The cell types that constitute the NVU (depicted in Figure 1) and collaborate to perform its functions are endothelial cells (ECs), pericytes, smooth muscle cells (SMCs), astrocytes, and microglia (Iadecola, 2017; Freitas-Andrade et al., 2020), which are functionally or physically connected to neurons (Andreone et al., 2015; Cortes-Canteli and Iadecola, 2020; Freitas-Andrade et al., 2020). The NVU is the morpho-functional unit including the BBB, which is important for the transport of nutrients and oxygen from the systemic circulation to the brain, for the clearance of toxic waste from the CNS, for the connection between blood flow and neuronal function, as well as for forming a physical barrier to prevent the entrance of pathogens and other harmful entities into the CNS (Sarvari et al., 2020).

ECs are essential components of the blood vessel wall. Cerebrovascular ECs form tight and adherent junctions with each other to limit the entry of molecules and cells from the peripheral circulation into the CNS (Reese and Karnovsky, 1967). Transporters expressed on the plasma membrane of ECs specifically regulate what enters and exits the CNS. For example, glucose enters the brain exclusively via transporters, despite the brain being a highly metabolic organ responsible for up to 25% of total glucose consumption in the body (Tang et al., 2017; Parodi-Rullán et al., 2019). ECs also regulate blood flow by releasing vasodilators and vasoconstrictors, such as nitric oxide (NO) and endothelin-1, respectively (Morikawa et al., 1994; Biernaskie et al., 2001; Freitas-Andrade et al., 2020). Due to the fact that mitochondria are very abundant in cerebrovascular ECs (Oldendorf et al., 1977; Sarvari et al., 2020), these cells are particularly sensitive to oxygen deprivation (Pun et al., 2009; Freitas-Andrade et al., 2020). Hence, pathological conditions which cause oxygen-glucose deprivation (OGD) and prompt excessive reactive oxygen species (ROS) production trigger cerebral endothelial dysfunction, cell death and BBB breakdown (Schreibelt et al., 2007; Pun et al., 2009; Lochhead et al., 2010; Ghiso et al., 2014; Fossati et al., 2016; Parodi-Rullán et al., 2019; Freitas-Andrade et al., 2020), pointing to the mitochondria as critical targets for EC function and BBB integrity. Efficient communication and exchange of materials between ECs and other cell types of the NVU is essential for CNS homeostasis (Andreone et al., 2015; Freitas-Andrade et al., 2020; Sarvari et al., 2020).

Cerebrovascular dysfunction has been observed in both AD and IS (Fossati et al., 2010, 2012a,b; Guo et al., 2010; Lochhead et al., 2010; Jiao et al., 2011; Shin et al., 2016; Freitas-Andrade et al., 2020; Parodi-Rullán et al., 2020; Quintana et al., 2021). Aging, as well as cardiovascular risk factors, such as hypertension and diabetes, contribute to cerebrovascular pathology (Huang et al., 1995; Girouard and Iadecola, 2006; Price et al., 2012; Cortes-Canteli and Iadecola, 2020; Ojo et al., 2021). The reduction of tight junction proteins such as zona occludin-1 (ZO-1) and occludin is observed in models of stroke and AD (Marco and Skaper, 2006; Jiao et al., 2011; Engelhardt et al., 2014; Freitas-Andrade et al., 2020; Parodi-Rullán et al., 2020). An increase in adhesion molecules expression, such as intercellular adhesion molecule-1 (ICAM-1) and vascular cell adhesion molecule-1 (VCAM-1), is observed in stroke, and recently, has been also associated with AD (Cruz Hernández et al., 2019; Sarvari et al., 2020). The reduction of tight junction proteins and the increase in adhesion molecules triggers the recruitment of peripheral immune cells into the brain and may be due to endothelial activation by danger associated molecular patterns (DAMPs), such as low glucose and oxygen, and aggregated proteins, like Aβ and hyperphosphorylated tau (Canepa and Fossati, 2020; Freitas-Andrade et al., 2020; Sarvari et al., 2020). An increase in BBB permeability allows peripheral substances to enter the CNS, leading to neuroinflammation and oxidative stress (Pun et al., 2009; Turner and Sharp, 2016; Yang et al., 2019). Increased oxidative stress in ECs exacerbates mitochondrial dysfunction, leading to apoptosis (Lochhead et al., 2010; Fossati et al., 2012b,2016; Solesio et al., 2018; Parodi-Rullán et al., 2019). During both ischemic injury and AD, ECs have been observed to decrease NO production, causing dysregulation of CBF (Morikawa et al., 1994; Biernaskie et al., 2001; Girouard and Iadecola, 2006; Austin et al., 2013; Parodi-Rullán et al., 2019; Freitas-Andrade et al., 2020). Interestingly, in endothelial nitric oxide synthase (e-NOS) knockout (KO) mice, deficiency of NO reduces the ability of ECs to neutralize ROS enhancing oxidative stress and neuroinflammation, likely exacerbating AD pathology (Austin et al., 2013). Accordingly, in models of AD and stroke, EC dysfunction has been shown to enhance neuroinflammation by increasing the production of ROS and enhancing BBB permeability, amongst other cellular mechanisms (Parodi-Rullán et al., 2019, 2020; Yang et al., 2019).

Pericytes and SMCs surround ECs within the vascular walls, wrapping around capillaries and arterioles/arteries, respectively. Both cell types have a vital role in the regulation of CBF and BBB integrity (Sagare et al., 2013; Hall et al., 2014; Freitas-Andrade et al., 2020). These cells regulate blood flow mainly through their contraction or dilation, which control the diameter of the blood vessel (Hall et al., 2014; Freitas-Andrade et al., 2020). Recently, the role of blood vessels in the clearance of CNS interstitial fluid and proteins has been thoroughly investigated (Aldea et al., 2019; Carare et al., 2020). Intramural periarterial drainage (IPAD) has been hypothesized as a route to drain the brain interstitial fluid and soluble proteins from the CNS into the cervical lymph nodes, by traveling along the capillaries and arterioles in the opposite direction of blood flow, within the basement membrane of the brain vasculature. The contraction of SMCs provides the motive force for IPAD (Carare et al., 2008; Aldea et al., 2019). This system is dysregulated in models of CAA and aging (Ojo et al., 2021). Another brain clearance pathway, the glymphatic system, has been observed to import cerebrospinal fluid along periarterial space and export interstitial fluid along perivenous spaces, in perivascular tunnels, formed by astroglial cells. This pathway has been proposed to be important in the clearance and regulation of toxic proteins in the brain such as tau and Aβ, and to be mediated by astrocytic aquaporin-4 (AQP4) (Braun and Iliff, 2020; Nedergaard and Goldman, 2020). Interestingly, the glymphatic system is most efficient during sleep (Boespflug and Iliff, 2018; Mestre et al., 2020).

During both stroke and AD, pericytes are observed to detach from the BBB (Sagare et al., 2013; Freitas-Andrade et al., 2020). In an AD mouse model, pericyte loss was observed to exacerbate AD pathology (Sagare et al., 2013). The loss of pericytes decreases vascular stability and decreases the ability for the NVU to regulate CBF (Sagare et al., 2013; Hall et al., 2014). On the other hand, a recent study concluded that Aβ induces pericyte-mediated capillary constriction, reducing CBF (Nortley et al., 2019). Pericytes have been observed to endure damage in models of diabetes as well as obesity, which are both risk factors of AD and IS (Price et al., 2012; Shah et al., 2013a). In cats exposed to hypoxia, pericytes exhibited detachment from the microvasculature (Gonul et al., 2002). Along with dysregulation of CBF in models of AD and stroke, pericyte and SMC loss also contribute to impaired clearance in models of AD and CAA (Sagare et al., 2013; Aldea et al., 2019; Carare et al., 2020; Kim et al., 2020; Ojo et al., 2021).

Astrocytes are essential cells for the NVU and provide a physical connection between blood vessels and neurons (Iadecola and Nedergaard, 2007; Braun and Iliff, 2020). Astrocytic end-feet wrap the blood vessels, helping to stabilize EC tight junctions and regulate CBF (Iadecola and Nedergaard, 2007). Astrocytes play a key role in providing metabolic and physical support to the CNS, along with the regulation of CBF, as well as brain clearance (Freitas-Andrade et al., 2020). AQP4 is a membrane protein that functions in water exchange within the CNS. In healthy individuals, AQP4 is a membrane channel localized at the astrocytic end feet, and it is important for clearance of toxic solutes from the brain (Smith et al., 2019). It is hypothesized that AQP4 is a part of the glymphatic system, mediating fluid exchange and the drainage of proteins, as well as the elimination of liquid from the CNS (Iliff et al., 2012; Boland et al., 2018; Braun and Iliff, 2020; Mestre et al., 2020). Despite the differential contributions of the glymphatic system and the IPAD pathway to the clearance of cerebral fluids and waste material are not completely grasped, it is accepted that astrocytes play a pivotal role in the exchange of fluids, based on studies showing that, in AQP4 KO mice, cerebrospinal fluid influx as well as CNS clearance were decreased (Iliff et al., 2012; Braun and Iliff, 2020). Another essential function of astrocytes is glutamate uptake and release, which is very important for the maintenance of CNS homeostasis (Dejakaisaya et al., 2021; van Putten et al., 2021; Verkerke et al., 2021). Astrocytes also have a critical role in the brain antioxidant system maintenance and in the production of glutathione, an important modulator of oxidative stress and aging (Bains and Shaw, 1997; Venkateshappa et al., 2012; Howarth et al., 2017; Verkerke et al., 2021).

In AD, astrocytes lose their polarization, detach from the BBB, become reactive and release inflammatory cytokines (Liddelow et al., 2017; Sweeney et al., 2019b; Cortes-Canteli and Iadecola, 2020). The crosstalk between astrocytes and microglia needs further understanding, although it has been extensively shown that they modulate each other’s activation state (Joshi et al., 2019; McConnell et al., 2019; McAlpine et al., 2021). Astrocytes have also been reported to lose their expression of AQP4 early in AD (Smith et al., 2019). Differently, in stroke, AQP4 expression correlates with cerebral edema, increasing neuronal damage (Manley et al., 2000). In IS, astrogliosis is very significant due to the presence of DAMPs during the initial/acute phase, as well as the secondary/later phase of injury. A major indicator of gliosis is an increase in the expression of glial fibrillary acidic protein (GFAP) by astrocytes, observed in stroke as well as AD models (Pekny and Lane, 2007; Oeckl et al., 2019; Freitas-Andrade et al., 2020). Upon ischemic injury, neurovascular coupling is lost, together with astrocyte mediated CBF (McConnell et al., 2019; Freitas-Andrade et al., 2020). This results in a loop of metabolic stress and inflammation, worsening mitochondrial dysfunction in vascular cells, and ultimately leading to BBB breakdown (Girouard and Iadecola, 2006; Freitas-Andrade et al., 2020; Radenovic et al., 2020; Sarvari et al., 2020).

Although not directly attached to ECs, microglia mediate BBB integrity (Eldahshan et al., 2019; Freitas-Andrade et al., 2020). They are considered the resident immune cells of the CNS, able to phagocytize neurotoxic substances, and, more recently, they have also been reported to participate in a series of cerebral homeostatic functions, including synaptic plasticity and brain development (Badimon et al., 2020; Willis et al., 2020). There are two well-studied activation phenotypes of microglia, referred as M1 and M2, which are classified as being pro-inflammatory and anti-inflammatory, respectively. The M1 phenotype is detrimental to the BBB, causing dysregulation of the NVU, while the M2 facilitates the endocytosis and the clearance of toxic substances and dying cells, limiting the amount of oxidative stress and promoting a more suitable environment for regeneration/healing, following CNS injury (Tang and Le, 2016; Freitas-Andrade et al., 2020; Jiang et al., 2020). Recently, a protein highly expressed in microglia, triggering receptor expressed on myeloid cells-2 (TREM2), has been observed to protect from neurodegeneration in models of AD. TREM2 is hypothesized to be neuroprotective, associated with the M2 phenotype, although there are variants that increase the likelihood of developing AD (Ulland and Colonna, 2018; Gervois and Lambrichts, 2019; Zhou et al., 2020). The protective mechanisms of TREM2 have also been observed in IS (Gervois and Lambrichts, 2019).

Following IS or HS, microglia transition into a reactive state, causing the release of pro-inflammatory cytokines, exacerbating neuroinflammation and neurovascular dysfunction (Eldahshan et al., 2019). Microglia activation is also observed in models of AD and is now considered one of the major hallmarks of the disease (Villegas-Llerena et al., 2016). Inflammasome activation has been observed to occur in aging, neurodegeneration, as well as in stroke (Mohamed et al., 2015; Liddelow et al., 2017; Hu et al., 2019; Yang et al., 2019). In the brain, inflammasome activation occurs in many different cell types. Particularly, microglia, being the most sensitive cells to pathogen associated molecular patterns (PAMPs) and DAMPs, trigger the activation of caspase-1 and the release of cytokines IL-1β along with IL-18 (Hu et al., 2019). The cytokine receptors on vascular cells are then activated, triggering detrimental processes which cause NVU dysfunction and disruption of neurovascular coupling (Eldahshan et al., 2019). Among other cells, microglia have also been observed to secrete metalloproteases (MMPs), such as matrix metallopeptidase-2 (MMP-2) and matrix metallopeptidase-9 (MMP-9), which have been reported to be activated in AD and IS, degrading the basement membrane, therefore increasing BBB permeability and NVU dysfunction (Lorenzl et al., 2003; Hernandez-Guillamon et al., 2010; Yang and Rosenberg, 2015; Turner and Sharp, 2016; Montaner et al., 2019; Freitas-Andrade et al., 2020; Sarvari et al., 2020; Carcel-Marquez et al., 2021).

The basement membrane is an essential part of the BBB and is composed of numerous proteins such as laminins, collagen, nidogen, and heparin sulfate proteoglycans (Carare et al., 2020). The different proteins that make up the basement membrane are secreted by the cell types composing the NVU such as ECs, pericytes and astrocytes. These different proteins support cell-cell interactions, and thus BBB integrity (Carare et al., 2020; Freitas-Andrade et al., 2020).

Destruction of the basement membrane is severely apparent after middle cerebral artery occlusion (MCAO) (Sarvari et al., 2020). This degeneration occurs through different mechanisms of neuroinflammation such as ROS production as well as MMP secretion (Mohamed et al., 2015; Yao, 2019; Freitas-Andrade et al., 2020). Furthermore, the secretion and activation of MMPs interferes with the composition of the BBB, causing it to become leaky, and further exacerbating already existing oxidative stress and neuroinflammation (Fujimura et al., 1999; Hernandez-Guillamon et al., 2010; Montaner et al., 2019; Kang and Yao, 2020). Basement membrane composition and structure has also been observed to change in models of CAA (Morris et al., 2014).

Blood vessel-coupled-neurons communicate with ECs and other cells of the NVU to modulate vascular structure, dilation or constriction, and to provide nutrients based on neuronal need (Andreone et al., 2015; Iadecola, 2017; Kaplan et al., 2020). Neurovascular coupling is essential to maintain the proper influx of nutrients and the proper efflux of toxic waste from the brain (Girouard and Iadecola, 2006; Andreone et al., 2015; Kaplan et al., 2020). Neuronal activity has been observed to participate in both angiogenesis and neurovascular coupling (Girouard and Iadecola, 2006; Huneau et al., 2015).

Upon CNS injury and cell death of NVU cells, dysregulation of neurovascular coupling occurs (Iadecola, 2017; Sweeney et al., 2019a; Freitas-Andrade et al., 2020; Kaplan et al., 2020). The absence of communication between the CNS and the systemic blood circulation ultimately leads to metabolic failure, oxidative stress, mitochondrial dysfunction and synaptic loss (Freitas-Andrade et al., 2020; Kaplan et al., 2020). It is therefore conceivable that, when cerebrovascular integrity and function are preserved, neurodegeneration is less severe and CNS homeostasis is better maintained.

CAIs have been studied for decades (Maren, 1967; Bertini and Luchinat, 1983; Silverman and Lindskog, 1988; Supuran et al., 2003; Krishnamurthy et al., 2008; Supuran, 2008; Winum et al., 2009; Alterio et al., 2012; Akocak and Ilies, 2014; McKenna and Supuran, 2014; Supuran and Winum, 2015; Ilies and Winum, 2019; Angeli et al., 2020; Supuran and Capasso, 2020). They were first developed as diuretics due to their function on the reabsorption of sodium and water within the kidney (DuBose, 1984; Purkerson and Schwartz, 2007; Supuran, 2018), and are currently used to treat glaucoma as they decrease intraocular pressure, and for the prevention of high altitude sickness (Bradwell et al., 1992; Mincione et al., 2007; Ritchie et al., 2012), as they diminish pulmonary vasoconstriction, increase CBF likely controlling cerebral oxygenation, and reduce cerebral edema (Mishra et al., 2020). However, the molecular mechanism responsible for the effects of CAIs, including some of the most used pan-CAIs such as ATZ and MTZ, are multiple, and still under investigation.

MTZ, for example, was recognized as one of a few compounds that had the ability to inhibit cytochrome c release from the mitochondria under oxidative stress (Wang et al., 2008). Over 1000 compounds of the NINDS drug library were first screened in isolated mitochondria from mouse liver and further confirmed in striatal cells in models of Huntington’s disease. MTZ, inhibiting cytochrome c release from challenged mitochondria, also resulted in the reduction of caspase-9 and caspase-3 activation (Wang et al., 2008; Fossati et al., 2016; Sekerdag et al., 2018). Following this study, FDA-approved CAIs such as MTZ, ATZ, topiramate and more recently developed non-FDA approved selective inhibitors, are starting to be applied in models of cerebrovascular pathology, ischemia, CAA and AD, and their positive effects have been attributed, at least in part, to their ability to prevent mitochondrial dysfunction in cerebrovascular cells (Wang et al., 2009; Shah et al., 2013a; Fossati et al., 2016; Solesio et al., 2018).

CA inhibition has been shown to contribute to cerebrovascular tone during transient phases of pH change, but not in steady-state conditions, in rat arteries (Rasmussen and Boedtkjer, 2018). This study used two different CAIs, ATZ and 4-aminomethylbenzenesulfonamide (AMB). It was revealed that only the pan-CAI ATZ, potent against most of intracellular CA isozymes, had an effect on intracellular acidification mechanisms. Based on the observed results, it was concluded that intracellular CAs are responsible for modifying the rate of intracellular pH and vascular tone within the arteries (Rasmussen and Boedtkjer, 2018). Other studies have observed the vasodilator abilities of ATZ, such as increased NO production leading to increased blood flow in the cortex of rats (Tuettenberg et al., 2001). It is likely that the production of NO is not the only mechanism by which ATZ mediates vasodilation (Kiss et al., 1999). Systemic administration of ATZ has also been widely used as a short test to increase CBF in human studies (Okudaira et al., 1995; Grossmann and Koeberle, 2000; Russell et al., 2008). Interestingly, its stimulatory effect on CBF is reduced in patients with AD or vascular dementia compared to healthy controls (Stoppe et al., 1995; Pavics et al., 1999).

The neuroprotective effects of MTZ, in both in vitro and in vivo models of ischemic injury, displayed in Table 1, were first explored by Wang et al. (2009). Exposing primary cortical neurons to OGD and H₂O₂ resulted in necrosis, which was rescued by MTZ treatment. Similarly, OGD increased cytochrome c release and apoptosis inducing factor (AIF) release from the mitochondria, as well as the activity of caspase-3 (Wang et al., 2009), and MTZ attenuated these effects (Wang et al., 2009), suggesting that MTZ does not only prevent necrosis, but also mitochondria-mediated apoptosis, induced by OGD. In the same study, primary cortical neurons exposed to OGD presented inflammasome activation, measured by the activation of caspase-1 and release of IL-1β. MTZ inhibited both IL-1β release as well as caspase-1 activation in vitro (Wang et al., 2009). This finding supports the hypothesis that CAs may mediate both apoptotic and inflammatory mechanisms. In a mouse model of MCAO, MTZ-treated mice (20 mg/kg) had a smaller infarct size, improved neurological score, and decreased cytochrome c release and caspase-3 activation, compared to non-treated mice (Wang et al., 2009). A more recent study performed in rats determined the effectiveness of ATZ alone, as well as in conjugation with head-down tilt, a way to physically promote CBF, in a transient MCAO rat model. The results indicated that ATZ was protective following transient MCAO, significantly decreasing infarct size (Han et al., 2020). Furthermore, ATZ reduced AQP4 expression, compared to rats with no treatment, suggesting a reduction in brain edema (Han et al., 2020). In a rat model of permanent MCAO, ATZ was used in comparison with selective inhibitors for CA-VII, CA-IX, and CA-XII to determine whether these isoforms are involved in the pathology. Interestingly, the pan-CAI ATZ did not improve neurological score 24 h following occlusion, however, the CAI with selectivity for CA-VII, as well as the compound more selective for membrane isoforms CA-IX and CA-XII, improved the neurological deficits observed in the occluded untreated rats. In the group treated with the CA-IX/CA XII potent and medium selective inhibitor, 6-(benzyloxy)benzo[d]thiazole-2-sulfonamide (BBT), infarct size was significantly reduced, while no reduction was observed in the CA-VII inhibitor group. However, rats treated with 50 mg/kg of ATZ also displayed diminished infarct volume. This study suggests that CA-IX and possibly CA-XII may exacerbate neuronal loss along with neurological and vascular deficits, following ischemic insult (Di Cesare Mannelli et al., 2016). Very recently, another study evaluated different CA-selective inhibitors, in both in vitro and in vivo models of IS (Dettori et al., 2021). This study analyzed a new generation of CAIs, which are lipophilic and selective for the hypoxia-associated (CA-IX and CA-XII) and cytosolic (CA-II and CA-I) isoforms. Importantly, ATZ was used as a reference compound at a much lower dose in this study (4.4 mg/kg). In vitro, the lipophilic CAIs and ATZ protected against OGD-induced anoxic depolarization, providing mechanistic information. In vivo, the CAIs reduced infarct volume, neurological deficits, neuronal damage and microglial activation (Dettori et al., 2021). This study also measured TNF-α and Il-10 plasma levels, showing differences between the sham surgery and the occluded groups, without any affect in the treated groups, likely due to the short time window of treatment (24-h). As proposed in the discussion, it is possible that, with a longer treatment, CA inhibition may mediate inflammatory pathways (Bulli et al., 2021; Dettori et al., 2021). Indeed, CAs have been also associated with inflammatory pathologies, such as rheumatoid arthritis and cancer metastasis (Supuran, 2008; Alver et al., 2011; Liu et al., 2012). Overall, both MTZ and ATZ are protective in models of IS, however, the development of CA-IX and potentially CA-XII inhibitors may be beneficial for the treatment of both neurological and vascular disorders, following IS. The protective mechanisms observed following CA inhibition in models of IS, summarized in Table 2, seem to be superior with compounds selective for membrane-bound isoforms CA-IX and CA-XII (Bulli et al., 2021).

The protective mechanisms of CA inhibition in models of HS are summarized in Table 2. Inhibition of CA reduced brain injury after ICH in Sprague-Dawley rats (Guo et al., 2012). This study focused on CA-I, which is highly expressed in red blood cells. Upon intracaudate injection of blood, CA-I levels were increased in the ipsilateral basal ganglia, for as long as 3 days post-injection. Moreover, following intracaudate injection of CA-I, increased brain water content, microglial activation and neuronal cell death were detected, suggesting that CA-I expression may contribute to cerebral edema and neuroinflammation. ATZ-treated rats exhibited reduced perihematomal edema, as well as sodium accumulation, together with decreased neuronal deficits (Guo et al., 2012). However, the cellular mechanisms responsible for the effects of CA-I on BBB integrity require further investigation. Lately, the scientific community is beginning to hypothesize that CA-I, in both the retina and brain, contributes to vascular permeability, suggesting that specific inhibitors of CA-I could be beneficial to minimize NVU dysfunction, and thus neuroinflammation around the barrier (Gao et al., 2007). Furtherly, in a SAH mouse model, mice treated with the pan-CAI, MTZ, had decreased caspase-3 activation and apoptosis in the hippocampus and cortex, compared to the non-treated group, ameliorating neurological deficits (Li et al., 2016). In primary cortical neurons, MTZ reduced hemoglobin and blood induced cell death along with production of ROS (Li et al., 2016). Topiramate, another CAI active on multiple CA enzymes, including CA-V, reduced SAH injury in rabbits (Seçkin et al., 2009).

Recently, the effects of CA inhibition on mitochondrial and neurovascular cell function have been tested in models of amyloidosis (Table 3). In AD and CAA, Aβ toxic aggregates accumulate around and within the cerebral vessel walls, as well as in parenchymal plaques in specific brain regions, such as the hippocampus and cortex (Ghiso et al., 2014; Cortes-Canteli and Iadecola, 2020). Our lab has shown that Aβ oligomers and protofibrils induce cytochrome c release from the mitochondria in multiple cell types of the NVU, including ECs and SMCs, neurons and glial cells, leading to caspase activation and cell death (Fossati et al., 2010, 2012a,b, 2016; Parodi-Rullán et al., 2020). In all of these neurovascular cell types, MTZ reduced mitochondrial mediated cell death triggered by Aβ (Fossati et al., 2010, 2016). Moreover, MTZ attenuated caspase-3 and caspase-9 activation in ECs, neuronal and glial cells in vitro, as well as caspase-3 activation in vivo (Fossati et al., 2016; Solesio et al., 2018). In an in vivo study, Aβ was injected into the hippocampus of wild-type mice, in the presence or absence of a previous intraperitoneal MTZ injection. Interestingly, MTZ treatment attenuated the activation of caspase-3 in microglia and increased the amount of NeuN positive neurons within the hippocampus, indicating that MTZ treatment had a protective effect against neurodegeneration following Aβ-injection (Fossati et al., 2016). To elucidate the specific mechanisms responsible for these protective effects, our group showed that Aβ-induced mitochondrial dysfunction, mitochondrial membrane depolarization, cytochrome C release, and H₂O₂ production were attenuated in neuronal and ECs, not only by MTZ, but also by ATZ, which was effective at lower concentrations (Solesio et al., 2018). Moreover, both ATZ and MTZ inhibited caspase-9 activation caused by Aβ in cerebrovascular EC, and the resulting apoptosis (Solesio et al., 2018). MTZ has been also observed to increase the activation of nuclear factor-related factor 2 (Nrf2), in models of high-altitude sickness and more interestingly, in human neuroblastoma cells and primary cortical neurons challenged with Aβ in vitro (Lu et al., 2020; Sotolongo et al., 2020). Nrf2-activation by MTZ increased the activity of antioxidant enzymes, such as superoxide dismutase-1 and heme-oxygenase-1, pointing to the potential downstream effects of MTZ (Sotolongo et al., 2020). These studies support the hypothesis that CA inhibition is protective to multiple cell types of the NVU, in models of amyloidosis, and highlight the necessity to test these and/or isoform-specific compounds in clinical trials for AD and CAA. New studies currently performed in our lab are also confirming the positive effects of CAIs on cognitive performance in mouse models of amyloidosis (Angiulli et al., 2018). Additional studies are in process to investigate chronic treatments with pan-CAIs in multiple models of CAA and AD, in parallel with the assessment of specific isoform inhibitors, to further elucidate the role of different CA isoforms in AD and CAA pathology.

Interestingly, although CAIs are proving to be beneficial in preventing CAA and AD pathology, CA activators have also showed beneficial effects for memory retention in acute models (Ilies et al., 2002; Canto de Souza et al., 2017; Sanku et al., 2018; Provensi et al., 2019; Blandina et al., 2020; Schmidt et al., 2020). These findings point to a relevant role of CA enzymes in the modulation of multiple pathways in contextual fear memory extinction, neurodegenerative and neurovascular pathology. They also highlight the need for a careful and specific modulation of CA enzyme’s activity, and for a further understanding of the effects of each isoform in different brain areas, cell types, and pathological conditions.

Some studies have also explored the specific inhibition of the mitochondrial isoforms CA-VA and CA-VB, which recently gained interest for their role in the prevention of cerebrovascular pathology, in different models of diabetes (Price et al., 2012, 2017; Shah et al., 2013a), which contributes to AD and stroke pathogenesis (Kuźma et al., 2018; Cortes-Canteli and Iadecola, 2020; Freitas-Andrade et al., 2020; Zlokovic et al., 2020). In a streptozotocin induced diabetic CD-1 mouse model it was reported that topiramate treatment for 3 weeks increased the levels of glutathione and reduced oxidative stress in the brain. It was also observed that diabetes-induced loss of cerebral pericytes was attenuated by topiramate (Price et al., 2012). The same group confirmed these mechanisms in the brains of CA-V double KO mice, where increased levels of glutathione and decreased oxidative stress was observed compared to wild-type (Price et al., 2012). Hyperglycemia-treated cerebrovascular pericytes increased ROS production, as well as the rate of mitochondrial respiration. The inhibition of CA (particularly CA-V), with topiramate, reduced ROS production (Shah et al., 2013a). Although topiramate has a high inhibitory activity on the CA-V isoforms, it also has affinity for the other isoforms, as well as multiple off-target effects in the brain, as shown by studies that analyzed topiramate’s functions on GABA and glutamate receptors (Johnson, 2005; Mao et al., 2015). Ideally, more specific inhibitors should be developed to further investigate the functions of CA-VA/B. The same group used a plasmid to overexpress CA-VA in cerebral pericytes exposed to high-glucose to confirm the hypothesized molecular mechanisms involving CA-VA, such as ROS production and cell death, observing that topiramate was protective against these mechanisms (Patrick et al., 2015). As a proof of concept, the same group genetically knocked down CA-VA and CA-VB in cerebral pericytes (Price et al., 2017). CA-VA and CA-VB knockdown improved cell viability, compared to the control, when exposed to hyperglycemia, although the specific knockdown of CA-VA was even more effective than CA-VB knockdown in decreasing ROS production and apoptosis, in a hyperglycemic environment (Price et al., 2017). These studies reveal potential differences between CA-VA and CA-VB function in brain pericytes, which need further investigation. In a streptozotocin induced diabetic mouse model, cerebrovascular pathology was also characterized using electron microscopy. The breakdown of the BBB and its dysfunction was attenuated by topiramate treatment in vivo (Salameh et al., 2016). More recently, a study using a high-fat-induced diabetes mouse model, focused on the disruption of the hippocampal BBB. BBB tight junction proteins, such as ZO-1 and claudin-12, were reduced with a high-fat diet, while topiramate treatment increased their expression, along with the attenuation of oxidative stress (Salameh et al., 2019).