Neuroglial ATP release through innexin channels controls microglial cell movement to a nerve injury

Abstract

Microglia, the immune cells of the central nervous system, are attracted to sites of injury. The injury releases adenosine triphosphate (ATP) into the extracellular space, activating the microglia, but the full mechanism of release is not known. In glial cells, a family of physiologically regulated unpaired gap junction channels called innexons (invertebrates) or pannexons (vertebrates) located in the cell membrane is permeable to ATP. Innexons, but not pannexons, also pair to make gap junctions. Glial calcium waves, triggered by injury or mechanical stimulation, open pannexon/innexon channels and cause the release of ATP. It has been hypothesized that a glial calcium wave that triggers the release of ATP causes rapid microglial migration to distant lesions. In the present study in the leech, in which a single giant glial cell ensheathes each connective, hydrolysis of ATP with 10 U/ml apyrase or block of innexons with 10 µM carbenoxolone (CBX), which decreased injury-induced ATP release, reduced both movement of microglia and their accumulation at lesions. Directed movement and accumulation were restored in CBX by adding ATP, consistent with separate actions of ATP and nitric oxide, which is required for directed movement but does not activate glia. Injection of glia with innexin2 (Hminx2) RNAi inhibited release of carboxyfluorescein dye and microglial migration, whereas injection of innexin1 (Hminx1) RNAi did not when measured 2 days after injection, indicating that glial cells' ATP release through innexons was required for microglial migration after nerve injury. Focal stimulation either mechanically or with ATP generated a calcium wave in the glial cell; injury caused a large, persistent intracellular calcium response. Neither the calcium wave nor the persistent response required ATP or its release. Thus, in the leech, innexin membrane channels releasing ATP from glia are required for migration and accumulation of microglia after nerve injury.

Figure 1.

Measurements of extracellular ATP after nerve cord injury. (A) An ATP luminescence assay (luciferin/luciferase) was used to measure the increase in extracellular ATP after nerve cord crush. Connectives were dissected and kept overnight in supplemented L-15. No (uncrushed control), one, two, or three crushes were made to the connectives that were then incubated at 18°C for 30 min in 150 µl of leech Ringer’s solution. 100 µl of the supernatant was analyzed for its ATP content. Connectives with two and three crushes showed statistically significant increases in extracellular ATP compared with the uncrushed control (*, P < 0.05; n = 3). (B) Extracellular ATP was elevated for at least 2 h after injury. Samples (100 µl) of supernatant of two connectives each crushed twice (total volume 150 µl) were collected at 15, 30, and 120 min after making the crushes, and samples were analyzed using a luciferin/luciferase luminescence ATP assay. Significant elevation of extracellular ATP remained for at least 2 h after injury (*, P < 0.05; n = 3). (C and D) Release of ATP >15 min after injury and the effect of 10 µM CBX. Connectives were crushed twice, incubated for 15 min at room temperature, washed three times with Ringer’s solution, resubmerged in 150 µl of fresh Ringer’s solution, and incubated at room temperature for 30 min. 100-µl aliquots of supernatant were analyzed with luciferin/luciferase luminescence assay to measure levels of extracellular ATP. In C, ATP was elevated 15 min after the injury (P < 0.05; n = 3). In D, in another experiment, the continued release of ATP was attenuated by treatment with 10 µM CBX. Control and CBX conditions are displayed as percentages of values for uncrushed tissues to control for the direct effect of CBX on the luciferase assay, for CBX may artificially increase the absolute luminescence values slightly (P < 0.05; n = 3).

Figure 2.

Effects of CBX and ATP on microglia migration and accumulation. (A) Schematic diagram of the leech nerve cord and, for the region depicted in the rectangle, micrographs of crushed connectives to show how cell accumulations are measured. Tissue was stained with Hoechst 33258 dye to show nuclei. The top and bottom micrographs are representative collapsed images (10 each through 18 µm) with and without drug treatment. The crushes are indicated, with squares (100 × 100 µm) outlining regions of measurement. Nuclei are mostly microglia but include some sheath cells. “Treatment” tissue in the micrograph was incubated in 10 µM CBX. (B) Cell accumulation at the site of a crush injury in CBX. Tissues were treated with CBX at 1-, 10-, and 100-µM concentrations at the time of crush, and the microglia were allowed to accumulate for 4 h. Tissues were then fixed and stained. Bars represent mean number of cells at the site of injury in a 100 × 100 × 18–µm³ volume. “Distal” is a measure of the number of cells in an uncrushed region, ∼1 mm from any injury, and is used as a baseline for the number of cells distributed randomly throughout the tissue. The difference between the “control” or other conditions and the distal is the true accumulation. Mean ± SEM are represented (P < 0.001; n = 6). (C) Microglia moving in 30 µM CBX and 100 µM ATP. Microglia nuclei were tracked using time-lapse recording, and those moving >30 µm were counted. CBX did not inhibit movement in ATP, although it did inhibit movement induced by crushing the connectives. Bars represent average number of microglia moving in 30 min ± SEM (P < 0.01; n = 5). (D) Microglia accumulation in CBX and ATP. 10 µM CBX significantly reduced the accumulation of cells at the crush. In the CBX washout condition, the connectives were incubated in CBX for 4 h and washed for 1 h before crushing. “Distal” is a measure of cell nuclei in an uncrushed region. The reduction seen in CBX did not occur after CBX was washed from the tissue, showing that the effect of CBX was reversible. It was also eliminated in the simultaneous presence of 100 µM ATP, consistent with an effect of CBX to block release of ATP, but not the direction of movement (n = 3; *, P < 0.5; **, P < 0.01). (E) Microglial migration toward a crush and the effects of 10 µM CBX and 100 µM ATP. In the ∼400-µm-long region adjacent to the crush, the number of microglia nuclei moving >50 µm in 2 h was determined from time-lapse images. Drugs were added 5 min before crushing. The CBX with ATP condition was similar to the control condition in both total movement and directional movement, but CBX alone decreased the movement significantly (P < 0.05; n = 3).

Figure 3.

The effect of apyrase on microglia migration and accumulation. (A) Tissues were treated with apyrase at 5, 10, and 20 U/ml beginning at the time of crush, and after 4 h, the tissue was fixed, stained with Hoechst 33258 dye, and photographed. “Distal” was a measure of microglia in an uncrushed region, ∼1 mm from any injury, and was used as a baseline for the number of cells distributed throughout the tissue before injury. The difference between the “control” and the distal is the measure of microglia accumulation, as sheath cells do not move. Bars represent mean number of cells at the site of injury in a 100 × 100 × 18–µm³ volume (P < 0.05; n = 3). (B) Hydrolysis of ATP and microglial migration toward a crush. In the ∼400-µm-long region adjacent to the crush, the number of microglia nuclei moving >50 µm in 2 h was determined from time-lapse images. The connectives treated with apyrase to hydrolyze ATP had approximately one fourth the number of microglia moving relative to the control; the fraction of those moving that moved toward the crush (i.e., the directional movement) was approximately the same as the control (P < 0.05; n = 3). (C) Diffusion of extracellular ATP released after a crush caused microglia to move in an adjacent tissue. Two sets of connectives were dissected, stained with Hoechst 33258 dye, and incubated in L-15 culture medium overnight. The next day, one set was imaged with time-lapse microscopy (every 2.5 min), whereas the other set was crushed and placed parallel to it, 40–50 µm away. The microglia moved in the intact connectives, which would not have occurred in isolation. This movement was blocked in 20 U/ml apyrase, indicating that extracellular ATP released from the crush was required to cause microglial movement. Movement was measured as translocation of microglial cell nuclei by >30 µm in 1 h (P < 0.05; n = 3).

Figure 4.

Microglia in connectives treated with isotonic potassium gluconate (KGlu) and the effect of 10 µM CBX. (A) A representative tracing of currents (bottom traces) from an oocyte injected with Hminx2 and voltage clamped at negative potentials with shifts to positive potentials (top traces). The oocyte was treated with oocyte saline in which 50% NaCl was replaced with potassium gluconate (KGlu) to open the innexon channels and with 100 µM CBX to close them. A CBX-sensitive current increased severalfold in KGlu, reflecting increased membrane conductance due to innexon channel opening. (B) Cell nuclei stained with Hoechst 33258 dye were imaged using time-lapse microscopy, and microglia counted moving 25 µm for 30 min beginning at 30 min. The leech Ringer’s solution bathing the tissue was then replaced with 140 mM KGlu for 30 min, and the nuclei moving were counted again for 30 min. Although CBX had no effect on the basal movement of the microglia, it blocked the KGlu-induced movement (t test, n = 3 and P < 0.05).

Figure 5.

Hminx2 RNAi, but not Hminx1 RNAi, inhibited dye loss from the connective glial cell. (A) Photomicrographs of representative glial cells containing 6-CF dye. One pair of glial cells in the connectives was injected with Hminx2 RNAi into the left cell and Hminx1 RNAi into the right cell 2 d before the cells were injected with 6-CF dye, which was allowed to diffuse throughout the cells. The tissue was then slightly stretched to open the innexon channels. The left panel shows the cells at time 0, and the right panel shows them 54 min later. The glial cell injected with Hminx2 RNAi retained substantially more dye than the control glial cell injected with the Hminx1 RNAi. (B) Time course of dye loss from the experiment shown in A. The Hminx2 RNAi–injected cell lost dye at a slower rate than its Hminx1 RNAi–injected neighbor did. A movie of this experiment (Video 1) can be seen in the supplemental material.

Figure 6.

Hminx2 RNAi injection into connective glial cells blocked microglial migration to nerve lesions. Connectives were injected with Hminx2 RNAi or Hminx1 RNAi 2 d before the experiments, as in Fig. 5 and Materials and methods, except that both connectives were injected with the same type of RNAi. Microglia accumulations after injury were performed according to normal laboratory methods. The Hminx2 RNAi condition had significantly less accumulation than the Hminx1 RNAi condition (P < 0.05; n = 3).

Figure 7.

The post-injury calcium response. The fluorescence intensity of a glial cell injected with Calcium Green 1 was measured before and after the cell was injured by crushing the connectives. An interval of >1 min (at break in time axis) occurred before the jump in fluorescence at 0 s. The calcium response in the area adjacent to the crush and up to several hundred micrometers away developed a calcium gradient during the first few minutes after the injury. As indicated in the inset, the calcium responses in regions 100-µm² square were measured every 25 µm from near the crush to 150 µm distant. All regions measured had a large increase in fluorescence immediately after the crush was made; intensity in the regions nearer to the crush continued to rise, whereas the regions farther away began to decline, thereby creating a larger spatial calcium gradient.

Figure 8.

Mechanically induced calcium waves. (A) Measurements of calcium waves in connective glia. A probe coming from the left approaches a Calcium Green 1–injected glial cell. Blue squares are 25-µm² regions of measurement at 20-µm intervals beginning at the site of initiation. (B) An example of a mechanically induced calcium wave in a glial cell. Glial cells were filled with Calcium Green 1 and stimulated with a piezoelectric controlled tungsten probe (refer to Materials and methods). ΔF/F was determined for six 25-µm² regions located as shown with squares; the baseline F was fixed. The wave properties of the calcium response were evaluated by the increased delay of the calcium response with distance from the site of stimulation.

Figure 9.

Mechanically induced calcium waves and the effect of CBX. (A) Mechanically induced calcium waves were induced with a sharp tungsten probe controlled by a piezoelectric device that briefly poked the connective, as shown in Fig. 8 A and described in Materials and methods. A 40-V piezoelectric touch was given after 7 s of recording to initiate the calcium wave in the Calcium Green 1–filled connective glial cell. Changes in calcium in three 25-µm² regions located at 20 µm intervals were measured and included in the graphs. The calcium response was calculated as the change in fluorescence divided by the resting fluorescence (ΔF/F). The representative time courses of mechanically induced calcium responses in a connective glial cell with and without CBX were not distinguishable. Arrows indicate onset of the 200-msec stimulus. (B) The peak calcium responses during mechanically induced calcium waves in connective glial cells were not altered by CBX. Bars represent the average maximum fluorescence intensities in a 25-µm² region at the initiation site of the calcium wave induced by a mechanical stimulus and are statistically indistinguishable. Mechanically induced calcium waves were induced with a sharp tungsten probe controlled by a piezoelectric device. The stimulus was a 40-V pulse, which produced an ∼50-µm displacement of the probe (n = 4). (C) Propagation speed of the calcium wave did not change in 10 µM CBX. These measurements were made for the same calcium waves as used in B. Five 25-µm² regions located at 20-µm intervals from the stimulus (“poke”) site were used. At each region, the time to reach the half-maximum response after the stimulus, averaged for five responses, was determined. The distance of each region to the site of stimulation divided by the times to half peak from several adjacent regions was averaged to give the speed of propagation, as shown in the inset. There was no significant difference between the control and CBX conditions for the speeds (n = 4).

Figure 10.

Diagram of signals postulated to control microglial cell migration to lesions. (A) The nerve cord before injury. (B) The cord from injury (on left) to several hours later. Note the presence of the NO and Ca²⁺ gradients, and the opening of the innexon channels. It is during this phase that the microglia become round and move toward the crush in response to ATP. (C) Recovery from injury after microglia have begun to accumulate at the site of injury. MG, microglia; inx, innexon; P2YR, purinergic receptor.