development but resulted in integrity defects manifested by by a thin layer of myocardial cells. Expression of heg, krit1 and cardiac rupture, vascular hemorrhage and lymphatic leakage. These ccm2 mRNA has been detected in the endocardium, but not the cardiovascular defects arose in conjunction with abnormal endothelial

1Department of Medicine and Cardiovascular Institute, University of Pennsylvania, 421 Curie Blvd., Philadelphia, Pennsylvania 19104, USA. 2Department of Medicine, University of California, San Diego, 9500 Gilman Drive, MC 0726, La Jolla, California 92093-0726, USA. 3Biozentrum, University of Basel, Klingelbergstrasse 50/70, CH-4056 Basel, Switzerland. 4Department of Pathology and Department of Obstetrics and Gynecology, Institute of Cancer Genetics and Herbert Irving Comprehensive Cancer Center, Columbia University Medical Center, New York, New York 10032, USA. Correspondence should be addressed to M.L.K. ([email protected]).

Received 8 October 2008; accepted 17 December 2008; published online 18 January 2009; doi:10.1038/nm.1918

1 ONLIN E PUBLICATIO N 7 ARTICLES 177

+/+ –/– Figure 1 Heart and blood vessel integrity defects in HEG1-deficient a Heg1 c Heg1 d Heg1 mouse embryos and neonates. (a,b) Heg1 was selectively expressed in the

nt developing cardiovascular system. Radioactive in situ hybridization revealed Heg1 expression in the vascular endothelium, cardiac endocardium and ao neural tube at E10.5 (a). By E14.5, Heg1 was expressed in both endothelial and smooth muscle cells in major arteries (b). ao, aorta; h, heart; nt, neural tube. (c–f) E15 Heg1–/– embryos showed deep invaginations of the

ventricular chamber into the compact zone of the ventricular wall and h into the septum, often associated with the presence of blood between e f the epicardial and myocardial cell layers of the heart (arrow in f). e,f show magnification of boxed regions in c,d. (g–i) Pulmonary hemorrhage in b Heg1–/– neonates (h,i) but not in Heg1+/+ littermates (g) was manifested ao by the presence of erythrocytes and fibrin in alveolar air spaces. (j) Cardiac

–/– h rupture in P4 Heg1 neonate. Hemopericardium (left, middle) associated Heg1 with cardiac rupture and formation of a transmural thrombus (right) was observed in Heg1–/– mice. VSD, ventricular septal defect. Scale bars, g h i 200 mm in a–d and j (middle), 20 mm in e–i and j (right).

showed cardiac defects characterized by invagination of the ventricular

cavity into, and often through, the compact layer of ventricular myocardium (Fig. 1c–f ). The septal myocardium was similarly honey-

ved. r combed by endothelial-lined extensions from the ventricular cavity j Hemopericardium (Fig. 1d), a defect accompanied by the presence of ventricular septal defects in most late-gestation embryos. TUNEL and Ki67 staining revealed that HEG1-deficient myocardial defects were not caused by VSD an increase in myocardial apoptosis or a decrease in myocardial proliferation (Supplementary Fig. 5 online and data not shown). All rights rese Although most HEG1-deficient mice survived to birth, approximately Fibrin clot half died before weaning as a result of pulmonary hemorrhage Inc. (Fig. 1g–i). Neonatal HEG1-deficient mice also showed defective cardiac integrity manifested by a blood-filled pericardial sac, a junctions similar to those observed in human CCMs15–17 and asso- phenotype that arose as a result of rupture of the low-pressure atrial 18

America, ciated with KRIT1 depletion in human endothelial cells in vitro , chamber of the heart (Fig. 1j). Cardiac or pulmonary integrity defects

suggesting a common mechanism by which loss of this signaling were observed in 50% of HEG1-deficient mice. These findings indicate pathway confers a spectrum of vertebrate cardiovascular defects in that HEG1 deficiency results in a loss of cardiac and pulmonary Nature

developing and mature animals. Our studies indicate that the signaling vascular integrity, resulting in lethality. pathway implicated in human CCM regulates endothelial cell-cell

2009 association and suggest that agents designed to promote endothelial

RESULTS Heg1 is expressed in the developing cardiovascular system

It has been unclear how CCM proteins function in cardiovascular cell +/+ –/– c Heg1 Heg1 types and whether it is loss of CCM protein function in cardio- d vascular or other cell types that causes human vascular disease. To Lyve1 determine whether HEG1 receptors have a direct role in regulating SMA CCM protein signaling, we cloned the mouse Heg1 gene (Supple- –/– mentary Fig. 1 online) and characterized Heg1 mRNA expression in h Heg1 mice. Using in situ hybridization, we detected Heg1 expression in the

endothelium of the developing heart and aorta and in the neural tube –/– –/– e Heg1 f Heg1 at E10.5 (Fig. 1a), and in the arterial endothelium, smooth muscle, endocardium of the heart and brain vasculature at E14.5 (Fig. 1b and Supplementary Fig. 2 online). In contrast to Heg1, Ccm2 expression was low and detected primarily in the developing neural tube at E10.5 Lyve1 (Supplementary Fig. 3 online). SMA

HEG1 deficiency results in lethal hemorrhage in mice Figure 2 Lymphatic vessel dilatation and leakage in Heg1–/– neonatal mice. To address the role of the HEG1 receptor in mammals, we generated (a,b) HEG1-deficient neonates showed chylous ascites, manifested by the HEG1-deficient mice by targeting Heg1 exon 1 in embryonic stem accumulation of white chyle in the peritoneal space. (c–f) Lymphatic –/– cells (Supplementary Fig. 4 online). In mice with genetic back- malformations in Heg1 mice. Neonatal mesenteric lymphatic vessels of Heg1–/– mice were dilated (arrows) and leaked chyle into the intestinal wall grounds of 50% SV129 and 50% C57Bl/6 or 495% C57Bl/6, and peritoneum. (g,h) Immunostaining for LYVE1 confirmed the lymphatic HEG1 deficiency resulted in embryonic and postnatal lethality (Sup- identity of the dilated mesenteric vessels (arrows). SMA, a-smooth muscle plementary Table 1 online). Midgestation HEG1-deficient embryos actin. Scale bars, 100 mm.

LI N PUBLICA TION 1 7 178 ARTICLES

Cranial Caudal Figure 3 Heg1–/–; Ccm2 lacZ/+ mouse embryos do not establish a patent blood vascular network. Transverse sections of E9 Heg1+/–; Ccm2 +/+

(control) and two Heg1–/–; Ccm2 lacZ/+ embryos at three levels are shown. H&E staining revealed the Control presence of blood-fi lled dorsal aortas (DA), cardinal veins (CV) and branchial arch arteries (BAA) of normal caliber in the Heg1+/–; Ccm2 +/+ control embryo (top) but not in Heg1–/–; Ccm2 lacZ/+ littermates (bottom). Staining for Flk1 in adjacent

sections at the level of the first two branchial #1 arch arteries is shown. Flk1+ endothelial cells lacZ/+ were present at the dorsal aortas, cardinal veins –/–

Ccm2 and branchial arch arteries (arrows) in Heg1 ; ; lacZ/+

–/– Ccm2 embryos, but these cells did not form vessels of normal caliber with patent lumens. AO Heg1

SAC, aortic sac; SV, sinus venosus. Scale bars,

#2 50 mm.

ved. r HEG1 deficiency causes dilated lymphatic vessel malformations mice, which showed no abnormalities before midgestation, both Neonatal mammals transport absorbed fat through the intestinal and Heg1–/–; Ccm2lacZ/+ and Ccm2lacZ/lacZ embryos died before E10 (Sup- mesenteric lymphatic vessels in the form of white chyle. A distinct plementary Table 2 online). Heg1–/–; Ccm2lacZ/+ and Ccm2lacZ/lacZ phenotype observed in approximately 10% of HEG1-deficient neo- embryos were indistinguishable from littermates of all other genotypes natal mice was the appearance of chylous ascites shortly after their first until E9, when the mutants showed identical phenotypes of growth feeding (Fig. 2a,b). Chylous ascites was invariably associated with retardation and marked pericardial edema despite the presence of a All rights rese

severely dilated intestinal and mesenteric lymphatic vessels that leaked visibly normal heartbeat (data not shown). Histologic examination chyle into the intestinal wall and peritoneal space (Fig. 2c–f ). The revealed normal development of the ventricular chamber and bulbus Inc. dilated mesenteric lymphatic vessels were lined with endothelial cells cordis (future right ventricle) but aberrant formation of a dilated that expressed the lymphatic molecular marker LYVE1 and were aortic sac (Fig. 3 and Supplementary Fig. 7 online). Compared to associated with smooth muscle cells typical of collecting lymphatic control embryos, the paired dorsal aortas of E9 Heg1–/–; Ccm2lacZ/+ lacZ/lacZ

America, vessels (Fig. 2g,h). These findings indicate that loss of HEG1 receptors and Ccm2 embryos were small or undetectable and completely

is associated with a loss of integrity of lymphatic vessels as well as of blood vessels and the heart. Because lymphatic vessels are not subject a Nature

to the hemodynamic forces generated by the beating heart, these Control findings further suggest that loss of integrity arises as an intrinsic

2009 defect in the heart and vessels of HEG1-deficient mice.

© heg MO Heg1 and Ccm2 interact genetically in mice Identical big-heart phenotypes arise in zebrafish embryos lacking heg, krit1 (also known as santa) or ccm2 (also known as valentine), and gene-knockdown experiments using morpholinos have shown strong ccm2 MO interactions among these genes in zebrafish12. In contrast, human CCMs have been linked to loss-of-function mutations in KRIT1 and b CCM2, but not HEG1, and HEG1-deficient mice do not experience Control the early embryonic lethality reported for KRIT1-deficient mice13. To determine whether—and to what extent—HEG1 interacts genetically with CCM genes in mammals, we intercrossed Heg1+/–; Ccm2lacZ/+ lacZ heg MO mice. The Ccm2 allele is predicted to be a null allele because the Ccm2lacZ mRNA transcript lacks the 3¢ half of the Ccm2 mRNA (exons 6–10, Supplementary Fig. 6 online), and Ccm2lacZ/lacZ embryos experience early embryonic lethality that phenocopies Krit1 / –/– +/+ ccm2 MO embryos (see below and ref. 13). In contrast to Heg1 ; Ccm2

Figure 4 Endothelial cells of zebrafish lacking heg or ccm2 form vessels c that are normally patterned but not patent. (a) Fli1-GFP–transgenic, heg- and ccm2-morphant (MO) embryos showed dilated hearts (arrows) and Wild type normal vascular patterning. (b) Angiography revealed a proximal circulatory block in heg- and ccm2-morphant zebrafish. Fluorescent microspheres were distributed throughout the vasculature of control, but not heg- or ccm2- morphant, zebrafish after venous injection. (c) FITC-dextran was distributed ccm2 throughout the vasculature in wild-type, but not ccm2-mutant, zebrafish mutant after venous injection. Scale bars, 250 mm.

1 ONLIN E PUBLICATIO N 7 ARTICLES 179

100% Figure 5 HEG1 is required to form normal a P < 0.0001 b 8,000 29% >2,500 nm endothelial junctions in vivo. Dilated mesenteric –/– +/+ 49% lymphatic vessels in Heg1 mice had severely Heg1 1,000–2,500 nm 6,000 –/– <1,000 nm shortened endothelial junctions and gaps between Heg1 P < 0.0001 40% 4,000 endothelial cells. (a) Endothelial cell junctions P < 0.0001 were measured as the length of two overlapping 42% 2,000 cell membranes. Junction lengths in Heg1+/+ and 31% Heg1–/– lymphatic vessels are shown divided into (nm) length Junction 9% 0 0% +/+ –/– terciles (1st, shortest third of junctions in each 1st 2nd 3rd Heg1 Heg1 Terciles group; 2nd, middle third of junctions in each c group; 3rd, longest third of junctions in each group; mean ± s.d. for each tercile). n ¼ 66 Heg1+/+ junctions and 123 Heg1–/– junctions.

The data shown are from analysis of 6 vessel +/+ cross-sections from one animal of each genotype.

Heg1 (b) Percentage of endothelial junctions o1,000 nm, 1,000–2,500 nm and 42,500 nm. 10 ∝m 500 nm 500 nm 500 nm The same junctions were analyzed in panels a and b. (c) Representative low-magnification (far left) and high-magnification images of endothelial

cells. Asterisks indicate sites of endothelial gaps,

–/– normally not present in collecting mesenteric ved. r lymphatics; arrowheads indicate endothelial Heg1 junction limits. Endothelial gaps were present in Heg1–/– but not Heg1+/+ vessels. 10 ∝m 500 nm 500 nm 500 nm

lacked luminal blood cells (Fig. 3 and Supplementary Fig. 7). Analysis dilated heart chambers but no abnormalities in vessel patterning All rights rese

of serial transverse sections revealed that the aortic sac did not connect (Fig. 4a). However, we noted visually that blood cells in the hearts to a lumenized first, second or third branchial arch artery in the of heg- and ccm2-morphant zebrafish did not circulate despite Inc. mutant embryos, suggesting that blood in the heart did not enter the detectable contraction of the dilated heart, a finding similar to the dorsal aortas despite the presence of a normal heartbeat (Fig. 3 and circulatory block we observed in E9 Heg1–/–; Ccm2lacZ/+ and Supplementary Fig. 7). When visible, the cardinal veins of E9 Heg1–/–; Ccm2lacZ/lacZ mouse embryos. lacZ/+ lacZ/lacZ America, Ccm2 and Ccm2 embryos were also devoid of blood, We used angiography to functionally test the patency of the

except at the point where they connected to the sinus venosus of vasculature of heg- and ccm2-morphant zebrafish. Fluorescent beads the heart (Fig. 3). In contrast to the lack of blood in the vessels, injected into the sinus venosus of zebrafish embryos treated with Nature

extravasated blood cells were frequently present in the dilated peri- control morpholinos circulated throughout the body and outlined the cardial cavity (Fig. 3 and data not shown), a finding that may reflect a developing vasculature of the head and tail (Fig. 4b). In contrast,

2009 primary defect in the integrity of the heart; alternatively, this finding injection of heg- or ccm2-morphant zebrafish revealed a complete or

© may arise from the heart beating against a closed circulatory system near-complete circulatory block immediately distal to the heart and/or from reduced embryonic viability at this point in development. (Fig. 4b), the same point at which circulation was blocked in E9 To determine whether the lack of blood-filled vessels in E9 Heg1–/–; Heg1–/–; Ccm2lacZ/+ and Ccm2lacZ/lacZ mouse embryos. We also Ccm2lacZ/+ and Ccm2lacZ/lacZ embryos is due to a lack of endothelial observed a complete circulatory block in all ccm2-mutant zebrafish cells, we used Flk1 immunostaining to identify endothelial cells. We embryos injected with FITC-dextran (Fig. 4c). These findings indicate found endothelial cells at sites where the branchial arch arteries and a conserved role for HEG1 and CCM2 in the formation of a patent dorsal aortas are normally located, but in all Heg1–/–; Ccm2lacZ/+ and vertebrate circulatory system. Ccm2lacZ/lacZ embryos studied these cells were not organized into lumenized vessels as they were in control embryos (Fig. 3). Thus, HEG1-CCM signaling regulates endothelial tube formation circulation in Heg1–/–; Ccm2lacZ/+ and Ccm2lacZ/lacZ embryos was The inability of nascent endothelial cells to form patent branchial arch blocked by an inability of mutant endothelial cells to form lumenized vessels in Heg1–/–; Ccm2lacZ/+ and Ccm2lacZ/lacZ mouse embryos and vessels capable of carrying blood. These findings indicate a strong heg- or ccm2-morphant zebrafish embryos suggested that HEG1-CCM genetic interaction between HEG1 receptors and the CCM protein protein signaling is required for endothelial cells to form lumenized signaling pathway during formation of the primary vasculature tubes during early vascular development. To assess the role of CCM2 in mammals. in endothelial tube formation, we used a fibrin bead assay to examine the ability of human umbilical vein endothelial cells heg and ccm2 are required for vessel patency in zebrafish (HUVECs) expressing CCM2-specific or scrambled control small In contrast to E9 Heg1–/–; Ccm2lacZ/+ and Ccm2lacZ/lacZ mouse hairpin RNAs to form tubes19. CCM2-specific small hairpin RNA embryos, zebrafish embryos lacking heg or ccm2 show primarily resulted in an 85%–90% reduction in CCM2 mRNA expression and cardiac defects11,12. To determine whether HEG1 and CCM proteins had no effect on endothelial cell size, proliferation or migration have differing roles in cardiovascular development in zebrafish and (Supplementary Fig. 7 and data not shown). Although control mice, we examined vascular patterning and function in heg-morphant, HUVECs generated multicellular structures with clearly visible mature ccm2-morphant and ccm2-mutant zebrafish embryos at 48 hours post lumens at 10 d, CCM2-deficient HUVECs formed branched cords fertilization (h.p.f.). Morpholino knockdown of heg or ccm2 in Fli1- similar to those of control cells in length but were frequently unable to GFP transgenic fish (which have GFP-labeled endothelium) resulted in form visible lumens (Supplementary Fig. 7).

LI N PUBLICA TION 1 7 180 ARTICLES

Figure 6 HEG1 receptor intracellular tails a Lysate IP: anti-Flag b associate with CCM2 through KRIT1. Lysate IP: anti-CCM2 (a,b) Coimmunoprecipitation of mouse HEG1 HA-CCM2 + + + + + + Flag-CCM2 + – – + + + – – + + Flag-HEG1 – + – – + – Flag-HEG1 – + – + – – + – + – and mouse CCM2 required the HEG1 receptor Flag-HEG1 C – – + – – + Flag-HEG1 C – – + – + – – + – + intracellular tail. Hemagglutinin (HA)-tagged IB: anti-HA HEG1 CCM2 or Flag-CCM2 was coexpressed with IB: anti-Flag Flag-HEG1 or Flag-HEG1DC (a mutant lacking IB: anti-Flag CCM2 the terminal 106 amino acids of the HEG1 〈 C-terminal tail) in HEK293T cells. Immuno- c d Lysate 2b HEG1 〈 precipitation was done using antibodies to Flag Lysate 2b HEG1 Flag-Krit1 + – + + + + – + + (a) or CCM2 (b). The results shown are Myc-KRIT1 + – + + + – + HA-Ccm2 – + + – + – + + – HA-CCM2 – + + + – + + HA-Ccm2-L197R – – – + – – – – + representative of 43 independent experiments. IB: anti-Myc (c) The HEG1 intracellular tail interacted with IB: anti-Flag KRIT1 and CCM2. HA-tagged mouse CCM2 and Myc-tagged mouse KRIT1 were expressed in IB: anti-HA HEK293T cells, and pull-down assays were done IB: anti-HA with affinity matrices containing the intracellular tail of either the human a integrin or human e f HEG1 2b Input Endothelium 〈 KRIT1 HEG1. The HEG1 tail efficiently pulled down 2b HEG1 (5%) NPxY CCM3 Vessel VE-cadherin -cat formation KRIT1 in the absence of CCM2, but not vice IB: anti-KRIT1 PTB versa. The results shown are representative of CCM2 Junctional stability 43 independent experiments. (d) HEG1 IB: anti-Flag CCM2 ved. PTB -cat Vessel r interacted with ccm2 through krit1. Flag-tagged NPxY CCM3 KRIT1 integrity zebrafish krit1, HA-tagged zebrafish ccm2 and HA-tagged zebrafish ccm2-L197R were HEG1 expressed in HEK293T cells, and pull-down ECM assays were done using human a2b or human HEG1 intracellular tail affinity matrices as in c. ccm2-L197R did not associate with the HEG1 tail (far right lane). The results shown are representative of 43 independent experiments. (e) The HEG1 tail efficiently bound endogenous KRIT1. Pull-down assays using a2b or HEG1 tail affinity matrices were done using lysates from CHO cells transfected with Flag-tagged human CCM2. Endogenous KRIT1 was All rights rese

detected with monoclonal antibody to KRIT1 (top) and heterologous CCM2 with monoclonal antibody to Flag (bottom). The far right lane shows immunoblotting of cell lysate equivalent to 5% of the input for the pull-down assays. The results shown are representative of 43 independent experiments. Inc. (f) HEG1-CCM protein signaling at endothelial cell junctions. HEG1 receptors couple to KRIT1 and CCM2 through HEG1 tail–KRIT1 and KRIT1-CCM2 interactions, respectively. KRIT1 also interacts with the junctional proteins VE-cadherin and b-catenin (b-cat) and CCM2 also interacts with CCM3. HEG1-CCM protein signaling is proposed to regulate junction formation and function. ECM, extracellular matrix; NPxY, NPXY amino acid sequences; PTB, phosphotyrosine binding domain.

America,

Studies of lumen formation by endothelial cells in vitro and in the vessels of Heg1–/– neonatal mice showed markedly shortened endo- Nature

intersegmental vessels (ISVs) of the zebrafish embryo in vivo have thelial junctions compared to vessels in control littermates (mean shown that endothelial cells can assemble lumenized vessels through length of junctions, 1,769 ± 506 nm for Heg1–/– versus 3,355 ± 20–23 +/+ 2009 1,009 nm for Heg1 ; P ¼ 0.01; Fig. 5a,b) and were accompanied the formation and fusion of intracellular vacuoles . We assessed

© the formation of vacuole-like structures in the ISVs of morphant by endothelial gaps not present in control collecting mesenteric and control zebrafish embryos using transgenic zebrafish expressing lymphatic vessels (Fig. 5c). Similarly shortened junctions were iden- GFP-Cdc42 fusion proteins that outline the intracellular vacuoles that tified in the endocardium of ccm2-morphant zebrafish hearts (Sup- form in endothelial cells22. Endothelial vacuole-like structures formed plementary Fig. 8 online). Finally, analysis of the constricted, normally in the ISVs of zebrafish embryos lacking heg or ccm2 bloodless aortas of E9 CCM2-deficient mouse embryos also revealed (Supplementary Fig. 7 and Supplementary Movies 1–3 online). shortened endothelial junctions despite the fact that the caliber of Injection of fluorescent quantum dots into the dorsal aorta of these vessels was markedly reduced (Supplementary Fig. 8). These morphant zebrafish to bypass the block at the level of the branchial findings indicate that defective endothelial junctions characterize both arch arteries confirmed that these vessels formed a patent lumen the dilated and constricted cardiovascular phenotypes of animals (Supplementary Fig. 7 and Supplementary Movies 4–6 online). lacking HEG1-CCM protein signaling as well as human CCMs. These findings suggest that defects in endothelial tube formation Notably, the levels of b-catenin, VE-cadherin and claudin-5, key observed in deficient endothelial cells do not arise because of loss of components of endothelial junctions, were preserved in the lymphatic intracellular endothelial vacuole-like structures. vessels of HEG-1 deficient mice, in the hearts and vessels of zebrafish embryos lacking heg or ccm2 and in CCM2-deficient HUVECs HEG1-CCM signaling regulates endothelial junctions in vivo (Supplementary Fig. 9 online). These findings suggest that HEG1- An ultrastructural characteristic of human CCMs is the presence of CCM protein signaling may regulate the function of these proteins abnormal endothelial cell-cell junctions and gaps between endothelial rather than their expression. cells15–17, and a recent study showed a role for KRIT1 in the dynamic regulation of endothelial cell-cell junctions in vitro18. Thus, one HEG1 receptors couple to CCM proteins through KRIT1 mechanism by which HEG1-CCM protein signaling might regulate Our findings suggest that HEG1 receptors interact with CCM proteins cardiovascular development and integrity is through regulation of in an endothelial cell–autonomous signaling pathway: HEG1 is endothelial cell association. To directly assess the role of HEG1 in expressed in endothelial cells in zebrafish and mice; Heg1–/–; regulating endothelial junctions in vivo, we used transmission electron Ccm2lacZ/+ and Ccm2lacZ/lacZ mouse embryos and heg- and ccm2- microscopy to examine cell-cell junctions in the endothelium of morphant zebrafish embryos show aberrant vessel formation at an neonatal mouse lymphatic vessels. The dilated mesenteric lymphatic early developmental time point when vessels are composed exclusively

1 ONLIN E PUBLICATIO N 8 ARTICLES 181

of endothelial cells; and CCM2-knockdown HUVECs show defects in Several lines of evidence suggest that the vascular defects observed tube formation. In addition, we detected heterologously expressed in zebrafish, mice and humans lacking HEG1 and CCM proteins mouse HEG1 at cell junctions in HUVECs (Supplementary Fig. 10 reflect the loss of an endothelial cell signaling pathway. Although online). To determine whether HEG1 can interact with intracellular Ccm2 gene expression was relatively low and not vascular specific, CCM2, we coexpressed Flag-tagged, wild-type mouse HEG1 or mouse Heg1 expression was detected specifically in vascular tissues and was HEG1 lacking most of its intracellular tail (HEG1DC) with hemag- restricted to endothelial cells at the stages at which deficient mouse glutinin-tagged mouse CCM2 in HEK293T cells and conducted and zebrafish embryos first develop cardiovascular defects. The coimmunoprecipitation experiments with antibodies to Flag and vascular defects observed in E9 Heg1–/–; Ccm2lacZ/+ and Ccm2lacZ/lacZ CCM2. CCM2 coimmunoprecipitated with full-length HEG1 using mouse embryos and 48-h.p.f. morphant zebrafish embryos lacking antibodies to Flag (to pull down HEG1) or CCM2 (Fig. 6a,b). In Heg or Ccm2 arose before the appearance of other vascular cell types, contrast, although HEG1DC was expressed at the cell surface at levels such as pericytes or smooth muscle cells, in which HEG1 might similar to those of full-length HEG1 (Supplementary Fig. 11 online), indirectly regulate endothelial cell function. Consistent with these CCM2 did not coprecipitate with HEG1DC (Fig. 6a,b). observations, mosaic analysis in zebrafish embryos has suggested that To further define how the HEG1 intracellular tail associates with krit1 functions cell autonomously during cardiovascular develop- CCM proteins, we used affinity matrices containing the intracellular ment24. Finally, CCM2-knockdown endothelial cells were defective tail of human HEG1 or of human a2b integrin to pull down in the formation of lumenized tubes, and biochemical studies showed hemagglutinin-tagged mouse CCM2 and Myc-tagged mouse KRIT1 that HEG1 receptors coupled to CCM proteins such as CCM2

heterologously expressed in HEK293T cells. The intracellular tail of primarily through the KRIT1 protein. These findings strongly support HEG1, but not that of a2b integrin, efficiently pulled down KRIT1 and a conserved cell-autonomous mechanism in which HEG1 receptors ved. r both KRIT1 and CCM2 when they were coexpressed (Fig. 6c). How- signal through CCM proteins to regulate endothelial cells during the ever, only small amounts of CCM2 were pulled down when it was formation and function of the heart and vessels. expressed alone (Fig. 6c), indicating that KRIT1 markedly facilitates Common ultrastructural defects in endothelial junctions suggest CCM2 interaction with the HEG1 intracellular tail. that the nonpatent branchial arch arteries that appear early in the To further assess the mechanism of CCM2 association with HEG1, development of mouse and zebrafish embryos lacking HEG1-CCM we compared the ability of human HEG1 to interact with zebrafish protein signaling, the diverse cardiovascular integrity defects in older All rights rese

krit1 coexpressed with wild-type zebrafish ccm2 or with ccm2-L197R, HEG1-deficient mice, and the vascular defects in human CCMs can all a ccm2 protein with an amino acid substitution in the phospho- be explained by varying degrees of loss of endothelial cell-cell Inc. tyrosine binding domain that blocks its binding to krit1 (ref. 8); the association. In our studies, the most severe cardiovascular defects corresponding mutant allele of CCM2 has been identified as a cause of were observed in 48-h.p.f. morphant zebrafish embryos and E9 human CCM6. Expression of wild-type ccm2 efficiently rescued the Heg1–/–; Ccm2lacZ/+ and Ccm2lacZ/lacZ mouse embryos. In these ani-

America, big-heart phenotype of ccm2-deficient zebrafish embryos, but expres- mals, the differentiation and proliferation of early endothelial cells and

sion of ccm2-L197R did not (Supplementary Fig. 12 online), indicat- their migration to sites of vessel formation (vessel patterning) were ing that this amino acid substitution leads to a loss of ccm2 function undisturbed (consistent with our observation that CCM2 deficient Nature

in zebrafish and humans. As observed with the mouse proteins, the HUVECs migrated normally in vitro), but endothelial cell assembly HEG1 receptor tail efficiently pulled down both krit1 and wild-type into a lumenized, patent circulatory system was blocked. Recent

2009 ccm2 when they were coexpressed in HEK293T cells, but only krit1 studies in zebrafish suggest that patent vessels arise either through

© was efficiently pulled down when ccm2-L197R was coexpressed the coalescence of vacuoles in single endothelial cells22 (that is, a (Fig. 6d). Finally, HEG1 receptor tails also efficiently pulled down mechanism within endothelial cells) or through the circumferential endogenous KRIT1 in CHO cells, despite the fact that the levels of arrangement of endothelial cells that are connected by cell-cell junc- endogenous KRIT1 were so low that they could not be detected in 5% tions21,23 (a mechanism between endothelial cells). Our finding that of the input cell lysate by immunoblotting with the same antibody endothelial cell vacuolization and lumenization were preserved in the (Fig. 6e). These findings support a model in which HEG1 receptors ISVs of zebrafish embryos lacking the HEG1-CCM protein pathway is couple to CCM proteins primarily through interaction with KRIT1 at most consistent with a defect in endothelial cell-cell association. endothelial cell-cell junctions (Fig. 6f). The defects observed in the integrity of the heart, blood vessels and lymphatic vessels of HEG1-deficient mice also support the idea that DISCUSSION impaired endothelial cell-cell association results from loss of HEG1- Human genetic studies have revealed that CCM is caused by haplo- CCM protein signaling. The dilated, leaky lymphatic vessels of HEG1- insufficiency of KRIT1, CCM2 or PDCD10, and genetic studies in deficient mice recapitulated many of the key structural and functional mice and zebrafish have implicated these proteins in cardiovascular defects observed in human CCMs and heg-deficient zebrafish hearts, development. How the loss of this signaling pathway leads to these including shortened endothelial cell junctions, abnormal coverage of diverse cardiovascular defects has not been clear. Our studies of the endothelial cells by adjacent cell types such as pericytes or cardio- expression and function of HEG1 and CCM2 support a mecha- myocytes, and vessel leakage and rupture. In addition, we observed nism in which HEG1 receptor signaling through CCM proteins is dilated lymphatic vessels in Heg1+/–; Ccm2lacZ/+ mice but not in required to regulate endothelial cell-cell association during formation Heg1+/– or Ccm2lacZ/+ mice (data not shown), a result consistent of the cardiovascular system and for its integrity thereafter. We with a genetic link between HEG1 and CCM2 in the formation of identified endothelial junction defects identical to those in human CCM-like dilated vessels and in early vessel formation in embryos. CCMs in the dilated hearts and vessels of zebrafish and mice deficient Finally, KRIT1 was recently found to localize at and functionally in HEG1 or CCM2, suggesting that CCM proteins regulate vertebrate regulate endothelial cell junctions18, and we found that heterologously cardiovascular development through control of endothelial cell expressed HEG1 receptors similarly colocalized with b-catenin at association and that defects in this regulation leads to human endothelial junctions. Thus, a unifying mechanism to explain CCM pathogenesis. these diverse vascular phenotypes is one in which HEG1 signaling

LI N PUBLICA TION 1 8 182 ARTICLES

through CCM proteins is required for endothelial cells to associate Immunoprecipitation and immunoblotting. We subcloned full-length mouse with each other to create and maintain the vertebrate cardio- HEG1 without endogenous signal peptide (amino acids 38–1,313) an d vascular system. C-terminally truncated HEG1 (amino acids 38–1,207) into pcDNA3.1 with A final question with important therapeutic implications is whether a human interleukin-1 signal peptide, an N-terminal Flag epitope and a C- terminal v5 epitope (see Supplementary Fig. 1b). We cloned His -recombinant and how these studies advance our understanding of human CCM 6 human HEG1 intracellular tail containing an in vivo biotinylation peptide tag at pathogenesis. Although molecular and genetic pathways are frequently the N terminus into pET15b as previously described for integrin intracellular-tail postulated to link embryonic cardiovascular development and adult model proteins30. We expressed and purified tail proteins from Escherichi a coli31. cardiovascular disease states, such links are often tenuous. It is there- For coimmunoprecipitation studies, we transfected plasmids driving expres- fore noteworthy that our studies of HEG1-CCM signaling in develop- sion of Flag-tagged full-length and truncated mouse HEG1 and hemagglutinin- ing mice and zebrafish support conservation of the role of this tagged mouse CCM2 into HEK293T cells. After 48 h, we incubated the cells for pathway in both development and disease. Our findings suggest that 10 min in lysis buffer (50 mM Tris-HCl, 150 mM NaCl, 25 mM EDTA, 1% the primary role of HEG1-CCM signaling is to control the association NP-40, 0.5% sodium deoxycholate, PhosStop (Roche) and complete protease of endothelial cells to create and maintain cardiovascular structures. inhibitor (Roche), pH 8) drawn through a 25-gauge needle and cleared by –1 Thus, CCMs are predicted to be a disease of defective endothelial centrifugation. We conducted Flag immunoprecipitation with 2 mg ml anti- association, and agents that positively regulate endothelial junction body to Flag (M2, Sigma) and CCM2 immunoprecipitation with polyclonal antibody to CCM2 (2055, see Supplemental Methods) for 42 h and then formation might provide a means of stabilizing human CCMs or incubated with protein G–agarose (Invitrogen) for 1 h. We detected proteins preventing their de novo formation. Alternatively, activation of HEG1- using the following antibodies: mouse monoclonal antibody to Flag (1:500; M2,

CCM signaling might provide a means of treating vascular diseases Sigma), mouse monoclonal antibody to hemagglutinin (1:2,000; Abcam) and characterized by vascular leakage or defective vessel integrity, such as horseradish peroxidase–conjugated goat antibody to mouse IgG (1:5,000; ved. r sepsis. Studies to define better the components and in vivo role of this Jackson ImmunoResearch). signaling pathway may therefore identify new strategies for treating For affinity matrix pull-down assays, we transfected HEK293T and CHO inherited and acquired vascular diseases. cells with plasmids driving expression of Myc-tagged mouse KRIT1, hemagglutinin-tagged mouse CCM2, Flag-tagged zebrafish krit1, hemagglutinin- METHODS tagged zebrafish ccm2, hemagglutinin-tagged zebrafish ccm2-L197R and Mice. We generated SV/129 embryonic stem cells heterozygous for the Heg1 Flag-tagged mouse CCM2. We incubated the cells for 10 min in lysis buffer

All rights rese (25 mM HEPES (pH 7.4), 100 mM NaCl, 0.5% NP-40 and complete protease

allele by deleting exon 1 using recombineering-based gene-targeting techniques, and we microinjected the stem cells into C57Bl/6 blastocysts. We intercrossed inhibitor (Roche)). After clarification, we incubated 350 mg of lysate with

Inc. +/– 10 mg of immobilized human HEG1 tail overnight. We detected proteins using F1-generation Heg1 mice (50% 129, 50% C57Bl/6) and conducted phenotypic analysis on Heg1–/– and wild-type littermates. We generated Heg1+/– mice the following antibodies: mouse monoclonal antibody to Flag (1:2,000; M2, on a 495% C57Bl/6 background by backcrossing for 5 generations. We Sigma), mouse monoclonal antibody to Myc (1:2,000; Abcam), mouse monoclonal antibody to hemagglutinin (1:2,000; Abcam), polyclonal antibody to generated Ccm2lacZ/+ mice from 129P2 embryonic stem cells in which an

America, KRIT1 (1:1,000; 6832, ref. 18), horseradish peroxidase–conjugated goat anti- IRES-bGeo cassette had been inserted into Ccm2 exon 6 using a retroviral gene trap (Bay Genomics clone RRG051; ref. 14). We obtained Tg(Tek-cre)12Flv and body to mouse IgG (1:5,000; Jackson ImmunoResearch) and horseradish Gt(ROSA)26Sor transgenic mice from Jackson Research Laboratories. Unless peroxidase–conjugated donkey antibody to rabbit IgG (1:5,000; Jackson ImmunoResearch). We carried out all steps at 4 1C. Nature

otherwise specified, we maintained all mice on a mixed genetic background. The University of Pennsylvania Institutional Animal Care and Use Committee Electron microscopy and cell-cell junction quantitation. We fixed neonatal

2009 approved all animal protocols. Genotyping primer sequences are available in mouse gut mesentery, 48-h.p.f. zebrafish embryos and E9.5 mouse embryos

© Supplementary Methods online. overnight, embedded them in Polybed 812 (Polysciences), sectioned them, In situ hybridization and immunostaining. Primer sequences used to generate stained them and examined them with a JEOL 1010 electron microscope fitted Heg1 and Ccm2 in situ probes are available in Supplementary Methods. We with a Hamamatsu digital camera. We imaged all identifiable endocardial cell- carried out radioactive in situ hybridization, immunostaining and TUNEL cell junctions from six Heg1+/+ and five Heg1–/– lymphatic vessel cross-sections staining on paraformaldehyde-fixed, paraffin-embedded sections. Detailed (66 and 123 junctions, respectively), and we quantitated junction lengths using protocols are available at http://www.med.upenn.edu/mcrc/histology_core/. ImageJ. We similarly imaged and quantitated endothelial cell-cell junctions We used the following antibodies for immunostaining of tissue sections: rabbit from three control and four ccm2 morpholino–injected zebrafish atria cross- monoclonal antibody to Ki67 (1:250; Vector Laboratories), mouse monoclonal sections (88 and 107 junctions, respectively) and four Ccm2+/+ and six antibody to a-smooth muscle actin (1:100; 1A4, Sigma), rabbit polyclonal Ccm2lacz/lacz mouse dorsal aorta cross-sections (128 and 107 junctions, antibody to LYVE1 (1:2,000; ref. 25), mouse monoclonal antibody to b-catenin respectively). We ordered the junctions by length and determined the mean (1:100; BD Transduction), rat monoclonal antibody to Flk1 (1:50; Phar- junction length for each tercile. We also calculated the percentage of junctions Mingen) and rabbit polyclonal antibody to claudin-5 (1:50; Zymed). o1,000 nm, 1,000–2,500 nm and 42,500 nm. Lastly, we calculated the overall mean junction length from the mean junction lengths for each heart Zebrafish studies. We maintained and bred Tuebingen long-fin wild-type or vessel cross-section. zebrafish and Tg(fli1a:EGFP)y1 (ref. 26), Tg(kdr:EGFP) (ref. 27), Tg(fli1a:EGFP-cdc42)y48 (ref. 22) and ccm2hi296aTg (ref. 28) mutant zebrafish. Statistics. We calculated P values using an unpaired two-tailed Student t test or We injected antisense morpholino oligonucleotides (Gene Tools) that interfere w2 analysis as indicated. with the splicing of ccm2 and heg11,12 into the yolks of one-cell-stage embryos at a dose of 5 ng. To rescue the big-heart phenotype caused by the ccm2 Note: Supplementary information is available on the Nature Medicine website. morpholino, we first injected the one-cell-stage embryos with 5 ng of the ccm2 morpholino and then injected half of those embryos with 100 pg of cRNA ACKNOWLEDGMENTS encoding ccm2 or ccm2-L197R. We carried out microangiography of zebrafish We thank C. Bertozzi, C. Chen, A. Granger, J. Lee, D. Li, P. Mericko, A. Schmaier, 29 E. Sebzda, N. Shanbhag, S. Sweeney and K. Whitehead for valuable insights; embryos as described . We injected red fluorescent (580 and 605 nm) 0.02-mm M. Pack and J. He for assistance with zebrafish studies; R. Meade for preparation –1 carboxylate-modified FluoSphere beads (Invitrogen) or 10 mg ml 70-kDa of the electron microscopy samples; and T. Branson for animal husbandry. This FITC-dextran (Sigma) into the sinus venosus of 48-h.p.f. zebrafish embryos. work was supported by the Swiss National Science Foundation, a Network of We mounted the embryos laterally in 2% methylcellulose and acquired images Excellence grant from the European Community (M.A.) and the US National using an Olympus MVX10 microscope. Institutes of Health grants T32 HL07439 (to B.K.), T32 HL07971 (to X.Z.),

1 ONLIN E PUBLICATIO N 8 ARTICLES 183

HL078784 and AR27214 (to M.G.), HL62454 (to J.K.) and HL075380 and 11. Mably, J.D., Mohideen, M.A., Burns, C.G., Chen, J.N. & Fishman, M.C. Heart of glass HL095326 (to M.L.K.). regulates the concentric growth of the heart in zebrafish. Curr. Biol. 13, 2138–2147 (2003). 12. Mably, J.D. et al. Santa and valentine pattern concentric growth of cardiac myocardium AUTHOR CONTRIBUTIONS in the zebrafish. Development 133, 3139–3146 (2006). B.K. designed and performed all of the studies involving mutant mouse lines, 13. Whitehead, K.J., Plummer, N.W., Adams, J.A., Marchuk, D.A. & Li, D.Y. Ccm1 is contributed to endothelial cell culture and biochemical studies and wrote the required for arterial morphogenesis: implications for the etiology of human cavernous manuscript; X.Z. designed and performed most of the zebrafish and biochemical malformations. Development 131, 1437–1448 (2004). studies; J.J.L., M.C., Z.Z. and L.G. contributed to the biochemical studies; 14. Plummer, N.W. et al. Neuronal expression of the Ccm2 gene in a new mouse model of Y.B. and M.C. contributed to the zebrafish studies; J.J.T. performed the cerebral cavernous malformations. Mamm. Genome 17, 119–128 (2006). endothelial cell culture studies; M.L. performed the immunohistochemistry 15. Wong, J.H., Awad, I.A. & Kim, J.H. Ultrastructural pathological features of cerebro- studies; D.Z. contributed to the generation of mutant mouse lines; J.K. designed vascular malformations: a preliminary report. Neurosurgery 46, 1454–1459 (2000). 16. Clatterbuck, R.E., Eberhart, C.G., Crain, B.J. & Rigamonti, D. Ultrastructural and the endothelial cell culture studies and wrote the manuscript; M.A. designed immunocytochemical evidence that an incompetent blood-brain barrier is related to some of the zebrafish studies and wrote the manuscript; M.H.G. designed the pathophysiology of cavernous malformations. J. Neurol. Neurosurg. Psychiatry 71, some of the biochemical studies and wrote the manuscript; M.L.K. contributed 188–192 (2001). to the design of mouse, zebrafish, endothelial and biochemical studies and wrote 17. Tu, J., Stoodley, M.A., Morgan, M.K. & Storer, K.P. Ultrastructural characteristics of the manuscript. hemorrhagic, nonhemorrhagic, and recurrent cavernous malformations. J. Neurosurg. 103, 903–909 (2005). Published online at http://www.nature.com/naturemedicine/ 18. Glading, A., Han, J., Stockton, R.A. & Ginsberg, M.H. KRIT-1/CCM1 is a Rap1 effector Reprints and permissions information is available online at http://npg.nature.com/ that regulates endothelial cell cell junctions. J. Cell Biol. 179, 247–254 (2007). 19. Nakatsu, M.N. et al. Angiogenic sprouting and capillary lumen formation modeled by reprintsandpermissions/ human umbilical vein endothelial cells (HUVEC) in fibrin gels: the role of fibroblasts and Angiopoietin-1. Microvasc. Res. 66, 102–112 (2003). 1. Rigamonti, D. et al. Cerebral cavernous malformations. Incidence and familial occur- 20. Folkman, J. & Haudenschild, C. Angiogenesis in vitro. Nature 288, 551–556 (1980). rence. N. Engl. J. Med. 319, 343–347 (1988). 21. Jin, S.W., Beis, D., Mitchell, T., Chen, J.N. & Stainier, D.Y. Cellular and molecular ved. 2. Revencu, N. & Vikkula, M. Cerebral cavernous malformation: new molecular and analyses of vascular tube and lumen formation in zebrafish. Development 132, r clinical insights. J. Med. Genet. 43, 716–721 (2006). 5199–5209 (2005). 3. Sahoo, T. et al. Mutations in the gene encoding KRIT1, a Krev-1/rap1a binding protein, 22. Kamei, M. et al. Endothelial tubes assemble from intracellular vacuoles in vivo. Nature cause cerebral cavernous malformations (CCM1). Hum. Mol. Genet. 8, 2325–2333 442, 453–456 (2006). (1999). 23. Blum, Y. et al. Complex cell rearrangements during intersegmental vessel sprouting 4. Eerola, I. et al. KRIT1 is mutated in hyperkeratotic cutaneous capillary-venous and vessel fusion in the zebrafish embryo. Dev. Biol. 316, 312–322 (2008). malformation associated with cerebral capillary malformation. Hum. Mol. Genet. 9, 24. Hogan, B.M., Bussmann, J., Wolburg, H. & Schulte-Merker, S. Ccm1 cell autono- 1351–1355 (2000). mously regulates endothelial cellular morphogenesis and vascular tubulogenesis in

All rights rese 5. Liquori, C.L. et al. Mutations in a gene encoding a novel protein containing a zebrafish. Hum. Mol. Genet. 17, 2424–2432 (2008).

phosphotyrosine-binding domain cause type 2 cerebral cavernous malformations. 25. Sebzda, E. et al. Syk and Slp-76 mutant mice reveal a cell-autonomous hematopoietic Am. J. Hum. Genet. 73, 1459–1464 (2003). cell contribution to vas cular development. Dev. Cell 11, 349–361 (2006). Inc. 6. Denier, C. et al. Mutations within the MGC4607 gene cause cerebral cavernous 26. Lawson, N.D. & Weinstein, B.M. In vivo imaging of embryonic vascular development malformations. Am. J. Hum. Genet. 74, 326–337 (2004). using transgenic zebrafish. Dev. Biol. 248, 307–318 (2002). 7. Bergametti, F. et al. Mutations within the programmed cell death 10 gene cause 27. Jin, S.W. et al. A transgene-assisted genetic screen identifies essential regulators of cerebral cavernous malformations. Am. J. Hum. Genet. 76, 42–51 (2005). vascular development in vertebrate embryos. Dev. Biol. 307, 29–42 (2007). 8. Zawistowski, J.S. et al. CCM1 and CCM2 protein inter actions in cell signaling: 28. Amsterdam, A. & Hopkins, N. Retroviral-mediated insertional mutagenesis in zebra- America, implications for cerebral cavernous malformations pathogenesis. Hum. Mol. Genet. fish. Methods Cell Biol. 77, 3–20 (2004). 14, 2521–2531 (2005). 29. Weinstein, B.M., Stemple, D.L., Driever, W. & Fishman, M.C. Gridlock, a localized 9. Voss, K. et al. CCM3 interacts with CCM2 indicating common pathogenesis for cerebral heritable vascular patterning defect in the zebrafish. Nat. Med. 1, 1143–1147 (1995). cavernous malformations. Neurogenetics 8, 249–256 (2007). 30. Lim, C.J. et al. Alpha4 integrins are type I cAMP-dependent protein kinase-anchoring Nature 10. Zhang, J., Rigamonti, D., Dietz, H.C. & Clatterbuck, R.E. Interaction between krit1 and proteins. Nat. Cell Biol. 9, 415–421 (2007). malcavernin: implications for the pathogenesis of cerebral cavernous malformations. 31. Pfaff, M., Liu, S., Erle, D.J. & Ginsberg, M.H. Integrin beta cytoplasmic domains

2009 Neurosurgery 60, 353–359 (2007). differentially bind to cytoskeletal proteins. J. Biol. Chem. 273, 6104–6109 (1998).

LI N PUBLICA TION 1 8