The pathophysiology of depression remains elusive and current treatments, which focus on traditional pathways (monoamine alterations), are only partially effective. Remission rates in the treatment of depression are only about 30% for those treated with traditional pharmacotherapy, and multiple agents are often required to achieve an adequate level of recovery [1] Evidence points to the involvement of neuroinflammation, oxidative and nitrosative stress pathways, mitochondrial dysfunction and neurotrophic signalling in depression [2].

Recently, the renin–angiotensin system (RAS) was proposed to be implicated in depression, and that blocking this system, either with angiotensin-converting enzyme inhibitors (ACEIs) or with angiotensin II type 1 receptor (AT1R) blockers, would translate into clinical benefits for the depression treatment [3–7]. Here, we review the literature so far on RAS-targeting drugs in depression.

A PubMed search was conducted for literature published between January 1974 and June 2017. Search terms included were: depression OR inflammation OR anxiety OR mood AND renin–angiotensin system, angiotensin, ATR1, ATR2, angiotensin receptor blockers, angiotensin-converting enzyme inhibitors, ATR3, ATR4, Mas, and aldosterone. Systematic reviews, randomised controlled trials (RCTs), observational studies, case series and animal studies with an emphasis on the angiotensin system and its role in depression were included. Articles not in English were excluded. The PubMed search was augmented by manually searching the references of key papers and related literature. The results were presented as a narrative review.

The RAS was discovered in the 19^th Century, after the blood pressure-raising agent renin was first identified in the rabbit kidney [8]. In time, the RAS became an established and extensively studied peripheral regulator of blood pressure and renal-mediated body fluid homeostasis, and was discovered to be a central target in clinical hypertension therapy. Renin, a protein synthesised by the juxtaglomerular cells of the kidney, cleaves the polypeptide angiotensinogen to generate angiotensin I (Ang I). This peptide is metabolised to angiotensin II (Ang II) by angiotensin I-converting enzyme (ACE).

It was surprising when renin was identified in the dog brain in 1971 [9, 10]. Subsequently, intracranial Ang II was shown to elevate blood pressure and to promote fluid intake [11–14], suggesting that angiotensin receptors were present in the brain. The actions of Ang II in the central nervous system are mediated mainly by two receptor types: AT1R and AT2R [15, 16]. Other receptors, including MAS [17], the (pro)renin receptor (PRR) [18] and AT4R [19], have also recently been identified but their roles remain less well characterised. AT3R was first reported as a new binding site for Ang II in mouse neuroblastoma cell cultures [20], but a separate gene for this receptor remains to be sequenced in humans.

AT1R mediates most of the peripheral and central actions of Ang II [21] and is implicated in multiple pathways related to regulation of the stress response. Stimulating AT1R contributes to the release of inflammatory markers [22]. Ang II interacts with AT1Rs, activating the NADPH–oxidase complex [23–25], the microglial RhoA/Rho kinase pathway [26–28], NF-kappa B, inducible nitric oxide synthase (iNOS) and cyclooxygenase-2 (COX-2). In turn, activated COX-2 forms an intermediate in several key aspects of central nervous system inflammation, and in oxidative and nitrosative stress (see Fig. 1). AT1R stimulation also releases tumour necrosis factor α (TNF-α) [29, 30], which is important in several neurodegenerative disorders [29, 31–33], and regulates activation of the hypothalamic–pituitary–adrenal axis. Stimulation of AT1R in the parvocellular hypothalamic paraventricular nucleus (PVN) by Ang II increases production of corticotrophin-releasing factor [34–36]. In turn, this spurs adrenocorticotropic hormone secretion in the anterior pituitary gland, starting the stress response cascade. Accordingly, in humans, AT1R blockade downregulates hypothalamic–pituitary–adrenal axis activation [37].

Ang II also stimulates the release of aldosterone via AT1R in the adrenal cortex of the kidney [38]. Thus, the acronym ‘RAAS’ (as in renin–angiotensin–aldosterone system) is often used. Besides being regulated by Ang II, aldosterone release is also stimulated by adrenocorticotropic hormone and the sympathetic nervous system. The role of aldosterone in the brain has previously been downplayed because its specific intracellular receptor, the mineralocorticoid receptor (MR), shares affinity with cortisol, which circulates at a ~1000-fold higher concentration than aldosterone [39]. For a tissue to be sensitive to aldosterone, it must express 11β-hydroxysteroid dehydrogenase type 2 (HSD-2) protein, which degrades cortisol, freeing the MR to the action of aldosterone. HSD-2 has been identified in the brain, mainly in the nucleus of the solitary tract, but also in the PVN [40]; regions that also express AT1R. Surprisingly – paralleling the history of angiotensin – aldosterone synthesis was also recognised in the amygdala, hippocampus and hypothalamus of the brain [41].

AT1R is particularly dense in the anterior pituitary; the circumventricular organs (area postrema; subfornical organ, the vascular organ of lamina terminalis and the median eminence); the lateral geniculate body; inferior olivary nucleus; the nucleus of the solitary tract and in the PVN, the preoptic and the supraoptic nuclei of the hypothalamus [42].

Modern molecular approaches have revealed that AT2R is also expressed in the adult brain [43, 44]. AT2R is involved in neurodevelopment [45–49] and participates in cell growth inhibition, fetal tissue development, extracellular matrix modulation, neuronal regeneration, apoptosis, cellular differentiation, and, possibly, vasodilation and left ventricular hypertrophy [50]. AT2R stimulation exerts neuroprotective effects in ischaemic stroke in rodents [51–55], and while the underlying mechanism remains to be fully characterised, it seems to partly involve an increase in the anti-inflammatory cytokine interleukin-10 [56]. AT2R is particularly dense in the amygdala, caudate putamen, medial geniculate body, globus pallidus, habenula, hypoglossal nucleus, inferior colliculus, inferior olivary nucleus, locus coeruleus, thalamus, and ventral tegmental area [42].

More components of the RAS such as ACE2, angiotensin-(1–7) and the Mas receptor have recently been identified in the brain. This alternative pathway is sometimes referred to as the non-classical RAS [57]. Originally identified in 1986 as an oncogene in mice [58], the tumorigenic power of Mas was later discredited and remained an orphan receptor until it was subsequently shown to bind with Ang (1–7) [17]. ACE2 can hydrolyse Ang II to produce Ang-(1–7). It can also cleave Ang I, producing Ang-(1–9) with subsequent Ang-(1-7) formation, although with much less efficiency. Mas is thus proposed to be a receptor for Ang-(1-7), with its highest expression in the brain [59]. The action of Ang-(1–7) through Mas is thought to influence arachidonic acid production and nitric oxide synthase activation [60] (see Fig. 2).

The recently discovered PRR is highly expressed in the brain [18]. Its large extracellular domain binds and captures renin and its almost inactive precursor prorenin, increasing their enzymatic activities [61], but it also mimics the actions of AT1R through intracellular signalling [62].

A specific receptor for angiotensin IV (Ang IV), another less active peptide than Ang II, was first identified in a guinea pig hippocampus [19]. It is thought that the identity of AT4R was established when it was discovered that Ang IV is a strong inhibitor of insulin-regulated aminopeptidase (IRAP) [63]. IRAP is responsible for oxytocin degradation and, as demonstrated when an injection of Ang IV abolished the antidepressant effects of oxytocin in mice [64], is apparently required for its mood effects to take place. Yet recently, discrepancies between Ang IV binding site-antagonist and IRAP inhibitors [60], or the unaltered cognitive response of Ang IV in IRAP knockout mice [65], have cast doubt on whether IRAP is the only AT4R receptor. Further candidates for the role of AT4R have been proposed [42].

Ang II is also involved in cerebral blood flow regulation [21, 22]. Rising circulating Ang II is free to cross into the subfornical organ. This is a circumventricular organ lacking the blood–brain barrier, which, via AT1R, signals the paraventricular nucleus of the hypothalamus to activate the rostral ventrolateral medullary neurons and peripheral sympathetic nerves, thereby raising blood pressure [66]. Overstimulation of AT1Rs can lead to endothelial dysfunction [67] and neuronal injury and vulnerability caused by cerebrovascular remodelling [68–72].

It is well established that angiotensin receptors are present in the brain, yet the origin of active angiotensin peptides in the brain remains somewhat controversial. Researchers are puzzled because while Ang II is too hydrophilic to cross the blood–brain barrier [73], expression of renin in the brain is too low to account for its local synthesis [74]. Among the hypotheses advanced to solve this apparent paradox are renin-independent synthesis of angiotensin peptides [75]; impaired blood–brain barrier in hypertension leading to Ang II leaking into the cerebrospinal fluid [73]; an intracellular form of renin in the brain [76] or undetectable renin caused by its sequestration by PRR [62]. Although uncertainties persist, targeting the brain RAS or the peripheral RAS cannot be equal because ACEIs that penetrate the blood–brain barrier are superior to non-centrally acting ones in preventing cognitive decline [77, 78].

Inflammation is essential for restoring homeostasis in stress, infection and injury [79]. Hormones and circulating pro-inflammatory cytokines, products of neuronal injury and bacterial endotoxins, activate transcription factors. Activated inflammatory cascades with brain parenchymal microglia and blood-derived infiltrating macrophages also participate [80]. A well-regulated central inflammatory chain is fundamental to restore homeostasis, but an exaggerated response can be responsible for chronic inflammation, neuronal damage and a decrease in brain-derived neurotrophic factor [81–86]. Thus, excess or sustained activation of immune responses augments the risk of disease in vulnerable individuals, and can be important in the pathophysiology of many neurological and psychiatric disorders [2, 81, 87–96].

The inflammatory hypothesis [97, 98] postulates that depression is the result of altered immune-inflammatory pathways. This leads to increased immune activation, inflammation, nitro-oxidative stress and alteration of the kynurenine pathway, which ultimately causes changes in monoamine levels. MDD is characterised by a low-grade inflammatory state with increased peripheral levels of inflammatory cytokines, and microglial activation [98–103]. Normalised levels of inflammatory markers are associated with remission of clinical depression [104], while persistently elevated levels are associated with a lack of response to antidepressants [105]. Elevated levels of inflammatory markers such as C-reactive protein (CRP) may increase the risk of a first episode of depression [106, 107]. However, a large Mendelian randomisation study found no causal association between increased CRP levels and depression in people with genetically elevated CRP [108], and also that inflammation may better stratify those who will or will not benefit from anti-inflammatory treatments [109]. More compelling is the strong observation of depressive symptoms induced by interferon-α treatment, both in humans and in animal models [110–113].

Consequently, it has been hypothesised that drugs with anti-inflammatory properties might also demonstrate antidepressant potential. Nonsteroidal anti-inflammatory drugs have shown benefits [114, 115], although no influence was observed in association with antidepressants [116]. Cytokine inhibitors were found to improve depression [117–119] and specific depressive symptoms, such as anxiety [120] and fatigue [117], among patients with psoriasis [117, 118, 120] or ankylosing spondylitis [119]. This finding is supported by evidence from animal models [121]. In an open-label report, aspirin exhibited antidepressive effects, even at low doses [122], and may have a more favourable benefit/risk ratio compared with selective COX-2 inhibitors [123, 124]. Epidemiological reports also support antidepressant effects of aspirin [106, 125]. N-acetylcysteine may also be useful in treating MDD [126–128]. Statins, which apart from their antiatherosclerotic and cardioprotective effects also display neuroprotective and anti-inflammatory effects [129–131], showed the potential to produce mood-related benefits [132] and are associated with a reduced risk of depression [133]. Clinical trials of statins seem to show antidepressant effects in aggregate [134]. In a meta-analysis, supplementing the treatment of severe MDD with polyunsaturated fatty acids (PUFAs) was found to be beneficial, even though its role in mild-to-moderate depression or prevention seems limited [135].

Studies attempting to link depression with genetic variations in the RAS provide additional evidence. Initial reports for the most studied ACE polymorphism (I/D) – the presence or absence of a 287-bp fragment in intron 16 related to ACE serum levels [136] – were inconsistent and a meta-analysis showed no significance [137, 138]. However, other single nucleotide polymorphisms have been associated with depression [139, 140], including the GG genotype of ACE A2350G, which also correlated with higher ACE serum activity [141]. Recently, seven single nucleotide polymorphisms were significantly tied to late-life depression and cortisol levels under stressful circumstances [142]. The AT1R genotype (A1166C) CC is also associated with depression and increased responsiveness to Ang II [6], as well as clinical response [143, 144]. Epigenetic mechanisms also appear to be important, as altered methylation of the regulatory region of the ACE gene has been associated with depression [145]. ACE polymorphisms even seem able to influence antidepressant response [145–147], cognitive function after a depression episode in the elderly [148, 149], or suicide behaviour [150, 151].

The role of aldosterone in depression is an emerging area of research, thus regulation of aldosterone by the RAS is another point to take into account. Patients with primary hyperaldosteronism have depressive symptoms [152, 153]. In animal models, administering aldosterone leads to depressive behaviour [154], anxiety [155] and anhedonia [156]. Eplerone, an aldosterone antagonist, had anxiolytic properties in rats [157]. Poorer clinical outcome in MDD is predicted by higher salivary aldosterone [158, 159]. Conversely, MDD patients with suicidal behaviour had lower concentrations of aldosterone compared to suicidal patients without MDD and non-suicidal depressive patients [160]. Spironolactone, another MR antagonist, induces a sleep pattern characteristic of melancholic depression and reduces the efficacy of amitriptyline [40]. This hints at a non-linear dynamic of aldosterone throughout the MDD episode, prompting its exploration as a biomarker that is able to differentiate depression duration. Indeed, at least in women, higher aldosterone levels are associated with a shorter duration of a depressive episode [159], and in an animal model were used to mark the onset of depression [161].

Taking the above evidence in aggregate, current understanding of the pathophysiology of depression supports the search for novel therapeutics affecting the pathways of inflammation, oxidative biology, apoptosis and neurogenesis. Besides their anti-inflammatory effects, angiotensin receptor blockers (ARBs) and ACEIs have good tolerability, limited side effects and are already widely used drugs approved by the US Food and Drug Administration [162, 163]. Their neuroprotective, anti-inflammatory, vasodilatory [164] and microglia activation inhibitory effects [29] make them candidates for novel therapeutic targets for inflammatory brain diseases and cognitive disorders [21, 29, 30, 165, 166]. In this regard, interesting data is emerging from animal models.

The body of evidence supporting the antidepressant and antianxiety effects of drugs targeting the RAS in animal models is increasing. Mutant mice lacking the angiotensin gene have less depressive-like behaviour in the forced swim test [167]. Pharmacologically decreasing the production of Ang II by administering captopril (an ACEI) produces an analogous result [168].

Blockage of Ang II also leads to antidepressant-like activity in the learned helplessness [169] and chronic mild stress paradigms [170, 171], both more valid models than the forced swim test. Preclinical data also suggests a link between the antidepressant effect and a decrease in Ang II activity; AT1R antagonism by its specific blockers losartan [3], valsartan [171], irbesartan [170] and telmisartan [172] has similar actions to that caused by ACEIs. As with most antidepressants, use of these blockers also seems to have antianxiety properties. Candesartan [21, 173], losartan [174, 175] and captopril [176] reduced anxiety behaviour (promoting exploration) in the elevated plus maze test. Nevertheless, enalapril (a non-centrally acting ACEI) was not effective in normotensive rats [175].

Remarkably, different phenotypes of anxiolytic response to ARBs across different mice strains may be explained by differences in AT1R expression levels [177]. Curiously, mood effects were also apparent in an amphetamine-induced model of mania in mice, which candesartan was able to prevent and treat with comparable efficacy to lithium [30]. Transgenic rats overexpressing Ang-(1-7) [178] or ACE2 [179] showed a reduced anxiety phenotype that is seemingly dependent on Mas signalling, since antagonism of Mas reversed the phenotype. Administering Ang-(1-7) was associated with decreased oxidative stress markers in the amygdala [180]. The same Mas antagonism also prevented the anxiolytic/antidepressant effect of enalapril in transgenic hypertensive rats [181, 182].

These agents seem to influence mood disorders independently of their blood pressure-lowering activity. A study exploring the effect of valsartan in a chronic mild stress model found no change in average blood pressure after a month of treatment, while at the same time registering antianxiety and antidepressant effects [171].

Animal experiments also support the anti-inflammatory and oxidative stress-reducing effects of these drugs as part of their mechanisms of action. Both irbesartan and fluoxetine decreased levels of thiobarbituric-reactive substances – oxidative stress markers – while increasing catalase and glutathione (antioxidants) and serotonin (5-HT) levels in the brain [170]. Valsartan also increased neurogenesis in mice [171]. Captopril and perindopril (both centrally acting ACEIs) [183], telmisartan [183, 184] and candesartan [21, 185, 186] all show anti-inflammatory effects by reducing microglial activation and levels of inflammatory markers such as nitric oxide and TNF-α.

To date, no RCT has assessed the effects of ACEIs or ARBs in depression. However, observational studies have established a bidirectional link between cardiovascular disorders and depression. Antihypertensive sympatholytic drugs such as reserpine or clonidine can induce depression [187–189], prompting some to propose that sympathetic nervous system hyperreactivity is a common substrate [190, 191]. It was unclear whether this association was caused by hypertension itself, its treatment, or both [192, 193].

A meta-analysis of prospective cohort studies [194] found no evidence that hypertension is a risk factor for depression. However, the contrary – that depression increases the risk of developing hypertension – has been suggested [195] and confirmed by a meta-analysis [196]. In light of all the evidence, the RAS now emerges as a major link between mood and the cardiovascular system.

In a case-control study of 972 patients from primary care practices, who had both diabetes and a new diagnosis of depression, those exposed to ACEIs in the last 6 months showed a lower odds ratio for depression (OR 1.3, 95% CI: 0.8–2.2) compared to those exposed to beta-blockers (BBs) (OR 2.6, 95% CI: 1.1–7.0) and calcium channel blockers (CCBs) (OR 2.2, 95% CI: 1.2–4.2) [201]. In a recent population cohort study, ACEIs decreased the incidence of MDD [202]. These results were replicated by Boal et al. [203], who examined mood-related hospital admissions of 144,660 patients treated with antihypertensive monotherapy for a five-year follow-up. Interestingly, ACEIs and ARBs were associated with the lowest risk of mood disorder admissions (log-rank P = 0.006), while CCBs (hazard ratio (HR) = 2.28, [95% CI 1.13–4.58]; P = 0.02) and BBs (HR = 2.11, [95% CI 1.12 –3.98]; P = 0.02) were associated with increased risk compared to ACEIs and ARBs. There was no significant difference in patients receiving no antihypertensive medication (HR = 1.63 [95% CI 0.94–2.82]; P = 0.08), or those taking thiazide diuretics (HR = 1.56 [95% CI 0.65–3.73]; P = 0.32).

However, in the CREATE trial, a randomised placebo-controlled trial of citalopram in 284 coronary heart disease patients with MDD, the use of ACEIs predicted a worse response to citalopram [204]. A possible caveat is that the use of ACEIs may cause bias towards more severe coronary disease, and thus a possible vascular, more refractory type of depression. Another interesting possibility, considering the antidepressant properties of ACEIs, is that their use may have prevented or even treated milder episodes of depression, creating a selection bias for more severe depression. Indeed, we know that an increasingly smaller percentage of patients respond or remit after trying a second or third drug after failing previous treatments [205], and that antidepressant-naïve patients improve their Hamilton Depression Rating Scale score more than those taking antidepressants in response to treatment [206].

The antidepressant effects of ACEIs can be further inferred both by mood effects in the population without a formal diagnosis of MDD, and in studies looking at quality of life. Mood elation was reported in healthy volunteers taking enalapril [207]. One RCT found a higher quality of life score was attained in patients taking captopril compared to other classes of antihypertensive drugs, despite similar blood pressure control [208]. A head-to-head comparison of captopril (a centrally acting ACEI) and enalapril (a non-centrally acting ACEI) reported no difference in antihypertensive efficacy, but that captopril had a superior effect on quality of life measurements [209].

In the Norwegian HUNT study [192], the depressive symptoms of a large population of 55,472 patients with systemic hypertension taking an ACEI were compared with those of patients with untreated systemic hypertension. Results showed an important trend in favour of the depressive symptom-reducing effects of ACEIs, as assessed by the Hospital Anxiety and Depression Rating Scale (OR 0.54, 95% CI 0.28–1.08). Interestingly, those on BBs (OR 1.20, 95% CI 0.78–1.83) or on CCBs (OR 1.04, 95% CI 0.70–1.53) showed no reduction in depressive symptoms compared to the untreated systemic hypertension group. Again, this suggests that the pharmacological benefits of ACEIs and ARBs in depression are independent of their antihypertensive effects. A small open-label trial of 17 type 2 diabetic patients taking candesartan for at least 3 months found that depression scores were improved [210].

Nonetheless, there are a few negative reports of the effects of RAS drugs on mood. A small (n = 8), 6-week, double-blind crossover trial found captopril to have no positive effects on mood [211]. Another study found the BB atenolol superior to captopril for self-reported anxiety [212]. However, BBs are known to affect somatic anxiety, so measuring anxiety might not be an appropriate proxy for mood in this case. In a double-blinded trial of 451 hypertensive patients taking either enalapril or the CCB amlodipine for 38 weeks, no differences were found between the two drugs in terms of quality of life measures [213]. Another 6-month double-blind trial with 540 hypertensive patients showed no superiority of cilazapril (an ACEI) over atenolol (a BB) [214]. Losartan was also not superior to nifedepine (a CCB) in a 12-week randomised double-blind trial with 223 hypertensive patients [215].

A growing body of evidence suggests a role for the angiotensin system in the pathophysiology of MDD. Drugs targeting the RAS reduce oxidative and inflammatory stress and enhance neurogenesis; all documented pathological markers in depression. Despite the heavy burden of depression, new drug development has been underwhelming. While RCTs providing definitive proof are yet to come, available preclinical and clinical data suggest the potential antidepressant properties of ACEIs and ARBs. The search for novel, effective, safe anti-inflammatory drugs that act centrally in the brain are of fundamental interest. Future clinical trials targeting the brain angiotensin system are necessary to verify the usefulness of these agents in treating depression.