ACTAUNIVERSITATIS

UPSALIENSISUPPSALA

2020

Digital Comprehensive Summaries of Uppsala Dissertationsfrom the Faculty of Medicine 1653

Cdc42, orchestrator of vascularmorphogenesis in the retina

ALBERTO ÁLVAREZ-AZNAR

ISSN 1651-6206ISBN 978-91-513-0910-1urn:nbn:se:uu:diva-407117

Dissertation presented at Uppsala University to be publicly examined in Rudbecksalen, DagHammarskjölds Väg 20, Uppsala, Friday, 15 May 2020 at 13:00 for the degree of Doctor ofPhilosophy (Faculty of Medicine). The examination will be conducted in English. Facultyexaminer: Professor Ondine Cleaver (University of Texas Southwestern Medical Center atDallas, US).

AbstractÁlvarez-Aznar, A. 2020. Cdc42, orchestrator of vascular morphogenesis in the retina. DigitalComprehensive Summaries of Uppsala Dissertations from the Faculty of Medicine 1653.48 pp. Uppsala: Acta Universitatis Upsaliensis. ISBN 978-91-513-0910-1.

Cdc42 is a small GTPase that controls many cellular functions related to cytoskeletal dynamics,such as migration, polarity, and proliferation. Despite what we know of Cdc42 in other cell types,not much research has been done on the vasculature. This thesis describes the consequences ofCdc42 deletion in two vascular cell types—endothelial and mural cells—during developmentalangiogenesis.

In paper I, we demonstrate through a combination of in vitro, in silico, and in vivo assays, thatCdc42-deficient endothelial cells migrate less and fail to distribute normally in areas of naturallyoccurring high proliferation during angiogenesis, causing vascular malformations with enlargedlumens. In addition, these cells present impaired filopodia formation, a disadvantage for the tipcell position, disturbed axial polarity and altered junctions.

With an in vivo approach, in paper III we demonstrate that the deletion of Cdc42 in muralcells has consequences on the morphogenesis of the retinal vasculature. Cdc42-deficient muralcells proliferate less and cannot keep up with the nascent angiogenic vasculature, which resultsin a complete pericyte loss at the sprouting front. Furthermore, we describe that mural cellscontribute to the remodeling of the vasculature, also after the initial phases of angiogenesis.

The CreERT2 system is frequently used for conditional gene deletion and lineage tracing.Tamoxifen administration allows spatiotemporally controlled recombination of fluorescentreporters, and tracing of the labeled cells. However, in the course of our studies, we observedtamoxifen-independent recombination. In paper II, we describe this phenomenon in detail, usingdifferent combinations of CreERT2 and fluorescent reporter lines. We conclude that tamoxifen-independent recombination is a widespread occurrence, and that fluorescent reporter linespresent varying levels of susceptibility to it.

In summary, the work presented here sheds new light on the role of Cdc42 in the vasculature.Additionally, this thesis describes in detail an important feature of CreERT2 and reporter linesthat should be taken into account when performing lineage-tracing experiments.

Keywords: Cdc42, endothelial cell, pericyte, mural cell, angiogenesis, CreERT2

Alberto Álvarez-Aznar, Department of Immunology, Genetics and Pathology, VascularBiology, Rudbecklaboratoriet, Uppsala University, SE-751 85 Uppsala, Sweden.

© Alberto Álvarez-Aznar 2020

ISSN 1651-6206ISBN 978-91-513-0910-1urn:nbn:se:uu:diva-407117 (http://urn.kb.se/resolve?urn=urn:nbn:se:uu:diva-407117)

“The most important step a man can take. It’s not the first one, is it? It’s the next one. Always the next step, Dalinar.”

- Brandon Sanderson, Oathbringer

A mis padres, quienes, a pesar de tantoslibros de fantasía, siguen siendo los mejores

héroes que he tenido.

List of Papers

This thesis is based on the following papers, which are referred to in the text by their Roman numerals.

I Laviña, B., Castro, M., Niaudet, C., Cruys, B., Álvarez-Aznar,

A., Carmeliet, P., Bentley, K., Brakebusch, C., Betsholtz, C., Gaengel, K. (2018) Defective endothelial cell migration in the absence of Cdc42 leads to capillary-venous malformations. De-velopment, 145(13):dev161182

II Álvarez-Aznar, A., Martínez-Corral, I., Daubel, N., Betsholtz, C., Mäkinen, T., Gaengel, K. (2019) Tamoxifen-independent recombination of reporter genes limits lineage tracing and mo-saic analysis using CreERT2 lines. Transgenic Research, doi:10.1007/s11248-019-00177-8

III Álvarez-Aznar, A., Vázquez-Liébanas, E., Orlich, M., Adams, R., Brakebusch, C., Betsholtz, C., Gaengel, K. (2020) Cdc42 is required in mural cells for proper patterning of the retinal vas-culature. Manuscript

Reprints were made with permission from the respective publishers.

Other papers by the author

Álvarez-Aznar, A., Muhl, L., Gaengel, K. (2017) VEGF Receptor Tyrosine Kinases: Key Regulators of Vascular Function. Current Topics in Develop-mental Biology, 123, 433–482 Castro, M., Laviña, B., Ando, K., Álvarez-Aznar, A., Abu Taha, A., Brake-busch, C., Dejana, E., Betsholtz, C., Gaengel, K. (2019) Cdc42 deletion elic-its cerebral vascular malformations via increased MEKK3-dependent KLF4 expression. Circulation Research, 124:1240-1252

Contents

Introduction to the vascular system .............................................................. 11 Mural cells: sidekicks of the vascular system .......................................... 12 Angiogenesis ............................................................................................ 14 

The sprout ............................................................................................ 15 Pro- and antiangiogenic signals guide the vasculature ........................ 17 Maturation and remodeling of the primitive vasculature ..................... 19 The mouse retina: angiogenesis model ................................................ 22 

Endothelium and disease .......................................................................... 24 

Cdc42, small Rho GTPase ............................................................................ 25 Cellular functions of Cdc42 ..................................................................... 25 Cdc42 in the vasculature .......................................................................... 27 Cdc42 in disease ....................................................................................... 28 

Fluorescent reporters in the Cre-loxP system ............................................... 30 

Present investigations .................................................................................... 33 Paper I ...................................................................................................... 33 Paper II ..................................................................................................... 34 Paper III .................................................................................................... 34 

Future perspectives ....................................................................................... 36 

Acknowledgements ....................................................................................... 38 

References ..................................................................................................... 40 

Abbreviations

Alk1 Activin receptor-like kinase 1 AMD Ang1/2

Age-related macular degeneration Angiopoietin 1/2

aPKC Atypical protein kinase C Arp2/3 AVM

Actin-related protein 2/3 Arteriovenous malformation

BBB Blood brain barrier BMP Bone morphogenetic protein bp BRB

Base pairs Blood retina barrier

CADASIL Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy

CCM Cerebral cavernous malformation Cdc42 Cell division cycle 42 Cre Cyclization recombination protein

CreERT2 Tamoxifen-inducible estrogen receptor Cre CNS Central nervous system Dll4 DR EndMT

Delta-like ligand 4 Diabetic retinopathy Endothelial-mesenchymal transition

Eph Ephrin receptor Foxo1 Forkhead box O1 GAP GTPase activating protein GDI Guanine-nucleotide dissociation inhibitor GEF Guanine-nucleotide exchange factor GTP Guanosine triphosphate HHT Hemorraghic hereditary telangiectasia Jag1 Jagged 1 JNK c-Jun N-terminal kinase LoxP Locus of X-over-P1 MAPK Mitogen-activated protein kinase

MMPs Matrix metalloproteinases MTOC Microtubule organizing center Nrp1 Neuropilin 1 Par3/6 Partitioning defective homolog PDGF-B Platelet-derived growth factor-B PDGFR-B Platelet-derived growth factor receptor-B Robo4 Roundabout guidance receptor 4 Sema3E Semaphorin 3E Smad Small, mothers against decapentaplegic SRF Serum response factor Tie1/2 Tyrosine kinase with immunoglobulin-like and EGF-like

domains 1 TNF-α Tumor necrosis factor alpha UNC5B Unc-5 Netrin Receptor B WASP Wiskott-Aldrich syndrome protein VEGF-A Vascular endothelial growth factor-A VEGFR1/2 Vascular endothelial growth factor receptor 1 VSMC Vascular smooth muscle cell

11

Introduction to the vascular system

Most animal cells need a constant supply of oxygen to live. This molecule can passively diffuse through tissues up to 100 to 200μm, meaning that, in some of the simplest animals such as sponges or jellyfish, oxygen and nutri-ents can reach the cells from the outer environment without any trouble. In contrast, organisms that are more complex require some kind of circulatory system to distribute oxygen in an efficient manner. Insects for example, have an open circulatory system, with a substance called hemolymph that flows throughout their bodies providing oxygen and nutrients. However, verte-brates are even more complex than that, and thus possess a closed circulatory system, in which the heart pumps blood into a network of tubules called blood vessels that extend all over the body, so that no cell is ever too far away from the oxygen supply. To put the intricacy of this network into per-spective, all the blood vessels in the human body are said to span a distance that would be enough to wrap around the Earth two and a half times.

When a tissue or organ develops, it does so by sending pioneer cells into the unknown, away from the safety of the cardiovascular system and its nu-trients. To increase their chances of survival, these pioneer cells must estab-lish a supply chain and thus, they light the beacons and recruit new blood vessels to their adventure. Occasionally, the development of the cardiovascu-lar system is abnormal, causing illnesses like cerebral cavernous malfor-mations (CCM) or hereditary hemorrhagic telangiectasia (HHT), in which vessel malformations lead to bleedings that cause suffering or even death. Frequently, blood vessels decline over the course of other diseases, worsen-ing the condition of those patients; such is the case of diabetic retinopathy (DR), in which diabetes may lead to blindness. Additionally, some tumors are able to hijack the vascular system, securing a nutrient supply for them-selves in order to keep growing, and we still struggle to inhibit this activity. On the other hand, no therapeutic effort to reperfuse ischemic tissues has been successful to date. For all these reasons, it is imperative to continue unravelling the vascular system, and understanding the mechanisms behind its development.

12

Mural cells: sidekicks of the vascular system The cardiovascular system is a closed circuit that consists of heart and blood vessels. The vessel network is highly organized and hierarchical, and it is formed by arteries, veins, and capillaries. Arteries carry oxygenated blood from the heart under high pressure and pulsatile flow, whereas veins transport it back to the heart under low-pressure conditions. In between them lie the capillaries, which account for 80% of the vasculature length and where most of the gas exchange actually occurs. Blood vessels are formed by two types of cells: endothelial and mural cells. Endothelial cells are flat and they are arranged on a single-cell layer to form the inner lining of the vessel. Mural cell is a collective term for two types of perivascular cells with shared ontogeny that are embedded in the endothelial basement membrane: vascular smooth muscle cells (VSMC), and pericytes. VSMC wrap around arteries and veins like a solenoid and possess contractile machinery, illus-trated by the expression of α-smooth muscle actin (α-SMA). Pericytes on the other hand, lie on top of the capillaries and present peg and socket arrange-ments, structures through which their processes penetrate endothelial invagi-nations, providing close signaling contact between the two cell types 1.

Unfortunately, characterization of pericytes has posed a challenge ever since their discovery due to the absence of unambiguous markers, a circum-stance that has paved the way of research with contradiction and controversy. Even today, appropriate pericyte and VSMC characterization depends on a combination of criteria: multiple markers, localization, and morphology; and cross-tissue comparison remains difficult, due to their organotypicity and mul-tiple developmental origins 2. Recently, single cell RNA-sequencing provided new proof of why they are difficult to distinguish without proper techniques. In the central nervous system (CNS), the expression profile of venous VSMC, although similar to that of arterial VSMC, actually forms a continuum with the RNA profile of pericytes 3, which indicates that at the expression level, peri-cytes and venous VSMC of the CNS are almost identical, and it blurs the clas-sical boundaries between these two cell types (Figure 1).

Endothelial cells recruit pericytes to the nascent sprout during angiogene-sis and establish adherens and tight junctions; perturbing this mechanism elicits vessel enlargement and microaneurysms 4, which was the first indica-tion that pericytes promote vascular maturation and restrict angiogenesis. This has been confirmed in several models such as during mouse brain and retinal development and is further described in the “Angiogenesis” section. Because it is vital to keep toxins out of the CNS, the brain and retina present continuous endothelium with barrier properties, called blood-brain barrier (BBB) and blood-retina barrier (BRB), respectively. Pericytes are especially important for these barriers, and their depletion during angiogenesis leads to endothelial hyperplasia and permeable vessels, which at least in the brain is due to upregulated transcytosis 5,6. Whether pericytes are required to keep

13

quiescence in the mature vasculature is more disputed, although work in the retina suggests that, at the very least, pericyte ablation sensitizes the adult vasculature to VEGF-A leakage 7. Pericytes also regulate proteins of the extracellular matrix (ECM), and can influence vascular architecture through matrix-dependent signaling gradients 8. Additionally, one of the differential elements in the RNA profile of brain pericytes with regard to other cells is the enrichment of numerous transmembrane transporters of the solute carrier (SLC), ATP, and ATP-binding cassette (ABC) families. These transporters are important for homeostasis, reinforcing the importance of pericytes in the maintenance of the neurovascular milieu across the BBB 3.

Actin can make up for up to 40% of the protein content of aortic VSMC 9, which in combination with their solenoid morphology, unsurprisingly points to contractibility as the main function of VSMC. Studies on the brain, which has been used as a model to assess blood flow control, show that contractile VSMC sitting on arteries and arterioles play a role in neurovascular coupling by regulating flow in a regional manner depending on neuronal demands 10,11. Opposing work claimed that contractile pericytes regulate blood flow instead 12, but that study was an example of misguided nomenclature, be-cause according to their own description of pericytes, they actually studied arteriolar VSMC.

The functions of pericytes described above have been explored in detail in vivo through knockout models based on validated markers. However, there is no shortage of literature ascribing other attractive functions to pericytes based on in vitro work that may be misinterpreted, or dubious markers. Some of this research has attributed pericytes roles in wound healing as mesenchymal stem cells or scar forming cells 13–15. In summary, what we know about mural cells is still quite limited due to a combination of reasons such as challenging characterization, and a traditionally secondary role on the vasculature; in addition, it remains indispensable to filter studies that use ambiguous or straight erroneous nomenclature.

14

Figure 1. Types of mural cells. Arterial and arteriolar VSMC wrap in a solenoid fashion around the vessel, although the coverage of arterial VSMC is denser. They are contractile and regulate blood flow. Venous VSMC also possess contractile machinery, but their stellate shape suggests that they do not control blood flow in the same way as arterial VSMC. Pericytes are in close contact with the vessels through gap junctions and peg/sockets. Pericytes participate in maturation and qui-escence of the vasculature, maintaining the BBB and BRB in the CNS. According to recent single cell sequencing data, venule VSMC and pericytes are indistinguishable from each other.

Angiogenesis During embryonic development, the first blood vessels are formed through vasculogenesis. This process consists in the assembly of angioblasts, progen-itor cells of mesodermal origin, into a cord structure. Subsequently, neigh-boring endothelial cells become polarized and flattened and form a lumen in between them, contributing to the first vessels in the embryo 16,17. Intussus-ception is another, less understood process, during which endothelial cells within an established vessel assemble inside the lumen to split the vessel in two and then mural cells contribute to its maturation 18,19. Angiogenesis, the development of new vessels from preexisting ones, is responsible for the vast majority of vascular development, both in physiological and pathologi-cal conditions.

15

Angiogenesis produces an immature network that eventually undergoes remodeling to form differentiated capillaries, veins, and arteries, which, in addition to the traditional classification based on location and morphology, in the era of single cell transcriptomics have been shown to have a differen-tiated gene expression pattern 3.

Pro-angiogenic signals such as vascular endothelial growth factor (VEGF-A), promote and guide endothelial cell sprouting from quiescent vessels, triggering angiogenesis 20. During this process, matrix metallopro-teinases (MMPs) support the nascent sprout by degrading the basement membrane that surrounds the quiescent vessel, and the extracellular matrix that lies beyond, thereby allowing the endothelium to invade the tissue. MMPs are expressed by sprouting endothelial cells and other cell types such as mural cells 21. Tip cells are highly specialized endothelial cells that are situated at the leading edge of the angiogenic sprout, and they are motile, invasive and extend filopodia towards the VEGF-A gradient. The tip cell position is transient, and endothelial cells constantly compete and overtake each other, driven by a fine balance of VEGF-A, Notch signaling, and me-tabolism 22,23. As endothelial cells proliferate and migrate, they recruit mural cells to the nascent vasculature though platelet-derived growth factor (PDGF-B) signaling; and the sprouts progress and interconnect with each other, forming the vascular network 24. Mural cells contribute to the matura-tion process by becoming invested onto the blood vessels, restricting angio-genic signaling as the vasculature matures to optimize its functionality. Here, some of the molecular and mechanical cues that influence vascular morpho-genesis will be reviewed, from the initial angiogenic sprout to the posterior remodeling.

The sprout The potent angiogenic factor, VEGF-A, binds two different receptors of the VEGF receptor family: VEGFR1 and VEGFR2. VEGFR2, which is ex-pressed in endothelial cells, is the main effector of VEGF-A signaling, medi-ating a myriad of vascular functions such as endothelial cell migration, pro-liferation or survival, as well as vascular permeability, and it is considered the main cue triggering angiogenesis 25,26. The main function of VEGFR1 is to work as a decoy receptor that traps VEGF-A, inhibiting VEGFR2 signal-ing and it is expressed mainly in endothelial cells. However, at least in the mouse retina, pericytes at the sprouting front also represent an important source of VEGFR1 and modulate VEGF-A signaling 6. Seminal work showed that VEGF-A promotes formation of filopodia in endothelial tip cells and their guidance in the mouse retinal vasculature 27. This work also characterized the angiogenic sprouts, defining them as consisting of a migra-tory tip cell that expresses high levels of VEGFR2 and extends filopodia towards the VEGF-A gradient, and a group of stalk cells that follow the tip

16

and proliferate, forming a lumenized and perfused vessel. In following years, Notch1 was pointed out as a negative regulator of tip cell formation in vivo 28, when mouse retina work showed that deletion of Notch1 or the Notch ligand, delta-like ligand 4 (Dll4) promotes filopodia formation, as opposed to the effect of pharmacological treatment with another Notch ligand, Jag-ged1 (Jag1). Additionally, Notch1-deficient endothelial cells display a high-er commitment to the tip cell position, reinforcing the notion that Notch1 is key in tip-stalk specification. Subsequently, a combination of in vivo, in vitro, and in silico models, demonstrated that the tip position is a dynamic one, rather than a committed cell fate. In these studies, Notch1 was revealed to orchestrate the tip-stalk cell balance by controlling VEGFR1 and 2 ex-pression 29,30, with Dll4 working as an agonist of Notch, and Jag1 behaving as a de facto antagonist of this interaction due to Dll4 displacement and a lower activating capacity 31. Essentially, tip cells with high levels of VEGF-A—VEGFR2 signaling express Dll4, activating Notch1 in the neighboring cells (Figure 2). Endothelial cells with activated Notch1 signaling increase the expression of VEGFR1 while decreasing VEGFR2, this alteration of the VEGFR2-VEGFR1 balance results in a downregulation of VEGF-A signal-ing and the tip potential in these cells, which remain as proliferating stalk cells. Remarkably, an oscillating level of Notch and a constant competition between heterogeneous tip and stalk cells is required for an optimal morpho-genesis of the vasculature, whereas Notch signaling synchronization impairs angiogenesis 32,33.

Strikingly, recent live imaging work in the zebrafish embryo claimed that stalk cells can activate Notch signaling in tip cells and that Notch is actually dispensable for tip cell competition, in clear opposition to the previous cur-rent of thought 34. This was further supported by a parallel study on the mouse retina demonstrating that, although Notch1-deficient tip cells indeed have a predisposition to remain at the leading position, surprisingly this is also the case for Dll4-deficient tip cells, which, in theory, should lose the competition instead 35.

In recent years, endothelial cells were found to be highly dependent on glycolysis, which brought metabolism to the spotlight. Blocking this path-way impairs tip cell selection and filopodia formation, pointing towards me-tabolism playing a role not only in angiogenesis but more specifically in tip-stalk cell balance, which may be due to metabolic exhaustion of the cells at the leading position 36. Additionally, in a fascinating zebrafish study, dele-tion of the adherens junction protein, vascular endothelial cadherin (VE-Cadherin), results in the breakdown of the connection between tip cells and stalk cells, which allows tip cells to migrate alone. In these conditions, the trailing stalk cells tend to acquire tip properties and become the new leader of the sprout. Interestingly, laser-mediated tip cell ablation does not elicit the same stalk to tip cell conversion, which suggests that VE-Cadherin not only

17

keeps tip cells attached to the rest of the sprout, but also plays an important function in tip-stalk transition 37.

Figure 2. The angiogenic sprout: endothelial cells express VEGFR2 and migrate towards the VEGF-A gradient. Dll4 signaling from filopodia-extending tip cells activates Notch in stalk cells, which downregulates VEGFR2 expression and tip cell competence. At the same time, endothelial cells express PDGF-B to recruit PDGFR-B expressing pericytes to the nascent endothelium. In turn, pericytes express Ang1, which contributes to vascular maturation and barrier function through the vascular Tie2 receptor. The stretch of vasculature not covered by pericytes at the sprouting front is leaky and presents immature junctions.

Pro- and antiangiogenic signals guide the vasculature It is difficult to imagine building a complex network system simply based on endothelial cells growing towards a VEGF-A gradient and overtaking each other based on Notch signaling. In fact, innumerable guidance molecules and pathways are required to achieve such level of intricacy, although only some of them will be mentioned here for the sake of brevity. Sphingosine 1-phosphate (S1P), is a lipid that is produced by endothelial and red blood cells mainly, and is transported in the blood. S1P binds the endothelial-specific receptor, S1PR1, and inhibits VEGFR2 internalization, a step re-quired for its downstream signaling, which results in an antiangiogenic effect 38. Strikingly, this S1PR1 can also mediate flow-dependent cell adhesion and orientation independently of S1P activation 39. Another receptor of the same

18

family, S1PR3, is expressed in mural cells, although its role in these cells remains unexplored.

In nature, efficient processes are conserved in between systems, so it is perhaps not surprising that vascular and axonal sprouting, which often de-velop in parallel, share some guidance pathways. One such case is neuropilin (Nrp1), which promotes angiogenesis in the endothelium although, interest-ingly, it restricts sprouting in neuronal tissue. Nrp1 is a key vascular co-receptor, and it elicits downstream signaling of VEGFR2 and plexin recep-tors. It regulates tip cell sprouting and motility 40, and it is required for prop-er arterial specification 41, a role that is probably related to the interplay be-tween tip cells and arterial fate in the developing vasculature, explored later in this text.

Several pairs of semaphorin ligand-plexin receptor combinations exist, and they influence vascular and neural patterning as well. Semaphorins are classified depending on if they are soluble or membrane bound, and some of them use Nrp1 as a co-receptor. Among them, Sema3E and Plexin-D1 are a well-known endothelial-specific ligand-receptor pair that mediates repulsion between endothelial cells and is essential for vascular development in the embryo, with their loss leading to disorganization and lack of normal vascu-lar hierarchy 42,43.

Another set of molecules that are present in both neural and vascular sys-tems are ephrins and Eph receptors. This pathway, which is particularly in-teresting because both the ligand and receptor are membrane-bound, has mostly been suggested to have proangiogenic potential, such as ephrinA1-EphA2, used by tumors to promote VEGF expression in their surroundings, although this proangiogenic function remains controversial in some settings 44,45. Perhaps the most studied couple of ephrin molecules is ephrinB2-EphB4, which are expressed in arteries and veins, respectively, and contrib-ute to arteriovenous specification in zebrafish by distinctively promoting endothelial progenitor migration 46. Additionally, mural cells require ephrinB2 signaling for motility and adhesion, and thus for their attachment onto the vasculature 47,48, which in turn contributes to the maturation of the endothelium. Interestingly though, this pair of molecules is also capable of reverse signaling from EphB4 to ephrinB2, acting as a modulator that sup-presses sprouting and promotes circumferential vessel growth instead 49, a key step in regulating the vessel network.

Slit proteins bind Robo4 and have an unclear role: Slit2 has been linked to vascular stabilization and Slit3 to vessel formation in the embryo 50,51. An-other set of signaling molecules, netrins, are expressed by mural cells, bind-ing the UNC5B receptor to restrict angiogenesis, in line with the role of mural cells working to promote vascular maturation 52,53. Several bone mor-phogenetic proteins (BMPs) have also been linked to vascular development. This family of ligands usually bind Alk receptors, which then lead to down-stream phosphorylation of Smad proteins with different outcomes. For ex-

19

ample, BMP2 behaves as a proangiogenic signal that is associated with arte-riovenous fate 54. On the other hand, circulating BMP9 and its receptor Alk1, synergize with Notch signaling to restrict angiogenesis 55,56.

In summary, a myriad of signaling molecules keep the delicate balance that contributes to proper angiogenesis, as illustrated above. The formation of the vascular network not only relays on tip cell specification and migra-tion, but also on the proliferating cells behind them that contribute to vessel extension and the diameter enlargement needed to allow blood flow. Fur-thermore, finely tuned branching is required to spread the vasculature evenly across the newly formed tissue, and the primitive vascular beds must under-go specification and remodeling to acquire a hierarchical pattern consisting of arteries, veins, and capillaries. This complexity partly explains why no proangiogenic therapy has been successful to date, and why antiangiogenic therapies are moderately effective at best 57,58.

Maturation and remodeling of the primitive vasculature During angiogenesis, the tip/stalk cell competition gets most of the attention because it is at the battlefront so to speak; however, a whole deal of action is going on behind the trenches. While the vasculature sprouts, the newly as-sembled and primitive vascular plexus undergoes remodeling and matures, a process controlled by several factors. Endothelial cells at the sprouting front secrete PDGF-B, a ligand that attracts PDGFR-B+ mural cells to cover the unstable endothelium. Seminal work by this lab showed that, without PDGF-B mediated recruitment, blood vessels are depleted of mural cells, which leads to microaneurysms 4. In addition, secreted PDGF-B must remain at-tached to the endothelial cell surface to achieve adequate recruitment 59. Without mural cell coverage, the embryonic vasculature undergoes diameter enlargement due to hyperplasia, as well as increased transcellular and para-cellular transport leading to edema 60. Recently, inducible depletion of mural cells showed a similar hyperplastic phenotype in the angiogenic retina, ex-plained by the fact that pericytes near the angiogenic front express VEGFR1, and use it to modulate the VEGF-A gradient in the vicinity of the tip cells 6. Another member of the Notch family, Notch3, is important in mural cell recruitment and survival. In zebrafish, it promotes PDGFR-B expression and mural cell differentiation in close proximity to arterial beds, from which they migrate towards the veins 61. Whether this model can be extrapolated to other organisms and tissues remains to be studied but, in mice, deletion of Notch3 impairs VSMC coverage 62.

Mural cells play an important role stabilizing the vasculature, which has traditionally been explained by the expression of angiopoietin 1 (Ang1), a paracrine ligand that activates the pan-endothelial receptor Tie2. Tie2 signal-ing promotes cell survival and mediates vascular stabilization and quies-cence (Figure 2), and both Tie2 and Ang1 are indispensable for vascular

20

development in the embryo 63–65. As a stabilizing factor, Ang1 modulates the angiogenic vasculature avoiding hypervascularization 66 and, in mature en-dothelium it can mitigate VEGF-A induced vascular leakage 67. As part of a negative feedback loop, Tie2 signaling downregulates Foxo1 activity, a pro-angiogenic transcription factor that promotes the expression of PDGF-B and MMPs that are necessary for sprouting 68. In addition, as part of its proangi-ogenic role, Foxo1 regulates metabolism, branching and vessel diameter 69. Interestingly, pharmacological inhibition of PDGF-B signaling hinders mu-ral cell recruitment to the vasculature, resulting in a highly disorganized endothelium, which can be restored through Ang1 administration 70, reaf-firming the fact that the combination of mural PDGFR-Β/Ang1 and endothe-lial Tie2/PDGF-B achieves a balance between both cell types that results in the maturation of the vasculature. It has recently been proposed that Tie1, an orphan receptor related to Tie2 and expressed mainly during development, competes with Tie2 at the cell membrane of tip cells, effectively modulating Tie2 signaling 71. There is a second ligand for Tie2, Ang2, which competes with Ang1, and behaves mainly as a proangiogenic and destabilizing factor. Ang2 is highly expressed in the tumor microenvironment and ischemic tis-sues, as well as in tip cells, where it acts in an autocrine manner 72–74. Re-cently, models of mural cell depletion in the mouse retina showed that Foxo1 activation in the absence of mural coverage can lead to Ang2 upregulation, causing inflammation and rendering the blood retinal barrier more sensitive to VEGF-A 7,75.

Part of the remodeling that is required to achieve a mature and functional vascular network consists in the specification of arteries and veins, establish-ing a clear route for blood to flow. Several molecules have been linked to arterial specification, for example ephrinB2 and EphB4 may guide angio-blast migration towards dorsal aorta or cardinal vein, respectively, in the zebrafish 46. The co-receptor Nrp1 is another of such molecules: when delet-ed from the endothelium during mouse limb development, only arteries fail to develop 41. Interestingly this protein is also required to guide the filopodia of tip cells in the embryonic hindbrain, which provides another indication of the existing coupling between tip cell and artery formation 40. The vascula-ture is far from being a static structure; in fact, tracing and live imaging techniques have revealed in recent years that endothelial cells are quite dy-namic throughout not only angiogenesis but also remodeling, and that endo-thelial migration is not limited to the tip position, in consequence, migratory defects in endothelial cells may originate vascular anomalies. A combination of zebrafish fin live imaging and mouse retina pulse-chase experiments pro-vided a useful model to visualize arteriovenous fate and proved that most endothelial cell proliferation happens within the veins rather than arteries. In addition, endothelial cells at the tip position of the sprouting front migrate towards the arteries, contributing to the longitudinal growth of these vessels 76 (Figure 3); and a series of mouse models of deletion or overexpression of

21

Notch1 signaling molecules determined that this pathway mediates the tran-sition from tip cell towards arterial fate in the retina 35. Furthermore, in zebrafish, arterial flow activates Notch1 signaling in the intersegmental ves-sels, maintaining arterial cells in place and keeping them from being re-placed by venous endothelial cells migrating against the flow 77.

Figure 3. Model of endothelial proliferation and migration: most endothelial prolif-eration occurs in veins (mitotic cells depicted in yellow), which contributes to their extension and thickening. Stalk cells near the sprouting front compete and overtake each other to reach the tip position. Tip cells extend filopodia and lead the sprout. Notch signaling controls the subsequent migration of cells at the tip towards the artery, where they contribute to the longitudinal growth of this vessel.

Flow plays yet another role in remodeling: early work pointed out that one of the early signs of vessel regression is blood stasis, or lack of flow. Capillar-ies through which blood does not flow are essentially useless and they re-gress, leaving behind empty sleeves of the collagen matrix that used to sur-round them 78. Some years later, the Notch pathway was linked to flow-mediated regression in physiological and pathological contexts: during hy-peroxia, blood vessels undergo vasoconstriction, and the lack of perfusion leads to regression of unnecessary capillaries; along the same lines, during physiological regression, redundant capillaries are pruned. During this pro-cess, Dll4-Notch signaling promotes negative selection and vessel regression by upregulating the expression of vasoconstrictive molecules 79. The mecha-

22

nism through which these obsolete capillaries regress in the mouse retina was unveiled recently, through an analysis based on cell polarity. This work studied the disposition of endothelial nuclei and Golgi apparatus, and con-firmed that endothelial cells orient themselves against the direction of blood flow. In capillaries with no flow, that is, redundant, endothelial cells are no longer all polarized in the same direction, but rather opposite directions, and they eventually migrate away from each other in reverse anastomosis, leav-ing an empty collagen sleeve 80.

The mouse retina: angiogenesis model The in vivo studies in this thesis are performed on the mouse retina, which is one of the most investigated models of angiogenesis. That is due to a number of reasons: firstly, mouse models have been around for a long time and there are plenty of genetic tools available for them. In addition, the eye is an organ that provides easy dissection access, but perhaps more importantly, the retina is a flat tissue that facilitates microscope imaging, and its vasculature grows in a stereotyped fashion that allows analyzing different developmental steps on the same tissue.

At birth, the mouse retina is avascular and the ocular blood supply comes from the hyaloid vessels, a transient vascular structure that eventually re-gresses. Immediately after birth, the retina becomes vascularized by angio-genic vessels that sprout from the optic disk 81. In this early postnatal stages, the endothelial cells crawl on top of a network of VEGF-A expressing astro-cytes that guide the nascent vasculature 82 to form the superficial plexus, which is characterized by a pattern of alternating arteries and veins that grow centrifugally towards the rim of the retina, with a capillary network extend-ing in between them. The vessels of the superficial plexus reach the rim by postnatal day seven to ten (P7-P10). After that, capillaries sprout down-wards, resulting in the formation of the deeper plexus, from which capillaries will subsequently grow upward again to generate the intermediate plexus. The three plexuses are fully developed by the third postnatal week, time during which the superficial plexus undergoes remodeling and pruning of excessive vessels 83 (Figure 4). All these steps can be easily visualized, thus, the mouse retina is a useful model to study several angiogenic processes such as sprouting, mural cell recruitment, vessel morphogenesis, and remod-eling.

23

Figure 4. Development of the murine retinal vasculature: depiction of one retinal wedge on postnatal days 2 (P2), P7, P15, and P30. The superficial vascular plexus grows centrifugally and after some days, alternating arteries and veins are distin-guishable from each other. After the vasculature reaches the rim of the retina, the dense superficial plexus undergoes remodeling and redundant capillaries are nega-tively selected and pruned. After the retina is fully remodeled, alternating veins have also regressed, and the remaining ones run along the rim of the retina. Below each wedge, there is a schematic representation of the development of the deeper and intermediate plexuses. When the superficial plexus is almost covering the whole surface, some endothelial cells sprout downwards to colonize the deeper plexus. After that, upward sprouting occurs to generate the intermediate plexus. This multi-layered vasculature is required to achieve a fine oxygenation of the neuronal retinal layers.

24

Endothelium and disease Like any other organ, the endothelium can undergo pathological processes if basic cellular processes such as proliferation, migration, remodeling, or mu-ral cell recruitment, fail. For example, angiomas are considered benign neo-plasms characterized by hyperplasia of endothelial cells, some of which can be present as birthmarks on the skin, as in the case of hemangiomas, or de-velop during ageing such as spider angiomas, also as visible marks on the skin 84. In addition, CCM are sometimes classified as angiomas, and they are characterized by vascular lesions in the brain consisting on enlarged lumens, which can be of familial or sporadic origin. Unlike hemangiomas or spider angiomas, these lesions may be fatal because they can rupture leading to brain bleedings. Familial cases have been linked to mutations in the CCM1/2/3 genes that cause upregulation of MAPK signaling, which might elicit clonal expansion of endothelial cells that undergo endothelial to mes-enchymal (EndMT) transition and become hyperproliferative 85,86.

Another type of vascular lesions are arteriovenous malformations (AVM), which consist of shunts: a direct connection between arteries and veins by-passing the capillary network. These type of malformations are frequent in HHT, a vascular disease linked to autosomal dominant mutations on the endoglin or Alk1 receptors that can cause chronic hemorrhages 87. Although the mechanisms behind this disease remain unclear, endoglin has been pro-posed to mediate flow-dependent endothelial migration and vascular remod-eling 88.

Cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy (CADASIL) is another vascular disease in which VSMC degradation leads to vascular lesions. This pathology is linked to Notch3 mutations, which regulates maturation, and recruitment of mural cells 89,90, and in mice, simultaneous deletion of Notch1/3 recapitulates CA-DASIL hallmarks and AVM 91.

In the eye, pericyte loss and aneurysm formation are early signs of DR, one of the complications of diabetes 92. In addition, neovascular age-related macular degeneration (AMD) is characterized by invasion of new blood vessels into the retina. These two conditions lead to BRB degradation and macular edema, and constitute two of the leading causes of blindness in the world 93.

Lastly, tumors can promote angiogenesis in their microenvironment to se-cure blood flow and enable their growth, which has led to numerous thera-peutic efforts based on antiangiogenic therapy, which has only been partially successful so far 57. The above-mentioned are only some of the examples that justify the continued efforts to unveil the intricacies of vascular devel-opment.

25

Cdc42, small Rho GTPase

Cellular functions of Cdc42 The discovery of Cdc42 was first published in 1990, after a screening of Saccharomyces cerevisiae mutants revealed a gene that rendered yeast una-ble to polarize and form the buds required for cell division, hence the name, cell division cycle 42 94. The sequence analysis of Cdc42 revealed that it is a small GTPase of 191 amino acids and 21kDa belonging to the family of Rho GTPases 95. As such, Cdc42 acts as molecular signaling switch that alter-nates between active (GTP-bound), and inactive (GDP-bound) states, and its activity is regulated by three types of factors: guanine-nucleotide exchange factors (GEFs), GTPase activating proteins (GAPs), and guanine-nucleotide dissociation inhibitors (GDIs). GEFs replace GDP by GTP, activating Cdc42. The signal transmitted by Cdc42 is shut off by GAPs, which acceler-ate the hydrolysis of GTP and elicit the GDP-bound state. Finally, Cdc42 is kept in the inactive state by GDIs, which prevent the release of GDP (Figure 5).

Cdc42 is highly conserved in eukaryotes and expressed in all cell types, and it was firstly hypothesized to play a role in cytoskeletal dynamics 96. This was promptly confirmed with the discovery that Cdc42 promoted filo-podia formation and focal adhesions in fibroblasts 97,98. However, unraveling the functions of Cdc42 soon proved to be a colossal task because, in subse-quent years, Cdc42 was shown to play roles in abounding cytoskeleton-associated cellular functions in vitro. Cdc42 binds Wiskott-Aldrich syn-drome protein (WASP) and Arp2/3, forming a complex that promotes actin polymerization by providing a nucleation site for the assembly of actin fila-ments 99–101. Work with Caenorhabditis elegans embryogenesis provided the first proof that Cdc42 is involved in polarity in multicellular organisms, which it achieves by regulating the Par3/Par6/atypical PKC (aPKC) complex 102,103. This pathway has an impact on both the actin filaments and the micro-tubules, and is important for epithelial apical-basal polarity and tubulogene-sis 104, as well as planar-cell polarity and maintaining directionality during migration 105,106. On the same note, Cdc42 regulates polarity-based collective migration 98, and is important during cytokinesis 107 and for vesicle traffick-ing 108. Apart from these functions, all of which are intertwined with cyto-skeletal dynamics, Cdc42 is involved in proliferation promoting G1 progres-sion during the cell cycle 109. To conclude, Cdc42 has been linked to tran-

26

scription factor regulation, as it plays part in the JNK/p38 and ERK MAPK signaling cascades 110–112, and activates actin-related serum response factor (SRF) 113.

As illustrated above, the role of Cdc42 as a molecular switch places it at the center of multiple signaling pathways, complicating the study of its func-tions. Moreover, most of the early investigations were based on dominant negative and constitutively active mutants in vitro. At the time, this greatly advanced our knowledge of the functions of Cdc42; however, many of the GEFs, GAPs and GDIs are promiscuous and interact with several Rho GTPases, at least in vitro. As a result, the use of this type of mutants might result in sequestration of a common factor, interfering with the normal func-tioning of the other GTPases, and making it difficult to discern their roles 114,115. With the advent of new molecular tools, it was soon possible to inacti-vate Cdc42 genetically, which led to some contradicting results regarding its relevance in filopodia formation, proliferation, migration or ERK signaling 116–118. Genetic deletion revealed that mouse embryonic stem cells need Cdc42 for normal actin assembly and cell morphology and that the global deletion of Cdc42 in vivo leads to lethality during early embryogenesis 116. Despite confirming the indispensability of Cdc42, embryonic lethality pre-vents us from learning anything else from Cdc42 in vivo. Conditional genetic deletions in mice through the Cre-loxP system bypass this problem, because they allow the study of a specific time point and/or cell type, which has led to corroborating most of the cell functions of Cdc42 in a wide range of living tissues.

Most of the functions of Cdc42 have been confirmed in different cell types in vivo: Cdc42 has been shown to promote apical-basal polarity through the Par3/Par6/aPKC complex in epithelial cells during tubulogenesis 119,120, and in neural progenitors regulates their post-mitotic differentiation 121. Additionally, defective polarity in stem cells with elevated Cdc42 activi-ty has been linked to stem cell aging 122. During heart development, Cdc42 is required for cardiomyocyte proliferation and adhesion 123, and in the adult heart, its deletion elicits cardiac hypertrophy after induction of stress because of deficient JNK signaling 124. Cdc42 deletion in neural crest cells elicits aberrant craniofacial development, which is caused by migration, but not proliferation nor apoptosis defects 125. In epithelial cells of the developing lens, Cdc42 mediates filopodia formation 126. Interestingly however, Cdc42 deletion in hematopoietic stem cells (HSCs) and T-cells promotes prolifera-tion 127,128, which seems to contravene most of the work on Cdc42, and re-flects how a protein may play different functions depending on cell type.

27

Figure 5. The cycle of Cdc42 activation: GEFs activate Cdc42 signaling facilitating the exchange of GDP for GTP. GAPs promote the GTPase activity of Cdc42, which turns promotes the inactive form of the protein, and GDIs ensure that Cdc42 remains in its inactive form, bound to GDP. The overall cellular functions of Cdc42 are listed on the lower part.

Cdc42 in the vasculature Soon, Cdc42 proved to be relevant also on endothelial cells. In vitro, Cdc42 mediates actin reorganization and cell junctions in response to external cues such as TNF-α, shear stress-dependent repositioning of the microtubule or-ganizing center (MTOC), and endothelial barrier function 129–132. It was ex-pected that a protein capable of affecting the architecture of several tissues and with proven roles in the endothelium would also have an impact on vas-cular morphogenesis. This was confirmed when studies in endothelial cells in vitro revealed that Cdc42 mediates the cytoskeletal rearrangements elicit-ed by VEGFR2 signaling 133, and regulates endothelial lumen formation via the Par3/Par6/aPKC complex 134,135. Furthermore, in cultures of embryoid bodies, endothelial cells present reduced chemotaxis and do not develop vascular networks in the absence of Cdc42 136. The study of Cdc42 in the vasculature was taken to a new level with the endothelial-specific deletion of

28

Cdc42 in mice. This experiment leads to embryonic death around E9.5 and, although it supported the relevance of Cdc42 in vasculogenesis, lumen for-mation and cell survival 137, such early lethality does not allow for a more in-depth analysis of the vascular functions of Cdc42 in vivo. Time-specific de-letion in mice using tamoxifen-dependent CreERT2 circumvented this obsta-cle and confirmed the importance of Cdc42 for polarity and lumen formation during mid-gestation, and for filopodia formation in the postnatal retina 138. Nevertheless, it also indicated that Cdc42 is necessary for endothelial prolif-eration but dispensable for cell survival, in contrast with the previous study, which reflects the complexity of the issue. In addition, Cdc42 regulates mi-gration and cell adhesion in vitro, and studies in zebrafish confirmed the importance of Cdc42 in filopodia and migration of endothelial cells by means of pharmacological inhibitors and morpholinos 139,140. For which of these numerous endothelial functions is Cdc42 responsible during retinal angiogenesis? Perhaps more importantly, which of these functions affect the morphogenesis of the vascular network and in which way?

Not surprisingly, literature on the role of Cdc42 in mural cells is scarce and, the few published works focus exclusively on VSMC rather than peri-cytes, despite the nomenclature they use. Cdc42 is downregulated in VSMC from pulmonary arterial hypertension patients, and these cells present im-paired polarity and migration in vitro 141. Additionally, Cdc42 activity has been linked to VSMC migration in an in vivo mouse model of intimal hyper-plasia 142. This lack of literature highlights the obscurity of mural cells, and provides a unique chance to explore the functions of Cdc42 in a novel con-text, as well as scrutinize the role of these cells in the angiogenic vascula-ture.

Cdc42 in disease Given the multitude of signaling pathways impinged by Cdc42, it should not come as a surprise that defects in this protein have ties to several types of diseases. For example, mutations in Fgd1, a GEF for Cdc42, are linked to Aarskog-Scott syndrome, a developmental disease 143. Recently, in vitro and in silico studies linked a series of activating and deactivating missense muta-tions of Cdc42 with an array of other developmental diseases, confirming the impact of both type of mutations on migratory and proliferative capabilities and reinforcing the central role of Cdc42 in signaling pathways of vital cel-lular functions 144.

As an important player in proliferation, migration, or adhesion, among others, Cdc42 has been linked to several types of cancer, where its expres-sion is abnormal, influencing malignant growth, and metastasis. Even though Cdc42 is not considered itself an oncogene, some of the molecules involved upstream and downstream of Cdc42 signaling promote tumorigenesis in

29

certain conditions 145, and several mutations of Cdc42 have been discovered in different tumor types through genome-wide sequencing 146.

30

Fluorescent reporters in the Cre-loxP system

Mice have always been one of the most used vertebrate animals in research, however, the number of mouse-related studies skyrocketed in the early 90s, and today they make up for 60% of the animals used in research in the Euro-pean Union 147,148. One of the main reasons for this inflection point in mouse popularity was the ability to perform targeted genetic manipulation that came with the advent of the Cre-loxP system 149. Cre (cyclization recombina-tion) protein is a recombinase that was first discovered in the bacteriophage P1, where it functions recombining genetic material flanked by two loxP (locus of X-over-P1) sequences, each of which consist of two inverted re-peats of 13 base pairs (bp) with an 8 bp central region 150. The Cre-loxP sys-tem as used nowadays in research is a powerful gene-targeting tool in which the genomic region of interest, usually a gene or part of it, is substituted by a modified sequence that is identical but flanked by loxP sites. When Cre ac-cesses the nucleus of the cell, it splices the genetic fragment flanked by loxP sites, effectively knocking out the gene.

The usefulness of the Cre-loxP system comes from the possibility to re-strict its action in time and place: when the recombinase is inserted under the control of a gene promoter, Cre expression becomes coupled to that of the gene. When this coupling is done with a cell type-specific marker, only said cell type will be targeted by Cre, which allows studying the role of a gene in a particular population, for example, Tie2-Cre targets endothelial cells 151. Constitutively active Cre lines have the disadvantage of often leading to embryonic death, if the gene of interest is of importance during develop-ment, which hinders the study of later stages. This led to the creation of in-ducible Cre lines, in which Cre is fused to a fragment of the human estrogen receptor 152. In baseline conditions, the receptor keeps the recombinase con-fined into the cytosol, and administration of tamoxifen, an estrogen receptor ligand, elicits its translocation to the nucleus (Figure 6). The system was quickly improved by the same group with the introduction of several muta-tions to avoid endogenous estrogen activation, and it resulted in the creation of the CreERT2 system, which is, in theory, tight and does not leak 153,154. CreER and CreERT2 lines are widely used nowadays, and they allow full spatiotemporal control, enabling deletion of a gene in a specific cell type at a desired time point.

31

Figure 6. Mechanism behind fluorescent reporter recombination in the CreERT2-loxP system and its impact on lineage tracing. Tamoxifen elicits translocation of CreERT2 to the nucleus, where it can recombine loxP sites. Without the STOP cas-sette, the fluorescent reporter becomes expressed. In a lineage tracing experiment, cells of interest become labelled and their progeny studied. In the case of back-ground recombination, cells that are fluorescently labelled in the absence of tamoxi-fen represent a confounding factor in the experiment.

32

Fluorescent reporter genes can be used in combination with the Cre-loxP to visualize which cells undergo recombination. In order to do this, a STOP cassette is placed within a loxP-flanked region, and a fluorescent gene is inserted downstream of the floxed sequence: in the absence of recombina-tion, the fluorescent reporter is not transcribed, whereas after tamoxifen in-duction and recombination, the STOP sequence is spliced and the fluorescent gene expressed. Fluorescent reporters are useful when studying a mosaic phenotype in which not all cells are mutants and thus, contribute differently to the phenotypic outcome 155. Furthermore, fluorescent labeling is frequent-ly employed to track cell populations. One example of this type of experi-ments is lineage tracing, which involves labeling of progenitor cells and analyzing their proliferation and fate acquisition patterns in vivo 156. In this context, a high-fidelity Cre-loxP system with limited leakiness is desirable, to minimize potential confounding results (Figure 6). Many reporter lines exist nowadays, created from combinations of different fluorescent proteins and scaffolds, and some of them have raised concerns of spontaneous activa-tion 157,158. Given the relevance of CreERT2 lines, detailed knowledge of their pitfalls and awareness of potential experimental problems would be helpful for the scientific community.

33

Present investigations

Paper I Despite the numerous functions of Cdc42 in different cell types, not much research has been done on the endothelium. For this reason, we wanted to investigate what of the known roles of Cdc42 are important for endothelial cells, especially in vivo during angiogenesis, and what the result would be for vascular morphogenesis carried out by Cdc42-deficient endothelial cells. In this study, we crossed Cdh5-CreERT2, Cdc42flox/flox (Cdc42iΔEC) mice and deleted Cdc42 in endothelial cells during retinal postnatal angiogenesis.

We found that the retinas of Cdc42iΔEC mice presented malformations that were restricted to veins and capillaries. These malformations were character-ized by enlarged lumens and were totally absent from arteries. To dissect this phenotype, and knowing the importance of Cdc42 for migration, we ana-lyzed the distribution of Cdc42-deficient cells in mosaic retinas through the presence of R26R-EYFP reporter and discovered that KO cells lagged be-hind the sprouting front. Further arteriovenous distribution analysis revealed that Cdc42 cells clustered around veins but were underrepresented in arter-ies, in contrast to control cells of wild type retinas. In addition, imaging of aortic explant sprouting and motility assays in vitro confirmed that the motil-ity and directed migration of Cdc42-deficient cells is impaired. In combina-tion with analyses demonstrating that proliferation in Cdc42iΔEC retinas is not increased, we discarded hyperplasia as a potential reason for the malfor-mations. Instead, we concluded that impaired migration in regions of natu-rally occurring high rates of proliferation (such as veins) impairs efficient redistribution of endothelial cells, causing their aggregation and subsequent vessel enlargement.

During the distribution analyses, we noticed that even in cases where re-combination was so high that R26R-EYFP cells reached the sprouting front for lack of competitors, these cells presented a disadvantage for the tip cell position and that, when they did reach the tip, they lacked filopodia. In addi-tion, axial but not apical-basal polarity was disturbed in Cdc42 cells, and adherens and tight junctions were abnormal. An in silico assay corroborated that impaired filopodia and axial polarity would contribute to the observed tip disadvantage. These results, in summary, indicate that Cdc42-deficient endothelial cells cause vascular malformations during angiogenesis in vivo due to impaired migration at normal proliferative levels.

34

Paper II During our research, we came across tamoxifen-independent activation of fluorescent reporters. In this work, we wanted to analyze in depth and report such downside that may be encountered while working with CreERT2 lines in combination with fluorescent reporter genes. In order to perform a thorough analysis and delineate the problem, we crossed three vascular CreERT2 lines: Prox1-, Pdgfb- and Cdh5-CreERT2 with the tdTomato-based Ai14 reporter. Immunohistochemistry of several organs and developmental stages of these mice confirmed that all three lines presented tamoxifen-independent back-ground activity that was capable of eliciting recombination. Importantly, background recombination required Cre expression and did not occur in other cell types.

After confirming that all studied CreERT2 lines may elicit background re-combination, we analyzed whether a variety of fluorescent reporter lines react differently to it. To assess this, we crossed four commonly used report-ers that are built differently: Ai14, Ai3, mTmG, and R26R-EYFP, to the vascular Cdh5-CreERT2 line and performed immunohistochemistry on brains and retinas. In both organs, the Ai14 and Ai3 reporters presented a lower recombination threshold than mTmG and R26R-EYFP, resulting in wide-spread endothelial labeling in the absence of tamoxifen. Furthermore, in some of our experiments, we detected Ai14+ and Ai3+ microglia, which are not known to express Cdh5, indicating that an inherited recombination event had happened in an earlier, Cdh5+ progenitor cell.

Thanks to the combination of four reporters that we used, we concluded that, of all their distinctive characteristics, distance between loxP sites was the most likely responsible to influence these results, rather than chromatin packing, promoter strength, or fluorescent brightness. This work showed loxP distance plays an important role in reporter recombination, and these lines should always be tested and compared to controls for each application. Experiments requiring accurate single-cell tracking, such as in lineage trac-ing, are especially sensitive, due to the risk of occurrence of tamoxifen-independent recombination events.

Paper III The role of mural cells: pericytes and vascular smooth muscle cells (VSMC) is largely unknown during angiogenesis in vivo. In consequence, to expand the research of Cdc42 in the vasculature, we decided to study its function in mural cells. In this study, we crossed Pdgfrb-CreERT2, Cdc42flox/flox (Cdc42iΔMC) mice and deleted Cdc42 in pericytes and VSMC during retinal postnatal angiogenesis. In this study, we found that VSMC and pericytes required Cdc42 during angiogenesis to acquire their normal morphologies,

35

and that Cdc42-deficient pericytes regained a normal shape when they ma-tured, whereas VSMC did not. More specifically, during angiogenic stages, venous VSMC and pericytes of Cdc42iΔMC retinas presented elongated bod-ies, and arterial VSMC displayed an abnormal arrangement of their cyto-skeleton, leading to impaired wrapping around the arteries.

Interestingly, we discovered that Cdc42-deficient pericytes presented de-layed collective migration towards the sprouting front and left a stretch of vasculature uncovered, which lead to increased erythrocyte extravasation. To explore this phenotype, we performed proliferation analyses, which demon-strated that Cdc42-deficient mural cells presented decreased proliferation capacity in vivo. Such decrease in proliferation contributed to the observed reduced coverage behind the tip cells, since we also found that most mural cell proliferation naturally occurred immediately behind the sprouting front during angiogenesis. In addition, the vasculature of angiogenic Cdc42iΔMC retinas presented a high degree of disorganization, with decreased branching points, increased number of empty collagen sleeves, and abnormal direction-ality of arterioles, among others. We hypothesized that Cdc42 deficient peri-cytes lagging behind the sprouting front create transient conditions of peri-cyte deletion, leading to several morphogenetic defects in the angiogenic vasculature.

During the analysis of mature retinas, we discovered that the superficial plexus presented strikingly reduced arteriolar branching angles, which could be related to the impaired orientation of Cdc42-deficient arterial VSMC, which remained throughout vascular remodeling and could have influenced the final shape of the vessels. To sum up, we showed that pericytes need Cdc42 to maintain the proliferative levels required to keep up with the angi-ogenic front. We also identified VSMC as an important regulator of post-angiogenic vascular remodeling.

36

Future perspectives

In this work, we unveil several aspects of Cdc42 biology in the endothelium, as well as shed some light on the relationship between endothelial and mural cells during angiogenesis. However, some questions remain unanswered that could be expanded with our genetic models.

In paper III, we hypothesized that proliferation contributed to the de-creased pericyte coverage in Cdc42iΔMC retinas. While this is most likely true for the reduction in pericyte coverage that we find overall in the capillary plexus, we believe that it only partially explains the striking phenotype at the sprouting front. If the decreased proliferation in Cdc42 pericytes were the sole reason for the lack of coverage at the front, pericytes would be there, only sparser. However, behind the tip cells, we find a stretch of vasculature that is totally void of pericytes, which is why we believe impaired migration in these cells also plays a role. To obtain evidence that Cdc42 mediates mi-gration in mural cells, we will perform several in vitro motility assays. Working with primary cells in vitro will allow us to study general motility and cytoskeleton dynamics using dyes that label actin, comparing control and Cdc42KO cells. Furthermore, we are planning to perform scratch wound assays to evaluate directed migration in mural cells. To do this, we are cur-rently performing mural cell isolations from brains of Cdc42iΔMC mice. These combined experiments will improve our knowledge of mural cell re-cruitment during angiogenesis in vivo.

Additionally, it would be interesting to perform a detailed analysis of the Cdc42iΔMC animals, which sometimes die during early adulthood. They fre-quently present craniofacial defects, short limbs and snout, and very small bodies compared to their littermates. We performed preliminary skeletal stainings that suggest calcification defects in the bone of these mice, and it would be worth it to follow it up.

In line with the current expertise of the lab, we are also planning to isolate mural cells from the brain of Cdc42iΔMC mice and perform single-cell RNA sequencing on them. Single-cell RNA sequencing is a powerful tool that allows studying the expression pattern of individual cells isolated from an organ. Through this technique, we can understand which genes become up- or downregulated in single cells of our population of interest, rather than studying whole tissues, and it constitutes an analogue innovation to that of the Cre-loxP system, which made the field of genetics take a leap forward. With this study, we would compare the mRNA expression pattern of normal

37

and Cdc42-deficient mural cells. In addition, this would open the door to comparing the mural and endothelial KO profiles. The latter, a continuation of paper I, is already underway, and it will shed more light on Cdc42 biolo-gy.

The boundary between pericytes and VSMC is not clear and is still often based on location on the vessels. Due to its stereotypic vasculature, the retina provides a perfect model for the differential study of pericytes and VSMC. An interesting experiment that would give us information about their shared origin, and whether one cell type can differentiate into the other, is lineage tracing. Crossing Pdgfrb-CreERT2 mice with a multicolor reporter line such as the Brainbow or Dual ifgMosaic mice 159,160 would provide clear answers to how both cell types are related in the retina.

38

Acknowledgements

I write these lines knowing that they are the only part that approximately 99% of you will read, so I hope I don’t mess it up... This has been a long but fruitful and fun journey, and that’s due to the people that surrounded me during my years in Uppsala, whom I would like to thank for their support.

First to my supervisor, Kon: we did it! Thank you for these 6 years, and for everything you have taught me along the way, which is really diverse and includes but is not limited to: your view of science, type fonts, or beers. You have been a source of positivity, and a boss with whom I could joke around; I believe we really formed a perfect team. To my other supervisor during these years: Christer, thank you for taking me under your wing, I am grate-ful for having experienced the kindness you instill in this lab first hand.

I would like to thank my co-supervisor, Taija, and my examiner, Pontus, who had my back even if thankfully no crisis (other than a pandemic) hap-pened. Also, I am grateful to the students that contributed along the years to my projects: María Gª., Mina, Andrea, Susmita, Claudia, and Valéria.

Not everyone is as lucky to have such a wonderful group of people in their workplaces, and the list in this one is quite long! Thanks to the CBZ lab for these years full of great moments: Maarja (who loved teasing me but I know that it was always in good faith), Michael (who shared musical and literary tastes with me), Johanna, Riikka, Liqun, Marie, Ying, Jana… Hélène (the coolest person I know), Cissi (who tried to make sure I learnt enough Swedish to not make her ashamed), and Pia (who taught me a couple of things of horses and cars). Colin, who was friend and mentor in and out of the lab, Koji, Rajesh, Bàrbara, Bong, and Tatti.

Despite Khayrun stealing my bench on our first day, she ended up being the best labmate I could have wished for, source of fun moments, gossip, and mutual “I might have done something stupid, no questions asked”. Towards the end, she turned into the wise one and made sure I didn’t stumble on any of the bureaucratic PhD steps, I might still be filling applications were it not for her. The other member of The Office trinity, Tove, was a great addition to the team and brought lots of fun, but also enough productivity levels to almost get our work done for us, in addition to hers. She really has every-thing under control; rumor has it that her only weak spot might be throwing a ball. Elisa has been coming and going for so long that I lost track, however she is here to stay and I know the lab is in good hands with her as the new PhD student. I don’t think Miguel ever said no to going out for a beer, like,

39

ever. And despite my poor excuses time and time again, he managed to drag me along most of the time, which I greatly appreciated. The rest of the Root-beck Peeps: Mami, Rubén, Sarah R., Maureen, and Magda, were always up for a good toilet paper discussion, and that is something to be cherished in these times.

I want to mention some of the people that used to be around but left for their next adventures: Marco was like my big brother in the lab, I inherited his projects, his techniques…he taught me a lot, scientifically and otherwise. Yang was a wholesome person to have around, always brightening every-one’s day. Leonor…well, she is still somewhat here, regardless, I have al-ways considered her an example of motivation and determination.

I also want to thank some people with whom I spent some quality gaming time, first at Rudbeck and nowadays online: Ross, Milena, Eric and Vero.

During my time at Rudbeck I have been lucky to meet a lot of other peo-ple, who have created a very nice environment in the building: Roberta, Sarah S., Di, Nina, Inés, Simon, Yan, Elin, Dom, Tor, Svea, Alba, Pat-rick, Matko, Emelie, Maria Gl., Alessandra, Kalyani…and the list goes on…sorry for not including everyone. Also Carlos, who never lost the chance to have a political discussion with pizza and beers in Flogsta.

As for the rugby club, I met many great people; Sam and Peter are two of the kindest guys I know. Moose helped me believe in myself, and I owe him becoming a better player, despite of what he might think. I didn’t really meet Mark at the club, but it was fun playing alongside him on the weekend while discussing science during the week.

I am grateful for the amazing people that started this journey with me, my knitting club: Marta taught me that madrileños do have an accent, and thought of me as a grandpa, but was always there for me. I still owe a visit to Krzysztof, who will no doubt pour some vodka in my honor. Christos took me in when I needed it, and on top of that showed me some grandma love worrying about my milk and cookie diet, dude we have lived a lot together. Sitaf, who didn’t make fun of me for being shorter than her; and Jessica, who remained my Swedish partner until I quit the end.

¿Qué sería de mí sin mi familia? Papá y mamá, me animasteis a venir a Uppsala (con lo indeciso que soy, tuvieron que ser meses bastante insoporta-bles) y encima me acabé quedando mucho más tiempo del planeado. Aun así, me habéis apoyado siempre, con eso y todo; gracias por sentar los ci-mientos (y las columnas, ladrillos...) de lo que somos hoy. Hermanos, o la distancia o la madurez me han ayudado a darme cuenta de cuánto os admiro y la suerte que tengo de teneros, en cualquier caso, menos mal que dejamos la adolescencia atrás…Bea, no sé qué hubiese sido de mí sin ti…ni quiero saberlo. Has tenido una paciencia infinita conmigo (en todos los sentidos) y sé que no ha sido fácil, especialmente estos últimos meses. Gracias por ser mi compañera de aventuras, para ésta, y tooodas las que vienen. Juntos.

40

References

1. Armulik, A., Genové, G. & Betsholtz, C. Pericytes: Developmental, Physio-logical, and Pathological Perspectives, Problems, and Promises. Developmen-tal Cell 21, 193–215 (2011).

2. Yamazaki, T. & Mukouyama, Y. Tissue Specific Origin, Development, and Pathological Perspectives of Pericytes. Front Cardiovasc Med 5, (2018).

3. Vanlandewijck, M. et al. A molecular atlas of cell types and zonation in the brain vasculature. Nature 554, 475–480 (2018).

4. Lindahl, P., Johansson, B. R., Levéen, P. & Betsholtz, C. Pericyte Loss and Microaneurysm Formation in PDGF-B-Deficient Mice. Science (1997).

5. Armulik, A. et al. Pericytes regulate the blood–brain barrier. Nature 468, 557–561 (2010).

6. Eilken, H. M. et al. Pericytes regulate VEGF-induced endothelial sprouting through VEGFR1. Nature Communications (2017).

7. Young Park et al. Plastic roles of pericytes in the blood–retinal barrier. Na-ture Communications (2017).

8. Diéguez-Hurtado, R. et al. Loss of the transcription factor RBPJ induces disease-promoting properties in brain pericytes. Nat Commun 10, 1–19 (2019).

9. Fatigati, V. & Murphy, R. A. Actin and tropomyosin variants in smooth mus-cles. Dependence on tissue type. J. Biol. Chem. 259, 14383–14388 (1984).

10. Fernández-Klett, F., Offenhauser, N., Dirnagl, U., Priller, J. & Lindauer, U. Pericytes in capillaries are contractile in vivo, but arterioles mediate func-tional hyperemia in the mouse brain. Proc Natl Acad Sci U S A 107, 22290–22295 (2010).

11. Hill, R. A. et al. Regional Blood Flow in the Normal and Ischemic Brain Is Controlled by Arteriolar Smooth Muscle Cell Contractility and Not by Capil-lary Pericytes. Neuron (2015).

12. Hall, C. N. et al. Capillary pericytes regulate cerebral blood flow in health and disease. Nature 508, 55–60 (2014).

13. Dore-Duffy, P., Katychev, A., Wang, X. & Van Buren, E. CNS microvascu-lar pericytes exhibit multipotential stem cell activity. J. Cereb. Blood Flow Metab. 26, 613–624 (2006).

14. Goritz, C. et al. A Pericyte Origin of Spinal Cord Scar Tissue. Science (2011).

15. Sava, P. et al. Human pericytes adopt myofibroblast properties in the micro-environment of the IPF lung. JCI Insight (2017).

16. Ambler, C. A., Nowicki, J. L., Burke, A. C. & Bautch, V. L. Assembly of Trunk and Limb Blood Vessels Involves Extensive Migration and Vasculo-genesis of Somite-Derived Angioblasts. Developmental Biology 234, 352–364 (2001).

17. Strilić, B. et al. The Molecular Basis of Vascular Lumen Formation in the Developing Mouse Aorta. Developmental Cell 17, 505–515 (2009).

41

18. Gianni-Barrera, R. et al. VEGF over-expression in skeletal muscle induces angiogenesis by intussusception rather than sprouting. Angiogenesis (2012).

19. Groppa, E. et al. EphrinB2/EphB4 signaling regulates non sprouting angio-genesis by VEGF. EMBO reports (2018).

20. Potente, M., Gerhardt, H. & Carmeliet, P. Basic and Therapeutic Aspects of Angiogenesis. Cell 146, 873–887 (2011).

21. Arroyo, A. G. & Iruela-Arispe, M. L. Extracellular matrix, inflammation, and the angiogenic response. Cardiovasc Res 86, 226–235 (2010).

22. Eilken, H. M. & Adams, R. H. Dynamics of endothelial cell behavior in sprouting angiogenesis. Curr. Opin. Cell Biol. 22, 617–625 (2010).

23. Vandekeere, S., Dewerchin, M. & Carmeliet, P. Angiogenesis Revisited: An Overlooked Role of Endothelial Cell Metabolism in Vessel Sprouting. Mi-crocirculation 22, 509–517 (2015).

24. Betsholtz, C. Cell–cell signaling in blood vessel development and function. EMBO Molecular Medicine 10, e8610 (2018).

25. Álvarez-Aznar, A., Muhl, L. & Gaengel, K. Chapter Thirteen - VEGF Recep-tor Tyrosine Kinases: Key Regulators of Vascular Function. in Current Top-ics in Developmental Biology (ed. Jenny, A.) vol. 123 433–482 (Academic Press, 2017).

26. Simons, M., Gordon, E. & Claesson-Welsh, L. Mechanisms and regulation of endothelial VEGF receptor signalling. Nat Rev Mol Cell Biol 17, 611–625 (2016).

27. Gerhardt, H. et al. VEGF guides angiogenic sprouting utilizing endothelial tip cell filopodia. The Journal of Cell Biology (2003).

28. Hellstrom, M. et al. Dll4 signalling through Notch1 regulates formation of tip cells during angiogenesis. Nature (2007).

29. Funahashi, Y. et al. Notch regulates the angiogenic response via induction of VEGFR-1. J Angiogenes Res 2, 3 (2010).

30. Jakobsson, L. et al. Endothelial cells dynamically compete for the tip cell position during angiogenic sprouting. Nature Cell Biology (2010).

31. Benedito, R. et al. The Notch Ligands Dll4 and Jagged1 Have Opposing Effects on Angiogenesis. Cell (2009).

32. Pontes-Quero, S. et al. High mitogenic stimulation arrests angiogenesis. Nature Communications 10, 2016 (2019).

33. Ubezio, B. et al. Synchronization of endothelial Dll4-Notch dynamics switch blood vessels from branching to expansion. eLife 5, e12167 (2016).

34. Hasan, S. S. et al. Endothelial Notch signalling limits angiogenesis via con-trol of artery formation. Nature Cell Biology (2017).

35. Pitulescu, M. E. et al. Dll4 and Notch signalling couples sprouting angiogen-esis and artery formation. Nat Cell Biol 19, 915–927 (2017).

36. De Bock, K. et al. Role of PFKFB3-Driven Glycolysis in Vessel Sprouting. Cell 154, 651–663 (2013).

37. Sauteur, L. et al. Cdh5/VE-cadherin Promotes Endothelial Cell Interface Elongation via Cortical Actin Polymerization during Angiogenic Sprouting. Cell Reports 9, 504–513 (2014).

38. Gaengel, K. et al. The Sphingosine-1-Phosphate Receptor S1PR1 Restricts Sprouting Angiogenesis by Regulating the Interplay between VE-Cadherin and VEGFR2. Developmental Cell 23, 587–599 (2012).

39. Jung, B. et al. Flow-Regulated Endothelial S1P Receptor-1 Signaling Sus-tains Vascular Development. Developmental Cell (2012).

40. Gerhardt, H. et al. Neuropilin-1 is required for endothelial tip cell guidance in the developing central nervous system. Dev. Dyn. 231, 503–509 (2004).

42

41. Mukouyama, Y.-S., Gerber, H.-P., Ferrara, N., Gu, C. & Anderson, D. J. Peripheral nerve-derived VEGF promotes arterial differentiation via neuro-pilin 1-mediated positive feedback. Development 132, 941–952 (2005).

42. Gu, C. et al. Semaphorin 3E and Plexin-D1 Control Vascular Pattern Inde-pendently of Neuropilins. Science 307, 265–268 (2005).

43. Zhang, Y. et al. Tie2Cre-mediated inactivation of plexinD1 results in con-genital heart, vascular and skeletal defects. Dev. Biol. 325, 82–93 (2009).

44. Brantley-Sieders, D. M., Fang, W. B., Hwang, Y., Hicks, D. & Chen, J. Ephrin-A1 facilitates mammary tumor metastasis through an angiogenesis-dependent mechanism mediated by EphA receptor and vascular endothelial growth factor in mice. Cancer Res. 66, 10315–10324 (2006).

45. Hunter, S. G. et al. Essential role of Vav family guanine nucleotide exchange factors in EphA receptor-mediated angiogenesis. Mol. Cell. Biol. 26, 4830–4842 (2006).

46. Herbert, S. P. et al. Arterial-venous segregation by selective cell sprouting: an alternative mode of blood vessel formation. Science 326, 294–298 (2009).

47. Foo, S. S. et al. Ephrin-B2 controls cell motility and adhesion during blood-vessel-wall assembly. Cell 124, 161–173 (2006).

48. Salvucci, O. et al. EphrinB reverse signaling contributes to endothelial and mural cell assembly into vascular structures. Blood 114, 1707–1716 (2009).

49. Erber, R. et al. EphB4 controls blood vascular morphogenesis during postna-tal angiogenesis. EMBO J. 25, 628–641 (2006).

50. Jones, C. A. et al. Robo4 stabilizes the vascular network by inhibiting patho-logic angiogenesis and endothelial hyperpermeability. Nat. Med. 14, 448–453 (2008).

51. Zhang, B. et al. Repulsive axon guidance molecule Slit3 is a novel angiogen-ic factor. Blood 114, 4300–4309 (2009).

52. Lu, X. et al. The netrin receptor UNC5B mediates guidance events control-ling morphogenesis of the vascular system. Nature 432, 179–186 (2004).

53. Nacht, M. et al. Netrin-4 regulates angiogenic responses and tumor cell growth. Exp. Cell Res. 315, 784–794 (2009).

54. Wiley, D. M. et al. Distinct Signaling Pathways Regulate Sprouting Angio-genesis from the Dorsal Aorta and Axial Vein. Nat Cell Biol 13, 686–692 (2011).

55. Larrivée, B. et al. ALK1 Signaling Inhibits Angiogenesis by Cooperating with the Notch Pathway. Dev Cell 22, 489–500 (2012).

56. Ricard, N. et al. BMP9 and BMP10 are critical for postnatal retinal vascular remodeling. Blood 119, 6162–6171 (2012).

57. Ferrara, N. & Adamis, A. P. Ten years of anti-vascular endothelial growth factor therapy. Nat Rev Drug Discov 15, 385–403 (2016).

58. Mac Gabhann, F., Qutub, A. A., Annex, B. H. & Popel, A. S. Systems biolo-gy of pro-angiogenic therapies targeting the VEGF system. Wiley Interdiscip Rev Syst Biol Med 2, 694–707 (2010).

59. Lindblom, P. et al. Endothelial PDGF-B retention is required for proper in-vestment of pericytes in the microvessel wall. Genes Dev. 17, 1835–1840 (2003).

60. Hellström, M. et al. Lack of Pericytes Leads to Endothelial Hyperplasia and Abnormal Vascular Morphogenesis. J Cell Biol 153, 543–554 (2001).

61. Ando, K. et al. Peri-arterial specification of vascular mural cells from naïve mesenchyme requires Notch signaling. Development (2019) doi:10.1242/dev.165589.

43

62. Domenga, V. et al. Notch3 is required for arterial identity and maturation of vascular smooth muscle cells. Genes Dev 18, 2730–2735 (2004).

63. Davis, S. et al. Isolation of Angiopoietin-1, a Ligand for the TIE2 Receptor, by Secretion-Trap Expression Cloning. Cell 87, 1161–1169 (1996).

64. Puri, M. C., Rossant, J., Alitalo, K., Bernstein, A. & Partanen, J. The receptor tyrosine kinase TIE is required for integrity and survival of vascular endothe-lial cells. EMBO Journal 14, 5884–5891 (1995).

65. Suri, C. et al. Requisite role of angiopoietin-1, a ligand for the TIE2 receptor, during embryonic angiogenesis. Cell 87, 1171–1180 (1996).

66. Jeansson, M. et al. Angiopoietin-1 is essential in mouse vasculature during development and in response to injury. J. Clin. Invest. 121, 2278–2289 (2011).

67. Thurston, G. et al. Leakage-Resistant Blood Vessels in Mice Transgenically Overexpressing Angiopoietin-1. Science 286, 2511–2514 (1999).

68. Potente, M. et al. Involvement of Foxo transcription factors in angiogenesis and postnatal neovascularization. J Clin Invest 115, 2382–2392 (2005).

69. Wilhelm, K. et al. FOXO1 couples metabolic activity and growth state in the vascular endothelium. Nature (2016).

70. Uemura, A. et al. Recombinant angiopoietin-1 restores higher-order architec-ture of growing blood vessels in mice in the absence of mural cells. J. Clin. Invest. 110, 1619–1628 (2002).

71. Savant, S. et al. The Orphan Receptor Tie1 Controls Angiogenesis and Vas-cular Remodeling by Differentially Regulating Tie2 in Tip and Stalk Cells. Cell Reports 12, 1761–1773 (2015).

72. del Toro, R. et al. Identification and functional analysis of endothelial tip cell–enriched genes. Blood 116, 4025–4033 (2010).

73. Oh, H. et al. Hypoxia and Vascular Endothelial Growth Factor Selectively Up-regulate Angiopoietin-2 in Bovine Microvascular Endothelial Cells. J. Biol. Chem. 274, 15732–15739 (1999).

74. Vajkoczy, P. et al. Microtumor growth initiates angiogenic sprouting with simultaneous expression of VEGF, VEGF receptor-2, and angiopoietin-2. J Clin Invest 109, 777–785 (2002).

75. Ogura, S. et al. Sustained inflammation after pericyte depletion induces irre-versible blood-retina barrier breakdown. JCI Insight 2, (2017).

76. Xu, C. et al. Arteries are formed by vein-derived endothelial tip cells. Nat Commun 5, 1–11 (2014).

77. Weijts, B. et al. Blood flow-induced Notch activation and endothelial migra-tion enable vascular remodeling in zebrafish embryos. Nat Commun 9, 1–11 (2018).

78. Baluk, P. et al. Regulated Angiogenesis and Vascular Regression in Mice Overexpressing Vascular Endothelial Growth Factor in Airways. Am J Pathol 165, 1071–1085 (2004).

79. Lobov, I. B. et al. The Dll4/Notch pathway controls postangiogenic blood vessel remodeling and regression by modulating vasoconstriction and blood flow. Blood 117, 6728–6737 (2011).

80. Franco, C. A. et al. Dynamic Endothelial Cell Rearrangements Drive Devel-opmental Vessel Regression. PLOS Biology (2015).

81. Mitchell, C. A., Risau, W. & Drexler, H. C. A. Regression of vessels in the tunica vasculosa lentis is initiated by coordinated endothelial apoptosis: A role for vascular endothelial growth factor as a survival factor for endotheli-um. Developmental Dynamics 213, 322–333 (1998).

44

82. Dorrell, M. I., Aguilar, E. & Friedlander, M. Retinal vascular development is mediated by endothelial filopodia, a preexisting astrocytic template and spe-cific R-cadherin adhesion. Invest. Ophthalmol. Vis. Sci. 43, 3500–3510 (2002).

83. Stahl, A. et al. The Mouse Retina as an Angiogenesis Model. Invest Oph-thalmol Vis Sci 51, 2813–2826 (2010).

84. George, A., Mani, V. & Noufal, A. Update on the classification of hemangi-oma. J Oral Maxillofac Pathol 18, S117–S120 (2014).

85. Bravi, L. et al. Endothelial Cells Lining Sporadic Cerebral Cavernous Mal-formation Cavernomas Undergo Endothelial-to-Mesenchymal Transition. Stroke 47, 886–890 (2016).

86. Malinverno, M. et al. Endothelial cell clonal expansion in the development of cerebral cavernous malformations. Nat Commun 10, 1–16 (2019).

87. Lebrin, F. et al. Thalidomide stimulates