The post-arteriole transitional zone: a specialized capillary region that regulates blood flow within the CNS microvasculature
Handling Editors: Laura Bennet & Justin Dean
The peer review history is available in the Supporting information section of this article (https://doi.org/10.1113/JP282246#support-information-section).
Abstract
The brain is an energy hog, consuming available energy supplies at a rate out of all proportion to its relatively small size. This outsized demand, largely reflecting the unique computational activity of the brain, is met by an ensemble of neurovascular coupling mechanisms that link neuronal activity with local increases in blood delivery. This just-in-time replenishment strategy, made necessary by the limited energy-storage capacity of neurons, complicates the nutrient-delivery task of the cerebral vasculature, layering on a temporo-spatial requirement that invites - and challenges - mechanistic interpretation. The centre of gravity of research efforts to disentangle these mechanisms has shifted from an initial emphasis on astrocyte-arteriole-level processes to mechanisms that operate on the capillary level, a shift that has brought into sharp focus questions regarding the fine control of blood distribution to active neurons. As these investigations have drilled down into finer reaches of the microvasculature, they have revealed an arteriole-proximate subregion of CNS capillary networks that serves a regulatory function in directing blood flow into and within downstream capillaries. They have also illuminated differences in researchers’ perspectives on the vascular structures and identity of mural cells in this region that impart the vasomodulatory effects that control blood distribution. In this review, we highlight the regulatory role of a variably named region of the microvasculature, referred to here as the post-arteriole transition zone, in channeling blood flow within CNS capillary networks, and underscore the contribution of dynamically contractile perivascular mural cell – generally, but not universally, recognized as pericytes – to this function.
Introduction
‘Science advances one funeral at a time.’ This oft-paraphrased quote by German physicist and Nobel Prize recipient Max Planck sums up the state of play in all areas of scientific research at one time or another in which progress is stymied by a dominant figure who exercises outsized influence over the contours or direction of research in the field. According to this sentiment, nothing short of the permanent departure of such figures from the scene will allow the field to resume its natural course. The brain microvascular field – especially as it applies to blood flow regulation within capillary networks – finds itself at a similar impasse. In this case, however, there are many equally prominent voices, so funerals do not present themselves as an obvious solution. However, there is actually reason to be optimistic about a congenial resolution because, at root, most persisting barriers to progress in the field revolve around how things are named rather than what they do.
In this brief review, we highlight functional themes relating to regulation of blood flow within capillary networks of the central nervous system (CNS) microcirculation, focusing on concepts that are generally held in common but are nonetheless argued about because of differences in nomenclature. In particular, we address the role of a functionally distinct – but variably named – region of the capillary bed in regulating the fine control of blood flow within capillaries and its intersection with neurovascular coupling (NVC) mechanisms that link regional neuronal activity to arteriole-mediated increases in bulk flow to localized target regions. The emphasis is on the involvement of ‘contractile pericytes’ – an oxymoron in some quarters – in this regulatory function.
The physiological context
The cerebrovascular system has evolved to meet the unique and everchanging demands of neurons. It is not only charged with maintaining adequate perfusion of brain tissue generally, it must also be capable of redirecting blood flow to regions of higher neuronal activity reflecting ongoing sensory-motor processing, learning and memory, and other functions. These unique demands on the cerebral vasculature imposed by the brain's computational functions are met by an incompletely understood hierarchy of mechanisms that serve to rapidly deliver blood to regions of higher neuronal activity. The need for such activity-dependent increases in blood perfusion (functional hyperaemia) arises because neurons have limited energy reserves and are sensitive to even relatively brief periods of oxygen and nutrient deprivation. To meet these spatially and temporally varying energetic demands, evolution has had to thread the needle between ensuring the delivery of nutrients precisely to areas where needs are acute and avoiding the risk of squandering resources on relatively quiescent areas.
Most early research on the NVC mechanisms that link neuronal activity to changes in brain vessel diameter and blood flow focused on the role of arterioles, coalescing around a model in which synaptic glutamate released by active neurons engages cognate metabotropic glutamate receptors (mGluRs) on astrocytes. According to this model, mGluR signalling-mediated elevations in calcium (Ca2+) in arteriole-enwrapping astrocytic end feet leads to Ca2+-dependent release of vasoactive factors onto the surface of arteriole smooth muscle cells (SMCs) and subsequent arteriole dilation (Anderson et al., 2004; Filosa et al., 2004, 2006; Girouard et al., 2010; Mulligan & MacVicar, 2004; Munoz et al., 2015; Zonta et al., 2003). Although more recent studies using genetically engineered mice have challenged this model (Bonder & McCarthy, 2014; Fiacco et al., 2007; Petravicz et al., 2008), these studies are not without their own issues (e.g. see Bazargani & Attwell, 2016), and it is likely that some variation on the neuron-to-astrocyte-to-arteriole signalling theme operates in vivo.
However, more recent studies have focused on another level of the cerebral vascular tree – the vast network of brain capillaries – ascribing to them both neuronal activity-sensing and blood flow-regulatory functions. Among the first such studies were those from Attwell and colleagues, who reported that, instead of signalling to arterioles, astrocytes translate neuronal activity into a pericyte-mediated capillary dilatory response, and thus regulate cerebral blood flow, via a mechanism that depends on an influx of extracellular Ca2+ through ATP-gated ion channels and Ca2+-dependent production and release of the arachidonic acid-derived vasodilator prostaglandin E2 (Hall et al., 2014; Mishra et al., 2016; Peppiatt et al., 2006). Our lab recently demonstrated the operation of a different capillary-centric NVC mechanism in which the focus is squarely on endothelial cells. In this study, Longden et al. showed that extracellular K+, released by neurons during action potentials, activates inwardly rectifying K+ (Kir2.1) channels in capillary endothelial cells, inducing a membrane hyperpolarization that propagates upstream via endothelial cell-to-endothelial cell signalling, ultimately reaching parenchymal arterioles and pial arteries, where it causes smooth muscle relaxation, vessel dilation, and increased blood flow to the site of signal initiation (Longden et al., 2017). These are but two of the many recent studies that have probed NVC mechanisms that operate at the capillary level (Alarcon-Martinez et al., 2020; Fernandez-Klett et al., 2010; Gonzales et al., 2020; Grubb et al., 2020; Hall et al., 2014; Harraz et al., 2022; Hartmann et al., 2021; Longden et al., 2021; Mishra et al., 2016; Rosehart et al., 2021; Sancho et al., 2022; Stefanovic et al., 2008; Thakore et al., 2021). Yet despite the proliferation of studies supporting the operation of capillary-based NVC mechanisms, even a question as straightforward as where the capillary bed begins has somehow managed to become a matter of contention.
Structural organization of the CNS microvasculature
The overall organization of the cerebral vasculature is depicted in Fig. 1A. Unless indicated otherwise, structural details are based on the mouse. Blood flows into the brain via two pathways: (1) the internal carotid arteries, which feed the anterior cerebral artery and the middle cerebral artery; and (2) vertebral arteries, which fuse to form the basilar artery and feed the posterior cerebral arteries and a number of lateral branches. These anterior and posterior circulations on the surface of the brain are connected via the Circle of Willis. Arteries coming off the Circle of Willis, collectively referred to as pial arteries, branch to form parenchymal (or penetrating) arterioles, which dive into the brain parenchyma and further branch to form a network of capillaries composed of diverging branches that ultimately converge and drain via venules (Fig. 1B) (See Schaeffer & Iadecola, 2021).
Pial arteries (diameter, typically 55−100 μm, although sometimes as small as 30 μm) (Ghanavati et al., 2014; Steinman et al., 2017; Xiong et al., 2017) consist of a single layer of endothelial cells surrounded by multiple layers of SMCs. Parenchymal arterioles, which are generally narrower than pial arteries (diameter, 15−40 μm) (Sweeney et al., 2018), are further distinguished by the presence of only a single layer of SMCs, with each adjacent SMC wrapping circumferentially around the underlying endothelial cell tube, creating a tight banding pattern. A boundary layer, termed the internal elastic lamina (IEL), separates endothelial and smooth muscle layers in both arteries and arterioles, forming a continuous fenestrated sheet that can be readily identified histologically by staining for elastin, a major component of the IEL. The end of the IEL layer marks the furthest reaches of the arterial/arteriolar circulation and the beginning of the capillary network (Grubb et al., 2020; Longden et al., 2021; Ratelade et al., 2020; Shaw et al., 2021; Shen et al., 2012). Capillaries, the smallest-bore vessels (diameter, 3−15 μm) (Sweeney et al., 2018) in the cerebrovasculature, typically branch at a steep (approximately right) angle from parenchymal arterioles, at least in the better-studied upper layers of the cortex. The first segment entering the capillary bed (1st order) diverges into two daughter branches (2nd order), each of which further diverges into daughter branches (3rd order), and so on, before converging in the same manner on draining venules. Although this branching pattern describes more than 90% of capillary junctions, trifurcations and higher-order arrangements also occur (Gould et al., 2017). A rigorous analysis of the brain microvasculature by Kleinfeld and colleagues (Ji et al., 2021) demonstrated a remarkable conservation in the branching structure of the capillary bed between mice and humans, showing that there are a total of 3−4 diverging bifurcations and 3−4 converging bifurcations linking the feeding arteriole with the draining venule in both species. Thus, on average, the shortest path between the penetrating arteriole and draining venule in brain capillary networks comprises ∼7 branches (Fig. 1B). Kirst and colleagues (Kirst et al., 2020) reported that the average capillary branch order and distance are larger to the nearest vein than artery, implying that the need for fine control of blood distribution is greater than that for collection of deoxygenated blood and metabolic waste. These two studies by Kirst et al. and Ji et al., each of which is a tour de force in its own right, used different approaches (brain clearing/antibody labelling and dye-filling calibrated by in vivo vessel imaging, respectively) to analyse vessel topology, with the latter study reporting a much higher overall vascular density. Despite these quantitative differences, these two studies provide complementary insights into structural properties of the vasculature, underscoring the importance of contextualizing properties of the whole brain using different approaches.
To our knowledge, no similarly exhaustive comparative structural analysis has been reported for the retinal vasculature – a preparation popular for studying light stimulus-induced neurovascular coupling and brain-related capillary blood flow regulatory mechanisms owing to its shared CNS developmental origin and easily accessible planar organization. Nonetheless, the general organization of the retinal circulation is well defined (Fig. 1C). It consists of three layers – superficial, intermediate and deep – with input via the central retinal artery (a branch of the ophthalmic artery) and outlet via the central retinal vein, both at the superficial level. In the superficial (i.e. most accessible) layer, the central retinal artery branches to form radiating arterioles (baseline diameter, ∼20–25 μm), from which 1st-order capillaries (baseline luminal diameter, ∼7–8 μm) branch and continue to divide into similar-size (∼5–7 μm) branches (up to ∼5th-order branches) before diving down to intersect with intermediate and deep layers. This three-tiered structure makes direct comparisons of the branching structure of the retinal capillary network with that of the brain capillary network difficult, but work by us and others (Alarcon-Martinez et al., 2020), suggests a more extended arteriole-to-venule capillary network in the retinal circulation (i.e. substantially greater than the ∼7 branches observed in brain capillary networks). Despite these differences, the general features of the retinal vasculature are structurally and functionally similar to those of the cerebral vasculature (Fig. 1).
The tubes that make up brain capillaries are formed from endothelial cells connected to each other via tight junctions, which collectively constitute the blood-brain barrier. Although endothelial cells in these capillaries express similar cell-type-defining molecular markers (e.g. CD31/PECAM-1, von Willebrand factor, VE cadherin) as those in pial arteries and penetrating arterioles, they show differences in structural arrangements. For example, our recent analysis showed that, unlike the cobblestone endothelial cell arrangement observed in pial/penetrating vessels, individual endothelial cells in capillaries are arranged in an end-to-end fashion (Longden et al., 2017) and adopt a ‘taco’-like appearance, although endothelial cells in 1st-order segments may not strictly conform to this description. The longitudinal length of a capillary endothelial cell is ∼35 μm, which is roughly equal to the length of most brain capillary segments (Longden et al., 2021).
Pericytes – a major bone of contention
CNS capillaries are not naked endothelial cell tubes. Their surfaces are variably covered by perivascular cells with a shared ‘bump on a log’ nuclear morphology and processes that wrap around or partially cover the underlying capillary segment. These cells, first described by Eberth (1871) and Rouget (1873) in the 1870s and later termed ‘pericytes’ by Zimmermann (1923), have long been a matter of controversy. They are found throughout the capillary bed but exhibit a decreasing density gradient from feeding arteriole to draining venule and show a marked preference for junctional locations (Gonzales et al., 2020; Longden et al., 2021). Shih and colleagues have proposed the terms ‘ensheathing’, ‘mesh’ and ‘thin-strand’ to describe the varying cellular morphologies of pericytes (Grant et al., 2019). Ensheathing and mesh pericytes, found at 1st- to 3rd/4th-order diverging branches of the capillary bed, can be distinguished by the patterns formed by their processes. The processes of an ensheathing pericyte wrap tightly around the underlying capillary multiple times, forming a tight banding pattern that superficially resembles the individual SMCs that surround the endothelium in arterioles. Mesh-type pericytes take their name from the pattern of their processes, which extensively cover capillary segments but do not exhibit the regular-repeating banding pattern of ensheathing pericytes. Thin-strand pericytes, which occupy more distal, generally converging, branches (4th/5th to ∼7th order in brain) of the capillary bed, typically extend long, thin processes longitudinally along the length of capillary segments. Note that, although some authors specifically reserve the term ‘capillary pericyte’ for these latter thin-strand pericytes, we consider all vessels between the feeding arteriole and draining venule to be capillaries and the overlaying perivascular mural cells to be capillary pericytes. Also, while the terms ‘ensheathing’, ‘mesh’ and ‘thin-strand’ are useful for discussion purposes, it is important to emphasize that they simply describe representative morphotypes along a continuum of pericyte morphologies.
Molecular and functional correlates of pericyte morphotypes
Molecular correlates
As pericytes transition from ensheathing to mesh type and ultimately to a thin-strand morphology, their molecular signatures and functional properties change. Ensheathing pericytes, found predominantly in the most arteriole-proximate region of the capillary network (∼1st and 2nd order), express α-smooth muscle actin (αSMA) at levels comparable to those of SMCs in penetrating arterioles based on a comparison of immunostaining intensity. Mesh-type pericytes, which bridge the morphological continuum between ensheathing and thin-strand pericytes, show variable, albeit generally reduced, immunostaining for αSMA. Thus, although αSMA is common to ensheathing and mesh pericytes, its expression is not uniform, decreasing progressively from 1st- to 3rd/4th-order capillary segments (Eltanahy et al., 2021; Gonzales et al., 2020; Nehls & Drenckhahn, 1991; Ratelade et al., 2020) and becoming virtually undetectable in pericytes associated with 5th-order and above capillary segments (Grant et al., 2019; Hartmann et al., 2021, 2022). Both of these types of pericytes, as well as thin-strand pericytes, also express smooth muscle myosin heavy chain (SMMHC) (Berthiaume et al., 2018; Gonzales et al., 2020; Hartmann et al., 2022). Like their ensheathing/mesh counterparts, thin-strand pericytes express high levels of desmin, neural/glial antigen 2 (NG2; also known as chondroitin sulphate proteoglycan 4 (Cspg4)), platelet-derived growth factor receptor-β (PDGFR-β) and CD13 – markers that are expressed at low levels in SMCs (Grant et al., 2019; Ratelade et al., 2020). Other useful markers for distinguishing pericytes from SMCs include vimentin (Bandopadhyay et al., 2001), regulator of G protein signalling 5 (RGS5) (Cho et al., 2003; Mitchell et al., 2008), and CD13/aminopeptidase N (APN) (Alliot et al., 1999). Among these markers, PDGFR-β, which is expressed in both CNS and peripheral pericytes, is widely employed as a definitive marker of pericytes (Hartmann et al., 2021; Lindahl et al., 1997; Winkler et al., 2010), although it is also expressed in some SMC, astrocyte and fibroblast populations (Armulik et al., 2011; Tong et al., 2021; Vanlandewijck et al., 2018). Strikingly, ensheathing, mesh and thin-stranded pericytes in CNS (retinal) capillaries (Gonzales et al., 2020) lack functional ryanodine receptors, a defining feature of all SMCs studied to date, suggesting that the absence of ryanodine receptor function could be another definitive pericyte marker. More recently, Atp13a5, a cation-transporting ATPase, was reported to be highly expressed in CNS pericytes (brain, retina and spinal cord), but not in SMCs or peripheral pericytes (Guo et al., 2021), although a close inspection of the data presented in this pre-print suggests that this marker may be more specific to thin-strand pericytes. These observations, taken together with the demonstrated expression of αSMA in other non-smooth muscle cell types, including myofibroblasts (granulation tissue fibroblasts) (Frangogiannis et al., 2000), activated lung fibroblasts (Zhang et al., 1996) and myoepithelial cells (Gugliotta et al., 1988), among others, argues against the view that the subset of pericytes that express αSMA are SMCs. The fact that capillary pericytes lie along a morphological continuum further complicates attempts to define a subset of these cells as a different cell type and lends an arbitrary quality to the effort.
Functional correlates
The expression of contractile proteins in ensheathing/mesh pericytes occupying 1st- to 3rd/4th-order capillary branches raises the obvious question of whether pericytes in this region are capable of dynamically contracting. In fact, it has long been assumed that this should be the case based on the expression of αSMA, SMMHC and other contraction-regulating proteins (e.g. tropomyosin) (Joyce et al., 1985) as well as ion channels (e.g. voltage-gated Ca2+ channels) (Borysova et al., 2013) important in contraction in other contractile cells. Even 19th century investigators, unincumbered by molecular insights of any kind, speculated about such a function based on the perivascular location and enwrapping morphology of these cells (Krogh, 1924; Zimmermann, 1923). Among the first to experimentally demonstrate that CNS capillary pericytes are capable of dynamically contracting in a native-like context were Attwell and colleagues, who showed that direct electrical stimulation caused Ca2+ elevation and contraction of pericytes in an intact rat retinal vasculature preparation and that bath-applied ATP or UTP constricted capillaries near pericytes in rat brain slices (Peppiatt et al., 2006). Lindauer and colleagues made similar observations in vivo, reporting that the thromboxane A2 analogue U46619 induced contraction of capillary pericytes in the mouse brain cortex (Fernandez-Klett et al., 2010). More recently, brain capillary pericytes were shown to constrict in response to amyloid beta oligomers, which were reported to act through generation of reactive oxygen species and subsequent endothelin-1 (ET-1)-dependent activation of pericyte ET-1 receptors (Nortley et al., 2019). Vascular function studies using an ex vivo retinal preparation have demonstrated contractile effects of endothelin-1 (Torring et al., 2014), ATP (Peppiatt et al., 2006), platelet-derived growth factor-B (PDGF-B) (Sakagami et al., 2001), angiotensin II (Kawamura et al., 2004) and acetylcholine (Wu et al., 2003) on pericytes. Using an ex vivo retinal preparation, our laboratory also recently confirmed rapid constriction of capillary pericytes by U46619 or a depolarizing concentration (60 mM) of K+ through activation of voltage-gated Ca2+ channels and inositol trisphosphate (IP3) receptors, further demonstrating that this response is specifically associated with pericytes in the first few (1st- to ∼4th-order) branches (Gonzales et al., 2020). This latter paper also presented a branch-by-branch analysis of the molecular properties of pericytes at capillary junctions in both ex vivo retinal preparations and the brain in vivo that clearly demonstrated a progressive decrease in αSMA expression from 1st- to ∼4th-order branches (Gonzales et al., 2020). These findings are consistent with previously presented immunostaining results (Alarcon-Martinez et al., 2018; Gonzales et al., 2020; Grant et al., 2019; Hartmann et al., 2021; Nehls & Drenckhahn, 1991; Ratelade et al., 2020), although it should be noted that some researchers have reported an abrupt decrease in αSMA expression across this region (Grant et al., 2019; Hartmann et al., 2021; Hartmann et al., 2022). Interestingly, αSMA-positivity (in a retinal preparation) was reported to depend on fixation techniques (Alarcon-Martinez et al., 2018); thus, the foregoing characterization may not be the last word on the subject of where αSMA expression ends. Although this apparent progressive decrease in αSMA might be expected to produce a graded contractile response (Hill et al., 2015), scant reports on the subject show a more digital all-or-none response, possibly reflecting difficulties accurately measuring changes in diameter in these small-bore vessels using existing optical techniques. It may also indicate that a minimum threshold concentration of αSMA is needed for contraction, although the degree of constriction depends on a number of factors beyond the amount of αSMA, including intracellular calcium, MLCK activity and, importantly, the degree of opposing pressure, which decreases throughout the capillary bed. Distal, thin-strand pericytes also contract to U46619; however, they do not respond to 60 mM K+ and their response to U46619 is much slower than that of transition zone pericytes (Gonzales et al., 2020). Thus, capillary pericytes fall into two broad categories – arteriole-proximate and distal – that regulate blood flow on fast and slow time scales, respectively, reflecting the corresponding rapid and slow kinetics of their contractile responses. In recognition of this ability of both classes of pericytes to contract, we use the term ‘dynamically contractile’ in referring to transition zone pericytes to distinguish their rapid response from the slower response of distal pericytes, which also respond to a more limited range of contractile stimuli.
A common thread running through these reports is that of progressive changes in the molecular features and morphology of CNS capillary pericytes. The structural-functional nexus of these properties is the arteriole-proximate region of the capillary bed corresponding to 1st- to ∼4th-order branches in both the brain and retinal microvasculature. As an alternative to the misleading terms that have come to be used to describe this functionally distinct entity (see What's in a name? below), we have opted for the terminology ‘post-arteriole transition zone (or region)’ – or simply ‘transition zone’ – which, although a bit ungainly, accurately positions this region within the capillary microcirculation and conveys a sense of its transitional quality. Without putting too fine a point on it, it's safe to say that that all ensheathing pericytes and some mesh-type pericytes are αSMA-positive and that the transition zone ends where αSMA expression becomes undetectable, typically at approximately 4th/5th-order branches.
The post-arteriolar transition zone: a functionally distinct region of the capillary bed that regulates capillary blood flow distribution
The post-arteriole transitional zone of the capillary network possesses an organization and functional properties consistent with a regulatory function, but what is the evidence that it actually regulates blood flow distribution and what are the underlying mechanisms? That blood flow in capillary beds is actually non-uniform, exhibiting a distribution that cannot be accounted for by the static structural properties of the network or that shows changes in response to various stimuli, has been amply demonstrated by a number of studies (Gonzales et al., 2020; Hutchinson et al., 2006; Jespersen & Ostergaard, 2012; Kleinfeld et al., 1998; Krogh, 1919; Villringer et al., 1994). Studies performed in the last decade have gone beyond observation and quantification of non-uniformity of flow within capillary beds to a consideration of cellular and molecular mechanisms.
Regulation of blood flow within the capillary network
In its most straightforward conceptualization, the role of the transition zone in blood flow regulation is to control the distribution of blood flow and/or RBC flux into and within the capillary network. There has actually been surprisingly little research that directly speaks to this function. The study that best encapsulates this regulatory perspective is one by Gonzales et al. (2020), who investigated the ability of junctional pericytes in the post-arteriole transition zone to regulate the branch-by-branch distribution of blood flow/RBC flux within the capillary bed. Using an ex vivo retinal preparation, these authors demonstrated that junctional pericytes in the transition zone rapidly contract in response to application of a depolarizing concentration of K+ (60 mM) or stimulation of Gq-coupled-protein receptors with the thromboxane A2 receptor agonist, U46619. They further showed that these contractile events are correlated with Ca2+ signalling activity in pericytes. Notably, a Pearson's correlation analysis showed that Ca2+ events in projections enwrapping different capillary branches could occur independently of each other, supporting the idea that each of these separate clusters of branch-surrounding projections is capable of acting as an independent functional unit. A similar analysis applied to the brain in vivo showed that, in cases where two daughter branches in the transition zone showed differences in baseline diameter, the frequency of Ca2+ signals in processes of the overlying junctional pericyte enwrapping the more constricted daughter branch was higher than that in pericytes surrounding the more dilated branch, consistent with pericyte-mediated, branch-specific, Ca2+-dependent contraction. These authors further showed that both IP3 receptors and L-type Ca2+ channels contribute to Ca2+-dependent contraction of transition zone pericytes. This study was the first to show that processes of transition zone junctional pericytes enwrapping different daughter branches in both retinal (ex vivo) and brain (in vivo) capillary networks are capable of independently contracting to differentially regulate blood flow (and RBC flux) to each branch, providing a mechanism for fine branch-by-branch control of blood flow and RBC distribution within the capillary bed (Gonzales et al., 2020). At about the same time, Lauritzen and colleagues (Grubb et al., 2020) identified a role for NG2-positive, αSMA-expressing mural cells (i.e. dynamically contractile pericytes) at 1st-order capillary branches in the brain cortex in regulating blood flow into the capillary bed. Specifically, they reported that, in a subset of structures examined, pericytes at the junctions of penetrating arterioles and first capillary branches in upper cortical layers of the brain were associated with ring-like indentations that they termed sphincters (Nakai et al., 1981); they further speculated that these structures served to increase plasma skimming and protect capillaries against high pressure. In work likely to have profound, but as yet ill-defined, implications for local blood flow within capillary beds, Longden et al. (2021) recently demonstrated the existence of an astounding array of Ca2+ signals in endothelial cells of the brain microvasculature that exhibit different spatiotemporal characteristics, ranging from focal events lasting a few milliseconds to cell-wide events with a duration of almost a minute. These signals, primarily reflecting intracellular Ca2+ release via IP3 receptors, are associated with Ca2+-dependent production/release of nitric oxide (NO) and relaxation of pericytes in the transition zone and thus provide an additional potential mechanism for fine branch-to-branch control of capillary blood flow.
Integration of transition zone regulatory mechanisms with NVC mechanisms
Importantly, a number of recent studies have investigated vascular dynamics in the transition zone initiated by stimuli that evoke a functional hyperaemic response, placing transition zone pericyte-mediated control of capillary blood distribution in the broader context of NVC mechanisms that regulate bulk blood flow. In the study by Gonzales et al. (2020) described above, transition zone pericyte-mediated regulatory functions were explicitly linked with our previously described capillary endothelial Kir2.1 channel-mediated, propagating-hyperpolarization (electrical) NVC mechanism (Longden et al., 2017). Specifically, Gonzales et al. demonstrated that pericytes in this region of the brain capillary bed relax in vivo in response to retrograde hyperpolarizing signals induced by focal application of K+ onto downstream capillary segments. This integration of capillary pericyte responses with K+-induced, Kir2.1-mediated propagating hyperpolarization reflects the ability of capillary endothelial cells to communicate electrically with pericytes through cell-cell junctions. In a similar vein, Lauritzen and coworkers (Cai et al., 2018) demonstrated that activation of the somatosensory cortex by whisker stimulation evoked a dilatory response in capillaries that began at the transition zone, specifically, 1st- and 2nd-order branches, including ‘sphincters’ (Grubb et al., 2020). Interestingly, they noted that this dilatory signal spread both up- and downstream, albeit rather slowly (5–20 μm/s), and suggested a role for the gliotransmitter ATP and P2 purinergic receptors in this process. In a subsequent paper (Zambach et al., 2021), these researchers extended their branch-order analysis, showing that the dilatory response of pericyte-occupied 1st-order capillary branches to whisker pad stimulation was comparable to that of penetrating arterioles, but was decreased at 3rd-order branches in parallel with decreased pericyte expression of α-SMA. They further showed that these changes in arteriole and capillary diameter were associated with decreases in Ca2+ signalling in SMCs and pericytes, respectively. Using pharmacological approaches, they also identified potential chemical mediators of changes in capillary diameter, showing that endothelial nitric oxide (eNOS)/NO signalling contributes to the basal tone of transition zone capillary segments (1st- and 2nd-order branches), and that KATP channel activation, but not eNOS/NO signalling, is involved in mediating dilatory responses to whisker stimulation in this region (sphincter/1st- to 3rd-order branches). Kornfield & Newman (2014) reported similar differential dilatory responses as a function of capillary branch order in the retina, showing that a flickering light stimulus evoked rapid dilatory responses that were centred on the central retinal artery and first two branches of the capillary bed (called ‘arterioles’ by Kornfeld and Newman). In an earlier study, Lindauer and colleagues (Fernandez-Klett et al., 2010) investigated the role of pericytes in dynamic changes in brain capillary diameter and blood flow, reporting that only ensheathing pericytes in the post-arteriolar transition zone (reported as pre-capillary arterioles), and not thin-strand pericytes associated with more distal capillaries (reported simply as ‘capillaries’), regulated CBF in vivo following sensory stimulation. Hill et al. (2015) subsequently reported similar findings, showing that Ca2+ transients in perivascular mural cells, including ensheathing pericytes (reported as precapillary SMCs) and arteriolar SMCs, but not thin-strand pericytes, were correlated with changes in vessel diameter under normal conditions. Whether these reports of the involvement of the transition zone in hyperaemic responses point to integration with our previously reported Kir-based electrical mechanism or instead reflect the involvement of other NVC mechanisms (e.g. glial/astrocyte-mediated release of vasodilators directly onto capillary endothelial cells and/or pericytes) remains untested. Similarly, but with a twist, Earley and colleagues (Thakore et al., 2021) recently suggested a novel role for the transition zone in capillary-based NVC, proposing that this region is an integral element in mediating large-scale increases in blood flow in response to neuronal stimulation. Specifically, they posited that slowly propagating Ca2+ signals mediated by the TRPA1 (transient receptor potential ankyrin 1) channel, initiated at more distal reaches of the capillary bed in response to neuronal activity, are converted to electrical signals in the transition zone and then spread rapidly upstream to induce arteriolar dilatation. Intriguingly, despite differences in nomenclature and some details, a 2012 review by Itoh and Suzuki anticipated much of the foregoing (Itoh & Suzuki, 2012). In it, they proposed a ‘proximal integration model’ that postulated that the arteriole-proximate region (i.e. transition zone) possessed features consistent with a regulatory function and speculated that capillaries directly sensed neuronal activity and communicated this upstream to the proximal region and on to parenchymal arterioles and pial arteries.
Few studies have investigated the potential involvement of transition zone pericytes in a pathological setting, but the recent work by Lauritzen and colleagues referred to above (Zambach et al., 2021) implicated pericyte endothelin 1 (ET-1) receptors in the contractile response of transition zone capillary segments (sphincter/1st- and 2nd-order branches) to global ischaemia induced by cardiac arrest. Hill et al. (2015), using multiple tools (e.g. optogenetic and whisker stimulation) and different pathological conditions (spreading depolarization, transient cerebral ischaemia), also reported that, in addition to their role under normal conditions, ensheathing pericytes and arteriolar SMCs are critical regulators of CBF under ischaemic conditions. Transition zone capillary pericytes can also contribute to pathological outcomes if they become hypercontractile. In a recent paper with significant clinical implications, Joutel and colleagues (Ratelade et al., 2020) reported that enhanced contractility (‘hypermuscularization’) of pericytes in the transition zone, together with upstream arteriolar SMC loss, causes spontaneous intracerebral haemorrhage, a devastating type of stroke.
What's in a name? – the hazards of nomenclature
Collectively, these observations reveal a growing consensus around the idea that the post-arteriole transition zone of brain and retinal capillary beds containing dynamically contractile pericytes is a functionally distinct entity involved in regulating capillary blood flow. However, differences in naming conventions used by different research groups have managed to obscure this general consensus. The convention that has done the most damage in this regard is the unfortunate descriptor, ‘pre-capillary arteriole’, which once meant what it sounds like – the feeding arteriole that precedes the capillary bed – but has morphed into something entirely different. As applied in the current context, both elements of this terminology have a tenuous relationship with near-universally accepted physiological reality: (1) The vessels in this region are not arterioles: they have a smaller diameter range than arterioles, lack an IEL, and are not covered by a single continuous layer of individual SMCs. (2) These vessels do not precede the capillary bed; they are an integral part of it. In fact, from a numerical standpoint, these vessels in the brain represent ∼50% of the capillary segments along the most direct route between the feeding arteriole and draining venule – the textbook definition of a capillary bed. Right behind this in terms of consensus undermining is the argument that the mural cells in this region are SMCs and not pericytes. On some level this argument may reflect the predisposition of researchers to approach this question as ‘lumpers’ versus ‘splitters’, but it over-applies certain defining criteria while under-applying others. While it may be prudent to caution against clinging too stridently to the view that these cells are wholly distinct from SMCs, given their shared neural crest cell origin and the absence of definitive fate-mapping experiments, for the moment the case for drawing a clear distinction between SMCs and pericytes is much stronger than that for lumping a subset of pericytes occupying a continuum of morphotypes into the SMC category.
In the final analysis, the primary victims in this nomenclature tug-of-war are uninitiated readers, who can be forgiven for giving up in the face of the seemingly contradictory assertions that populate the literature. The obvious solution would be for researchers in different labs to get on the same page regarding what to call things. Barring that, authors should at least let readers know what colour the sky is in their world by explicitly defining the terminology they use.
Parting thoughts
Accumulating evidence has established an important role for the post-arteriole transition zone (by any name) in regulating the distribution of blood within capillary networks. The evolutionary logic behind positioning a dynamic regulatory region at the entryway into a network of diverging branches is clear. After all, ‘decisions’ made in the first few branches disproportionately affect blood flow through the network; thus, this is where the action is. It is also clear that the continuum of pericyte morphotypes – from ensheathing/mesh to thin-stranded – supports this logic, with the former mediating relatively rapid, dynamic contractile responses and the latter, despite their absence of αSMA, managing to influence blood flow, albeit with much slower kinetics. But what are the developmental processes that create this continuum? How do the different types of pericytes come to occupy their positions in the capillary network to establish the observed morphological and functional continuum? Do signals received from the underlying endothelium provide guidance and developmental cues? Beyond such a possible developmental role, there is growing evidence that the endothelium, which has largely been ignored in this review, is a major contributor to the blood flow regulatory properties of the post-arteriole transition zone. For example, the work of Earley and colleagues (Thakore et al., 2021) demonstrating that slowly propagating Ca2+ signals in distal reaches of the capillary network are converted to rapid electrical signals in the transition zone implies that these arteriole-proximate endothelial cells possess functional attributes lacking in endothelial cells occupying more distal branches. In addition, endothelial cells and pericytes clearly work in synchrony with respect to chemical mediators such as NO, but the potential implications of electrical communication between endothelial cells and pericytes have only begun to be investigated. Signals initiated by sensorimotor or visual stimuli that induce functional hyperaemic responses – whether via Kir2.1 channel-dependent electrical signalling or other incompletely defined mechanisms – have been shown to converge on transition zone pericytes to cause these latter cells to dilate and facilitate blood flow along a specific path, but how these signals are communicated from endothelial cells to pericytes remains unclear. Also unclear is whether endothelial cell-pericyte communication is bidirectional. Ultimately, the regulatory properties of the transition zone serve the needs of neurons, so it will be important to integrate the body of research relating the activity of individual neurons (or groups of neurons) to local changes in blood flow – the ‘how many neurons does it take to initiate a hyperaemic response’ question – with ‘decisions’ made in the transition zone.
Biographies
Amreen Mughal, Mark Nelson and David Hill-Eubanks are part of a group at the University of Vermont studying mechanisms that regulate blood flow in the brain, particularly those involving vascular endothelial and mural cell contributions to neurovascular coupling mechanisms linking changes in blood flow to spatiotemporally varying neuronal energy demands. Amreen Mughal (Assistant Professor) is an early-stage scientist whose personal research interests extend to investigations of early functional deficits in neurovascular coupling in Alzheimer's disease. Mark Nelson (Distinguished University Professor, Chair of the Department of Pharmacology, member of the National Academy of Sciences) has overseen an illustrious research programme that has continued to maintain itself at the leading edge of the cardiovascular field. David Hill-Eubanks (Faculty Scientist), resident writer/editor and 20+-year veteran of the Nelson group, brings his eclectic scientific background to the enviable task of helping convey Nelson lab research to the scientific community. Photos (from left to right): Drs Amreen Mughal, Mark Nelson, and David Hill-Eubanks.
References
Additional information
Competing interests
The authors declare no competing financial interests.
Author contributions
D.H.E. and A.M. wrote the manuscript; M.T.N. provided scientific and editorial input. All authors have approved the final version of the manuscript and agree to be accountable for all aspects of the work. All persons designated as authors qualify for authorship, and all those who qualify for authorship are listed.
Funding
This study was supported by an American Heart Association postdoctoral fellowship (20POST35210155 to A.M.), an American Heart Association Career Development Award (856791 to A.M.), the Totman Medical Research Trust (to M.T.N.), EC Horizon 2020 (to M.T.N.), and the National Institutes of Health (1K99AG075175 to A.M., R35HL140027, R01NS110656, RF1NS128963, and P20GM135007 to M.T.N.)
Acknowledgements
We thank Drs Anne Joutel, Thomas Longden and Nicholas Klug for their comments and discussion.
