Arachidonic acid closes innexin/pannexin channels and thereby inhibits microglia cell movement to a nerve injury

Abstract

Introduction

Arachidonic acid (ArA) is enzymatically freed from membrane-bound phospholipids by phospholipase A2. This action occurs in response to a variety of physiological and pathophysiological conditions, such as increases in intracellular calcium (Sun et al., 2005). ArA and its products, via the cyclooxygenase (COX) and lipoxygenase (LOX) signaling pathways, have a well documented presence at sites of neuronal and brain injury (Sun et al., 2004; Gabryel et al., 2007; Lopez-Vales et al., 2011). They are thought to be involved in diverse processes including vasodilatation, recruitment of immune cells, release of cytokines, and platelet aggregation (Minghetti and Levi, 1998; Minghetti, 2004; Phillis et al., 2006). Invertebrates are also able to produce ArA (Watts and Browse, 2002) and process it by COX and LOX pathways (De Petrocellis and Di, V, 1994; Jarving et al., 2004), and the products may control membrane processes, including in leeches (Scuri et al., 2005). In this study, we investigated a novel role for ArA as an inhibitor of pannexon/innexon channels. We also examined consequences of this inhibition for responses of microglial cells in the leech. Leeches are favorable for in situ study of control of microglial cell migration to nerve lesions (von Bernhardi and Muller, 1995; Duan et al., 2005).

ArA has long been recognized as a potent inhibitor of gap junctions in a variety of tissues (Giaume et al., 1989; Fluri et al., 1990; Miyachi et al., 1994; Boger et al., 1999; Martinez and Saez, 1999). However, its effect on gap junction “hemichannels” (see Sosinsky et al., 2011, for nomenclature) has not been assessed. Pannexin 1 (Panx1) is a member of a recently discovered mammalian gap junction family homologous to the invertebrate gap junction family, the innexins. Panx1 forms exclusively non-junctional membrane channels, which have the related name pannexons, on cells including neurons and glia. Recently, Panx1 has been implicated in the activation of the inflammasome (Kanneganti et al., 2007; Pelegrin et al., 2008; Silverman et al., 2009).

The overlapping pharmacology of pannexons and connexons, the latter forming the mammalian gap junction channels, has made it difficult to distinguish the two functionally. Most invertebrates, including the leech, do not have connexins but only pannexin-like proteins, the innexins, making them useful for the study of these channels. We previously found that Hminx2, a leech pannexin homologue present in glia (Dykes and Macagno, 2006) and important for selective formation of gap junctions (Firme, III et al., 2012; Yazdani et al., 2013), forms ATP permeable membrane channels (innexons) that when expressed in oocytes are activated by intracellular calcium and inhibited by carbenoxolone (CBX, ~50% inhibition at 1 μM) (Bao et al., 2007). In the same study, loss of 6-carboxyfluorescein and Lucifer Yellow dyes from glial cells decreased under conditions that closed innexon channels, including acidification or CBX (10 μM). Moreover, RNA knockdown of Hminx2 showed that that innexin was specifically involved (Samuels et al., 2010).

Glial cells are intermingled in the nervous system with an unrelated class of cells, the microglia, which are immune cells of the central nervous system (CNS) (Aguzzi et al., 2013). Extracellular ATP is a well recognized activator of microglia (Honda et al., 2001; Inoue, 2002; Davalos et al., 2005; Samuels et al., 2010). There is evidence that glia are the cellular source of this extracellular ATP released at a distance after injury (Salter and Hicks, 1995; Guthrie et al., 1999; Verderio and Matteoli, 2001; Schipke et al., 2002; Bianco et al., 2005; Scemes and Giaume, 2006). Moreover, in vertebrates the molecular mechanism of ATP release from the glial cells is apparently through pannexon channels in vertebrates (Dahl and Locovei, 2006; Iglesias et al., 2009).

Leech microglia resemble mammalian microglia in their cytology, physiology, histochemical staining, and phagocytic ability (Coggeshall and Fawcett, 1964; Kuffler and Nicholls, 1966; Morgese et al., 1983; von Bernhardi and Muller, 1995). Because of this and their ready identification in the living nerve cord, they have been useful for understanding migration of microglia to nerve lesions (Morgese et al., 1983; McGlade-McCulloh et al., 1989; Duan et al., 2005; Ngu et al., 2007). Experiments closing or specifically knocking down innexons formed by Hminx2 have shown that ATP release from glia via the innexon channels is required for the migration of microglia to nerve injuries (Samuels et al., 2010). But no endogenous regulator of innexon or pannexon channels, other than intracellular calcium, low pH and extracellular ATP itself (Qiu and Dahl, 2009), has been identified. In particular, it is of interest to know what might inhibit the channels and thereby reduce the concentration of extracellular ATP, thus slowing or stopping microglia when they reach lesions.

The present study aimed to determine whether ArA affects microglial movement and whether pannexons/innexons are involved in this process. Moreover, it was important to know whether any effects were produced directly by ArA, as determined by whether its non-metabolizable analogue eicosatetraynoic acid (ETYA) could also produce them. First, we tested whether ArA reduces currents of pannexons and glial innexons expressed in oocytes. Since the migration of microglia to injury sites in the leech CNS is dependent on functional innexon channels releasing ATP, a second aim was to determine if ArA and ETYA reduce or block microglia migration to lesions in the leech.

Methods

LEECH NERVE CORD AND MICROGLIA

Preparation

Adult leeches (Hirudo sp., Leeches USA, Westbury, NY), 3-4g, were maintained in artificial pond water (Forty Fathoms, 0.5 g/L H₂O; Marine Enterprises, Towson, MD) at 15°C. Nerve cords were dissected from a total of 30 leeches anaesthetized on ice. Cords were pinned in a physiological saline solution (Ringer's (Kuffler and Potter, 1964)) in 35 mm Petri dishes with 1 mL Sylgard 184 (Dow Corning, Midland, MI) silicone rubber on the bottom. Nerve cords were maintained up to several days in Leibowitz-15 culture medium supplemented with 2% fetal bovine serum, 1% glutamine, 0.06% glucose, and 5 μg/mL gentamicin (L-15).

Nerve cords were prepared as previously described (Samuels et al., 2010). Connectives were crushed for 8-12 seconds with fine forceps (Dumont #5, ground to have parallel edges) and incubated in leech Ringer's solution at room temperature for 4 hours with or without drugs; controls were not crushed. The tissue was then fixed overnight at 4°C in PBS with 4% paraformaldehyde (pH 7.2), washed in PBS, and mounted on a microscope slide in 85% glycerol with Hoechst 33258 nuclear stain (10 μg/mL). Mounted preparations were photographed through a 20X objective, NA 0.7, with a Spot-RT camera (Diagnostic Instruments, Sterling Heights, MI) mounted on a Leica DM RZA2 fluorescence microscope. To identify the location of the crush without bias, an image of the connective's autofluorescence was made using a fluorescein filter. Cell nuclei were imaged using a UV filter set. To measure microglia accumulation in each preparation, a series of 10 optical sections was photographed at 2 μm intervals through a combined depth of 18 μm. Each stack of images was combined and collapsed using the Metamorph® program (Molecular Devices, Downingtown, PA), and the microglia were counted in a 100 μm² region at the edge of each crush. The counts were statistically analyzed with an ANOVA and post-hoc Fisher Test (Statistica®, StatSoft Inc., Tulsa, OK). Although a few nuclei in each stack of images were of sheath cells, including in the “distal” counts in Figure 3, only microglia moved (McGlade-McCulloh et al., 1989) and only microglia accumulated at the lesion.

Open in a separate window

Figure 3

The inhibition of microglia migration and accumulation by ArA and ETYA was reversed by exogenous ATP. Microglia accumulation was measured in connectives fixed 4 hours after crushing and bathing in indicated solutions using standard methods in which tissues were stained with Hoechst 33258 dye, photographed and cell nuclei counted (see Figure 2A in reference Samuels et al., 2010). Bars represent mean numbers of cells at the lesions in a 100×100×18 μm³ volume. “Distal” was a measure of microglia in the same volume in an uncrushed region, approximately 1 mm from any injury; it was used as a baseline for the number of cells distributed throughout the tissue as a background level. Control accumulations were in 0.1% ethanol. The difference between the “ethanol” and “distal” values was the accumulation of microglia, the only cells that moved. (a) ArA reduced the accumulation of microglia at lesions in a dose dependent manner (N=3, p<0.01). (b) The ArA non-hydrolysable analogue eicosatetraynoic acid, ETYA, reduced the accumulation of microglia at lesions in a dose dependent manner (N=3, p<0.01). (c) Microglia accumulation at sites of injury, inhibited by ArA, occurred in the combined presence of ArA (100 μM) and ATP (100 μM) (N=3, p<0.01). (d) ATP reversed the ArA inhibition of microglia movement toward a crush. Time lapse images of crushed connectives with Hoechst-stained nuclei, treated with either ArA alone (100 μM) or ArA and ATP together (100 μM for each), were collected and made into a movie. About 3 times more microglia moved when extracellular ATP was added, but the directional movement was approximately the same (N=3, p<0.05). Therefore, decreased movement rather than misdirection caused decreased accumulation at lesions.

Low light video microscopy

Nerve cords dissected as described above, incubated for 30 minutes in Hoechst 33258 nuclear dye (0.001%), and rinsed 3 times for 5 minutes were kept overnight in L-15 medium at 15°C. The following day the tissue was crushed, or not crushed for controls. It was then treated as for microglia accumulation studies described above, and imaged on a Zeiss WL epifluorescence microscope (40X water immersion objective, NA 0.75) equipped with a 12V, 100 W tungsten-halogen lamp operated at 8-10V and a Zeiss 02 (UV) filter set. Metamorph® imaging software controlled a shutter and CCD camera (Hamamatsu XC-77, Hamamatsu City, Japan) to acquire time-lapse images (averages of 256 frames at 35 frames/sec, 512×512 pixels in size, collected at 2.5 minute intervals for up to 4 hours). In the region adjacent to the crush (approximately 100-500 μm away), the number of microglia nuclei moving more than 50 μm in two hours was counted. Microglia movement during the time-lapse recording was measured double-blind by assistants who tracked individual microglia nuclei, noted their direction of movement, and counted the number moving more than 30 or 50 μm, depending on the experiment. The counts were statistically analyzed with an ANOVA and post-hoc Fischer test (Statistica, StatSoft Inc., Tulsa, OK).

Dye release from living nerve cord

A 1 mm piece of leech nerve cord was dissected in physiological saline, pinned to a Sylgard-lined dish, and incubated overnight in L-15 medium (Samuels et al., 2010). The following day the nerve cord was washed in Ringer's solution and the connective glial cell was ionotophoretically injected through a microelectrode filled with 6-carboxyfluorescein (6-CF, MW 376.32, 5 μM, -2 nA for 15 minutes) that impaled the cell through a cut end. The dye was allowed to diffuse throughout the cell for 20 minutes required for equilibration. Images were then collected every 5 minutes on a Leica fluorescence microscope with a 10X objective (0.3 NA) equipped with a Spot-RT camera (Diagnostic Instruments, Sterling Heights, MI) operating in a linear range. Drugs were bath applied at their final concentrations. Washes consisted of briefly removing the tissue from the microscope and rinsing the tissue three times with leech Ringer's before returning it to the microscope in the same location, determined optically from micrographs.

Fluorescence intensities of the paired connectives were measured with Metamorph® (Molecular Devices, Sunnyvale, CA) using a region analysis tool. Pixel intensities were averaged to obtain an average intensity. The percent dye lost during treatment was calculated by dividing the intensity of each sequential image by the previous image's intensity. A repeated measures ANOVA followed by a post-hoc Fischer test was used for comparison and statistical analysis of the dye loss.

Results

ArA EFFECTS ON CURRENTS THROUGH INNEXIN AND PANNEXIN CHANNELS

Mouse Panx1 or leech glial-associated innexin (Hminx2) mRNA was injected into Xenopus oocytes and 2 to 4 days later, oocytes were penetrated with a pair of microelectrodes electrodes for voltage clamp experiments. The cells were held at depolarized or positive potentials to open the pannexons and innexons on the membrane surface, as indicated by the increased currents during 5 sec pulses at 0.1 Hz. Such currents are not present in uninjected cells or before there has been adequate time for expression.

The conductance associated with Hminx2 was quickly and reversibly inhibited by 100 μM ETYA and by 100 μM ArA (Figure 1a, b). Similar inhibition occurred for innexon channels that had been opened by treating the cells with oocyte saline in which the NaCl was replaced with half the molar amount of potassium gluconate (50% KGlu) while the membrane potential was clamped at the resting potential (Figure 1b,c). Figure 1d shows an alternative pulse protocol (from –30 mV to +60 mV), which was also known to open innexons (Bao et al., 2007). Again, ArA blocked the innexons. The current measured during the intracellular acidification of the cell in panel on the right was a positive control for the closing of the innexon channels.

Open in a separate window

Figure 1

Hminx2 and mouse Panx1 channels expressed by oocytes were closed by ArA and ETYA. (a) Hminx2 channels were closed by ArA (100 μM) and ETYA (100 μM). The membrane potential of an Hminx2 expressing oocyte was shifted to positive potentials, as indicated in the top trace which also shows +15 mV voltage pulses applied at a rate of 0.1 Hz. The increased membrane currents during pulses, reflecting the opening of the innexons, were attenuated by applying ETYA or ArA in the bath. As a control, 0.1% ethanol, the vector in which both ArA and ETYA were dissolved, was applied before each drug as indicated by a horizontal line. Same scales for (a), (b) and (c). (b) Hminx2 channels, opened with KGlu, closed in response to ArA. Oocytes expressing Hminx2 were voltage clamped at –30 mV and treated with a modified OR2 solution in which 50% of NaCl was replaced with potassium gluconate (50% KGlu) to open the innexons. The increased conductance of the oocyte with 50% KGlu, shown by increased current during 15 mV pulses, was inhibited by treatment with ArA (50 μM). As a control, 0.1% ethanol was applied before ArA. (c) Hminx2 channels, opened with KGlu, closed in response to ETYA and CBX. Oocytes expressing Hminx2 were voltage clamped at –30 mV and, to open innexons, treated with 50% KGlu. The increased conductance of the oocyte in response to the 50% KGlu was blocked with a treatment of ETYA (50 μM), the non-metabolizable ArA analogue. As a control, 0.1% ethanol was applied before ETYA. 100 μM CBX also reduced the 50% KGlu-induced increase in conductance. (d) Oocytes injected with Hminx2 mRNA were voltage clamped and the membrane potential was shifted between -30 and +60 mV, a technique known to open pannexon and innexon channels. The large conductance of this oocyte was nearly blocked by ArA. The current was also measured during the intracellular acidification of this cell with CO₂, in the panel on the right, as a positive control for the closing of the innexon channels. (e) ArA inhibited mouse pannexin channels expressed in oocytes. Mouse Panx1 expressing oocytes were voltage clamped and the membrane potential shifted between -50 and +50 mV at 0.1 Hz to open the pannexon channels. ArA (100 μM) caused a decrease in conductance (relative to CBX) of approximately 10%, and its non-metabolizable analogue ETYA (100 μM) caused a decrease in the cellular conductance of approximately 50% (n=4). As controls, the vector 0.1% ethanol was applied before each drug.

To determine ArA effects on the vertebrate pannexons, ArA and ETYA were applied to oocytes expressing mouse Panx1. As shown in the example in Figure 1e, the pannexon currents were significantly and reversibly decreased in either 100 μM ArA (by approximately 15-20%) or 100 μM ETYA (by approximately 45-50%). Thus, ArA itself causes innexon and pannexon channels to close.

To determine if the reduction in pannexon current by ArA corresponded to a significant reduction in ATP release, ATP release was measured from individual oocytes expressing Panx1. As background, channel blockers may produce a small reduction of current through pannexons expressed in Xenopus oocytes when those same drugs cause a large reduction in permeability to dye or ATP (Qiu and Dahl, 2009). Generally, agents or conditions known to close innexon channels have also been found to block ATP passage through pannexons expressed in Xenopus oocytes. Therefore, the effects of ArA and ETYA on ATP release from Xenopus oocytes injected 4 days earlier with mouse Panx1 mRNA and known to express functional channels were measured. As shown in Figure 2, potassium gluconate (KGlu) solution (see Methods) used to open the pannexons caused release of ATP, as measured by increased luminance in the luciferin-luciferase assay. The ATP release was stopped by 100 μM ArA and by 100 μM ETYA (p<0.001 below the KGlu control) and reduced by 10 μM ArA (p<0.02). Thus, both ArA and ETYA reduced release of ATP through pannexons in a dose-dependent manner.

Open in a separate window

Figure 2

ArA and ETYA inhibited the release of ATP from oocytes expressing mouse pannexons opened by treatment with potassium gluconate solution (KGlu). For each bar, 4 oocytes were injected with Panx1 mRNA, treated as indicated, and the luciferinluciferase assay of the supernatant, averaged for each condition, gave ATP release as luminescence on a linear, arbitrary scale. Saline was a control, with significance of decreases of ATP release compared to KGlu alone indicated by * for p<0.02 and ** for p<0.001.

ArA EFFECT ON 6-CF DYE LOSS FROM THE CONNECTIVE GLIAL CELL

ArA and ETYA resembled CBX in their inhibition of microglia migration. CBX, like CO₂, closes glial innexons and prevents the release of ATP and of dyes 6-carboxyfluorescein (6-CF) and Lucifer Yellow from the connective glial cell (Bao et al., 2007; Samuels et al., 2010). It was, therefore, of interest to know if the physiological compound ArA and its non-metabolizable analogue ETYA could also inhibit dye release from glial cells. ArA significantly reduced 6-CF loss from the glial cell as compared to its vector (ethanol) control, although there appeared to be a delay in the effect (Figure 4a). Therefore, the normalized rate of loss was higher in the first 30 minutes than in the second 30 minutes. A similar assay with the non-metabolizable form of ArA, ETYA, demonstrated a more rapid reduction in dye loss that persisted for about half an hour after the ETYA was removed and then also recovered (Figure 4b). The basis for the delay in onset with ArA compared to ETYA is unknown, but unlike ETYA, ArA was likely catabolized by tissue, which could delay its reaching peak levels along the involuted membrane of the glial cell (Coggeshall and Fawcett, 1964; Elliott and Muller, 1981). These studies demonstrated that ArA itself directly blocked dye loss from the connective glial cell in a manner similar to CBX, the pannexon/innexon channel blocker.

Open in a separate window

Figure 4

ArA and ETYA reduced 6-CF dye loss from a dye-injected glial cell. (a) The graph shows the rate of normalized dye loss per 5 minutes from the glial cell, averaged during the first 30 minutes and the second 30 minutes of ArA treatment. As indicated by the asterisk, the inhibited rate of dye loss in the ArA treated tissue was significantly different from the ethanol-treated control only in the second 30 minutes (n=4, p<0.05). (b) ETYA, a non-metabolizable analogue of ArA, significantly inhibited 6-CF dye loss from the connective glial cell throughout the period of measurement. Graph shows normalized rate of dye loss, measured as in (a), from a 1 mm piece of connective glial cell every 5 minutes. In saline alone and in saline with 0.1% ethanol vector there was a substantial loss of dye (~12%), which was significantly reduced during the first half hour after ETYA was added (“ETYA”, n=4, p<0.05). This effect had reached a peak by the first half hour of wash (“ethanol wash, 0-30”), with a significant reduction compared to controls throughout the ethanol wash at p<0.001 (double asterisk shown only for the first time point, n = 4). During the wash, the rates of dye loss at the first and last time points were significantly different from each other at p<0.05.

Discussion

Now in addition to ATP itself, calcium and voltage, ArA can be considered to be an endogenous regulator of glial innexon and pannexon channels, and it may play an important role in modulating the microglial response to nerve injury. It is not known whether ArA inhibits the channels only extracellularly, as is evidently true for voltage-dependent sodium channels (Fang et al., 2011), whether it acts intracellularly, as for arachidonate-regulated Ca²⁺-selective (ARC) channels (Shuttleworth, 2009), or whether it acts within the membrane as for K⁺–channels (Börjesson and Elinder, 2011). If it passes out of cells through channels in the membrane, such as through unpaired innexon or pannexon channels, it may also regulate its own release. There is evidence for such passage of ArA through channels (Jiang et al., 2007) that are exclusively pannexons, in light of what is now known about glibenclamide's block of pannexons (Qiu et al., 2011).

Other agents that reduce pannexon conductance by only 15 to 20% also can block uptake or release of ATP and dye. This is because pannexons and innexons have multiple conductance states (Bao et al., 2004), with different dye permeabilities for different states. Thus, even a small reduction in total conductance in shifting from one state to another may completely block the ability of dyes and ATP to pass through the channel. This phenomenon was observed for the effects of different drugs on current and flux of ATP or dyes through Panx1 channels, including BBG (Qiu and Dahl, 2009; Wang and Dahl, 2010), bongkrekic acid and artemisinin [Qiu and Dahl, unpublished]. Taken together, these experiments provide evidence for a novel function of ArA as an inhibitor of the innexon and pannexon family of channels, and as an inhibitor of the microglial response to injury in the leech CNS. The reversibility of the microglial response when ATP is added to ArA is consistent with earlier results with the blocker CBX (Duan et al., 2009; Samuels et al., 2010). This indicates that ArA inhibits microglial migration by blocking glial innexons rather than by directly affecting microglia. It is not known if it affects other innexons that might be important for other functions. The similar effectiveness of ETYA and ArA in all assays is evidence that ArA itself and not its downstream signaling products, such as produced through the COX and LOX pathways, is responsible for the actions of ArA. Our data do not exclude the possibility that ArA metabolites also inhibit the pannexons and innexons, but they indicate that ArA may act as an endogenous regulator of pannexon and innexon channels, reducing their open probability or changing the single channel conductance.

The expected rise in ArA after nerve injury may have additional physiological effects that were not examined in the present study. For example, there is evidence for a direct effect of ArA on stretch-activated potassium channels (Maingret et al., 1999; Monaghan et al., 2011). And ArA may inhibit on astrocytes a large conductance channel, termed a maxi-anion channel, that is permeable to ATP in response to oxygen-glucose deprivation, but is not blocked by classic blockers of pannexons such as carbenoxolone (Liu et al., 2008).

Mass spectroscopy has indicated that levels of ArA and its metabolites increase in the CNS after injury (Westcott et al., 1987; Miller et al., 2009). In the leech CNS, ArA may act as an endogenous inhibitor and regulator of the innexon channels with injury. High local levels of ArA may occur close to the site of the crush, closing innexons, reducing the release of ATP, and slowing the movement of microglia approaching the injury so that they accumulate. The immediate rise in glial [Ca²⁺]_i and production of nitric oxide at the injury (Kumar et al., 2001; Samuels et al., 2010) may cause a rise in cytosolic phospholipase A2 (PLA2) activity (Xu et al., 2008).

In mammals, there are many downstream products of ArA that are involved in inflammation, and many of the COX and LOX enzymes are induced upon injury and activation of PLA2 (Gronert, 2008). Pannexon channel activation is known to be part of a cascade of inflammatory signals in neurons and glia, including activation of the inflammasome, a multiprotein complex involved in innate immunity (Silverman et al., 2009; MacVicar and Thompson, 2010). The early activation of PLA2 by calcium synthesizes high levels of ArA that will be present immediately upon injury. These high levels of ArA would initially inhibit the activity of the inflammasome while the inducible COX2 would increase its expression. When large amounts of COX2 are made as a result of the injury, the balance would shift in favor of the ArA metabolites, as ArA would be metabolized as soon as it is made. This would release the inhibition on the innexons or pannexons, which could allow for the simultaneous actions of the prostaglandins and inflammasomes. The initially high levels of ArA that inhibit the pannexon channels may act as an endogenous checkpoint for determining whether the cells will have a full inflammatory response. ArA's inducible catabolic enzymes may overcome this ArA inhibition, which could be the trigger for the cells to have a strong response. However, in milder injuries that produce less COX2, ArA would inhibit the pannexon channels, and the cells would only have a mild inflammatory response to injury. The ArA effect on pannexons may therefore be a type of molecular switch that determines the level and quality of the inflammatory response by damaged cells.

In the leech, from the present results and those that have been previously published, the following model emerges for control of microglia migration to lesions. Within seconds of nerve injury a calcium wave culminating in a maintained calcium gradient is generated in the ensheathing glial cell (Samuels et al., 2010), contributing to the opening of innexons and release of ATP from the glia (Bao et al., 2007), which activates microglia up to a millimeter from the lesion (Duan et al., 2005). Activated microglia migrate toward the lesion, directed by NO (Duan et al., 2009) generated by the gradient of intracellular calcium that activates a constitutive nitric oxide synthase (Shafer et al., 1998; Chen et al., 2000). The higher levels of NO at the lesion itself may slow or stop the microglia (Chen et al., 2000; Kumar et al., 2001). ArA generated as a consequence of the injury would inhibit the release of ATP, helping to account for the progressive decline in numbers of migrating microglia and their accumulation at the lesion (McGlade-McCulloh et al., 1989).

It remains to be determined whether ArA controls mammalian microglia as it can in leeches, but the similar responses and roles reported for microglia across species suggest that ArA may be an intrinsic signal that slows and stops microglia by reducing release of ATP locally from glial cells.

Acknowledgements

Abbreviations used in this paper

References