Microvascular plasticity in the healthy and diseased mouse cortex

(1)

by Patrick Reeson

B.Sc., University of Calgary, 2005

A Dissertation Submitted in Partial Fulfillment of the Requirements for the Degree of

DOCTOR OF PHILOSPHY

in the Division of Medical Sciences (Neuroscience)

 Patrick Reeson, 2018 University of Victoria

(2)

Supervisory Committee

Microvascular plasticity in the healthy and diseased mouse cortex by

Patrick Reeson

B.Sc., University of Calgary, 2005

Supervisory Committee

Dr. Craig E. Brown, Division of Medical Sciences Supervisor

Dr. Patrick C. Nahirney, Division of Medical Sciences Departmental Member

Dr. Bob Chow, Department of Biology Outside Member

(3)

Abstract

The brain relies on a properly functioning vasculature system to deliver oxygen and nutrients and remove metabolic waste. However as in all biological systems, the brain is sometimes challenged by small or large-scale failures in the vascular system, which threaten the neuronal networks they support. Cerebral capillaries are uniquely prone to spontaneous obstructions, randomly stopping flow on a moment to moment basis. While not surprising given that capillaries are narrow, low pressure tubes that pass relatively large and adherent cells and debris, the ultimate outcomes of these obstructions are unknown. The vascular response to these events could have profound effects on brain health, as these random events accumulate over time. Similarly, while much research has studied the neural and vascular responses to large vessel obstructions (ischemic stroke), how common comorbidities which also afflict the vasculature, like diabetes, alters vascular plasticity and in turn neuronal rewiring and functional recovery, is not understood. This dissertation furthers our understanding of how microvascular plasticity, in response to either small or larger interruptions to blood flow, affect brain health.

In the first aim I examined the fates of cortical capillaries in the mouse somatosensory cortex to either spontaneous or experimentally induced obstructions. Using in vivo 2 photon imaging of cortical blood flow, I found that ~0.12% of cortical capillaries become obstructed each day. Tracking natural or microsphere induced obstructions in anesthetized or awake mice revealed that most capillaries recanalize. Remarkably, 30% of all obstructed capillaries failed to recanalize and were pruned by 21 days. This loss was not compensated for by any angiogenic sprouting in any imaging area. Using this

(4)

information, I was able to predict capillary loss over time that closely matched experimental estimates. From a mechanistic perspective, endothelium specific genetic knockdown or pharmacological inhibition of VEGF-R2 signaling was a critical factor in promoting capillary re-canalization and mitigating subsequent pruning. Thus, this work reveals the incidence, mechanism and long-term outcome of capillary obstructions and contributes to our understanding of age related capillary rarefaction.

In the second aim, I examined the vascular adaptions that accompany large scale disruptions to the cerebral blood flow in the form of ischemic stroke. Using a mouse model of type 1 diabetes, I revealed that ischemic stroke leads to an abnormal and persistent increase in Vascular Endothelial Growth Factor Receptor 2 (VEGF-R2) expression in peri-infarct vascular networks. Correlating with this, BBB permeability was markedly increased in diabetic mice which could not be prevented with insulin treatment after stroke. Imaging of capillary ultrastructure revealed that BBB permeability was associated with an increase in endothelial transcytosis rather than a loss of tight junctions. Pharmacological inhibition or endothelial-specific knockdown of VEGF-R2 after stroke attenuated BBB permeability, loss of synaptic structure in peri-infarct regions, and improved recovery of forepaw function. However, the beneficial effects of VEGF-R2 inhibition on stroke recovery were restricted to diabetic mice and appeared to worsen BBB permeability in non-diabetic mice. These results showed that aberrant VEGF signaling and BBB dysfunction after stroke plays a crucial role in limiting functional recovery in an experimental model of diabetes.

Overall this dissertation demonstrates that the structure, integrity, and function of mature cerebrovascular networks undergoes substantial changes in the face of both small

(5)

and large scale vascular insults. Furthermore, I have revealed a critical role for endothelial VEGF-R2 signalling in mediating many of these vascular changes, which in turn, had important consequences for brain aging and the recovery of function after stroke.

(6)

List of Abbreviations

α-SMA – Alpha Smooth Muscle Actin ACA – Anterior Cerebral Artery

AGE – Advance Glycation End products BBB – Blood-Brain Barrier

CBF – Cerebral Blood Flow ECM – Extra-Cellular Matrix GFP – Green Fluorescent Protein MCA – Middle Cerebral Artery

MCAO – Middle cerebral Artery Occlusion NO – Nitric Oxide

NVC – Neuro-vascular Coupling NVU – Neuro-vascular Unit PCA – Posterior Cerebral Artery ROS – Reactive Oxygen Species T1D – Type 1 Diabetes

T2D – Type 2 Diabetes

TJC – Tight Junctional Complex TPA – Tissue Plasminogen Activator WSS – Wall Shear Stress

VCI – Vascular Cognitive Impairment VEGF – Vascular Endothelial Growth Factor VEGF-R1 – VEGF Receptor 1

VEGF-R2 – VEGF Receptor 2 VSD – Voltage Sensitive Dye

(10)

List of Figures

Figure 1. Blood supply to the brain. ... 8

Figure 2. Major cerebral artery territories and functional lobes ... 9

Figure 3. Structure of the cortical vasculature. ... 10

Figure 4. The Blood Brain Barrier. ... 13

Figure 5. Vascular Endothelial Growth Factor Receptor 2 (VEGF-R2) Signalling. ... 20

Figure 6. Mouse photothrombotic model of ischemic stroke. ... 34

Figure 7 Cortical capillaries are prone to spontaneous obstruction. ... 77

Figure 8 Work flow and validation of automated estimates of vessel density. ... 81

Figure 9 Fluorescent microspheres as a model of spontaneous naturally occurring capillary obstructions. ... 83

Figure 10 . Microsphere based obstruction and pruning did not induce a microglial response or cell death. ... 85

Figure 11 Microsphere obstructions are distributed across major cerebral vascular territories. ... 86

Figure 12 Fates of obstructed cortical capillaries. ... 90

Figure 13 Capillary pruning does not alter adjoining capillary position. ... 91

Figure 14 Additional examples of capillary recanalization and pruning. ... 93

Figure 15 Recanalization correlates with obstruction location but not Local blood flow. 95 Figure 16 Blood flow in recanalized capillaries does not predict later pruning. ... 96

Figure 17 Obstructed capillaries that recanalized had a higher risk for subsequent obstruction... 97

Figure 18 Capillary pruning leads to altered blood flow in adjacent connected capillaries. ... 99

Figure 19 Lower capillary density in aged mice is predicted by obstruction and pruning rates. ... 103

Figure 20 Modelling capillary loss over time. ... 105

Figure 21 VEGF-R2 signaling dictates capillary recanalization. ... 109

Figure 22 Vascular specific knockdown of VEGF-R2 does not affect cardiovascular health or blood flow. ... 113

Figure 23 Inhibition of VEGF-R2 signaling with SU5416 did not affect cardiovascular health or blood flow. ... 115

Figure 24 Experimental outline. ... 150

Figure 25 Diabetes amplifies the upregulation of VEGF-R2 after stroke. ... 152

Figure 26 Diabetes exacerbates the loss of BBB integrity 3 days after stroke. ... 156

Figure 27 Excessive BBB permeability in diabetic mice is not related to peri-infarct angiogenesis. ... 158

Figure 28 Loss of BBB integrity at 3 days is due primarily to increased transcytosis and not tight junctional complex disassembly. ... 162

Figure 29 Inhibiting VEGF-R2 signalling attenuates stroke induced BBB permeability in diabetic mice. ... 166

Figure 30 SU5416 treatment does not affect infarct volume. ... 167

Figure 31 Inhibiting VEGF-R2 signalling mitigates excessive spine loss in peri-infarct cortex in diabetic mice. ... 171

(11)

Figure 32 Inhibiting VEGF-R2 signalling improves functional recovery in diabetic mice. ... 175 Figure 33 Summary of obstructed cortical capillary fates and the effects of VEGR-R2 inhibition. ... 211 Figure 34 Summary of changes to peri-infarct blood vessels and dendritic structure after stroke. ... 216

(12)

Acknowledgments

I would like to acknowledge, as an uninvited settler to these lands, the Songhees, Esquimalt and WSÁNEĆ peoples. It was upon their lands this research was

performed and this dissertation was written.

First and foremost, I would like to acknowledge my twin sister Kristina, who’s love, kindness and support sustained me. And her husband Mark, the hardest working and most honorable man I know. You have built a warm, loving, and endlessly entertaining family. To have been welcomed into this haven has meant more to me than I will ever be able to express.

To Emma, Lizzie, and Teddy, you have kept me young and mindful of the astonishing joy in curiosity and the power of imagination. More importantly your love has given me strength and hope when I have needed it. Science is meaningless unless done in the service of others. I was lucky enough to love and be loved in return by three of the most extraordinary human beings to ever be. If I can spend my life making your world just a little better, if I can inspire you, help you find your passion in life, it will all have been worth it.

To my brothers Michael and Andrew, your friendship, love, and mentorship (as only older brothers can) has been a profound force in shaping my life. You have shown me different perspectives and exposed me to vast worlds of art, literature, philosophy, music, and politics. I officially forgive you for tying me up and pouring hot sauce down my throat until I vomited.

To my Mum and Dad, without your support and love this PhD would not have been possible. Our family life was not always perfect, but for better or worse I am the person I am because of your love and guidance, and I am grateful.

To my friends, if I only I could reach back in time to the shy little boy of my youth, who would eat his lunch hiding in stairwells alone, and tell him what stupendous people I would one day be honored to call my friends. To Amanda, Lena, Alex, Abdul, Scott, Ben, Aaron, Christine, and Essie. There is so much you have done for me, so many little moments of kindness, compassion, support, and friendship, in it’s totality, it is truly staggering. If I were to be judged by the company I keep, my value would be grossly overestimated. And a special gratitude to Amanda and Lena, who have always done a better job at loving me than I have myself, and it has made all the difference.

(13)

I have had the privileged of being mentored by a few extraordinary scientists. To my supervisor Craig who gave me ever opportunity to try, to succeed, and to fail. You started my career in Neuroscience, gave me the tools and the knowledge to start down this extraordinary path, and the trust and guidance to follow my own directions. For that I am eternally in your debt. To my committee members Bob Chow and Pat Nahirney, I thank you for the years of guidance. To my external Grant Gordon, for many pieces of great advice over the years. To Kerry Delaney, your mentorship and support has always been an honor. You seem to think I’m not entirely an idiot, I find this very encouraging. To my fellow members of the Brown Lab, Andrew, Angela, Akram, Kim, Essie, Emily, Mohammed, and Ben, you made a few benches in MSB a joyous place to work. And of special mention, to Abdul, your friendship and support, your intellect, humor, creativity, and passion is amazing, your inquisitive and extraordinary mind is an example I aspire to. And to the volunteers who labor I shamelessly exploited and took credit for, thank you Natalie, Jessie, Charmaine, Kevin, and Patrick. To my first mentor in science, who will likely never read this, but to Dr. Benjamin Silverberg, who in a high school biology class challenged me to be exceptional. Your belief that I was capable of something more than I aspired to, that the lofty goal of discovering something new about the world around me was within my grasp, allowed me to stumble upon the extraordinary joy of science and started me towards my true passion in life.

To the many support staff at the University of Victoria, who’s contribution to our success is criminally under appreciated. You are the bedrock of this institution. I am deeply grateful for the help of our DMS graduate secretaries Karen and fellow Dr. Who aficionado Erin, Lab Mangers Evelyn, and Sara, and especially to Facilities Manager Robin, who’s opera recommendations were spot on, and who’s many engrossing early morning political debates I have greatly enjoyed.

And lastly to the many dogs I have known and loved, this life would suck hard without you, to Charlie, Scooter, Duffy, Maggie and Sita.

(14)

Dedication

To Emma, Lizzie, and Teddy.

I will always be more honored to have been called Pat-Pat than

any other title I may ever earn.

(15)

Chapter 1 1.0 Introduction

All complex systems depend on the infrastructure that feeds in raw materials and removes refuse. The brain is no exception, it’s function relies on an encompassing transportation network that delivers a constant influx of oxygen and nutrients, while removing metabolic waste. The human brain contains ~650 km of blood vessels, the majority of which are capillaries (Cipolla 2009; Schmid, Barrett, et al. 2017). However, to date the focus on the cerebral vasculature has been somewhat myopically centered on large vessels, such as the major cerebral arteries. For example one of the most widely used textbook on neuroscience barely mentions capillaries (Kandel 2013) (the vasculature itself is relegated to an appendix). However, almost all the exchange of gases and macromolecules occurs at the level of capillaries (Cipolla 2009). A more complete understanding of this relationship, between the vascular and neuronal systems, requires a renewed focus on the principle functional unit of the cerebral vasculature, the capillary.

Plasticity is the ability for networks to change either structurally or functionally over time. In the brain it has traditionally been considered in the context of neural circuits, however most biological systems exhibit some level of plasticity in response to environmental cues. While larger vessels are relatively fixed (structurally) throughout life, capillaries display remarkable plasticity in development, and possibly into adulthood (Harb et al. 2013; Chen et al. 2012; Whiteus, Freitas, and Grutzendler 2014). Thus, capillaries are both the main site of exchange, and the primary substrate for vascular plasticity in the brain. How cerebral vascular networks evolve over time is likely

(16)

dominated by changes at the level of capillaries. The relationship between the structure and function of neural circuits and the vasculature is reciprocal, neural activity shapes the development of vascular networks (Whiteus, Freitas, and Grutzendler 2014) and locally influences vascular function by Neural Vascular Coupling (NVC) (Attwell et al. 2010; Roy and Sherrington 1890). Conversely, dysfunction of the vascular system quickly impacts neural structure and function (Iadecola 2004; Iadecola 2013; Brown, Wong, and Murphy 2008). A critical question is therefore to what extent can the cerebral vasculature respond and adapt to challenges. In otherwise healthy animals, cerebral capillaries are lost with aging (Brown and Thore 2011; Riddle, Sonntag, and Lichtenwalner 2003), yet mechanistically why this should be the case is not clear. Given that anatomically age-related changes to neuronal or synaptic density does not strongly correlate with cognitive decline, while cerebral capillary rarefaction does, it has been speculated that capillary loss may be a key driver of age related declines in brain function and some forms of vascular dementia (Riddle, Sonntag, and Lichtenwalner 2003; Brown and Thore 2011; Iadecola 2013). Likewise, after ischemic stroke the vascular system plays a role in supporting adaptive neural plasticity to partially restore lost brain function (Brown et al. 2007; Murphy and Corbett 2009). A compromised vascular system could however impair recovery, particularly in diseased states that injure the vasculature, such as diabetes.

Plasticity occurs in response to a changing environment, in the adult cortex this could be changes to blood flow and/or endothelial cell health. In the cerebral vasculature, the microcirculation is uniquely prone to spontaneous stalls and obstructions (Santisakultarm et al. 2012; Kleinfeld et al. 1998; Villringer et al. 1994; Erdener et al. 2017;

(17)

Santisakultarm et al. 2014). Spontaneous obstructions, while unpredictable and often transient, happen with enough frequency to be observed in the rodent cortex (Erdener et al. 2017; Kleinfeld et al. 1998; Santisakultarm et al. 2014; Villringer et al. 1994). Some obstructions persist, halting blood flow for hours, and potentially longer (Santisakultarm et al. 2014; Erdener et al. 2017). Despite the potential impact on vascular function, the long-term fates of these obstructions are unknown (Erdener et al. 2017; Santisakultarm et al. 2014). How the microvasculature adapts to these obstructions could over time reshape vascular trees, with functional consequences. For example, if the microcirculation maintains a developmental like capacity for angiogenesis, these obstructions would be benign. Likewise, if all capillaries have similar mechanisms of recanalizing, such as angiophagy seen in large capillaries and penetrating arterioles (Lam et al. 2010), then obstructions may only briefly interrupt blood flow. However, if some capillaries are unable to recanalize, then the accumulation of spontaneous obstructions over years could lead to changes in cerebral perfusion and cognitive performance.

In the disease afflicted brain, the challenges facing the vasculature are even greater. For example, vascular endothelial cells in diabetics are bathed in hyperglycemic blood. These cells, due to the nature of their glucose transporters, are unable to regulate the influx of sugar into the cell (Lu et al. 2013; Benarroch 2014). The constant influx of glucose drives endothelial cells into metabolic overdrive, altering cell biochemistry and signalling pathways (Brownlee 2001; Tomlinson 1999; Spitaler and Graier 2002; Pieper 1997; Pieper and Peltier 1995). The end result of glucose overload is the compromise of endothelial function. The diabetic brain is also highly prone to obstructions, leading to ischemic strokes at much higher rates than euglycemic counterparts (Baird et al. 2002).

(18)

Following stroke, the brain undergoes adaptive rewiring in the surviving adjacent brain tissue, partially restoring lost brain functions (Murphy and Corbett 2009; Brown et al. 2009; Brown et al. 2007). This process is however dependent on the surviving vascular network in the same tissue. Diabetes significantly worsens prognosis for functional recovery following stroke (Baird et al. 2002; Megherbi et al. 2003; Wei, Heeley, Wang, Huang, Wong, Li, Heritier, Arima, and Anderson 2010). Impaired functional recovery after stroke has also been shown in rodent models of type 1 diabetes, which correlated with a failure for surviving peri-infarct (adjacent to infarct) circuits to rewire (Sweetnam et al. 2012). The underlying mechanism behind this diminished recovery is however unknown, but likely involves the intersection of an impaired vasculature with rewiring cortical circuits.

The molecular signalling pathways that regulate the various manifestations of microvascular plasticity remain incompletely mapped, yet if elucidated, could offer a promising substrate for improving cerebral vascular health. Vascular Endothelial Growth Factor (VEGF) signalling through its canonical receptor, VEGF-R2 (Ferrara and Henzel 1989; Ferrara, Gerber, and LeCouter 2003; Olsson et al. 2006), is a probable starting point. VEGF-R2 signalling is a master regulator of endothelial function and feeds into most major signalling pathways (Olsson et al. 2006). VEGF-R2 plays a key role in sensing shear stress (Tzima et al. 2005) and is upregulated after stroke (Hermann and Zechariah 2009) , so it is ideally situated to initiate vascular changes following a spontaneous obstruction or after ischemic stroke.

Given the vital role the microvasculature plays in maintaining brain health and supporting recovery following injury, the following thesis will examine the intersection

(19)

of the vasculature and the brain in 2 aims. First, I will investigate microvascular obstructions, their prevalence in mouse cortex, microvascular recanalization strategies, and structural plasticity following these sub-ischemic obstructions. The first aim will also look at what effects microvascular obstructions have on the mature vascular network over years, and what endothelial signaling pathways influence can be exploited to improve recanalization. The second aim will explore how the microcirculation in the diabetic brain reacts to focal ischemic stroke in the somatosensory cortex, and how an impaired diabetic vasculature affects circuit remapping and recovery of lost brain functions. Both aims will focus on the role of VEGF-R2 in regulating endothelial cell function and response to small or large vascular insults. VEGF-R2 is well established as a master regulator of most endothelial signaling pathways, and a strong component of the vascular response to large ischemic events. However, how VEGF-R2 signaling relates to sub-ischemic events, or within the diabetic brain will be addressed by both aims.

1.1 On the paradox of cerebral blood flow

The brain consumes vastly more energy than any other organ, ~20% of resting oxygen consumption (despite being ~2% of body weight), at a rate of around 75mL / 100g / min (Cipolla 2009). The metabolic demands of brain tissue fluctuate spatially and temporally with neural activity, leading to corresponding increases in blood flow (Roy and Sherrington 1890; Attwell et al. 2010). Neuronal signalling is energetically expensive, an increase in activity of one action potential / cortical neuron raises oxygen consumption by 3% of resting rate (Attwell and Laughlin 2001). Despite this range of energetic requirements, and the fine spatial and temporal requirements of neural signalling (and thus fluctuating blood flow), neurons paradoxically lack any significant

(20)

energy stores and the cerebral vasculature is fully perfused, without any reserve capacity beyond vessel dilation (Iadecola 2013). While a system of redundancies such as collaterals and pial surface vascular loops exist to shield neurons against large changes in flow, there are no safeguards once blood flow enters the cortex (Shih et al. 2013). This is the fundamental paradox of the cerebral blood supply, the tissue with one of the greatest ranges of energy consumption, and arguably the highest sensitivity to local hemodynamic changes (to maintain fidelity of neural signaling) lacks nearly any safety net beyond the pial vasculature to interruptions of blood flow. Why this paradox exist is a mystery, but given it is ubiquitous across mammalian brains, it likely reflects a fundamental compromise between the different demands and limitations of the nervous and vascular systems. This susceptibility also reflects the evolutionary forces that shaped the brain (such as being contained in the skull), which in turn shaped the structure and function of the vasculature. This is even more remarkable given that brain function is nearly synonymous with life, thus maintaining proper blood flow to the brain is literally a matter of survival.

(21)

1.2 The anatomy of the cerebral blood supply

Cerebral blood flow starts with oxygenated blood returning from the lungs to the left atrium and then to the left ventricle of the heart. Oxygenated blood leaves the ascending aorta and travels up the two major pairs of arteries supplying the brain, the Internal carotid arteries (supply the anterior brain) and vertebral arteries (supplying the brainstem and posterior brain) (Kandel 2013; Cipolla 2009) (Figure 1). The carotid arteries supply approximately 80% of the blood supply to the brain. The first major safeguard in maintaining perfusion of the brain is the Circle of Willis, the network of bilateral communicating arteries that connect the major cerebral arteries fed by either the Internal carotid arteries or vertebral arteries (Figure 1) (Cipolla 2009). Thus, when one of the main feeding arteries is interrupted, the Circle of Willis provides collateral flow between the anterior and the posterior circulation along the floor of the cerebral vault, providing blood to tissue that would otherwise become ischemic (Figure 1). The Internal carotid arteries supply the Anterior cerebral circulation. These large arteries are the medial branches of the common carotid arteries in the neck which enter the skull (Cipolla 2009). The internal carotid artery branches into 2 of the 3 major cerebral arteries, the anterior cerebral artery (ACA) and continues to form the middle cerebral artery (MCA) (Kandel 2013) (Figure 1). The Vertebral arteries are smaller arteries that branch from the subclavian arteries. Within the cranium the two vertebral arteries fuse into the basilar artery and supply blood to structures in the brainstem and cerebellum through pontine and cerebellar arteries (Cipolla 2009). The basilar artery then bifurcates into the Posterior Cerebral Artery (PCA) (Cipolla 2009) (Figure 1). The blood flow to cortical regions is through the three major cerebral arteries, ACA, MCA, PCA (Figure 2) (Kandel 2013).

(22)

Cortical blood flow starts with the surface network of pial vessels that form an interconnected honeycomb like mesh of vessels running parallel to the cortical surface (Blinder et al. 2013; Shih et al. 2015) (Figure 3). From surface

Figure 1. Blood supply to the brain.

Diagram showing the major routes of blood from the heart into brain as well as the Circle of Willis, the network of collateral connections between major cerebral arteries to provide rerouted perfusion in the case of a major obstruction

(23)

pial arteries, branches perpendicular to the surface, penetrating arterioles, travel straight down through most layers of the cortex (Shih et al. 2015; Blinder et al. 2013). Capillaries (3-8 μm in diameter) branch off from penetrating arterioles, and branch many times before reaching the venous side, often with interconnected loops forming a mesh like lattice (Figure 3) (Blinder et al. 2013; Scallan, Huxley, and Korthuis 2010). The capillary bed feeds into penetrating venules which mirror penetrating arterioles and direct deoxygenated blood out of the cortex towards the pial network of interconnected surface venules. The venous drainage of the cerebrum can be separated into two subdivisions, superficial and deep. The superficial system is made up of the dural venous sinuses, whose walls are formed from the dura mater as opposed to veins (Cipolla 2009). The major dural sinus is the superior sagittal sinus, which flows along the surface of the

Figure 2. Major cerebral artery territories and functional lobes

The three major cerebral arteries supply blood to distinct regions of the cortex. Distinct vascular territories mean occlusions of a major cerebral artery leads to distinct functional deficits, as beyond the Circle of Willis there is minimal collateral connections between the 3 major cerebral arteries.

(24)

cortex, between the hemispheres, anterior to posterior. At the posterior base of the cerebrum the superior sagittal sinus bifurcates into two transverse sinuses which travel laterally and go on to form the two jugular veins (Cipolla 2009). The deep venous drainage comes from traditional veins inside deep structures of the brain, which join behind the midbrain to form the vein of Galen.

1.3 The structure of cerebral blood vessels

One of the most important concepts in cerebrovascular research to emerge in the last 20 years is that of the Neuro-Vascular Unit (NVU), the idea that neurons and blood vessels act as an integrated functional unit to regulate cerebral blood flow (CBF) (Iadecola 2013). The NVU is comprised of the endothelial cells that form the luminal lining of the blood vessel, contractile cells around the endothelial cells, smooth muscle

Figure 3. Structure of the cortical vasculature.

Branching off the major cerebral arteries, the surface of the cortex is covered in a honeycomb like mesh of pial arteries, which are highly interconnected allowing robust redistribution of flow if obstructed. Surface arteries give rise to penetrating arterials the branch perpendicular to the pial surface and dive down most layers of the cortex. Off penetrating arterials is the highly interconnected network of capillaries. Scalebars (x,y,z) 50 μm.

(25)

cells (SMC) or pericytes, and astrocytes that envelope blood vessels with end feet and signal increases in neural activity (and thus energetic demands), and along with neurons to the NVU (Iadecola 2017; Gordon et al. 2008; Gordon, Mulligan, and MacVicar 2007). As penetrating arterioles enter the cortex the Virchow-Robin space surrounds them, this then gives way to an encompassing sheath of astrocyte end foot process, the glial limitans (Iadecola 2013). Penetrating arterioles are wrapped in α smooth muscle actin (α-SMA) positive smooth muscle cells which give way to (α-SMA) positive ensheathing pericytes in the first 1-4 branches off the penetrating arteriole (Grant et al. 2017). Higher order (more branches away from the arteriole) capillaries are more sparsely covered by α-SMA negative pericytes (Grant et al. 2017; Hill et al. 2015). The exact relationship between the different types of mural cells and how they relate to active dilation or constriction of blood vessels, let alone the proper nomenclature, is still hotly debated (Attwell et al. 2016). While some studies have suggested that capillaries initiate vessel dilation during neurovascular coupling (NVC) (Hall, Reynell, Gesslein, Hamilton, Mishra, Sutherland, O/'Farrell, et al. 2014; Mishra et al. 2016), others have suggested this only occurs in α-SMA positive mural cells (defined as pericytes by some but not all) (Hill et al. 2015). Endothelial cells themselves play a critical role in conducting NVC responses across the vascular tree (Longden et al. 2017; Chen et al. 2014). Endothelial cells are electrically coupled through gap junctions, as well as to mural cells like pericytes (Cuevas et al. 1984; Komarova et al. 2017; Iadecola 2013), and conduct calcium waves likely vital to spatial and temporal propagation of NVC (Longden, Hill-Eubanks, and Nelson 2015).

(26)

1.4 The Blood-Brain Barrier

Perhaps the most critical structural feature of the NVU is the Blood-Brain Barrier (BBB) (Figure 4). The BBB is the property of cerebral blood vessels to restrict or filter the passive movement of most molecules in the blood from entering the brain (Figure

(27)

Figure 4. The Blood Brain Barrier.

The blood brain barrier (BBB) separates the brain from the global circulation. A) The BBB was first identified as the propensity for vascular dyes to be excluded from entering the brain unlike other organs, as shown with Evans blue dye. Injected i.v. through the tail vein and circulating for 30 min, after removing the blood through cardiac perfusion, the brain is devoid of any Evans blue staining while non-barrier organs such as skin and liver are stained by Evans blue (that permeated from blood into the tissue). B) There are two major routes through the BBB. Paracellular, through the clefts between endothelial cells or Transcellular, through vesicles moving through the endothelial cells themselves. C) The cells of the Neurovascular Unit (NVU), comprised of endothelial cells, mural cells, and astrocyte end feet. Endothelial cells are tightly bound by Tight Junctional Complexes (TJCs) and surrounded by the extracellular matrix. And Neurons (not shown). D) Electron micrograph of a cortical capillary. The endothelium is noticeably sparsely pockmarked with vesicles and the insert shows the connection of two leaflets of endothelial cells, bound by TJCs, creating a strong barrier to paracellular movement. E) Diagram of TJC structure. Right panel shows how TJCs are arranged along the sides of the two endothelial cell membranes. Left panel shows the major proteins that comprise each TJC (Claudins, Occludins and JAMs) and accessory proteins that link TJC to the cytoskeleton and regulate TJC assembly (Zona Occludens).

(28)

4A-B). The BBB regulates the delicate balance of ions, proteins and sugars required for

proper brain function (e.g. Insulin, glucose, albumin etc.) and prevents potentially harmful constituents of the blood, such as high concentrations of glutamate, from entering the brain (Abbott et al. 2010; Ballabh, Braun, and Nedergaard 2004; Dyrna et al. 2013; Hawkins and Davis 2005). The BBB is however not an intrinsic property of brain endothelial cells but rather induced and maintained by cells of the NVU (Figure 4C) (Armulik et al. 2010; Daneman et al. 2010). Functionally the BBB is not one single property but several factors, including tight junctions between endothelial cells, the spectrum and polarity of endothelial cell transporters and channels expressed, and the suppression of transcytosis (Obermeier, Daneman, and Ransohoff 2013; Andreone et al. 2017; Brightman and Reese 1969; Reese and Karnovsky 1967). The importance of the BBB for brain health can easily be seen by the extent of damage to neural circuits that occurs when the BBB fails. BBB disruption correlates with inflammation, cell death, loss of synaptic structure and disruption to normal brain function (Chen et al. 2009; Tomkins et al. 2007; Shinnou et al. 1998; Hawkins and Davis 2005; Zlokovic 2008). Not surprisingly BBB disruption is associated with many pathological brain states, notably Alzheimer’s and dementia (Attems and Jellinger 2014; Bell and Zlokovic 2009; Claudio 1996; Davies and Hardy 1988; Masters and Beyreuther 1988; Taguchi 2009; Iadecola 2004; Iadecola 2013). BBB disruption is also a significant driver of secondary injury following ischemic stroke, which is reviewed in section 1.11.

1.4.1 Endothelial Tight Junctions

The first barrier is the Tight Junctional Complexes (TJC), the most apical structure, that connect endothelial cells together forming seals as close as 4 Å (Figure 4C-D)

(29)

(Anderson and Van Itallie 2009). Endothelial TJCs are considered the core structure of the BBB, forming a physical and charge selective barrier to paracellular diffusion (Figure 4B) (Bauer et al. 2014). TJC are multi protein complexes that also act as signaling complexes, scaffolding to cluster membrane components and as anchors to the cytoskeleton (Figure 4E) (Dejana 2004; Bauer et al. 2014; Cipolla 2009). TJC are dynamic, capable of rapidly assembly and disassembly, and in turn regulate many cell functions including polarity, proliferation, and gene expression (Bauer et al. 2014; González-Mariscal, Tapia, and Chamorro 2008; Dejana 2004). The main backbone of TJCs are three integral membrane proteins, claudin, occludin and junction adhesion molecule (JAM) (Figure 4E) (Cipolla 2009; Luissint et al. 2012). Claudins are 22 kDa phosphoproteins that bind other endothelial claudins on adjacent cells to form the primary seal of TJC (Luissint et al. 2012; Cipolla 2009; Furuse, Sasaki, and Tsukita 1999). Occludins are 65 kDa proteins with 4 trans membrane domains that bind adjacent occludins and / or claudins (Cipolla 2009; Bauer et al. 2014). Lastly JAMs, a 40 kDa membrane protein, bind adjacent JAMs to form the last component of the TJC seal (Bauer et al. 2014). All three transmembrane proteins have cytoplasmic ends that associate with accessory proteins, the Zona Occludens (ZO), which link the TJC to the actin cytoskeleton (Figure 4E) (Bauer et al. 2014; Luissint et al. 2012). ZOs connect TJC with the cytoskeleton and are regulated by their phosphorylation states (Abbott et al. 2010). Therefore, ZOs provide both structural stability and a substrate for the regulation of TJC permeability through their association with other TJC proteins.

(30)

1.4.2 BBB endothelial transporters

Another feature of the BBB is the selective expression of luminal and abluminal transporter in barrier endothelial cells. The BBB freely passes oxygen, carbon dioxide and small lipophilic molecules, however everything else must be transported in or out of the brain (Cipolla 2009; Abbott et al. 2010). Vital hydrophilic molecules are passively moved down concentration gradients through carrier-mediated transport (facilitated diffusion), such as glucose, lactose, amino acids, and vitamins (Zlokovic 2008; Masaki 2007). Other substances, either larger, or with weaker concentration gradients (insufficient to drive diffusion), or strictly regulated signalling molecules, are moved across the BBB through receptor mediated transport (Cipolla 2009; Zlokovic 2008; Abbott et al. 2010). This includes proteins, chemokines, cytokines, growth factors and insulin to name a few. Lastly, as the BBB helps maintain the delicate microenvironment of the brain, moving things out of the brain is equally important. Efflux out of the brain is primarily done by an ATP dependent class of transports, the ATP-Binding Cassette (ABC) transporter super family (Zlokovic 2008; Cipolla 2009; Masaki 2007). ABC transporters facilitate the removal of potentially toxic materials out of the brain, such as metabolites and certain chemical compounds.

1.4.3 BBB endothelial transcytosis

There are 2 major routes across a vascular wall, paracellular movement (between endothelial cells) or transcellular / transcytotic movement (through endothelial cells) (Figure 4B). A down regulation of endothelial cell transcytosis is an important pillar of the BBB (Abbott et al. 2010; Andreone et al. 2017; Villegas and Broadwell 1993). Under normal conditions endothelial cells in the BBB have low levels of vesicular transport

(31)

across the endothelium, as well as a near complete absence of larger pore like structures of vesicles such as vesiculo-vacuolar organelles (VVOs) and trans-endothelial pores (Figure 4D) (Olsson et al. 2006). Endothelial cells of the BBB have relatively low numbers of pinocytotic vesicles, clathrin-coated vesicles and caveolae compared to peripheral vasculature (Elizabeth 2012). However, some transport does occur through these vesicles, the most numerous being clathrin-coated vesicles and, caveolae (formed from lipid raft domains) as well as larger fluid engulfing macropinocytotic vesicles (Elizabeth 2012; Abbott et al. 2010; Villegas and Broadwell 1993). Transcytosis can also be divided between non-specific bulk fluid transport or Receptor Mediated Transcytosis, in which vesicular associated receptors give ligand specificity for molecules such as insulin, transferrin, and lipoproteins (Elizabeth 2012; Villegas and Broadwell 1993). Plasmalemmal vesicles can be identified in endothelial ultrastructure as ~50 nm omega shaped invagination on the cell membrane, the majority of which are open to the extracellular environment (Predescu, Predescu, and Malik 2007; Komarova et al. 2017). Plasmalemmal vesicles are particularly sensitive to the local hemodynamic environment through caveolae, and plasmalemmal vesicles are often colocalized with the insulin receptor, suggesting they could be involved in the pathogenesis of diabetes (Predescu, Predescu, and Malik 2007).

1.4.4 Formation and regulation of the BBB

The BBB is a product of many cells creating a unique environment. Brain endothelial cells quickly lose BBB properties when transplanted outside the brain, while endothelial cells transplanted into the brain develop BBB properties (Abbott et al. 2010). However, determining the exact timing of the development of the BBB has been

(32)

challenging (Blanchette and Daneman 2015; Obermeier, Daneman, and Ransohoff 2013). In the mammalian brain angiogenesis precedes barrier development by a few days embryonically, however some BBB features are present in invading endothelial cells (Obermeier, Daneman, and Ransohoff 2013). The BBB then matures after angiogenesis occurs in the brain, with a down regulation of fenestrations and transcytosis in endothelial cells being one of the last features of the BBB to develop (Obermeier, Daneman, and Ransohoff 2013; Andreone et al. 2017). While the initial angiogenic sprouting of endothelial cells to form nascent blood vessels is highly dependent on classical angiogenic signals, such as VEGF and Wnt (Obermeier, Daneman, and Ransohoff 2013; Felmeden, Blann, and Lip 2003; Carmeliet and Jain 2011), the maturation of the BBB is a multistep and multi cellular process. The association of both astrocytes and pericytes with newly formed vessels is required for full BBB development (Daneman et al. 2010; Obermeier, Daneman, and Ransohoff 2013). Once formed, the BBB requires continued association with, and signalling from, associated cells to be maintained, and can be rapidly opened, particularly in cases of injury, inflammation or disease (Obermeier, Daneman, and Ransohoff 2013). A major regulator of BBB permeability in the brain is VEGF signaling through VEGF-R2 (Figure 5), and depending on the circumstance, VEGF can affect paracellular or transcellular permeability (Cipolla 2009; Olsson et al. 2006; Fischer et al. 2002). VEGF signaling can promote paracellular permeability by promoting cytoskeleton stress fiber contraction which decrease cell-cell contact and increase intercellular junctional spaces (Lum and Malik 1994). In endothelial cells, VEGF can promote the disassembly of TJC through altering the phosphorylation state of ZO-1 (Weis and Cheresh 2005). However, most

(33)

research indicates high VEGF signalling increases transcytosis, fenestrations and VVOs in endothelial cells (Feng et al. 1999; Kamba et al. 2006; Liu et al. 1999; Weis and Cheresh 2005; Esser et al. 1998; Bates and Harper 2002; Roberts and Palade 1995). For example VEGF application to isolated blood vessels increased permeability to albumin 3 to 4 fold, which suggested a 2.5 fold increase in transcellular transport (Wu et al. 1996). While VEGF signalling in vitro has been shown to interact with almost all known endothelial cell signaling pathways (Figure 5), PLC, PLA, AKT, PI3K, MEK, FAK, Calcium influxes, and eNOS have all been implicated in increasing transcytosis (Bates and Harper 2002).

(34)

The interplay between these distinct mechanisms of permeability and the web of

downstream signaling from VEGF receptors is not clear, nor is it clear if and when each route dominates in any given instance of pathological BBB disruption.

1.5 The nature of blood flow in cortical capillaries

Capillaries are inherently narrow high resistance tubes that must pass large and adherent cells and plasma constituents. Blood flow through the cerebral vasculature can be approximated by Ohm’s law in that flow is proportional to the difference in inflow and outflow pressure (ΔP) divided by the resistance (Cipolla 2009), and by Poiseuille’s law that flow is proportional to ΔP, and thus vessel length (L), blood viscosity (μ), volumetric flow rate (Q) and vessel radius (r) to the fourth power (ΔP = (8μLQ)/πr4 ) (Sumpio 1993).Flow in the cerebral capillaries is generally laminar with a blunted parabolic cross section profile and also free of inertia effects (Hirsch et al. 2012). Blood is not a pure liquid, but a suspension of cells and flow depends on both vascular geometry and serum constituents. Thus the distribution of RBC along different capillaries define each segments’ contribution to total vascular resistance (Hirsch et al. 2012). Several phenomena of flow, including plasma skimming, RBC migration to the luminal center,

Figure 5. Vascular Endothelial Growth Factor Receptor 2 (VEGF-R2)

Signalling.

VEGF-R2 is the major receptor for VEGF-A and a member of the Receptor Tyrosine Kinase (RTK) superfamily. Like most RTK binding of the ligand to the extracellular immunoglobulin like domains, or physical force, causes the receptor to dimerize where each half phosphorylates the other in the Kinase Regulatory Domain, which leads to phosphorylation of tyrosine residues downstream, notably Y-1175. Phosphorylation of these c-terminal tyrosine sites allows secondary messenger proteins to interact with the receptor and activate signal cascades. VEGF-R2 intersects with most major signal transduction pathways, including MAPK, AKT, PKC and Src.

(35)

unequal distribution of RBCs at bifurcations, and vascular topology affect RBC distribution (Hirsch et al. 2012). Thus, the major determinants of blood flow are luminal diameter, pressure gradients and RBC distribution. While RBC velocities are relatively consistent across larger penetrating arterioles (~ 10 mm/ sec). Flow across microvessels have significantly greater variation, ranging from 0.01 to 10 mm/sec, across several orders of magnitude (Shih et al. 2013). This despite a much smaller range of diameters compared to arterioles (Shih et al. 2013; Blinder et al. 2013). Therefore, in the cerebral capillary bed, while changes in diameter are determinants to flow, pressure gradients and RBC density are likely equally important. The capillary bed is also the largest source of hemodynamic resistance in the brain, and capillaries experience the largest pressure differential across any vessel type in the cortex (Gould et al. 2016). The capillary network can be regarded as a 3 dimensional mesh with a width of approximately 50 to 75 μm wide (average capillary length) (Schmid, Barrett, et al. 2017). The architecture is both homogenous and highly interconnected, with substantial branching and interconnected loops, making any unifying topological order difficult to extract (Schmid, Barrett, et al. 2017; Blinder et al. 2013; Hirsch et al. 2012). However, just as vascular topology influences RBC distribution along different capillary paths, RBCs may play a role in molding the development of vascular topology. Computer modeling of simple capillary networks predict frequent branches of very slow or stalled RBCs (Obrist et al. 2010), which can significantly impair RBC oxygen delivery, suggesting developing capillary networks are structurally refined to increase efficiency (Hirsch et al. 2012).

Alongside the surprising heterogeneity in average RBC transit time (Jespersen and Østergaard 2012), an RBC moving through the cortical capillary bed, from an

(36)

penetrating arteriole to a penetrating venule, has a number of possible paths (on average 8) to choose from (Schmid, Barrett, et al. 2017). However, computer modelling suggest that for most capillary starting points (>50%) there is a preferred RBC path (of the ~8) that is chosen over 50% of the time (Schmid, Tsai, et al. 2017). Further, it has been suggested that the preferential path is most often “in-plane”, that the capillary bed is primarily designed to facilitate horizontal RBC movement (Schmid, Tsai, et al. 2017). The capillary network also displays remarkable input redundancy, with ~63% of all capillary paths being fed by more than 1 penetrating arteriole (Schmid, Barrett, et al. 2017; Guibert et al. 2012). Functionally, for permeable substances, diffusion across the BBB in capillaries is through the forces of hydrostatic pressure, concentration gradients, and osmotic pressures. It has been suggested that the unique capillary tree structure is to maximize exchange while allowing for a resting baseline that increases the dynamic range of flow rates during NVC (Schmid, Tsai, et al. 2017; Jespersen and Østergaard 2012). However, this structural / functional arrangement has potential drawbacks, namely a vulnerability to flow stalling, reversals, and obstructions. As well excessively high flow through one capillary branch can generate hemodynamic steal, leading to abnormally high RBC transit through one path at the expense of others, leading to overall lower tissue oxygenation (Jespersen and Østergaard 2012). Maximally efficient delivery of oxygen by RBCs likely requires for each capillary (based on length and morphology) a “goldilocks” range of RBC velocity, with either abnormally fast or slow RBC transit times reducing the efficiency of gas exchange (Schmid, Barrett, et al. 2017). It is speculated that aberrant capillary flow rates and reduced CBF may play significant roles in numerous pathological states such as ageing, diabetes, stroke, and Alzheimer’s disease

(37)

(Jespersen and Østergaard 2012; Girouard and Iadecola 2006; Iadecola 2004; Iadecola 2013).

1.6 The effects of shear stress on endothelial cell function

Like all cells, endothelial cells sense their external environment and are shaped by it. A major source of environmental signals to endothelial cells are the forces exerted by the flow of blood on the vessel wall, indicating a functional vascular segment connected to the global circulation. The primary force applied by flowing blood to the luminal endothelial surface is wall shear stress (WSS). WSS stress influences many aspects of endothelial cells, from polarity, cytoskeleton structure, intracellular signalling pathways, and gene expression (Wragg et al. 2014; Chiu and Chien 2011; Zhou, Li, and Chien 2014). The start of flow, and application of WSS to endothelial cells, is a critical developmental step in forming a mature blood vessel (Zhou, Li, and Chien 2014; Wragg et al. 2014). Much like the initial over production and subsequent pruning of synapses in development, initially the vascular system is overly redundant and contains capillary segments with little to no blood flow, or turbulent and highly variable flow (Obrist et al. 2010; Chen et al. 2012). In the developing zebrafish this initial excess of capillary segments is followed by the refinement of the network by an orderly pruning of capillary segments (Chen et al. 2012; Franco et al. 2015). Pruning is however not random, low flow segments are specifically removed (Chen et al. 2012). Vessel pruning is achieved by a coordinated migration of endothelial cells into adjacent vascular branches, not through cell death (Chen et al. 2012; Franco et al. 2015; Kochhan et al. 2013; Lenard et al. 2015). Segment pruning could also be induced in functioning (normal flow) segments by stopping blood flow through the injection of polystyrene microspheres (obstructing flow

(38)

in a single segment) (Chen et al. 2012). Thus, vessel pruning wasn’t likely due to any pre-existing state of the vascular segment, such as signalling gradients or ectopic factors, but was the effect of the loss of flow. Therefore, in development there exist a coordinated program for endothelial cells in which the loss of flow leads to the collapse of the vessel segment and endothelial cells migration into adjacent segments, presumably improving the overall efficiency of the vascular tree. In the adult vasculature WSS continues to provide signalling to endothelial cells, supporting a stable quiescent state (Zhou, Li, and Chien 2014). Turbulent flow, such as at bifurcations, lead to altered gene expression, abnormal endothelial cells physiology, and higher risk for plagues and disease (Chiu and Chien 2011). However, the effects of the complete loss of WSS in a mature capillary is essentially unknown. While in vitro models can provide some clues, they exist independent of the effects of the larger vascular system and associated NVU. Meanwhile

in vivo studies of the loss of flow has been restricted to that of many vessels

simultaneously, associated with much larger events such stroke, or when isolated to a single vessel, in much larger arterioles and capillaries (8-20 μm) (Shih et al. 2013; Lam et al. 2010; Grutzendler et al. 2014). In both cases it is difficult to separate the effects of the lose of flow and the accompanying hypoxia.

1.7 VEFG-R2 is a master regulator of endothelial cell function

As early as the 1800s many, including famed anatomist Rudolf Virchow, had observed the capacity for tumours to induce the growth of new blood vessels (Ferrara 2002). By the 1940s it was widely speculated that tumours secreted a soluble factor that initiated angiogenesis, and at the same time researchers proposed some soluble “factor x” was also responsible for pathological neovascularization of the retina in diabetic

(39)

retinopathy (Ferrara 2002). By 1971 Judah Folkman had made the now famous hypothesis that inhibiting angiogenesis could cure cancer, specifically by identifying the mysterious soluble protein he called Tumour Angiogenesis Factor. While cancer labs around the world hunted this angiogenic factor, in an unrelated lab Donald Senger in 1983 discovered a secreted protein that could induce hyperpermeability in blood vessels, which he named Vascular Permeability Factor (VPF) (Ferrara 2002). In 1989 Napoleone Ferrara discovered a powerful endothelial cell mitogen that he named Vascular Endothelial Growth Factor (VEGF), which would be found to be the same VPF discovered in 1983 (Ferrara 2002). Thus, the history of the discovery of VEGF highlights the duality of its nature, both a critical growth factor and a powerful inducer of vascular permeability (~50,000x more potent than histamine). Further sequencing of the VEGF gene would also reveal a surprising degree of homology across all mammals, cementing VEGF’s place as a “master” vascular endothelial cell regulator. The VEGF family is comprised of 5 secreted, dimeric glycoproteins, around 40 kDa (Olsson et al. 2006). VEGF-A is both the most abundant and functionally relevant to vascular endothelial cells and is by convention (and throughout here) simply referred to as VEGF. VEGF-A however has several isoforms, differing in length, specifically in the heparan binding domain, thus affecting solubility in the extracellular matrix (ECM) (Olsson et al. 2006; Ferrara, Gerber, and LeCouter 2003). Thus VEGF120 (mouse variant, human VEGFs are all 1 amino acid longer) has weak binding affinity to Heparan Sulfate Proteoglycans (HSPGs) and high solubility (Olsson et al. 2006). The VEGFs then increase in size and decrease in solubility from VEGF144, VEGF164, VEGF188 and VEGF205 (Olsson et al. 2006; Ferrara, Gerber, and LeCouter 2003). Larger VEGF isoforms (188 and 205) are

(40)

exclusively found bound to the ECM and require proteolytic cleavage to be freed, while the most common isoform VEGF165 is still soluble (1993; Ferrara, Gerber, and LeCouter 2003). Thus, the varying solubility of VEGFs allow for complex angiogenic patterning imbedded and releasable from the ECM, as well as secretable and diffusible forms for establishing angiogenic gradients. While each isoform plays an important role in development, in the adult vasculature VEGF164 is by far the most abundant, and thus by convention, unless otherwise stated, referred to in this work as VEGF (meaning VEGF-A164).

Shortly after the discovery of VEGF/VPF, its receptors were identified. VEGF receptors are members of the Receptor Tyrosine Kinase (RTK) super family, with 7 immunoglobulin like domains extracellularly and a split tyrosine kinase domain and c-terminus intracellularly (Olsson et al. 2006). Dimerization of the receptor (through ligand binding or force) leads to activation of the receptor through auto phosphorylation in the kinase regulatory domain and further phosphorylation of tyrosine residues in the c terminal tail to facilitate second messenger interactions (Olsson et al. 2006; Ferrara, Gerber, and LeCouter 2003). There are 3 main receptors for VEGF identified, VEGF-R1, VEGF-R2 and VEGF-R3, each varying in affinity, kinase activity, and functional roles (Ferrara, Gerber, and LeCouter 2003). While VEGF-3 is mainly associated with lymphatic tissue, VEGF-R1 and VEGF-R2 are the primary vascular VEGF receptors, however with a unique dichotomy. While VEGF-R1 has a strong affinity for VEGF, it has low kinase activity, conversely VEGF-R2 has lower affinity for VEGF but much higher kinase activity (Ferrara, Gerber, and LeCouter 2003). This has lead to the general hypothesis that VEGF-R1 acts as a VEGF sink, while VEGF-R2 is the primary receptor

(41)

for all or most endothelial VEGF signal transduction (Figure 5). Supporting this notion, while both VEGF and VEGF-R2 knockout mice are embryonically lethal due to lack of vascular development (Shalaby et al. 1995), VEGF-R1 knockouts are lethal due to a pathological over growth of blood vessels (Fong et al. 1995; Kearney et al. 2002). In the developing brain both autocrine (from endothelial cells) and paracrine sources (mostly neurons and astrocytes) of VEGF are required for vascular development and homeostasis (Lee et al. 2007; Haigh et al. 2003). Furthermore, a substantial degree of the regulation of VEGF signalling occurs at the level of the receptor, rather than the ligand. VEGF-R2 surface expression is strictly controlled, with rapid endocytosis and inactivation being critical for accurate responses to local VEGF gradients (Olsson et al. 2006; Nakayama et al. 2013). The main phosphorylation site of activated VEGF-R2 is Tyr1175 (in mice) which leads to downstream activation of PLC, MAPK cascades, PKC, AKT and eNOS (Figure 5) (Olsson et al. 2006; Ferrara, Gerber, and LeCouter 2003).

Interestingly while normally considered pro-angiogenic, VEGF is upregulated by WSS (Zhou, Li, and Chien 2014). Downstream of VEGF signalling and calcium influx, eNOS activation is crucial for EC alignment, polarity, and cytoskeleton regulation through regulation the Rho family GTPases (Cdc42, Rho and Rac) (Kuchan, Jo, and Frangos 1994; Tzima 2006). Specifically, WSS has been shown to lead to activation of AKT (also activated by VEGF through VEGF-R2), AKT phosphorylates eNOS at Ser 1177 increasing NO production (Dimmeler et al. 1999). An important discovery was of an integrin WSS sensing complex linked with VEGF-R2 (Tzima et al. 2005; Wang et al. 2002). WSS forces the dimerization and activation of VEGF-R2, independent of ligand binding (Jin et al. 2003; Tzima et al. 2005). This provides a direct link between the

(42)

physical force of blood flow to VEGF-R2 and in turn to a plethora of signalling pathways such as PI3K, AKT and eNOS. Therefore, the loss of WSS would be expected to have a significant impact on endothelial VEGF-R2 signaling, specifically a decrease in luminal VEGF-R2 activation. The loss of basal levels of activated VEGF-R2 and downstream ATK, eNOS and PLC (and many others, Figure 5) likely shifts endothelial cells out of quiescence and alters intracellular calcium, cytoskeletal rearrangements, and potentially initiating larger scale structural plasticity.

1.8 Vascular plasticity in the developing and adult brain

Since the beginnings of human anatomy, the structural similarities and spatial correlations between the vascular and nervous systems has been apparent, and it is now clear that their development is tightly interwoven (Gelfand, Hong, and Gu 2009; Carmeliet 2003). Early neural development involves an over production of synapses followed by a period of refinement through synaptic pruning, and this similar architectural paradigm is also present in the cerebral vasculature. Early pre-and post natal cortical vasculature show extensive turnover, both angiogenic sprouting and vessel pruning (Harb et al. 2013; Chen et al. 2012; Franco et al. 2015; Phng et al. 2009). Initial angiogenesis and vascular topology is driven by VEGF / Notch signaling and guidance cues such as Semaphorin and Plexins (Kur et al. 2016). Blood flow then dictates subsequent refinement of vasculature through pruning, specifically stable WSS signals a functional vascular branch, turbulent or no WSS indicating a non-functioning vascular segment requiring pruning (Chen et al. 2012; Franco et al. 2015; Kochhan et al. 2013; Lenard et al. 2015). Early vascular plasticity is not limited to VEGF signalling or flow related cues. Recent studies have shown that altering synaptic activity in the developing

(43)

cortex though sensory stimulation or deprivation, can adjust capillary density, (Whiteus, Freitas, and Grutzendler 2014; Lacoste et al. 2014; Lacoste and Gu 2015). Thus, the early postnatal cerebral capillary bed displays a high degree of plasticity in response to angiogenic signals, shear stress and local neural activity.

Like most developmental processes in the brain, there appears to be a critical period for extensive endothelial plasticity. For example, repeated exposure to hypoxic conditions (i.e. 10% Oxygen in ambient air) significantly increased the rate of angiogenesis in mice under 1.5 months of age, but had little impact in mice >3 months (Harb et al. 2013). Furthermore, capillary pruning was unaffected by hypoxia exposure. Of note, the authors reported a low level of capillary pruning that persisted under normal conditions even into old age. These findings are consistent with a study from our own lab where adult cortical capillaries were repeatedly imaged in vivo for weeks after focal ischemic stroke (Tennant and Brown 2013). Contrary to current dogma that stroke serves as a powerful stimulant of angiogenesis, Tennant and Brown found little to no evidence of capillary sprouting in the peri-infarct cortex (Note: focusing on vessels below the cortical surface). In fact, the only common structural event observed after stroke was the regression and elimination of capillary segments after stroke. Both in vivo time lapse imaging studies indicate that the most dominant form of vascular plasticity in the adult brain is vessel pruning. However, what events were responsible for continued pruning throughout postnatal life or following stroke, is unclear (Harb et al. 2013).

Another form of vascular plasticity that was only recently recognized, is called angiophagy (Lam et al. 2010; Grutzendler et al. 2014). Microvessels (8 - 20 μm) occluded by emboli (protein, cholesterol or induced with i.v. injected microspheres)