Fetal Cerebrovascular Maturation: Effects of Hypoxia

Abstract

The human cerebral vasculature originates in the fourth week of gestation and continues to expand and diversify well into the first few years of postnatal life. A key feature of this growth is smooth muscle differentiation, whereby smooth muscle cells within cerebral arteries transform from migratory to proliferative to synthetic and finally to contractile phenotypes. These phenotypic transformations can be reversed by pathophysiological perturbations such as hypoxia, which causes loss of contractile capacity in immature cerebral arteries. In turn, loss of contractility affects all whole-brain cerebrovascular responses, including those involved in flow-metabolism coupling, vasodilatory responses to acute hypoxia and hypercapnia, cerebral autoregulation, and reactivity to activation of perivascular nerves. Future strategies to minimize cerebral injury following hypoxia-ischemic insults in the immature brain might benefit by targeting treatments to preserve and promote contractile differentiation in the fetal cerebrovasculature. This could potentially be achieved through inhibition of RTK mediated growth factors, such as VEGF and PDGR, which are mobilized by hypoxic and ischemic injury and which facilitate contractile dedifferentiation. Interruption of the effects of other vascular mitogens, such as endothelin and angiotensin-II, and even some miRNA species, also could be beneficial. Future experimental work that addresses these possibilities offers promise to improve current clinical management of neonates who have suffered and survived hypoxic, ischemic, asphyxic, or inflammatory cerebrovascular insults.

Introduction

Recent innovations in high resolution imaging and genomic analysis have dramatically expanded studies of vascular biology, and these new approaches are now beginning to find applications in the study of the dynamic relations between structure and function in the immature cerebral circulation. This review summarizes some of these advances, with emphasis on the functional competence of the term fetal cerebral circulation, how it is affected by an adverse intrauterine environment, and possible new strategies to ameliorate fetal and neonatal cerebrovascular compromise.

Origins of the Fetal Cerebrovasculature

The cerebral vasculature originates at about 24 days of gestation, and grows from an initial leptomeningeal plexus into identifiable vascular structures by 28 days¹. These early vessels continuously grow and differentiate in waves that spread radially from the base of the brain towards the midbrain convexity. Between 6 and 7 weeks of gestation, portions of the dorsal aorta, first aortic arch, and third aortic arch merge to form the internal carotid artery. From this point onward, the cerebrovascular bed arborizes rapidly to facilitate perfusion of the developing cortex². Human cerebral arteries continue to grow and diversify well into the third postnatal month at which time vessel density begins to stabilize³. In contrast, capillary density levels off near term in telencephalic white matter but continues to increase in cortical gray matter throughout the first three or four years of postnatal life⁴. This regional and temporal heterogeneity emphasizes the highly dynamic character of cerebral vasculogenesis, which in turn helps explain why immature cerebral vessels are so vulnerable to metabolic and mechanical injury. Importantly, throughout life the entire cerebral vasculature continues to undergo remodeling and angiogenesis, and can dedifferentiate in response to injury or stress even in the adult brain.

One of the most common and important pathophysiological stresses on the developing cerebrovasculature is hypoxia. When oxygen delivery cannot sustain local increases in metabolic activity, tissue hypoxia results. In turn, hypoxia stimulates multiple cell types within the brain to produce and release HIF-1α (Hypoxia Inducible Factor), which is a transcription factor that can translocate to the cell nucleus. Within the nucleus, HIF-1α can combine with its coactivator, HIF-1α, and bind the promoter regions of numerous genes that help mediate cellular responses to hypoxia. These include genes for Vascular Endothelial Growth Factor and Erythropoietin, as well as for glycolytic enzymes and proteins regulating mitochondrial function^5,6. In combination with other angiogenic factors such as angiopoietin⁷, transforming growth factor-α⁸, and Wnts⁹, these factors govern the formation and differentiation of new cerebral vessels. This diversity of factors that regulate angiogenesis enables a highly robust and redundant regulation that assures close matching between vascular architecture and local metabolic activity. Such a system is essential to support the high rates of growth and expansion characteristic of the immature brain.

Smooth Muscle Differentiation

In contrast to cardiac myocytes, which undergo terminal differentiation, smooth muscle does not¹⁰. Instead, smooth muscle, and in particular vascular smooth muscle, exists in a continuum of different phenotypes that include broad variations in the capacity for migration, proliferation, secretion and contraction^10,11. In addition, these phenotypes can transform into one another under the influence of numerous growth factors, including Vascular Endothelial Growth Factors, Platelet-Derived Growth Factor, Fibroblast Growth Factors, amines such as norepinephrine and serotonin, peptides such as angiotensin and endothelin, and many others¹⁰. Correspondingly, the receptors for these vascular growth factors include both Receptor Tyrosine Kinases and G-Protein Coupled Receptors that convey information and signals from outside the smooth muscle cell into the vascular smooth muscle cytoplasm, and in many cases, on into the nucleus. Indeed, many of these cytosolic signaling pathways involve selective phosphorylation and nuclear translocation of key vasoregulatory transcription factors such as myocardin, serum response factor, ternary complex factor, and Elk-1¹². Another critically important influence on smooth muscle differentiation is “excitation-transcription coupling,” as first articulated by Wamhoff¹³. This idea couples cyclical changes in cytosolic calcium concentration, as occur during smooth muscle contraction and activation, with changes in gene expression through the actions of calcium-dependent kinases and phosphatases that target transcription factors. As such, these mechanisms help smooth muscle phenotype adapt to its history of activation by physiological stimuli such as changes in vasoactive agonist concentrations, sheer stress, and pressure-induced stretch. Together, these diverse influences constitute a near continuous barrage of factors that work together to govern the phenotype, structural, and functional characteristics of all cerebrovascular smooth muscle.

Owing to the diversity of factors that govern cerebrovascular smooth muscle phenotype, the dynamic and variable nature of their influences, and regional differences in their production, clearance, and/or activity, smooth muscle phenotype in immature cerebral arteries is highly heterogeneous. This is important when comparing cerebral arteries from different brain regions, but even within the same artery, adjacent smooth muscle cells may be of different phenotypes such that the artery wall includes cells in proliferative, migratory, synthetic, and contractile phenotypes at any given time. A key marker for smooth muscle phenotype in immature vascular smooth muscle is the myosin heavy chain isoform it expresses. At the beginning of smooth muscle differentiation, the cells initially express non-muscle myosin heavy chain. As differentiation proceeds, this non-muscle myosin is gradually replaced with smooth muscle myosin heavy chain of the S1 type¹⁴. As differentiation continues further, the cells begin to express the S2 isoform of smooth muscle myosin heavy chain, which is approximately 34 amino acids shorter than the S1 isoform and is required for maximal contractile efficiency. At full contractile differentiation, the S1 and S2 isoforms of myosin heavy chain are approximately equal in abundance, and these complex together to form the myosin hexamers that mediate contraction^14,15. Similarly, many other contractile proteins also exhibit graded expression during differentiation, including smoothelin, calponin, caldesmon, and many others, each on an independent time course^10,11. In aggregate, these changing patterns of expression and intracellular organization reveal that smooth muscle phenotype is a continuum of many overlapping characteristics that together define not just the state of differentiation at any moment in time, but also smooth muscle structure and function. In general, smooth muscle cells in fetal cerebral arteries exhibit a smaller proportion of fully contractile cells than found in adult arteries, which helps explain their reduced overall contractility, and differential reactivity to pathophysiological stimuli such as infection, inflammation, ischemia, and hypoxia^15–17.

Effects of Hypoxia on Cerebrovascular Structure and Function

One of the most common and important pathophysiological factors influencing cerebrovascular development is hypoxia^18,19. At the level of the microcirculation, hypoxia acts through HIF-1a to promote angiogenesis, increase capillary density, and thereby reduce inter-capillary distances over which oxygen must diffuse to reach every cell within the brain parenchyma^20–28. Given that hypoxia is a prominent aspect of ischemia, cerebral ischemia also promotes many of the same microcirculatory responses²⁹. In cerebral arteries upstream of the microcirculation, hypoxia causes significant changes in composition, including increases in wall thickness and protein content in a manner that is graded with the severity and duration of hypoxia^18,30–32. Age is also a critical factor in remodeling responses to hypoxia^33–35 due in large part to developmental and maturational differences in the populations of cell surface receptors on the smooth muscle cells for both RTK-dependent and GPCR dependent vasotrophic factors such as VEGF¹⁵ and endothelin-1^19,36. For VEGF, receptor levels for both the FLT and FLK types are expressed at lower levels in fetal than adult arteries, and hypoxia dramatically increases both receptor types in both age groups^37,38. Hypoxia also increases expression of endothelin-A receptors in fetal cerebral arteries³⁶.

Commensurate with hypoxic changes in composition, cerebrovascular function is also modulated by hypoxia. Long-term hypoxia depresses contractile capacity through depressed expression of GPCR receptors that stimulate contraction^30,39. Chronic hypoxia also can depress cerebrovascular IP3 synthesis and IP3 receptor density⁴⁰. At the level of the contractile proteins, chronic hypoxia attenuates calcium-dependent myosin light chain phosphorylation but simultaneously enhances the ability of phosphorylated myosin to generate contractile force. The net result of these opposing effects is that myofilament calcium sensitivity is increased by chronic hypoxia in fetal cerebral arteries³⁴. Chronic hypoxia also increases the agonist binding affinity of serotonergic receptors⁴¹ and decreases the ability of calcium-sensitive and ATP-sensitive potassium channels to mediate vasorelaxation^42,43. The integrated result of these effects is that reactivity to both contractile and relaxant stimuli is attenuated by chronic hypoxia in fetal cerebral arteries. As a result, fetal cerebral arteries respond more slowly and less forcefully to vasoactive stimuli. An obvious consequence of this pattern is that under conditions of hypoxia, fetal cerebral arteries are less able to match cerebral perfusion to local metabolic activity, constrict more slowly in response to rising arterial pressure, and overall are less efficient at maintaining cerebrovascular homeostasis.

Chronic hypoxia also influences the density and efficacy of the perivascular innervation of cerebral arteries. In perivascular nitridergic nerves, chronic hypoxia reduces the content of neuronal nitric oxide synthase and the ability of these nerves to release nitric oxide⁴⁴. Given that NO released from nitridergic nerves can enhance norepinephrine release from nearby adrenergic nerve endings, the effects of chronic hypoxia on nitridergic nerves secondarily decrease the release of norepinephrine from perivascular adrenergic nerves^35,45. In opposite fashion, chronic hypoxia upregulates the ability of perivascular adrenergic nerves to activate adrenergic receptors on fetal cerebral arteries^32,46,47. These increases in adrenergic activation potently influence phenotypic differentiation in fetal vascular smooth muscle, which in turn contribute to increased contractile differentiation in hypoxic fetal cerebral arteries⁴⁸.

Just as gestational hypoxia can dramatically alter the pattern and timing of cerebrovascular development, it also powerfully influences neuronal and glial development and many of these effects have important secondary effects on cerebrovascular development. Antenatal maternal hypoxia can alter the expression of numerous genes involved in neuronal growth and death, and as such hypoxia can precipitate neurological disorders in the neonate if it acts during key developmental time periods^49–51. For example, hypoxia can act through epigenetic mechanisms to alter the renin-angiotensin system in both extracerebral^52,53 and cerebral tissues^54–58. Owing to the potent influences of angiotensin on smooth muscle differentiation and phenotype⁵⁹, it follows that hypoxic modulation of neuronal and glial angiotensin pathways will also influence local cerebrovascular development. Chronic hypoxia modulates the hypothalamo-pituitary-adrenal (HPA) axis, and alters the influence of glucocorticoids on fetal cerebrovascular maturation^60,61; changes in circulating concentrations of glucocorticoids can program fetal cerebral arteries, with long-term consequences for cerebrovascular structure and function, including depression of myogenic reactivity and regulation of cytosolic calcium within cerebrovascular smooth muscle⁶². These myriad effects of hypoxia vary with the intensity and duration of hypoxia, and are heavily dependent upon developmental age and brain region. Although this heterogeneity can greatly complicate understanding of how a specific insult will affect the immature brain and its vascular supply, it also emphasizes that growth and development of the fetal brain as a whole is highly integrated across multiple cell types, is robust due to redundancies in the mechanisms regulating growth, and that together these systems normally enable highly coordinated patterns of functional maturation.

Cerebrovascular Endothelium and the Blood-Brain Barrier

As in all vascular beds, the endothelium of fetal cerebral arteries serves four main functions important for: 1) angiogenesis; 2) hemostasis; 3) blood brain barrier function; and 4) vascular tone. The first of these to mature is the ability of the endothelium to initiate capillary angiogenesis through the release of vascular growth factors such as Platelet-Derive Growth Factors, Fibroblast Growth Factors, Thrombospondin, and Insulin-Like Growth Factors, all of which promote angiogenesis⁶³. Conversely, the endothelium can release angiostatic factors that inhibit angiogenesis, including nitric oxide, prostacyclin, TGF-β, herparin,and heparin sulfate. As a whole, this diversity of factors enables the immature endothelium of fetal cerebral arteries to closely control the high rates of vascular growth and expansion needed to support the rapid growth of the brain during late fetal and early neonatal life.

The hemostatic functions of the cerebrovascular endothelium become functionally competent in the 3^rd trimester, at which time intact endothelial cells begin to continuously release nitric oxide and prostacyclin, which in turn inhibits initiation of hemostasis.⁶⁴. Healthy endothelial cells also contain multiple pro-coagulant factors, including tissue factor (thromboplastin), von Willebrand factor, factor V, factor X, and platelet activating factor. Endothelial injury causes the release of these factors and initiation of hemostasis. Throughout late fetal development and early postnatal maturation, endothelial content and the ability to release these factors increases steadily⁶⁵, as does production and release of nitric oxide^66–68.

A critically important and unique characteristic of cerebrovascular endothelium is its ability to efficiently facilitate or restrict the movement of substances between circulating blood and the brain interstitial space⁶⁹. This endothelial blood-brain-barrier function is due in large part to tight junctions that bind adjacent endothelial cells to one another. Cerebrovascular endothelium also expresses the integral membrane protein, occludin, which spans intracellular gaps and further binds adjacent endothelial cells together⁷⁰. In addition, cerebrovascular endothelium also expresses claudins, a family of cell-cell adhesion molecules that increase electrical and hydraulic resistance across the cerebrovascular endothelial layer^71,72. The expression and localization of all of these components of the blood-brain-barrier continues throughout fetal life, and attain full functional maturity, as characterized by low hydraulic permeability, not until well after birth^73–75.

A major limit on the rate at which the immature blood-brain-barrier tightens to reduce its overall permeability is the rapid pace of angiogenesis in the fetal brain. In newly formed capillaries, tight junctions, occludins, and claudins appear gradually over several days or weeks and during this initial maturation period, initial permeability is high as is electrical conductivity. The astrocytes that envelop cerebral capillaries are not fully mature at birth^76,77.

Similarly, subendothelial pericytes, which help tighten and maintain the blood-brain-barrier, also are more sparse in fetal than in adult cerebral arteries⁷². Together, these characteristics render the immature blood-brain barrier less efficient and more vulnerable to stress and injury than the fully mature blood-brain barrier that begins to appear only in early postnatal life⁷⁸.

Another important role of the cerebrovascular endothelium is the regulation of local vascular tone through the release of vasoactive factors, one of the most important of which is nitric oxide⁷⁹. The short half-life of nitric oxide restricts its influence to nearby smooth muscle, in which it activates soluble guanylate cyclase and promotes vasodilatation⁸⁰. The physiological release of nitric oxide is stimulated by shear stress, but this mechanism is attenuated in fetal compared to adult cerebral arteries^66–68. Even so, this effect of shear stress, in turn, implies that any influence that increases blood viscosity (polycythemia, hyperproteinemia, dehydration, etc.) or blood flow velocity, will correspondingly enhance nitric oxide release from a healthy cerebrovascular endothelium. The fetal cerebrovascular endothelium also releases the vasodilator prostacyclin⁸¹, but through pathways that are fully developed in fetal cerebrovascular endothelium⁸². Fetal cerebrovascular endothelium also can release a broad variety of other vasoactive molecules, including the vasodilator Endothelium-Derived Hyperpolarizing Factor⁸³, and vasoconstrictors such as endothelin, thromboxane A2, and superoxide⁸⁴. Overall, however, the structural and functional immaturity of the fetal cerebrovascular endothelium generally attenuates its influence on vascular tone in fetal, compared to adult, cerebral arteries.

Hypoxia of even short duration can modulate multiple aspects of endothelial function in cerebral arteries, including dysregulation of initial vessel formation and architecture (vasculogenesis) secondary to increased VEGF activity. Thus, sustained in utero hypoxia during human gestation can increase fetal blood-brain permeability⁸⁵ and attenuate endothelium-dependent vasodilatation³⁰ through mechanisms that appear to be conserved across many vertebrate species, including birds⁸⁶. The depressive effects of chronic hypoxia on endothelium-dependent vasodilatation involve depression of transcription of the eNOS gene, and inhibition of synthesis of functional eNOS protein⁸⁷. In parallel, hypoxia also inhibits expression of soluble guanylate cyclase in cerebrovascular smooth muscle, and thus reduces the ability of endothelial NO to stimulate cGMP synthesis and promote vasodilatation⁸⁸. In contrast to chronic hypoxia, acute hypoxia can enhance the release of endothelium-dependent relaxant factors such as nitric oxide⁸⁹, although the direct contribution of prostanoids in this effect appears to be minimal⁹⁰. The effects of hypoxia, either acute or chronic, on the release of vasoactive molecules other than nitric oxide or prostanoids from the fetal cerebrovascular endothelium remain largely unexplored.

Fetal and Neonatal Whole Brain Cerebrovascular Reactivity

A main theme of the fetal and neonatal brain is heterogeneity: structural, functional, and regional. Cerebral arteries of different size also exhibit different characteristics, variations in the size and phenotype of smooth muscle cells, extent and type of innervation at the adventitial-medial junction, and patterns of receptor and ion-channel expression, etc. A cerebral artery with a diameter of 150 μm from the brainstem also has different characteristics than a similarly-sized artery from the cortex or pial circulation; local factors heavily influence vascular differentiation. Developmental age further accentuates this heterogeneity such that cerebral arteries more proximal to the heart appear to mature more quickly than those distal to the heart. Another important type of cerebrovascular heterogeneity is chronological heterogeneity, which reflects the recent history of activation of smooth muscle by local vasoactive influences. These influences typically change patterns of phosphorylation of contractile proteins, membrane receptors, and other determinants of contractility, such that vasoreactivity to any given stimulus depends heavily upon the recent past. Such diversity in heterogeneity related to artery size, age, region of origin, and history of activation, complicates appreciation of global patterns of regulation, but is the very feature that enables the immature brain to maintain cerebrovascular homeostasis despite constantly changing and highly variable conditions.

Flow-Metabolism Coupling

At its most fundamental level, cerebrovascular homeostasis requires a tight coupling between local cerebral metabolic rate and local perfusion. From this perspective, local metabolic rate is the primary variable driving changes in local cerebral perfusion, regardless of all other aspects of cerebrovascular heterogeneity. Given that brain metabolism is uniquely reliant on the catabolism of glucose, the coupling between blood and cerebral metabolism typically involves simultaneous coupling between the rates of glucose oxidation, oxygen consumption, and local cerebral perfusion⁹¹. Compared to adult values, human neonatal cerebral blood flow and metabolic rate are both low at birth but increase gradually through the first few years of postnatal life and reach a maximum at about age six, after which rates of cerebral blood flow and metabolism begin a gradual decline that continues throughout juvenile, adolescent, and adult life⁹². At all ages, however, the rates of glucose catabolism, oxygen consumption, and cerebral perfusion are tightly coupled in virtually all mammalian species, including humans^93–95.

Decades of intensive research have revealed that coupling between cerebral metabolism and perfusion is mediated by a broad variety of factors, one of the most important of which is adenosine⁹⁶. As the oxygen supply-demand ratio begins to fall, ATP becomes progressively converted to ADP, AMP and then adenosine, which is released from all cell types in the brain parenchyma⁹¹. The metabolic pathways that mediate this coupling between the oxygen supply-demand ratio and adenosine release are fully functional in the fetal brain, and as the overall rate of cerebral metabolism increases throughout late fetal and early postnatal life, so too does the brain interstitial concentration of adenosine⁹⁷. The adenosine A2a receptor subtype that promotes vasodilatation when bound to adenosine, is expressed on cerebrovascular smooth muscle cells early in fetal brain development and the presence of these receptors in cerebral arteries persists throughout adult live⁹⁸. In turn, A2a receptors enable feedback regulation that ultimately couples increased extracellular adenosine concentrations to increased cerebral perfusion and increased oxygen supply in the immature brain^99–101. This coupling is critically important during postnatal life as fetal hemoglobin is gradually replaced with adult hemoglobin, which in turn increases the mass of oxygen that can be extracted from each unit of hemoglobin^101,102. The adenosine pathway is also a critical mediator of hypoxic vasodilatation in cerebral arteries of all ages, which implies that any pathophysiological disturbance that interferes with the release of adenosine, or its binding to vascular A2a receptors, will interrupt hypoxic flow-metabolism coupling¹⁰³.

Hypercapnic Vasodilatation

Carbon dioxide is a universally recognized cerebral vasodilator that can increase cerebral blood flow in many different species of all ages, from fetus to adult^93,104. Importantly, the magnitude of the vasodilator response to hypercapnia gradually increases during postnatal life^105–107, which has been attributed to parallel increases in reactivity to reduced pH as cerebral arteries mature¹⁰⁸. Age-related decreases in hematocrit have also been suggested to contribute to age-related increases in cerebral CO2 reactivity¹⁰⁹. Hypercapnia also appears to stimulate the synthesis and release of vasodilator prostaglandins^110–112, although their cellular origin remains uncertain and may include astrocytes¹¹³ and cerebrovascular endothelial cells¹¹⁴ as well as other cell types of the cerebral parenchyma. Other mechanisms may also contribute to hypercapnic vasodilatation, including release of nitric oxide from either neurons or endothelial cells^115,116, and activation of ATP-sensitive and calcium-sensitive potassium channels^117,118. Conversely, hypercapnic vasodilatation in the immature cerebral circulation appears not to depend upon changes in cerebral metabolic rate^106,119,120 or activation of perivascular nerves¹²¹. Without doubt, multiple mechanisms mediate the cerebrovascular response to hypercapnia, which appears due in large part to the ability of carbon dioxide to readily diffuse into all cell types of the brain, where it can differentially disturb intracellular acid-base balance.

Acute Hypoxic Vasodilatation

Given that the human cerebral circulation is largely inaccessible to investigation, particularly in the fetus, most understanding of hypoxic cerebral vasodilatation in the immature brain has originated from studies in animals¹²².

Across all mammalian species, acute or short-term hypoxia produces a marked cerebral vasodilatation in all age groups including fetuses and neonates^{97,108,123,124}. Investigations of these responses in premature human neonates have largely corroborated the results from animal models¹²⁵. As might be expected, the areas of the brain that exhibit the highest metabolic rates also demonstrate the strongest immediate responses to hypoxia^75,126–129. The magnitude of the cerebrovascular response to hypoxia is also graded with the severity of hypoxia, with the important result that hypoxic increases in cerebral blood flow at least initially maintain the oxygen supply-demand ratio and the oxygen extraction fraction over a broad range of blood oxygen tension values¹³⁰. If hypoxia becomes more severe and arterial oxygen tensions fall below about 50 mm Hg, rates of cerebral perfusion can become maximal, in which case the oxygen extraction fraction begins to increase. Once both cerebral perfusion and oxygen extraction are maximal, sustained or more severe hypoxia will decrease cerebral oxygen consumption, increase rates of anaerobic glycolysis, and ultimately cause lactic acidosis¹³¹. In essence, this sequence of events is an extrapolation of the adenosine-dependent mechanisms that govern physiological maintenance of the oxygen supply-demand ratio and are fully functional in neonates of most vertebrate species^97,123,132. Aside from adenosine, acute hypoxia promotes release of many other vasoactive factors, including vasodilator prostaglandins¹³³ and endothelial factors⁸⁹. Hypoxia also has potent direct effects on cerebrovascular smooth muscle that attenuate the influx and mobilization of calcium essential for contraction¹³⁴. Again, the relative importance of each of these mechanisms varies significantly amongst arteries of different size, from different regions, and at different ages. Even so, this pronounced redundancy of mechanisms constitutes a strong resistance to hypoxic injury and fails to maintain flow-metabolism coupling only under extreme conditions.

Autoregulation

The efficiency of baroreceptor-mediated regulation of blood pressure improves with fetal weight and age¹³⁵, which explains why slow oscillations in blood pressure are more common in fetuses and neonates than adults. These dynamic swings in blood pressure require that cerebral autoregulatory mechanisms, which help maintain flow-metabolism coupling despite changes in blood pressure, are intact in the fetal and neonatal brain. When assessed using indirect methods, cerebral autoregulation in human fetuses appears in the third trimester, perhaps as early as 23 weeks, and improves steadily thereafter^93,136. Strictly defined, however, cerebral autoregulation maintains a relatively constant rate of cerebral blood flow in the face of changes in cerebral perfusion pressure¹³⁷. Cerebral perfusion pressure, in turn, is calculated as the difference between mean arterial pressure and the greater of either intracranial pressure or cerebral venous pressure. This is an important distinction, because intracranial hypertension directly compresses the cerebral veins, obstructs venous outflow, and decreases cerebral perfusion pressure^138,139.

In the neonates of most species, including humans, autoregulatory mechanisms promote a progressive vasodilatation as perfusion pressure falls to about 30 to 40 mmHg, which is the lower limit of autoregulation^{75,108,140,141}. At perfusion pressures below the lower limit of autoregulation, cerebral blood flow falls in direct proportion to perfusion pressure.

Just as blood pressure gradually rises with developmental age, so too does the upper limit of autoregulation, which typically falls between 70 and 100 mm Hg in neonates. Under some circumstances, however, infant blood pressure oscillations can drive blood pressure above the upper limit of cerebral autoregulation, which increases the risk of vessel injury and rupture with a subsequent intracranial bleed^141–143. This upper limit extends as high as 140 to 150 mm Hg in adults¹³⁷, due in part to reflex sympathetic vasoconstriction¹⁴⁴; this mechanism is not fully functional in immature cerebral arteries¹⁴⁵.

Despite decades of intensive investigation, the cellular mechanisms that mediate cerebral autoregulation remain uncertain, and appear to involve a blend of myogenic, metabolic, and also neurogenic mechanisms^137,146. As the name implies, myogenic reactivity originates within the smooth muscle itself as a direct response to stretch, which mobilizes intracellular calcium^147,148. Numerous mechanisms have been suggested to help mediate this response, including the activation of cell-surface integrins¹⁴⁹ or stretch activation of TRP calcium channels¹⁵⁰. In immature cerebral arteries, most of these cellular mechanisms are not yet fully developed, due in part to the relatively low fraction of smooth muscle cells that have undergone complete contractile differentiation. Correspondingly, any perturbation that promotes contractile dedifferentiation will also attenuate autoregulatory capacity, which helps explain why autoregulation is so vulnerable to injury in immature cerebral arteries.

Mechanisms that couple flow to metabolism reinforce myogenic mechanisms of neonatal cerebral autoregulation. Indeed, cerebral autoregulation can be conceived as a family of blood flow-perfusion pressure curves, one for each level of cerebral metabolic rate. In support of this view, decreases in cerebral perfusion of the neonatal brain increase interstitial concentrations of adenosine^151,152, as well as prostanoids^153,154, opioids¹⁵⁵, and other vasodilator molecules. To an extent limited by the functional maturation of the perivascular innervation, sympathetic adrenergic mechanisms also can contribute to cerebral autoregulation, and extend the upper limit in neonates^156,157, as in adults¹⁵⁸. Given that cerebral autoregulation depends on mechanisms that are not fully mature in fetal and neonatal cerebral arteries, such as the extent of contractile differentiation of cerebrovascular smooth muscle and the functional efficiency of the perivascular innervation, it follows that neonatal cerebral autoregulation is somewhat fragile and highly vulnerable to many insults¹⁵⁹ including acidosis¹⁶⁰, sustained or severe hypoxia¹⁶¹, asphyxia¹⁶², cerebral ischemia¹⁵³, and intracranial hemorrhage¹⁶³.

Neurovascular Mechanisms

Fully mature cerebral arteries are innervated by a rich plexus of nerves that terminate near the adventitial-medial junction within the arterial wall¹⁶⁴. This perivascular plexus includes contributions from adrenergic, cholinergic, and peptidergic nerves, all of which originate during a late embryonic outgrowth phase followed by a synthesis phase during which the pathways essential for neurotransmitter synthesis and storage are developed. Lastly, the synaptic varicosities of these nerves differentiate and become functional¹⁶⁵. Typically, fetal arterial smooth muscle cells express post-synaptic receptors well before the nerves become functional with the result that reactivity to autonomic neurotransmitters is often greater in fetal than adult arteries¹⁶⁶, due, in part, to the absence of presynaptic neuronal reuptake¹⁴⁵. As for many other aspects of fetal cerebrovascular development, the pace of maturation of the perivascular innervation is highly heterogeneous and depends on artery size, nerve type, region, and developmental age.

Adrenergic nerves are the most widely studied of all nerve types that make up the cerebrovascular perivascular innervation. Sympathetic fibers first develop in arteries supplying rostral brain regions and begin to establish functional varicosities between about 19 and 23 weeks of gestation in the human¹⁶⁴. By term, exogenously applied norepinephrine can activate α1 adrenergic receptors in small cerebral arteries¹⁶⁷ and α2 adrenergic receptors in pial arterioles¹⁶⁸, which elicits contraction^168–170. Noradrenergic activation can also stimulate the synthesis and release of the vasodilator prostaglandin PGE2, which can attenuate vasoconstrictor responses to norepinephrine¹⁶⁸. Electrical activation of perivascular sympathetic nerves can constrict cerebral arteries and decrease cerebral perfusion up to 25%, depending upon brain region and developmental age^171,172. Together, these findings emphasize the regulatory importance of noradrenergic mechanisms in immature cerebral arteries¹⁷⁰.

In parallel with adrenergic nerves, cholinergic nerves also begin to appear in the perivascular innervation somewhere between 19 and 23 weeks of gestation in human cerebral arteries, and become functional at term¹⁶⁴. In contrast to adult cerebral arteries, fetal and neonatal cerebral arteries dilate in response to low concentrations of acetylcholine¹⁷³, but contract in response to high concentrations¹⁶⁶. Given that indomethacin can markedly attenuate acetylcholine-induced contractions in immature cerebral arteries, these contractions appear dependent upon the synthesis and release of vasoconstrictor prostaglandins such as thromboxane A2 or PGF2α¹⁷⁴. In turn, these findings suggest that the vasoactive effects of indomethacin, or any other cyclooxygenase inhibitor, when these drugs are administered to neonates.

Nerves that contain and release peptide neurotransmitters are the third component of the perivascular innervation of cerebral arteries¹⁷⁵. Among this class of perivascular nerves, those that release neuropeptide-Y or vasoactive intestinal polypeptide are most common in human fetal cerebral arteries, particularly in arteries of the Circle of Willis and pial circulation¹⁷⁶. The fetal perivascular innervation also includes sensory fibers that originate from the trigeminal ganglion and can release vasoactive intestinal polypeptide, calcitonin gene-related peptide, substance-P, somatostatin, and cholecystokinin¹⁷⁷. Whereas the functions of the vasodilator peptides have been extensive studied in the adult cerebral circulation due to their possible involvement in migraine¹⁷⁸, almost nothing is known of the functions of these neurotransmitters in the fetal cerebral circulation. In turn, studies of the involvement of perivascular peptides in the fetal cerebral circulation seem warranted, particularly in light of findings that these peptides contribute significantly to adult cerebrovascular responses to ischemia and intracranial hemorrhage^{175,179–181}.

Future Directions

One of the most common pathophysiological challenges for a developing fetus is in utero hypoxia, particularly if hypoxia persists over the final days or weeks before birth. In fetal cerebral arteries, chronic hypoxia increases water content, total protein content, and wall thickness³⁰. Accompanying these significant changes in structure and composition are depressed overall contractility and depressed endothelial vasodilator capacity^30,31 that is attributable both to a decreased capacity for endothelial NO synthesis, reduced soluble guanylate cyclase activity within cerebrovascular smooth muscle, and an attenuated ability of PKG to act on BK channels^43,182,183. Chronic hypoxia also modulates the abundances, organization and function of contractile proteins, due partially to the effects of VEGF, resulting in a less efficient contractile apparatus^15,34,184. Conversely, hypoxia also promotes the growth and expansion of the perivascular adrenergic innervation, which helps maintain contractile differentiation of cerebrovascular smooth muscle^48,185. However, in aggregate these hypoxic changes yield a cerebrovasculature that is generally less reactive and slower to respond to changes in blood pressure, blood gases and other contractile stimuli than typical of a healthy normoxic fetus^36,41,161. What then, is the best strategy to manage an infant that has suffered but survived in utero or perinatal hypoxia?

Fortunately, the fetal and neonatal cerebrovasculature is endowed with a broad variety of cellular mechanisms that enable compensation for stress and injury. When injury is severe but does not cause extensive apoptosis or death of local arterial smooth muscle, this compensation can cause the phenotypic dedifferentiation of cerebrovascular smooth muscle resulting in loss of contractile capacity. This is an important factor contributing to the long-recognized pattern of “pressure passive” cerebral blood flow following severe injury in the neonate¹⁸⁶. In turn, this pattern of compensatory dedifferentiation implies that a “pressure passive” cerebral circulation does not always indicate irreversible loss of neonatal cerebrovascular function. The challenge then is to know how to rescue injured cerebrovascular smooth muscle and restore its contractile function and capacity for flow-metabolism coupling. Assuming that some assessment of cerebrovascular reactivity, perhaps using perfusion-weighted MRI techniques¹⁸⁷, reveals persistence of at least some vascular reactivity to perturbations such as mild hypocapnia or hypercapnia, the strategy would be to promote contractile differentiation in the surviving cerebrovascular smooth muscle.

One possible approach to stimulating contractile differentiation might be to administer nitric oxide, which is a standard clinical therapy¹⁸⁸. Nitric oxide stimulates the synthesis of cGMP, which in turn activates Protein Kinase G and promotes contractile differentiation¹⁸⁹. Another strategy may be to antagonize VEGF, whose synthesis and release are directly stimulated by hypoxia¹⁹⁰. VEGF, in turn, promotes not only angiogenesis, but also contractile dedifferentiation in arterial smooth muscle¹⁵. A molecule with about 60% homology to VEGF is PDGF, which can also promote contractile dedifferentiation^191,192. Correspondingly, antagonism of the PDGF receptor following intracerebral hemorrhage has proven to help preserve contractile differentiation and reduce overall cerebral injury¹⁹³. A variety of G-Protein Coupled Receptors, including those activated by endothelin³⁶ and angiotensin II⁵⁹, also potently influence contractile differentiation, although the effects of inhibitors of these peptides have yet to be investigated in the context of hypoxic-ischemic injury in the immature brain. Finally, recent discoveries that multiple miRNA molecules, including miR-1¹⁹⁴, miR-29c¹⁹⁵, miR-145¹⁹⁶, and potentially many others¹⁹⁷ can significantly influence smooth muscle phenotype. Future therapies may be possible through selective administration of these molecules, or their antagonists, to preserve or promote contractile differentiation in the immature cerebrovasculature. This therapeutic potential is currently motivating intensive interest in the vascular actions of these molecules. Meanwhile, in the immediate future it seems reasonable to focus on the potential therapeutic benefit of treatments already approved for human use, including nitric oxide administration and antagonism of VEGF and PDGF receptors.