Stress-induced microglial activation occurs through β-adrenergic receptor: noradrenaline as a key neurotransmitter in microglial activation

Background The involvement of microglia in neuroinflammatory responses has been extensively demonstrated. Recent animal studies have shown that exposure to either acute or chronic stress induces robust microglial activation in the brain. In the present study, we investigated the underlying mechanism of brain microglial activation by acute stress. Methods We first looked at the spatial distribution of the noradrenaline (NA)-synthesizing enzyme, DBH (dopamine β-hydroxylase), in comparison with NA receptors—β1, β2, and β3 adrenergic receptors (β1-AR, β2-AR, and β3-AR)—after which we examined the effects of the β-blocker propranolol and α-blockers prazosin and yohimbine on stress-induced microglial activation. Finally, we compared stress-induced microglial activation between wild-type (WT) mice and double-knockout (DKO) mice lacking β1-AR and β2-AR. Results The results demonstrated that (1) microglial activation occurred in most studied brain regions, including the hippocampus (HC), thalamus (TM), and hypothalamus (HT); (2) within these three brain regions, the NA-synthesizing enzyme DBH was densely stained in the neuronal fibers; (3) β1-AR and β2-AR, but not β3-AR, are detected in the whole brain, and β1-AR and β2-AR are co-localized with microglial cells, as observed by laser scanning microscopy; (4) β-blocker treatment inhibited microglial activation in terms of morphology and count through the whole brain; α-blockers did not show such effect; (5) unlike WT mice, DKO mice exhibited substantial inhibition of stress-induced microglial activation in the brain. Conclusions We demonstrate that neurons/microglia may interact with NA via β1-AR and β2-AR.

In our previous study, we demonstrated that stressinduced microglial activation occurs within 30 min of exposure to restraint/water immersion stress [57]. This indicates the involvement of fast signals, such as those conveyed by neurotransmitters.
NA is the best documented neurotransmitter in stress experiments. For instance, NA has been reported to increase in the brain in response to various types of stresses including immobilization, foot shock, tail pinch [17,44,62]. In addition, administration of β AR agonist, isoproterenol, significantly increased interleukin-1β (IL-1β) in the brain [30,74], and cultured microglia [64]. Besides, microglial activation induced by repeated social defeat is completely blocked by propranolol, an antagonist of β1 and β2 ARs (β1-AR and β2-AR) [72]. Furthermore, the induction of IL-1β in the hypothalamus (HT) by foot shock stress is blocked by propranolol [8,9], which also inhibits proinflammatory cytokine production in microglial cells isolated from rats [71]. Collectively, these results suggest that microglia may receive noradrenergic signals in stressed brains. Therefore, we hypothesize that the sympathetic nervous system, most likely noradrenergic neurons, may control the microglial activation status. Here, we demonstrate a possible mechanism for stress-induced microglial activation.

Animals and treatments
All procedures were approved by the Institutional Animal Care and Use Committee of the Nippon Medical School (permission no. 27-052; Tokyo, Japan) and were performed in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals, aiming to minimize the number of animals used and their suffering. Fischer rats (F344, males, 250-280 g), known to be a stress-sensitive strain, were purchased from Japan Laboratory Animals, Inc. (Tokyo, Japan). For stress experiments, the animals were restrained with wire nets for 1-4 h. Restraint stress (RS) was started at 10:00 a.m. and ended at 11:00 a.m., 12:00 a.m., and 2:00 p.m. (Fig. 1a). The rats were pretreated with the β-blocker propranolol (10 mg/kg) (P0884-1G; Sigma-Aldrich, St. Louis, MO, USA) and the α-blockers prazosin (0.5 mg/kg) (P7791-50MG; Sigma-Aldrich) and yohimbine (3.0 mg/kg) (Y3125-1G; Sigma-Aldrich), which were administered intraperitoneally, 1 h before the experiments. All animals were sacrificed immediately after the RS. Age-matched, unstressed animals, sacrificed immediately after being removed from the animal room, were used as controls. All animals were housed, with Fischer rats two to three and mice two to five in a cage, in a room with controlled temperature (21 ± 1°C), on a 12:12 light:dark cycle with lights on at 7:00 a.m, with food and water provided ad libitum access.
For immunofluorescence, the following primary antibodies and dilutions were used

Quantification of immunoreactivity and ISH
In order to quantify immunohistochemical signals and ISH, the mean optical densities (ODs), defined as the average of the ODs within the target area, were measured from each section using the image analysis software Win-ROOF (Mitani Corporation, Tokyo, Japan) [51].

Cell counting, cell size measurement, and Sholl analysis
The Iba1-immunoreactive (Iba1-ir) microglial cells of the hippocampus (HC) (comprising the unilateral DG area), the thalamus (TM), and the HT were counted in a 200 × 200 μm square using WinROOF. For the cell size measurement, all pixels with gray level values below the threshold value were treated as belonging to cell image, and other pixels were treated as background. The appropriate threshold value was determined as the level at which the binary overlay completely covered the entire cell body and processes. Cell surface area was measured using binary images of cells, microglia and astrocyte, with image analysis software (WinROOF). To further evaluate the morphological changes, we performed Sholl analysis for microglial cells as well as astrocytes [13]. Briefly, circles of diameter ranging from 0 to 40 μm, with 10 μm interval, were placed on the imaged cells, with each ring centered on the soma of each single cell. Intersections between the ring and the cell dendrites were counted using the image analysis software LuminaVision (Mitani Corporation, Tokyo, Japan), and the intersection count were averaged for all cells in images from each animal on different experimental conditions [67] (Fig. 6C).

Enzyme-linked immunosorbent assay (ELISA) for corticosterone measurement
Blood was collected into a tube containing EDTA. After 30 min of centrifugation, the plasma samples were stored at − 80°C until used for the measurements. Corticosterone levels were measured using an ELISA kit (Cayman Chemical Company, Ann Arbor, MI, USA).

Statistical analysis
Results are presented as the mean ± standard error of the mean (SEM). All statistical analyses were performed with SPSS software (IBM, Chicago, IL, USA). Statistical significance was determined by Student's t test and one-way and two-way analysis of variance (ANOVA). ANOVA was followed by Bonferroni's test for multiple comparisons. A level of p < 0.05 was considered statistically significant. All results were obtained from four rats and experiments under each condition.

Results
Monitoring stress levels with CORT in the plasma of rats exposed to acute stress First, the stress levels of Fischer rats exposed to acute RS were monitored via the quantification of plasma corticosterone, following the stress procedure (Fig. 1b).

DBH-ir neuronal fibers in the HT, TM, and HC
Microglia possess ARs [41,61], suggesting communication with noradrenergic neurons. Therefore, we investigated the immunoreactivity of DBH, the specific enzyme that yields NA from dopamine, in the locus coeruleus (LC), HT, TM, and HC.
Double IHC, using OX-42 and DBH, revealed that the fibers of DBH-ir neurons were extended, surrounding microglial cells in the HC, TM, and HT (data not shown). Furthermore, confocal immunofluorescence revealed that, in the HC, OX-42-ir microglia were meticulously surrounded by DBH-ir fibers (Fig. 3b, upper panels). Z-stack analysis, which investigates 3D structures, clearly demonstrated DBH-ir fibers framing the OX-42-ir microglial cells (Fig. 3b, lower panels). Furthermore, Z-stack analysis demonstrated the co-localization of DBH-ir fibers and OX-42-ir microglial cells in other regions such as HC and CCx (Fig. 3c). This finding was also confirmed with a different combination with Iba1-DBH staining which demonstrated the DBH-ir fibers surrounding Iba1-ir microglial cells in the HT (Fig. 3d).

Noradrenergic neuronal activation in the LC, but not in the SN
It is well known that the cell bodies, projecting DBH-ir axons to a variety of brain regions as shown in Fig. 2A, are located in the LC. To demonstrate the noradrenergic neuronal activation, we measured TH mRNA in the LC and SN. TH mRNA significantly increased in the LC following acute stress (n = 4, F(3,12) = 21.236, p < 0.001; one-way ANOVA) (Fig. 3e, f). In contrast to the LC, no significant change of TH mRNA levels was observed in the SN following acute stress (n = 4, F(3,12) = 1.126, p = 0.377; one-way ANOVA) (Fig. 3e, f). These results demonstrated the activation of noradrenergic neurons in the LC following the acute stress in Fischer rats.

ARs in the brain
Although both β-ARs (β1-AR, β2-AR, and β3-AR) and α-ARs (α1-AR and α2-AR) are expressed in cultured microglia, the former has been well demonstrated to induce c-AMP elevation [41,47,61]. We therefore investigated the expression and distribution of β-ARs in the brain.
RT-PCR analysis showed a clear band of β1-AR and β2-AR in each brain region, including the HC, TM, and HT. However, β3-AR was not detected in any brain region (Fig. 4a). ISH showed the expression of β1-AR and β2-AR, but not β3-AR, in the HT (Fig. 4b), which was further confirmed by IHC, showing the existence of β1-AR and β2-AR in glia-like cells (Fig. 4b).

β-Blockers significantly suppress Iba1-ir microglia
In order to study the functional involvement of β-ARs, the β-blocker propranolol (10 mg/kg) was administered intraperitoneally 1 h before each stress procedure (Fig. 5a) [72]. Corticosterone levels significantly increased during exposure to stress (n = 4, F(3,31) = 27.401, p < 0.001; two-way ANOVA), and no significant effect of drug treatment was observed (n = 4, F(1,31) = 0.575, p = 0.456; two-way ANOVA) (Fig. 5b). Furthermore, we evaluated cells morphology by employing parameters such as cell size, cell count, and intersections. In contrast to saline-treated rats, the microglial density looked sparse in the HT of propranolol-treated rats (Fig. 6A, B). The morphological activation of microglia differed significantly between saline-and propranolol-treated rats (n = 4, F(1,31) = 193.082, p < 0.001; two-way ANOVA). Propranolol significantly suppressed the microglial density at CTRL (n = 4, p < 0.01), 1 h RS (n = 4, p < 0.01), 2 h RS (n = 4, p < 0.01), and 4 h RS (n = 4, p < 0.01), compared with saline-treated rats (Fig. 7a). In addition, there was a significant effect of propranolol on the number of microglia (n = 4, F(1,31) = 374.025, p = 0.001; twoway ANOVA) in that the HT of propranolol-treated rats had a significantly lower microglial count at CTRL (n = 4, p < 0.01), 1 h RS (n = 4, p < 0.01), 2 h RS (n = 4, p < 0.01), and 4 h RS (n = 4, p < 0.01) than that of saline-treated rats (Fig. 7a). There was no significant effect of propranolol treatment on astrocyte cell size (n = 4, F(1,31) = 0.060, p = 0.809; two-way ANOVA) or cell count (n = 4, F(1,31) = 0.053, p = 0.819; two-way ANOVA) (Fig. 7b) Moreover, we analyzed morphological changes with Sholl analysis. As shown in Fig. 8a, there was some difference on the count of intersections between 10 and 20 μm on the distance from soma. For instance, at 10 μm from soma, there was no difference on the intersection numbers in the control conditions between saline and β-blocker treatment. In the stressed conditions, such as 1 h RS or 2 h RS, there was decrease on the intersection numbers both in saline and β-blockertreated rats. At 20 μm from soma, there were significant differences both in control and stressed conditions between saline and β-blocker treatment. In particular, in saline-treated rats, there was a significant decrease of the count of intersections on microglial cells in stressed conditions (n = 4, F(3,12) = 18.000, p < 0.001; one-way ANOVA). However, in propranolol-treated rats, there was no significant change between control and stressed conditions in propranolol-treated rats based on Sholl analysis (n = 4, F(3,12) = 0.425, p = 0.739; one-way ANOVA) (Fig. 8a, c). In fact, there was already a significant decrease of intersection counts even in control conditions (Fig. 8c). The decrease of intersections of microglia in control conditions in propranolol treatment is likely to be due to cellular shrinkage as shown in Fig. 6B, not to the hypertrophic changes.
Curiously, during the course of the experiment we found that there are two types of microglia: OX-42 + -Iba1 + and OX-42 + -Iba1 − (Fig. 6D). As to the cause behind the drastic decrease of Iba1-ir microglial cells following propranolol, we found no evidence of cleaved caspase-3 staining (6E), suggesting that the change may be not through apoptosis.

Discussion
The main finding of the present study was that microglial activation, as represented by enlarged Iba1-ir cell surface areas in the HT, HC, and TM, was significantly inhibited by pretreatment with the β-blocker propranolol. This finding was further confirmed by impaired, stress-induced, microglial activation in DKO mice, which contrasted sharply with that of WT mice.
Regarding the mechanism of neuronal-glial interaction, we demonstrated that fibers immunoreactive to DBH extend into wide brain regions, including the HC, TM, and HT. This result is consistent with those of previous studies which demonstrated extensive branched axons providing the main source of NA throughout the brain including neocortex, amygdala, cerebellum and spinal cord [37,40]. In addition to a variety of neuropeptides including neuropeptide Y, somatostatin, cholecystokinin, and galanin, the LC contains two enzymes, DBH and TH, which are critically involved in NA biosynthesis [5,51]. Therefore, it is crucial to know how DBH and TH respond to the acute RS. Importantly, we found the density of DBH-ir neurons in the LC increased significantly after 2 h RS (data not shown). In addition, mRNA of TH was shown to significantly Fig. 8 a, b Scholl analysis of microglia and astrocytes in the HT at CTRL, 1 h RS, and 2 h RS, respectively, as compared to β-blocker treatment. c, d Histograms demonstrating the intersections count of Iba1-ir microglia (c) and GFAP-ir astrocytes (d) at the distance of 20 μm from soma. The asterisks indicate a statistical difference between saline-treated and β-blocker-treated rats at each time point ( * p < 0.05, ** p < 0.01, n = 4) increase following the RS only in the LC. Thus, these results suggest the activation of the noradrenergic neurons. On the other hand, it has been reported that microglial cells possess receptors for NA, such as β1-AR, β2-AR, β3-AR, α1-AR, and α2-AR [41,61,72]. In the present study, confocal microscopic Z-stack analysis revealed that DBH-ir fibers edged microglial cells. In previous reports based on electron microscopy, it has been reported that noradrenergic synapses release NA into the extracellular fluid, diffusing the neurotransmitter into the surrounding synaptic clefts [2,3]. Moreover, confocal microscopy showed that both β1-AR and β2-AR, but not β3-AR, were co-localized with OX-42-ir microglia in the brain. Taken together, these results suggest that microglial cells possessing β1-AR and β2-AR may receive adrenergic signals from noradrenergic neurons.
In the present study, we investigated the role of β-ARs using the β-blocker propranolol prior to acute RS and observed that microglial activation triggered by acute RS was substantially inhibited in the HT, HC, and TM. In addition, pretreatment with propranolol significantly decreased the number of Iba1-ir microglial cells in those regions. These findings are consistent with those of previous studies demonstrating that propranolol inhibits microglial activation following various stresses, such as social disruption stress [72] and inescapable foot shock [30]. It was also reported that the stress-associated increase of IL-1β mRNA is inhibited by propranolol in the central nervous system [8]. Therefore, we suggest that stress-induced microglial activation may occur through β-ARs.
On the other hand, and against our expectations, microglia were further activated by pretreatment with the α2-AR blocker yohimbine. This effect was stronger than that of the α1-AR blocker prazosin. In this study, it was also found that the levels of plasma corticosterone were elevated in animals treated with those α-AR blockers, which supports the fact that the hypothalamic pituitary adrenal (HPA) axis may be activated by pretreatment with α-AR blockers. In fact, this finding is consistent with those of previous studies showing upregulated corticosterone levels in prazosin-or yohimbinetreated rats [38,55]. Importantly, the LC receives afferent neurons from the HT [5], causing NA to increase through the activated HPA axis. In addition, α2-AR blockade in chronic, unexpected, mild stress increases the level of NA in the brain [70]. It is well established that the α2-AR plays a role as a presynaptic inhibitory receptor regulating neurotransmitters' release [1,12,22]. Therefore, it is possible that the enhanced microglial activation by yohimbine treatment may be induced by the increased release of NA from the synaptic terminals. The results obtained with α-blockers, such as Fig. 9 a A schematic depiction of the α-blocker treatment protocol with prazosin and yohimbine. b Plasma corticosterone levels (ng/mL) of saline-and α-blocker-treated rats in acute RS. The asterisks indicate a statistical difference between saline-and yohimbine-treated rats following acute RS at each time point ( * p < 0.05, ** p < 0.01, n = 4). Results are presented as means ± SEM Fig. 11 a A schematic depiction of the stress protocol for WT and DKO mice. b Plasma corticosterone levels (ng/mL) of WT and DKO mice following acute RS (n = 4). Data are presented as means ± SEM prazosin and yohimbine, further suggest that stressinduced microglial activation may be achieved through NA upregulation.
Although the involvement of α-ARs in stress-induced microglial activation cannot be excluded, the effect of propranolol, a β1-AR and β2-AR blocker, was more predominant than those of blockers for α1-AR and α2-AR. Therefore, we used DKO mice lacking β1-AR and β2-AR in order to study the functional involvement of β1-AR and β2-AR in microglial activation. Microglial activation was first confirmed in WT mice exposed to 2 h RS. IHC demonstrated that, in DKO mice, the intensities of CD11b immunoreactivities, as a marker of morphological microglial activation in mice, were significantly suppressed in the HT, HC, and TM following acute RS. The level of corticosterone was relatively decreased in DKO mice, as compared to WT mice. Since corticosterone limits microglial activation [58], it is unlikely that the microglial suppression observed in DKO mice may have derived from the corticosterone levels. In addition, the stress-induced microglial activation observed in WT mice was significantly suppressed by propranolol. Intriguingly, there was no significant change in microglial morphology between WT and DKO mice under control conditions, suggesting that β-ARs may not be involved in regulation of resting microglia. As a whole, the present study demonstrates that deletion of β1-AR and β2-AR genes substantially suppresses acute stressinduced microglial activation in the brain. Thus, β1-AR and β2-AR may directly transmit signals, presumably NA, resulting in stress-induced microglial activation.
In several studies, the anti-inflammatory effects of NA have been reported, which contradicts the results of our study. For instance, in cultured microglia, NA and isoproterenol, a β-AR agonist, inhibit proinflammatory markers, such as IL-1β, IL-6, and iNOS mRNA, following LPS treatment (1.0 μg/mL), through the inhibition of NF-κB translocation [27]. In addition, NA depletion in aged rats' HC has been found to aggravate inflammation following LPS treatment (0.75 mg/kg) [6], whereas the proinflammatory effects of NA have been demonstrated in several studies. For instance, in cultured microglia, administering isoproterenol significantly increases the induction of IL-1β mRNA [64]. β2-AR stimulation leads to an increase of IL-1β and IL-6 [60]. These anti-and proinflammatory effects of NA comprise an intriguing paradox. One aspect common to the abovementioned studies is that experiments showing an anti-inflammatory effect of NA are mostly conducted in combination with LPS treatment, which induces NF-κB translocation into the nucleus. Unlike those experiments, we did not use LPS in the present study. Considering the existence of an alternative mechanism to the conventional PKA/NF-κBdependent pathway [47], such differential pathway might contribute to the differential effects of NA. It is conceivable that such alternative pathway may underlie the proinflammatory effects of NA.
Microglia play distinctive roles depending on their morphologies. Under resting conditions, microglia with ramified, long, thin processes play constructive roles, such as debris clearance, neuronal support, and synaptic monitoring and remodeling [14,15]. In contrast, under morphological activation, microglia become hypertrophic with short and thick processes and play various harmful roles, such as cytokine and superoxide release, neurotoxicity, and phenotypic polarization [10,18,36]. It is, therefore, critical to keep microglia quiescent and at a resting condition throughout the brain. Importantly, the β-blocker propranolol has been shown to effectively ameliorate neurodegenerative disorders, such as AD and PD. In particular, propranolol (5 mg/kg) has been shown to reduce cognitive deficits and amyloid and tau pathology in a model of AD [16]. In addition, microglial inhibition during sleep is reported to be critical for preventing cognitive dysfunction, via clearance of metabolic waste [4, 24-26, 43, 73]. Furthermore, β-blockers are reported to prevent anxiety-like behaviors through microglial inhibition [72]. Intriguingly, microglial inhibition by minocycline was found to lead to emotional stability [31]. Thus, controlling the microglial status may contribute to a variety of brain functions, ranging from cognition to mental activity. We therefore suggest that β-blockers, rather than α-blockers, are critical for maintaining the microglial status.

Conclusions
In the present study, we demonstrated that neurons/ microglia may interact with NA throughout the brain via β1-AR and β2-AR. NA may therefore be one of the neurotransmitters regulating microglial activation in the brain. β-blockers may effectively treat neurological disorders associated with microglial activation.