Research ArticleCell Migration

ATP promotes the fast migration of dendritic cells through the activity of pannexin 1 channels and P2X7 receptors

See allHide authors and affiliations

Science Signaling  21 Nov 2017:
Vol. 10, Issue 506, eaah7107
DOI: 10.1126/scisignal.aah7107

ATP increases dendritic cell speed

When dendritic cells (DCs) in peripheral tissues encounter danger-associated signals, such as microbial products or ATP released from damaged cells, they migrate to lymph nodes to activate T cells and initiate the adaptive immune response. Sáez et al. found that ATP stimulated P2X7 receptors in DCs, which resulted in the opening of pannexin 1 (Panx1) channels and the release of ATP as part of an autocrine loop that increased DC migration speed. DCs from Panx1-deficient mice migrated more slowly than did DCs from wild-type mice. When injected into the footpads of mice, ATP-treated Panx1-deficient DCs exhibited defective migration to draining lymph nodes. Together, these data suggest that P2X7 receptors and Panx1 channels facilitate the speedy migration of DCs to lymph nodes in response to danger signals.

Abstract

Upon its release from injured cells, such as infected, transformed, inflamed, or necrotic cells, extracellular adenosine-5ʹ-triphosphate (ATP) acts as a danger signal that recruits phagocytes, such as neutrophils, macrophages, and dendritic cells (DCs), to the site of injury. The sensing of extracellular ATP occurs through purinergic (P2) receptors. We investigated the cellular mechanisms linking purinergic signaling to DC motility. We found that ATP stimulated fast DC motility through an autocrine signaling loop, which was initiated by the activation of P2X7 receptors and further amplified by pannexin 1 (Panx1) channels. Upon stimulation of the P2X7 receptor by ATP, Panx1 contributed to fast DC motility by increasing the permeability of the plasma membrane, which resulted in supplementary ATP release. In the absence of Panx1, DCs failed to increase their speed of migration in response to ATP, despite exhibiting a normal P2X7 receptor–mediated Ca2+ response. In addition to DC migration, Panx1 channel– and P2X7 receptor–dependent signaling was further required to stimulate the reorganization of the actin cytoskeleton. In vivo, functional Panx1 channels were required for the homing of DCs to lymph nodes, although they were dispensable for DC maturation. These data suggest that P2X7 receptors and Panx1 channels are crucial players in the regulation of DC migration to endogenous danger signals.

INTRODUCTION

Dendritic cells (DCs) are bone marrow–derived cells that are sparsely but widely distributed in peripheral tissues. Upon encounter with danger-associated signals, DCs migrate to draining lymph nodes to activate T lymphocytes and initiate adaptive immunity (1). DC migration is therefore instrumental to the onset of the immune response. Danger signals include pathogen-associated molecular patterns (PAMPs), such as bacterial lipopolysaccharide (LPS), and damage-associated molecular patterns (DAMPs), such as adenosine-5′-triphosphate (ATP) (2, 3). The case of extracellular ATP is particularly intriguing because it regulates DC migration at different stages (4). Upon tissue damage, ATP released into the extracellular milieu induces the rapid recruitment of immune cells (4, 5). Simultaneously, but with reduced kinetics, ATP activates DCs and stimulates the cell surface expression of CCR7, a chemokine receptor that facilitates DC migration to lymph nodes (6). ATP therefore emerges as a danger signal that links local sensing to adaptive immunity.

In tissues, the steady-state extracellular concentration of ATP is maintained in the nanomolar range but markedly increases (up to the millimolar range) upon tissue damage, inflammation, infection, and other pathological conditions (79). The presence of extracellular ATP is sensed through purinergic receptors (P2), which are located at the cell surface and classified into two families, P2X or P2Y, which correspond to ion channels and metabotropic receptors, respectively (10). P2 receptors are differentially activated and are involved in a plethora of cellular responses depending on the exposure time and the concentration of ATP (11, 12). In particular, ATP released from damaged cells to the extracellular milieu promotes cell migration through paracrine signaling (4, 5). In addition, ATP released from migrating cells acts in an autocrine manner, enhancing the motility of the ATP-releasing cells (4, 12, 13); however, whether such signaling mechanisms function in DCs is unclear.

Different mechanisms of ATP release might contribute to autocrine signaling that is stimulated during cell migration, including membrane channels constituted of pannexin 1 (Panx1) (4, 1215). Panx1 channels enable the passage of small metabolites, signaling molecules, and fluorescent dyes, and their opening is associated with the activation of P2X7 (16) and some P2Y receptors (17). Panx1 is ubiquitously expressed in the immune system and is involved in various cellular processes, such as cell migration, T cell activation, and cell death (14). In DCs, the presence of functional Panx1 channels is suggested from experiments in which dye uptake was observed after the ATP-dependent activation of P2X7 receptors (1824). However, functional evidence of the activity of those channels and their potential influence on immune surveillance by DCs is still missing.

Here, we combined the use of microfabricated devices and live-cell microscopy to show that the ATP-stimulated activation of DCs promoted their fast migration, which likely occurred through a sustained autocrine purinergic signaling loop. This loop depended on the opening of Panx1 channels, P2X7 receptor activation, and subsequent Ca2+ signaling and resulted in the rearrangement of the actin cytoskeleton to maintain fast migration. Our results identify Panx1 channels and the P2X7 receptor as players in the regulation of cytoskeletal organization and DC locomotion. Thus, modulation of Panx1 channels and P2X7 receptor activity in DCs might help modulate the immune sentinel functions of these cells in a pathological context, such as asthma or cancer, in which ATP concentrations are dysregulated.

RESULTS

ATP-stimulated membrane permeability in DCs involves Panx1 channels

Upon activation, Panx1 channels and other large pore–forming proteins enable the diffusion of ions, small molecules, and also fluorescent dyes, such as 4′,6-diamidino-2-phenylindole (DAPI) (16, 25). After passive diffusion across the cell membrane (from now on referred to as uptake), DAPI fluoresces when it intercalates nucleic acids, and it can therefore be used to evaluate the membrane permeabilization that results from the opening of Panx1 or other large pore–forming proteins (17). We found that treatment with extracellular ATP (500 μM) substantially increased DAPI uptake by wild-type (WT) DCs (Fig. 1A and movie S1). As previously observed in other immune cells (16), this response presented two components (fig. S1A) and was only observed at high ATP concentrations [median effective concentration (EC50), 473.4 μM ATP; fig. S1B]. The first component corresponded to a rapid increase in DAPI uptake, whereas the second component increased more slowly than the first and was not due to DAPI fluorescent signal saturation because detergent-induced membrane permeabilization further increased DAPI uptake (fig. S1C). DCs derived from Panx1 knockout (Panx1−/−) mice displayed a marked reduction in the first, but not in the second, component of the response (Fig. 1, B to D, fig. S1D, and movie S1). We next performed experiments with carbenoxolone (Cbx; 50 μM), a nonselective Panx1 channel blocker (Fig. 1E) (26). Pretreatment of cells with Cbx had no further effect on ATP-induced DAPI uptake in Panx1−/− DCs or on the second component of the response (Fig. 1D and fig. S1E). Panx2 and Panx3 also form functional membrane channels in other cellular systems (27), but they were equally abundant in WT and Panx1−/− DCs, as were the connexins (Cxs) Cx43 and Cx45, which are also expressed in DCs (fig. S2, A to C) (14, 28). Pharmacological inhibition of Panx1, but not of Panx3, reduced the ATP-induced uptake of DAPI by WT DCs (fig. S2D) but did not affect the response of Panx1−/− DCs (fig. S2E), suggesting that Panx3 is not involved in this process. The use of 18β-glycyrrhetinic acid, a broad pharmacological blocker of Panx1 channels and Cx hemichannels, altered ATP-induced DAPI uptake by WT but not Panx1−/− DCs (fig. S2, B to E), ruling out any contribution of Cx hemichannels. Finally, specific inhibition of transient receptor potential cation channel type A1 channels, another large pore–forming protein detected in immune tissues (29, 30), did not affect ATP-induced DAPI uptake by WT or Panx1−/− DCs (fig. S2, F and G). Together, these data suggest that Panx1 channels are selectively required for ATP-induced membrane permeabilization in DCs.

Fig. 1 Compared to WT DCs, Panx1−/− DCs exhibit reduced ATP-induced uptake of DAPI.

(A and B) Sequence of fluorescence images of WT (A) and Panx1−/− (B) DCs in a representative DAPI uptake experiment. Times, treatments, and some representative cells are indicated. Scale bar, 100 μm. Under each panel, the DAPI fluorescence plot profile is shown. (C) Analysis of the DAPI uptake experiments shown in (A) and (B) for WT DCs (open circles) and Panx1−/− DCs (green circles). After 5 min of recording, 500 μM ATP was added. Each point corresponds to the mean ± SEM of 30 cells. (D) Analysis of DAPI uptake by WT and Panx1−/− DCs under resting conditions and after treatment with ATP in control cells and in cells pretreated with 50 μM Cbx. Only the first component is depicted. Data are means ± SEM of nine experiments for control WT DCs, four experiments for Cbx-treated WT DCs, nine experiments for Panx1−/− DCs, and four experiments for Cbx-treated Panx1−/− DCs, with at least 30 cells analyzed per experiment. Data were analyzed by Kruskal-Wallis test, followed by Dunn’s multiple comparison test. ***P < 0.001 compared to basal uptake; #P < 0.05 when comparing the indicated treatments to ATP-treated WT DCs. (E) Percentages of cells (black) that exhibited ATP-induced uptake of DAPI or Etd. Data are from nine experiments for DAPI uptake by WT and Panx1−/− DCs and six experiments for Etd uptake by WT and Panx1−/− DCs. Fluo., fluorescence; A.U., arbitrary units.

To extend these results, we analyzed the uptake of different fluorescent dyes because large pore–forming proteins show different permeability properties (16, 25, 29, 31). DCs exhibited low baseline uptake rates for DAPI, ethidium (Etd), YO-PRO-1, and propidium iodide (PI) (fig. S3A). Treatment with ATP (500 μM) stimulated the uptake of only DAPI and Etd (DAPI more so than Etd), a response observed in different number of cells (Fig. 1E and fig. S3A). As observed for DAPI uptake, Panx1−/− DCs showed reduced ATP-stimulated Etd uptake compared to that of WT DCs (fig. S3B). No substantial uptake of YO-PRO-1 or PI by DCs was observed after treatment with ATP (fig. S3A). However, bone marrow–derived macrophages (BMMs) showed effective YO-PRO-1 uptake in response to ATP (fig. S3C), consistent with a previous report (32), suggesting that macrophages and DCs exhibit different dye selectivity. These results suggest that ATP stimulates a Panx1-dependent selective increase in the membrane permeability of DCs.

Panx1-dependent signaling is required for the ATP-induced fast migration of DCs

Given previous reports suggesting a role for ATP and Panx1 channels in the locomotion of myeloid cells, such as microglia and neutrophils (12, 14, 33, 34), we investigated their possible role in DC motility. For this, we used microfabricated channels that facilitate DC migration by mimicking the confined environment of peripheral tissues and further enable the quantitative analysis of cell speed and cytoskeletal organization (3537). We treated DCs with ATP (500 μM) for a short time period (30 min to trigger Panx1 opening), carefully washed the cells, and imaged them while they migrated in the microchannels (see Fig. 2A and Materials and Methods). We found that this brief pulse of ATP was sufficient to increase the amount of DAPI uptake by WT, but not by Panx1−/−, DCs (Fig. 2B), as was observed for two-dimensional (2D) cultures (Fig. 1A), indicating that Panx1 channels are functional in confined, migrating DCs. Note that this transient pulse of ATP was also sufficient to stimulate the fast migration of WT DCs (Fig. 2C and movie S2). This response was dependent on ATP concentration (EC50, 419.1 μM ATP; fig. S4A), similar to ATP-stimulated DAPI uptake (fig. S1B). This ATP pulse did not change the abundance of Panx1 in WT DCs (fig. S4B). Similar results were obtained when DC migration was analyzed in the continuous presence of ATP (Fig. 2C).

Fig. 2 Functional Panx1 channels and ATP release are required during the ATP-stimulated migration of DCs.

(A) Protocol to evaluate the effect of ATP [pulse or continuous (Cont.)] on cell migration. (B) Left: Confocal fluorescence images of DAPI in migrating WT and Panx1−/− DCs under control conditions or treated with ATP at 1 and 15 min of recording. The average (AVG) DAPI fluorescence after 15 min in >20 cells is shown in pseudocolor from a representative experiment. Scale bar, 5 μm. Right: DAPI uptake rate in untreated and ATP-treated WT and Panx1−/− DCs. Data are means ± SEM of three experiments. Data were analyzed by Kruskal-Wallis test, followed by Dunn’s multiple comparison test. ***P < 0.001 compared to WT. (C) Migration of WT DCs was evaluated after a 30-min pulse (Pulse) or continuous exposure (Cont.) to 500 μM ATP. ns, not significant. (D) Effect of apyrase (5 U/ml) on the instantaneous speed of WT DCs. (E) Migration of Panx1−/− DCs under the same conditions described in (C). In (C) to (E), the bars show 90% of the points and the median from one experiment that is representative of three experiments, with at least 100 cells analyzed per condition. The horizontal dashed line denotes the median in untreated DCs. Data were analyzed by Kruskal-Wallis test, followed by Dunn’s multiple comparison test. ***P < 0.001 and **P < 0.01 when compared to resting conditions. (F) ATP measurement in the culture medium of WT and Panx1−/− DCs under resting conditions or after an ATP pulse. Data are means ± SEM of three experiments. Data were analyzed by Kruskal-Wallis test, followed by Dunn’s multiple comparison test. *P < 0.05 when compared to WT cells.

To evaluate the possible contribution of an ATP-mediated autocrine loop during DC motility, we used apyrase to degrade extracellular ATP. We found that apyrase (5 U/ml) did not affect the spontaneous migration of DCs (Fig. 2D). However, apyrase abolished the ATP-stimulated migration of DCs in a concentration-dependent manner (Fig. 2D and fig. S4C), suggesting that ATP-induced fast DC migration requires ATP release and sustained autocrine purinergic signaling. Panx1−/− DCs did not substantially increase their speed of migration in response to a pulse of ATP (Fig. 2E, fig. S4D, and movie S2), although their speed of spontaneous migration was similar to that of WT DCs. This migration defect of Panx1−/− DCs was reverted when exogenous ATP was maintained during the entire duration of the experiment (Fig. 2E). These results suggest that the initial pulse of ATP triggered the release of intracellular ATP through Panx1 channels and that the released ATP likely acted in an autocrine manner to sustain fast migration. Consistent with this idea, we found that the pulse of ATP induced further ATP release by WT, but not by Panx1−/−, DCs (Fig. 2F). We conclude that Panx1 channels control ATP-induced DC migration by releasing intracellular ATP.

ATP-induced membrane permeabilization and migration depend on P2X7 receptor activation

P2 receptors have different affinities for nucleotides (10). The concentration range at which ATP induced membrane permeabilization in DCs and their fast migration is compatible with the involvement of P2X7 receptors. A study suggested that there is a correlation between P2X7 receptor abundance and the migration of cancer cells (38). Consistent with this finding, we found that the abundance of P2X7 receptors was increased in ATP-pulsed WT, but not in ATP-pulsed Panx1−/−, DCs (Fig. 3A), suggesting that the P2X7 receptor contributes to fast DC migration. Accordingly, treatment of DCs with oxidized ATP, a P2X7 receptor antagonist, reduced ATP-induced fast migration (fig. S5A). In addition, A-740003, a selective antagonist of the P2X7 receptor (39), prevented ATP-induced fast migration and DAPI uptake, without affecting the spontaneous migration speed or the baseline DAPI uptake of DCs (Fig. 3, B to E). Moreover, a pulse of BzATP [2′-3′-O-(4-benzoylbenzoyl)-ATP], a nonhydrolyzable P2X7 receptor agonist, increased the migration speed of WT DCs, a response that was abolished upon apyrase treatment (Fig. 3C). In contrast, inhibition of P2X7 receptors did not alter the speed of Panx1−/− DCs (Fig. 3B).

Fig. 3 P2X7 receptor activation is required for the ATP-induced migration of DCs and membrane permeabilization.

(A) Left: Western blotting analysis of the relative abundance of P2X7 receptor in total homogenates of WT and Panx1−/− DCs under resting conditions or 12 hours after an ATP pulse. Vinculin was used as loading control. Right: Normalized P2X7 receptor abundance expressed as a percentage of the abundance of the corresponding control cells (Ctrl). Data are means ± SEM of four experiments and were analyzed by unpaired t test. *P < 0.05. (B) Effect of 10 μM A-740003 (A74) on the instantaneous speed of untreated and ATP-treated WT and Panx1−/− DCs. (C) Effects of a pulse of 200 μM BzATP and treatment with apyrase (5 U/ml) on the instantaneous speed of WT DCs. For (B) and (C), the bars show 90% of the points and the median from one experiment that is representative of three experiments, with at least 100 cells analyzed per condition. The horizontal dashed line denotes the median in untreated WT DCs. Data were analyzed by Kruskal-Wallis test, followed by Dunn’s multiple comparison test. ***P < 0.001 when compared to the untreated condition. #P < 0.05 between the indicated treatments. (D) Left: Effect of preincubation with 10 μM A74 on DAPI uptake by WT and Panx1−/− DCs. Data are means ± SEM of 30 different cells. Right: Representative DAPI fluorescence plot profiles are shown for the indicated treatments. (E) Effect of 10 μM A74 on DAPI uptake by WT and Panx1−/− DCs. Only the first component is depicted. Data are means ± SEM of three experiments. Data were analyzed by Kruskal-Wallis test, followed by Dunn’s multiple comparison test. ***P < 0.001 when compared to the corresponding basal uptake. #P < 0.05 between the indicated treatments.

In contrast to the inhibition of P2X7 receptors, the inhibition of other P2X and P2Y receptors did not affect the migratory response of DCs to ATP (fig. S5A). Accordingly, activation of P2Y receptors with adenosine-5′-diphosphate did not induce membrane permeabilization (fig. S5B), although it effectively induced a Ca2+ response in WT and Panx1−/− DCs (fig. S5, C and D). Together, these data suggest that P2X7 is the main purinergic receptor involved in the Panx1-dependent signaling loop that is needed for ATP-induced membrane permeabilization and fast migration. This effect seemed to be specific to the P2X7 receptor because the P2Y receptors did not contribute to ATP-induced motility or membrane permeability in DCs. Note that although prolonged P2X7 receptor stimulation eventually leads to cell death (18), no cell death was detected in the concentration range used in this study, as evaluated by measuring the cell incorporation of EthD-1, a dye that does not permeate Panx1 channels (fig. S6).

Ca2+ influx is required for the ATP-stimulated fast migration of DCs

P2 receptor stimulation leads to an increase in cytosolic Ca2+ concentration, which stimulates cell migration (4044). As predicted, treatment of WT DCs with extracellular BAPTA (Glycine, N,N′-[1,2-ethanediylbis(oxy-2,1-phenylene)]bis[N-(carboxymethyl)]), a Ca2+ chelator, reduced the ATP-stimulated fast migration such that the cells migrated at a speed similar to that of ATP-treated Panx1−/− DCs (Fig. 4A). Consistent with this result and with the role of Ca2+ signaling in cell migration, inhibition of Ca2+/calmodulin-dependent protein kinase type II (CaMKII) with KN-62 abolished the ATP-induced migration of WT DCs (Fig. 4A). The abundance of CaMKII was similar in WT and Panx1−/− DCs, suggesting that the cell migration defect observed in Panx1−/− DCs was not due to the altered abundance of this signaling protein (fig. S7A). Note that BAPTA had no substantial effect on the spontaneous migration of WT DCs or on the migration of Panx1−/− DCs (fig. S7B). ATP elicited a similar biphasic Ca2+ response in WT and Panx1−/− DCs (Fig. 4B). Detailed analysis of this response showed no differences between WT and Panx1−/− DCs in the baseline, peak, or plateau phases (Fig. 4C). Removal of extracellular Ca2+ or inhibition of P2X7 receptors prevented Ca2+ influx, and only the initial peak was detected (fig. S7, C and D), suggesting that the P2X7 receptor–dependent entry of extracellular Ca2+ was needed for ATP-induced fast migration and that Panx1 channels were not permeable to Ca2+.

Fig. 4 Panx1-independent Ca2+ influx is required for the ATP-induced migration of DCs.

(A) Effect of extracellular Ca2+ chelation with 2 mM BAPTA and CaMKII inhibition with 10 μM KN-62 (KN) on the migration of untreated and ATP-treated WT and Panx1−/− DCs. The bars show 90% of the points, and the line corresponds to the median from one experiment that is representative of three independent experiments, with at least 100 cells analyzed per condition. The dashed line denotes the median instantaneous speed of WT DCs under resting conditions. Data were analyzed by Kruskal-Wallis test, followed by Dunn’s multiple comparison test. ***P < 0.001 compared to the control condition. ###P < 0.001, ##P < 0.01, and #P < 0.05 when compared to ATP-treated WT cells. (B) Top: Sequence of fluorescence images of WT and Panx1−/− DCs loaded with Fura-2 at indicated times after treatment with 500 μM ATP. Scale bar, 20 μm. Bottom: Ca2+ signal traces in WT and Panx1−/− DCs before and after ATP treatment. Each point corresponds to the means ± SEM of 40 different cells in one experiment that is representative of three experiments. (C) Maximal Ca2+ signal at baseline (1 min; blue circle), at the peak (3 min; red circle), and at the plateau (9 min; green circle). Data are means ± SEM of 40 different cells from one experiment that is representative of three independent experiments. Data were analyzed by Mann-Whitney test. (D) Confocal images showing the intracellular distribution of Panx1 as evaluated by immunofluorescence in WT and Panx1−/− DCs under control conditions or after ATP treatment, as indicated. Scale bar, 5 μm. Images are representative of three independent experiments.

In neutrophils, Panx1 polarizes toward the leading edge of the cells during migration, where it might contribute to the P2-dependent increase in the intracellular free Ca2+ concentration (12). However, in WT DCs, Panx1 showed no changes in its intracellular distribution in response to ATP (Fig. 4D). Similarly, Cx43, Panx2, and Panx3 did not exhibit changes in polarization in WT DCs in response to ATP (fig. S7E). Moreover, specific blockade of Cx43 hemichannels with a mimetic peptide (Gap26) (26) did not affect the ATP-induced migration of WT DCs (fig. S7F), whereas blockade of Panx1 channels (with the 10Panx1 mimetic peptide) (16) inhibited migration (fig. S7F). Thus, these data suggest that the activity of Panx1 channels, but not of the hemichannel-forming protein Cx43, was required for the ATP-induced fast migration of DCs. We conclude that, in response to extracellular ATP and downstream of P2X7 receptors, intracellular ATP released through Panx1 channels increases DC migration by stimulating Ca2+ signaling.

Panx1-dependent signaling controls the organization of the actin cytoskeleton

We previously showed that similar to ATP, LPS stimulates the fast migration of DCs by inducing their maturation (45). This response correlates with the appearance of polymerized actin (F-actin) at the rear of the migrating DCs (45). We therefore investigated whether ATP sensing was associated with dynamic changes in the actin cytoskeleton of migrating DCs in experiments with cells derived from LifeAct-GFP (green fluorescent protein) transgenic mice (46). The LifeAct peptide binds to F-actin without impairing its nucleation, enabling the monitoring of F-actin dynamics when fused to GFP. We found that a pulse of ATP was sufficient to stimulate substantial remodeling of the actin cytoskeleton of WT DCs, as revealed by the appearance of an F-actin–enriched structure at the rear of the cells (Fig. 5, A and B, and movies S3 and S4). These actin filaments were localized at the cell cortex (Fig. 5A), as was previously observed during fast DC migration (45). No differences in F-actin distribution were found in the middle or top planes of control or ATP-treated DCs (Fig. 5A). Unbiased analysis of the mean behavior of the entire cell population with LifeAct-GFP density maps showed that both the front-to-back F-actin ratio and the time spent by cells with F-actin concentrated at the front were decreased in ATP-treated DCs (Fig. 5B and fig. S8A). These data suggest that ATP induces fast DC migration through reorganization of the actin cytoskeleton at the cell rear.

Fig. 5 ATP stimulates reorganization of the actin cytoskeleton to enable the fast migration of DCs.

(A) Different confocal images of LifeAct-GFP showing the indicated planes of F-actin in untreated (Control) and ATP-pulsed (500 μM) DCs. Scale bars, 5 μm. (B) Left: Density maps of LifeAct-GFP at the cortical plane showing the average distribution in control and ATP-treated DCs. Right: Analysis of the fraction of the time that LifeAct-GFP spent at the first third of the cell (front) in LifeAct-GFP DCs. Data are means ± SEM of three experiments, with at least 30 cells analyzed per condition. Data were analyzed by Mann-Whitney test. ***P < 0.001. (C) Actin density maps (left) and graph (right) showing the effects of apyrase (Apy; 5 U/ml), 10Panx1 (200 μM), A74 (10 μM), and KN (10 μM) on the mean fraction of time that LifeAct-GFP spent at the front in control (Ctrl) and ATP-treated DCs. Data are means ± SEM of three experiments and were analyzed by Kruskal-Wallis test, followed by Dunn’s multiple comparison test. ***P < 0.001 compared to Ctrl.

Next, we evaluated the role of the Panx1 channel– and P2X7 receptor–dependent signaling in actin cytoskeletal reorganization. We found that the degradation of extracellular ATP with apyrase or the inhibition of Panx1 channels, P2X7 receptors, or CaMKII prevented the appearance of the predominant F-actin pool at the rear of ATP-pulsed DCs (Fig. 5C) but had no effect on F-actin distribution in untreated DCs (fig. S8B). Together, these data suggest that the Panx1 channel– and P2X7 receptor–dependent autocrine loop induced by ATP increases the speed of DCs, at least in part, by promoting the reorganization of the actin cytoskeleton.

Panx1 is required for DC homing to lymph nodes, but not for DC maturation

Long-term exposure of DCs to ATP increases the abundance of costimulatory molecules associated with DC maturation (11). We found that a transient pulse of ATP was sufficient to induce DC maturation in a time-dependent manner (fig. S9A), supporting the role of ATP as a danger signal. The maturation of DCs increases their migration within tissues before undergoing chemotactic-driven migration to lymph nodes (1, 45). To evaluate the migration of DCs in a 3D confined environment, we used collagen gels. In agreement with data obtained from our experiments with microchannels, we found that a short pulse of ATP was sufficient to increase the migration of WT, but not of Panx1−/−, DCs (Fig. 6, A and B, and movie S5). Accordingly, the homing of Panx1−/− DCs from their injection site in the mouse footpad to the popliteal lymph nodes was impaired (Fig. 6C and fig. S9B). This defect was not due to a decrease in the surface expression of the CCR7 (the receptor for CCL21) or to impaired chemokine gradient detection (fig. S9, C and D). These data suggest that ATP-pulsed DCs derived from Panx1−/− mice harbor a defect in their intrinsic capacity to migrate. Accordingly, we found that LPS- and ATP-induced maturation was similar in WT and Panx1−/− DCs and that ATP-induced maturation was not affected by P2X7 receptor inhibition (Fig. 6D and fig. S9E). Together, these data suggest that signaling dependent on Panx1 channels and P2X7 receptors controls the ability of DCs to migrate and reach lymph nodes in response to ATP without compromising their maturation.

Fig. 6 Panx1 contributes to the 3D migration of DCs and their homing to lymph nodes but not to their ATP-induced maturation.

(A) Scheme showing the 3D migration of DCs confined in collagen gels. Cell tracks (>150 cells each condition from one experiment that is representative of three experiments) of WT and Panx1−/− DCs migrating in collagen gels under resting conditions or after ATP treatment. Cells were imaged for 6 hours. Scale bars, 50 μm. (B) Cell displacements under the conditions shown in (A). Data are means ± SEM of three experiments and were analyzed by Kruskal-Wallis test, followed by Dunn’s multiple comparison test. ***P < 0.001. (C) Left: Flow cytometric analysis showing the presence of WT [5-(and-6)-{[(4-chloromethyl)benzoyl]amino}tetramethylrhodamine (CMTMR), red] and Panx1−/− [5-(and-6)-carboxyfluorescein diacetate succinimidyl ester (CFSE), green] DCs in the popliteal lymph nodes of mice 16 hours after these cells were injected in the footpad. Positive populations are depicted. Data are from one experiment and are representative of eight experiments. Right: Homing index (see Materials and Methods) indicating the ratio of Panx1−/− DCs compared to WT DCs that migrated to the lymph nodes. Data are means ± SEM of eight experiments. (D) Flow cytometric analysis of the cell surface abundance of CD86 in WT (left) and Panx1−/− (right) DCs under the indicated conditions. The dotted lines indicate the gate for positive cells. Data are from one experiment and are representative of three independent experiments.

DISCUSSION

DCs not only play a pivotal role during the onset of the immune response against pathogens but also respond to endogenous danger signals (2, 3), such as extracellular ATP and other DAMPs that increase in abundance under pathological conditions (79). In addition, autocrine signaling mediated by ATP released into the extracellular microenvironment from migrating immune cells has been previously reported (12). This ATP release has been linked to the opening of membrane channels, such as Panx1 channels (12, 14), but their expression and role in DC migration have not been described. Here, we showed that ATP stimulated the fast migration of DCs through Panx1 channel– and P2X7 receptor–dependent signaling, likely through an autocrine loop. This signaling was needed for the appropriate reorganization of the actin cytoskeleton that was associated with fast DC migration (Fig. 7).

Fig. 7 Proposed model for the contribution of Panx1 channels and P2X7 receptors to the migration of DCs.

Under resting conditions (top), the Panx1 channels and P2X7 receptors exhibit low levels of activity, but these are increased after the concentration of extracellular ATP is increased (bottom). The opening of both Panx1 channels and P2X7 receptors leads to the activation of the DCs, which then release ATP through Panx1 channels to further activate P2X7 receptors. P2X7 receptor–mediated Ca2+ signaling modifies the actin cytoskeleton, thus increasing the speed of migration of the DCs.

The exposure of DCs to extracellular ATP induced a selective increase in plasma membrane permeability to different dyes and the influx of Ca2+. Nevertheless, the latter occurred independently of Panx1 channels. This is consistent with reports that challenge the idea that Panx1 acts as a Ca2+-permeable channel (4752). These observations suggest a functional separation of Panx1 channels from Cx hemichannels, which are also expressed in DCs (14). However, Cx hemichannels present different permeability to dyes (25) and are Ca2+-permeable (5357). Because P2X7 receptor–mediated membrane permeabilization enables the uptake of DAPI, Etd, YO-PRO-1, or PI depending on the cell type (31, 58), we propose that Panx1 is permeable only to small molecules (those <400 Da). The lack of Panx1 did not completely abrogate ATP-dependent dye uptake, which might be explained by the presence of a cryptic protein that contributes to DAPI uptake or by the P2X7 receptor itself that enables DAPI uptake, as was previously proposed (31, 59).

Extracellular ATP is a well-known inducer of the chemotaxis of immune cells (4, 5), although its effect is transient (60) and might be cell-dependent (61). In neutrophils and microglia, Panx1 channels contribute to cell migration during chemotaxis (33, 6264). Here, we showed that a transient pulse of ATP, which is a DAMP, was sufficient to induce DC activation and sustained the fast migration of DCs in the absence of an ATP gradient. These data reveal the dual effects of ATP on DC migration, which stimulates chemotaxis early upon its release into the microenvironment and later induces fast migration. A study showed the differences between DAMP-induced migration and migration induced by a PAMP (60), suggesting that these different stimuli have specific effects. However, we found that the late effect of ATP was associated with the reorganization of the actin cytoskeleton, as was previously observed with a PAMP stimulus, such as LPS (45), which suggests that there are common downstream effectors for different stimuli (45) and in different cells (65).

The experiments showing that the degradation of extracellular ATP impaired the ATP-induced fast migration of DCs suggests the existence of an autocrine loop in which the stimulation of DCs with ATP induces the release of ATP through Panx1 channels. This loop is reminiscent of that described in neutrophils, macrophages, and microglia (13, 33, 34), although the pathway used for ATP release might depend on the cell type and the stimulus (61). However, the ATP-induced ATP release in other cellular systems occurs over a shorter time scale (66) compared to that in our observations. We thus hypothesize that the initial ATP pulse triggers an increase in cytosolic Ca2+ that leads to the activation of Ca2+-decoding proteins, such as CaMKII (43), which sustain fast DC migration. CaMKII might generate a long-lasting effect on DC locomotion either through the direct regulation of the actin cytoskeleton (67, 68) or indirectly by activating the myosin II light chain kinase (43). Non-muscle myosin IIA and RhoA, which are required for fast DC migration (35, 36, 45), are thought to be needed to sustain cell polarity in migrating neutrophils (69).

Note that we cannot exclude the idea that additional mechanisms might be at work. For example, Panx1 might directly modify cytoskeletal proteins (70), including actin (71) and Arp3 (72), a subunit of Arp2/3 complex, which controls actin nucleation and DC migration (45). Furthermore, we cannot rule out the possibility that ATP degradation products, such as adenosine, might also affect DC migration because their downstream signaling pathways also increase the cytosolic Ca2+ concentration and modify the actin cytoskeleton (73). Alternatively, intracellular signaling through RhoA (74), non-muscle myosin IIA (75), or CaMKII activation (76) might control the opening of Panx1 channels or activation of the P2X7 receptor. In this context, Ca2+ signaling triggered upon the pulse of ATP might generate cross-talk between these membrane channels and the cytoskeleton (70, 77). This would maintain the activity of these channels and the polarization of the cytoskeleton, respectively, enabling sustained fast DC migration.

DCs link innate immunity to adaptive immunity mainly through their migratory capacity (1). ATP contributes to this link because immediately after it is detected, it stimulates chemotaxis (78) but with a slower kinetics. ATP induces DC maturation and the switch to fast migration, which enables DCs to reach the lymph nodes. This long-term effect of ATP on DC migration might play a pivotal role under pathological conditions in which the extracellular ATP concentration is increased (79). Whether other DAMPs also induce the fast migrations of DCs or other immune cell types through the mechanisms described here shall now be addressed.

MATERIALS AND METHODS

DC culture

Mouse bone marrow cells from precursor cells in the femur and tibia were obtained from WT mice, Panx1−/− C57BL/6 mice (79), or LifeAct-GFP transgenic mice (46). After being flushed out of the bone, the precursors were seeded in nonadherent plastic dishes (Greiner Bio-One) and cultured for 11 days in Iscove’s modified Dulbecco’s medium (IMDM) supplemented with 10% fetal bovine serum and granulocyte-macrophage colony-stimulating factor–containing cultured medium obtained from transfected J558 cells to generate mouse bone marrow–derived DCs. For passages at days 4 and 7, the cells were detached with 1 mM EDTA in phosphate-buffered saline (PBS) for 5 min at room temperature. Panx1−/− C57BL/6 and LifeAct-GFP mice were provided by H. Monyer (University of Heildelberg, Germany) and M. Sixt (Institute of Science and Technology, Austria), respectively.

BMM culture

Mouse bone marrow cells derived from femur and tibia precursors were obtained from WT and Panx1−/− C57BL/6 mice and cultured in IMDM supplemented with macrophage colony-stimulating factor (40 ng/ml). The medium was replaced at days 3 and 5. The differentiated BMMs were used at day 7 at the same cell density and conditions as for DCs.

Time-lapse fluorescence experiments

DCs (8 × 105 cells) were seeded on coverslips or in glass dishes and exposed to Locke-HEPES solution or culture medium containing 5 μM DAPI (Life Technologies) or one of the following dyes: Etd bromide, YO-PRO-1, or PI. Fluorescence was recorded every 30 s (exposure time, 30 ms; gain, 0.5) with a 40× objective lens in a water immersion Olympus BX51WI upright microscope equipped with a QImaging model Retiga 13001 fast-cooled monochromatic digital camera (12-bit; QImaging). Metafluor software (version 6.2R5; Universal Imaging) was used for image analysis and fluorescence quantification (80). Cells were preincubated with individual blockers for 10 to 20 min, as appropriate. Briefly, two independent background fluorescence (BF) intensity measurements at each time were averaged and subtracted from the fluorescence intensity of each cell at each time interval (CF). The results of this calculation (CF-BF) at each time interval for each of the 30 cells were averaged and plotted against time (expressed in minutes). Dye uptake rates were calculated with Microsoft Excel software and expressed as arbitrary units per minute. For Ca2+ experiments, DCs seeded as previously described were loaded with 5 μM Fura-2 AM in IMDM without serum for 30 min at 37°C, washed, and then analyzed as previously described (81). To evaluate the uptake of DAPI during migration, DCs were seeded as described for the microchannel migration experiments. Once the cells were inside the microchannels, 5 μM DAPI was added to the bath solution 1 min before the recording started. Fluorescent images were acquired every 5 min with a 20× objective lens. The data were analyzed with Fiji ImageJ software (82), and the dye uptake rate was calculated as previously described.

Migration experiments in microchannels

Microchannels were prepared as previously described (45). Briefly, the microfluidic device (dimensions, 8 μm × 5 μm) was fabricated in polydimethylsiloxane with rapid prototyping and soft lithography. The surface was then coated with bovine plasma fibronectin (10 μg/ml; Sigma-Aldrich) for 1 hour and then washed several times with PBS before it was seeded with untreated (control) or ATP-pulsed DCs in culture medium. DCs (1 × 106 cells/ml) were treated with 500 μM ATP for 30 min at 37°C and then washed three times with fresh medium. The drugs (A-740003, BAPTA, KN-62, or apyrase) were added after confirmation that DCs were inside the microchannels. Phase-contrast images were recorded from multiple fields for 12 hours with a time frame acquisition of 2 min using an epifluorescence video-microscope Nikon TiE microscope equipped with a cooled charge-coupled device camera (HQ2, Photometrics) with a 10× objective lens equipped with an environmental chamber for temperature, humidity, and CO2. For kymograph extraction and analysis, we used a program written in C++ and custom routines in MATLAB, as previously described (83).

Measurement of extracellular ATP

DCs pulsed with ATP were cultured overnight at a density of 1 × 106 cells/ml. Cell culture medium was then collected, and two successive centrifugation steps were performed to ensure the absence of cells. Supernatant (100 μl) was then collected to measure the concentration of ATP with the luciferin-luciferase bioluminescent assay kit (Sigma-Aldrich) according to the manufacturer’s instructions. Standard curves were generated with defined ATP concentrations.

Immunofluorescence analysis

DCs were allowed to migrate for 12 hours in microchannels that were adhered to coverslips, and then, the cells were fixed with 4% paraformaldehyde for 30 min at 37°C. The microchannels were removed, and the cells that were attached to the coverslip were incubated with blocking solution [PBS, 1% bovine serum albumin (BSA), and 0.05% saponin] for 2 hours and then incubated with Fc block (anti-mouse CD16/CD32, eBioscience) for 1 hour at room temperature. Primary antibodies [rabbit anti-Panx1, custom-made (51); goat anti-P2X7 receptors, Santa Cruz Biotechnology] were incubated overnight. F-actin was then stained with phalloidin, and secondary antibodies were added for 2 hours. Finally, slides were mounted with DAPI Fluoromount-G (SouthernBiotech). Images were acquired with an inverted Spinning Disk Confocal Roper/Nikon microscope with a 100× oil immersion objective lens (numerical aperture, 1.4).

Flow cytometry analysis

Untreated or ATP-pulsed DCs were seeded in 12-well dishes overnight before the cells were detached as previously described. The cell suspension was stained with a mixture of appropriate antibodies (against CD11c and CD40 or CD86; BD Pharmingen). To analyze the cell surface abundance of CCR7, we used a mouse CCL19-Fc (eBioscience) and a secondary anti-Fc antibody (Life Technologies). Finally, the cell suspension was analyzed with a BD Accuri C6 flow cytometer (BD Biosciences), and data were obtained and analyzed with FlowJo v10 software (BD Biosciences).

Migration in collagen gels

The migration experiments in collagen gels were performed as previously described (45). Briefly, a mix of 1.6% collagen type I (Advanced BioMatrix) and IMDM medium was prepared at 4°C and basic pH. WT or Panx1−/− DCs (5 × 105 cells/ml) that were or were not pulsed with ATP were then added to the collagen mixture, which was deposited under a 12-mm coverslip and incubated at 37°C for 20 min to enable collagen polymerization. Images were then recorded as was previously described for the microchannel migration experiments. The cell trajectories were analyzed with the TrackMate v2.8.1 plugin from Fiji ImageJ software (82). For chemotaxis experiments, CCL21 (200 ng/ml) was added to the bath solution.

Lymph node homing

The labeling of DCs was performed as previously described (83). Briefly, ATP-pulsed WT or Panx1−/− DCs (1 × 107 cells/ml) were loaded for 10 min at room temperature with 5 μM CMTMR or CFSE, respectively. The cells were then washed several times with DC medium, and the WT and Panx1−/− DCs were mixed in a cell suspension at a 1:1 ratio containing about 2 × 106 cells of each cell type. An aliquot (20 μl) of the mixed cell suspension was injected into the footpads of recipient C57BL/6 mice. Sixteen hours later, popliteal lymph nodes were collected, dissected, and then digested with collagenase D (1 mg/ml; Roche) for 30 min. The tissue was then disrupted and filtered, and the collected cells were washed twice with PBS and 2% BSA before being analyzed by flow cytometry.

Western blotting analysis

Confluent DC cultures were gently rinsed twice with cold PBS (pH 7.4) at 4°C and harvested by scraping with a rubber policeman in a solution containing 5 mM EDTA, Halt, and M-PER protein extraction cocktail, which was prepared according to the manufacturer’s instructions. The cellular suspension was lysed in a Microson ultrasonic cell disruptor (Heat Systems-Ultrasonics) on ice. Proteins were measured in aliquots of cell lysates using the Bio-Rad protein assay. Aliquots of cell lysates (50 μg of protein) were resuspended in Laemmli sample buffer, resolved by 8% SDS–polyacrylamide gel electrophoresis, and electrotransferred to nitrocellulose sheets. Equivalent loading of samples was confirmed by protein staining with Ponceau S red [2% (w/v) in 30% trichloroacetic acid] and detection of actin. Nonspecific protein binding was blocked by incubation of nitrocellulose sheets in PBS-BLOTTO (5% nonfat milk in PBS) for 1 hour at room temperature before overnight incubation with the appropriate primary antibodies at 4°C. After several washes with PBS, blots were incubated with the secondary antibody conjugated to horseradish peroxidase for 45 min at room temperature. Immunoreactivity was detected by enhanced chemiluminescence using the SuperSignal kit according to the manufacturer’s instructions.

Cell death quantification

DCs cultured on glass coverslips were used. Dead cells were identified with the LIVE/DEAD viability/cytotoxicity assay kit according to the manufacturer’s instructions and as previously described (51). Briefly, cells were incubated for 10 min at 37°C with 1 μM calcein-AM and then were carefully washed several times to remove any remaining calcein. The cells were then incubated with 10 μM Etd homodimer-1 (EthD-1), and treatment was performed at room temperature. Cells were preincubated with blockers for a few minutes before extracellular ATP was added. Treated cells were left at 37°C in the dark for different times according to the stimulation. Finally, labeled cells were observed with a fluorescence microscope (BX51WI, Olympus). The fluorescence emissions were acquired separately (calcein at 530 nm and EthD-1 at 645 nm). Cells were analyzed using the ImageJ cell counter (National Institute of Mental Health).

Statistical analysis

All values for the dye uptake and Ca2+ signaling experiments are expressed as means ± SEM. For migration experiments, the mean instantaneous speed was expressed as a median with the 10th to 90th percentile. Most analyses and graphs were performed and prepared with Microsoft Office Excel Professional Plus 2010 (Microsoft) and Prism 5.0 software (GraphPad Prism Software Inc.). Normality was tested with Kolmogorov-Smirnov test. Statistical analysis was performed using nonparametric Mann-Whitney test and Kruskal-Wallis test, followed by Dunn’s multiple comparison test in case of significance. Differences between groups were considered statistically significant when P < 0.05.

SUPPLEMENTARY MATERIALS

www.sciencesignaling.org/cgi/content/full/10/506/eaah7107/DC1

Fig. S1. Extracellular ATP increases the permeability of the plasma membrane in DCs to DAPI.

Fig. S2. The abundances of Panx2, Panx3, Cx43, and Cx45 are similar between WT and Panx1−/− DCs.

Fig. S3. ATP-induced dye uptake by DCs is selective.

Fig. S4. Extracellular ATP contributes to migration of DCs.

Fig. S5. WT and Panx1−/− DCs have functional P2Y receptors that do not contribute to ATP-induced migration or membrane permeabilization.

Fig. S6. ATP stimulates the migration, but not the death, of DCs.

Fig. S7. Role of extracellular Ca2+ and Cx hemichannels in DC migration.

Fig. S8. ATP-induced reorganization of the actin cytoskeleton.

Fig. S9. The ATP-induced cell surface expression of CD40 and CD86 is Panx-independent.

Movie S1. WT and Panx1−/− DCs in a representative DAPI uptake experiment.

Movie S2. WT and Panx1−/− DCs migrating inside microchannels under resting conditions or after exposure to ATP.

Movie S3. WT LifeAct-GFP DCs migrating inside microchannels.

Movie S4. ATP-pulsed WT LifeAct-GFP DCs migrating inside microchannels.

Movie S5. ATP-pulsed WT and Panx1−/− DCs migrating randomly in collagen gels.

REFERENCES AND NOTES

Acknowledgments: We acknowledge the Nikon Imaging Center@CNRS-Institut Curie and PICT-IBiSA, Institut Curie, Paris (a member of the France-BioImaging national research infrastructure) for the support in image acquisition. We also acknowledge P. Fernández and T. Vergara for their excellent technical assistance and H. Monyer and M. Sixt for providing the Panx1 knockout and LifeAct-GFP mice, respectively. Funding: This work was funded by Comisión Nacional de Investigación Científica (CONICYT) y Tecnológica 24100062 and European Molecular Biology Organization (EMBO) ASTF 458-2010 (to P.J.S.), European Research Council ERC-Strapacemi-GA 243103 (to A.-M.L.-D.), and Fondo Nacional de Desarrollo Científico y Tecnológico 1150291, Chilean Science Millennium Institute Grant P09-022-F (to J.C.S.). Author contributions: P.J.S. designed, performed, and analyzed most experiments, conceived the project, and drafted the manuscript. P.V. participated in article drafting, the setup, and assistance with the microchannels, collagen, and in vivo experiments. K.F.S. genotyped the mice and assisted with dye uptake and immunofluorescence experiments. P.A.H. assisted with the cell culture and the dye uptake experiments. A.-M.L.-D. drafted the manuscript. J.C.S. designed the experiments, conceived the project, and drafted the manuscript. Competing interests: The authors declare that they have no competing interests.
View Abstract

Navigate This Article