Stress-induced glucocorticoid signaling remodels neurovascular coupling through impairment of cerebrovascular inwardly rectifying K⁺ channel function

Significance

When neurons become active, they signal to local arterioles via intermediate glial cells, called astrocytes, to evoke dilation. This increases local blood flow and provides the oxygen and glucose necessary to support ongoing neuronal function. This process is termed neurovascular coupling. We demonstrate that chronic stress—which is a contributing factor for many diseases—impairs neurovascular coupling in the amygdala, a region involved in stressor processing. Our results further indicate that this dysfunction is due to the loss of arteriolar inwardly rectifying potassium (K⁺) channel function, which makes vessels less able to respond to vasodilatory K⁺ ions released by astrocytes during periods of increased neuronal activity. This neurovascular coupling impairment may contribute to the pathology of a range of brain disorders.

Abstract

Studies of stress effects on the brain have traditionally focused on neurons, without considering the cerebral microcirculation. Here we report that stress impairs neurovascular coupling (NVC), the process that matches neuronal activity with increased local blood flow. A stressed phenotype was induced in male rats by administering a 7-d heterotypical stress paradigm. NVC was modeled by measuring parenchymal arteriole (PA) vasodilation in response to neuronal stimulation in amygdala brain slices. After stress, vasodilation of PAs to neuronal stimulation was greatly reduced, and dilation of isolated PAs to external K⁺ was diminished, suggesting a defect in smooth muscle inwardly rectifying K⁺ (K_IR) channel function. Consistent with these observations, stress caused a reduction in PA K_IR2.1 mRNA and smooth muscle K_IR current density, and blocking K_IR channels significantly inhibited NVC in control, but not in stressed, slices. Delivery of corticosterone for 7 d (without stressors) or RU486 (before stressors) mimicked and abrogated NVC impairment by stress, respectively. We conclude that stress causes a glucocorticoid-mediated decrease in functional K_IR channels in amygdala PA myocytes. This renders arterioles less responsive to K⁺ released from astrocytic endfeet during NVC, leading to impairment of this process. Because the fidelity of NVC is essential for neuronal health, the impairment characterized here may contribute to the pathophysiology of brain disorders with a stress component.

Stressor exposure has been linked to detrimental health effects, and the activation of stress responses is recognized as a contributory factor in many pathologies. Chronic exposure to stressors can alter neuronal physiology, and signaling in stress-associated networks, including those underlying the processing of fearful and anxiogenic stimuli such as the amygdala, the bed nucleus of the stria terminalis (BNST), and the hippocampus, may become aberrant after chronic stress (1–3). Underlying this in part are morphological changes at the cellular level (including changes in dendritic arborization in hippocampal, BNST, and amygdalar neurons) (4) and altered expression of stress-related proteins, such as corticotropin-releasing factor (CRF) (5), tissue plasminogen activator (tPA) (6), and polysialylated neural cell adhesion molecule (7). It has been suggested that such changes contribute to psychopathologies including depression, anxiety, and posttraumatic stress disorder (8).

Stress also impacts the cardiovascular system, and a developing body of literature has linked repeated/chronic stress exposure to cardiovascular disease. Indeed, high perceived job stress is associated with an increased risk of adverse cardiovascular events (9), and psychological distress has been associated with an increased risk of fatal ischemic stroke (10), for which impaired cerebrovascular endothelial function resulting from enhanced glucocorticoid release may be a contributory factor (11). Recent reports have also explored the acceleratory effects of stress on the progression of dementia and Alzheimer’s (12–15)—disease states with an important cerebrovascular component.

Active regions of the brain require rapid and precise delivery of nutrients through a local elevation of blood flow, a phenomenon known as functional hyperemia. This moment-to-moment matching of blood flow to metabolic demand is ensured by the signaling mechanisms of neurovascular coupling (NVC). Astrocytes seem to play a central role in NVC by generating a calcium (Ca²⁺) wave in response to neuronal activity (16) that propagates to specialized endfoot projections, which almost completely encase intracerebral (parenchymal) arterioles (PAs) (17). This increase in endfoot Ca²⁺ evokes relaxation of juxtaposed arteriolar smooth muscle (SM) through the release of potassium (K⁺) ions and other vasodilators such as epoxyeicosatrienoic acids (EETs) (18). The elevation of K⁺ in the perivascular space activates SM inwardly rectifying K⁺ (K_IR) channels to cause PA vasodilation and increased blood flow (19). Because NVC provides a communicative bridge between the central nervous and cardiovascular systems—both of which are susceptible to stressor exposure—we reasoned that this vitally important process might also be affected by stress. Indeed, it is possible that prolonged periods of stressor exposure (e.g., 1 wk of chronic stress) could alter NVC in stress-affected regions, such as the amygdala, which responds to and is shaped by stressful events (20).

In this study we provide the first examination of NVC in the amygdala and explore the mechanistic basis underlying the impairment of this process by chronic heterotypic stressor exposure. This NVC impairment may be a contributory factor in pathologies with a stress component, such as depression, anxiety, posttraumatic stress disorder, and Alzheimer’s disease. These studies may therefore pave the way for the development of novel treatment options for such disorders.

Results

Heterotypic Stressor Exposure for 7 d Produces an Anxious Behavioral Phenotype.

During exposure to an established 7-d heterotypic stress paradigm (21), weight gain was significantly attenuated in stressed rats compared with control rats (Fig. 1A). One day after the final stressor, we analyzed behavior using the elevated plus maze, a test that capitalizes on the tendency of anxious rodents to prefer closed spaces to open spaces. Stressed rats spent significantly less time exploring the open arms of the elevated plus maze (Fig. 1 B and C). To rule out possible influences of changes in locomotion between groups (e.g., due to increased freezing behavior in stressed animals), we assessed the percentage of open arm entries relative to the total number of arm entries for each animal (Fig. 1D). This analysis indicated that, regardless of overall locomotor activity, stressed rats preferentially avoided the open arms of the maze during the 5-min testing period. Taken together, these data indicate that our stress paradigm produced an anxious behavioral phenotype.

Fig. 1.

Heterotypical chronic stress causes decreased weight gain and anxiety-like behavioral changes. (A) Weight gain over the 7-d stress paradigm was significantly attenuated in rats receiving stressors (control: 72 ± 3 g; stressed: 45 ± 2 g, n = 7 each). (B) Typical elevated plus-maze exploration paths over a 5-min period for a control rat (black, Left) and a stressed rat (blue, Right). (C) Time spent exploring the open arms of the maze during a 5-min test period was significantly reduced after repeated variate stress (control: 87 ± 9 s; stressed: 44 ± 16 s, n = 8 each). (D) Open arm entries expressed as a percentage of the total arm entries during a 5-min test period. Stressed rats chose to make fewer entries into the open arms (24% ± 5% of total entries were into open arms, n = 8) compared with controls (58% ± 4%, n = 8). Groups were compared using Student’s unpaired t test. *P ≤ 0.05, **P ≤ 0.001, ***P ≤ 0.0001.

Stress Impairs NVC in the Amygdala.

The morphology, neurochemistry, and neurophysiology of neurons in subregions of the amygdala are altered by chronic stress exposure in ways consistent with increased activation, and some of the behavioral consequences of chronic stress are thought to be mediated by these changes (2, 4). Because stress may lead to sustained activation of some amygdala subregions [e.g., lateral (2) and basolateral (22) nuclei], we hypothesized that stress might also affect arteriolar function and NVC in the amygdala. We thus chose to focus our studies on arterioles serving the amygdala, including the lateral, basolateral, cortical, central, and medial nuclei.

Arterioles within the brain normally exist in a partially constricted state, from which they dilate to deliver blood on demand. To simulate this, we preconstricted arterioles in amygdala brain slices with the thromboxane-mimetic 9,11-dideoxy-9α,11α-methanoepoxy prostaglandin F_2α (U46619; 125 nM); the degree of constriction produced by this agent was not affected by stress (Fig. 2E, Inset). We used electrical field stimulation (EFS) to depolarize neurons and initiate NVC. This manipulation evoked an increase in astrocytic endfoot Ca²⁺ that immediately preceded vasodilation (Fig. 2 C and D). In control arterioles, the EFS-evoked vasodilation was robust and sustained throughout the 1-min recording period after the onset of EFS. In contrast, vasodilation was both smaller in magnitude and shorter in duration in vessels in slices from stressed rats (Fig. 2 A–D and Movies S1 and S2). Indeed, we observed a 66% reduction in the peak arteriolar luminal diameter change evoked by EFS in slices from stressed rats compared with those from controls (Fig. 2E).

Fig. 2.

NVC is impaired by stress. (A) Representative baseline diameter and post-EFS peak diameter in an amygdala PA in a brain slice from a control rat. Red dashed lines indicate the edges of the vessel lumen. White measure bars highlight the change in diameter before and after EFS. Topographic coloration represents F/F₀ values of Fluo-4 AM fluorescence indicated by the chart at right, corresponding to astrocytic endfoot Ca²⁺. Both endfoot Ca²⁺ and vessel diameter increased in response to EFS. Because the endfoot Ca²⁺ wave precedes the peak of vasodilation, these images do not represent peak Ca²⁺ responses. This recording is shown in full in Movie S1. (B) As in A for a vessel from a stressed rat. See also Movie S2. (Scale bars in A and B, 10 µm.) (C) Averaged time-courses showing EFS-evoked vasodilations and endfoot Ca²⁺ transients in 31 slices from 21 control rats. EFS evoked an increase in endfoot Ca²⁺ followed by a robust vasodilation. Light blue rectangle represents the duration of the EFS pulse. Data are displayed as change in diameter/fluorescence relative to 20 baseline images. (D) As in C, for 36 slices from 23 stressed rats. After stress, the magnitude and duration of the vasodilation were decreased. (E) Summary data relating EFS-evoked vasodilation of PAs to their passive diameters. Tone evoked by preconstriction with 125 nM U46619 (Inset) was unchanged (control: 41% ± 2% tone, n = 31 vessels; stressed: 41% ± 2% tone, n = 36 vessels), indicating no difference in the vasodilatory capacity of arterioles between groups. Despite this, stressed arterioles dilated less than control arterioles (control: 38% ± 4% dilation relative to passive diameter, n = 31; stressed: 13% ± 2%, n = 36). Groups were compared using Student’s unpaired t test. (F) Summary data of astrocytic endfoot [Ca²⁺]_i before and after EFS calculated using the F_max equation (25). The evoked Ca²⁺ response in stressed astrocytic endfeet (644 ± 103 nM, n = 10) was larger than that in control astrocytes (328 ± 23 nM, n = 7). There was no significant difference in resting endfoot Ca²⁺ (control: 72 ± 11 nM; stressed: 105 ± 14 nM). Groups were compared using two-way ANOVA with post hoc Bonferroni’s multiple comparisons test. **P ≤ 0.01, ****P ≤ 0.0001.

Because the level of astrocytic endfoot Ca²⁺ is a key determinant of the neurovascular response (23), a reduction in EFS-evoked vasodilation in slices from stressed animals could reflect a blunted elevation of endfoot Ca²⁺. However, according to our Ca²⁺ analysis (SI Materials and Methods), the peak endfoot [Ca²⁺]_i evoked by neuronal stimulation was significantly higher after stress, and we observed a strong trend (P = 0.11) suggesting that resting endfoot [Ca²⁺]_i was also elevated (Fig. 2F).

Collectively, these observations suggest that, despite elevated endfoot Ca²⁺ signaling, communication between astrocytic endfeet and vascular myocytes is disrupted by stress.

Decreased K_IR Channel Activity Contributes to Neurovascular Coupling Impairment in Situ.

Barium (Ba²⁺) ions are relatively potent blockers of K_IR2 channel members (24). We previously reported that, in cortical brain slices, inhibition of K_IR channels with Ba²⁺ (100 μM) reduced NVC by ∼70% (19). In amygdala slices from control rats, 100 μM Ba²⁺ decreased the magnitude of EFS-evoked vasodilation by ∼50% (Fig. 3 A and C), but this manipulation had no effect on evoked dilations in slices from stressed rats (Fig. 3 B and C). These results indicate that a K_IR-dependent component of NVC is lost after 1 wk of stress. Ba²⁺ had no overall effect on U46619-induced tone (Fig. 3D).

Fig. 3.

Ba²⁺ inhibits NVC in brain slices from control, but not stressed, rats. (A) Averaged diameter time-courses from control rat brain slices indicating that 100 μM Ba²⁺ decreased the magnitude of EFS-evoked vasodilation (n = 7 paired experiments, 6 rats. Black line: control; red line: 100 µM Ba²⁺). Light blue rectangle represents the duration of the EFS pulse. Data are displayed as change in diameter relative to 20 baseline images. (B) As in A, for vessels in slices taken from stressed rats. The impaired vasodilation evoked by EFS was not affected by 100 µM Ba²⁺ (n = 6 paired experiments, 5 rats. Blue line: control; red line: 100 µM Ba²⁺). (C) Summary data showing the magnitude of EFS-evoked vasodilation relative to passive vessel diameter. In control brain slices, Ba²⁺ (100 µM) reduced the magnitude of EFS-evoked dilations from 36% ± 6% to 18% ± 6% of passive diameter (n = 7), but had no effect on the impaired dilations observed in stressed rats (control: 14% ± 3%; Ba²⁺: 13% ± 4%; n = 6). (D) In control and stressed rats, Ba²⁺ (100 µM) had no effect on tone in the presence of 125 nM U46619 (control: 39% ± 3%; control + Ba²⁺: 39% ± 6%; stressed: 42% ± 3%; stressed + Ba²⁺: 42% ± 5%; n = 6 or 7 paired experiments each). Groups were compared using Student’s paired t test. ***P ≤ 0.001.

Elevation of bath K⁺ to 10 mM also causes arteriolar dilation in slices through SM K_IR channel activation, without engagement of local neurons or astrocytes (19, 23). Raising K⁺ to 10 mM to directly test SM K_IR function dilated PAs by 52% in control slices but had little effect on arterioles in slices from stressed animals (Fig. S1). This finding indicates that PAs in situ from stressed animals are less responsive to small increases in extracellular K⁺ concentration ([K⁺]_o), despite a substantial vasodilatory reserve.

Responses to Extracellular K⁺ Manipulation Are Diminished in Isolated Parenchymal Arterioles from Stressed Rats.

To examine the effects of stressor exposure on the cerebral vasculature more closely, we explored the responses of isolated amygdalar PAs from stressed and nonstressed rats to elevated [K⁺]_o. Elevation of intravascular pressure to 40 mm Hg constricted arterioles from both control and stressed groups by 45% (Fig. 4B, Inset). In control arteries, elevating [K⁺]_o from 3 mM to 8 mM caused a substantial vasodilation, which was sustained during subsequent elevations to 15 and 20 mM. Further elevations in [K⁺]_o to 25, 30, and 60 mM resulted in substantial vasoconstriction (Fig. 4A, Upper, and Fig. 4B). In contrast, in arterioles from stressed rats, the magnitude of vasodilation evoked by increasing [K⁺]_o to 8 mM was significantly smaller than that in control arterioles, and a similar effect was seen at 15 mM. At 20 mM [K⁺]_o the responses of control and stressed arterioles were opposite, with controls exhibiting a substantial dilation and stressed arterioles exhibiting a modest constriction. Furthermore, the magnitude of vasoconstriction evoked by 60 mM [K⁺]_o was significantly reduced in arterioles from stressed rats (Fig. 4A, Lower, and Fig. 4B). These results indicate that the defect in the K_IR channel-mediated dilatory responses of PAs to elevated [K⁺]_o resides in the arterioles.

Fig. 4.

Isolated pressurized amygdalar PAs from stressed rats exhibit impaired vasodilation to increases in [K⁺]_o. (A) Typical traces demonstrating vascular diameter responses to raising [K⁺]_o. In control animals (black trace, Upper) increasing [K⁺]_o to less than 20 mM caused substantial vasodilation. Further increases in [K⁺]_o caused constriction. In stressed animals (blue trace, Lower), the vasodilatory response to increases in [K⁺]_o was impaired. At the end of each experiment, vessels were perfused with 0 [Ca²⁺]_o aCSF to obtain the passive diameter. (B) Summary data from 12 control vessels (12 rats) and 18 stressed vessels (12 rats). The vasodilation to 8 mM [K⁺]_o was significantly impaired in stressed vessels (control: 59% ± 10% dilation from baseline; stressed: 30% ± 7%); this same phenomenon was observed in a subsequent step to 15 mM [K⁺]_o (control: 57% ± 12% dilation; stressed: 19% ± 8%). At 20 mM [K⁺]_o, dilation was sustained in control vessels, whereas stressed vessels were slightly constricted relative to baseline (control: 23% ± 7% dilation; stressed: −5% ± 5% constriction). Upon raising [K⁺]_o to 25 mM, both control and stressed vessels were constricted to a similar extent (control: −10% ± 7%; stressed: −9% ± 5%). At 30 mM [K⁺]_o, controls were constricted by −34% ± 6% relative to baseline, and stressed vessels constricted by −19% ± 5%. Constriction to 60 mM [K⁺]_o was less in stressed vessels (−34% ± 6%) compared with controls (−60 ± 6%). Groups were compared using a two-way ANOVA with post hoc Bonferroni’s multiple comparisons test. (Inset) Myogenic tone developed in response to 40 mm Hg intravascular pressure was not affected by stress (control: 45% ± 3% tone, n = 12; stressed 45% ± 3%, n = 18, Student’s unpaired t test). *P ≤ 0.05. ***P ≤ 0.001.

Vasodilation of PAs to K⁺ is likely caused by activation of K_IR2.1, but not K_IR2.2, channels (25). Consistent with this, we observed a lower abundance of K_IR2.1 (but not K_IR2.2) mRNA in PAs after stress, indicating decreased expression of the K_IR2.1 channel gene (Fig. S2).

K_IR Channel Current Density Is Decreased in Isolated Myocytes from Amygdalar PAs of Stressed Rats.

An elevation of [K⁺]_o increases the K⁺ conductance of K_IR channels to drive the myocyte resting membrane potential (∼−35 mV) (26) to the new potassium equilibrium potential (E_K). Provided that E_K is negative to −35 mV (i.e., [K⁺]_o is <25 mM), elevation of [K⁺]_o will cause membrane potential hyperpolarization and thereby deactivate L-type voltage-dependent Ca²⁺ channels, decreasing Ca²⁺ influx to cause vasorelaxation (27). Our observations that PA dilations to small increases in [K⁺]_o are impaired and K_IR2.1 transcript is reduced in stressed animals suggested that myocyte K_IR channel function may be compromised by stress. To directly address this possibility, we measured whole-cell currents in response to a voltage ramp protocol in myocytes isolated from amygdalar PAs of control and stressed rats, using 60 mM [K⁺]_o to amplify inward K_IR currents (E_K = −23 mV). In stressed myocytes, inward currents were significantly smaller at voltages negative to E_K, consistent with a loss of K_IR channel currents (Fig. 5 A and B). However, outward currents at +40 mV (Control: 10.94 ± 1.21 pA/pF, n = 24; Stress: 8.03 ± 0.88 pA/pF, n = 13) were unaffected, suggesting a lack of an effect on voltage-dependent K⁺ currents. To test whether the K_IR current density differed between stressed and control rats, we superfused cells with 100 μM Ba²⁺. We observed that the Ba²⁺-sensitive current was reduced by 90% in stressed myocytes compared with controls (Fig. 5 C and D), indicating that stress reduces the number of functional K_IR channels in the SM membrane of PAs.

Fig. 5.

PA myocyte K_IR current density is reduced in stressed animals. (A) Representative raw current density traces from a control (black line) and a stressed (blue line) SM myocyte recorded during a voltage ramp from −140 to +40 mV. (B) Summary data for raw current density at −100 mV. Current density in myocytes from stressed animals (−4.26 ± 0.69 pA/pF, n = 12 observations from 6 rats) was significantly lower than that in myocytes from control animals (−7.76 ± 1.11 pA/pF, n = 20 observations from 11 rats). (C) Representative traces of Ba²⁺ (100 µM)-sensitive currents from a control (pink line) and a stressed (red line) myocyte between −140 and −20 mV. (D) Summary of K_IR current density in control and stressed SM myocytes at −100 mV. Ba²⁺-sensitive currents were substantially smaller in SM cells from stressed animals (−0.59 ± 0.30 pA/pF, n = 6 observations from 5 rats) than in their control counterparts (−6.09 ± 1.67 pA/pF, n = 10 from 8 rats). Groups were compared using Student’s unpaired t test. *P ≤ 0.05.

Glucocorticoid Signaling Mediates NVC Impairment by Stress.

It is well established that exposure to a range of stressors evokes a rapid and sustained increase in circulating corticosterone, and this has been linked to changes in neuronal function in the amygdala (28). This corticosterone response may habituate in response to chronic homotypic stressor exposure, but habituation does not occur in response to the presentation of heterotypic stressors, such as those used in our paradigm (29) (Fig. S3).

To exert effects on neurons, corticosterone must first pass through the endothelial and SM cells of the vasculature, where it may interact with ubiquitously expressed glucocorticoid receptors (GRs) (30, 31). Therefore, we hypothesized that NVC impairment is mediated by stress-evoked corticosterone–GR signaling. To test this hypothesis, we sought to mimic the stressor-induced increase in corticosterone (in the absence of stressor delivery) by injecting corticosterone s.c. at a dose (2.5 mg/kg) (32) designed to match the increase in circulating corticosterone observed after exposure to the type of stressors used in our paradigm (Fig. S3). Corticosterone delivery once per day for 7 d in lieu of stressor exposure produced an impairment of NVC similar to that produced by stress. Under these conditions, EFS-evoked vasodilations were reduced by ∼43% compared with those observed in slices from rats treated with vehicle (Fig. 6 A and B). Vehicle delivery itself had no effect on the magnitude of EFS-evoked vasodilations compared with responses evoked in slices from noninjected animals (Figs. 2E and 6B). The level of preconstriction induced by U46619 was the same in vessels from both vehicle-treated and corticosterone-treated rats (Fig. 6B, Inset).

Fig. 6.

Corticosterone signaling via GRs is a key mediator of NVC impairment. (A) Averaged diameter time-courses for vessels in slices from rats treated daily with s.c. corticosterone (cort) at 2.5 mg/kg (n = 10 slices, 6 rats; blue line) or vehicle (n = 12 slices, 6 rats; black line) for 7 d instead of stressor exposure. Light blue rectangle represents the duration of the EFS pulse. Data are displayed as change in diameter relative to 20 baseline images. (B) Summary data showing the magnitude of EFS-evoked vasodilation relative to vessel passive diameter. Vehicle treatment had no effect on the magnitude of vasodilation evoked by EFS (39% ± 6%, n = 12), whereas cort reduced the magnitude of EFS-evoked dilation to 22% ± 4% of passive diameter (n = 10). (Inset) Tone in the presence of 125 nM U46619 was unaffected by vehicle or cort treatment (vehicle: 44% ± 4% constriction, n = 12; cort: 38% ± 4, n = 10). (C) As in A, for vessels in slices from rats treated daily with i.p. RU486 at 30 mg/kg (n = 9 slices, 4 rats; purple line) or vehicle (n = 11 slices, 4 rats; blue line) 1 h before stressor exposure. (D) As in B, for vehicle- or RU486-treated rats. Vehicle-treated rats developed NVC impairment in response to stressors (15% ± 3% dilation relative to passive diameter, n = 11), an effect that was prevented by RU486 treatment (38% ± 6% dilation, n = 9). (Inset) Tone in the presence of 125 nM U46619 was unaffected by vehicle or RU486 treatment (vehicle: 47% ± 3% constriction, n = 11; RU486: 46% ± 6%, n = 9). Groups were compared using Student’s unpaired t test. *P ≤ 0.05, **P ≤ 0.01.

We also examined the effect of inhibiting GR signaling with the antagonist RU486. Intraperitoneal injection of 30 mg/kg RU486 1 h before stressor exposure abolished the effect of stress on NVC. In slices from treated animals, EFS evoked a vasodilation (Fig. 6 C and D) that was no different from that evoked in normal control animals (Fig. 2 C and E). In contrast, EFS-evoked vasodilations in slices from animals that received vehicle 1 h before stressor exposure were significantly smaller than those in slices from animals treated with RU486 (Fig. 6 C and D) and were comparable to responses evoked in slices from noninjected stressed rats (Fig. 2E). U46619-induced preconstriction was not different between groups (Fig. 6D, Inset).

Discussion

Traditionally, studies of the effects of stress on the central nervous system have focused on neurons. Here we demonstrate a previously unidentified relationship between stressor exposure and the astrocytes and vascular myocytes of the amygdalar neurovascular unit, and provide evidence that NVC is compromised by stress. Our results show that stress leads to a glucocorticoid-mediated disruption of K⁺-driven PA vasodilation during NVC, through impaired SM K_IR channel function. These studies unveil a multifaceted effect of stress on the cells of the neurovascular unit and suggest that blood flow to the amygdala may be compromised during NVC in the stressed brain.

Using a previously characterized chronic stress paradigm (21) that produced an anxiety-like behavioral phenotype, we found that stress caused a robust impairment of NVC and made the key observation that EFS-evoked vasodilation is attenuated in brain slices from stressed rats (Fig. 2). This effect was not due to a blunted elevation of endfoot Ca²⁺ in response to neuronal activity; in fact, EFS elicited an enhanced astrocytic endfoot Ca²⁺ response after stress. These larger magnitude endfoot Ca²⁺ transients in stressed slices may be related to an increase in amygdalar neuronal activity in chronically stressed animals (2, 22). Our previous work showed that elevation of astrocytic endfoot Ca²⁺ by EFS or Ca²⁺ uncaging can cause either dilation or constriction, and that the polarity of the response is determined by the endfoot Ca²⁺ concentration (23). In that study, we did not observe attenuated dilations at endfoot Ca²⁺ levels intermediate between those causing dilation (mean 350 nM) and constriction (mean 730 nM), that is, the response was bimodal (23). Thus, at the peak endfoot Ca²⁺ measured after stress in this study (644 nM), we would have predicted a vasoconstrictor response, but we actually observed blunted vasodilations. This finding may suggest that the Ca2²⁺-dependent release of vasoactive substances from endfeet is altered by stress and furthermore, that the signal transduction machinery of the underlying SM may be affected.

Our observation that vasodilatory responses to small increases in [K⁺]_o—mediated by the activation of myocyte K_IR channels—were severely reduced in isolated vessels prompted us to investigate the effects of stress on this channel in more detail. We found that stress down-regulated PA K_IR2.1 mRNA expression, an effect that was accompanied by a sevenfold reduction in K_IR channel current density in PA myocytes isolated from stressed rats. This latter effect is similar to that induced in healthy SM by pharmacological blockade of K_IR channels with Ba²⁺ (Fig. 5D and ref. 33), and this maneuver normally inhibits EFS-induced vasodilation in slices (Fig. 3A and ref. 19). However, we found no effect of Ba²⁺ ions on the residual EFS-evoked vasodilations in brain slices from stressed rats. Moreover, dilatory responses to raising [K⁺]_o were significantly blunted in slices from stressed rats compared with those in controls. Taken together, these observations are consistent with a model in which K_IR channel function is impaired after stress.

We then sought to further define the molecular pathway linking stress and NVC impairment. Stress-related neurotransmitter and/or neuropeptide release (for example, CRF and tPA) (34), inflammatory processes (35), sympathetic nervous system activation, and/or activation of the hypothalamic–pituitary–adrenal (HPA) axis (36) could each contribute to the NVC impairment we observed. Activation of the HPA axis involves the secretion of adrenocorticotropic hormone from the anterior pituitary, which results in the downstream release of glucocorticoids from the adrenal cortex (36). These glucocorticoids can interact with mineralocorticoid receptors (MRs) and GRs to mediate their effects. MRs are almost saturated at basal corticosteroid levels (37), whereas the lower affinity GRs represent a substantial receptor pool that is available for mediating the effects of the increased circulating steroids resulting from stressor exposure (37). In the brain, GRs are broadly expressed in neurons in a range of subregions, including the amygdala (38, 39), and are also expressed to a lesser extent in astrocytes (39). GRs are also found in SM (30) and endothelial cells (31). Because many effects of stressor exposure are mediated by glucocorticoid signaling through GRs, we chose to study the effects of manipulating GR signaling and corticosterone—the principal glucocorticoid hormone in rat—on NVC. Not only were we able to mimic NVC impairment by delivering corticosterone in place of stressors, but we were also able to abrogate the effects of stress on NVC by antagonizing GRs during stressor exposure. These data thus imply that corticosterone signaling via GRs is a key mediator of NVC impairment after chronic stress. The simplest interpretation of these results is that corticosterone exerts a direct effect on abundant SM GRs (30). Upon binding corticosterone, GRs translocate to the nucleus, where they mediate transactivation or transrepression of target genes (36). According to our data, it seems likely that, after stress, corticosterone/GR complexes in the SM directly down-regulate K_IR2.1 gene expression, leading to fewer functional channels in the myocyte membrane. However, the transcriptional regulation of other genes whose products interact with the K_IR channel is also possible.

Collectively, our observations suggest a model (Fig. S4) in which stress-induced glucocorticoid signaling causes a decrease in functional K_IR channels in the SM membrane. This impairs the ability of the vessel to respond to local increases in [K⁺]_o during NVC, leading to suboptimal vasodilation, which could lead to decreased regional cerebral blood flow in stressed animals during a hyperemic response. Although our results implicate this mechanism as a major contributor to NVC impairment after chronic stress, it remains possible that interactions of corticosterone with GRs in other cell types in the neurovascular unit could lead to further remodeling of neurovascular communication, and other NVC mediators could also be altered. For example, EETs signaling, which plays a role in NVC under normal circumstances (18), could be affected by stress through effects on the expression and/or function of the enzymes responsible for the production of these molecules. Another unexplored possibility is that NVC in brain regions other than the amygdala may also be impaired. These remain questions for in-depth future studies.

Accumulating evidence implicates impaired NVC and cerebral blood flow in a range of pathologies, many of which have a stress component. These include hypertension (40), stroke (41), Alzheimer’s disease (42), and affective disorders such as depression (43) and schizophrenia (44). It has been suggested that cerebral hypoperfusion in some of these disorders leads to metabolic stress and eventual neuronal dysfunction and cognitive impairment (e.g., ref. 42). In slices from stressed rats, we observed a 66% reduction in the magnitude of the vasodilation evoked by neuronal activity. Our data thus predict a substantial reduction in the volume of blood delivered to the tissue during functional hyperemia and therefore, after chronic stress, brain regions in which NVC is impaired may experience functional hypoperfusion. Stress induces synaptic plasticity and neuronal hypertrophy in the amygdala, which leads to neuronal hyperactivity and behavioral changes (22). Presumably, neurons in this state have even higher energy demands than those in the nonstressed condition, but our results suggest that the delivery of substrates critical for energy production will be decreased after repeated stress. It is thus possible that stress precipitates a state of metabolic imbalance that, if persistent, could have detrimental effects on neuronal function over time. Further experiments will aim to explore this possibility.

In conclusion, our study demonstrates that—in addition to effects on neurons—stress impacts the cells of the neurovascular unit, leading to a marked impairment of NVC in the amygdala. We show that stress impairs NVC by reducing the number of functional K_IR channels in the membrane of PA myocytes. This renders arterioles less sensitive to small increases in local K⁺ released by astrocytes, thereby impairing K⁺-mediated vasodilation during NVC and, by extension, blunting the increase in blood flow needed to support local neuronal activity. This mechanism for the disruption of neurovascular communication may be important in diseases in which NVC is defective. Moreover, because stress is a contributory factor in many disease states, the NVC impairment elucidated here may be an important early step on the path to neuronal dysfunction and as such may represent a novel target for preventing or treating diseases associated with stressor exposure.

Materials and Methods

Male Sprague-Dawley rats weighing 225–250 g (Charles River) were housed in separate cages and kept on a 12-h light:dark cycle with ad libitum access to food and water. All procedures were approved by the Institutional Animal Care and Use Committee of the University of Vermont.

For detailed information on stress procedures, behavioral testing, multiphoton imaging of brain slices, pressure myography, quantitative PCR, electrophysiology, corticosterone ELISA, data analysis, and reagents, please refer to SI Materials and Methods.