Angiotensin AT1 and AT2 receptors heteromer expression in microglia correlates with Parkinson’s disease progression in the hemilesioned rat model of the disease


 Background/Aims : The renin-angiotensin system (RAS) is altered in Parkinson’s disease (PD), a disease due to substantia nigra neurodegeneration and whose dopamine-replacement therapy, using the precursor levodopa, leads to dyskinesias as the main side effect. Angiotensin AT 1 and AT 2 receptors, mainly known for their role in regulating water homeostasis and blood pressure and able to form heterodimers (AT 1/2 Hets), are present in the central nervous system. We assessed the functionality and expression of AT 1/2 Hets in Parkinson Disease (PD). Methods: Immunocytochemistry was used to analyze the colocalization between angiotensin receptors, bioluminescence resonance energy transfer was used to detect AT 1/2 Hets. Calcium and cAMP determination, MAPK activation and label-free assays were performed to characterize signaling. Proximity ligation assays was used to quantify receptor expression in microglial cells and brain striatal slices. Results: We confirmed that AT 1 and AT 2 receptors form AT 1/2 Hets that are expressed in cells of the central nervous system. AT 1/2 Hets are novel functional units with particular signaling properties. Importantly, the coactivation of the two receptors in the heteromer reduces the signaling output of angiotensin. Remarkably, AT 1/2 Hets that are expressed in both striatal neurons and microglia show a cross-potentiation, i.e. candesartan, the antagonist of AT 1 increases the effect of AT 2 receptor agonists. In addition, the level of expression in the unilateral 6-OH-dopamine lesion rat PD model increases upon disease progression and is maximal in dyskinetic animals. Conclusion: The results indicate that boosting the action of neuroprotective AT 2 receptors using an AT 1 receptor antagonist constitutes a promising therapeutic strategy in PD.


Introduction
The renin/angiotensin system (RAS) is composed of enzymes that produce angiotensin (Ang) peptides and of cell surface receptors that convey cytocrin signals to achieve specific cell responses. There are two angiotensin receptors (AT 1 R and AT 2 R) that belong to the superfamily of G-protein-coupled receptors. RAS has been abundantly studied in the periphery, mainly in relation with the control of arterial blood pressure. However, different laboratories have provided solid evidence on the relevant role of RAS in the central nervous system (CNS). Ang is an important regulator of motor control, and AT 1 R and AT 2 R have been suggested as targets to combat Parkinson's disease (PD) and related conditions such as levodopa (L-DOPA)-induced dyskinesias [1,2].
Age is a main risk factor for sporadic PD, which is characterized by dysregulation of the dopaminergic function due to the death of dopaminergic neurons of the substantia nigra (SN). A local RAS has been reported in the SN [3,4], in which overactivity of AT 1 R correlates with aging-related alterations, neuronal death [5,6] and neuroinflammation [7,8]. Microglial cells are the main mediators of neuroinflammation and despite once activated they are considered as detrimental, it is now known that they may undertake the pro-inflammatory (M1) or the neuroprotective (M2) phenotype. The search for pharmacological tools targeting GPCR to convert M1 into M2 phenotype is an active field of research [9]. The role of AT 2 R and the interplay between the two receptors in the above-mentioned changes due to Ang action in the aged or in the pathological brain is still unclear.
The cognate proteins for coupling to AT 1 R and AT 2 R are, respectively, Gq (also Gi) and Gi.
Accordingly, agonists of AT 1 R may mobilize calcium ion from intracellular stores, whereas agonists of AT 2 R decrease the activity of adenylyl cyclase thus depressing the cAMP/PKA signaling (https://www.guidetopharmacology.org). Interestingly, the two receptors may interact, leading to the formation of receptor heteromers with particular properties: pharmacological, functional or both [9,10]. On the one hand, heteromerization modifies receptor trafficking and ß-arrestin recruitment [10].
On the other hand, Ang II induces the formation of heteromers of the two receptors (AT 1/2 Hets) in luminal membranes of kidney tubular epithelial LLC-PK1 cells. In these cells, the peptide activates a calcium channel, sarco/endoplasmic reticulum Ca 2+ -ATPase (SERCA), that in kidney cells participates in the control of blood pressure [11].
The main target for pharmacological anti-parkinsonian interventions is the striatum that receives the SN dopaminergic input needed for motor control. What it is important is to know is whether AT 1 and AT 2 interact in the CNS, which is their physiological function and how their expression is alter in the course of a neurodegenerative disease. Accordingly, the aims of this paper were to i) get further insight into the properties of AT 1/2 Hets in a heterologous expression system, ii) investigate the expression and function of AT 1 R, AT 2 R and AT 1/2 Hets in striatal neurons and iii) investigate the expression and function of AT 1 R, AT 2 R and AT 1/2 Hets in striatal microglia in resting and activated states. The results show that AT 2 R are expressed in neurons and in activated microglia where they interact with AT 1 R to form AT 1/2 Hets. Accordingly, a final aim was to discover differential expression of AT 1/2 Hets in microglia striatal samples from parkinsonian and dyskinetic animals.
To prepare mice striatal primary microglial cultures, brain was removed from C57BL/6 mice of 2-4 days of age. Microglial cells were isolated following protocols described elsewhere [12][13][14] and grown in DMEM medium supplemented with 2 mM L-glutamine, 100 U/ml penicillin/streptomycin, MEM Non-Essential amino acids preparation (1/100) and 5% (v/v) heat inactivated Fetal Bovine Serum (FBS) (Invitrogen, Paisley, Scotland, UK). Briefly, striatum tissue was dissected, carefully stripped of its meninges and digested with 0.25% trypsin for 20 min at 37 ºC. Trypsinization was stopped by washing the tissue. Cells were brought to a cell suspension by repeated pipetting followed by passage through a 100 µm pore mesh. Glial cells were resuspended in medium and seeded at a density of 1 × in situ proximity ligation assays (PLA) and in 96-well plates for mitogen-activated protein kinase (MAPK) experiments. Cultures were maintained at 37ºC in humidified 5% CO 2 atmosphere and, unless otherwise stated, medium was replaced once a week.
Parkinson's disease (PD) model generation, levodopa treatment and dyskinesia assessment All experiments were carried out in accordance with EU directives (2010/63/EU and 86/609/CEE) and were approved by the Ethical committee of the University of Santiago de Compostela. Animals, male Wistar rats, were divided into three groups as follows: non-lesioned rats, 6-hydroxydopamine (6hydroxy-DA)-lesioned animals receiving vehicle, and 6-hydroxy-DA-lesioned animals receiving a chronic treatment with levodopa. A total of 15 animals were used. Details of model generation, protocol of drug administration and behavioral analysis, performed by a blinded investigator, are given elsewhere [1,15]. In brief, surgery was performed on rats anesthetized with ketamine/xylazine (1% ketamine − 75 mg/kg-, and 2% xylazine − 10-mg/kg). Lesions were produced in the right medial forebrain bundle to achieve complete lesion of the nigrostriatal pathway. The rats were injected with 12 µg of 6-hydroxy-DA (to provide 8 µg of 6-hydroxy-DA free base; SigmaAldrich) in 4 µl of sterile saline containing 0.2% ascorbic acid.
Some animals were treated with levodopa by receiving a daily subcutaneous injection with levodopa methyl ester (6 mg/kg) plus benserazide (10 mg/kg) for 3 weeks (such treatment reliably induces dyskinetic movements). In order to discriminate dyskinetic from non-dyskinetic animals, the manifestation of levodopa-induced AIMS (abnormal involuntary movements) was evaluated according to the rat dyskinesia scale described in detail previously [15][16][17]. The severity of each AIM subtype (limb, orolingual, and axial) was assessed using scores from 0 to 4 (1, occasional, i.e., present < 50% of the time; 2, frequent, i.e., present > 50% of the time; 3, continuous, but interrupted by strong sensory stimuli; 4, continuous, not interrupted by strong sensory stimuli). Rats were classified as "dyskinetic" if they displayed a ≥ 2 score per monitoring period on at least two abnormal involuntary movement (AIM) subtypes. Animals classified as "non-dyskinetic" exhibited either no AIMs or very mild/occasional ones. Animals with low scores, i.e., either non-dyskinetic or dyskinetic, were discarded.

Fusion proteins
Human cDNAs for AT 1 , AT 2 and σ 1 receptors cloned into pcDNA3.1 were amplified without their stop codons using sense and antisense primers harboring either BamHI and HindIII restriction sites to amplify AT 1 R and AT 2 R or BamHI and EcoRI restriction sites to amplify σ 1 receptor. Amplified fragments were then subcloned to be in frame with an enhanced yellow fluorescent protein (pEYFP-N1; Clontech, Heidelberg, Germany) or a Rluc (pRluc-N1; PerkinElmer, Wellesley, MA) on the Cterminal end of the receptor to produce AT 1 R-YFP, AT 2 R-RLuc, AT 2 R-YFP and σ 1 R-RLuc fusion proteins.
Cell transfection HEK-293T cells were transiently transfected with the corresponding cDNA by the PEI (PolyEthylenImine, Sigma-Aldrich, St. Louis, MO) method [18,19]. Briefly, the corresponding cDNA diluted in 150 mM NaCl was mixed with PEI (5.5 mM in nitrogen residues) also prepared in 150 mM NaCl for 10 min. The cDNA-PEI complexes were transferred to HEK-293T cells and were incubated for 4 hours in a serum-starved medium. Then, the medium was replaced by fresh supplemented culture medium and cells were maintained at 37 ºC in a humid atmosphere of 5% CO 2 . 48 hours after transfection, cells were washed, detached, and resuspended in the assay buffer.

Bioluminescence resonance energy transfer (BRET) assays
For BRET assays, HEK-293T cells were transiently cotransfected with a constant amount of cDNA encoding for AT 2 R-RLuc (0.9 µg) and with increasing amounts of cDNA corresponding to AT 1 R-YFP (0. The analysis of cAMP levels was performed in HEK-293T cells transfected with cDNA for AT 1 (1 µg) and/or AT 2 (1 µg) receptors in primary cultures of striatal neurons or glia using the Lance Ultra cAMP kit (PerkinElmer). The optimal cell density to obtain an appropriate fluorescent signal was first established by measuring the TR-FRET signal as a function of forskolin concentration using different cell densities. Forskolin dose-response curves were related to the cAMP standard curve in order to establish which cell density provides a response that covers most of the dynamic range of cAMP standard curve. 2 hours before the experiment the medium was substituted by serum-starved DMEM medium. Cells (2,000 HEK-293T cells, 4,000 striatal neurons or glial cells by well in 384-well microplates) growing in medium containing 50 µM zardaverine were pre-treated with the AT 1 R or AT 2 R antagonists (Candesartan and PD123319 respectively) or the corresponding vehicle at 24 °C for 15 min, and stimulated with the AT 1 R and/or AT 2 R agonists (Ang II and CGP-42112A respectively) for 15 min before adding 0.5 µM forskolin or vehicle, and incubating for an additional 15 min period. After 1 hour, fluorescence at 665 nm was analyzed on a PHERAstar Flagship microplate reader equipped with an HTRF optical module (BMG Labtech). A standard curve for cAMP was obtained in each experiment.
Extracellular signal-regulated kinases 1/2 (ERK1/2) phosphorylation To determine ERK1/2 phosphorylation, 40,000 HEK-293T cells transfected with cDNA for AT 1 R (1 µg) and/or AT 2 R (1 µg) or 50,000 striatal neurons or glial cells primary cultures were plated in each well of transparent Deltalab 96-well microplates. Two hours before the experiment, the medium was substituted by serum-starved DMEM medium. Then, cells were treated or not for 10 min with the selective antagonists Candesartan or PD123319 in serum starved DMEM medium followed by 7 min treatment with the selective agonists Ang II and/or CGP-42112A. Cells were then washed twice with cold PBS before the addition of lysis buffer (15 min treatment). 10 µL of each supernatant were placed in white ProxiPlate 384-well microplates and ERK1/2 phosphorylation was determined using Z-planes with a step size of 1 µm were acquired. The Andy's Algorithm [21], a specific ImageJ macro for reproducible and high-throughput quantification of the total PLA foci dots and total nuclei, was used for data analysis.

Statistical analysis
The data in graphs are the mean ± SEM (n = 5, at least). GraphPad Prism software version 7 (San Diego, CA, USA) was used for data fitting and statistical analysis. One-way ANOVA followed by posthoc Bonferroni's test were used when comparing multiple values. When a pair of values were compared, the Student's t test was used. Significant differences were considered when the p value was < 0.05.

Functionality of AT 1/2 Hets in a heterologous expression system
Interactions between AT 1 and AT 2 receptors have been previously reported [10,11]. Hence, we first investigated whether in HEK-293T cells and in our assay conditions AT 1 and AT 2 receptors may form heteromers. We analyzed the colocalization of angiotensin receptors at the plasma membrane by using HEK-293T cells coexpressing AT 1 R and AT 2 R fused to, respectively, the yellow fluorescent protein (YFP) and Renilla luciferase (Rluc). The proper traffic of fusion proteins to the cell membrane was confirmed by immunocytochemical analysis (Fig. 1A-B). The high degree of colocalization between AT 1 R-YFP and AT 2 R-Rluc in the plasma membrane and in the cytosol is shown in yellow ( Fig. 1C). To know whether a direct interaction between AT 1 and AT 2 receptors is possible, BRET assays were performed in HEK-293T cells expressing a constant amount of a fusion protein consisting of AT 2 R and Renilla Luciferase (AT 2 R-Rluc) and increasing amounts of AT 1 R fused to YFP (AT 1 R-YFP).
The saturation curve in Fig. 1D indicates close proximity between the two Ang receptors. The BRET max and BRET 50 values were 42 ± 1 mBU and 6 ± 2, respectively. When HEK-293T cells were transfected with a constant amount of cDNA for σ1-Rluc and increasing amounts of cDNA for AT 1 R-YFP, a linear response was observed indicating a nonspecific interaction of this negative control (Fig. 1D). These results confirm that the two Ang receptors may form heteromers in living HEK-293T cells. The proper functionality of fusion proteins was confirmed by cAMP assays (data not shown). A schematic representation of the technique is shown in Fig. 1E.
To characterize the AT 1/2 Het functionality, signaling assays were performed in single-transfected HEK-293T cells and in cotransfected AT 1/2 Hets-expressing cells. Cytosolic calcium levels, cAMP determination, ERK1/2 phosphorylation and label-free DMR assays were performed after treatment with AT 1 R and/or AT 2 R ligands. Consistent with Gq coupling of AT 1 R, treatment of AT 1 R-expressing cells with Ang II led to a marked increase in cytosolic calcium levels. The effect was receptor-mediated, as it was blocked by candesartan, the AT 1 R antagonist ( Fig. 2A). In AT 2 R expressing cells, the AT 2 R agonist CGP-42112A did not induce mobilization of the ion. These results agree with AT 2 R not engaging Gq proteins. In cotransfected cells, CGP-42112A did not produce any effect but reduced the response peak produced by Ang II. Hence, within the AT 1/2 Het, AT 2 R stimulation inhibits the AT 1 receptor signaling. Unlike the selective AT 1 R antagonist, candesartan, which blunted the agonist effect, the selective AT 2 R antagonist, PD123319, potentiated the AT 1 R-mediated effect (Fig. 2C).
Consistent with AT 2 R coupling to Gi, CGP-42112A reduced the forskolin-induced cAMP cytosolic levels in AT 2 R single transfected cells. Interestingly, in AT 1 R expressing cells, Ang II treatment also reduced the forskolin-induced cAMP levels. These effects can be explained by AT 1 R coupling not only to Gq, but also to Gi proteins. These effects were receptor-mediated and specific as they were blocked by the corresponding antagonist ( Fig. 2D-E). Analysis of cAMP-PKA signaling in cotransfected cells is more complex. On the one hand, all agonists reduced forskolin-induced [cAMP] and each antagonist blocked the effect of the corresponding agonist. Simultaneous administration of Ang II and CGP-42112A did not result in an additive effect. On the other hand, the antagonist of the AT 2 R enhanced the effect induced by the AT 1 R agonist and vice-versa, the antagonist of the AT 1 R enhanced the effect of the AT 2 R agonist (Fig. 2F). These latter results fit with data from calcium release experiments assays where the antagonist of the AT 2 R potentiated the action of Ang II. Then, it seems that both angiotensin receptors regulate one another via heteromerization. The negative regulation can be reversed and the antagonist of one receptor can even reinforce the output due to activation of the partner receptor.
In AT 1 R-expressing cells a marked increase in agonist-induced ERK1/2 phosphorylation that was blocked by candesartan was observed, while in AT 2 R-expressing cells a mild non-significant effect was obtained by treatment with the agonist, CGP-42112A (Fig. 3A-B). Remarkably, AT 1/2 Hets expression in cotransfected cells led to a significant increase in ERK1/2 phosphorylation upon Ang II treatment. The selective AT 2 R agonist was still unable to produce any significant effect but, in combined treatments, it markedly reduced the effect induced by Ang II (Fig. 3C). Therefore, these results are in the same line with that observed in calcium release and cAMP accumulation, indicating that both receptors inhibit one another in the AT 1/2 Het. DMR was modified by agonists in both types of single-transfected cells. In AT 1 R-expressing cells the antagonist could not revert totally the effect of the agonist, indicating that part of the output signal is due to off-site action of Ang II (Fig. 3D). In contrast, the effect of CGP-42112A on AT 2 R-expressing was totally blocked by PD123319 (Fig. 3E). In cotransfected cells, the results indicate that both angiotensin receptors show a characteristic signal that is blocked by selective antagonists. However, coactivation in these cells produced an important decrease in the signal. This effect is named negative crosstalk and it is similar to that observed in MAPK phosphorylation assays (Fig. 3E). The enhancement by candesartan of the effect of CGP-42112A, the AT 2 R agonist, was also noticed.
In summary, these results indicate a particular functionality of the AT 1/2 Het, namely Ang is able to engage two different signaling pathways. This dual effect is regulated by activation of the partner receptor within the AT 1/2 Het as i) activation of AT 1 R blocks AT 2 R agonist induced effect and vice-versa and ii) AT 1 R antagonist releases the inhibition exerted by AT 2 R over AT 1 R and vice-versa.

Functionality of AT 1 R and AT 2 R in primary cultures of striatal neurons
On the basis of the above-described relevance of Ang receptors in motor control and as potential targets to combat PD, we investigated Ang receptor-mediated signaling in primary cultures of neurons isolated from mouse striatum. Reduction of forskolin-induced [cAMP] increases were obtained using Ang II, while no sign of Gi-coupled AT 2 R was detectable (Fig. 4A). Interestingly, in simultaneous treatment with agonists, a lower decrease of forskolin-induced [cAMP] indicated the presence of AT 2 R whose activation reduced AT 2 R-mediated signaling. In the presence of the AT 1 R antagonist, candesartan, the AT 2 R agonist, CGP-42112A, induced a significant decrease in forskolin-induced cAMP levels. These results indicate that in the AT 1/2 Het, the AT 2 R signal is inhibited by AT 1 R expression and this effect is counteracted by AT 1 R antagonists. Similar results were confirmed in experiments of ERK1/2 phosphorylation determination in cultured striatal neurons, where cells responded (moderately) to Ang II and non-significantly to CGP-42112A (Fig. 4B). However, candesartan pretreatment potentiated AT 2 R signal. In agreement with the results in HEK-293T cells, cotreatment with agonists induced a lower signal to that induced by angiotensin II, indicating a negative crosstalk effect in both cAMP and MAPK phosphorylation signaling pathways. Overall, these cells express AT 1/2 Hets with similar functional characteristics to that observed in cotransfected HEK-293T cells.
Functionality of AT 1 R and AT 2 R in primary cultures of striatal microglia. Expression of AT 1/2 Hets.
Before assessing the functionality of Ang receptors in resting and LPS + IFNγ activated microglia isolated from striatum, we used in situ PLA to assess the expression of AT 1/2 Hets; PLA is instrumental to detect clusters of interacting proteins in natural sources. Figure 5B shows that resting cells have a few number of red dots due to AT 1/2 Hets, while the negative control show a negligible number of red dots (Fig. 5A). Interestingly, the red label was markedly increased in activated cells, as shown in Fig. 5C and in the bar graph of Fig. 5D indicating that AT 1/2 Hets expression increases in activated microglia. We then performed cAMP level determination and ERK1/2 phosphorylation assays in resting and activated microglial cells. The functionality of AT 1 R, AT 2 R and/or AT 1/2 Hets in resting cells was very low. But in cAMP assays, pretreatment with candesartan potentiated AT 2 R induced signaling, and PD123319 potentiated AT 1 R functionality (Fig. 5E, G). In contrast, the angiotensin-receptor-mediated signaling was more robust in LPS + IFNγ activated cells (Fig. 5F, H). In cAMP assays, the agonist of the two receptors reduced the forskolin-induced levels of this second messenger, although receptor costimulation did not lead to any additive effect. As it occurred in HEK-293T cells and neuronal primary cultures, antagonist pretreatments potentiated the partner receptor signal. On the other hand, the agonist of any of the two receptors activated the MAPK signaling pathway, while simultaneous stimulation completely blunted the ERK1/2 phosphorylation effect in activated microglia. In this signaling pathway antagonists blocked the cognate receptor and did not potentiate the activation of the partner receptor in the heteromer. Taken together, these results indicate that AT 1/2 Hets are significantly expressed in activated microglia showing the same properties than those displayed in heterologous system. Expression AT 1/2 Hets in the striatum of parkinsonian and dyskinetic rats.
Striatal sections of well-characterized 6-OH-dopamine hemilesioned rats, treated or not with levodopa (L-DOPA) and divided into dyskinetic or resistant to dyskinesia were prepared as described in methods. PLA assays were performed simultaneously and in identical conditions to detect the occurrence and the amount of AT 1/2 Hets. A representative image of each of the conditions is shown in all the drugs are able to cross the blood-brain barrier complicates data analysis. Anyhow, some of the reported results appear promising and using ß-blockers as a control and 65001 hypertensive patients for > 4 years [22] showed that the use of certain antihypertensives (angiotensin receptor blockers included) associates with reduced PD risk. The still unsolved question is whether such drugs may be of benefit in delaying the progression of the disease.
First of all, it has been described that the global effects of angiotensin on AT 1 and AT 2 receptors are opposite. In several tissues overactivity of the AT 1 receptor has been linked to aging-related proinflammatory changes [7,23]. Different mechanisms have been proposed to explain counteracting effects of AT 2 R on AT 1 R signaling (see [24][25][26]for review). Although the issue is complex, the expression of AT 2 R in brain suggests a relevant role in the regulation of neuroinflammation. On the one hand, as earlier indicated, the two angiotensin receptors may interact leading to receptor heteromers with particular properties such as regulating SERCA activity [10,11]. On the other hand, several heteromers formed by angiotensin receptors have been described. Among other examples, [27] it has been reported that apelin and AT 1 receptors interact and that the resulting heteromers mediate the apelin inhibition of AT 1 R-mediated actions. MAS protooncogene, the novel player in AT research, rescues a defective AT 1 R, likely by forming heterodimers [28]. The discovery of interaction between AT 1 R and the most abundant receptor in the CNS, namely the cannabinoid CB 1 receptor, has led to hypothesize that these functional units may lead to pathogenic actions when AT is produced.
However, the potential toxicity was studied in hepatic cells in relation to ethanol consumption but it was not assayed in neuronal cell models [29]. The AT 1 R may form complexes with a variety of adrenergic receptors [30,31] although their relevance to CNS physiology seems scarce. Less data are available for the AT 2 R receptor that may interact (in the periphery) with MAS and with bradykinin B2 receptors [25,[32][33][34]. In addition, AT 1/2 Hets may form dimers that, may eventually interact with MAS or with bradykinin B2 receptors in cells in which three of these receptors are expressed together [35,36]. One of the relevant results in this study is the demonstration in a rat model of microglial AT 1/2 Hets expression and its upregulation correlating with PD progression.
What is relevant for PD pathophysiology and disease progression is to delay the death of the approximately 30% nigral dopaminergic neurons that are left at the time of PD diagnosis. Glia in general, and microglia in what concerns to neuroinflammation, are key to preserve neurons from death. AT 1 Rs have been proposed as targets to reduce chronic neuroinflammation [37][38][39] as that occurring in PD. The general idea is that activation of the AT 1 R is detrimental, for instance by a microglia-mediated enhancement of neuronal loss in status epilepticus induced in rats [40]. In a previous study performed in the SN of rats we showed that angiotensin-induced Rho-kinase activation was involved in NADPH-oxidase activation, which, in turn, was involved in angiotensininduced Rho-kinase activation [41]. In addition, a prevention of astrocyte activation and promotion of hippocampal neurogenesis has been attributed to AT 1 R antagonization and subsequent prevention of NFкB and MAP kinase signaling and activation of Wnt/β-catenin signaling [42]. In our experimental conditions performed with already activated striatal microglia, MAP kinase signaling is suppressed by combined treatment with AT 1 and AT 2 receptor agonists. In the retina, AT 1 R activation results in regulating microglial activation thus suggesting that AT may have important implications in diabetic retinopathy [43]. To our knowledge the expression of AT 1/2 Hets in retinal cells has not been addressed yet. In agreement with the occurrence of a RAS opposite arm consisting of AT 2 R (also of Mas receptor) [44], AT 2 R activation attenuates microglial activation in an autoimmune encephalomyelitis rodent model [45]. Activation of the receptor is neuroprotective in a model of ischemia induced in conscious rats [46]. Similarly, the report by Bennion et al. [47] proves that AT 2 R activation in neurons and glial cells affords long-term neuroprotection in stroke, by both direct and indirect mechanisms. Recent studies show that the receptor prevents/attenuates pro-inflammatory microglial activity via protein phosphatase 2A-mediated inhibition of protein kinase C [48].
To our knowledge the AT 1 -bradikinin B2 heteroreceptor complex was the first to be associated to a peripheral-affecting disease. In fact the expression of the heteromer was increased thus mediating a higher Ang responsiveness in preeclampsia, a disease that markedly alters blood pressure in pregnancy [49]. The heteromer physiological function involves diverse signaling pathways and a variety of cells events such as phosphorylation of c-Jun terminal kinase and enhanced production of nitric oxide and a second messenger, cGMP [32]. An unbalanced proportion of AT 1 -bradikinin B2 receptor heteromers alters activation of cognate G-proteins and receptor desensitization [50,51]. We have collected data on differential expression of dopamine-receptor-containing heteromers in PD and the conclusion is that very often the expression of heteromers is altered in one or different stages of the disease. Usually the expression of those heteromers in dyskinesia is lower than in PD animals not displaying dyskinesia. Hence, this is the first example in which expression of heteromers that is already enhanced in parkinsonian conditions is further increased in non-dyskinetic conditions. These results are relevant as antagonists of angiotensin are considered as having potential to both improve PD symptoms and minimize levodopa-induced dyskinesias.
Limitations of the study are related with the variety of cell types in neurons and in glial cells. It is a challenge to know the relative expression of receptors in projection neurons, in choline acetyltransferase (ChAT), parvalbumin (PV), calretinin (CR) or nitric oxide synthase interneurons and in astroglia and microglia, which may be at different degrees of activation depending on the disease status. Furthermore, receptor functionality in response to a given receptor ligand may vary from cell to cell [52] and at onset of disease when comparing naïve versus levodopa-treated individuals.
But in any case, pharmacological manipulation of RAS components presents potential in PD [44]. One of the relevant findings in this paper is the insensitivity of AT 2 R to agonist treatment of striatal neurons. Indeed those neurons express both Ang receptor types as previously demonstrated [53].
However, it is more remarkable that it becomes functional in the presence of candesartan. There are few examples of similar findings and a recent one consists of progressive decrease of adenosine A 2A receptor functionality upon coexpression of another adenosine receptor, A 2B , and formation of A 2A /A 2B heteroreceptor complexes. This phenomenon is due to allosteric inter-protomer interactions in the heteromer, i.e. the presence of one receptor blocks the signaling of the partner receptor in the complex [54]. Interestingly, the presence of an antagonist could make, as in the case of Ang receptor heteromers in striatal neurons, appear AT 2 R functionality back. Therefore, AT 1 R antagonists in these neurons may achieve two benefits, which are repressing the detrimental actions mediated by the AT 1 R while making that the AT 2 R becomes functional and provides the benefits associated to its activation. Furthermore, in terms of looking for interventions to prevent PD disease progression there is enough information to agree that microglial cells may be key if there is a way to skew the physiology to acquire the M2 neuroprotective phenotype. On the one hand, neuronal alpha-synuclein produces an upregulation of AT 1 R while increasing in microglia the proportion of pro-inflammatory M1 versus neuroprotective M2 markers; accordingly it is suggested that antagonists already used in hypertension and able to cross the blood-brain barrier, may be repurposed for the therapy of PD [8].
On the other hand, microglial AT 2 Rs, constituent in AT 1/2 Hets, show promise as i) they are upregulated in both parkinsonian conditions and in levodopa-induced dyskinesias and ii) their activation is seemingly neuroprotective. Remarkably, AT 1/2 Hets do not show cross-antagonism, a property displayed by many heteromers and that would lead to a therapeutic dead end in terms of neuroprotection; instead the antagonist of one receptor releases the brake on activation of the partner receptor. Taken together, the opposite action of AT 1 and AT 2 receptors, their expression in microglia and the marked upregulation of AT 1/2 Hets lacking cross-antagonism but displaying antagonist-mediated cross-potentiation, suggest that interventions aimed at antagonizing central AT 1 Rs to potentiate AT 2 R-mediated actions may be neuroprotective in PD.

Conclusions
The present study demonstrates that AT 1 and AT 2 receptors form AT 1 / 2 Hets that are expressed in cells of the central nervous system. AT 1 / 2 Hets are novel functional units with particular signaling properties. Importantly, the coactivation of the two receptors in the heteromer reduces the signaling output of angiotensin. Remarkably, AT 1 / 2 Hets, which are expressed in both striatal neurons and microglia, show a cross-potentiation, i.e. candesartan, the antagonist of AT 1 increases the effect of AT 2 receptor agonists. In addition, the level of expression in the unilateral 6-OH-dopamine lesion rat PD model increases upon disease progression and is maximal in dyskinetic animals.
Importance and relevance of the study reported These findings reported are potentially important because they indicate that boosting the action of strategy in PD. The strategy may consist of designing AT 1 R receptor antagonists able to readily cross the blood-brain barrier and effective in releasing the brake on the AT 2 receptor. Compostela (#2016/0345) whose resolutions are supervised by regional and National regulatory bodies.

Competing interests
Author declares no conflict of interest Funding This work was supported by grants of the Spanish Ministry of Health (PI17/00828 and CIBERNED) and from the Spanish Ministry of Science, Innovation and Universities (RTI2018-098830-B-I00 and RTI2018-094204-B-I00; they include EU FEDER funds).
Authors' contributions RF, GN and JLL designed, supervised the work in the different laboratories and validated the data in the manuscript. RRS and IRR performed biophysical, signaling and immunohistochemical assays. RRS analyzed proximity ligation assay data. AM and AIRP did the lesions, characterized the different animal groups and prepared brain sections. RF wrote the first draft that was edited first by GN and JLL and, subsequently, by all co-authors.

Figure 1
Human AT1 and AT2 receptors interact in a heterologous expression system. Panels A-C: Immunocytochemistry assays were performed in HEK-293T cells expressing AT1R-YFP (1 µg cDNA), which was detected by its own yellow fluorescence (green), and AT2R-Rluc (1 µg cDNA), which was detected by a mouse anti-Rluc antibody and a secondary Cy3 anti-mouse antibody (red). Colocalization is shown in yellow. Cell nuclei were stained with Hoechst