Activation of the P2X7 receptor induces migration of glial cells by inducing cathepsin B degradation of tissue inhibitor of metalloproteinase 1


Address correspondence and reprint requests to Niamh Murphy, Trinity College Institute of Neuroscience, Trinity College Dublin, Dublin 2, Ireland. E-mail:


The P2X7 receptor is an ion-gated channel, which is activated by high extracellular concentrations of adenosine triphosphate (ATP). Activation of P2X7 receptors has been shown to induce neuroinflammatory changes associated with several neurological conditions. The matrix metalloproteinases (MMPs) are a family of endopeptidases that have several functions including degradation of the extracellular matrix, cell migration and modulation of bioactive molecules. The actions of MMPs are prevented by a family of protease inhibitors called tissue inhibitors of metalloproteinases (TIMPs). In this study, we show that ATP-treated glial cultures from neonatal C57BL/6 mice release and increase MMP-9 activity, which is coupled with a decrease in release of TIMP-1 and an increase in activated cathepsin B within the extracellular space. This process occurs independently of NLRP3-inflammasome formation. Treatment with a P2X7 receptor antagonist prevents ATP-induced MMP-9 activity, inhibition of active cathepsin B release and allows for TIMP-1 to be released from the cell. We have shown that cathepsin B degrades TIMP-1, and inhibition of cathepsin B allows for release of TIMP-1 and inhibits MMP-9 activity. We also present data that indicate that ATP or cell damage induces glial cell migration, which is inhibited by P2X7 antagonism, depletion of MMP-9 or inhibition of cathepsin B.

Abbreviations used

apoptotic speck-like protein


danger-associated molecular pattern


Dulbecco's Modified Eagle's Medium


matrix metalloproteinases




stress-inducible protein 1


tissue inhibitors of metalloproteinases


urokinase-type plasminogen activator

Microglia are regarded as the immunocompetent cells of the central nervous system and thus play an important role in response to any brain injury. Under resting conditions, microglia are found in a motile ramified state where they constantly extend processes to sample the surrounding parenchyma (Nimmerjahn et al. 2005). Upon stimulation, for example in response to injury, microglia adopt an activated phenotype. They retract their processes and migrate towards the site of injury, where they release pro- and anti-inflammatory cytokines with the objective of limiting damage by phagocytosing cellular debris and subsequently, to exert a restorative effect by initiating tissue repair.

The purine adenosine 5′-triphosphate (ATP) has previously been shown to act as a potent chemoattractant for microglia when present in the extracellular milieu (Choi et al. 2010). Under normal physiological conditions, extracellular concentrations of ATP are maintained between 400 and 700 nM. However, under pathological conditions, for example ischaemia, damaged cells release ATP and increase extracellular levels to concentrations in the high millimolar range (Melani et al. 2005). At these concentrations, ATP acts as a danger-associated molecular pattern (DAMP) and can act on members of the purinergic receptor family. The P2X7 receptor is a member of this family of receptors (Bianchi 2007). It has previously been shown to be present on immune cells including macrophages (Qu et al. 2007) and microglia (Murphy et al. 2012).

For cell migration to occur, proteases must cleave cell–cell interactions or cell–matrix interactions. Matrix metalloproteinases (MMPs) are a family of zinc-dependent proteolytic enzymes that have previously been shown to be involved in cell migration (Palmisano and Itoh 2010). MMPs have several functions including cell migration, morphogenesis, cell proliferation, apoptosis and modulation of the activity of enzymes or other biologically active molecules such as growth factors. As proteases, matrix metalloproteinases have the potential to be destructive to surrounding tissue and therefore their expression and activity is tightly regulated. They are expressed as inactive zymogens and must be processed to expose their active site. They are also regulated by a family of endogenous inhibitors called tissue inhibitors of matrix metalloproteinases (TIMPs). The family consists of four members, TIMP-1, -2, -3 and -4. TIMPs are 20–30kDa in size and interact directly with MMPs through a small number of amino acids. All MMPs can be inhibited by TIMPs once activated, but the MMPs can also form complexes with TIMPs in their latent form.

Another family of proteases, the cathepsins, which are cysteine peptidases, have also been implicated in the process of cell migration. The cathepsins are a family of 15 proteins, which are located predominantly within lysosomes. However, in response to neurological insults (Fogarty et al. 2010) (Kilinc et al. 2010), the lysosomal membrane can become destabilized and result in the release of cathepsins into the surrounding extracellular space. Their function in the extracellular space is still debated, but a role in cell migration has been suggested (Veeravalli et al. 2010).

The data here propose a mechanism by which ATP or cell injury induces cell migration. We demonstrate that ATP, acting through the P2X7 receptor, induces release of cathepsin B into the extracellular space where it degrades TIMP-1, permitting the MMP-9-dependent migration of glial cells.

Experimental procedure

Primary mixed glial cultures

Mixed glia were prepared from the cortices of 1-day-old C57Bl6 mice (Trinity College, Dublin, Ireland). Cortical tissue was cross-chopped, incubated for 25 min at 37°C in Dulbecco's Modified Eagle's Medium (DMEM, Invitrogen, Paisley, UK) supplemented with 10% foetal bovine serum (FBS; Invitrogen) and 50 U/mL penicillin/streptomycin (Invitrogen) and plated (2.5 × 105) as previously described (Nolan et al. 2004). Cells from each mouse were treated separately. After 13 days, cells were treated with ATP (Sigma-Aldrich, Dublin, Ireland) for 2 or 4 h. In some experiments, cells were incubated in the presence of the P2X7 inhibitor, GSK1370319A [20 μM GlaxoSmithKline, Harlow, UK; (Murphy et al. 2012)], the specific cathepsin B inhibitor II (Ac-LVK-CHO; 1 μM; Merck, Middlesex, UK) or the caspase 1 inhibitor (Z-WEHD-FMK; 300 nM; Merck) for 1 h prior to addition of ATP. It should be noted that in addition to inhibiting caspase 1, Z-WEHD-FMK also inhibits caspase 5; importantly the evidence suggests that glia do not express caspase 5. In one experiment, glia were treated with ATP in a Ca2+-free solution containing 1 mM BAPTA-AM to chelate any free intracellular calcium. In another series of experiments to knock down the MMP9 gene, mixed glia were incubated in the presence or absence of MMP9 siRNA (Dharmacon, Lafayette, CO, USA) and DharmaFECT 1 transfection reagent (Dharmacon) for 48 h; western blot analysis was used to confirm knockdown of the MMP-9 protein. In all experiments, supernatants were collected and cells were harvested for later analysis. An MTS viability assay (Promega, Southampton, UK) was performed on cells that had been treated with ATP (5 mM) for 2 and 4 h.

To prepare purified microglia and astrocytes, cells were grown in T25 flasks in DMEM as described above. After 12 days, the flasks were shaken for 2 h at 0.22 g at 21°C and tapped several times to remove microglia. The supernatants were removed from the flask and centrifuged at 400 g for 3 min at 21°C. The pellet was resuspended in DMEM and the cells were counted. Cells were pipetted into six-well plates at a density of 1 × 105 cells/mL. To prepare astrocytes, the flasks containing the adherent astrocytes were washed with phosphate-buffered saline and 1 mL of 0/05% w/v trypsin–EDTA was added at 37°C until the cells just began to detach; DMEM containing FBS, was then added to the flask to inhibit the action of trypsin. The cells were centrifuged at 2000 rpm for 3 min. The pellet was resuspended in DMEM and the cells were plated in nine-well plates at a density of 1 × 105 cells/mL. Isolated microglia and astrocytes were treated in the same manner as mixed glial cultures. All experiments were approved by the Trinity College Dublin Ethics Committee.

Analysis of TIMP-1

Supernatant concentrations of TIMP-1 obtained from glial cultures were measured using ELISA (R&D systems, Abingdon, UK). TIMP-1 concentrations in the test samples were evaluated with reference to the standard curve prepared using recombinant TIMP-1 protein of a known concentration.

Analysis of MMP-9 activity

MMP-9 activity was determined using a fluorometric MMP-9 assay kit (Anaspec, Freemont, CA, USA). Samples of supernatant were incubated with a fluorogenic MMP-9 substrate (5FAM/QXL FRET5 peptide) On cleavage, the substrate emits a fluorescent signal, which was monitored at excitation/emission wavelengths 490 nm/520 nm.

Analysis of cathepsin B activity

Cathepsin B activity was assessed using a fluorometric cathepsin B assay kit (Abcam, Cambridge, UK). Samples of supernatant were incubated with the synthetic cathepsin B substrate RR-AFC. Upon cleavage, the substrate emits a fluorescent signal, which was monitored at excitation/emission wavelengths 380 nm/500 nm.

Analysis of protein expression by western immunoblotting

Western blotting was performed as previously described (Lyons et al. 2007). Protein was precipitated from cell culture supernatants using tetraacetic acid (5%) and the protein pellets were homogenized in buffer containing Tris-HCl (0.01 M) and EDTA (1 mM). The lysate (20 μg) was boiled in gel-loading buffer, applied to 12% sodium dodecyl sulphate–polyacrylamide gels and separated by electrophoresis. Proteins were transferred to nitrocellulose membranes and incubated with an antibody against MMP-9 (Abcam). Membranes were incubated with horseradish peroxides-conjugated secondary antibodies (Jackson ImmunoResearch, Westgrove, PA, USA) and bands were visualized using Supersignal West Pico Chemiluminescent Substrate (Pierce, Rockford, IL USA). Images were captured using a Fujifilm LAS-3000 (Brennan and Co, Dublin, Ireland). Densitometry was performed using ImageJ software (

Protein degradation assay

Mouse recombinant cathepsin B (R&D systems) was reconstituted in activation buffer (50 mM sodium acetate, 2 mM dithiothreitol , 2 mM ethylenediaminetetraacetic acid). Some activated cathepsin B was heat denatured by incubating it at 100°C for 5 min. Activated or heat-denatured cathepsin B (10 ug/mL) was incubated with recombinant TIMP-1 (20 ug/mL; Sigma Aldrich, Dublin, Ireland) for 1 h at 37°C. The reaction was stopped using 1x protease inhibitor (Sigma Aldrich). Sample buffer (2x sodium dodecyl sulfate) was added to each sample, which were run out on a 15% polyacrylamide gel and the proteins were visualized by silver staining.

Cell migration assay

Primary mixed glial cells (14-day old) were seeded onto a Millipore cell culture insert (8-μm pore size; 1 × 105 cells per insert). Inserts were placed into a 24-well plate containing DMEM with or without 5 mM ATP, 20 μM GSK1370319A or 1 μM cathepsin B inhibitor. In some experiments, mixed glial cells were previously treated with MMP9 siRNA, as described above. Cells were incubated at 37°C for 3 h. Inserts with cells were fixed with 4% formaldehyde for 30 min. Non-migrating cells from the upper side of the cell culture insert were removed and migrating cells from the underside of the insert were stained using 0.1% Coomassie Blue (Sigma Aldrich). Cells were counted under a light microscope under 10x magnification. The cell migratory index was calculated by expressing the number of migratory cells in each treatment group as a fold change from control (Choi et al. 2010).

Scratch assay

Mixed glial cultures were plated on a six-well plate at a density of 2 × 105 cells per well. Some wells were pre-treated with 20 μM GSK1370310A, 10 μM cathepsin B inhibitor or MMP-9 siRNA or 5 U/mL potato apyrase (Sigma Aldrich),. A 200 μL sterile pipette tip was scratched across the middle of each well to produce a denuded area. Cytosine arabinfuranoside (10 μM; Sigma Aldrich) was added to each well after the scratch wound to prevent cell proliferation. A digital photograph was taken of each well immediately after the scratch (0 h) and at 24 h. Image J, the image processing program, was used to measure the area of damage at each time point from the digital images.

Statistical analysis

Data were analysed as appropriate using either Student's t-test for independent means or analysis of variance (anova) followed by post-hoc Tukey's test to determine which conditions were significantly different from each other. Data are expressed as means ± SEM.


ATP inhibits TIMP-1 release and activates MMP-9

Mouse mixed glial cultures were treated with a range of concentrations of ATP from 1 μM to 5 mM and TIMP-1 release into supernatants was assessed by ELISA at 2 h. Higher concentrations of ATP (1 and 5 mM) significantly reduced the amount of released TIMP-1 compared with that released from untreated cells at 2 h (*< 0.05; ***< 0.001; anova; Fig. 1a), whereas lower concentrations of ATP (1 μM, 50 μM and 100 μM) had no effect. TIMP-1 is the endogenous inhibitor of MMP-9, therefore, next we measured MMP-9 activity in the samples of supernatant obtained from these mixed glia using a fluorometric assay. ATP (1 mM) increased activity of MMP-9 (***< 0.001; anova; Fig. 1b), however, the most profound effect was seen in cells treated with 5 mM ATP (***< 0.001; anova; Fig. 1b).

Figure 1.

Mixed glial were stimulated with 1, 50 and 100 μM, and 1 and 5 mM concentrations of ATP for 2 h. Concentrations of 1 and 5 mM induce a decrease in released tissue inhibitors of metalloproteinase (TIMP)-1 (*< 0.05; ***< 0.001; anova; Control vs. ATP) (a) and increased activity of matrix metalloproteinase (MMP)-9 (*< 0.05; ***< 0.001; anova; Control vs. ATP) (b) from mixed rat glial cells.

Inhibition of the P2X7 receptor attenuates the ATP-induced changes in TIMP-1 and MMP-9

The P2X purinoreceptors are a family of ligand-gated ion channel-linked receptors that are activated in response to extracellular ATP. The family member, P2X7, is of particular interest, as it is only activated by concentrations of ATP in the millimolar range (1–5 mM). As changes in TIMP-1 and MMP-9 activity were seen with millimolar concentrations of ATP, we investigated whether activation of the P2X7 receptor was required for these changes.

Mixed glial cultures were treated with 5 mM ATP and 20 μM GSK1370319A, a specific P2X7 antagonist (Murphy et al. 2012). ATP has previously been shown to induce cell death; however, treatment of mixed glial cells, with 5 mM ATP, in this setting, did not induce significant cell death at either 2 or 4 h after stimulation as assessed by MTS assay [cell viability (percentage from control) at 2 h post ATP stimulation was 96.99 ± 8.47% and at 4 h post ATP stimulation was 96.99 ± 2.74%]. As previously seen, 5 mM ATP decreased the TIMP-1 release at 2 h and 4 h (*< 0.05; ***< 0.001; anova; Fig. 2a) compared with control supernatants from 4 h. However, supernatant concentration of TIMP-1 returned to control levels by 24 h (data not shown). Treatment with GSK1370319a inhibited this ATP-induced decrease in TIMP-1 at both 2 and 4 h (#< 0.05; ##< 0.01; anova; Fig. 2a).

Figure 2.

Inhibition of the P2X7 receptor attenuates the ATP (5 mM)-induced changes in tissue inhibitors of metalloproteinase (TIMP)-1 at 2 and 4 h after ATP stimulation [*< 0.05; ***< 0.001; anova; Control vs. ATP; ##< 0.01; #< 0.05; anova; ATP vs. ATP + GSK1370319A (GSK319)] (a) and attenuates the ATP-induced activity of matrix metalloproteinase (MMP)-9 at 2 and 4 h (**< 0.01, anova; Control vs. ATP; ###< 0.01; anova; ATP vs. ATP + GSK319) (b) in mixed glial cells. P2X7 inhibition does not affect ATP-induced release of MMP-9 from mixed glial cells (c).

Next, we investigated whether MMP-9 activity was affected by inhibition of the P2X7 receptor and show that activity was increased in response to ATP at both 2 and 4 h (**< 0.01, anova, Fig. 2b); mirroring the change in supernatant concentration of TIMP-1, MMP-9 activity had returned to control levels by 24 h (data not shown). Treatment with GSK130319A decreased the activity of MMP-9 at both time points (###< 0.001; anova; Fig. 2b).

Whereas ATP significantly increased total released MMP-9 (***< 0.001, anova, Fig. 2c), GSK130319A exerted no effect on this ATP-induced change, indicating that activation of the P2X7 receptor plays a role in triggering activation of MMP-9, but not its release.

ATP inhibits release of TIMP-1 and activation of MMP-9 from both microglia and astrocytes

To assess whether TIMP-1 and MMP-9 were being released exclusively from one glial cell type, purified cultures of microglia and astrocytes were prepared. TIMP-1 was released from both microglia (Fig. 3a) and astrocytes (Fig. 3b) and release from both cell types was inhibited by treatment with 5 mM ATP for 2 and 4 h, but the effect of ATP was greater in the microglial cells (**< 0.01; ***< 0.001; anova; Fig. 3a and b). Blocking the P2X7 receptor by incubating the cells in the presence of GSK130319A prevented the ATP-induced inhibition of TIMP-1 release in both cell types (#< 0.05; ##< 0.01; anova; Fig. 3a and b).

Figure 3.

ATP (5 mM) inhibits the release of tissue inhibitors of metalloproteinase (TIMP)-1 from both microglia (a) and astrocytes (b) (**< 0.01; ***< 0.001; anova; Control vs. ATP) at 2 and 4 h after stimulation. Inhibition of the P2X7 receptor also attenuates this change in TIMP-1 release in both cell types, but to a greater extent in the isolated microglia (a and b) [##< 0.01, #< 0.05; anova; ATP vs. ATP + GSK1370319A (GSK319)]. ATP induces activation of matrix metalloproteinase (MMP)-9 in both microglia (c) and astrocytes (d) (***< 0.001; anova; Control vs. ATP), which is attenuated by P2X7 antagonism in both cell types (##< 0.01; ###< 0.001; #< 0.05; anova; ATP vs. ATP + GSK319).

ATP induced an increase in the activity of MMP-9 in samples of supernatant prepared from microglia and astrocytes (***< 0.001; anova; Fig. 3c and d). GSK1370319A treatment significantly inhibited the ATP-induced MMP-9 activity in both microglia and astrocytes (#< 0.05, ##< 0.01, ###< 0.001; anova; Fig. 3c and d).

ATP-induced changes in TIMP-1 and MMP-9 activation are not dependent on the NLRP3 inflammasome

Activation of the P2X7 receptor has previously been shown to initiate assembly of the NLRP3 inflammasome, which results in activation of caspase 1 and processing of the cytokine interleukin-1β (Murphy et al. 2012). As inhibition of the P2X7 receptor prevents the ATP-induced activation of MMP-9, we investigated whether this process was dependent on activation of the NLRP3 inflammasome by inhibiting caspase 1 activity with the inhibitor Z-WEHD-FMK. As previously seen, ATP decreased supernatant concentration of TIMP-1 (***< 0.001; anova; Fig. 4a), however, this effect was not reversed by treatment with Z-WEHD-FMK. Similarly, ATP induced an increase in MMP-9 activity, which was also unaffected by inhibition of caspase 1 (Fig. 4b).

Figure 4.

The ATP (5 mM)-induced changes in tissue inhibitors of metalloproteinase (TIMP)-1 (a) (*< 0.05; **< 0.01; anova; Control vs. ATP) and matrix metalloproteinase (MMP)-9 (b) (***< 0.001; anova; Control vs. ATP), 2 and 4 h after stimulation, were not affected by treatment with the caspase 1 inhibitor Z-WEHD-FMK.

Extracellular cathepsin B activity is increased with millimolar concentrations of ATP; modulation by antagonism of the P2X7 receptor

It has previously been shown that cathepsins are able to inactivate TIMPs by degradation in human articular chrondocytes (Kostoulas et al. 1999). Therefore, we investigated the activity of cathepsin B in response to ATP treatment at 2 h post stimulation and report that its activity was significantly increased in the supernatants obtained from mixed glial cells treated with 1 mM, and to a greater extent 5 mM, ATP but not with any of the lower concentrations (0.001, 0.05, 0.1 mM) of ATP (*< 0.05; ***< 0.001; anova; Fig. 5a).

Figure 5.

Cathepsin B activity is increased extracellularly in response to millimolar concentrations of ATP (1 and 5 mM) at 2 h post stimulation (a) (*< 0.05; ***< 0.001; anova; Control vs. 1 mM and 5 mM ATP). This ATP (5 mM)-induced increase in cathepsin B activity, in the extracellular space, is modulation by antagonism of the P2X7 receptor and also requires the presence of Ca2+ (b) [++< 0.01; anova; ATP vs. ATP + GSK1370319A (GSK319) and ATP vs. ATP + Ca2+-free media].

Next, we investigated whether cathepsin B activity was inhibited by antagonism of the P2X7 inhibitor. Mixed glial cells were treated with or without ATP (5 mM) and GSK1370319A (20 μM); the data show that ATP significantly increased cathepsin B activity (***< 0.001; anova; Fig. 5b) and that antagonism of the P2X7 receptor inhibited activity in the supernatants obtained from ATP-treated cells (++< 0.01; anova; Fig. 5b). Lopez-Castejon et al. (2010) have previously shown that ATP-induced release of cathepsin B from bone marrow-derived macrophages is Ca2+ dependent; therefore, we investigated whether this was the case in glial cells. Glial cells were treated with ATP in a Ca2+-free solution containing 1 mM BAPTA–AM. Active cathepsin B in the supernatants of ATP-treated cells was significantly reduced in Ca2+-free supernatants (++< 0.01; anova; Fig. 5b).

Cathepsin B degrades TIMP-1 and inhibiting its activation prevents ATP-induced changes in TIMP-1 release and MMP-9 activation

To assess whether cathepsin B was capable of degrading TIMP-1, recombinant active or heat-denatured cathepsin B and TIMP-1 were coincubated at 37°C for 1 h. The resulting lysates were run out on a gel and silver stained (Fig. 6a). In the samples where TIMP-1 was incubated with cathepsin B (lane 4; Fig. 6a), full-length TIMP-1 decreased and a large band of small molecular weights appeared at the bottom of the gel, but the cathepsin B band did not decrease (lane 4; Fig. 6a). In the sample containing heat-denatured cathepsin B and TIMP-1, neither TIMP-1 nor cathepsin B were degraded (lane 5, Fig. 6a).

Figure 6.

The silver-stained gel indicates that recombinant cathepsin B degrades recombinant tissue inhibitors of metalloproteinase (TIMP)-1, 1 h after incubation. However, heat-inactivated cathepsin B did not degrade TIMP-1 (a). Pre-treatment of mixed glial cells with a cathepsin B inhibitor (1 μM) prevents the ATP-induced (5 mM) attenuation of TIMP-1 release (b) (**< 0.01, ***< 0.001; anova; Control vs. ATP; ###< 0.001; ##< 0.01; anova; ATP vs. ATP + cathepsin B inhibitor) and activation of matrix metalloproteinase (MMP)-9 (c) (**< 0.01, ***< 0.001; anova; Control vs. ATP; ###< 0.001; anova; ATP vs. ATP + Cathepsin B inhibitor) at 2 and 4 h post ATP stimulation.

We next investigated whether inhibition of cathepsin B affected the ATP-induced changes in TIMP-1 release or MMP-9 activity. Figure 6b shows that the ATP-induced inhibition of released TIMP-1 (**p < 0.01; ***< 0.001; anova) was attenuated by treating cells with a cathepsin B inhibitor (##< 0.01; ###< 0.001; anova; Fig. 6b). Similarly, the ATP-induced activity of MMP-9 (***< 0.001; anova) was inhibited by the cathepsin B inhibitor at 2 and 4 h post ATP stimulation (###< 0.001; anova; Fig. 6c).

Inhibition of the P2X7 receptor or cathepsin B or knockdown of MMP-9 inhibit ATP or cell damage-induced cell migration

Previous studies have suggested a role for matrix metalloproteinases in cell migration (Shin et al. 2010; Wang et al. 2010; Zhao et al. 2010). Therefore, we measured the movement of glial cells across a semiporous membrane in response to 5 mM ATP (Fig. 7a). ATP-treated glial cells had an average cell migratory index of 3.2 (***< 0.001; Fig. 7b). When cells were treated with the P2X7 inhibitor, the migratory index was significantly decreased to 1.3 (##< 0.01; anova; Fig. 7b) and migration of cells was clearly reduced (Fig. 7a). Knocking down MMP-9 using an siRNA similarly decreased the cell migratory index to 0.95 (##< 0.01; anova; Fig. 7b). In addition, the cathepsin B inhibitor significantly decreased the cell migratory index to 1.5 (#< 0.05; anova; Fig. 7b).

Figure 7.

Photomicrographs of coomassie-stained mixed glial cells show ATP induces cell migration across a semiporous membrane, which was inhibited in cells treated with 20 μM GSK1370319A (GSK319), matrix metalloproteinase (MMP)-9 siRNA or 1 μM cathepsin B inhibitor (a). Knockdown of MMP-9 was confirmed by western blot with actin used as a loading control (b). The cell migratory index was quantified by counting migratory cells in five fields of view (x10 magnification) from n = 8 cultures for each treatment (c) ***< 0.001; anova; Control vs. ATP, #< 0.05; anova; ATP vs. ATP + Cathepsin B Inhibitor, ##< 0.01; anova; ATP vs. ATP + GSK319 or ATP + mmp9 siRNA. Cell damage induced by a scratch assay elicits migration of cells into the area denuded by the scratch (d). Photomicrographs of the scratch area at 0 h and 24 h indicate that migration of cells in response to damage was inhibited by 20 μM GSK319, MMP-9 siRNA, 1 μM cathepsin B inhibitor or 5 U/mL apyrase. Cell migration was quantified by the measuring the area of the denuded area at both time points. Treatment with GSK319, MMP-9 siRNA, cathepsin B inhibitor and apyrase significantly inhibit the percentage change in the denuded area (n = 6 per treatment group) (***< 0.001; anova; Untreated vs. Apyrase or GSK319, **p,0.01; anova; Untreated vs. mmp9 siRNA or Cathepsin B inhibitor) (e).

One of the functions of cell migration is to facilitate repair of damaged cells. Here, we investigated whether inhibition of the P2X7 receptor would inhibit the ability of glia to migrate to cells in response to damage induced by a scratch made on a monolayer of mixed glial cells. The extracellular concentration of ATP in the supernatants of the scratched cells was increased 10-fold compared with control cells (data not shown). Figure 7d shows the scratches made and imaged at 0 h and then imaged again at 24 h. The cells were either untreated or treated with 5 U/mL Apyrase, 20 μM GSK1370319A, cathepsin B inhibitor or MMP-9 siRNA. In control cells after 24 h, the scratch area was decreased in size by 36.43 ± 10.5%. Inhibition of the P2X7 receptor by GSK1370319A prevented the migration of glial cells into the damaged area so that the average decrease in the size of the scratch area after 24 h was 15.0 ± 2.9%. Cathepsin B inhibition and knockdown of MMP9 also prevented migration of the glial cells in response to cell damage; the average decrease in the size of the scratched areas were 12.7 ± 4.0% and 12.1 ± 4.4% respectively. Hydrolysis of ATP by potato apyrase inhibited migration of the glial cells into the scratched area. After 24 h, the scratch area in apyrase-treated cultures was decreased by 12.9 ± 7.1%. (Fig. 7d and e).


In this study, we have demonstrated that ATP, acting via the P2X7 receptor induces increased activity of the matrix metalloproteinase, MMP-9, in glial cells. The activation of MMP-9 occurs as a result of degradation of its endogenous inhibitor, TIMP-1, by cathepsin B and this series of events provides a mechanism for the ATP-induced migration of cells. The data indicate that the specific P2X7 receptor antagonist inhibits migration of cells by blocking the ATP-induced activity of MMP-9.

Extracellular ATP is a ‘find me’ signal that dying cells release to recruit phagocytic cells to the area of damage (Elliott et al. 2009), and here we demonstrate that its ability to influence migration of glia is inhibited by a specific antagonist at the P2X7 receptor. Importantly, the data show that the ATP-induced migration is dependent on MMP-9 activity. The link between MMP-9 activity and cell migration has been widely studied in endothelial cells (Regina et al. 2003), leucocytes (Khandoga et al. 2006) and monocytes (Zhou et al. 2012). To date, relatively few studies have investigated the role of MMP-9 in glial migration, although it was suggested that the migration of microglial cells in response to urokinase-type plasminogen activator (uPA) is mediated by MMP-9 (Shin et al. 2010). Similarly, stress-inducible protein 1 (STI1) activates MMP-9 and induces cell migration in microglial cultures (da Fonseca et al. 2012). Several processes are necessary for cell migration including extension and projection of the cell membrane, as well as cell-to-extracellular matrix adhesion, which relies on ligation of adhesion receptors by components of the extracellular matrix. Although the mechanism by which MMP-9 modulates cell migration is not well understood, it has been shown to modulate the interaction between the receptor CD44, which is present in astrocytes and microglia, and hyaluronan in the ECM leading to increased motility of glioblastoma cells (Chetty et al. 2012).

Our data indicate that MMP-9 activation occurs downstream of P2X7 receptor activation and is evident in mixed glia as well as isolated microglia and isolated astrocytes; this broadly supports previous data, which indicate that extracellular ATP induces MMP-9 release from peripheral blood mononuclear cells (Gu and Wiley 2006) and increases its expression in mesangial (Huwiler et al. 2003) and epithelial (Wesley et al. 2007) cells. ATP-induced release of MMP-9 from microglia and BV-2 cells has also been reported (Choi et al. 2010). The authors reported that ATP-induced migration of BV2 cells was MMP-9 dependent, and argued that the effect was mediated by activation of P2Y1 and P2Y12 receptors as it was blocked by pyridoxal-phosphate-6-azophenyl-2′,4′-disulphonate (PPADS) and clopidogrel. However, it has been suggested that PPADS is not a specific P2Y1 receptor antagonist and that it can also act as an antagonist of P2X receptors (Oliveira et al. 2011) (Prasai et al. 2011). These other purinergic receptors, however, may play a role in the secretion of MMP-9 from the cell into the extracellular space. Although our data indicate that inhibition of the P2X7 receptor inhibits MMP-9 activation, there is no evidence of any change in the secretion of MMP-9 from the cell when this receptor is antagonized. Activation of P2Y2 by ATP has been shown to induce release of the chemokine MCP-1 from immune cells (Stokes and Surprenant 2007) and this, in turn, can induce the secretion of MMP-9 (Cross and Woodroofe 1999). Similarly, another inducer of MMP-9 secretion, RANTES (Cross and Woodroofe 1999), is released from the cell in response to ATP activation of P2Y1 (Pastore et al. 2007). These reports, taken together with the evidence presented here, suggest that purinergic receptors other than P2X7, play a role in MMP-9 secretion.

In an effort to understand the events leading up to MMP-9 activation, we examined ATP-induced release of the endogenous inhibitor of MMP-9, TIMP-1, from mixed glial cells. The data indicate that ATP decreased its release and that this effect was attenuated by inhibition of the P2X7 receptor with GSK1370319A. Similar effects were observed in both isolated microglia and isolated astrocytes, however, the effect of the P2X7 antagonist was more profound in microglia. This difference may be accounted for by our previously reported finding that the P2X7 receptor is expressed to a much greater extent on microglia than astrocytes (Murphy et al. 2012). Mirroring the changes observed here, it has been reported that release of activated MMP-9 was associated with decreased TIMP-1 release in peripheral blood mononuclear cells (Gu and Wiley 2006), although the underlying mechanism was not identified. Here we report, for the first time, that active cathepsin B is released from primary mouse glial cells in a P2X7-dependent manner. In a similar manner, activation of the P2X7 receptor by ATP has been shown to release cathepsin B from cancer cells, mouse bone marrow-derived macrophages and human lung macrophages (Lopez-Castejon et al. 2010; Jelassi et al. 2011). Lopez-Castejon et al. report that cathepsin B release in response to P2X7 receptor activation occurs in a calcium-dependent manner in bone marrow-derived macrophages. This study indicates that this process is also calcium dependent in glia. The evidence presented here suggests that cathepsin B degrades TIMP-1 and that ATP-dependent activation of MMP-9 is blocked by inhibiting cathepsin B. Therefore, we conclude that cathepsin B may play a key role in triggering the events leading to ATP-induced cell migration. Interestingly, TIMP-1, and also TIMP-2, have been identified as substrates for cathepsin B in human chondrocytes (Kostoulas et al. 1999), and a role for both cathepsin B and MMP-9 in migration of human glioma cells has been reported (Veeravalli et al. 2010). Cathepsin B is a protease, which is normally found in lysosomes. However, in response to insults, for example ischaemic injury (Qin et al. 2008), where release of ATP from damaged cells is significant, cathepsin B leaks into the cytoplasm. Indeed, the evidence suggests that its release, as a result of increased lysosomal membrane permeability, may contribute to the pathogenesis of neurodegenerative diseases (Wang and Qin 2010).

Activation of the P2X7 receptor in glial cells has been shown to result in maturation and release of the proinflammatory cytokine IL-1β from lipopolysaccharide-primed cells (Murphy et al. 2012). In this setting, activation of the P2X7 receptor results in recruitment association of three of proteins, apoptotic speck-like protein (ASC), the nod-like protein, NLRP3 and the cysteine protease, caspase 1. This assembly of proteins is known as the NLRP3 inflammasome and it drives the proteolytic cleavage of the cysteine protease, caspase 1 which, in turn, cleaves pro-IL-1β to the mature form enabling its release from the cell. We investigated whether the P2X7-mediated decrease in TIMP-1 and subsequent activation of MMP-9 was dependent on formation of this inflammasome. The data indicate that the caspase 1 inhibitor, Z-WEHD-FMK, exerted no effect on ATP-induced TIMP-1 release or MMP-9 activation, and therefore we can conclude that these ATP-associated changes which lead to cell migration are independent of inflammasome formation. Recently, it has been reported that P2X7-dependent release of cathepsin B is independent of the NLRP3 inflammasome-processed IL-1β (Lopez-Castejon et al. 2010).

We propose a novel mechanism for the migration of glia in response to injury or cell damage in the brain that is dependent on activation of the P2X7 receptor. Receptor activation induces a sequence of events initiated by release of cathepsin B that results in cell migration. Using a model of cell damage, as well as application of ATP to glia, the evidence indicates that migration is blocked by the P2X7 receptor antagonist, GSK1370319A and also by inhibiting cathepsin B and MMP-9. Targeting this pathway has the potential to provide novel pharmacological tools, which may enable repair in certain neurological disorders.


This work was funded by GlaxoSmithKline. The authors of this paper declare no conflicts of interest. NM designed and carried out all of the experiments and wrote the manuscript. MAL revised the manuscript and gave final approval on the version to be published.