Introduction

Alterations in synaptic plasticity are thought to underlie numerous stress-linked psychiatric disorders, including major depressive disorder (MDD) [1, 2]. For instance, patients suffering from MDD have fewer synapses in the prefrontal cortex (PFC) and this correlates with MDD severity and working memory dysfunction [3]. Clinical and preclinical findings suggest that microglia, the brain’s resident macrophages, may contribute to synaptic alterations in the PFC and MDD-associated symptoms. In fact, chronic stress has been shown to induce microglia-mediated remodeling of synapses in the medial PFC (mPFC) of mice and subsequent deficits in PFC-mediated behaviors, including working memory function and cognitive flexibility [4, 5]. Despite these findings, relatively little is known regarding the pathways that drive microglia-neuron interactions in stress and microglial interventions in stress-linked disorders remain elusive.

Studies indicate that initial stress exposure increases pyramidal neuron activity and glutamatergic signaling, and disturbs neuronal homeostasis in the PFC [6, 7]. Microglia respond dynamically to changes in neuronal activity by detecting the extrasynaptic release of purines, including adenosine-diphosphate and triphosphate (ADP and ATP) [8, 9]. Purine release by neurons attracts microglial processes via P2Y12, a purinergic receptor exclusively expressed by microglia in the brain [10,11,12]. These activity-dependent interactions have functional consequences as mice with genetic deletion of P2Y12 (P2ry12–/–) show impaired plasticity in the visual cortex following monocular deprivation during neurodevelopment and this was mediated, in part, by reduced uptake of synaptic material by microglia [10]. Thus, it is plausible that P2Y12 plays an important role in guiding stress-induced changes in microglia-mediated synaptic remodeling in the mPFC and associated behavioral consequences. To investigate this, we disrupted P2Y12 signaling during chronic stress using either genetic or pharmacological tactics, with analyses addressing microglia-neuron interactions in the mPFC, synapse density in this region, stress physiology, and cognitive-behavioral function.

Methods and materials

See online Supplementary Methods and Materials for full experimental details and product/reagent information.

Animals

Transgenic P2ry12–/– and Thy1-GFP(M) mice were obtained from in-house breeders backcrossed to C57BL/6 J mice (6–8 weeks old). Thy1-GFP(M)/P2Y12 transgenic mice were generated by crossing lines. Pharmacological experiments were performed in both wild-type C57BL/6 J mice (behavior and flow cytometry) and Thy1-GFP(M) mice (histological analyses). Primary studies used male mice as microglia-mediated neuronal remodeling in the PFC occurs in stressed male- but not female- mice [5, 13]. Behavioral and physiological outcomes were examined in P2ry12–/– females. Animal procedures were approved by the University of Cincinnati Institutional Animal Care and Use Committees (IACUC) and were in accordance with guidelines established by the National Institutes of Health.

Chronic unpredictable stress (CUS)

CUS was performed as previously described [5, 14]. In brief, mice were exposed to two random intermittent stressors per day for 14 days, including: cage rotation, isolation, restraint, radio noise, food or water deprivation, light on overnight, light off during day, rat odor, stroboscope overnight, crowding, wet bedding, no bedding, or tilted cage. To validate this manipulation, body weight was measured at the start and end of CUS, with percent weight gain calculated. Additional physiological analyses were conducted in P2ry12–/– mice.

Drug administration

Mice received either clopidogrel (50 mg/kg) or ticagrelor (10 mg/kg) via daily intraperitoneal (i.p.) injection [10]. The bioactive metabolite of clopidogrel can cross the blood-brain barrier, where it irreversibly binds to microglial P2Y12 [10, 15]. In contrast, the P2Y12 antagonist ticagrelor is not blood-brain barrier permeant, with studies showing no effect of peripheral ticagrelor treatment on microglial motility or injury responsivity [10]. Control mice received equal volume injections of vehicle.

Behavior and cognitive tests

Open field test (OFT), forced swim test (FST) and temporal object recognition testing (TOR) were conducted as previously described [5, 14, 16]. The mPFC contributes to both stress coping behavior in the FST and discrimination in the TOR, making these behavioral assays particularly relevant to the neurobiological outcomes examined in this study [17, 18]. The open field test was used to assess animal mobility in a novel context and thigmotaxis. For the OFT, mobility and the total time spent in the center zone of the open field were assessed. Time spent immobile (2–6 min phase) in the FST was measured. For the TOR, the time spent exploring objects was measured and a discrimination index was calculated.

Percoll gradient enrichment of microglia and flow cytometry

Dissected frontal cortex was homogenized and cells were enriched using Percoll. Cells were incubated with conjugated antibodies (FITC-CX3CR1 or AlexaFluor488-CD115; PE-P2Y12 or PE-CD68; PerCp-Cy5-CD11b; PE(594)-CD45). An average of 25,434 ± 971 microglia were examined per sample based on CD11b/CD45lo expression using a BioRad S3e four-color cytometer/cell sorter (Hercules, CA). Data were analyzed using FlowJo software (Ashland, OR).

Immunohistology

Mice were perfused approximately 4 h after the final stressor or 2 h after behavioral testing. Brains were frozen and sectioned (40 µm). Free-floating sections containing the PFC were washed, blocked, and incubated with primary antibody: rabbit anti-IBA1, rabbit anti-P2Y12, rat anti-CD68, or rabbit anti-FOSB. Sections were washed and incubated with conjugated secondary antibody (Alexa Fluor 488, 546, and 647).

Quantitative immunofluorescence

Confocal images were captured from adjacent brain sections in both hemispheres of the mPFC (3-4 sections/sample spanning the rostral-caudal axis, 2.10–1.34 mm Bregma). FosB immunolabeled tissue sections were imaged using a 10× objective. To quantify FosB+ cells, bright immunolabled cell bodies were thresholded and then counted using the ImageJ (NIH) Analyze Particles function. FosB+ cell density was calculated as cells per mm2. For analysis of IBA1 + material and P2Y12 intensity, tissue sections were imaged using a 20×. Individual IBA1 + and P2Y12 + cell counts were obtained by hand and cell density was calculated (cells per mm2). For microglial area, images were thresholded, total area was recorded (µm2), and area per microglial cell was calculated (µm2/cell). To assess microglial clustering, the distance between each individual microglia and their closest neighboring microglia was measured using the Nearest Neighbor Distance plugin. For analysis of microglial P2Y12 expression, images were thesholded, fluorescence intensity was measured (integrated density, A.U.), and P2Y12 expression per microglial cell was calculated as fluorescence intensity/total number of IBA1 + cells per image (A.U., relative to control).

In Thy1-GFP(M) mice, layer I of the mPFC was identified in multiple adjacent sections and apical dendritic segments were imaged using either a 40x or 60x objective. Given the expression profile of GFP in this transgenic line, apical segments in layer I most likely originate from pyramidal cells in layer V and VI of the mPFC [19]. Dendritic segments were traced and measured in NeuronStudio with spines counted by hand (6–8 segments/sample). To quantify GFP + inclusions (total number and volume), confocal images were obtained in multiple adjacent sections and microglia with complete morphological profiles were identified. Images were collected with sufficient resolution to detect synaptic inclusions (average inclusion diameter: 0.25 µm) [20]. Inclusions were identified both within microglial processes and in microglial somata. In addition to GFP + inclusions, the total volume of microglial CD68 was measured in these images, with data expressed as CD68 volume (µm3)/total number of IBA1 + cells (µm3/IBA1 + cell). Of 629 analyzed GFP + inclusions, 619 (98.4%) colocalized with immunolabeled CD68.

To examine autofluorescent lysosomal inclusions, microglia were visualized in P2ry12–/– mice lacking the Thy1-GFP(M) transgene with images collected as described above. Laser settings were maintained across studies. Autofluorescent inclusions were imaged at 488 nm excitation with an emission spectra of 510–574 nm.

Statistical analysis

Data were analyzed using GraphPad Prism 8.1.2 (La Jolla, California). Global main effects and interactions were determined using two-way ANOVA (all factors between-subjects; stress×genotype, stress×drug treatment). Following a significant ANOVA finding (main effect or interaction), a series of hypothesis-informed, planned comparisons were conducted using Sidak’s multiple comparisons test [21,22,23,24]. For P2ry12–/– studies, the following comparisons were made: wild-type control vs wild-type CUS, wild-type control vs P2ry12–/– control, wild-type control vs P2ry12–/– CUS, and P2ry12–/– control vs P2ry12–/– CUS. For pharmacological experiments comparing clopidogrel and ticagrelor, animals treated with either 0.9% saline or DMSO:saline did not differ across any metrics and were thus collapsed into one vehicle-treated group for analysis. For these studies, the following planned comparisons were made: vehicle control vs vehicle CUS, vehicle control vs clopidogrel control, vehicle control vs ticagrelor control, vehicle control vs clopidogrel CUS, vehicle control vs ticagrelor CUS, clopidogrel control vs clopidogrel CUS, and ticagrelor control vs ticagrelor CUS. Pearson correlation coefficients were computed and analyzed for select variables. The number of animals examined in each analysis is noted in Table S1, with complete omnibus statistics and pairwise comparisons (including confidence intervals) reported in Table S2 and Table S3, respectively.

Results

Genetic loss of P2Y12 attenuates stress-induced behavioral consequences and induces significant alterations in frontal cortex microglia

To determine the role of P2Y12 in stress-induced microglia responses, wild-type and P2ry12–/– mice were exposed to 14 days of CUS and then physiological, behavioral, and molecular outcomes were assessed (Fig. 1A). Both unstressed and CUS-exposed mice moved a comparable distance- and spent a similar amount of time in the center zone in the OFT (Fig. 1B, C). Loss of P2Y12 caused a modest reduction in time spent in the center zone in the OFT (F(1,39) = 6.898, p = 0.012), however, no pairwise differences were detected following planned comparisons. Consistent with prior studies, CUS increased immobility in the FST (F(1,40) = 7.784, p = 0.008), with pairwise comparisons indicating greater immobility in wild-type (p = 0.049), but not P2ry12–/– mice (Fig. 1D). CUS also reduced object discrimination in the TOR (F(1,40) = 23.83, p < 0.0001), with planned comparisons indicating significant differences in wild-type mice only (p < 0.0001; Fig. 1E). The total amount of time exploring objects in the TOR was unaffected by stress or genotype (Fig. 1F). Following behavioral testing, we examined whether P2Y12 ablation disrupts physiological stress responsivity. Stressed animals were subjected to a 1 h acute restraint challenge followed by analysis of plasma corticosterone, body weight, and organ weight. Acute restraint elevated plasma corticosterone levels in both wild-type and P2ry12–/– mice (F(1,38) = 57.34, p < 0.0001; Fig. 1G). Additional analyses show that CUS reduced weight gain (Fig. 1H) and caused adrenal hypertrophy in mice regardless of genotype (Fig. 1I); spleen weight was unaffected by CUS (Fig. 1J). Parallel studies in female mice found no effects of stress or genotype on behavior, despite CUS-induced alterations in weight gain and stress-sensitive organ weight (Fig. S1A–G). Together, these findings indicate that genetic loss of P2ry12 in male mice attenuated behavioral responses to stress – despite a pronounced physiological response.

Fig. 1: Constitutive ablation of P2Y12 attenuates stress effects on coping behavior and working memory, and induces significant alterations in frontal cortex microglia.
figure 1

A Male wild-type or P2ry12–/– mice were exposed to 14 days of chronic unpredictable stress (CUS) or were handled as controls (CON). One cohort was subjected to behavioral testing, an acute stress challenge, and analysis of stress-sensitive organ weight (n = 10–12/group). B Average distance moved in the open field test (OFT). C Average time spent in the center zone in the OFT. D Average time spent immobile in the forced swim test (FST). E Discrimination index in the temporal object recognition task (TOR). F Average time spent exploring objects in the TOR. G Levels of plasma corticosterone in mice subjected to an acute restraint stress challenge. H Body weight was measured prior to stress and post-CUS exposure, and percent weight change was calculated for all animals regardless of experimental endpoint (n = 24–29/group). I Adrenal weight relative to animal body weight. J Spleen weight relative to animal body weight. K In a separate cohort of mice, brains were extracted, frontal cortex was dissected out, and microglia were isolated and characterized using flow cytometry (n = 8/group). Representative dot plots of FSC-H/CD11b-H and CD45-H/CD11b-H are shown. Histogram depicts levels of P2Y12 fluorescence in frontal cortex microglia in representative WT and P2y12–/– mice. L Normalized mean fluorescence intensity of P2Y12, CSF1R, CD11b, CX3CR1, and CD68 in frontal cortex microglia. Normalized mean side scatter profile is shown. Bars represent mean ± S.E.M. *p < 0.05, pairwise comparison indicated (significant interaction). #p < 0.05, significant main effect of genotype. ‡ Stressp < 0.05, significant main effect of stress.

Full size image

In subsequent experiments, surface markers were characterized on frontal cortex microglia using flow cytometry. Analyses of cell surface expression confirmed the absence of P2Y12 in P2ry12–/– mice (F(1,28) = 295.0, p < 0.0001; Fig. 1K, L). Further panels revealed that frontal cortex microglia from P2ry12–/– mice are phenotypically different from wild-type microglia. In particular, frontal cortex microglia in P2ry12–/– mice had increased protein levels of CSF1R (F(1,28) = 7.719, p = 0.001), CD11b (F(1,28) = 137.6, p < 0.0001), and CD68 (F(1,28) = 102.2, p < 0.0001), as well as decreased CX3CR1 expression (F(1,28) = 244.4, p < 0.0001). Moreover, microglia from P2ry12–/– mice showed higher side scatter (SSC), suggesting an increase in intracellular complexity (F(1,28) = 7.905, p = 0.009). In addition, CUS increased microglial CSF1R, with pairwise comparisons indicating that this increase was only significant in P2ry12–/– mice (p < 0.001). Alongside CSF1R, exposure to CUS increased SSC (p = 0.008) in P2ry12–/– mice. These findings indicate that loss of P2Y12 leads to significant alterations in microglial phenotype and responsivity to CUS.

Loss of P2Y12 causes dysregulation of microglial phagocytosis in the medial prefrontal cortex and prevents microglia-mediated dendritic remodeling in chronic stress

To examine neuron-microglia interactions, P2ry12–/– and wild-type mice were crossed with Thy1-GFP(M) mice. Following 14 days of CUS brains were processed for confocal microscopy (Fig. S2A). Similar to prior work, exposure to CUS increased the number of FosB+ cells/mm2 in the mPFC regardless of genotype (F(1,26) = 46.54, p < 0.0001; Fig. S2B, C). Interestingly, P2ry12–/– mice had increased microglial density in the mPFC (F(1,26) = 83.95, p < 0.0001; Fig. S2B, D), contributing to a decrease in the distance between neighboring microglia (F(1,26) = 85.55, p < 0.0001;Fig. S2E). Loss of P2Y12 had little effect on basal levels of IBA1 + area per microglia in the mPFC (Fig. S2F). However, CUS increased microglial area in the mPFC in wild-type, but not P2ry12–/– mice (F(1,26) = 11.27, p = 0.002).

Assessment of microglia-neuron interaction in layer I of the mPFC (Fig. 2A, B) revealed a CUS-induced increase in the proportion of microglia with GFP + inclusions in wild-type mice (F(1,25) = 80.05, p < 0.0001). Surprisingly, we also observed high levels of GFP + inclusions in PFC microglia in both unstressed and stressed P2ry12-/- mice. Control measures indicate a similar amount of GFP + neuronal material in layer I of the mPFC across groups, regardless of genotype or CUS (Fig. S3A–C). Thus, it is unlikely that basal differences in the abundance of GFP + material contributed to the increase in microglial inclusions detected here. To expand on this finding, we quantified the cellular location of these inclusions in microglia and determined the proportion of GFP + inclusions exclusively in their soma, exclusively in their processes, and within both their soma and processes (Fig. 2B). These analyses showed that CUS reduced the proportion of microglia with GFP + inclusions only in their soma (F(1,25) = 17.74, p < 0.001), and increased the proportion of microglia with both somatic and process inclusions (F(1,25) = 17.42, p < 0.001) in wild-type, but not P2ry12–/– mice. The basal accumulation of GFP + inclusions in P2ry12–/– mice was primarily somatic, with very few process inclusions detected (F(1,25) = 24.03, p < 0.0001). Regardless of cellular localization, both stress and genotype had significant effects on the number of GFP + inclusions per phagocytic microglia (F(1,25) = 4.275, p = 0.049, Fig. 2C), and the volume of CD68 + material per microglia (F(1,25) = 13.82, p = 0.001, Fig. 2D, E) in the mPFC. In line with these findings, exposure to CUS reduced spine density on apical dendritic arbors in layer I of the mPFC in wild-type, but not P2ry12–/– mice (F(1,27) = 7.000, p = 0.013, Fig. 2F).

Fig. 2: Loss of P2Y12 increases lysosome markers in the medial prefrontal cortex and prevents dendritic remodeling in chronic stress.
figure 2

Male Thy1-GFP(M) wild-type or P2ry12–/– mice were exposed to 14 days of chronic unpredictable stress (CUS) or were handled as controls. Approximately 4 h after the final stressor, mice were perfused and brains were collected, sectioned, immunostained, and imaged (n = 5–9/group). A Confocal images of microglia (IBA1, red), dendritic segments (Thy1-GFP, green), and microglial lysosomes (CD68, blue) were obtained from lamina I of the mPFC. Crosshairs note either a microglial process in close proximity to dendritic elements or a dendritic element localized within a microglial cell body or process extension. Inset shows a magnified view of dendritic material in the noted location, with lysosome area marked (blue dashed line). Alongside merged channels, orthogonal cross-sections matching the noted location are depicted for each group. White scale bar represents 10 µm. B Left: Proportion of microglia with GFP + inclusions in the mPFC. Right: Proportion of phagocytic microglia with GFP + inclusions exclusively in their soma (bottom panel) or processes (middle panel), or within both their soma and processes (top panel). C Number of GFP + inclusions within microglia with dendritic elements. D Average GFP + inclusion volume within microglia. E Average volume of CD68 per microglial cell. F Representative images of dendritic segments (left) and average dendritic spine density (right). White scale bar represents 5 µm. Bars represent mean ± S.E.M. *p < 0.05, pairwise comparison indicated (significant interaction). #p < 0.05, significant main effect of genotype. ‡ Stressp < 0.05, significant main effect of stress.

Full size image

Prior studies indicate that accumulation of lysosomes in the cell soma may present as autofluorescent material in microglia (e.g., lysosomal bodies containing lipid droplets, cholesterol crystals) [25]. To test this, PFC microglia with CD68 + lysosomes were visualized in wild-type and P2ry12–/– mice lacking the Thy1-GFP(M) transgene. In wild-type mice there were few autofluorescent CD68 + lysosomes, however, autofluorescent material was observed in somatic lysosomes in P2ry12–/– mice (Fig. S4A, B). The mean intensity of these somatic accumulations was significantly lower than that of inclusions observed in mice carrying the Thy1-GFP(M) transgene (Fig. S4C). These results indicate that GFP + inclusions in microglia are composed of neuronal material in wild-type Thy1-GFP(M) mice. In contrast, it appears that GFP + inclusions in microglia from P2ry12–/– Thy1-GFP(M) animals are composed of both autofluorescent and neuronal material. Altogether, P2Y12 appears to play a critical role in regulating microglial density and lysosome function in the mPFC regardless of stress. These data further suggest that P2Y12 is important in guiding stress effects on microglial morphology and microglia-neuron interactions in this region.

Pharmacological blockade of microglial P2Y12 attenuates stress effects on behavior and shifts microglial phenotype in frontal cortex

To complement previous experiments, P2Y12 was manipulated pharmacologically using either clopidogrel, a P2Y12 antagonist that can access the brain, or ticagrelor, an antagonist which cannot cross the blood-brain barrier. Unstressed and CUS-exposed mice treated with either vehicle, clopidogrel, or ticagrelor were assessed in the FST and TOR (Fig. 3A). CUS increased immobility in the FST (F(2,71) = 5.239, p = 0.008) in vehicle- (p = 0.012) and ticagrelor-treated mice (p = 0.010), but not clopidogrel-treated animals (Fig. 3B). Similar to the FST, CUS caused deficits in temporal object recognition (F(1,72) = 23.92, p < 0.0001) which were only detected in vehicle- (p < 0.0001) and ticagrelor-treated mice (p = 0.006; Fig. 3C). Drug treatment affected the total time spent exploring objects in the TOR (F(2,72) = 4.212, p = 0.019), however, pairwise comparisons uncovered no specific group level differences (Fig. 3D). Exposure to CUS reduced weight gain in both vehicle- and clopidogrel-treated animals (Table S2, S3, Fig. S5). Collectively, these findings indicate that blocking P2Y12 on microglia (i.e., clopidogrel) but not peripheral cells (i.e. ticagrelor) attenuates stress effects on behavior.

Fig. 3: Pharmacological blockade of microglial P2Y12 attenuates stress effects on behavior and shifts microglial phenotype in the frontal cortex.
figure 3

A Male wild-type or Thy1-GFP(M) mice were exposed to 14 days of chronic unpredictable stress (CUS) or were handled as controls. During this time, animals received daily injections of either vehicle, clopidogrel, or ticagrelor. One cohort was subjected to behavioral testing (n = 9–20/group). B Average time spent immobile in the forced swim test (FST). C Discrimination index in the temporal object recognition task (TOR). D Average time spent exploring objects in the TOR. E In a separate cohort of mice, brains were extracted, frontal cortex was dissected out, and microglia were isolated and characterized using flow cytometry (n = 7–22/group). Normalized mean fluorescence intensity of P2Y12, CSF1R, CD11b, and CX3CR1 in frontal cortex microglia. Normalized mean side scatter profile is shown. Bars represent mean ± S.E.M. *p < 0.05, pairwise comparison indicated (significant interaction or main effect).

Full size image

In a separate cohort, we analyzed the phenotype of frontal cortex microglia using flow cytometry (Fig. 3E). Similar to genetic loss, treatment with clopidogrel reduced expression of P2Y12 (F(2,79) = 98.59, p < 0.0001) and CX3CR1 (F(2,62) = 30.39, p < 0.0001). There was an overall effect of drug treatment on microglial CD11b (F(2,83) = 3.390, p = 0.038); however, no significant differences emerged with pairwise comparisons. Increased levels of microglial CSF1R were observed in mice treated with clopidogrel (F(2,60) = 5.964, p = 0.004) and those exposed to CUS (F(1,60) = 10.66, p = 0.002). Planned comparisons showed that CUS increased microglial CSF1R in both vehicle- (p = 0.044) and ticagrelor-treated mice (p = 0.040), but not clopidogrel-treated mice. Neither drug treatment nor CUS had an effect on microglial SSC. Thus, pharmacological blockade of P2Y12 caused changes in microglial phenotype that resembled genetic loss, while also preventing stress effects on CSF1R expression in frontal cortex microglia.

Inhibiting microglial P2Y12 blocks stress-induced phagocytosis of neuronal elements and associated dendritic spine loss in the mPFC

We next analyzed neuronal function, microglial phenotype, and microglia-neuron interaction in the mPFC using confocal microscopy (Fig. 4A). CUS increased the number of FosB+ cells/mm2 in the mPFC in both vehicle- and clopidogrel- treated animals (F(1,26) = 24.18, p < 0.0001; Fig. 4B, C). Neither CUS nor treatment with clopidogrel affected the number of IBA1 + cells/mm2, P2Y12 + cells/mm2, or microglial nearest neighbor distance in the mPFC (Fig. 4D, Table S2). However, CUS increased microglial morphological area, and this was blocked by treatment with clopidogrel (F(1,26) = 13.59, p = 0.001; Fig. 4E). Similar to our flow cytometry findings, treatment with clopidogrel reduced microglial P2Y12 intensity in the PFC (F(1,26) = 22.59, p < 0.0001; Fig. 4F).

Fig. 4: Administration of clopidogrel reduces levels of microglial P2Y12 and prevents stress effects on microglial morphology.
figure 4

A Male Thy1-GFP(M) mice were exposed to 14 days of chronic unpredictable stress (CUS) or were handled as controls. During this time, animals received daily injections of either vehicle or clopidogrel. Approximately 4 h after the final stressor, mice were perfused and brains were collected, sectioned, immunostained, and imaged (n = 6–8/group). B Confocal images of FosB (top panel, green), microglial IBA1 (bottom panel, red) and microglial P2Y12 (bottom panel inset, cyan) in the mPFC. White scale bar represents 100 µm. C Average number of FosB+ cells/mm2 in the mPFC. D Average number of IBA1 + cells/mm2 in the mPFC. E Average IBA1 + area per microglial cell. F Average intensity (A.U.) of P2Y12 staining per microglial cell (relative to vehicle treated control group). Bars represent mean ± S.E.M. *p < 0.05, pairwise comparison indicated (significant interaction). #p < 0.05, significant main effect of drug treatment. ‡ Stressp < 0.05, significant main effect of stress.

Full size image

Exposure to CUS increased the proportion of microglia with GFP + inclusions in the mPFC in wild-type, but not clopidogrel-treated mice (F(1,22) = 13.52, p = 0.001; Fig. 5A, B). Analysis of inclusion location revealed a CUS-induced reduction in the proportion of microglia with soma-only inclusions (F(1,22) = 4.343, p = 0.049) and an increase in the proportion of microglia with both somatic and process inclusions (F(1,22) = 8.018, p < 0.001) in vehicle-treated animals (Fig. 5B). Further analyses indicate that treatment with clopidogrel prevented CUS-induced increases in the number of GFP + inclusions (F(1,22) = 5.606, p = 0.027; Fig. 5C), the average volume of GFP + inclusions per phagocytic microglia (F(1,22) = 15.94, p = 0.001; Fig. 5D), and the accumulation of CD68 + lysosomes in microglia (F(1,26) = 12.55, p = 0.002; Fig. 5E). In line with these data, blocking microglial P2Y12 with clopidogrel prevented spine loss on apical dendrites in the mPFC of CUS-exposed mice (F(1,21) = 12.98, p = 0.002; Fig. 5F). Correlational analyses indicate a significant association between FosB induction in the mPFC and an increased proportion of microglia with neuronal inclusions (r = 0.896, p < 0.0001) in vehicle-, but not clopidogrel-treated animals (Fig. 5G). This increase in the proportion of microglia with neuronal inclusions was strongly associated with reductions in dendritic spine density in vehicle-treated mice (r = −0.845, p < 0.0001). Together, these findings suggest that CUS increases neuronal activity in the mPFC, and this engages microglia and subsequently triggers microglia-mediated dendritic remodeling through a P2Y12-dependent pathway.

Fig. 5: Antagonism of microglial P2Y12 prevents stress-induced phagocytosis of dendritic elements and subsequent dendritic spine loss in the mPFC.
figure 5

Male Thy1-GFP(M) mice were exposed to 14 days of chronic unpredictable stress (CUS) or were handled as controls. During this time, animals received daily injections of either vehicle or clopidogrel. Approximately 4 h after the final stressor, mice were perfused and brains were collected, sectioned, immunostained, and imaged (n = 6–8/group). A Confocal images of microglia (IBA1, red) and dendritic segments (Thy1-GFP, green) were obtained from lamina I of the mPFC. Crosshairs note either a microglial process in close proximity to dendritic elements or a dendritic element localized within a microglial cell body or process extension. Inset shows a magnified view of dendritic material in the noted location, with inclusion area marked (green dashed line). Alongside merged channels, orthogonal cross-sections matching the noted location are depicted for each group. White scale bar represents 10 µm. B Left: Proportion of microglia with GFP + inclusions in the mPFC. Right: Proportion of phagocytic microglia with GFP + inclusions exclusively in their soma (bottom panel) or processes (middle panel), or within both their soma and processes (top panel). C Number of GFP + inclusions within microglia with dendritic elements. D Average GFP + inclusion volume per phagocytic microglia. E Average volume of CD68 per microglial cell. F Average dendritic spine density. G Left: Linear association between the density of FosB+ cells and the proportion of microglia with dendritic inclusions in the mPFC. Right: Linear association between dendritic spine density and the proportion of microglia with dendritic inclusions in the mPFC. Solid line represents linear fit in vehicle-treated animals (left), dashed line represents this in clopidogrel-treated animals (right). Pearson correlation coefficients (r) are shown for each dataset. Bars represent mean ± S.E.M. *p < 0.05, pairwise comparison indicated (significant interaction). #p < 0.05, significant main effect of drug treatment. ‡ Stressp < 0.05, significant main effect of stress.

Full size image

Discussion

Various stress-linked psychiatric disorders, including MDD, are marked by dendritic atrophy and synapse loss in the PFC. Recent preclinical studies suggest that microglia contribute to stress-induced changes in neuronal structure and associated behavioral consequences [4, 5, 14, 26], however, the pathways that direct microglia-neuron interactions in this context remain unclear. Our primary findings indicate that P2Y12 signaling is required for stress-induced microglial phagocytosis of neuronal elements and mediates spine loss on pyramidal neurons in the PFC, which underlies deficits in working memory and stress-coping behavior. Our data also demonstrate that P2Y12 is essential for establishing microglial phenotype in the PFC, regardless of stress.

P2Y12 is a critical regulator of microglial phenotype and function

Studies have identified P2Y12 as a critical receptor in the microglial ‘sensome’ [27], guiding microglial surveillance and physical microglia-synapse contact [8, 10, 28, 29]. In line with these findings, our data show that P2Y12 broadly regulates microglial phenotype and function in the frontal cortex. Genetic loss or pharmacological blockade of P2Y12 shifted the homeostatic phenotype of microglia with increased surface protein levels of CSF1R and CD11b, and decreased expression of CX3CR1. This is relevant because CSF1R and CD11b have been shown to drive microglia-neuron interactions and synaptic remodeling [5, 30], whereas CX3CR1-signaling can either promote- or suppress- aspects of microglial function [31, 32]. These phenotypic alterations are likely a compensatory mechanism through which microglia attempt to survey and modulate synaptic structures with deficient P2Y12 signaling.

Another important finding is that genetic loss, but not pharmacological blockade, of P2Y12 increased the density and morphology of microglia in the PFC of unstressed adult mice. This aligns with reports showing that P2Y12 guides rearrangement of the broad microglial landscape and microglial translocation [28]. In addition, microglia in unstressed P2ry12–/– mice accumulated large somatic CD68 + inclusions composed of both autofluorescent and neuronal material. These inclusions may have contributed to an increase in microglial SSC in P2ry12–/– mice, and suggest impairments in microglial lysosome regulation and increased levels of phagocytosis and/or autophagy [25]. Interestingly, this increase in microglial phagocytosis was not associated with a loss of dendritic spines in P2ry12–/– mice. Other studies have established that ablation of P2Y12 reduces microglial phagocytosis of GluR1+ synapses and disrupts experience-dependent synaptic plasticity in the developing visual cortex [10]. Considering this, alongside the discordant microglial phenotype observed (i.e., heightened levels of CSF1R, CD11b, CD68), it is possible that microglia are remodeling other dendritic structures not examined in this study or that microglia are contributing to a greater turnover of dendritic material in mice lacking P2Y12. Future studies will be needed to determine the composition of microglial inclusions and mechanisms that lead to their accumulation in P2ry12–/– mice. Despite these robust neurobiological effects, no behavioral alterations were detected in mice lacking P2Y12. This is consistent with a recent report showing only modest shifts in cognition and behavior in P2ry12–/– mice [33].

Stress-induced microglia-mediated dendritic remodeling, synapse loss, and behavioral adaptations in stress are dependent on P2Y12 signaling

Stress exposure increases neuronal activity, alters excitatory neurotransmission, and purine metabolism in the PFC [34, 35]. Prior work indicates that elevated neuronal activity ‘attracts’ microglial processes via extracellular purines acting on microglial P2Y12 [9]. This purinergic signaling pathway promotes microglia-neuron interactions, which shape neuroplasticity and circuit-level function in various contexts [10, 12, 29, 36, 37]. In this study, CUS increased expression of the neuronal activation marker FosB in the PFC and disrupted both stress-coping behavior and working memory. Administration of the brain permeant P2Y12 antagonist clopidogrel attenuated these CUS-induced deficits, whereas treatment with ticagrelor, which cannot cross the blood-brain barrier, did not prevent behavioral consequences [10, 15]. Thus, blocking P2Y12 on microglia, specifically, reduces stress effects on behavioral and cognitive outcomes mediated by the PFC.

Chronic stress increases microglia-neuron interaction and microglial process-mediated synapse elimination in the mPFC [4, 38]. Consistent with these findings, we show that CUS causes microglia in the mPFC to accumulate neuronal elements in both their processes and cell body, suggesting process-mediated actions (e.g., synaptic pruning) and somatic debris compartmentalization (e.g., in lysosomes) [4, 5, 31, 39]. This accumulation of neuronal material within microglia was strongly associated with reductions in dendritic spine density, further implicating microglia in CUS-induced neuronal remodeling [4]. Genetic knockout or pharmacological blockade of microglial P2Y12 prevented microglia-mediated neuronal remodeling and attenuated synapse loss in the mPFC. Considering this, alongside our prior data connecting neuronal activity and microglial function [26], it is plausible that CUS-induced activation of excitatory neurons in the mPFC engages P2Y12 signaling in microglia, which increases the frequency of microglia-synapse interactions and subsequent microglia-mediated dendritic remodeling (Fig. S6). This would be in line with other studies showing that P2Y12 is necessary for microglia to respond to fear- and seizure-associated activity in the hippocampus [29, 40], and that microglial P2Y12 can regulate experience-dependent synaptic plasticity [10]. It is important to note that the stress response is dynamic and likely engages other microglial pathways that are capable of modulating neuronal function [41]. Interestingly, a recent study showed that neuronal hyperactivity triggered a signaling cascade that involved P2Y12 and resulted in microglia providing negative regulatory feedback through release of the inhibitory molecule adenosine [36]. In this context, it is possible that P2Y12 signaling drives varied microglia functions across the duration of CUS. Additional studies will be needed to directly examine these temporal dynamics.

Exposure to stress differentially alters microglia in the mPFC in male and female mice and rats. For instance, chronic stress increases microglial morphological area in males yet has little effect on- or decreases microglial area in female rodents [13, 42, 43]. Moreover, microglial actions have been associated with stress-induced synapse loss in the mPFC and subsequent deficits in behavior in male, but not female, mice [13]. In this study, genetic ablation of P2Y12 had no effect on physiological stress responsivity, stress coping behavior, or temporal object recognition in female mice exposed to CUS. These findings, in concert with other reports, suggest a sex-specific role for microglia in regulating neuroplasticity and behavior in stress [44,45,46,47,48]. Notably, post-mortem transcriptomic studies indicate similar sex-dependent microglial and synapse-associated patterns in the dorsolateral PFC (dlPFC) in MDD; men show increased expression of microglial genes and decreased synapse-associated transcripts, whereas women show the opposite [49]. Future experiments characterizing sex differences in microglia and mechanisms mediating these effects are warranted [50].

The findings presented here may be translationally relevant, as recent studies show that microglia isolated from individuals diagnosed with MDD have heightened expression of ‘sensome’ markers, including P2Y12 and TMEM119 [51, 52]. Intriguingly, these same studies found no differential expression of inflammation-associated molecules, suggesting that MDD is not marked by microglial ‘activation’ or neuroinflammation, but rather a more nuanced shift in microglial phenotype. This parainflammatory phenotype is observed in several models of chronic stress [53]. Given our data, it is interesting to speculate that microglial P2Y12 may play an important role in regulating synapse loss in MDD.

Conclusion

Collectively, our findings demonstrate a role for microglial P2Y12 in guiding stress effects on prefrontal structure, cognition, and behavior. These data may be clinically pertinent, as both prefrontal synapse loss and altered levels of microglial P2Y12 have been found in individuals diagnosed with MDD. This work is significant as it provides insight into pathways that direct microglia-neuron interactions, and how these interactions modulate the synaptic structure and behavioral responses in stress-linked psychopathology.