Bridging regional neurovascular unit heterogeneity and cognitive function: a review

Abstract

The blood-brain barrier (BBB) is a critical central nervous system interface that tightly regulates the exchange of substances between the blood circulation and the brain parenchyma. The BBB, glia, mural cells and neurons form the neurovascular unit (NVU), which modulates cerebral homeostasis. Traditionally considered a uniform and selective barrier, emerging research has unveiled significant heterogeneity in BBB properties across various brain structures. This review synthesizes current knowledge of the heterogeneity of the NVU elements of different brain structures. The structural and functional differences of the endothelial cells, glial cells, and neurons of different brain structures are pinpointed. Finally, BBB heterogeneity and its link to cognition in physiology and pathology are discussed. Understanding such relationships could help effectively target and cure neurological diseases.

Introduction

The normal functioning of the central nervous system (CNS) requires a highly controlled microenvironment. The latter is provided by the blood-brain barrier (BBB), a highly selective semipermeable border that separates circulating blood and extracellular fluid in the CNS from the brain [1]. The BBB is a dynamic interface comprised of brain capillaries formed by specialized endothelial cells (ECs) that coexist and intricately communicate with mural cells, glia and neurons. Together, these components form the neurovascular unit (NVU) [2]. Even though ECs form the structural backbone of the BBB, it is important to understand that all the cellular components of the NVU constantly interact, contributing to their respective tasks and maintaining the final function of the CNS, cognitive function integrity [3, 4].

Several areas of the CNS, known as circumventricular organs, including the subfornical organ, the area postrema, the vascular organ of lamina terminalis, the median eminence, the pituitary neural lobe, and the pineal gland, feature fenestrated capillaries that do not exhibit BBB properties. This unique vascular permeability facilitates the exchange of sensory or secretory signalling molecules between the brain and the bloodstream, supporting the regulation of body homeostasis through circumventricular organs. In the rest of the CNS, the BBB has traditionally been seen as a uniform interface across all brain regions. This perception likely stemmed from the initial discovery of the BBB by Paul Ehrlich (1885) and Edwin Goldman (1913), who demonstrated its protective function but did not have the tools or data to explore regional differences [5, 6]. The concept of uniformity persisted for much of the 20th century due to limited imaging and molecular techniques, which made it challenging to investigate subtle variations in the BBB’s structure and functions across different brain regions. It was only with the advent of modern techniques like RNA sequencing, advanced microscopy, and molecular profiling that scientists began to uncover the BBB’s heterogeneity [7–9]. BBB heterogeneity refers to the variation in the structural and functional properties of the BBB across different brain regions. First, BBB heterogeneity has been considered in terms of ECs. However, the heterogeneity may rely also on differences in the morphology, gene and protein expression, properties, and functions of the other cellular components of the NVU (glia, mural cells, neurons) [7, 10]. The intricate properties that characterize the BBB are frequently disrupted in various disease states, with BBB dysfunction recognized as a key factor in numerous neurological disorders. Many CNS diseases affect to specific brain regions. Consequently, changes in cellular signalling associated with these diseases may be influenced by variations in the local composition or interactions of NVU cells. Given the interplay between the ECs of the BBB and the neurons in the context of the NVU, heterogeneity may play a role in the specialized function of a specific brain region. Consequently, it may explain why some regions are more susceptible to disease and could imply a region-dependent modulation of cognitive functions.

In this review, recent evidence will be highlighted showing the differences between the various NVU cell types within different brain regions such as the cortex, hippocampus, striatum and cerebellum and their involvement in specific cognitive functions. In more detail, the BBB heterogeneity of the cerebral elements including ECs, glial cells, mural cells and neurons will be discussed. Their morphological and functional differences will be pinpointed based on their location in the different structures, but also within the regions of the same structure. Finally, the influence of the different BBB components on cognition in physiology and in pathology across the brain will be addressed.

There are gaps in the literature concerning NVU heterogeneity throughout the different brain regions and its influence on behavioral and cognitive tasks. By addressing these shortcomings, we aim in this review to integrate cellular, vascular, and cognitive neuroscience to provide a clearer picture of the way NVU heterogeneity supports cognition and how its dysfunction contributes to diseases.

Based on location, brain cell heterogeneity influences the structural and functional cerebral endothelium

The cerebral vascular architecture varies significantly between regions, reflecting the unique metabolic and functional demands of each area. The comparison of three of the NVU elements between the white matter and the grey matter has already been reviewed by Bernier and collaborators (Bernier et al. 2021). They describe the molecular and therefore functional differences of astrocytes, mural cells and ECs in the white versus the grey matter highlighting useful tools to study the BBB heterogeneity throughout the brain. Although the cited work effectively highlights the differences in NVU elements between white and grey matter, there remains a pressing need to explore and compare BBB heterogeneity across a broader range of brain regions.

Brain vascular network and endothelial cells

The primary roles of the NVU are to preserve ionic homeostasis, assure the brain’s nutrition, and protect the neurons. To do so, all the NVU components have a crucial role. To begin with, the ECs are the component forming the brain blood vessel walls.

Until recently, the vasculature has been viewed as a homogeneous entity throughout the brain. Nevertheless, new research has revealed differences in the vascularization of different brain regions, but also within the same region [11–13]. To start with there are density differences between the capillaries in different cerebral areas. Interesting research shows vascularization differences in the hippocampus compared to the cortex, demonstrated by a quantitative confocal microscopy study. Important BBB vascularization differences have been identified not only between the cortex and the hippocampus, but also between neighbouring CA1 and CA3 hippocampal regions. Within the hippocampus, the CA1 region presents lower capillary density compared to the CA3 (Table 1). This can be justified by the anatomical differences, as the CA1 area is perfused by one arterial branch, the ventral artery, whereas the CA3 is fed by two major vascular afferent dorsal arteries [11]. Two older studies have demonstrated intra-regional heterogeneity in the vascularization of the human cortex depending on the depth, noting a lack of capillarity in the superficial (pial) layer [14, 15]. This information was verified by another study conducted on adult chinchillas that not only proved the same thing but also showed a heterogenous capillary bed distribution throughout the temporal auditory cortex [12]. The role of the capillary beds is to enable the exchange of nutrients between the blood and the brain. This could suggest that the deeper the brain region the greater the need it has for nutrients. The striatum is another brain region that is not uniformly vascularized and has different compartments [16] (Table 1). These distinct internal compartments are called the striosomes and the extrastriosomal matrix, which can be characterized by their differing immunoreactivity patterns for substances related to neurotransmission. A study on human brains showed differences in both the number and the density of the capillaries within the striatal compartments [16]. Furthermore, the vascular density in the striatal matrix is increased compared to the striosomes [16]. In contrast, a previous study highlighted an important relationship between capillaries and striosomes compared to a less vascularized matrix in the striatum of rats. This may be highlighted by interspecies differences between humans and rodents.

Table 1.

Cellular characteristics in cerebellum, striatum, hippocampus and cortex, underlying the regional heterogeneity of the neurovascular unit

	Endothelial cells	Astrocytes	Microglia	Pericytes	Neurovascular coupling
Cerebellum		Fibrous astrocytes [43]	Lower cluster 1 gene expression compared to cortex [62] Bigger soma and higher ramifications [49, 50]	Lower BBB permeability compared to cortex, striatum and hippocampus [35]	Non-linear neurovascular coupling frequency [66]
Striatum	Different number and density of capillaries within the striatal compartments [16]	High expression of GABA metabolizing enzyme [46]	Smaller soma and fewer ramifications compared to cerebellum [49, 50]	Higher BBB permeability compared to cerebellum [35]	Decreased neurovascular coupling compared to cortex [65]
Hippocampus	Fragile vascularization system [23] Higher permeability compared to other brain structures [25] CA1 region presents lower capillary density compared to CA3 [11]	Gene expression predominance of AST1 & AST4 [44] High expression of glutamate metabolizing enzyme [46]	High fibre tract density and more populated than the cerebellum [49] Smaller soma and fewer ramifications compared to cerebellum [49, 50] CA1: higher density than in CA3 [56]	Higher BBB permeability compared to cerebellum [35]	Decreased neurovascular coupling compared to cortex [37]
Cortex	Variation in the amount of capillaries throughout the cortex [13] Different biomarkers in each cortical area [13]	Protoplasmic astrocytes [42] Gene expression predominance of AST2 [44] Different expression of glutamate uptake transporter level compared to the hippocampus [44]	Higher cluster 1 gene expression compared to the cerebellum [62] Average fibre tract density [49] Smaller soma and fewer ramifications compared to cerebellum [49, 50]	Layers II/III have fewer pericytes than layer I [33] Higher BBB permeability compared to cerebellum [35]	Increased neurovascular coupling compared to hippocampus [37] and striatum [65]

In human research using postmortem cortical tissues, variation was noted in the number of capillaries throughout five cortical regions (entorhinal cortex, inferior temporal gyrus, prefrontal cortex, visual association cortex, and primary visual cortex) (Table 1). The primary visual cortex presented more capillaries than the other four regions [13]. This data could suggest that the primary visual cortex needs more nutrients and oxygen compared to the other regions so the vascular network is locally denser.

In addition to the heterogeneous vascular network inside the brain, vasculature heterogeneity is also present on a molecular level in the ECs as it seems that there is a genetic diversity throughout the various cerebral regions.

The brain ECs regulate all necessary nutrient entries from the blood to the brain parenchyma [17]. The brain ECs also modulate the paracellular and transcellular trafficking of other molecules, ions, toxins and drugs. This regulation is provided by tight junctions (TJs) that firmly seal neighbouring ECs, low rates of transcytosis, expression of a wide variety of transporters, and expression of metabolic enzymes [1, 18]. This specific phenotype allows maintenance of a stable microenvironment for optimal brain function and protection from harmful agents present in the bloodstream. To extend understanding of the roles of TJs, research has provided evidence showing differences in interindividual and regional variability in transporter abundance in humans between the occipital and the parietal grey matter [19]. These differences could potentially imply unequal drug distribution throughout the brain, but more studies are needed in order to confirm this hypothesis. Another interesting aspect of BBB heterogeneity is the enzyme distribution throughout the different brain structures (whole human brain, cerebellum, dura mater, and 16 other brain regions.). One study of the xenobiotic-metabolizing enzymes and transporters in the human brain demonstrated an uneven distribution of major cytochrome enzyme transcripts (CYP1B1, CYP2D6, CYP2E1, CYP2J2, CYP2U1, and CYP46A1) across brain regions [20]. Two other teams have described differences in enzyme distribution through different cerebral structures influencing neurotransmission and physiological processes, but this research, being old, should be reconfirmed by modern science [21, 22].

A cerebral regional specificity of gene expression has been demonstrated that shows distinct areas exhibit unique molecular signatures that influence their functional properties. In the study of Bryant et al. it was highlighted that each of the five regions (entorhinal cortex, inferior temporal gyrus, prefrontal cortex, visual association cortex, and primary visual cortex) had specific biomarkers that were upregulated compared to the other four. As a result, for example in the visual association cortex, the upregulation of marker genes such as RNASE1, ICAM2, and SLC2A1 enhances cytoplasmic translation and vasculogenesis, but also decreases cell migration and in this specific structure (13). In addition, a recent study compared the gene expression of multiple endothelial junctional proteins, ABC transporters and BBB receptors of mouse ECs in the hippocampus vs. the cortex that present heterogenous expression profiles. These findings suggest that permeability is due to the tightness of the junctions, but also expressions of transporters vary inter-regionally. Consequently, drug delivery is not uniform throughout the brain structures (7).

Interestingly, the vascular system of the hippocampus presents an inherent fragility compared to other regions [23, 24] (Table 1). In a review, Davidson and Stevenson explain that in cerebral insults the hippocampus is generally one of the first regions to be impaired. The studies mentioned in this paper indicate that for some of these pathologies the hippocampal BBB is the first to be affected compared to other regions, which could explain the vulnerability of the hippocampus [24]. For this reason, in humans under anoxic conditions or hypoperfusion, the BBB of the hippocampus is less capable of handling the reduced blood flow compared to other regions [23]. This data correlates with a human study proving a higher permeability of the hippocampal BBB compared to different brain areas using dynamic contrast-enhanced magnetic resonance imaging (MRI) (Ivanidze et al., 2019) (Table 1). This suggests that regions like the hippocampus that need a high metabolic rate to correctly function in the case of disease tend to be more vulnerable, as the BBB does not meet their needs, compared to other brain regions that can function with lower metabolic rates.

Pericytes

Pericytes are mural cells that are wrapped around the cerebral capillaries. In the past, it was believed that they were only implicated in hemodynamic regulation due to their contractile nature and their ability to secrete vasoactive agents. However, their functions extend beyond contractility to include supporting BBB integrity, angiogenesis, and intercellular signalling within the NVU [26].

In general, based on their structure and location, there are three cerebral pericyte categories: ensheathing, mesh, and thin-strand pericytes [27, 28]. This morphological diversity could imply a heterogenous regulation of BBB activity depending on the pericyte type of each brain area. The localization of the different pericyte types differs. The ensheathing pericytes are found around pre-capillary arterioles, the mesh pericytes around true capillaries and lastly the thin-strand pericytes around post-capillary venules [29]. There is a lack of research on the precise localization of each type of pericyte across the different cerebral structures, as the majority of the studies explaining pericyte morphology are limited to the cortex [29–31].

The density of pericytes differs intra-regionally. The upper cortical layers present an increased pericyte density, compared to the cell-body-rich layers, where the density is decreased to regulate vessel relaxation [32]. The study of Hartmann et al. proves that, in mice, the cortical layer I has around 40% more pericytes compared to layers II and III [33] (Table 1).

Based on the current literature, two of the brain areas with high pericyte coverage are the cortex and the hippocampus [34]. Furthermore, a similar model of pericyte deficiency revealed regional heterogeneity in BBB permeability following pericyte loss [35]. The cortex, striatum, and hippocampus presented a higher BBB permeability, while regions such as the cerebellum and midbrain exhibited less pronounced effects [35] (Table 1). In the paraventricular nucleus, the density of the endothelium is 3–5 times higher compared to the cortex, and the pericytes engulfing vessels are also quite elevated [36]. Another study highlights differences in pericyte coverage between the hippocampus and visual cortex [37] (Table 1). Some studies have also shown that pericytes release vasoactive agents in response to synaptic transmission [38]. Heterogeneity in pericyte coverage further contributes to the differential barrier properties, such as permeability and transport dynamics, which are tailored to the specific functional and metabolic demands of different brain structures. This highlights the nuanced and region-specific roles of pericytes in maintaining BBB integrity and facilitating neurovascular coupling.

Altogether, these observations suggest that pericyte density may correlate with the specialized functions and demands of different brain regions. More research in the field could help develop efficient strategies to treat diseases related to BBB breakdown caused by pericyte loss, such as vascular dementia and Alzheimer’s disease [39].

Astrocytes

Astrocytes are star-shaped cells that enwrap the brain capillaries with subcellular structures called endfeet that help maintain ionic homeostasis, provide metabolic support and regulate neurotransmitters [40, 41]. The heterogeneity of the astrocytes starts with differences in morphology, which varies depending on the brain region in mammals. The first category is the protoplasmic astrocytes that are located in the cortex [42] (Table 1). The other subtype is named fibrous astrocytes and these are found in the cerebellum [43] (Table 1). In their review, Lee et al. explain in detail the differences in astrocyte morphology and gene expression within the same brain region and between different regions such as the cortex, hippocampus and cerebellum [43]. This phenotypic astrocyte heterogeneity could have different impacts on overall BBB function. Given the key roles astrocytes play in helping maintain BBB homeostasis, their diversity manifests in region-specific characteristics that could modulate BBB integrity and signalling.

Other evidence supports morphological and gene expression differences between the hippocampal and cortical astrocytes using single-cell RNA sequencing in mice [44]. A detailed roadmap of the morphological and physiological astrocyte subtype (AST) differences was obtained. For example, they demonstrated a predominance of AST1 and AST4 in the hippocampus and of AST2 in the cortex [44] (Table 1). In the same study, it was found that the expression of solute transporters varies depending on the AST. This means that neurotransmitters such as gamma-aminobutyric acid (GABA) and glutamate are differently metabolized across the brain. More specifically, RNA sequencing and spatial transcriptomics have proved that a certain AST specialized in GABA uptake highly expresses Slc6a1 and Slc6a11, but expresses glutamate uptake transporters Slc1a2 and Slc1a3 less than other ASTs [44] (Table 1). Similarly, a different transcriptomic study highlighted higher levels of Slc1a2 and Slc6a11 in the cortical astrocytes compared to lower expression of Slc1a3 in the cerebellum [45]. This suggests that the heterogeneity of astrocytes throughout the brain tailors them to meet the functional demands of their local environment and to regulate BBB functions non-uniformly. There must be region-specific neurotransmitter regulation in order to ensure precise control of inhibitory GABA signalling, which is critical for local circuit balance. It would be interesting to study in more depth the interaction between astrocytes and ECs in a region-specific way and to identify the dysregulations in neurodegeneration and other neurological diseases to ultimately pave the way for targeted and region-specific therapeutic strategies.

Other data show that astrocytes present circuit-specialized properties. A study in mice suggests that distinctions exist between astrocytes in the hippocampus and striatum, allowing them to specialize in the specific functions of regional neuronal circuits [46]. In more detail, the gene Aldh5a1, which encodes the GABA metabolizing enzyme, is expressed more in striatal astrocytes than the gene Glul, which encodes the glutamate metabolizing enzyme and is expressed more in hippocampal astrocytes [46]. It would be interesting to regionally correlate the different ASTs to determine whether circuit activity corresponds to BBB regulation and to compare the differences based on the ECs of each brain region.

Microglia

Microglia are the immune cells of the brain that play a role in maintaining BBB integrity. They secrete signalling factors that influence the expression of TJ proteins of the BBB. During systemic inflammation there is a migration of microglia to brain blood vessels in response to chemokine release by ECs in order to preserve BBB integrity [47].

Fixed and real-time imaging techniques have allowed the morphological classification of microglia as ramified, primed, reactive, or amoeboid [48]. In most brain regions, microglia present normal ramifications with extended branches that can differ significantly from region to region or within different areas of the same region [49–53]. For example, striatal, hippocampal and cortical microglia present smaller areas of soma and cytoplasm, with a less ramified complex and covered area than those of the cerebellum [49, 50] (Table 1). Moreover, within the different cortical layers, microglia present different sizes and ramification patterns [52, 54].

Several studies have highlighted different microglial densities across the CNS. A first study used lipocortin 1 antibody and revealed the lowest microglial density in the brainstem and the cerebellum, a higher density in the midbrain, and the highest in the forebrain [55]. Another study, in a mouse model, found a low density in the fibre tracts, cerebellum, and brainstem, an average density in the cerebral cortex and a higher density in the hippocampus, basal ganglia and substantia nigra [49] (Table 1). This could suggest that the brain’s most vulnerable regions, such as the hippocampus, need more protection as it is known that the BBB is locally more fragile compared to other areas. Nevertheless, other evidence showed a lower microglial density in the cerebellum and a higher one in the frontal cortex by using ionized calcium-binding adapter molecule-1 [50, 51]. These morphological differences could imply region-specific microglial specialization and BBB maintenance, or even localized regulation of the immune response to inflammation. They could influence how effective protection of the BBB is in response to various challenges and pathological conditions.

Apart from the inter-regional differences in microglia density, there are also intra-regional variations. For example, in mice, the hippocampus presents a higher density within the CA1 and the dentate gyrus area compared to the CA3 region, and there are no differences in the dorsal hippocampus [56] (Table 1). The reason for these local density variations is still not clear. It is interesting that the different densities of the microglia throughout the CNS are in accordance with the glia-to-neuron ratio. Humans present a cortical neuron-to-brain ratio of around 3.72 (72% glia), compared to a ratio of about 0.23 (19% glia) in the cerebellum. This data suggests a coordination between glial migration and proliferation [57]. Furthermore, considering the microglia-neuron interactions, a coordination of the EC responses could be possible and may vary regionally depending on microglial phenotype and density.

A study on the mouse somatosensory cortex unveils microglial transcriptional sub-states that are cortical layer-specific. This suggests that neuronal diversity drives the microglial states by potential neurotransmitters and signalling [58]. Even though there is transcriptional homogeneity, microglia present heterogeneity across the different CNS regions concerning their morphology, density, and epigenome, and this is implicated in their lysosomal function and debris clearance role [59–61]. It seems that microglial identification of different brain areas shows a link between the identity of microglia and the neuronal environment around them.

Microglia present patterns of gene co-expression with region-dependent phenotypes [62]. Research has unveiled different gene cluster expression profiles. For instance, one of these gene clusters is expressed more in the cortical microglia than in cerebellar microglia [62].

More studies are needed to shed light on BBB regulation by the microglia across the different brain structures and on the way microglia respond to disease locally. It would be quite interesting to understand the way microglial phenotypes vary spatially and influence BBB properties across the brain. This could be done using techniques such as spatial transcriptomics to correlate microglial phenotypes with regional BBB transporter expression, TJ proteins, and vascular markers.

Neurovascular coupling

The heterogeneity in BBB permeability directly influences the neuronal microenvironment and vice-versa. An increase in neuronal electrical activity affects BBB function, which quickly responds to neuronal demands by rapidly increasing blood flow. Neuronal function affects BBB activity differently across brain regions causing variation in nutrient and metabolite availability, as well as exposure to circulating neuroactive substances. This is described by the neurovascular coupling that leads to the transfer of oxygen and nutrients to brain-activated areas [63]. Even though the intracellular pathways regulated by neurovascular coupling are not clear, it has been shown that glutamate induces the release of several arachidonic acid derivatives and ATP, which have an effect on the microcirculation [64]. In addition, a study in rats using fMRI and electrophysiology revealed further regional differences in neurovascular coupling [65]. The aim was to understand if a specific glutamate receptor presents regional differences in neurovascular coupling. This work underscores differences in the fMRI signal, with an 18% decrease of the signal in the primary cortex compared to a 66% reduction in the striatum [65] (Table 1). On the other hand, a recent MRI study found that the mouse cerebellum has a non-linear neurovascular coupling frequency [66] (Table 1).

Even though a lot of research correlates the increase of neuronal activity with an increase in blood flow, other work proves otherwise. There are several areas and conditions in which neuronal activity does not reflect blood flow changes, meaning that the various brain regions present neurovascular coupling differences [67].

Cognitive functions in physiology

The brain is considered as the central organ of cognition as it is characterized by extraordinary regional specialization. Cognitive functions encompass executive abilities, sensory integration, language, and memory processes. They are facilitated by the neuronal network and the NVU elements that interact with each other in order to keep the brain to its homeostatic conditions.

In general, each cognitive function is mainly associated with the activation of a distinct brain region. The frontal lobe is at the forefront of cognitive control and executive function. This region is in charge of executive functions such as working memory, decision-making, problem-solving, and inhibition of inappropriate behaviours [68]. Other roles of the frontal lobe include motor coordination, personality and emotion regulation, mental flexibility and speech production [68]. Furthermore, the parietal lobe is considered as the centre of sensory information such as somatosensory processing and spatial cognition in order to integrate sensory input to create a spatial map for navigation and reasoning [69]. A third lobe is the occipital lobe, which is in control of vision, reading and recognition [70]. In addition, the temporal lobe supports auditory processing and memory retrieval with language understanding [71]. In this area, the hippocampus, in particular, is known to play a crucial role in the consolidation of memory processes. Lastly, the cerebellum manages balance and movement [72]. Numerous studies have also shown its involvement in several cognitive functions such as social cognition, some specific learnings and spatial navigation [73].

Even though there is data explaining the regional heterogeneity of the BBB and the NVU in terms of transporter expression, TJ complexity, and metabolic enzyme activity, no studies have directly investigated the link between cognitive functions and the heterogeneity of the NVU in physiology. This knowledge gap raises critical questions about the potential role of BBB regional variability in shaping cognitive processes and their dysfunction (Fig. 1). On the other hand, in pathological conditions such as neurodegenerative diseases, brain tumours, and cerebrovascular disorders, BBB dysfunction is closely linked to cognitive decline. This suggests that BBB heterogeneity may influence regional brain vulnerability in disease states.

Fig. 1.

Schematic representation of the known neurobehavioral roles of the striatum, cortex, hippocampus and cerebellum as well as described heterogeneity in the NVU in these regions. Further research will be needed to establish the link between NVU regional specificity and neurobehavior

The link between NVU elements and cognition in pathology

NVU alterations have been described in many pathological contexts, including neurodegenerative diseases, brain injuries, epilepsy, and depression. A review by Yuan et al. summarizes the way brain endothelium disruption leads to neurodegeneration and cognitive impairment [86].

Alzheimer’s disease

Alzheimer’s disease (AD) is a neurodegenative disease causing memory loss and cognitive impairment [87]. In a human study research showed a correlation between BBB permability and dementia in a very large cohort of patients (1015 participants) [78].

In AD, the TJs of the brain endothelial cells of the cortex are disrupted leading to elevated BBB permeability [88] (Table 2). These alterations lead to cognitive impairments such as memory loss, and impaired judgement and problem-solving ability [75] (Table 2).

Table 2.

Diseases related to the regional heterogeneity of the neurovascular unit

Disease	Region mainly affected	Described BBB alteration	Described cognitive dysfunction	References
Alzheimer’s disease	Hippocampus, frontal cortex	Disruption of tight junctions, increased permeability	Memory loss, impaired judgment, reduced problem-solving abilities	[74–76]
Parkinson’s disease	Basal ganglia, frontal cortex, substantia nigra	Altered transporter expression, microvascular damage	Impaired decision-making, reduced cognitive flexibility	[77–79]
Stroke	Cortical and subcortical regions	Focal BBB breakdown, vasogenic oedema	Language deficits (aphasia), motor impairments, memory issues	[80]
Epilepsy	Temporal lobe, hippocampus	Regional BBB opening, inflammation	Memory impairment, executive dysfunction	[81, 82]
Depression	Prefrontal cortex, hippocampus	Altered transporter function, reduced BBB integrity	Impaired memory, emotional dysregulation	[83–85]

Yamazaki et al. also proved that loss of claudin-5 and occludin occurs predominantly in the neocortex during the progression of AD [88]. This implies that the BBB is actively involved in the memory process as the breakdown of its TJs could leads to memory decline. It could also suggest that restoration of TJ integrity could be a potential therapeutic strategy to slow memory decline and cognitive impairment in AD. Research conducted on BBB permeability in regions such as the medial temporal lobe, hippocampus, parahippocampal gyrus, and caudate nucleus has revealed compromised episodic memory and the potential to act as early indicators of AD and other neurodegenerative conditions.

AD strongly impacts microglial function and how microglia shape disease progression. At the beginning of the disease microglia can be neuroprotective by clearing amyloid-beta (Aβ) plaques. Chronic activation leads to sustained inflammation impaired function and aggravation of tau pathology [89–92]. As microglia are responsible for engulfing and clearing Aβ deposits and damaged synapses, in AD, their phagocytic capacity becomes impaired, contributing to the accumulation of toxic proteins. Overactive microglia can also excessively remove healthy synapses, likely via complement-dependent mechanisms, directly contributing to cognitive decline [89–91].

The integrity of the BBB has been directly correlated with memory. Studies have shed light on the link between polymorphisms and methylation patterns of claudin-5, an essential component of the TJs of ECs in the human prefrontal cortex, and cognitive impairment [93, 94]. In more details, Hüls et al. have demonstrated that in humans, increased methylation of the CLDN5 gene is linked to cognitive impairment, implying that the BBB is a key regulator in preserving cognitive function as individuals age [93]. These findings are coherent with another study that examined genome-wide associations in order to connect them with vascular risk factors for cognitive impairment, implicating genes like CLDN5 in the process [94]. This research shows the direct relation between BBB breakdown and memory impairment. Claudin-5 could potentially be proposed as a biomarker for BBB dysfunction and neurodegenerative progression.

In AD, astrocytes become reactive and they lose their ability to clear neurotransmitters like glutamate leading to synaptic dysfunction and impaired long-term potentiation, a mechanism essential for learning and memory [95]. In addition, neuroinflammation caused by AD leads reactive astrocytes to release pro-inflammatory cytokines also contributing to synaptic loss and playing a major role in neuronal damage. This neuroinflammation is closely associated with accelerated cognitive decline in patients. In a review paper the link between the physiological astrocytical heterogeneity and the different AD-related astrocytes is explained. In more details, the authors compare the proinflammatory A1 astrocytes and the anti-inflammatory A2 astrocytes and they explain the characteristics of the apoptotic phenotype of astrocytes. They give an insight on the transcriptomic and proteomic features of the different AD-related astrocytes that lead to a better understanding of the disease [96].

Another NVU element that is pericytes as they regulate the cerebral blood flow. Their dysfunction causes hypoperfusion, depriving neurons of nutrients and oxygen leading to cognitive deterioration [97]. Given the fact that the major pericyte role is regulating the vascular tone and the blood flow most research done correlation pericytes to AD shows that pericyte dysfuntion accelerates cognitive impairment. It would be interesting to investigate other pericytes roles and correlate them to cognitive decline.

A review paper by Procter et al., describes the way pericyte loss and furthermore disruption of the pericyte-endothelial cell crosstalk leads to dementia. In physiology pericyte-endothelial cosstalk control the cerebral blood flow and angiogenesis. Endothelial-pericyte signalling modulates transcription in cells and transcytosis. BBB breakdown leads to pericyte loss and furthermore to pericyte-endothelial interaction reduction with cerebral flow impairment, neuroinflammtion and transcytosis and trasncritpion alterations contributiong to dementia [98].

Parkinson’s disease

Parkinson’s disease (PD) is a neurodegenerative disorder leading to predomenently motor symtoms but also non-motor symptoms mostly affecting the basal ganglia the substantia nigra [79]. In PD, the basal ganglia and frontal cortex show BBB dysfunction characterized by altered transporter expression and microvascular damage, contributing to deficits in decision-making and cognitive flexibility [77, 78] (Table 2). For instance, in animal studies heightened permeability of the BBB in the basal ganglia is associated with PD [99–101]. Furthermore, PD is characterized by age-related BBB breakdown as the TJs are weakened, allowing the entry of various harmful factors to the brain. This leads to symptoms such as motor problems (bradykinesia, tremors and changes in posture) modulated by the basal ganglia [102]. This suggests that the BBB around the basal ganglia is impaired allowing the accumulation of different agents in the area.

Astrocytes are impacted during PD contributing to impairment. In PD, astrocytes fail to fully degrade aggregated alpha-synuclein, which is the main component of Lewy bodies. Consequently, there is accumulation of these aggregates within astrocytes that translate to inflammation and cellular stess [103].According to PET imaging In the early stages of PD patients presented increased reactive astrocytes in the brainstem [104]. In the middle and late stages of the disease decreased reactive astrocytes were found in the cortex and brain stem [104]. This research implys that astrocytes present different conditions during PD evolution. Astrocytes display altered metabolism and impaired gluatamate uptake in PD, causing excitotoxicity and synaptic dysfunction [103]. This could mean that loss of astrocytes in the substantia nigra affects the neurons in this area and could play a role in cognitive decline.

In rotenone-induced mouse PD model it was proved that microglia activation leads to cognitive impairment. Several cognitive-related tests were performed (novel objective recognition, passive avoidance, and Morris water maze) highlighting neurodegeneration implicating microglia depletion [105]. In addition, a human study showed increased cortical microglial activation in 8 non-demented and 11 demented PD patients, suggesting that microglia activation may contribute to pregressive cognitive dysfunction [106]. Another clinical study proved that there is a negative correlation between cortical microglial activation and cerebral glucose metabolic rate in PD demented patients [107]. These findings suggest that cortical neuroinflammation could be leading to neuronal alteration and therefore to PD dementia.

Concerning pericytes, it has been found an alpha-synuclein accumulation during PD. Consequently, pericytes are incapable to proceed to its degradation, leading to inflammation and cell death. The latter triggers an apoptotic cascade reaction on the BBB level speeding up degeneration [108]. This suggests that alpha-synuclein accumulation in pericytes could potentially contribute to the cognitive decline and dementia-related PD.

Stroke

Stroke, affecting both cortical and subcortical regions, results in focal BBB breakdown and vasogenic oedema, leading to cognitive deficits such as aphasia, motor impairments, and memory issues [80] (Table 2). Data demonstrate that pericyte loss leads to cerebral small vessel disease, acute stroke and Alzheimer’s disease [39]. To give an example in the case of stroke, Zhao et al. developed a map of “strategic brain regions” impaired after stroke insult in humans and found a direct relation between the stroke location and the impact on cognitive tasks [109].

After stroke astrocytes rapidly become reactive, modifying their gene expression, morphology and function leading to astrogliosis [110, 111]. After stroke, reactive astrocytes alter neuronal repair and synaptic reconstruction as they form glial scars that physically and chemically inhibit axon regeneration [112, 113]. This suggests that glial scarring interfere with the restoration of neuronal circuits necessary for cognitive recovery. Several preclinical studies in rodents showed that post-stroke, ABCA1 pathway is activated to transform astrocytes into a phagocytic phenotype and engulf synaptic components. By inhibiting this pathway there could be a reduction of astrocytical phagocytosis improving behavioural impairments [114–116].

Microglia play a dual role in stroke pathology and recovery. Their activation states and interactions with the NVU components significantly influences cognitive outcomes after stroke. Neuronal damage mediated by microglia presents an important role in post-stroke cognitive impairment involving different pathways such as TLR4, p25/CDK5, nuclear factor kappa-B and CX3CR1 [117–119]. More studies should be done on the cognitive decline and the role of microglia post-stroke.

Pericytes in stroke survivors are lost in CA1 hippocampal region [120]. This could be because of perfusion modifications or reconstruction of the capillary network. The hippocampus being the major brain region related to memory pericyte loss post-stroke could be correlated to memory loss after the injury. More studies should be conducted in order to clarify the relationship of local hippocampal pericytes and memory alterations after stoke.

Epilepsy

Epilepsy is a brain disorder leading to repetitive seizures as a result of abnormal electrical signals in the brain impacting the hippocampus, amygdala and severeal cortical brain regions leading to learning and memory deficits [121]. In epilepsy, BBB opening and inflammation are observed mainly in the temporal lobe and hippocampus, which correlate with memory impairment and executive dysfunction [81, 82] (Table 2). On a BBB level it has been demonstrated that seizures increase its permeability [122], suggesting a possible cognitive decline due to BBB’s important role in cognition. One of the other NVU elements that is altered in epilepsy is the astrocytes that release gliotransmitters, enhance synaptic transmission and network excitability contributing to seizure propagation [123]. To our knowledge not a lot of studies have investigated the direct cognitive impact of astrocytes impairment in epilepsy. Concerning microglia and inflammation, it has been showed that pro-inflammatory cytokines are present in increased concentrations in epilepsy patients correlated to cognitive impairment [124]. Gaining more knowledge into the molecular mechanisms and the link to inflammation and cognitive decline could facilitate the identification of novel treatments for epilepsy.

Depression

Depression has been linked to BBB dysfunction in the prefrontal cortex and hippocampus, where altered transporter function and reduced BBB integrity may underlie memory deficits and emotional dysregulation [83–85] (Table 2).

An increase in BBB permeability, due to a breakdown of the TJs or dysfunction of the other elements, results in a so-called “leaky” barrier that impacts different cognitive functions. For example, changes in BBB permeability are key factors in the regulation of emotional states, which are altered in stress-induced depression. One study demonstrated the role of vascular endothelial growth factor in BBB permeability in depressive behaviour in mice [125]. In addition, the astrocytic factor thrombospondin-1 (TSP-1) participates in the regulation of synaptogenesis and BBB permeability and has been correlated with depression in women [126]. BBB disruption in the prefrontal cortex, which is responsible for emotion regulation and executive function, is associated with depressive behaviour in mice [125, 127, 128].

Huntintgton disease

Huntington Disease (HD) is an autosomal dominant pathology with a cognitive, motor and behavioural impact mostly on the basal ganglia [129]. The putamen and basal ganglia, which are involved in movement regulation and executive functions, are affected in HD. Changes in brain vessel density affecting BBB permeability are seen in mouse models of HD [130]. This suggests direct involvement of the BBB in executive functions. Furthermore, the key protein whose gene is mutated in HD (huntingtin gene) is expressed in astrocytes and may modulate neurological deficits in HD and could be implicated in the impairment of the BBB [130]. More studies should be conducted on other NVU elements to investigate their involvement in cognitive decline. Also, the identification of more region-specific biomarkers and the development of molecules that cross the BBB depending on its permeability could help to target more efficiently the factors that are altered in neurodegeneration.

Cognitive flexibility impairment

A study in rats showed that altered cognitive flexibility is linked to Ca²⁺ signalling by astrocytic activation in the medial prefrontal cortex [131]. Boender et al. report evidence that the striatal astrocytic excitatory amino acid transporter 2 plays a role in behavioural flexibility. Mice with upregulated excitatory amino acid transporter 2 presented a loss of cognitive flexibility that returned to normal once the upregulation was inhibited [132]. Elevated BBB permeability and cognitive function decline were found to be linked with the brain’s resting-state functional connectivity across various cognitive tasks. The connectivity patterns in the brain were specifically associated with cognitive flexibility and executive control. The integration of multiple modalities, including phase contrast MRI and fMRI, offers fresh perspectives on comprehension of the brain mechanisms underlying cognitive dysfunction [133–135].

The fact that the NVU is distributed in a rather non-uniform way throughout the brain could impact how each region responds to diseases and damage affecting cognitive functions. However, the heterogeneity of the NVU still needs to be directly linked to cognitive impairment.

Concluding remarks and future perspectives

The emerging understanding of BBB heterogeneity challenges the traditional view of a uniform barrier and highlights its profound impact on brain structure and function. Understanding the specific ways in which heterogeneity of the different BBB elements influences neuronal activity is an active area of research. Astrocytes have been thoroughly investigated to link neuronal and vascular activity. Emerging interest in ECs and mural cells has helped us understand their crucial role in the modulation of behaviour. A future perspective would be to integrate different levels from cellular to cognitive with multimodal approaches. Such combinations hold out the promise of helping to find novel treatments and prevention strategies in terms of cognitive dysfunction regulation in different brain disorders.

Abbreviations

AD

Alzheimer’s Disease

AST

astrocyte subtype

BBB

blood-brain barrier

CNS

central nervous system

ECs

endothelial cells

fMRI

functional magnetic resonance imaging

GABA

gamma-aminobutyric acid

HD

Huntington’s disease

MRI

magnetic resonance imaging

NVU

neurovascular unit

PD

Parkinson’s Disease

TJs

tight junctions

Author contributions

A.M was responsible for project conceptualization, administration, NIH and ANR funding acquisition, data analysis and directed the writing of the manuscript. A.T contributed to the drafting of the article and the corrections. R.P, R.H, A.K, C.D, R.H, RiH contributed with thorough editing of the manuscript. All authors have read and approved the final version of the paper. All authors reviewed the manuscript.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.