Antidepressant-like effect of losartan involves TRKB transactivation from angiotensin receptor type 2 (AGTR2) and recruitment of FYN

Renin-angiotensin system (RAS) is commonly associated to peripheral fluid homeostasis and cardiovascular function, but recent evidence has also drawn its functional role in the brain. RAS has been described to regulate physiological and behavioral parameters related to stress response, including depressive symptoms. Apparently, RAS can modulate levels of brain derived neurotrophic factor (BDNF) and TRKB, which are important to neurobiology of depression and antidepressant action. However, interaction between BDNF/TRKB system and RAS in depression has not been investigated before. Accordingly, in the forced swimming test, we observed an antidepressant-like effect of systemic losartan but not with captopril or enalapril treatment. Moreover, infusion of losartan into ventral hippocampus (vHC) and prelimbic prefrontal cortex (PL) mimicked systemic losartan effect, whereas K252a, a blocker of TRK, infused into these brain areas impaired the systemic effect of losartan. PD123319, an antagonist of AT2 receptor (AGTR2), infused into PL but not into vHC, also prevented systemic losartan effect. Cultured cortical cells of rat embryos indicate angiotensin II (ANG2) binding to AGTR2 activates TRKB, possibly by recruiting FYN, a SRC family kinase. The higher levels of AGTR2 in cortical cells were inverted after insult with glutamate, and under this condition an interaction between losartan and ANG2 was achieved. Occurrence of TRKB/AGTR2 heterodimers was also observed, since GFP-tagged AGTR2 co-immunoprecipitated with TRKB. Therefore, antidepressant-like effect of losartan is proposed to occur through a shift of ANG2 towards AGTR2, followed by coupling of TRK/FYN and putative TRKB transactivation. Thus, AGTR1 show therapeutic potential as novel antidepressant drug therapy.

ABSTRACT Renin-angiotensin system (RAS) is commonly associated to peripheral fluid homeostasis and cardiovascular function, but recent evidence has also drawn its functional role in the brain. RAS has been described to regulate physiological and behavioral parameters related to stress response, including depressive symptoms. Apparently, RAS can modulate levels of brain derived neurotrophic factor (BDNF) and TRKB, which are important to neurobiology of depression and antidepressant action. However, interaction between BDNF/TRKB system and RAS in depression has not been investigated before. Accordingly, in the forced swimming test, we observed an antidepressant-like effect of systemic losartan but not with captopril or enalapril treatment. Moreover, infusion of losartan into ventral hippocampus (vHC) and prelimbic prefrontal cortex (PL) mimicked systemic losartan effect, whereas K252a, a blocker of TRK, infused into these brain areas impaired the systemic effect of losartan. PD123319, an antagonist of AT2 receptor (AGTR2), infused into PL but not into vHC, also prevented systemic losartan effect. Cultured cortical cells of rat embryos indicate angiotensin II (ANG2) binding to AGTR2 activates TRKB, possibly by recruiting FYN, a SRC family kinase. The higher levels of AGTR2 in cortical cells were inverted after insult with glutamate, and under this condition an interaction between losartan and ANG2 was achieved. Occurrence of TRKB/AGTR2 heterodimers was also observed, since GFP-tagged AGTR2 co-immunoprecipitated with TRKB. Therefore, antidepressant-like effect of losartan is proposed to occur through a shift of ANG2 towards AGTR2, followed by coupling of TRK/FYN and putative TRKB transactivation. Thus, AGTR1 show therapeutic potential as novel antidepressant drug therapy.

INTRODUCTION
Renin-angiotensin system (RAS) functional role has been historically implicated in cardiovascular and fluid homeostasis. Firstly, precursor molecule angiotensinogen is cleaved by renin into angiotensin I, which is then converted into Angiotensin II (ANG2) by angiotensin-converting enzyme (ACE) 1 . Main actions of the ANG2 are mediated by angiotensin II receptors type 1 and 2 -AGTR1 and AGTR2, respectively 1 .
Other reports, however, has pointed out all components of renin-angiotensin being produced inside central nervous system (CNS) 2 . Thus, AGTR1 and AGTR2 in circumventricular organs and in cerebrovascular endothelial cells may respond to circulating ANG2 of peripheral origin, whereas receptor lying in neurons inside blood brain barrier respond to RAS of brain origin 2 .
AGTR1 and AGTR2 have been found expressed inside blood brain barrier structures such as hippocampus and frontal cortex 3,4 , both considered crucial limbic structures associated to the neurobiology of depression 5 .
In fact, several piece of evidences introduce ANG2 as a hormone regulator of peripheral and central physiological changes regarding stress exposure, including behavioral consequences. For instance, both acute and chronic stress increased ANG2 and AGTR1 expression levels in the hypothalamic-pituitary-adrenal axis (HPA axis) [6][7][8] . Moreover, candesartan (AGTR1 antagonist) treatment prevented stress effect of increasing pituitary adrenocorticotrophic and adrenal corticosterone hormone levels 9 , and treatment with ACE inhibitors (ACEi) or AGTR1 antagonists reversed or prevented animal behavioral responses to stress [10][11][12][13][14][15] . In the same way, animals lacking angiotensinogen showed antidepressant-like phenotype 16 .
The neurotrophin brain-derived neurotrophic factor (BDNF), found mostly in the central nervous system, is important for neural plasticity, including synapse formation, neuronal differentiation and growth 17 . Functional role of BDNF and its receptor (TRKB, tropomyosin-related kinase B receptor ) has been linked to the pathophysiology of several psychiatric disorders, such as depression, and to the mechanism of action of antidepressant drugs 18 . Apparently, RAS may modulate BDNF and TRKB brain levels. For instance, candesartan treatment prevented both infarct volume and neurological deficit in animals suffering middle cerebral artery occlusion while increased protein and mRNA levels of TRKB in the brain 19 . In addition, telmisartan (AGTR1 antagonist) chronic treatment was able to prevent retinal damage and decrease of BDNF levels, such as observed in diabetic animal model 20 . Valsartan, another AGTR1 antagonist, counteracted the consequences of stress on depressive and anxiogenic-like behavior, as well as on BDNF levels in hippocampus and frontal cortex 12 .
Moreover, some case reports describe relief of depressive symptoms in hypertensive patients treated with the ACEi captopril [21][22][23] .
Despite scarce evidence, it is plausible to consider that drugs acting on RAS promote antidepressant-like effects. However, such properties have not been linked to the modulation of BDNF/TRKB system. In this sense, the present work aimed at investigating behavioral effects of AGTR1 antagonist losartan and ACEi in a model predictive of antidepressant-like effect, i.e. forced swimming test, and the requirement of BDNF/TRKB for such effect. Since AGTR1 activation was related to brain injury 24 and activation of AGTR2 has been supposed to employ balancing neuroprotective outcomes, especially when AGTR1 are blocked 25,26 , we hypothesized that activation of AGTR2 could underlie the antidepressant-like effects of losartan. In vitro analysis from cultured cortical cells were also performed to provide a mechanistic insight to the behavioral data.

Animals
Male Wistar rats used in behavioral studies (250-350g) were housed in pairs in a temperature
After behavioral tests, rats were anesthetized with chloral hydrate and 200nL of methylene blue was injected through the guide cannula. The brains were removed and injection sites verified.
All histological sites of injection were inserted in diagrams ( Figure S2) based on the atlas of Paxinos and Watson 31 .

Forced Swimming Test (FST)
Animals were placed individually to swim in a Plexiglas cylinder (24cm diameter by 60cm with 28cm of water at 25±1ºC) for 15min (pretest). Twenty-four hours later, animals were replaced in the cylinder for 5min swim test session and immobility time was measured. Water was changed between each test. After swimming, animals were towel-dried before returning to home cages.
Test was videotaped and analyzed by a trained observer blind to treatment.

Overexpression of GFP-AGTR2
MG87 cells were transfected to express GFP-tagged AGTR2 using lipofectamine 28 . Briefly, at a confluence of 70%, the cells were incubated with a mixture of 2.5% lipofectamine 2000 (Thermo Scientific, #11668019) and 5ug/mL of the plasmid in OptiMEM medium. Following 48h after transfection, cells were treated, lysed and submitted for immunoprecipitation as described below.
Lysate was collected in a clean tube and incubated with chloroform for 3min at RT. After centrifuged at 15200g for 10min at 4 o C, the aqueous phase was mixed with isopropanol for 10min at RT and centrifuged at 15200g for 10min at 4 o C. The pellet was washed with 75% EtOH two times, than with 100% EtOH once, air-dried and dissolved in 20µL MQ-water.

Polymerase Chain Reaction (qPCR)
Concentration and purity of each RNA sample were determined using NanoDrop (Thermo   Chemiluminescence emitted after addition of ECL was detected by a plate reader (Varioskan Flash, Thermo-Fisher). Signal from each sample, discounted blank, was normalized and expressed as percentage of control-group.

Surface expression of TRK
Cells from rat E18 cortex were cultivated in 96-well plates as described above (DIV8).
Detection of surface TRKB was performed by ELISA (Zheng et al, 2008). Cells were fixed with 4%PFA for 20min at RT. After washing with PBS wells were blocked with 5% non-fat dry milk and 5% normal goat serum in PBS for 1h at RT. Then, primary antibody against extracellular portion of TRK (1:500, Santa Cruz, #sc8316) was incubated overnight at 4 o C. Following wash with PBS, cells were incubated with HRP-conjugated antibody (1:5000; Bio-Rad, #170-5046) for 2h at RT. Signal detected after addition of ECL, discounted blank, was normalized by the average of vehicle-treated samples and expressed as percentage of control-group.

Data Analysis
Statistical analyses were carried out using two-tailed Student's t-test, one-way analysis of variance (ANOVA) followed by Fisher´s LSD post-hoc test, or two-way ANOVA test. Criteria for statistical significance was p<0.05.

Interaction between losartan and TRK or AGTR2: in vivo data
In this experimental set, requirement of TRK and/or AGTR2 activation for antidepressant effect of losartan was examined. As shown in figure 2a-  antagonist PD123319 into (c) PL (n=7-9/group) attenuated the effect of systemically injected losartan, but no change was observed in vHV (n=4-6/group). Losartan was administered 1h before FST and either K252a or PD was bilaterally infused 20min before swimming test. Data are expressed as mean ± SEM of immobility time (s). *p<0.05 compared to ctrl/ctrl group, unless otherwise stated; #p<0.05 compared to ctrl/losartan group. @p=0.06 compared to ctrl/losartan group.

Interaction between TRK and AGTR2: in vitro data
Since losartan antidepressant-like effect may depend mutually on TRKB and AGTR2 signaling in PL, we hypothesized losartan treatment would allow a shift of ANG2 from AGTR1 towards AGTR2 to forward TRKB activation. In order to test this possibility, losartan, PD123319 or K252a (1 st factors) was added to the primary cell cultures previously to ANG2 (2 nd factor) and levels of pTRK were analyzed. In fact, two way ANOVA indicated an interaction between both factors (F 3,28 =6.77, p<0.05) and pairwise comparisons assert that both PD123319 and K252a, prevented ANG2 effect of increasing pTRK levels (Fisher's LSD, NS for both; Figure 3a) and no additive effect was observed in the combination of ANG2 and losartan. Next step was to check if ANG2 effect was dependent on BDNF release. With this intent, a soluble BDNF scavenger -TRKB.Fc (1 st factor) -was added to the medium of the cell culture before ANG2 (2 nd factor). Two way ANOVA indicated no interaction between both factors (F 1,20 =0.82, NS), suggesting an effect of ANG2 on pTRK levels is independent of BDNF release (Figure 3b).
To further investigate that, PD123319 (1 st factor) was added to culture medium before BDNF

AGTR2-dependent interaction with FYN, surface positioning of TRKB and co-immunoprecipitation of AGTR2 and TRKB
Inasmuch as ANG2 increased pTRK levels independent of BDNF release, we decided to verify if ANG2 and BDNF are able to influence TRK/FYN coupling, since FYN is described as a SRC member responsible for transactivation of TRK 33 . For this purpose, PD123319 (1 st factor) was added in the culture medium before ANG2 or BDNF (2 nd factors) and TRK/FYN coupling was analyzed. Interestingly, two-way ANOVA indicated a significant interaction between ANG2 and PD123319 (F 1,19 =5.08, p<0.05) and, surprisingly, between BDNF and PD123319 (F 1,20 =6.74, p<0.05). Respectively, pairwise comparisons showed PD123319 abolished ANG2 and BDNF effect of increasing TRK/FYN coupling (Fisher's LSD, NS for both; Figure 4a).
Next, we measured if PD123319 or ANG2 was able to modulate surface levels of TRKB. In fact, cultured cortical cells exposed to PD123319 presented decreased, while ANG2 increased, surface levels of TRKB (F 2,30 =24.33; Figure 4c).
Altogether, these results point to the existence of a heterodimer TRKB/AGTR2, since previous studies have described cross-antagonism (ability of both antagonists of each receptor units in the heterodimer to block signaling of each other agonist) as a fingerprint of heterodimerization 34,35 .
In agreement with this idea, as observed in Figure 4d, a labeled GFP-tagged AGTR2 was co-precipitated with TRKB, however no apparent effect of BDNF or ANG2 was found in the levels of such complex.

Challenge with glutamate inverts the ratio between AGTRs in cortical cells
Preliminary comparison between the presented in vitro and in vivo analysis indicates an incompatibility regarding the effects of losartan. This compound, although effective when injected systemic or into mPFC, did not exerted any effect per se in cultured cortical cells.
Therefore, we considered the putative role of pretest stress in animals. In this scenario, as previously described 36 , a single exposure to inescapable stress, in addition to a peak in corticosterone production (lasting for 2h), increases glutamate levels for up to 24h. Therefore, firstly we determined the levels of AGTR1 and AGTR2 mRNA in our cultured cells.  Figure 3e).  Preliminary analysis showed that AGTR2 mRNA levels is 5 times higher than AGTR1 in our primary cultures, and this ratio is inverted to 2.5 times more AGTR1 after an insult with glutamate, and this later feature seems to allow a cooperative effect of losartan and ANG2.

DISCUSSION
Using a model of retinal ischemia, it was observed increased expression of AGTR1 receptor mRNA peaked 12h after reperfusion, while the treatment with candesartan was able to prevent ischemia-induced glutamate release 45 . Taken together, these data indicate a possible positive feedback between AGTR1 signaling and glutamatergic transmission. Moreover, in line with our in vitro observations, the levels of AGTR1 was increased while AGTR2 was decreased in medulla of stress-induced hypertensive rats 46 . However, an opposed effect of glutamate on AGTR2 mRNA have also been described 47 . In this study, the insult with glutamate led to an increase in AGTR2 mRNA. The precise mechanism where stressful events or excessive glutamate release reduces the levels of AGTR2 receptors are still not comprehended and these apparent discrepancies could rest on methodological differences. For example, the culture method of Makino et al. rely on cortical cells cultivated for 14 days, supplemented with calf serum and mitosis inhibitors; while our cultures were serum-free (substituted by B27), cultivated for 8 days without any drugs to prevent cell proliferation.
In conclusion, as depicted in figure 5, according to our findings we speculate that losartan-induced antidepressant-like effect is possibly mediated by AGTR2 and TRKB transactivation in the mPFC. We disclose a previously unknown TRKB transactivation by AGTR2, involving recruitment of SRC family kinase FYN. Therefore, TRKB and AGTR2 could form a heterodimer that probably docks FYN kinase to promote a crosstalk, putatively inducing pTRKB/PLCγ1 signaling. Considering the high comorbidity between depression and cardiovascular disorders, drugs such as losartan, could be an interesting therapeutic tool or even available as a novel class of antidepressant drugs.