The existence of a so-called brain renin-angiotensin system (RAS) is controversial. Given the presence of the blood–brain barrier, angiotensin generation in the brain, if occurring, should depend on local synthesis of renin and angiotensinogen. Yet, although initially brain-selective expression of intracellular renin was reported, data in intracellular renin knockout animals argue against a role for this renin in angiotensin generation. Moreover, renin levels in brain tissue at most represented renin in trapped blood. Additionally, in neurogenic hypertension brain prorenin up-regulation has been claimed, which would generate angiotensin following its binding to the (pro)renin receptor. However, recent studies reported no evidence for prorenin expression in the brain, nor for its selective up-regulation in neurogenic hypertension, and the (pro)renin receptor rather displays RAS-unrelated functions. Finally, although angiotensinogen mRNA is detectable in the brain, brain angiotensinogen protein levels are low, and even these low levels might be an overestimation due to assay artefacts. Taken together, independent angiotensin generation in the brain is unlikely. Indeed, brain angiotensin levels are extremely low, with angiotensin (Ang) I levels corresponding to the small amounts of Ang I in trapped blood plasma, and Ang II levels at most representing Ang II bound to (vascular) brain Ang II type 1 receptors. This review concludes with a unifying concept proposing the blood origin of angiotensin in the brain, possibly resulting in increased levels following blood–brain barrier disruption (e.g. due to hypertension), and suggesting that interfering with either intracellular renin or the (pro)renin receptor has consequences in an RAS-independent manner.
Introduction: critical issues
The existence of a so-called brain renin-angiotensin system (RAS) has been discussed for decades, and numerous studies have addressed different aspects of this topic. Many of these studies relied on the overexpression/infusion of one or more RAS component(s) in specific brain nuclei, which of course does not necessarily represent normal physiology. A local RAS in the brain would imply the occurrence of angiotensin generation at brain tissue sites, i.e. outside the blood compartment. In most, if not all, of these cases, the renin required for such local angiotensin generation is derived from the kidney [1,2], and either diffuses from blood into the interstitial fluid or binds to putative renin receptors [3,4]. An identical concept is true for renin’s substrate, angiotensinogen, which is synthesized in the liver. Yet, in the brain, given the existence of the blood–brain barrier, renin and angiotensinogen diffusion is unlikely, and thus brain RAS activity should either depend on the local synthesis of renin and angiotensinogen and/or active uptake of these two proteins. Evidence for such active transport is lacking, and thus the former assumption (local synthesis) seems the most likely (Figure 1). Indeed, an intracellular, non-secreted form of renin has been shown to occur particularly in the brain [5–7]. Second, it has been advocated that the brain synthesizes large quantities of prorenin, the inactive precursor of renin [8,9]. If true, a mechanism is required to allow this prorenin to display activity, since prorenin-renin conversion (by proteolytic cleavage of the prosegment) normally occurs in the kidney only . In the brain, it is assumed that prorenin binds to the so-called (pro)renin receptor, which results in prosegment unfolding, thus allowing angiotensin (Ang) I-generating activity without actual prosegment removal (‘non-proteolytic activation’) .
Potential locations of prorenin and renin (together denoted as (pro)renin) in the brain
We recently quantitated renin and prorenin in the brain , and were unable to detect levels that supported local synthesis: at most, renin and prorenin in brain tissue represented renin and prorenin in trapped blood plasma. Since this challenges the concept of local angiotensin generation in the brain, the present paper critically reviews the above concepts of intracellular renin and prorenin–(pro)renin receptor interaction. Without local renin/prorenin, where does brain angiotensin come from? Do circulating angiotensins enter the brain? Might intracellular renin and the (pro)renin receptor exert RAS-independent effects? The paper ends with a unifying proposal.
The intracellular renin concept
Intracellular renin represents a renin isoform derived from an alternative transcript of the renin gene, lacking the signal peptide and part of the prosegment (Figure 2) [5,13]. Without a signal peptide, it cannot leave the cell, and due to the truncated prosegment (insufficiently long to cover the active site) it should display activity without requiring prorenin activation. Convincing evidence for enzymatic activity of intracellular renin is still lacking. Assays often neither applied a renin inhibitor (to correct for non-renin-mediated Ang I-generating activity) nor corrected for the presence of renin in the angiotensinogen preparation (purified from plasma) .
Intracellular compared with secreted murine renin
Curt Sigmund’s group extensively studied intracellular renin in the brain. Yet, in contrast with the report by Lee-Kirsch et al., they also observed ‘secreted’ renin expression in the brain, albeit at levels corresponding to ≈1% of the expression of intracellular renin [5,,14]. In the hands of Sigmund’s group, overexpressing intracellular (human) renin plus (human) angiotensinogen in the mouse brain resulted in the same degree of blood pressure elevation as overexpression of secreted renin plus angiotensinogen. In both cases, the blood pressure rise could be prevented by losartan , suggesting that intracellular renin, like secreted renin, generates Ang I. Where this occurred intracellularly (and how angiotensinogen reached this site) was not investigated. Moreover, the double transgenic mice were compared with wild-type controls, while single human angiotensinogen-overexpressing mice might have been more appropriate.
Selectively deleting secreted renin from the brain by Xu et al.  had no detectable (cardiovascular) consequences, implying that all brain RAS activity, if present, relies on intracellular renin. Yet, preservation of intracellular renin expression in the brain in whole body secreted renin knockout mice did not compensate for the genetic loss of secreted renin: the renal lesions, low blood pressure, impaired aortic contraction, and inability to generate a concentrated urine under such conditions were unaltered after re-introducing intracellular renin . Apparently, secreted (renal) renin is indispensable for kidney development and blood pressure regulation, and intracellular (brain) renin cannot replace this. Finally, Shinohara et al.  generated a mouse lacking intracellular renin in the brain. Unexpectedly, these mice displayed low plasma renin levels and neurogenic hypertension, and their blood pressure could be normalized by intracerebroventricular application of either losartan or captopril. The latter is suggestive for up-regulation of brain renin. This led the authors to conclude that intracellular renin after all does not react with angiotensinogen, but rather suppresses the expression of secreted renin in the brain via an unknown mechanism . Indeed, when quantitating total (i.e. secreted + intracellular) renin mRNA they observed a modest renin increase in the rostral ventrolateral medulla, and no change in renin expression in other brain nuclei. Given their earlier observation that secreted renin expression is far below (≈100 times) intracellular renin expression , if expressed at all , this implies that deletion of intracellular renin was compensated by a huge (100-fold or more!) up-regulation of secreted renin. Unfortunately, no attempts were made to quantitate this massive brain renin up-regulation at the protein level. These data are difficult to reconcile with the earlier observations by the same group that deletion of secreted renin from the brain had no functional consequences , and that intracellular renin overexpression elevated blood pressure to the same degree as overexpression of secreted renin . An alternative explanation might be that intracellular renin has effects that are entirely unrelated to the RAS.
Non-RAS functions of intracellular renin?
Clausmeyer et al.  originally located intracellular renin in mitochondria. The same group reported that overexpressing cytosolic renin in the heart exerted antinecrotic and cardioprotective effects . This was unrelated to angiotensin generation, since a renin inhibitor (CH732) did not alter this outcome . The authors speculated that intracellular renin acted protective by binding to N-acyl-d-glucosamine 2-epimerase (previously known as ‘renin-binding protein’ or RnBP ), which is involved in carbohydrate metabolism. Indeed, such binding inhibits epimerase activity , although this has been demonstrated for secreted renin only, not intracellular renin or prorenin. Whether the epimerase link also explains the consequences of intracellular renin up- or down-regulation (i.e. the inverse relationship with secreted renin proposed by Sigmund et al.) is unknown . Furthermore, although it has been suggested that a so-called ‘mitochondrial RAS’ exists , data in angiotensin receptor knockout mice ruled out the occurrence of intracellular angiotensins in the absence of angiotensin receptors . Intracellular angiotensin generation is also in disagreement with the fact that angiotensinogen is a secretory protein, and thus is unlikely to react with renin in the cytosol or mitochondria. Thus, the only source of intracellular angiotensin is angiotensin receptor-mediated internalization . Taken together, intracellular angiotensin generation is highly unlikely, and thus a role of intracellular renin, if present, must be RAS independent. This may involve effects through RnBP.
The prorenin–(pro)renin receptor concept
One condition where endogenous brain RAS activation is believed to occur is neurogenic hypertension, for instance induced by deoxycorticosterone acetate (DOCA)-salt treatment. During such treatment, circulating RAS activity is suppressed. Unexpectedly, Sigmund et al. reported that this approach lowers intracellular renin expression, and doubles the expression of the secreted form of renin in brain tissue . Yet, this is in-line with the concept that intracellular renin inhibits secreted renin expression . But doubling secreted renin, given its baseline expression at ≈1% of the expression of intracellular renin, would still not yield substantial amounts of renin. To overcome the problem of low renin levels, Li et al. proposed that DOCA-salt selectively increases brain prorenin [8,9]. They simultaneously assumed that this prorenin would display activity through interaction with the (pro)renin receptor [8,9]. However, the low (nanomolar) affinity of the (pro)renin receptor implies that the prorenin levels in the brain after DOCA-salt treatment should have been many orders of magnitude above its plasma levels, to allow this mechanism to become true . Evidence for such massive up-regulation is not available.
Li et al. observed that DOCA-salt increased Ang II levels in brainstem, cortex, and hypothalamus ≈five-fold in wild-type mice, but not in neuron-specific (pro)renin receptor knockout mice. Yet, DOCA-salt did not alter the low brain renin/prorenin levels detected in the mouse brain by enzyme kinetic assay . As expected, DOCA-salt decreased plasma and kidney Ang II levels, confirming that circulating (i.e. kidney-derived) renin was suppressed. Intracerebroventricular infusion of the putative (pro)renin receptor antagonist, PRO20 (corresponding with the first 20 amino acids of the prosegment of mouse prorenin), reduced blood pressure . Although this was attributed to a modest (≈25%) reduction in the Ang II levels in the cortex and hypothalamus, in reality, due to the five-fold rise after DOCA-salt, brain Ang II levels were still three- to four-fold elevated above wild-type levels when blood pressure had normalized. This suggests that non-RAS mechanisms are more likely to underlie the antihypertensive effect of (pro)renin receptor blockade. Indeed, when the same authors overexpressed the human (pro)renin receptor in neurones, they observed hypertensive responses following intracerebroventricular infusion of human prorenin that could not be blocked by RAS inhibitors . Rather, this response was NADPH oxidase mediated. A complicating factor in the above studies is the species-dependency of the prorenin–(pro)renin receptor interaction. Hence, results obtained with concomitantly occurring prorenin, PRO20 and the (pro)renin receptor of both mouse of human origin should be interpreted with caution.
Non-RAS functions of the (pro)renin receptor?
The (pro)renin receptor was originally proposed to be a novel component of the RAS . Yet, the prorenin levels required to activate this receptor are many orders above its in vivo concentrations , both under normal and pathological conditions. Given this requirement, the idea that the (pro)renin receptor has a function within the RAS is now being abandoned [27,28]. Instead, research is focussing on the (pro)renin receptor as an accessory protein of vacuolar H+-ATPase (V-ATPase), potentially determining its integrity [29,30]. Acting as an adaptor between Frizzled co-receptor LRP6 and V-ATPase, the (pro)renin receptor appears to be indispensable for Wnt/β-catenin signaling , thus explaining why (pro)renin receptor deletion (unlike renin deletion) is lethal, even when restricted to specific cells, such as cardiomyocytes and podocytes [32,33]. Moreover, in a non-biased proteomics approach, the (pro)renin receptor has recently been identified as an interacting protein of sortilin-1 . Sortilin-1 is abundantly expressed in the central nervous system, plays an important role in regulating intracellular protein sorting and neurone survival, and acts as a clearance receptor for low-density lipoprotein (LDL) [35–38]. Inhibiting the (pro)renin receptor reduced the protein abundance of sortilin-1 and the LDL receptor, and decreased the cellular uptake of LDL . Interestingly, sortilin-1 deficiency was reported to protect sympathetic neurones from age-dependent degeneration . Additionally, mice deficient for the LDL receptor display increased locomotor activity in combination with decreased learning and memory abilities, despite normal sensory and motor function . The (pro)renin receptor also interacts with pyruvate dehydrogenase (PDH), the key enzyme that converts pyruvate into acetyl-CoA . Knocking down the (pro)renin receptor in mouse retina pigment cells reduced PDH, thus resulting in pyruvate and lactate accumulation . Similarly, hepatic (pro)renin receptor inhibition in mice resulted in impaired pyruvate metabolism as a consequence of reduced PDH abundance . Surprisingly, inhibiting the hepatic (pro)renin receptor in mice prevented diet-induced obesity and hepatosteatosis, and improved glycemic control, suggesting that the (pro)renin receptor is a central regulator of overall energy metabolism . Genetic mutations that cause PDH deficiency are associated with progressive neurological and neuromuscular degeneration, indicating the importance of PDH in maintaining normal neurone function [43,44]. Finally, inhibiting the (pro)renin receptor reduced acetyl-CoA carboxylase (ACC) abundance . This led to reduced lipid synthesis and increased fatty acid oxidation, since ACC is the rate-limiting enzyme that catalyzes the first step in fatty acid biosynthesis. ACC is also abundantly expressed in oligodendrocytes , which synthesize large amounts of lipid during myelination. Taken together, these findings suggest that the (pro)renin receptor may play an important role not only in brain development, but also in maintaining brain function via its regulation of glucose and lipid metabolism.
Importantly, the (pro)renin receptor is widely expressed in the brain [46,47], the highest expression being found in the pituitary and front lobe. It co-localizes with vasopressin and oxytocin in the paraventricular and supraoptic nuclei, thus potentially linking it to the control of water-electrolyte homeostasis and blood pressure. Indeed, overexpression of the human (pro)renin receptor in the supraoptic nucleus of normal rats resulted in increases in plasma and urinary vasopressin, and decreases in water intake and urine output, without any effect on blood pressure or heart rate . (Pro)renin receptor knockdown in the supraoptic nucleus of spontaneously hypertensive rats attenuated the age-dependent increases in blood pressure, and decreased heart rate and plasma vasopressin , while knockdown in the nucleus of the solitary tract induced blood pressure elevation in an Ang II-independent manner . Thus, the brain (pro)renin receptor appears to exert both pro- and antihypertensive effects, depending on its location, and these effects are unlikely to involve the RAS. Rather, they concern the link between the (pro)renin receptor and V-ATPase, Wnt/β-catenin signaling, and/or energy metabolism as discussed above. From this point of view, it is very well possible that the brain (pro)renin receptor contributes to neurogenic hypertension, albeit in an RAS-independent manner.
What are the renin and prorenin levels in the brain?
Ang I-generating activity in the dog brain was originally described by Ganten et al. . Anti-hog renin blocked ~33% of this activity, suggesting that the majority was not renin-dependent. Of interest, aldosterone treatment lowered dog brain Ang I-generating activity by >50%. Hirose et al.  performed similar experiments in the brain of nephrectomized rats, and also observed Ang I-generating activity that could be partially blocked by anti-hog renin antibodies. These experiments were performed 40 h after nephrectomy. To what degree this is long enough to wash away all kidney-derived renin from the rat brain (e.g. bound to the vascular wall) is unknown. No detailed comparisons with plasma renin levels were made in these studies. As mentioned above, we recently set out to fill this gap . In our studies, we made use of mice, since the renin levels in mice are up to 1000-fold higher (on ng Ang I/ml.h basis) than in rats, humans, pigs, and dogs.
We were able to detect Ang I-generating activity in every brain region we collected (brainstem, thalamus, cerebellum, striatum, midbrain, hippocampus, and cortex), and by making use of a renin inhibitor and renin knockout mice we could confirm that the majority of this activity (up to 90%) was due to renin (Figure 3). However, brain renin levels were extremely low, and corresponded with the amount of renin in 1–25 μl blood plasma per g brain tissue. Since this volume mimics the amount of blood plasma in various brain regions determined with tritiated inulin or Evans blue dye [52,53], these data do not support local synthesis of renin in the brain. Indeed, perfusing the brain with saline prior to the collection of the various regions, reduced brain renin by >60%, confirming the intravascular location of renin in brain tissue. Importantly, DOCA-salt, like aldosterone in the studies by Ganten et al. , lowered brain renin in parallel with plasma renin, and plasma and brain renin levels correlated significantly under multiple condition . Of course, a correlation does not prove causation, but clearly DOCA-salt (a widely used inductor of neurogenic hypertension, assumed to involve brain RAS up-regulation) does not result in brain renin up-regulation.
Renin in plasma and brain nuclei of the mouse
Prorenin activation modestly increased brain Ang I-generating activity, supporting the presence of small amounts of prorenin in brain tissue. The brain prorenin/renin ratio was comparable with that in blood plasma, even in the DOCA-salt treated mice. Thus, there was no selective prorenin up-regulation in the brain, nor did this occur after DOCA-salt. This argues strongly against the concept of prorenin-(pro)renin receptor interaction: the endogenous prorenin levels are simply far too low to allow such interaction. Of course, when infusing exogenous prorenin in excessive amounts [8,9], it is possible that such interaction occurs.
Virtually all other studies on brain renin relied on the detection of renin mRNA in the brain. They often reported poor renin expression [54–56]. We also attempted to detect renin mRNA, using either specific assays for secreted renin or intracellular renin, or a non-specific assay that detected both renins. Under no condition could we show renin expression (secreted or intracellular), and our results indicated that, if renin is expressed in the brain, its expression is >218-fold lower than that in the kidney (no signal after 40 cycles, with renin detection in the kidney at Ct≈22) .
Taken together, our data argue against local synthesis of renin or prorenin in the brain. Brain (pro)renin simply represents the small amounts of (pro)renin in trapped blood plasma. In fact, the levels are so low that even accumulation/uptake of circulating (pro)renin at brain tissue sites outside the blood compartment can be ruled out. This is entirely different from other organs like the heart, kidney, and adrenal. In such organs, renin diffuses freely into the interstitium and/or binds to a receptor, thus reaching tissue levels that are, on a gram basis, at least as high as the renin levels in blood plasma (on ml basis) [3,57,58]. Most likely the presence of the blood–brain barrier prevents such distribution.
Angiotensinogen synthesis in the brain
Many investigators have investigated the presence of angiotensinogen in brain tissue and cerebrospinal fluid (CSF). In most cases, the angiotensinogen levels per gram brain tissue or milliliter CSF corresponded with 5–10% of the levels in blood per ml [52,59,60]. Importantly, in CSF, this percentage is higher than the percentage for protein , suggesting selective enrichment of angiotensinogen compared with other proteins. Surprisingly, some papers reported the occurrence of intracellular angiotensinogen (in glial cells) , in contrast with the idea that angiotensinogen is a secretory protein. Of course, here one should keep in mind that all secretory proteins are intracellular at some moment (within secretory vesicles). Furthermore, results obtained by direct angiotensinogen immunoassay (relying on antibodies that recognize angiotensinogen) were >100-fold lower than those by enzyme kinetic assay (relying on the conversion of angiotensinogen into Ang I by exogenous renin) . This is suggestive for the presence of angiotensinogen in the renin preparation used to convert CSF angiotensinogen into Ang I, a well-known cause of artefacts . This artefact may explain why CSF angiotensinogen was ‘enriched’ compared with other proteins. Such artefacts can only be corrected for by simultaneously incubating a blank control with the renin preparation. Changes in brain and plasma angiotensinogen levels, e.g. after nephrectomy and adrenalectomy, occurred in parallel, although not necessarily in the same time frame [52,59]. Two important arguments for local angiotensinogen synthesis in the brain are the difference in isoelectric point (reflecting differences in glycosylation) compared with plasma angiotensinogen, and the presence of angiotensinogen mRNA in the brain . Moreover, selectively deleting brain angiotensinogen reduced the blood pressure response to intracerebroventricular renin infusion, and induced a diabetes insipidus-like syndrome .
Mice, given their massive renin levels, have circulating angiotensinogen levels that are less than 5% of the levels in rats and humans . As a consequence, their brain angiotensinogen levels, if corresponding to <5–10% of their plasma levels (like in rats and humans), would be <1/200th of the normal angiotensinogen plasma levels in humans and rats. Such low levels are close to the detection limit of current assays. Not surprisingly therefore, we were unable to detect angiotensinogen in the mouse brain, and could at most predict that brain angiotensinogen, if present, would be <3% of the angiotensinogen levels in mouse blood plasma, i.e. within the range observed in humans and rats. We did detect angiotensinogen mRNA expression in different regions of the mouse brain, in full agreement with previous work [65,66], although expression was up to 500-fold lower than in the liver. Thus, findings on brain angiotensinogen in mice do not seem to differ from those in other species, although obviously, given its low angiotensinogen levels, the mouse is not the ideal species to study brain angiotensinogen.
In conclusion, angiotensinogen is detectable by enzyme kinetic assay in brain tissue at levels that correspond to <10% of the levels in blood plasma. If limited to an intracellular (glial) compartment, this angiotensinogen is unlikely to result in local angiotensin generation. If this percentage partly represents an assay-artefact , it is possible that the real brain angiotensinogen levels are as low as those of renin, i.e. they might simply represent angiotensinogen in trapped blood. However, the detection of angiotensinogen mRNA argues against this concept. Studies in animals with selectively suppressed hepatic angiotensinogen expression (e.g. after exposure to liver-targetted angiotensinogen antisense oligonucleotides)  will help to unravel the contribution of brain-derived angiotensinogen.
Origin of brain angiotensin
All major angiotensin receptors (the Ang II type 1 and type 2 receptors and the Mas receptor) are expressed in the brain, and thus presumably they are to some degree activated by their agonists: Ang II and Ang-(1-7). However, as discussed above, renin does not seem to occur outside the blood compartment, and angiotensinogen, if present, is very low or occurs at intracellular sites that may have no relevance. This raises the question where the angiotensin originates that is seen by these receptors. Obviously, it is possible to artificially elevate Ang II and Ang-(1-7) in the brain, e.g. by co-expressing renin + angiotensinogen , by intracerebroventricular infusion or microinjection of Ang II or Ang (1-7) [69,70,71], or by administering the ACE2 activator diminazene (which generates Ang-(1-7) from Ang II) . The first approach relies on the simultaneous presence of ACE, while the last approach assumes the presence of Ang II. These studies resulted in the conclusion that Ang II in the brain induces thirst  and increases sympathetic outflow and blood pressure , while the opposite is true for Ang (1-7) . Yet, where brain Ang II and Ang-(1-7) normally originate cannot be deduced from these studies.
We therefore set out to measure all angiotensin metabolites in the brain region that contained the highest renin levels, i.e. the brainstem (Figure 3). These studies were done in rats (given their larger brains as compared with mice), both under control conditions and after RAS blockade with the Ang II type 1 receptor antagonist olmesartan or the ACE inhibitor lisinopril . Brain Ang I was detectable only during RAS blockade, when circulating Ang I levels were greatly elevated (Figure 4). Under those conditions, brainstem Ang I levels corresponded with ≈1% of plasma Ang I levels. Like for renin, this low percentage represents the amount of Ang I in trapped blood in brainstem tissue (which was not buffer-perfused before collection for the measurement of angiotensins), in full contrast with other organs (e.g. heart, kidney and adrenal) where Ang I is easily detectable and often far above the level of circulating Ang I [73–76].
Plasma and brain Ang I and II levels in rats under control conditions and after RAS blockade with the Ang II type 1 receptor antagonist olmesartan or the ACE inhibitor lisinopril
Without RAS blockade, brain Ang II occurred at levels that were ≈25% of the levels in plasma (Figure 4). This also contrasts with most other organs, where Ang II at least equals, but often far exceeds the plasma levels of Ang II [73–76]. Brainstem Ang II became undetectable after lisinopril, and olmesartan reduced the brain/plasma Ang II ratio by >80%. This suggests that normally circulating Ang II accumulates in brain tissue via binding to AT1 receptors. Such uptake occurs in multiple organs , and facilitates the intracellular accumulation of Ang II . Ang-(1-7) and other angiotensin metabolites were undetectable under all conditions. It should be emphasized that we employed LC-MS/MS to quantitate the individual angiotensin metabolites, a highly sensitive method with little or no background noise . This contrasts sharply with the RIAs and ELISAs that have been employed by others, and that, when used without prior HPLC separation, result in levels that are many orders of magnitude above the levels reported here. For instance, Li [8,9] reported brainstem Ang II levels that were above those in the kidney in their mice. This most likely relates to the detection of a significant background signal, which multiplies rapidly when converting the levels detected in milligrams of brainstem into a pg/g basis in order to allow an immediate comparison with renal levels expressed in pg/g.
Based on our angiotensin data, we conclude that there is no evidence for Ang I generation in the brain, and that brain Ang II represents plasma-derived Ang II bound to AT1 receptors. The concept of accumulation of circulating Ang II in multiple brain regions is not new . Here we should consider that Ang II itself disrupts the blood–brain barrier, thus facilitating its own access to brain regions that are normally inaccessible to angiotensins. Additionally, Ang II will obviously accumulate at sites outside the blood–brain barrier, like the subfornical organ and area postrema. Potentially, from there information might be conveyed to centers within the blood–brain barrier, for instance the centers that contribute to neurogenic hypertension, like the paraventricular nucleus, rostral ventrolateral medulla, and the nucleus tractus solitarii in the hypothalamus and brainstem. A unifying concept is therefore that circulating Ang II gains access to the brain, particularly when the blood–brain barrier is disturbed (e.g. after DOCA-salt). Under such conditions it is even possible that brain Ang II levels increase, despite the reduction in circulating Ang II, simply because a (much) larger percentage of circulating Ang II gains access to brain areas [8,9,14]. This concept does not imply that circulating Ang II under all conditions (e.g. in healthy, normotensive animals) continuously destroys the blood–brain barrier, most likely such effects involve a threshold and/or occur in conjunction with elevated blood pressure only.
Conclusion: a unifying hypothesis
A brain RAS, i.e. a local system allowing independent angiotensin generation at brain tissue sites, is unlikely to exist, both under normal and pathological conditions. Molecular approaches have yielded conflicting results, as discussed in this review, but sensitive biochemical assays now provide clear answers regarding the actual levels of renin, prorenin, Ang I, and Ang II in the brain. Intracellular renin has no role as an Ang I-generating enzyme, prorenin is not present in large amounts nor selectively up-regulated after DOCA-salt (‘neurogenic hypertension’), and brain renin at most equals the amount of renin in a few microliters of blood in brain tissue. Prior buffer perfusion washes away this small amount of renin. It has been argued that buffer perfusion would also wash away brain (pro)renin located outside the blood compartment (e.g. in CSF) . This proposal not only rules out a role for intracellular renin, but also implies that renin-synthesizing cells in the brain, in contrast with those in the kidney, would not store renin. Why this would be is unknown. Here we have to remember that the brain (pro)renin levels before buffer perfusion were already too low to support the presence of any (pro)renin outside the blood compartment. Moreover, washing away interstitial renin from the heart took 20–30 min , i.e. much longer than the acute brain perfusion with saline . Such rapid washout of extracellular (pro)renin from the brain would in fact require the immediate destruction of the blood–brain barrier. This is very hard to believe.
The absence of renin in the brain outside the blood compartment implies that angiotensin generation in the brain cannot involve renin. One possibility is that non-renin enzymes take over renin’s role. However, the exceptionally low Ang I levels in brain tissue strongly argue against the occurrence of Ang I generation outside the blood compartment. Thus, non-renin enzymes do not compensate for the absence of renin.
Angiotensinogen mRNA can be detected in brain tissue, but brain angiotensinogen levels are still very low and the substrate often occurred at unusual intracellular locations where it cannot be cleaved by secreted renin. As discussed above, the concept that intracellular renin is an Ang I-generating enzyme is now abandoned. Importantly, the angiotensinogen measurements reported so far are hampered by methodological problems and insufficient sensitivity, and it is therefore very well possible that the true brain angiotensinogen levels are (much) lower than reported earlier. Thus, to truly establish the contribution of brain angiotensinogen, one needs to quantitate brain Ang II levels in the absence of liver-derived angiotensinogen. At present, our data support that brain Ang II originates in the blood compartment, i.e. it represents circulating Ang II that binds to brain angiotensin receptors that are either outside the blood–brain barrier (e.g. in the circumventricular organ or the brain vasculature), or possibly behind this barrier, e.g. reached under conditions where blood–brain barrier permeability is compromised, like in (DOCA-salt) hypertension [79,81]. By disturbing the blood–brain barrier, high levels of Ang II might even facilitate its own access. In contrast, in organs like heart, kidney, and vascular wall, renin and angiotensinogen diffuse freely into the interstitial space, subsequently resulting in substantial local angiotensin production [3,82,83]. Intracerebroventricular application of losartan or brain-selective ACE2 overexpression in DOCA-salt hypertension will, respectively, block AT1 receptor-mediated internalization or enhance Ang II-Ang-(1-7) conversion of sequestered circulating Ang II, thus explaining the beneficial consequences of this approach . Finally, both intracellular renin and the (pro)renin receptor may well affect blood pressure in a manner that is entirely RAS-independent. A unifying figure (Figure 5) summarizes these insights.
Unifying scheme explaining all findings related to the brain RAS so far
The authors declare that there are no competing interests associated with the manuscript.