Report

Resting Microglial Cells Are Highly Dynamic Surveillants of Brain Parenchyma in Vivo

See allHide authors and affiliations

Science  27 May 2005:
Vol. 308, Issue 5726, pp. 1314-1318
DOI: 10.1126/science.1110647

Abstract

Microglial cells represent the immune system of the mammalian brain and therefore are critically involved in various injuries and diseases. Little is known about their role in the healthy brain and their immediate reaction to brain damage. By using in vivo two-photon imaging in neocortex, we found that microglial cells are highly active in their presumed resting state, continually surveying their microenvironment with extremely motile processes and protrusions. Furthermore, blood-brain barrier disruption provoked immediate and focal activation of microglia, switching their behavior from patroling to shielding of the injured site. Microglia thus are busy and vigilant housekeepers in the adult brain.

Microglial cells are the primary immune effector cells in the brain. In response to any kind of brain damage or injury, microglial cells become activated and undergo morphological as well as functional transformations. They are critically involved in lesions, neurodegenerative diseases, stroke, and brain tumors (1-4). Resident microglial cells in the healthy brain are thought to rest in a dormant state, whereas activation is associated with structural changes, such as motile branches or migration of somata (5, 6). However, because most tissue preparations represent traumatic injuries by themselves, key aspects of microglia function have remained elusive.

Here, we investigated microglia behavior in intact adult brains both during the resting state and immediately after local injury by using in vivo two-photon microscopy (7). We used transgenic mice showing specific expression of enhanced green fluorescent protein (EGFP) in resident microglia of the central nervous system (CNS). EGFP expression is achieved through placement of the EGFP reporter gene into the Cx3cr1 locus encoding the chemokine receptor CX3CR1 (8). Fluorescence images were acquired transcranially by using a thinned-skull preparation (fig. S1A) (9), except for cases that required direct access to the brain. Microglial cells had small rod-shaped somata from which numerous thin and highly ramified processes extended symmetrically (fig. S1B). Their three-dimensional distribution in vivo was rather homogeneous, displaying a territorial organization with typical cell-to-cell distances of 50 to 60 μm and volume densities of 6.5 × 103 ± 0.6 × 103 cells/mm3 and 6.4 × 103 ± 0.4 × 103 cells/mm3 in layer 1 and layer 2/3, respectively (n = 6 animals).

Time-lapse imaging experiments of up to 10 hours showed that somata of microglial cells generally remained fixed with only few signs of migration (5% of somata shifted their position by 1 to 2 μm per hour; 99 total cells; n = 12 animals). In contrast, microglial processes were remarkably motile, continuously undergoing cycles of de novo formation and withdrawal. These structural dynamics occurred on a time scale of minutes, leading to comprehensive changes in cellular morphologies within 1 hour except for a small scaffold of stable branches (Fig. 1A and movies S1 and S2). To quantify motility, we measured the velocity of length changes of individual processes. On average, extensions and retractions had similar velocities of 1.47 ± 0.10 μm/min and 1.47 ± 0.08 μm/min, respectively (Fig. 1, B and C) (range from 0.4 to 3.8 μm/min; 95 extensions and 147 retractions in 14 cells; n = 8 animals; typically, thick branches were on the lower end of this range). Branch additions and losses occurred at every branch order and balanced each other (Fig. 1D).

Fig. 1.

Microglial cells are highly dynamic in the resting state in vivo. (A) Maximum-intensity projections of an individual microglial cell (45 to 75 μm below the pia) at the beginning (left) and 1 hour after (center) the start of a transcranial time-lapse recording. (Right) Overlay showing extensive formation (green) and deletion (red) of microglial processes. (B) Extensions (green) and retractions (red) of processes over the time course of 20 min. (C) Length changes of the processes shown in (B) as a function of time. (Right) Mean motility values in μm/min for extensions and retractions. (D) Branch motility occurred at every branch order. (E) Example images of microglial protrusions (arrowheads) from a time-lapse recording. (F) Length changes over time of the two protrusions P1 and P2 indicated in (E). Vertical dashed lines mark the acquisition times of the images shown in (E). Arrows indicate protrusion lifetime (9). (G) Lifetime histogram of protrusions.

Microglia processes also displayed highly motile filopodia-like protrusions of variable shape, typically forming bulbous endings (Fig. 1E and movie S3). Such protrusions transiently and sometimes repeatedly appeared at various locations along the main processes and at their terminal endings. Often protrusive activity stalled for several minutes before further extension (or retraction) occurred (Fig. 1F). Time-lapse imaging at high temporal and spatial resolution revealed a high turnover of protrusions with velocities of up to 4.1 μm/min (extensions and retractions had similar rates: total average of 2.2 ± 0.2 μm/min, range from 0.6 to 4.1 μm/min, 22 extensions and 23 retractions on two cells in two animals). The average lifetime of such protrusions was 3.9 ± 0.2 min (Fig. 1, F and G) (72 protrusions in three cells; n = 3 animals; range from 1.7 to 8.3 min). Despite the constantly changing decoration of microglial processes with protrusions, the number of sites per cell showing protrusive activity remained rather constant over time (fig. S2, A and B) (mean of 19.3 ± 5.3 for n = 8 cells), as did the average total length of microglial processes (fig. S2B).

Microglial processes and protrusions sampled the extracellular space in a seemingly random fashion and at a high turnover rate. To quantify the volume fraction surveyed by microglia per time, we analyzed cumulative maximum-intensity projections through time-lapse recordings (9) (Fig. 2A), yielding a progressive filling rate of 14.4 ± 1.6% per hour (n = 8 animals) (Fig. 2B). Considering that the volume fraction of extracellular space is estimated to be about 20% (10), this suggests that the brain parenchyma is completely screened by resting microglia once every few hours. In doing so, microglial cells vary their territories. Border zones between neighboring microglial cells were mutable, and changes in favor of adjacent cells often occurred after retraction of thick processes in another cell. When processes of neighboring microglial cells encountered one other, endings mutually repelled each other.

Fig. 2.

Resting microglia continuously sample their microenvironment and dynamically interact with other cortical elements. (A) Example images illustrating cumulative volume sampling. (Left) Initial image at time t0. (Right) Cumulative projection at a later time point, t1 (9). (B) Quantitation of volume sampling. Dashed lines, percentage increases for individual cells; solid line, average trace. (C) Example images from a time-lapse recording showing spontaneous engulfment and subsequent evacuation of tissue components by microglial processes (yellow arrowheads). (D) Example images showing how microglial processes and protrusions contact neighboring astrocytes (left), neuronal cell bodies (center; unstained dark areas), and the astrocytic sheath around a microvessel (right). Images are overlays of the green microglia and red SR101 stain.

This high resting motility may serve a housekeeping function, enabling microglial cells to effectively control the microenvironment and to clear the parenchyma of accumulated (low diffusible) metabolic products and deteriorated tissue components. Indeed, branch protuberances of microglial cells were short-lived and typically showed bulbous endings, indicating that tissue material had been collected. In a few cases, we observed spontaneous engulfments of tissue components, which subsequently were transported toward the soma (Fig. 2C and movie S4). To further reveal the interaction between microglia and other cortical elements, we counterstained astroglia with the red fluorescent dye sulforhodamine 101 (SR101) (11). Control imaging experiments before, during, and after SR101 application showed no adverse effects of SR101 itself on microglia motility (n = 4 animals). Unlike microglial cells, astrocytes showed no comparable restructuring of their processes. The SR101 counterstain also enabled us to visualize neuronal cell bodies and cortical blood vessels, which appear as unstained dark areas (Fig. 2D). Microglia processes and protrusions directly contacted astrocytes, neuronal cell bodies, and blood vessels, suggesting that in the healthy brain microglia dynamically interact with other cortical elements (Fig. 2D and movie S5). Because microglia are thought to monitor neuronal well-being through molecular changes in their microenvironment (12), we tested whether a change in the level of neuronal activity might affect microglia behavior. Surface application of the ionotropic γ-aminobutyric acid (GABA) receptor blocker bicuculline (BCC, 50 μM) was found to significantly increase microglia volume sampling, whereas application of the sodium channel blocker tetrodotoxin (TTX, 25 to 50 μM) had no significant effect [Supporting Online Material (SOM) Text, fig. S3, and movie S6].

Another likely function of the high resting microglia motility is to facilitate prompt reactions to brain injury (5). We therefore characterized microglia activation immediately after targeted disruption of the blood-brain barrier (BBB) at the level of individual capillaries (Fig. 3). Vessel outlines were visualized with the use of SR101 application. After a baseline imaging period, individual capillaries of about 6 μm diameter were damaged by using highly localized laser lesions either through the thinned skull or through a small cranial window (Fig. 3A and movie S7). Disruption of the BBB was indicated by local tissue expansion and detachment of astroglial end feet. Laser lesions caused extravasation of dye in three experiments in which blood plasma was stained via tail-vein injection of a dextran-conjugated fluorescent dye (movie S8). Laser lesions induced an immediate microglia response, indicated by a switch from undirected to targeted movement of nearby microglial processes toward the injured site (Fig. 3B and movies S7 and S9). The average velocity of extensions radially impinging on the injured site was similar to extension rates during the resting state (mean of 1.8 ± 0.3 μm/min for n = 5 animals). Processes on the far side of activated microglial cells subsequently started to retract.

Fig. 3.

Microglia are rapidly activated after local BBB disruption. (A) Fluorescence images of EGFP-expressing microglial cells (Ch1) and SR101 counterstained astrocytes (Ch2) before, 30 min, and 60 min after a targeted laser-induced microlesion. The disrupted blood vessel is apparent in Ch2 (yellow flashes indicate the site of injury). (B) Rapid shielding of a lesioned blood vessel section. (Top) Overlay of green microglia and red-stained astrocytes before and 10 min after the laser-induced lesion. (Bottom) Microglia morphology at intermediate time points showing rapid, targeted movement of microglial processes toward the injured blood vessel (outlined in red; yellow flash indicates the site of injury). (C) Activated microglia processes at the site of laser lesion about 4 hours after injury. Several spherical engulfments are visible in the vicinity of the lesioned blood vessel arborization. Arrowhead points to an engulfment that collapsed within a few minutes. (D) Histogram of the diameter distribution of 33 postlesion engulfments. (E) (Left) Example time courses of spherical shaped engulfments. Diameters are normalized to initial values. (Right) Schematic illustrating the collapse of an engulfment.

The number of responding microglial cells depended on the severity of the injury. In general, only microglial cells in the immediate vicinity of the microlesion were activated, whereas cells farther away (>90 μm) did not or did not immediately respond. In two cases, laser lesions caused a transient activation of only a single microglial cell. In those cases, no measurable tissue expansion was observed, indicating only mild damage to the BBB. Yet in all lesion experiments, shielding of the injured area through accumulation of microglial extensions was observed (Fig. 3B and movie S10). In cases of severe BBB disruption, multiple spherical-shaped inclusions started to form around 10 to 15 min after the lesion, indicating phagocytic activity by microglial processes. Inclusions were found within 15- to 25-μm radial distances of the injured site, showing diverse dimensions with an average diameter of 4.6 ± 0.3 μm (Fig. 3, C and D and movie S11) (range from 4.6 to 11.1 μm; n = 34 inclusions in two animals). Inclusions were stable for several minutes (mean of 11.6 ± 1.9 min and range from 1.8 to 23.9 min) before they rapidly collapsed (mean of 2.0 ± 0.5 min; n = 14 inclusions in two animals) to around 40% of their initial size (Fig. 3E). Notably, the larger the inclusions, the shorter their lifetimes. Within the observation period (up to 5.5 hours), somata of microglial cells became more rounded. They did not, however, migrate toward the injured site. Interestingly, SR101-colabeled astrocytes showed no morphological response to the laser-induced microstroke. A switch from undirected surveillance behavior to targeted movement of microglial processes was also observed in response to local lipopolysaccharide application (SOM Text).

Our results demonstrate that microglial cells are highly dynamic structures during the “resting” state in vivo and not only after activation. The extent of ongoing structural changes far exceeds what has been described for both neurons (13, 14) and astrocytes (15) on a similar time scale (SOM Text). The pronounced and ongoing structural changes of resting microglial cells presumably serve an immune surveillance function (SOM Text). In particular, microglia can sense subtle changes in their microenvironment through a variety of surface receptors, such as purino- and fractalkine receptors (12), receptors for complement fragments, immunoglobulins, adhesion molecules, and inflammatory stimuli (16). In addition, microglia can respond to these changes, for example, through expression of neurotrophic factors or release of pro- and anti-inflammatory cytokines upon activation (12). Our experiments suggest that microglia perform this surveillance function by continuously sampling their environment with highly motile protrusions. These protrusions may also be involved in collecting tissue debris. Microglia motility most likely has its basis in actin, a cytoskeletal protein shown to be critically involved in growth and motility in many cells. Indeed, microglia contain high amounts of filamentous actin (17), and inhibitors of actin polymerization have been shown to affect the motility and migration of activated microglial cells (5).

Activated microglia are thought to exert neuroprotective as well as neurotoxic functions on neurons. Overall this effect may depend on both pathologic conditions and injury severity (12, 16). In our microlesion experiments, the shielding of injured sites indicated a neuroprotective role for microglia. Furthermore, the early formation of spherical-shaped inclusions suggests immediate phagocytic engulfment and removal of damaged tissue or leaked blood components. Together, this is consistent with the idea that microglia constitute the first line of defense against invading pathogens (12, 18). In conjunction with animal models of brain disease, our in vivo imaging approach presents the opportunity to study the role of microglia in various pathologies in the intact brain.

Supporting Online Material

www.sciencemag.org/cgi/content/full/1110647/DC1

Materials and Methods

SOM Text

Figs. S1 to S3

References and Notes Movies

S1 to S12

References and Notes

View Abstract

Navigate This Article