Abstract

Recently, it has been discovered that the working memory deficits induced by exposure to chronic stress can be prevented by treating stressed animals with minocycline, a putative inhibitor of microglial activity. One of the pressing issues that now requires clarification is exactly how exposure to chronic stress modifies microglial morphology, this being a significant issue as microglial morphology is tightly coupled with their function. To examine how chronic stress alters microglial morphology, we digitally reconstructed microglia within the rat medial prefrontal cortex. Our analysis revealed that stress increased the internal complexity of microglia, enhancing ramification (i.e. branching) without altering the overall area occupied by the cell and that this effect was more pronounced in larger cells. We subsequently determined that minocycline treatment largely abolished the pro-ramifying effects of stress. With respect to mechanisms, we could not find any evidence of increased inflammation or neurodegeneration (interleukin-1β, MHC-II, CD68, terminal deoxynucleotidyl transferase dUTP nick end labeling, and activated caspase-3). We did, however, find that chronic stress markedly increased the expression of β1-integrin (CD29), a protein previously implicated in microglial ramification. Together, these findings highlight that increased ramification of microglia may represent an important neurobiological mechanism through which microglia mediate the behavioral effects of chronic psychological stress.

Introduction

Chronic stress is recognized to be capable of producing profound alterations in mood and cognitive function (Pittenger and Duman 2008). In accordance with these serious consequences, a considerable effort has been directed toward understanding the specific neurobiological mechanisms involved (Krishnan and Nestler 2008). While the vast majority of research has focused on stress-induced alterations in neuronal structure and function (McEwen 2000), it has recently become apparent that glia, and in particular microglia, are also altered by exposure to chronic stress (Tynan et al. 2010; Hinwood, Morandini et al. 2011).

We have recently shown that chronic stress produces substantial changes in microglia within several brain regions critical to the regulation of mood state and cognition. Specifically, we identified that stress-induced microglial alterations coincide with an increase in depression-like behaviors (Tynan et al. 2010). Subsequently, we established that the anti-microglial activation drug minocycline significantly reduced the level of working memory disturbance induced by chronic stress (Hinwood, Morandini et al. 2011). Together, these findings constitute the first demonstration that a drug treatment that modifies microglial activity can improve a stress-induced cognitive impairment.

While previous research has clearly implicated microglial alterations in mediating stress-induced cognitive disturbance, one of the fundamental questions that remain unanswered is how stress alters microglial morphology. This is a highly significant issue as it well recognized that a tight coupling exists between microglial morphology and function. Rio-Hortega (1932) was the first to recognize this relationship, and over the ensuing years, the phases of microglial transformation, and their coupled functional states, have been significantly elaborated (Streit et al. 1999; Stence et al. 2001). It is now recognized that under non-pathological conditions, microglia are frequently found to possess numerous thin processes emerging from the soma, with each primary process exhibiting some secondary and tertiary branching. This form of microglia (often referred to as ramified microglia) do not typically engage in injury-related activities but rather appear to be involved in monitoring synaptic integrity (Tremblay et al. 2010). In contrast, when the central nervous system (CNS) has sustained an injury, microglial processes are frequently observed to thicken and shorten, or retract altogether transitioning into what are referred to as amoeboid microglia. Microglia possessing this type of morphology are commonly found to engage in phagocytosis, and release of free radicals and proinflammatory factors (Graeber and Streit 2010).

To date, all published studies that have examined the effects of stress on microglia have used densitometry. This technique quantifies the changes in the intensity of microglial-specific immunohistochemical or immunofluorescent labeling and reveals population-wide changes within a given region of interest (Sugama et al. 2009; Tynan et al. 2010; Hinwood, Morandini et al. 2011; Wohleb et al. 2011, 2012; Jurgens and Johnson 2012). The limitation of this procedure, however, is that it does not provide specific information on how individual cells are altered. Therefore, at present, it is not known whether the morphological alterations induced by chronic stress are more consistent with an injury or a non-pathological phenotype.

As we have shown that chronic stress drives substantial changes in microglia within the medial prefrontal cortex (mPFC) and that inactivating microglia using minocycline reduces these changes and improves stress-induced working memory deficits, we undertook a study whereby we reconstructed and analyzed microglia of the mPFC to determine exactly how chronic stress alters their phenotypes. Specifically, we focused our attention on those layers of the mPFC that have previously been shown to be most influenced by exposure to chronic stress (Radley et al. 2006; Shansky et al. 2009). Our data indicates that chronic stress significantly increased the internal complexity of microglia, enhancing ramification without altering the area occupied by the cell. We found that these effects were restricted to the secondary microglial processes and were more pronounced in larger cells and that minocycline abolished the pro-ramifying effects of stress. In terms of identifying the potential molecular mechanisms involved in driving these alterations, we first examined changes in several proteins linked to CNS injury and inflammation [MHC-II, CD68, activated caspase-3, terminal deoxynucleotidyl transferase dUTP nick end labeling (TUNEL), and interleukin-1β (IL-1β)]. We found no evidence that exposure to stress (or stress plus minocycline) elevated the levels of any of these markers. This result suggested that the microglial alterations occurring in response to stress were unlikely to be due to CNS injury. Additionally, we examined the expression of β1-integrin (CD29), as recent evidence has demonstrated that this protein is involved in promoting ramification and that its levels are significantly reduced with minocycline treatment (Nutile-McMenemy et al. 2007). Our investigation revealed that chronic stress significantly increased microglial levels of β1-integrin, with greater levels being found in larger cells. We further observed that minocycline dramatically reduced β1-integrin expression in both stress and control groups. Collectively, these results clearly indicate that exposure to chronic stress induces a form of non-injury-related hyper-ramification that appears to be mediated by an increase in β1-integrin.

Materials and Methods

Ethics

Experiments were approved by the University of Newcastle Animal Care and Ethics Committee and were conducted in accordance with the NSW Animal Research Act and the Australian Code of Practice for the use of animals for scientific purposes.

Experimental Design

We used a 2 (stress condition: chronic restraint stress vs. handled controls) × 2 (drug treatment: minocycline vs. no minocycline) between-group experimental design. This protocol yields 4 experimental groups (n = 10 per group): 1) 21 days of restraint stress (STR); 2) 21 days of brief handling (CON); 3) 21 days of restraint stress and administration of minocycline (STR + M); and 4) 21 days of brief handling and administration of minocycline (CON + M). All animals were sacrificed 24 h after the final episode of stress or handling.

Experimental Animals and Treatments

Adult male Sprague–Dawley rats (350–450 g; 70 days old at the commencement of the experiment) were obtained from the Animal Resource Centre (Perth, Western Australia). Animals were held in individual cages in temperature-controlled animal holding rooms (21 ± 1°C) on a 12 h reversed light–dark cycle (lights on at 19:00 h). All experimental procedures were conducted during the dark phase of the light cycle. Animals were adapted to individual housing for 7 days prior to any experimental manipulation and were maintained on standard rat chow and water provided ad libitum.

Restraint stress was conducted as described previously (Tynan et al. 2010). Rats were placed in wire mesh restrainers secured with butterfly clips for 6 h during the dark period of the light cycle. The restraint stress procedure was repeated once daily for 21 days. Control groups were handled twice daily for 2 min.

Administration of Minocycline

As previously described (Hinwood, Morandini et al. 2011), minocycline hydrochloride (PCCA, Australia) was administered orally via drinking water at a dosage of 40 mg/kg/day. Minocycline was administered for the duration of the stress exposure period. We have previously found this to be an effective form of drug administration in rats, with the administered dose reaching a clinically effective level (Hinwood, Morandini et al. 2011).

Tissue Processing and Immunohistochemistry

Twenty-four hours after the final day of the repeated stress protocol, animals were deeply anesthetized with sodium pentobarbital and transcardially perfused via the ascending aorta with 2% sodium nitrite followed by 4% ice-cold paraformaldehyde. Brains were removed and post-fixed overnight in the same fixative and then placed into 12.5% sucrose in phosphate-buffered saline (PBS) for storage and cryoprotection. Brains were sectioned into 30 μm slices using a freezing microtome (Leica).

As previously described (Hinwood, Tynan et al. 2011), coronal sections (1 in 6 series) were incubated with a rabbit polyclonal antiserum directed against the ionized calcium-binding adapter molecule 1 (Iba-1; 1:10 000; Wako Bioproducts; catalogue #019-1974) with 2% normal horse serum overnight. Iba-1 is a constitutively expressed calcium-binding protein which is specific to microglia (Ahmed et al. 2007; Imai and Kohsaka 2002). The specificity of this antibody has been verified by the manufacturer and in a recent study using western blot analysis (Horvath and DeLeo 2009). Sections were washed and incubated with biotinylated donkey anti-rabbit secondary antibody (1:500; Jackson Immunoresearch; catalogue #711-005-152). Iba-1 immunolabeling was developed using a nickel-enhanced 3′3′-diaminobenzidene reaction. Brain regions were identified anatomically in accordance with a stereotaxic rat brain atlas (Paxinos and Watson 2005). Negative control sections, in which one primary antibody was omitted at a time, were performed for all experiments.

Immunolabeling to exclude the possibility of microglial activation as a result of injury or apoptosis was performed for several markers of injury or inflammation. Specifically, we measured levels of MHC-II, a protein involved in the presentation of a processed antigen; CD68, a marker of microglial phagocytosis; and the apoptotic indicators activated caspase-3 and TUNEL. The pro-inflammatory cytokine IL-1β was also examined, as it has been previously reported that other neuroinflammatory conditions (e.g. peripheral injection of lipopolysaccharide) increase amounts of IL-1β in the brain (Corona et al. 2010) and that microglia are a primary producer of IL-1β in the CNS (Summers et al. 2009). These were each performed on a single mPFC section per animal. Mouse monoclonal antibodies were used for the identification of MHC-II and CD68 (Serotec; catalogue #MCA46G and #MCA341GA, respectively), which are surface antigens expressed by microglia during the response to tissue damage. For the identification of cells undergoing apoptosis, we used a rabbit polyclonal antiserum directed against the cleaved p17 fragment of activated caspase-3 (Chemicon International; catalogue #AB3623) and a TUNEL assay (Invitrogen). In addition to the experimental tissue, positive control tissue for apoptosis from rat mammary gland post-lactation was labeled to verify antibody effectiveness. For IL-1β, we used a rabbit polyclonal antibody raised against the C-terminus of IL-1β (Santa Cruz Biotechnology, catalogue #SC-7884).

Reconstruction of Microglia and Associated Processes

Slides were coded and data analysis was performed by an experimenter blind to the treatment condition, with the code not broken until all analyses were complete. Microglia in cortical layers II/III of the infralimbic mPFC were reconstructed using a computer-assisted morphometry system consisting of a Zeiss Axioskop photomicroscope equipped with an MAC 6000 XYZ computer-controlled motorized stage and joystick with focus control (Ludl Electronic Products), a Q Imaging video camera (MBF Biosciences), a PC running Windows Vista (Dell Australia), and Neurolucida morphometry software (MBF Biosciences). Microglia were visualized and reconstructed under a Zeiss Axio Plan NEOFLUAR ×100 objective with a numerical aperture of 1.3 under oil immersion, using Neurolucida software (MBF Biosciences). Inclusion for analysis required that microglia were located in layer II/III of the infralimbic mPFC and exhibited intact microglial processes unobscured by either background labeling or other cells. A separate sample of cells in layer II/III of the secondary motor cortex, a region not known to display a microglial response to stress, was reconstructed for comparison to those in the mPFC. Microglia were traced throughout the entire thickness of the section, and trace information was then rendered into a 2D diagram of each cell (Fig. 1). Four cells per region were randomly selected, giving a total of 148 reconstructed microglia included for analysis from the mPFC: 40 from each of the STR and CON groups, and 34 from each of the STR + M and CON + M groups.

Figure 1.

(A) Photomicrograph taken at ×100 and (B) the corresponding 2D trace. At high magnification, much of the 3D structure of the cell is lost from the plane of focus. The Neurolucida tracings demonstrate the entire appearance of individual cells in a 2D image. (C) Anatomical localization of microglial cells used for analysis in the infralimbic mPFC. The location for each cell is represented by a colored point, CON, green, STR, yellow, CON + M, blue, and STR + M, red. (D) Image depicting concentric circles imposed upon cells at 5 μm intervals for the Sholl analysis. ML, midline. Scale bar = 20 μm.

Figure 1.

(A) Photomicrograph taken at ×100 and (B) the corresponding 2D trace. At high magnification, much of the 3D structure of the cell is lost from the plane of focus. The Neurolucida tracings demonstrate the entire appearance of individual cells in a 2D image. (C) Anatomical localization of microglial cells used for analysis in the infralimbic mPFC. The location for each cell is represented by a colored point, CON, green, STR, yellow, CON + M, blue, and STR + M, red. (D) Image depicting concentric circles imposed upon cells at 5 μm intervals for the Sholl analysis. ML, midline. Scale bar = 20 μm.

Double Label Immunofluorescence and Confocal Laser Scanning Microscopy

Double labeling for Iba-1 (goat polyclonal antibody to Iba1; 1:500; Wako Biosciences; catalogue #ab5076) and β1-integrin [rabbit monoclonal antibody to β1-integrin (CD29); 1:100; Abcam; catalogue #ab52971] was used in order to quantify the amount of β1-integrin within microglial cells of layer II/III of the IL mPFC. Iba1 and β1-integrin were visualized using AlexaFluor-labeled secondary antibodies (488 or 594, respectively; 1:400; Molecular Probes). Briefly, free-floating sections were rinsed using PBS, incubated with a 3% bovine serum albumin (BSA)/0.3% Triton X-100 (TX-100) blocking solution and then incubated overnight at 4°C with a primary antibody cocktail. Antibodies were diluted in a 1% BSA/0.1% TX-100 blocking solution. Sections were rinsed and then incubated with appropriate secondary antibodies for 2 h. Sections were rinsed and mounted using anti-fade mountant (Gelvatol). Localization of protein expression was performed using a Nikon C1 confocal laser scanning microscope, with a ×100 objective. The double labeling was visualized by acquiring a Z-stack of confocal images (step size 0.5 μm), each with a resolution of 1024 × 1024 pixels. A minimum of 3 randomly selected cells per animal were imaged, from layer II/III of the IL mPFC.

Analysis of Reconstructed Cells

All analyses were conducted using NeuroExplorer software (MBF Biosciences). Initially, we examined several soma and process characteristics including cell body perimeter, number of processes, number of nodes (branch points), and total length and volume of cellular processes. A convex hull analysis was performed, which measures the area of the field defined by the processes, calculated as the area enclosed by a polygon that joins the most distal aspects of the cellular processes. A fractal analysis involved undertaking a box counting procedure (Fernandez and Jelinek 2001), which determines the fractal dimension (k-dim). This value provides a measure of how completely the cell fills the space defined by its boundaries. Due to the considerable range in cell size, we also examined morphological characteristics for large (>2000 μm2) and small cells (<1800 μm2) consistent with the size-based approach to the morphometric analysis approach used by Radley et al. (2008). As the complexity of microglial cells varies as function of the distance from the soma, we also undertook the Sholl analyses, for both the entire group, and separately for large and small cells. Specifically, this involved creating a series of increasing 5 μm concentric circles (radii) around the soma (Sholl 1956). Nodes (branch points), the number of processes that intersect the concentric radii, and process length, surface area, volume and average diameter were quantified as a function of the distance from the cell soma for each radius (see Fig. 1 for details).

For the quantification of β1-integrin (CD29) colocalized within microglial cells, Z-stacks of individual cells were imported into Neurolucida and condensed to a single tif file of the maximal intensity profile of the entire image stack. Each level of the Z-stack was checked visually to ensure that β1-integrin labeling was located within the boundaries of the microglial cell. Metamorph imaging software was used to create a region around the cell of interest, and a thresholding procedure was used to quantify the amount of β1-integrin within each cell. In this way, we ensured that only double-labeled Iba-1/β1-integrin was counted.

Statistical Analysis

Morphometric parameters were averaged for each experimental group, and between-group differences were analyzed using one-way ANOVAs with post hoc tests. For the Sholl analyses, group means were compared using mixed-design between-subject (treatment group: CON, STR, CON + M, STR + M) and within-subject (radial distance from the cell soma) ANOVAs. This was followed by planned comparisons of simple effects and post hoc testing to determine specific points of statistical significance. Additionally, we determined whether changes in process morphology occurred equally across the entire spectrum of cells, or whether they varied according to the cell size (similar to the technique used by Radley et al., 2008, in order to determine whether spine density varied according to the spine size). We examined all process characteristics for the lower 25th and upper 75th percentiles of cell area as calculated using a convex hull analysis for each experimental group. Thus, differences in morphometry of cells below the 25th percentile, and above the 75th percentile, were examined between groups, using the same analysis as described above.

An independent samples Kruskal–Wallis non-parametric ANOVA was used to determine whether stress has caused a change in the overall numbers of small and large cells within the PFC. The number of small and large cells was compared in each of the 4 experimental groups.

For the analysis of β1-integrin labeling, raw counts of positive immunoreactive material were analyzed using one-way ANOVA to determine differences between the 4 treatment groups. Pearson's correlations were run to determine the size of the association between the cell size and the amount of β1-integrin. The amount of β1-integrin in the lower 25th and upper 75th percentiles of cells according to the size was also examined.

In all cases, ANOVA assumptions, including homogeneity of variance, homogeneity of regression, and sphericity, were satisfied. Data are expressed as the mean ± SEM, and all statistical analysis was conducted using an α criterion of 0.05.

Results

The precise location of the reconstructed microglia was evaluated using cortical layer depths that have been established for the adult Sprague–Dawley rat (Gabbott et al. 2005) and used in several recent publications (Morshedi and Meredith 2007; Hinwood, Morandini et al. 2011; Hinwood, Tynan et al. 2011). Layer II/III of the infralimbic mPFC ranges from 17.8% to 46.6% of the distance from the pial surface to the underlying white matter (see Fig. 1 for details).

Stress Increases the Number of Branch Points of Microglial Cells, and Minocycline Causes a Reduction in Cellular Processes and Area Occupied by the Cell

We assessed the average size of the cell body; number, volume and length of processes; branch points; and the total area of the cell as assessed by convex hull analysis for microglial cells from each experimental group using one-way ANOVAs. These analyses revealed that there was a significant effect of the group on branch points (F3,141 = 3.39, P = 0.02). Post hoc tests revealed that stress increased the degree of branching by approximately 21% relative to handled controls and by approximately 39% relative to both minocycline-treated groups (all P < 0.05). There was a significant effect of the group on the number of processes (F3,141 = 4.26, P < 0.01), process length (F3,141 = 4.15, P < 0.01), and convex hull area (F3,141 = 4.14, P < 0.01). Post hoc tests revealed that these metrics were all reduced in minocycline-treated groups compared with both STR and CON (all P < 0.05). Note that across groups, the k-dim value, cell body size, and total volume of processes remained unchanged by any treatment (see also Table 1).

Table 1

Morphological characteristics of total microglial cell sample

 CON STR CON + M STR + M 
Cell body perimeter (μm) 32.03 (0.86) 33.63 (1.47) 30.07 (0.65) 32.35 (1.02) 
k-dim 1.029 (0.003) 1.035 (0.004) 1.028 (0.002) 1.025 (0.003) 
Number of processes originating from soma 5.29 (0.25)* 5.25 (0.22)* 4.65 (0.19) 4.34 (0.20) 
Branch points 12.66 (0.75)* 15.30 (1.55)** 10.38 (1.06) 11.53 (0.97) 
Total process length (μm) 309.62 (14.5)* 317.86 (21.40)* 253.02 (19.49) 246.38 (15.58) 
Total process volume (μm357.78 (3.93) 64.31 (5.32) 44.11 (4.53) 52.46 (4.28) 
Convex hull area (μm22151.4 (105.76)* 2042.3 (111.68)* 1734.6 (136.77) 1667.6 (102.11) 
 CON STR CON + M STR + M 
Cell body perimeter (μm) 32.03 (0.86) 33.63 (1.47) 30.07 (0.65) 32.35 (1.02) 
k-dim 1.029 (0.003) 1.035 (0.004) 1.028 (0.002) 1.025 (0.003) 
Number of processes originating from soma 5.29 (0.25)* 5.25 (0.22)* 4.65 (0.19) 4.34 (0.20) 
Branch points 12.66 (0.75)* 15.30 (1.55)** 10.38 (1.06) 11.53 (0.97) 
Total process length (μm) 309.62 (14.5)* 317.86 (21.40)* 253.02 (19.49) 246.38 (15.58) 
Total process volume (μm357.78 (3.93) 64.31 (5.32) 44.11 (4.53) 52.46 (4.28) 
Convex hull area (μm22151.4 (105.76)* 2042.3 (111.68)* 1734.6 (136.77) 1667.6 (102.11) 

*P < 0.05 compared with STR + M and CON + M groups.

**P < 0.05 compared with CON, STR + M, and CON + M groups.

The range of cell sizes as assessed by convex hull area across all groups varied between 401.56 and 4158.73 μm2. We have therefore included descriptive statistics for the 25th (<1800 μm2) and 75th (>2000 μm2) percentiles of cells according to their convex hull area (μm2), henceforth referred to as the cell's “footprint” in order to describe the heterogeneity of the sample. In both stress and control groups, 50% of the surveyed cells were over 2000 μm2. Note that stress predominantly affects the top 25% of cells. An independent samples non-parametric Kruskal–Wallis ANOVA was performed on all 4 groups to determine whether the distribution of cell sizes was significantly affected by either stress or minocycline administration. This was not significant (P = 0.33), which suggests that neither stress nor minocycline affects the ratio of small versus large microglia (Fig. 2).

Figure 2.

Representative Neurolucida tracings rendered from (A) small CON cell; (B) small STR cell; (C) large CON cell; (D) large STR cell; (E) small CON + M cell; (F) small STR + M cell; (G) large CON + M cell; and (H) large STR + M cell. Note the hyper-ramified appearance of the large STR cell (D). Scale bar = 20 μm.

Figure 2.

Representative Neurolucida tracings rendered from (A) small CON cell; (B) small STR cell; (C) large CON cell; (D) large STR cell; (E) small CON + M cell; (F) small STR + M cell; (G) large CON + M cell; and (H) large STR + M cell. Note the hyper-ramified appearance of the large STR cell (D). Scale bar = 20 μm.

The Effect of Stress on Microglial Cells is Magnified in the Upper 75th Percentile of Cells, an Effect Blocked by Minocycline

Analysis revealed that there was a significant effect of group on k-dim values (F3,39 = 4.54, P = 0.008) branch points (F3,39 = 6.6, P = 0.001), and process length (F3,39 = 7.82, P < 0.001). Post hoc tests showed that stress increased the k-dim value (all P < 0.01), branch points (all P < 0.001), and total process length (all P < 0.005) in comparison to all other treatment groups. This suggests that stress increases the internal complexity (degree of branching) of the cells. There was a significant effect of group on convex hull area (F3,39 = 5.89, P = 0.002). This was due to minocycline administration resulting in reduced cell area when compared with both STR and CON groups (all P < 0.01). Note that there are no differences between groups for soma size, number of processes, and process volume (see also Table 2).

Table 2

Morphological characteristics of large microglial cells

 CON STR CON + M STR + M 
Cell body perimeter (μm) 30.8 (1.43) 31.52 (1.67) 31.25 (1.45) 33.62 (1.79) 
k-dim 1.031 (0.003) 1.065 (0.01)* 1.038 (0.005) 1.038 (0.005) 
Number of processes originating from soma 5.1 (0.53) 5.4 (0.54) 4.4 (0.45) 4.4 (0.31) 
Branch points 15.2 (1.25) 27.2 (3.43)* 15.9 (1.96) 16.0 (1.71) 
Total process length (μm) 394.04 (13.68) 509.02 (35.82)* 365.72 (34.42) 332.87 (18.64) 
Total process volume (μm379.56 (7.04) 95.86 (9.53) 67.13 (9.46) 75.90 (6.64) 
Convex hull area (μm22946.2 (114.7) 3082.24 (113.8)** 2650.1 (208.0) 2330.5 (80.2) 
 CON STR CON + M STR + M 
Cell body perimeter (μm) 30.8 (1.43) 31.52 (1.67) 31.25 (1.45) 33.62 (1.79) 
k-dim 1.031 (0.003) 1.065 (0.01)* 1.038 (0.005) 1.038 (0.005) 
Number of processes originating from soma 5.1 (0.53) 5.4 (0.54) 4.4 (0.45) 4.4 (0.31) 
Branch points 15.2 (1.25) 27.2 (3.43)* 15.9 (1.96) 16.0 (1.71) 
Total process length (μm) 394.04 (13.68) 509.02 (35.82)* 365.72 (34.42) 332.87 (18.64) 
Total process volume (μm379.56 (7.04) 95.86 (9.53) 67.13 (9.46) 75.90 (6.64) 
Convex hull area (μm22946.2 (114.7) 3082.24 (113.8)** 2650.1 (208.0) 2330.5 (80.2) 

*P < 0.05 compared with CON, STR + M, and CON + M groups.

**P < 0.05 compared with STR + M and CON + M groups.

There is no Effect of Stress on the Microglial Cells in the Lower 25th Percentile of Cells

There was a significant effect of group on the number of processes (F3,39 = 3.4, P = 0.03), branch points (F3,39 = 3.27, P = 0.03), process length (F3,39 = 9.82, P < 0.001), and convex hull area (F3,39 = 13.3, P < 0.001). While there was no effect of stress on the bottom 25% of cells, there was an effect of minocycline, with branch points, process length, and convex hull area reduced in minocycline groups compared with both STR and CON groups. Again, there is no difference between groups in the k-dim values, cell body size, process number, or process volume (see also Table 3).

Table 3

Morphological characteristics of small microglial cells

 CON STR CON + M STR + M 
Cell body perimeter (μm) 32.44 (0.98) 29.64 (1.32) 27.93 (0.7) 29.24 (1.83) 
k-dim 1.031 (0.006) 1.021 (0.003) 1.019 (0.004) 1.017 (0.003) 
Number of processes originating from soma 5.5 (0.31) 4.8 (0.47) 4.6 (0.22) 4.0 (0.29) 
Branch points 9.4 (0.65)* 8.8 (0.81)* 7.2 (0.57) 6.9 (0.62) 
Total process length (μm) 224.74 (8.32)* 215.51 (7.25)* 181.56 (10.14) 161.63 (11.30) 
Total process volume (μm338.59 (5.27) 42.87 (4.35) 28.58 (2.03) 30.69 (3.01) 
Convex hull area (μm21469.43 (70.07)* 1420.75 (31.59)* 1196.14 (45.48) 1071.33 (51.38) 
 CON STR CON + M STR + M 
Cell body perimeter (μm) 32.44 (0.98) 29.64 (1.32) 27.93 (0.7) 29.24 (1.83) 
k-dim 1.031 (0.006) 1.021 (0.003) 1.019 (0.004) 1.017 (0.003) 
Number of processes originating from soma 5.5 (0.31) 4.8 (0.47) 4.6 (0.22) 4.0 (0.29) 
Branch points 9.4 (0.65)* 8.8 (0.81)* 7.2 (0.57) 6.9 (0.62) 
Total process length (μm) 224.74 (8.32)* 215.51 (7.25)* 181.56 (10.14) 161.63 (11.30) 
Total process volume (μm338.59 (5.27) 42.87 (4.35) 28.58 (2.03) 30.69 (3.01) 
Convex hull area (μm21469.43 (70.07)* 1420.75 (31.59)* 1196.14 (45.48) 1071.33 (51.38) 

*P < 0.05 compared with STR + M and CON + M groups.

Sholl Analysis for the Entire Sample Reveals That Stress Increases Branch Points in the Proximal Aspect of the Cell, and Minocycline Causes a Reduction in Primary Cellular Processes

The following parameters were quantified by the Sholl analysis: length, volume, and surface area of processes, average diameter of processes, intersections, and nodes (branch points; Fig. 3). Most effects of stress and drug administration occur in the proximal aspect of the cell. The most striking finding arising from the group-based Sholl analysis was the increase in branch points in cells from stressed animals between 18 and 28 µm from the cell soma. Other cellular properties overall appeared to be unaffected by stress. Minocycline had some effects on intersections, process length, surface area, and volume, reducing all of these measures typically between 8 and 28 μm from the cell soma, when compared with untreated groups. The average diameter of the processes was largely unaffected by either stress or minocycline (Fig. 3).

Figure 3.

Bar graphs depicting results from the Sholl analysis of total sample as a function of radius. Animals exposed to chronic stress exhibit a greater number of branch points than all other experimental groups at 18, 23, and 28 μm from the soma. Error bars represent SEM. *STR > all other groups (CON, CON + M, and STR + M). ^STR and CON > STR + M and CON + M. #STR > STR + M and CON + M (P < 0.05).

Figure 3.

Bar graphs depicting results from the Sholl analysis of total sample as a function of radius. Animals exposed to chronic stress exhibit a greater number of branch points than all other experimental groups at 18, 23, and 28 μm from the soma. Error bars represent SEM. *STR > all other groups (CON, CON + M, and STR + M). ^STR and CON > STR + M and CON + M. #STR > STR + M and CON + M (P < 0.05).

Branch Points

The proximal aspect of the cell contained the most branch points, with 92% of branch points occurring between 8 and 28 µm from the soma, regardless of treatment. Analysis of differences between the groups across each of the radii revealed a significant interaction between the radius and the group (F24,1104) = 161.66, P < 0.001). Planned comparisons revealed that this interaction was driven primarily by group differences at 18, 23, 28, and 38 µm from the soma. Post hoc comparisons revealed that there was a clear effect of stress, with STR possessing a greater number of branch points than any other group at 18, 23, 28, and 38 µm from the soma (all P < 0.05).

Intersections

The analysis revealed that independent of treatment, 89% of the intersections between cellular processes and the radii superimposed upon the cell for the Sholl analysis occur 8 and 28 µm from the cell soma. Analysis of differences between the groups across each of the Sholl radii revealed a significant interaction between the radius and the group (F24,1104 = 2.44, P < 0.001). Planned comparisons revealed that this interaction was driven primarily by differences between treatment groups at radii 13, 18, and 23 µm from the soma (all P< 0.05). Post hoc comparisons revealed that the stress group was not different from control at any radius. There was, however, a clear effect for minocycline, with both drug-treated groups (STR + M and CON + M) exhibiting fewer intersections compared with the stress and control groups at 13, 18, and 23 µm (all P < 0.05).

Length

The analysis showed that independent of any treatment, 77% of the process length was between 13 and 28 µm from the cell soma. Analysis of differences between the groups across each of the radii revealed a significant interaction between the radius and the group (F24, 104 = 2.64, P < 0.001). Planned comparisons revealed that this interaction was driven primarily by group differences at 8, 18, 23, and 28 µm from the soma (all P< 0.05). Post hoc comparisons revealed that the stress group was not different from control at any radius. There was, however, a clear effect for minocycline, with both drug-treated groups (STR + M and CON + M) exhibiting cells with shorter processes compared with the stress and control groups at 18 and 23 µm (all P < 0.05).

Surface Area

Analysis showed that independent of any treatment, 87.5% of the surface area of the cellular processes was between 8 and 28 µm from the cell soma. Analysis of differences between the groups across each of the radii revealed a significant interaction between the radius and the group (F24,1104 = 2.64, P < 0.001). Planned comparisons revealed that this interaction was driven primarily by group differences at 8, 18, 23, and 28 µm from the cell soma. Post hoc comparisons revealed that the stress group was not significantly different from control at any radius. However, there was a clear effect of minocycline administration, with both drug-treated groups (STR + M and CON + M) possessing a smaller process surface area at each of these levels than either the STR or the CON group (all P < 0.05).

Volume

Eighty-one percent of the volume of the cellular processes was between 8 and 23 µm from the cell soma. Analysis of differences between the groups across each of the radii revealed a significant interaction between the radius and the group (F24,1104 = 2.43, P < 0.001). Planned comparisons revealed that this interaction was driven primarily by group differences at 8, 13, 18, 23, and 28 µm from the soma. Post hoc comparisons revealed that there was an effect of stress at 18 µm (STR significantly greater than all other treatment groups, all P < 0.05) and an effect of minocycline, with STR possessing a significantly greater volume than both minocycline-treated groups at 23 and 28 µm from the soma (all P < 0.05).

Average Diameter

The average diameter of the cellular processes decreases distally from the soma, with processes 42% thicker between 0 and 23 µm from the soma than the remainder. Analysis of differences between groups across each of the radii revealed a significant main effect of radius (F8,1104 = 161.66, P < 0.001).

Sholl Analysis for Large Cells Reveals That Exposure to Stress Increases Branch Points, Intersections and Total Process Length in the Proximal Aspect of the Cell, an Effect Reversed by Minocycline

In large cells (top 25%), we found no difference in cell size or soma size, or average diameter of processes. The most prominent differences were contained to branching indices only; stress increased branch points (nodes), intersections, and process length 18–28 µm from the soma (Fig. 4).

Figure 4.

Bar graphs depicting results from the Sholl analysis of large cells (top 75th percentile of sample) as a function of radius. Animals exposed to chronic stress exhibited greater numbers than other experimental groups of branch points, intersections, and process length, all indicators of secondary branching. Error bars represent SEM. *STR > CON, CON + M, STR + M. STR > CON and CON + M (P < 0.05).

Figure 4.

Bar graphs depicting results from the Sholl analysis of large cells (top 75th percentile of sample) as a function of radius. Animals exposed to chronic stress exhibited greater numbers than other experimental groups of branch points, intersections, and process length, all indicators of secondary branching. Error bars represent SEM. *STR > CON, CON + M, STR + M. STR > CON and CON + M (P < 0.05).

Branch Points

Analysis of differences between groups across radii revealed a significant interaction between the radius and the group (F24,288 = 2.27, P < 0.001). Planned comparisons revealed that group differences were significant at 18, 23, and 28 μm from the soma (all P < 0.05). Post hoc tests indicated that this was due to STR possessing a greater number of branch points at these levels than CON, CON + M, or STR + M (P < 0.05).

Intersections

Analysis of differences between the groups across each radii revealed a significant main effect of both the radius (F8,288 = 145.88, P < 0.001) and the group (F3,36 = 6.85, P < 0.001). Planned comparisons revealed that these effects were driven primarily by significant differences between groups at 18, 23, and 28 μm from the cell soma (all P < 0.05). Post hoc comparisons revealed that there was a clear effect of stress, with STR possessing a greater number of intersections than any other group at 18, 23, and 28 μm (all P < 0.05).

Total Process Length

Analysis of differences between the 4 groups across each of the radii revealed a significant interaction between the group and the radius (F24,288 = 1.95, P = 0.006). Planned comparisons indicated that group differences were significant at levels 8, 18, 23, 28, and 33 μm from the cell soma (all P < 0.05). Post hoc comparisons revealed that there was a clear effect of stress, with STR possessing a greater process length at levels 18, 23, 28, and 33 μm than the other 3 groups (all P < 0.05).

Surface Area

Analysis of differences between groups across each radii revealed that there was a significant interaction between the group and the radius (F24,288 = 1.85, P = 0.011). Planned comparisons revealed that group differences were significant at 18, 23, and 28 μm from the cell soma (all P < 0.05). Post hoc tests indicated that there was a clear effect of stress, with STR possessing a significantly larger surface area than other groups at 18, 23, and 28 μm (all P < 0.05).

Volume

Analysis of differences between groups across each radii revealed a significant interaction between the group and the radius (F24,288 = 1.93, P = 0.007). Planned comparisons revealed that group differences were significant at 18 μm from the soma (P < 0.05). Post hoc tests indicated that this was due to an increased process volume in the STR group compared with CON and CON + M groups at this level (all P < 0.05).

Average Diameter

Analysis of differences between groups across the radii revealed that there was a significant main effect of the radius on average diameter (F8,288 = 23.52, P < 0.001). While planned comparisons revealed a group difference, this was at 43 μm from the cell soma only and was due to CON + M group possessing a greater diameter than the other 3 groups (all P < 0.05).

Sholl Analysis for Small Cells Reveals That There is No Effect of Stress on any Index

Branch Points

Analysis of differences between the 4 treatment groups across radii revealed a significant main effect of the radius (F8,288 = 61.71, P < 0.001) and the group (F3,36 = 3.27, P = 0.032). Planned comparisons revealed that there were significantly greater numbers of branch points at 23 μm from the cell soma across all groups (P < 0.05). Post hoc tests indicated that there were no significant differences between groups at this level (Fig. 5).

Figure 5.

Bar graphs depicting results from the Sholl analysis of small cells (lower 25th percentile of sample) as a function of radius. There was no effect of stress on any measure in this group of cells. Error bars represent SEM. #STR > CON + M and STR + M (P < 0.05).

Figure 5.

Bar graphs depicting results from the Sholl analysis of small cells (lower 25th percentile of sample) as a function of radius. There was no effect of stress on any measure in this group of cells. Error bars represent SEM. #STR > CON + M and STR + M (P < 0.05).

Intersections

Analysis of differences between the 4 treatment groups across radii revealed a significant interaction between the group and the radius (F24,288 = 1.97, P = 0.005). Planned comparisons revealed that this interaction was due to a significant effect of the group at 18 μm from the cell soma (all P < 0.05). Post hoc tests indicated that this was due to CON possessing a greater number of intersections than STR + M, and STR possessing a greater number of intersections than CON + M.

Length

Analysis of differences between the 4 treatment groups across radii revealed a significant interaction between the group and the radius (F24,288 = 2.35, P = 0.001). Planned comparisons revealed that this interaction was due to a significant effect of the group at 18, 23, and 28 μm from the cell soma (all P < 0.05). Post hoc testing indicated that this was due mainly to a small minocycline effect, with STR + M being shorter than STR and CON at 18 and 23 μm.

Surface Area

Analysis of the 4 treatment groups across radii revealed a significant interaction between the group and the radius (F24,288 = 2.29, P < 0.001). Planned comparisons revealed that this interaction was due to a significant effect of the group at 18, 23, and 28 μm from the cell soma (all P < 0.05). Post hoc tests revealed that this was due to an effect of minocycline administration, with STR possessing a greater surface area than STR + M at 18, 23, and 28 μm, and STR possessing a greater surface area than CON + M at 28 μm (all P < 0.05).

Volume

Analysis of differences between the 4 treatment groups across radii revealed a significant main effect of the radius (F8,288 = 97.27, P < 0.001). Planned comparisons revealed that this was due to a significant effect of the group at 23 and 28 μm from the cell soma (all P < 0.05). Post hoc tests indicated that this was due to an effect of minocycline administration, with STR possessing a greater process volume than CON + M and STR + M at both 23 and 28 μm (all P < 0.05).

Average Diameter

Analysis of differences between the 4 treatment groups across radii revealed a significant main effect of the radius (F8,288 = 91.55, P < 0.001). Planned comparisons however revealed no significant effect of group at any level.

Sholl Analysis of Secondary Motor Cortex Reveals no Effect of Stress on Microglia

The following parameters were quantified by the Sholl analysis for the secondary motor cortex: length, volume, and surface area of processes, average diameter of processes, intersections, and nodes. There was no effect of stress on any of these characteristics.

Exposure to Stress Causes Significant Reduction in Animal Weight

At the completion of the protocol, stressed rats weighed significantly less (ca. 15%) than handled controls (P < 0.001). This difference was primarily due to weight loss in the first week followed by a slower rate of gain over the remainder of the restraint period and is comparable to weight changes reported in other studies using the same protocol (Radley et al. 2006).

Immunolabeling Indicates That There Was Neither Evidence of Programmed Cell Death Nor Microglial Markers of Tissue Insult/Injury

We found no MHC-II, CD68, or activated caspase-3 immunoreactivity nor any cells positive for the TUNEL labeling in the mPFC of animals from any experimental group (images not shown). We also examined sections for pro-inflammatory cytokine IL-1β, and while there was inter- and extracellular IL-1β present in the tissue, there was no significant effect of stress (P > 0.05). This suggests that the changes we observe in microglial activity are not due to inflammation, tissue damage, or programmed cell death.

Stress Increases β1-Integrin (CD29) Immunoreactivity

Overall, stress significantly increases the level of β1-integrin immunoreactive material. A one-way ANOVA showed that there was a significant effect of the group on the levels of β1-integrin (F3,122 = 30.74, P < 0.01). Post hoc tests revealed that this was due to an effect of both stress and minocycline administration, with STR possessing a greater amount of β1-integrin than either the CON, STR + M, or CON + M group (all P < 0.01). The CON group also possesses a significantly greater amount of β1-integrin immunolabeling than either the STR + M or CON + M group (all P < 0.05).

For the lower 25th percentile of cells according to the size, there was a significant effect of the group on CD29 immunoreactivity (F3,39 = 5.03, P < 0.01). Post hoc testing revealed that this was due to the STR group possessing greater levels of CD29 immunoreactivity than either the CON, STR + M, or CON + M group (all P < 0.05). There were no significant differences between other groups (Fig. 6).

Figure 6.

Stress increases levels of β1-integrin immunoreactivity specific to microglia. (A). Levels of β1-integrin immunoreactivity in small microglia. (B). Levels of β1-integrin immunoreactivity in large microglia. In both groups of cells, stress significantly increases levels of β1-integrin relative to control, and minocycline reverses this effect. The effect of minocycline is more significant in large cells. The panel of images below depicts representative photomicrographs of double-labeled cells (Iba-1, green; and β1-integrin, red) from small and large cells of each experimental group. *P < 0.05. Scale bar = 10 µm.

Figure 6.

Stress increases levels of β1-integrin immunoreactivity specific to microglia. (A). Levels of β1-integrin immunoreactivity in small microglia. (B). Levels of β1-integrin immunoreactivity in large microglia. In both groups of cells, stress significantly increases levels of β1-integrin relative to control, and minocycline reverses this effect. The effect of minocycline is more significant in large cells. The panel of images below depicts representative photomicrographs of double-labeled cells (Iba-1, green; and β1-integrin, red) from small and large cells of each experimental group. *P < 0.05. Scale bar = 10 µm.

For the upper 25th percentile according to the cell size, there was a significant effect of group on CD29 immunoreactivity (F3,39 = 16.79, P < 0.01). Post hoc testing revealed that this was due to the STR group possessing greater levels of immunoreactivity than either the CON, STR + M or CON + M group (all P < 0.01). The CON group also had significantly greater levels of immunoreactivity than the CON + M group (P = 0.012; Fig. 6).

Discussion

To determine how exposure to chronic stress altered microglia, we undertook a study, the first of its type, to reconstruct microglia within the PFC, and quantitatively analyze changes in their morphology. These analyses revealed 6 major results: 1) chronic stress increased the level of microglial process branching, and this enhanced ramification was restricted to the secondary branches and above of the cell; 2) these effects were magnified in a subpopulation consisting of the largest but not the smallest cells in our sample, with stress increasing secondary ramification without altering the overall footprint of the cell; 3) administration of minocycline, an inhibitor of microglial activation, attenuated the effects of stress and appeared to induce mild de-ramification of microglial cells, an effect that was independent of experimental condition; 4) the stress-induced morphological alterations were relatively be specific to the mPFC, as no observable changes occurred within the secondary motor cortex; 5) the stress-induced enhancement of microglial ramification is not driven by CNS inflammation or injury as we could find no evidence of increased expression of IL-1β, MHC-II, CD68, TUNEL, or activated caspase-3; 6) we did, however, find that exposure to stress significantly increased the microglial-specific expression of β1-integrin and that this was blocked by minocycline treatment. Together, these results provide the first quantitative evidence that chronic stress promotes a specific form of microglial hyper-ramification and that this effect is associated with a stress-induced alteration in β1-integrin expression. Finally, the microglial phenotype induced by exposure to chronic stress is markedly different from those that have traditionally been observed in response to injury and most likely represents an adaptive response to this form of environmental challenge.

In the current study, we focused specifically on changes in the mPFC, as we had previously observed significant microglial changes in this region (Tynan et al. 2010; Hinwood, Morandini et al. 2011). In these previous studies, we detected microglial alterations by quantifying region-wide changes in immunoreactive (Iba-1) material. While this approach is the most extensively used to examine changes in glial populations, it is based on a thresholding procedure (Sugama et al. 2007; Tynan et al. 2010), which is insensitive to specific changes in the phenotype. Accordingly, when differences are observed using this technique, microglia are simply referred to as having become “activated”. Certainly, inferences can be made about the specific state of activation (reactive, amoeboid, etc.) when combined with qualitative observations of labeled cells; however, it is not clear how reliable these inferences are. The reconstruction approach undertaken in the current study circumvents the difficulties associated with an empirically establishing microglial phenotype.

While microglia respond to and interact with other cells within the CNS, they are unique, in that under non-pathological conditions, they are uncoupled from one another (Graeber 2010; Graeber and Streit 2010). Indeed, stereological evidence concerning the spatial distribution of microglia within the hippocampus has demonstrated that microglia appear to be repulsed from one another, with each cell maintaining its own independent footprint (Jinno et al. 2007). Our observations support this and have also revealed certain other characteristics of microglia not elsewhere reported. Microglia within the mPFC, independent of treatment (and excluding those treated with minocycline), cover approximately 2000 μm2 based on their convex hull area, exhibit 5 processes extending from their soma (primary processes), and on average possess 12 branch points. The Sholl analyses further revealed that almost all features of microglia (length, thickness, and surface area of processes, as well as process branching) reached their maximum levels between 8 and 18 μm from the cell soma.

Our initial analysis of the morphology data gathered in the current study involved examining the cells in each of the 4 groups. This investigation revealed that stress significantly increased the number of process branch points by approximately 20% in relation to all other groups. The magnitude of this difference is consistent with what we have previously observed when quantifying region-wide changes in immunoreactive material (Tynan et al. 2010) and stands in marked contrast to the very substantial changes (upwards of 400%) that are frequently reported in response to tissue insults such as facial nerve axotomy (Kreutzberg 1996; Graeber et al. 1998). Additional investigation of the branching data revealed one further intriguing difference. Specifically, we observed that there was no effect of stress on the number of primary processes (i.e. those emerging from the soma). This finding indicates that the overall difference in branching is driven primarily by alterations in secondary branching (or above), which suggests that alterations in microglial ramification can be restricted to particular components of the microglial process. Also of note was the fact that these stress-induced changes were not evident in the secondary motor cortex, an area not widely recognized to be altered by exposure to stress, a result indicating that the stress-induced alterations are somewhat circuit-specific.

Minocycline clearly blocked the effect of stress on microglial branching. Indeed, we observed that minocycline, independent of treatment, reduced the number of primary processes (20%), the degree of microglial branching (21%), the combined process length (19%), and the overall area that the cell occupied within the CNS (23%), relative to handled controls. As minocycline decreased several morphological parameters, including primary branching, and did so in both the stress and control groups, it would seem unlikely that minocycline achieves its effect by simply inhibiting the pro-ramification effect of stress. While this is the first in vivo study to quantitatively examine the morphological alterations induced by minocycline, our findings appear to align well with an extensive literature, indicating the ability of minocycline to enhance microglial quiescence (Yrjanheikki et al. 1999; Tikka et al. 2001). One final point concerning the effectiveness of minocycline in blocking the effects of stress is that it remains to be determined how effective the drug is at attenuating stress-induced changes when the commencement of treatment is delayed with respect to the onset of stress. From a clinical perspective, this is a highly significant issue and one that merits further investigation.

In our second analysis, we investigated whether the effects of stress were distinct for large and small microglia. This analysis was motivated by a previous study which demonstrated that stress-induced changes in neuronal architecture are size-dependent (Radley et al. 2008). The results from this additional analysis revealed that the morphology of microglia was similar for small cells from the stress and control groups. For large cells, however, we observed pronounced changes, with stress substantially increasing branching and total process length relative to all other groups. Given that the overall area of large microglia from the stress and control groups was not different, the stress-induced increase in branching and process length indicates that the “internal complexity” of the cell had increased. Indeed, this suggestion is supported by the fact that stress resulted in a significantly higher k-dim, a measure that reflects how completely a fractal object (i.e. the microglial cell) fills the area defined by its boundaries. These alterations are intriguing as it suggests that stress produces a form of morphological alteration that would allow microglial processes to make more contacts, or scan more efficiently, within its microenvironment. Again we observed that minocycline reversed the effects of stress, significantly reducing the complexity of the cell branching.

One additional issue that emerges with regard to the ability of stress to increase the complexity of larger cells is whether stress has caused a change in the overall numbers of small and large cells within the PFC. To address this issue, we examined the ratio of small to large cells in each of the 4 experimental groups. This analysis revealed that the ratio of small to large did not differ significantly across treatment groups. Accordingly, it appears that the ability of stress to enhance ramification is quite specific to the large microglial cells within the PFC.

The transition of ramified microglia to a hyper-ramified state has frequently been described in the literature (Streit et al. 1999). In the earliest study we could identify, Wilson and Molliver (1994) demonstrated that p-chloroamphetamine, which causes the degeneration of fine serotonergic axon terminals, appeared to increase the number of microglial processes. It is worthwhile noting, however, that this study did not directly quantify whether processes had in fact increased. Indeed, none of the studies described within the literature that we could identify, which referred to hyper-ramification directly quantified whether microglial processes (primary or secondary) had increased (Hurley and Coleman 2003; Roberts et al. 2004; Herber et al. 2006). The question of what drives hyper-ramification is a significant one, as it has previously been proposed that hyper-ramification represents one of the earliest microglial transformations in response to injury (Streit et al. 1999). This proposition, however, appears to have been predicated upon those studies which have “observed” ramification without quantifying it. In one of the only documented studies to observe microglial responses to injury in real time, Stence et al. (2001) found no evidence that microglia transitioned into a hyper-ramified state, either initially or at any other time. Given this situation, it will be of great interest to establish whether hyper-ramification should be considered as part of the injury schema or whether in fact it is a response to intense neuronal activity.

In investigating the signals responsible for driving the stress-induced hyper-ramification, we first examined a variety of markers associated with inflammation and neurodegeneration. Specifically, we examined changes in MHC-II expression, a marker of antigen presentation; CD68 (ED-1), a marker frequently associated with microglial phagocytosis; IL-1β, a pro-inflammatory cytokine; and activated caspase-3 and TUNEL, both commonly deployed markers of apoptosis. This analysis revealed that none of these markers were elevated in animals exposed to chronic stress (or stress plus minocycline). This result, in combination with our previous densitometry data indicating that the magnitude of changes induced by stress is quite moderate, suggests that stress-induced changes in microglial morphology are unlikely to be due to CNS injury.

We next examined changes in the expression of β1-integrin (CD29), a heterodimeric cell adhesion transmemberane protein that has been shown to play a role in microglial ramification (Kloss et al. 2001). Specifically, it has been shown that microglial-specific β1-integrin expression is dramatically up-regulated in a variety of pathological conditions. Further, Ohsawa et al. (2010) have provided compelling in vitro evidence, demonstrating that β1-integrin mediates adenosine triphosphate-induced process extension. In the current study, we observed that chronic stress significantly increased microglial β1-integrin expression. While this was significant at the group level, we wished to determine whether this effect was size-dependent, in line with our other results. This analysis revealed that β1-integrin expression was increased in both small and large microglia, a result that indicates if β1-integrin does mediate the increased ramification of large cells it does so in concert with yet unidentified factors that are not present in smaller cells. We further observed that minocycline treatment significantly reduced the expression of β1-integrin following stress in both small and large cells. These results suggest that increased ramification observed following chronic stress may be mediated by stress-induced disruption of interactions between the cell and the extracellular matrix.

Summary and Future Directions

The major finding to emerge from the current study is that exposure to chronic stress induces several readily quantifiable alterations in microglial morphology. Specifically, we have observed that stress induces a unique form of hyper-ramification, whereby only branching in secondary processes (and above) is increased and this occurs without any significant change in process diameter. Strikingly, chronic stress does not alter the overall area occupied by the cell within the brain (i.e. its footprint), a result, which when taken together with the observed change in branching, indicates that exposure to stress significantly increases the internal complexity of PFC microglia. This result significantly extends existing knowledge concerning the effects of chronic stress on microglia. Previously, densitometric analysis of microglial changes in microglial-specific cytoskeletal-linked proteins had only been able to indicate that the microglia had changed, and now, it is clear at the cellular level exactly how they have been altered.

With respect to signals driving the stress-induced secondary hyper-ramification, we have again confirmed that chronic stress, using the repeated restraint paradigm, does not induce any measurable alterations in proteins associated with inflammation or neurodegeneration (Mattsson et al. 2006; Graeber et al. 2011). This finding, rather than being surprising, is consistent with observations that microglial de-ramification is associated with an increased level of inflammatory signaling within the brain (Wynne et al. 2009; Kettenmann et al. 2011).

In contrast to the rich literature concerning microglial de-ramification, research on hyper-ramification is sparse. Certainly, several in vitro investigations have investigated this issue, but they have primarily addressed the question of what conditions are necessary for amoeboid microglia (the forms that are typically observed in standard culture conditions) to develop processes equivalent to those seen in vivo. This research has identified a variety of factors that promote ramification such as astrocyte conditioned medium (Wilms et al. 1997), gangliosides (Park et al. 2008), and antioxidants (Heppner et al. 1998). Among the most consistently identified molecules associated with ramification has been β1-integrin (Kloss et al. 2001; Ohsawa et al. 2010). In the current study, the effect of stress on β1-integrin was unequivocal. This finding is extremely interesting as it suggests that stress may be involved in disrupting either the extracellular matrix, and/or the proteins that microglia employ to adhere and interact with the matrix. This is a completely novel area of stress-related microglial research, and accordingly, many issues are yet to be investigated. Perhaps, most pressing is the temporal profile of alterations induced in microglial specific β1-integrin expression, and changes in the expression of other integrins that are also known to contribute to extracellular matrix adhesion.

Finally, in terms of placing the current set of results within a broader neurobehavioral context, we have previously reported that the chronic stress paradigm used in the current study produces a significant working memory impairment and that this impairment occurs in conjunction with an increase in the density of PFC microglial labeling (Hinwood, Morandini et al. 2011). We also reported that the microglial alterations and working memory deficit could be reversed by minocycline treatment (Hinwood, Morandini et al. 2011). Merging our current results with our prior findings, we now have reasonable evidence to suggest that stress-induced microglial hyper-ramification is associated with the emergence of working memory deficits and that reversal of this hyper-ramification is associated with the restoration of working memory performance. Accordingly, we propose that the increased ramification of microglia represents an important neurobiological mechanism mediating the neurobiological effects of chronic psychological stress on the PFC.

Funding

These studies were supported by funding from the Australian National Health and Medical Research Council, the Hunter Medical Research Institute, and the University of Newcastle's Centre for Translational Neuroscience and Mental Health Research.

Notes

We thank Ms Britt Saxby for her technical assistance and our peer reviewers for their constructive and insightful comments. Conflict of Interest: None declared.

References

Ahmed
Z
Shaw
G
Sharma
VP
Yang
C
McGowan
E
Dickson
DW
Actin-binding proteins coronin-1a and IBA-1 are effective microglial markers for immunohistochemistry
J Histochem Cytochem
 , 
2007
, vol. 
55
 (pg. 
687
-
700
)
Corona
AW
Huang
Y
O'Connor
JC
Dantzer
R
Kelley
KW
Popovich
PG
Godbout
JP
Fractalkine receptor (CX3CR1) deficiency sensitizes mice to the behavioural changes induced by lipopolysaccharide
J Neuroinflammation
 , 
2010
, vol. 
7
 (pg. 
93
-
107
)
Fernandez
E
Jelinek
HF
Use of fractal theory in neuroscience: methods, advantages, and potential problems
Methods
 , 
2001
, vol. 
24
 (pg. 
309
-
321
)
Gabbott
PL
Warner
TA
Jays
PR
Salway
P
Busby
SJ
Prefrontal cortex in the rat: projections to subcortical autonomic, motor, and limbic centers
J Comp Neurol
 , 
2005
, vol. 
492
 (pg. 
145
-
177
)
Graeber
MB
Changing face of microglia
Science
 , 
2010
, vol. 
330
 (pg. 
783
-
788
)
Graeber
MB
Li
W
Rodriguez
ML
Role of microglia in CNS inflammation
FEBS Lett
 , 
2011
, vol. 
585
 (pg. 
3798
-
3805
)
Graeber
MB
Lopez-Redondo
F
Ikoma
E
Ishikawa
M
Imai
Y
Nakajima
K
Kreutzberg
GW
Kohsaka
S
The microglia/macrophage response in the neonatal rat facial nucleus following axotomy
Brain Res
 , 
1998
, vol. 
813
 (pg. 
241
-
253
)
Graeber
MB
Streit
WJ
Microglia: biology and pathology
Acta Neuropathol
 , 
2010
, vol. 
119
 (pg. 
89
-
105
)
Heppner
FL
Roth
K
Nitsch
R
Hailer
NP
Vitamin E induces ramification and downregulation of adhesion molecules in cultured microglial cells
Glia
 , 
1998
, vol. 
22
 (pg. 
180
-
188
)
Herber
DL
Maloney
JL
Roth
LM
Freeman
MJ
Morgan
D
Gordon
MN
Diverse microglial responses after intrahippocampal administration of lipopolysaccharide
Glia
 , 
2006
, vol. 
53
 (pg. 
382
-
391
)
Hinwood
M
Morandini
J
Day
TA
Walker
FR
Evidence that microglia mediate the neurobiological effects of chronic psychological stress on the medial prefrontal cortex
Cereb Cortex
 , 
2011
, vol. 
22
 (pg. 
1442
-
1454
)
Hinwood
M
Tynan
RJ
Day
TA
Walker
FR
Repeated social defeat selectively increases deltaFosB expression and histone H3 acetylation in the infralimbic medial prefrontal cortex
Cereb Cortex
 , 
2011
, vol. 
21
 (pg. 
262
-
271
)
Horvath
RJ
DeLeo
JA
Morphine enhances microglial migration through modulation of P2X4 receptor signaling
J Neurosci
 , 
2009
, vol. 
29
 (pg. 
998
-
1005
)
Hurley
SD
Coleman
PD
Facial nerve axotomy in aged and young adult rats: analysis of the glial response
Neurobiol Aging
 , 
2003
, vol. 
24
 (pg. 
511
-
518
)
Imai
Y
Kohsaka
S
Intracellular signaling in M-CSF-induced microglia activation: role of Iba1
Glia
 , 
2002
, vol. 
40
 (pg. 
164
-
174
)
Jinno
S
Fleischer
F
Eckel
S
Schmidt
V
Kosaka
T
Spatial arrangement of microglia in the mouse hippocampus: a stereological study in comparison with astrocytes
Glia
 , 
2007
, vol. 
55
 (pg. 
1334
-
1347
)
Jurgens
HA
Johnson
RW
Dysregulated neuronal-microglial cross-talk during aging, stress and inflammation
Exp Neurol
 , 
2012
, vol. 
233
 (pg. 
40
-
48
)
Kettenmann
H
Hanisch
UK
Noda
M
Verkhratsky
A
Physiology of microglia
Physiol Rev
 , 
2011
, vol. 
91
 (pg. 
461
-
553
)
Kloss
CU
Bohatschek
M
Kreutzberg
GW
Raivich
G
Effect of lipopolysaccharide on the morphology and integrin immunoreactivity of ramified microglia in the mouse brain and in cell culture
Exp Neurol
 , 
2001
, vol. 
168
 (pg. 
32
-
46
)
Kreutzberg
GW
Microglia: a sensor for pathological events in the CNS
Trends Neurosci
 , 
1996
, vol. 
19
 (pg. 
312
-
318
)
Krishnan
V
Nestler
EJ
The molecular neurobiology of depression
Nature
 , 
2008
, vol. 
455
 (pg. 
894
-
902
)
Mattsson
P
Delfani
K
Janson
AM
Svensson
M
Motor neuronal and glial apoptosis in the adult facial nucleus after intracranial nerve transaction
J Neurosurg
 , 
2006
, vol. 
104
 (pg. 
411
-
418
)
McEwen
BS
The neurobiology of stress: From serendipity to clinical relevance
Brain Res
 , 
2000
, vol. 
886
 (pg. 
172
-
189
)
Morshedi
MM
Meredith
GE
Differential laminar effects of amphetamine on prefrontal parvalbumin interneurons
Neuroscience
 , 
2007
, vol. 
149
 (pg. 
617
-
624
)
Nutile-McMenemy
N
Elfenbein
A
Deleo
JA
Minocycline decreases in vitro microglial motility, beta1-integrin, and Kv1.3 channel expression
J Neurochem
 , 
2007
, vol. 
103
 (pg. 
2035
-
2046
)
Ohsawa
K
Irino
Y
Sanagi
T
Nakamura
Y
Suzuki
E
Inoue
K
Kohsaka
S
P2Y12 receptor-mediated integrin-beta1 activation regulates microglial process extension induced by ATP
Glia
 , 
2010
, vol. 
58
 (pg. 
790
-
801
)
Park
JY
Kim
HY
Jou
I
Park
SM
GM1 induces p38 and microtubule dependent ramification of rat primary microglia in vitro
Brain Res
 , 
2008
, vol. 
1244
 (pg. 
13
-
23
)
Paxinos
G
Watson
C
The rat brain in stereotaxic coordinates
 , 
2005
San Diego (CA)
Elsevier Academic Press
Pittenger
C
Duman
RS
Stress, depression, and neuroplasticity: A convergence of mechanisms
Neuropsychopharmacology
 , 
2008
, vol. 
33
 (pg. 
88
-
109
)
Radley
JJ
Rocher
AB
Miller
M
Janssen
WGM
Liston
C
Hof
PR
McEwen
BS
Morrison
JH
Repeated stress induces dendritic spine loss in the rat medial prefrontal cortex
Cereb Cortex
 , 
2006
, vol. 
16
 (pg. 
313
-
320
)
Radley
JJ
Rocher
AB
Rodriguez
A
Ehlenberger
DB
Dammann
M
McEwen
BS
Morrison
JH
Wearne
SL
Hof
PR
Repeated stress alters dendritic spine morphology in the rat medial prefrontal cortex
J Comp Neurol
 , 
2008
, vol. 
507
 (pg. 
1141
-
1150
)
Rio-Hortega
PD
Penfield
W
Microglia
Cytology and cellular pathology of the nervous system
 , 
1932
New York
Hoeber
(pg. 
483
-
534
)
Roberts
ES
Masliah
E
Fox
HS
CD163 identifies a unique population of ramified microglia in HIV encephalitis (HIVE)
J Neuropathol Exp Neurol
 , 
2004
, vol. 
63
 (pg. 
1255
-
1264
)
Shansky
RM
Hamo
C
Hof
PR
McEwen
BS
Morrison
JH
Stress induced dendritic remodeling in the prefrontal cortex is circuit specific
Cereb Cortex
 , 
2009
, vol. 
19
 (pg. 
2479
-
2484
)
Sholl
DA
The measurable parameters of the cerebral cortex and their significance in its organization
Prog Neurobiol
 , 
1956
, vol. 
2
 (pg. 
324
-
333
)
Stence
N
Waite
M
Dailey
ME
Dynamics of microglial activation: a confocal time-lapse analysis in hippocampal slices
Glia
 , 
2001
, vol. 
33
 (pg. 
256
-
266
)
Streit
WJ
Walter
SA
Pennell
NA
Reactive microgliosis
Prog Neurobiol
 , 
1999
, vol. 
57
 (pg. 
563
-
581
)
Sugama
S
Takenouchi
T
Kitani
H
Fujita
M
Hashimoto
M
Activin as an anti-inflammatory cytokine produced by microglia
J Neuroimmunol
 , 
2007
, vol. 
192
 (pg. 
31
-
39
)
Sugama
S
Takenouchi
T
Kitani
H
Fujita
M
Hashimoto
M
Differential microglial activation between acute stress and lipopolysaccharide treatment
J Neuroimmunol
 , 
2009
, vol. 
15
 (pg. 
24
-
31
)
Summers
L
Kielty
C
Pinteaux
E
Adhesion to fibronectin regulates interleukin-1 beta expression in microglial cells
Mol Cell Neurosci
 , 
2009
, vol. 
41
 (pg. 
148
-
155
)
Tikka
T
Fiebich
BL
Goldsteins
G
Keinanen
R
Koistinaho
J
Minocycline, a tetracycline derivative, is neuroprotective against excitotoxicity by inhibiting activation and proliferation of microglia
J Neurosci
 , 
2001
, vol. 
21
 (pg. 
2580
-
2588
)
Tremblay
Lowery
RL
Majewska
AK
Microglial interactions with synapses are modulated by visual experience
PLoS Biol
 , 
2010
, vol. 
8
 pg. 
e1000527
 
Tynan
RJ
Naicker
S
Hinwood
M
Nalivaiko
E
Buller
KM
Pow
DV
Day
TA
Walker
FR
Chronic stress alters the density and morphology of microglia in a subset of stress-responsive brain regions
Brain Behav Immun
 , 
2010
, vol. 
24
 (pg. 
1058
-
1068
)
Wilms
H
Hartmann
D
Sievers
J
Ramification of microglia, monocytes and macrophages in vitro: influences of various epithelial and mesenchymal cells and their conditioned media
Cell Tissue Res
 , 
1997
, vol. 
287
 (pg. 
447
-
458
)
Wilson
MA
Molliver
ME
Microglial response to degeneration of serotonergic axon terminals
Glia
 , 
1994
, vol. 
11
 (pg. 
18
-
34
)
Wohleb
ES
Hanke
ML
Corona
AW
Powell
ND
Stiner
LM
Bailey
MT
Nelson
RJ
Godbout
JP
Sheridan
JF
β-adrenergic receptor antagonism prevents anxiety-like behavior and microglial reactivity induced by repeated social defeat
J Neurosci
 , 
2011
, vol. 
31
 (pg. 
6277
-
6288
)
Wohleb
ES
Fenn
AM
Pacenta
AM
Powell
ND
Sheridan
JF
Godbout
JP
Peripheral innate immune challenge exaggerated microglia activation, increased the number of inflammatory CNS macrophages, and prolonged social withdrawal in socially defeated mice
Psychoneuroendocrinology
 , 
2012
Wynne
AM
Henry
CJ
Godbout
JP
Immune and behavioural consequences of microglial reactivity in the aged brain
Integr Comp Biol
 , 
2009
, vol. 
49
 (pg. 
254
-
266
)
Yrjanheikki
J
Tikka
T
Keinanen
R
Goldsteins
G
Chan
PH
Koistinaho
J
A tetracycline derivative, minocycline, reduces inflammation and protects against focal cerebral ischemia with a wide therapeutic window
Proc Natl Acad Sci USA
 , 
1999
, vol. 
96
 (pg. 
13496
-
13500
)