Research ArticleCell Migration

Autocrine Purinergic Receptor Signaling Is Essential for Macrophage Chemotaxis

See allHide authors and affiliations

Science Signaling  27 Jul 2010:
Vol. 3, Issue 132, pp. ra55
DOI: 10.1126/scisignal.2000588


Chemotaxis, the movement of cells along chemical gradients, is critical for the recruitment of immune cells to sites of inflammation; however, how cells navigate in chemotactic gradients is poorly understood. Here, we show that macrophages navigate in a gradient of the chemoattractant C5a through the release of adenosine triphosphate (ATP) and autocrine “purinergic feedback loops” that involve receptors for ATP (P2Y2), adenosine diphosphate (ADP) (P2Y12), and adenosine (A2a, A2b, and A3). Whereas macrophages from mice deficient in pannexin-1 (which is part of a putative ATP release pathway), P2Y2, or P2Y12 exhibited efficient chemotactic navigation, chemotaxis was blocked by apyrase, which degrades ATP and ADP, and by the inhibition of multiple purinergic receptors. Furthermore, apyrase impaired the recruitment of monocytes in a mouse model of C5a-induced peritonitis. In addition, we found that stimulation of P2Y2, P2Y12, or adenosine receptors induced the formation of lamellipodial membrane protrusions, causing cell spreading. We propose a model in which autocrine purinergic receptor signaling amplifies and translates chemotactic cues into directional motility.


Macrophages and neutrophils form the first line of defense against potentially invading pathogens and are recruited to sites of inflammation by gradients of chemokines (chemoattractant cytokines), complement components, or bacterial chemoattractants. Thus, motility and the ability to navigate along gradients are key functions of immune cells. Cellular movement is achieved by coordinately generating membrane protrusions (lamellipodia) at the front of the cell and contractions at the rear (1, 2), whereas chemotactic navigation requires that the dominant membrane protrusion (the leading edge) is directed toward the source of chemoattractant (3). Until recently, the molecular workings of the navigation system that underlie chemotaxis appeared to have been identified. The translation of extracellular gradients of chemoattractants into internal gradients of phosphatidylinositol 3,4,5-trisphosphate (PIP3) was assumed to be the backbone of the “compass mechanism” of migrating cells (47); that is, the local generation of PIP3, catalyzed by phosphatidylinositol 3-kinase (PI3K), was thought to act as a compass needle that points the cell toward the source of chemoattractant. However, studies have shown that the migration of neutrophils toward the archetypical chemoattractant N-formyl-methionyl-leucyl-phenylalanine (fMLP) is not dependent on PI3K activation or on the accumulation of PIP3 at the leading edge (812). Likewise, Hoeller and Kay observed that disruption of PIP3 signaling by deletion of the genes that encode PI3K and the phosphatase PTEN (phosphatase and tensin homolog deleted from chromosome 10) did not impair chemotaxis by the amoeba Dictyostelium discoideum (13). On the other hand, Kubes and colleagues elegantly demonstrated that PI3K plays a role in chemotaxis mediated by chemokines, whereas migration toward fMLP and the complement component C5a is strongly dependent on the signaling of the mitogen-activated protein kinase (MAPK) p38 (14, 15). In neutrophils that face opposing gradients of fMLP (or C5a) and chemokines, such as CXCL2 [also known as macrophage inflammatory protein–2α (MIP-2α)] or interleukin-8 (IL-8), the p38-dependent pathway dominates, providing a means by which neutrophils can prioritize among chemoattractant gradients (14, 15).

Because the role of PIP3 signaling in chemotaxis toward “end-target” chemoattractants such as bacterial and complement components has become less certain, the outstanding question of how cells amplify chemotactic cues and direct their migration becomes more pressing. Chen et al. (16) showed that fMLP induces the release of adenosine triphosphate (ATP) from neutrophils and, with micropipette chemotaxis assays, demonstrated that neutrophils deficient in the gene encoding the purinergic receptor P2Y2 have impaired gradient sensing. However, immune cells typically have multiple purinergic P2 (P2X and P2Y) receptor subtypes, and extracellularly released ATP is rapidly and sequentially hydrolyzed by ectonucleotidases [in the order ATP→ADP (adenosine diphosphate)→AMP (adenosine monophosphate)→adenosine], which gives rise to multiple signaling scenarios. P2X receptors (P2X1 through P2X7) are ATP-gated nonselective cation channels, whereas the P2Y receptors are heterotrimeric guanine nucleotide–binding protein (G protein)–coupled receptors (GPCRs), which are activated (in mouse) by ATP and uridine triphosphate (UTP) (P2Y2 and P2Y4), uridine diphosphate (UDP) (P2Y6), or ADP (P2Y1, P2Y12, and P2Y13). Four additional GPCRs are activated by adenosine: the purinergic P1 receptors A1, A2a, A2b, and A3. The extent to which purinergic signaling is involved in chemotaxis and how it functions is unclear. Here, we investigated macrophage chemotaxis in vitro and in vivo, and showed that autocrine feedback signaling through multiple purinergic receptor subtypes was a fundamental mechanism underlying the chemotactic navigation of macrophages toward C5a.


The chemotactic navigation of macrophages requires ATP signaling

We investigated the chemotaxis of macrophages in experiments with chemotaxis chambers that consist of two 40-μl reservoirs connected by a narrow channel into which the cells are seeded. The chemoattractant complement component C5a (~9 kD) was added to one of the reservoirs (at a final concentration of 10 nM) to establish a spatially well-defined chemotactic gradient. Resident peritoneal macrophages from wild-type mice migrated robustly toward the chemoattractant (Fig. 1A and movie S1); however, in the presence of the adenosine triphosphatase (ATPase) and adenosine diphosphatase (ADPase) apyrase, chemotactic navigation was virtually abolished (Fig. 1B and movie S2). The chemotaxis index, which is defined as the displacement along the y axis divided by the accumulated distance, was decreased from 0.45 ± 0.03 (n = 125 cells) to 0.02 ± 0.03 (n = 75 cells) by apyrase (Fig. 1C) without any significant effect (P < 0.05) on the mean velocity of the cells (Fig. 1D). This finding indicated that ATP signaling played a critical role in chemotactic navigation.

Fig. 1

The scavenging of extracellular nucleotides with apyrase blocks chemotaxis. Migration plots of wild-type (WT) macrophages in the absence (A) or presence (B) of apyrase (40 U/ml) obtained by tracking 25 cells between 6 and 12 hours after adding chemoattractant (complement component C5a) to one of the reservoirs of an Ibidi μ-slide chemotaxis chamber. The start point of each track was normalized to x = 0 and y = 0, and positive y-axis values represent movement in the direction of the source of chemoattractant. (C) Mean chemotaxis index of untreated WT (n = 100 cells, four independent experiments) macrophages relative to that of WT macrophages treated with apyrase (n = 75 cells, three independent experiments). *P < 0.05 by one-way ANOVA. (D) Plot of the mean velocities of untreated WT macrophages relative to those of WT macrophages treated with apyrase. Velocity data were analyzed with the Kruskal-Wallis test. (E) Release of ATP from macrophages stimulated with C5a (10 nM). ATP was detected with a luciferase-based chemiluminescence assay (representative of three independent experiments). (F) Reversed-phase HPLC analysis. Chromatograms (representative of three independent experiments) of samples injected before and after the addition of hexokinase and myokinase or of apyrase to a solution of ATP (10 μM) in PBS containing glucose.

To detect the release of ATP from the cells, we used the luciferin-luciferase chemiluminescence assay in the presence of the ecto-ATPase inhibitor ARL-67156 (100 μM) (17). From experiments with densely plated macrophages and a highly sensitive photon detection system, we found that the direct application of C5a evoked a transient light signal consistent with the release of ATP from the cells (Fig. 1E). In control time-lapse imaging experiments, the addition of C5a did not cause cell lysis. From reversed-phase high-performance liquid chromatography (HPLC) assays, we found that the potato apyrase used in our experiments did not specifically hydrolyze ATP to AMP, as would be expected for this enzyme (E.C. number Instead, the apyrase preparation degraded ATP to an unidentified product without the accumulation of either AMP or adenosine (Fig. 1F and fig. S1). Therefore, inhibition of purinergic signaling mediated by ATP, ADP, or adenosine could be implicated to explain the “apyrase effect” (Fig. 1B).

Apyrase decreases the recruitment of monocytes in vivo

Next, we tested whether apyrase blocked the chemotaxis of monocytes in vivo. Mice were intraperitoneally injected with phosphate-buffered saline (PBS), C5a, or C5a and apyrase. The peritoneum is highly permeable to proteins (18), including albumin, and gaps exist between the mesothelial cells that line the inner peritoneal surface (19), which permits the diffusion of apyrase into the submesothelial space. Initially, a rapid influx of neutrophils prevails in acute peritonitis (which peaks at ~4 hours), but later, through a steady increase, the recruitment of monocytes predominates (20, 21). We harvested peritoneal cells 16 hours after injection and analyzed them by flow cytometry (Fig. 2A). Apyrase dramatically decreased the C5a-induced accumulation of F4/80+ cells (monocytes and macrophages) in the peritoneal cavity compared to that in mice injected with C5a alone (Fig. 2B).

Fig. 2

Apyrase inhibits the recruitment of monocytes and macrophages in vivo. (A) Flow cytometric analysis of the binding of antibody against F4/80 to cells harvested from the peritoneal cavity 16 hours after intraperitoneal injection of mice with C5a alone or with apyrase. Each dot plot is representative of four independent experiments. Numbers beside the quadrants indicate the percentage of cells that have F4/80+ on the surface. (B) Mean number of total cells and of those cells positive for F4/80 that were isolated from the peritoneums of mice injected with C5a alone, with C5a and apyrase, or with PBS (n = 4 mice per group). *P < 0.05 by one-way ANOVA.

Purinergic receptor feedback loops underlie chemotaxis

The repertoire of purinergic P2X and P2Y receptors found on purified mouse peritoneal macrophages has been explored with knockout mice, functional assays, and reverse transcription polymerase chain reaction (RT-PCR) analyses (22). We extended this work and found that purified resident F4/80+ cells (macrophages) additionally had the purinergic P1 (adenosine) receptors A2a, A2b, and A3, as well as the ADP-selective P2Y12 receptor (Fig. 3A). We generated a working model (Fig. 3A) of the most likely autocrine purinergic feedback loops that could underlie the chemotactic compass mechanism. The UDP-selective receptor P2Y6 and the ADP-selective receptor P2Y1 are shown in gray because these receptors are low in abundance on freshly isolated resident macrophages (22). Moreover, P2X7 (shown in gray) is also an unlikely candidate because it is activated by unusually high (millimolar) concentrations of ATP (23). We systematically blocked one or two of the feedback loops in experiments with macrophages from wild-type, P2ry2−/−, or P2ry12−/− mice, with or without the P2Y12 inhibitor AR-C69931MX (24) or the nonselective P1 inhibitor 8-(p-sulfophenyl)theophylline (8-SPT) (Fig. 3, B and C, fig. S2, and movies S3 to S8). Genetic deletion of P2ry2 had little effect on the efficiency of chemotaxis, whereas combinations involving genetic deletion or pharmacological inhibition of P2Y12 receptors modestly impaired chemotaxis (Fig. 3D). The velocity of migration of the cells in the P2ry2−/−, P2ry12−/−, P2ry2−/− + AR-C69931MX, P2ry2−/− + 8-SPT, and P2ry12−/− + 8-SPT groups was not significantly different (P < 0.05) compared to that of cells under control conditions. Thus, inhibition of one or two of the three feedback loops that converge on P2Y2, P2Y12, or adenosine (A2a, A2b, and A3) receptors (Fig. 3A) did not explain the dramatic inhibitory effect of apyrase on chemotactic gradient sensing.

Fig. 3

Dissection of the multiple purinergic feedback loops that are implicated in chemotaxis. (A) RT-PCR analysis of P2Y and P1 receptor subtypes found in purified F4/80+ cells (macrophages), and a schematic diagram of candidate purinergic feedback loops required for gradient sensing. (B and C) Migration plots of macrophages in a gradient of complement component C5a after the blocking of one or two of the feedback loops that terminate at P2Y2, P2Y12, and adenosine (A2a, A2b, and A3) receptors by genetic deletion and specific inhibitors for P2Y12 (AR-C69931MX) and adenosine (8-SPT) receptors. (D) Summary of chemotaxis assays for cells in which single receptors (P2ry2−/− or P2ry12−/−, each n = 75 cells, three independent experiments) or double receptors were inhibited (n = 50 to 75 cells; two to three independent experiments). *P < 0.05 relative to the untreated WT control, by one-way ANOVA and post hoc Scheffé test.

We next asked whether the compass mechanism of chemotaxis was amplified by multiple redundant purinergic feedback loops. This was indeed the case, because we could reproduce the apyrase effect by blocking multiple purinergic feedback loops simultaneously. Through the use of macrophages from P2ry2−/− mice in combination with a cocktail of inhibitors of P2Y12 (AR-C69931MX), A2a and A3 (8-SPT), P2Y1 (MRS-2179), and P2X1 and P2X4 (NF449) receptors (25), chemotaxis was virtually abolished (Fig. 4A and movie S9). In further work, we found that inhibition of P2Y2 (P2ry2−/−), P2Y12 (AR-C69931MX), and adenosine (8-SPT) receptors (a condition we refer to as “triple block”) was sufficient to disorient cells in a chemotactic gradient (Fig. 4B and movie S10). In addition, inhibition of ecto-ATPase activity in P2ry2−/− macrophages with ARL-67156 impaired their chemotaxis (fig. S3). Similar to the case of apyrase, targeted inhibition of multiple purinergic receptors decreased the chemotaxis index (Fig. 4C) without having any significant effect (P < 0.05) on the mean velocity of cells (Fig. 4D). Next, we confirmed that F4/80+ cells had the surface enzymes CD39 [NTPDase1 (ectonucleoside triphosphate diphosphohydrolase 1)] and CD73 (ecto-5′-nucleotidase) (Fig. 4E), which are required to hydrolyze extracellular nucleotides (26). Thus, the chemotaxis of macrophages in a gradient of C5a required release of ATP, sequential nucleotide hydrolysis (ATP→ADP→AMP→adenosine), and at least one of three main effector positive feedback loops (Fig. 4F). The mechanism by which ATP is released from macrophages is currently unknown. Although pannexin-1 (Panx1) has emerged as a strong candidate for the ATP release pathway in nonneuronal cells (27), we found no difference in the efficiency of chemotaxis of macrophages from wild-type and Panx1−/− mice (fig. S4 and movie S11).

Fig. 4

The blocking of multiple purinergic feedback loops uncovers the workings of the chemotactic compass. (A) “Cocktail block” (P2Y2, P2Y12, adenosine, P2Y1, P2X1, and P2X4 receptors). Migration plot of P2ry2−/− macrophages in a chemotactic gradient and in the presence of specific inhibitors for the P2Y12 (AR-C69931MX), adenosine (8-SPT), P2Y1 (MRS-2179), and P2X1 and P2X4 (NF449) receptors. (B) Triple block (P2Y2, P2Y12, and adenosine receptors). Migration plot of P2ry2−/− macrophages in a chemotactic gradient and in the presence of inhibitors for the receptors P2Y12 (AR-C69931MX) and adenosine (8-SPT). (C) Summary plots comparing the efficiency of chemotaxis of untreated WT macrophages (n = 100 cells) relative to that of WT macrophages under conditions of cocktail block (n = 75 cells, three independent experiments) and triple block (n = 75 cells, three independent experiments). *P < 0.05, significantly different to control WT cells by one-way ANOVA. (D) Summary plots of the mean velocities of untreated WT cells relative to those of WT cells under cocktail block or triple block conditions. (E) Immunofluorescence detection of surface CD39 and CD73 in living macrophages (representative of two independent experiments). (F) Proposed model of the chemotactic compass of mouse macrophages. Chemoattractant gradient sensing is amplified by three major purinergic feedback loops that terminate at P2Y2, P2Y12, and adenosine receptors. Minor auxiliary loops that converge on P2Y1, P2X1, and P2X4 receptors cannot be excluded.

Purinergic feedback loops are coupled to short-term membrane dynamics

We next investigated the effects of stimulating individual purinergic receptors on cytoskeletal dynamics, as assessed by phase-contrast, time-lapse imaging. The P2Y2 agonists ATP (100 μM) and uridine 5′-O-(3-thiotriphosphate) (UTP-γ-S; 100 μM) induced rapid and marked extension of lamellipodia in wild-type cells (Fig. 5, A to C, and movie S12). We also found that the P2Y12 (and P2Y1) agonist ADP, and to a lesser extent adenosine, similarly evoked the formation of lamellipodial membrane protrusions (Fig. 5, B and C). Typically, the agonists ATP, ADP, and adenosine extended preexisting lamellipodia. Bordetella pertussis toxin (PTX) (0.5 μM) inhibited the formation of ATP-induced lamellipodial membrane protrusions (Fig. 5B), implying that mouse P2Y2 receptors, which are coupled to G proteins of the Gq family, were additionally coupled to members of the Gi/o family, as suggested by others (28). Signaling by Gi proteins is probably a functionally important common denominator of chemoattractant receptors and the chemotactic compass mechanism, and we found that purified F4/80+ cells (macrophages) contained messenger RNAs (mRNAs) for Gαi2 and Gαi3 (Fig. 5B, inset). Topographic scanning of fixed control macrophages (Fig. 5D) and ATP-stimulated macrophages (Fig. 5E) by atomic force microscopy (AFM) confirmed that the membrane protrusions apparent in phase-contrast images were <200 nm in height, consistent with the definition of lamellipodia (29). ATP failed to stimulate the formation of lamellipodia in macrophages from P2Y2−/− mice (Fig. 5F and movie S13), and adenosine 5′-O-(2-thiodiphosphate) (ADP-β-S) (or ADP) exerted no effect in the presence of the P2Y12 antagonist AR-C69931MX (fig. S5) or in macrophages from P2Y12−/− mice (Fig. 5F).

Fig. 5

Nucleotide-induced formation of lamellipodia through P2Y2, P2Y12, and adenosine receptors. (A) Phase-contrast images of macrophages before and 60 s after the application of ATP (100 μM). (B) Summary of the effects of various nucleotides on the projected (2D) surface area of a macrophage. In paired experiments, the cell surface area ~180 s after the application of nucleotide was divided by the initial area to obtain the fold increase. *P < 0.05 compared to WT control by one-way ANOVA (n > 15 cells per group). PTX was used to inhibit the α subunits of the G proteins Go and Gi. RT-PCR analysis of the Go/i α isoforms found in F4/80+ cells is shown in the inset image. (C) Dose-response curves of the effects of agonists for P2Y2 (UTP-γ-S), P2Y12 (ADP), and adenosine (Adeno) receptors on cell surface area (n > 15 cells per group). (D) 3D surface topography of a typical WT macrophage fixed and scanned (in contact mode) by AFM in PBS. (E) Surface topography of a macrophage fixed ~180 s after the application of ATP (100 μM) and imaged by AFM in contact mode. (F) Summary plot showing the lack of effects of P2Y2 and P2Y12 agonists on P2Y2- and P2Y12-deficient macrophages, respectively (n > 15 cells per group). (G) Proposed model showing how the release of ATP and autocrine stimulation of purinergic receptors amplify chemoattractant-induced cell polarity and directional motility.


Unraveling the molecular workings of chemotaxis has proven difficult, and it has been widely speculated that there must still be unidentified navigational components (10, 12). Autocrine ATP signaling is emerging as an important mechanism to control various cell functions (27), and a study identified a role for autocrine P2Y2 receptor signaling in the chemotaxis of neutrophils (16). Here, we showed that ATP release and autocrine purinergic signaling were essential for the chemotaxis of macrophages. We found that the migration of mononuclear phagocytes toward C5a in vitro and in vivo was blocked by the nucleotide-scavenging enzyme apyrase. Together with another study (22), we deduced that resident peritoneal macrophages had the receptors P2X1, P2X4, P2X7, P2Y1, P2Y2, P2Y6, P2Y12, A2a, A2b, and A3, as well as the ectonucleotidases CD39 and CD73, which give rise to at least nine possible autocrine feedback loops. The mechanism by which ATP is released from macrophages is unclear. Chemotactic navigation was not impaired in Panx1−/− macrophages, which suggests that Panx1 either is a redundant component of the ATP release pathway or is not involved in autocrine purinergic signaling.

In experiments with macrophages from P2Y2−/− or P2Y12−/− mice in combination with specific inhibitors, we found that one of three redundant feedback loops terminating on P2Y2, P2Y12, or adenosine receptors was sufficient for efficient chemotaxis. We did not further dissect the adenosine receptor feedback loop, although we speculate that Gi-coupled A3 receptors, rather than Gs-coupled A2a and A2b receptors, function in concert with Gq- and Gi-coupled P2Y2 and Gi-coupled P2Y12 receptors to amplify chemotactic signaling. In neutrophils, A3 receptors are implicated in the regulation of the velocity of migration, but the published data are conflicting. Chen et al. (16) reported that neutrophils deficient in Adora3, which encodes the adenosine A3 receptor, were slower than wild-type macrophages, whereas van der Hoeven et al. (30) found the opposite effect in experiments with cells from the same knockout mice. In the case of macrophages, we found that the nonselective adenosine receptor antagonist 8-SPT had no substantial effect on cell velocity.

We subsequently identified a mechanism by which purinergic receptor signaling may translate chemoattractant gradient sensing into directional motility. Application of ATP rapidly induced lamellipodial extensions in macrophages from wild-type, but not P2ry2−/−, mice, and time-lapse images revealed that the ATP-P2Y2 pathway preferentially extended preexisting lamellipodia rather than generating new ones. ADP and adenosine also extended preexisting lamellipodia, and the lack of effect of ADP on P2ry12−/− macrophages excluded a role for P2Y1 receptors, which are low in abundance. Our data suggest that purinergic receptors generate positive feedback signals to amplify the lamellipodial membrane responses initiated by chemoattractant receptor signaling. This “inside-out” signaling amplification mechanism would fit with chemotaxis models that assume either the presence of a chemical compass (31) or the biased maintenance of randomly generated lamellipodia (32). When confronted with diverging gradients of p38- and PI3K-dependent chemoattractants, phagocytes “choose” the p38-dependent path (14, 15). We speculate that autocrine purinergic signaling may support such hierarchical chemoattractant signaling because stimulation of P2Y2, P2Y12, or adenosine receptors activates p38 (3336).

In addition to their role in autocrine signaling, as delineated earlier, adenine nucleotides act as chemoattractants. For example, Haynes et al. (37) demonstrated that mouse microglia (brain-resident macrophages) migrate toward a pipette source of ATP inserted into a tissue slice, whereas no effect is seen in brain slices from P2ry12−/− mice. These data suggest that ADP (derived from the hydrolysis of ATP) can evoke chemotactic activity; however, chemokinesis (increased nondirectional motility in response to a chemical substance) is thought to account for the apparent chemotactic activity of ATP in experimental models with human neutrophils (16) and monocytes (38). Elliott et al. (39) reported that ATP released from apoptotic cells acts as a “find me” signal to recruit motile phagocytes such as monocytes and macrophages. The authors showed that the recruitment of cells to a deposit of apoptotic supernatant was reduced by apyrase. In light of our findings, as well as those of a related study (16), the impaired recruitment of monocytes by apyrase could be explained, at least in part, by the interruption of autocrine purinergic signaling. We suspect that ATP does not act as an efficient long-range chemoattractant in mouse macrophages, although the chemotactic activity of ATP and nonhydrolyzable analogs needs to be vigorously tested in robust two-dimensional (2D) or 3D chemotaxis assays. Instead, we speculate that ATP released by an apoptotic or necrotic cell acts as a local “touch me” signal to nearby macrophages by inducing transient cell spreading (lamellipodial membrane protrusion) through P2Y2 receptors.

In analogy to macrophage chemotaxis, autocrine purinergic receptor signaling amplifies the action of platelet activators (4042). ADP released by activated platelets induces cell shape changes and aggregation through the stimulation of P2X1, P2Y1, and P2Y12 receptors, and P2Y12 antagonists have been used clinically to prevent or treat thrombotic diseases, such as stroke and myocardial infarction (42, 43). Both chemoattractant and purinergic receptor signaling are viable targets for modulating chemotaxis and the recruitment of leukocytes. For example, inhibition of chemokine signaling is an effective approach to inhibit monocyte recruitment and treat inflammatory diseases in animal models of atherosclerosis and autoimmune encephalitis (44). Moreover, leukocyte infiltration and airway inflammation are abrogated by apyrase or nonselective P2 receptor antagonists in a mouse model of allergic asthma (45). Thus, a better insight into purinergic signaling and chemotactic navigation may lead to innovative therapies to augment host defense against tumors and infections or to blunt the unwanted recruitment of inflammatory cells.

In conclusion, the ability of single cells to sense and migrate along chemical gradients is not only a fascinating facet of biological systems but also essential for embryogenesis, wound healing, and a functional immune system (46). How cells navigate in chemical gradients is unclear. Our data suggest a model (Fig. 5G) in which the sensing of a chemoattractant by macrophages is translated to gradient navigation by the release of ATP and the autocrine stimulation of purinergic receptors coupled to the membrane protrusion machinery. In simple terms, our data, together with studies of neutrophils (16, 47), suggest that the amplification of chemotactic “outside-in” signaling by purinergic inside-out signaling may be a basic principle that underlies the chemotaxis of immune cells.

Materials and Methods


Recombinant mouse complement component C5a was obtained from R&D Systems. Apyrase (grade VII), hexokinase, myokinase, ATP, ADP, AMP, adenosine, 8-SPT, MRS-2179, d-luciferin, and luciferase were obtained from Sigma. The nucleotide analogs ATP-γ-S [adenosine 5′-O-(3-thiotriphosphate)] and UTP-γ-S [uridine 5′-O-(3-thiophosphate)] were specially prepared to >95% purity by Jena Bioscience. Alexa Fluor 488–conjugated mouse monoclonal antibody against F4/80 was obtained from Cell Signaling Technology; Alexa Fluor 647–conjugated mouse antibody against CD39 (NTPDase1) was from eBioscience; and Alexa Fluor 647–conjugated mouse antibody against CD73 (ecto-5′-nucleotidase) was from Biozol.

Knockout mice

The generation of P2ry2−/− and P2ry12−/− mice has previously been described (48, 49). Both knockout strains were backcrossed onto the C57BL/6 genetic background. Panx1−/− mice (50) were supplied by A. Herb and H. Monyer (Universität Heidelberg, Germany).

Isolation of peritoneal macrophages

Mice were killed by an overdose of isoflurane in air, and the peritoneal cavity was lavaged through a 24-gauge plastic catheter (B. Braun) with two 4-ml aliquots of ice-cold Hanks’ balanced salt solution without Ca2+ or Mg2+ (PAA). After centrifugation at 360g for 5 min, cells were resuspended in RPMI 1640 containing HCO3 (2 g/liter; Biochrom AG) and supplemented with 10% heat-inactivated fetal calf serum (FCS), penicillin (100 U/ml), and streptomycin (100 μg/ml; pH 7.4). All procedures and protocols met the guidelines for animal care and experiments in accordance with national and European (86/609/EEC) legislation.

Purification of macrophages and RT-PCR analysis

Live macrophages were labeled with Alexa Fluor 488–conjugated antibodies against F4/80, and F4/80+ cells (~30% of total cells) were isolated with a BD FACSCalibur flow cytometer (BD Biosciences), a dual cell analyzer and sorter. RNA was isolated with the RNeasy Mini kit (Qiagen) according to the manufacturer’s instructions. Complementary DNA (cDNA) was synthesized with SuperScript III RT, and after denaturation at 94°C for 5 min, PCR was performed with the following primers (product sizes shown in parentheses): P2Y12 [500 base pairs (bp)]: forward, 5′-CCGGAGACACTCATATCCTTC-3′; reverse, 5′-GCCCAGATGACAACAGAAAG-3′; P2Y13 (525 bp): forward, 5′-ATTCGTGGGTTGAGCTAGTAA-3′; reverse, 5′-ATCAGGGACCAGACGGAAAT-3′; A1 (188 bp): forward, 5′-TGGCTCTGCTTGCTATTG-3′; reverse, 5′-GGCTATCCAGGCTTGTTC-3′; A2a (163 bp): forward, 5′-TCAGCCTCCGCCTCAATG-3′; reverse, 5′-CCTTCCTGGTGCTCCTGG-3′; A2b (189 bp): forward, 5′-TTGGCATTGGATTGACTC-3′; reverse, 5′-TATGAGCAGTGGAGGAAG-3′; A3 (110 bp): forward, 5′-CGACAACACCACGGAGAC-3′; reverse, 5′-GCTTGACCACCCAGATGAC-3′ (51); Gαo (183 bp): forward, 5′-GGAGCAAGGCGATTGAGAA-3′; reverse, 5′-CAGGCTTGCTGTAGACCATGTA-3′; Gαi1 (135 bp): forward, 5′-CACGTCCATCATCCTTTTCCTC-3′; reverse, 5′-TGACATTGAATATACGCGGCC-3′; Gαi2 (201 bp): forward, 5′-GTGCTGGCTGAGGATGAGGA-3′; reverse, 5′-TGCCTCGTCGTACTTGTTGG-3′; Gαi3 (218 bp): forward, 5′-AGAAAGCGGCCAAAGAAGTG-3′; reverse, 5′-TCTGGCAGATTCCCCAAAAT-3′ (52).

Chemotaxis assays

Cells obtained by peritoneal lavage of a single mouse were resuspended in 100 to 150 μl of RPMI 1640 containing bicarbonate and 10% FCS, and 8 μl of the suspension was seeded into the narrow 1000 × 2000 × 70–μm channel of an uncoated (IbiTreat) μ-slide chemotaxis chamber (Ibidi). The narrow channel (observation area) connects two 40-μl reservoirs. In chemotaxis experiments, bicarbonate-free RPMI 1640 containing 20 mM Hepes (Biochrom AG) and 10% FCS was used. A final concentration of 10 nM C5a together with 0.003% Patentblau V (Chroma Gesellschaft), a blue dye, was added to one of the reservoirs, and cells were imaged by phase-contrast microscopy with a 10×/0.3 objective. The blue dye was used as a visual indicator of gradient formation. Images were captured every 2 min for 14 hours, and cell migration tracks between 6 and 12 hours were analyzed by ImageJ (National Institutes of Health) with a manual tracking plug-in and the chemotaxis and migration tool from Ibidi. Twenty-five randomly selected cells were manually tracked in each chemotaxis experiment.

Detection of ATP release

Macrophages were densely seeded onto the glass bottom (0.17-mm thickness) of a custom-made 200-μl Perspex well and imaged with a 20×/1.3 oil immersion objective on the stage of an inverted Zeiss AxioObserver microscope. The 50-μl reaction mixture contained d-luciferin (500 μM), recombinant firefly luciferase (2 mg/ml), and ARL-67156 (100 μM), and chemiluminescence was detected with a D104 microscope photometer and FeliX32 software (Photon Technology International). The photometer shutter was briefly closed before the chemoattractant was gently applied with a pipette.

Reversed-phase HPLC

In initial experiments, analytical HPLC was performed with an HP1050 system (Hewlett Packard), which incorporated a Bischoff C18 column (Nucleosil). The mobile phase consisted of PBS (pH 5.5), and the flow rate was 1 ml/min. Subsequently, samples were analyzed with an Agilent 1200 series HPLC system (Agilent Technologies), which included an autosampler coupled to a quaternary pump and a photodiode array detector. HPLC analysis was performed on a 2.0 × 125–mm C18 analytical column (Nucleodur C18, 3 μm). After equilibration of the column with 3% acetonitrile in 10 mM potassium phosphate (pH 5.0) and 2 mM tris-borate-EDTA (TBA) buffer and injection of the sample, ATP and its degradation products were eluted in a 50-min gradient with 50% acetonitrile in 10 mM potassium phosphate (pH 7.5) and 2 mM TBA buffer. Detection was carried out at 259.8 nm. Additionally, ultraviolet spectra (200 to 400 nm) were obtained.

Time-lapse imaging of membrane dynamics

Freshly isolated macrophages were seeded into the 100-μl channel of fibronectin-coated μ-slide I chambers, placed in a humidified incubator (at 37°C with 5% CO2), and, 2 hours later, washed at least 10 times with RPMI 1640 containing bicarbonate (2 g/liter) and 10% FCS. Experiments were performed on day 2 after overnight incubation or on day 3 in the same complete medium, except that bicarbonate buffer was replaced by 20 mM Hepes (pH 7.4). At least 20 min was allowed to equilibrate the pH of the medium. μ-Slide I chambers were placed on the stage of an inverted semimotorized Zeiss AxioObserver microscope controlled by AxioVision software. The microscope was fitted with a temperature-controlled stage incubator (Zeiss), and the temperature was maintained at 37°C. Cells were imaged with a 40× objective, and phase-contrast images were captured every 15 s by an AxioCam MRm camera (Zeiss).

Immunofluorescence imaging of CD39 and CD73

Live macrophages were colabeled with Alexa Fluor 488–conjugated antibody against F4/80 and Alexa Fluor 647–conjugated antibody against either CD39 or CD73, each of which was diluted 1:20. After a 5-min incubation at 37°C, the chamber was washed with medium, and fluorescence images were obtained with a 63×/1.4 oil immersion objective. Hepes-buffered RPMI 1640 supplemented with 10% FCS and 5% fatty acid–free bovine serum albumin was used for the immunofluorescence experiments.

In vivo model of C5a-induced monocyte infiltration

Mice were intraperitoneally injected with PBS or recombinant mouse C5a (100 ng) in the presence or absence of potato apyrase (80 U). After 16 hours, the peritoneal cavity was washed with 4 ml of ice-cold PBS. The recovered cells were incubated with the appropriate antibodies and analyzed by flow cytometry. Four mice were used in each group (PBS, C5a, and C5a with apyrase).

AFM analysis

Macrophages were seeded onto WillCo dishes (WillCo Wells) with a glass bottom of 40-mm diameter and 0.17-mm thickness, and cells were fixed with 4% paraformaldehyde in PBS. Optically guided AFM imaging was performed with a BioScope II atomic force microscope (Veeco Instruments) and NanoScope V controller (Veeco Instruments) integrated with a Leica DMI6000 B inverted microscope (Leica Microsystems). Measurements were performed in PBS with MLCT (microlever contact) cantilevers (Veeco Probes) with nominal spring constants of 0.01 N/m. Contact-mode imaging was performed at scan rates of 0.5 to 1 Hz with minimal contact forces. AFM data were collected and analyzed with NanoScope v7.30 software.

Statistical analysis

Normality and homoscedasticity were tested with the Kolmogorov-Smirnov and Levene tests, respectively. When the criteria of normality and homogeneity of variance were met, a one-way analysis of variance (ANOVA) was performed with an α value of 0.05. The post hoc Scheffé test was used to compare groups. Alternatively, when the criteria for an ANOVA were not met, the nonparametric Kruskal-Wallis test and post hoc Nemenyi test were used. Statistical analyses were performed with SPSS software, and data are presented as the mean ± SEM.


Acknowledgments: We thank N. Tschentscher and H.-J. Heinecke for help with statistical analyses. Funding: This study was supported by grants (HA110710 and KR620803) from the Innovative Medizinische Forschung program at the Westfälische Wilhems-Universität Münster (to P.J.H. and M.K.), the Deutsche Forschungsgemeinschaft (Schw407/9-3 to A.S.), the IZKF (Interdisziplinäres Zentrum für Klinische Forschung) Münster (Schw2/030/08 to A.S.), and the Danish Medical Research Council (to J.L.). Author contributions: M.K., J.S., K.I., H.-C.K., S.S., P.S., and P.J.H. performed the experiments; M.K., L.S., T.S., J.L., B.R., P.B.C., A.S., and P.J.H. designed the experiments; and M.K. and P.J.H. wrote the paper. Competing interests: P.B.C. is an employee of and a shareholder in Portola Pharmaceuticals, which has a P2Y12 inhibitor in phase II clinical trials. The use of P2ry2−/−, P2ry12−/−, and Panx1−/− mice requires materials transfer agreements.

Supplementary Materials

Fig. S1. Reversed-phase HPLC analysis of ATP degradation by potato apyrase.

Fig. S2. Inhibition of the adenosine feedback loop with 8-SPT.

Fig. S3. Chemotaxis is impaired by blocking ecto-ATPase in P2ry2-deficient macrophages.

Fig. S4. Chemotaxis is not impaired in Panx1-deficient macrophages.

Fig. S5. The P2Y12 receptor antagonist AR-C69931MX blocks ADP-β-S–induced lamellipodial membrane protrusions.

Movie S1. Wild-type macrophages in a C5a chemotactic gradient.

Movie S2. Wild-type macrophages in a C5a chemotactic gradient with apyrase (40 U/ml).

Movie S3. P2ry2−/− macrophages in a C5a chemotactic gradient.

Movie S4. P2ry12−/− macrophages in a C5a chemotactic gradient.

Movie S5. P2ry2−/− macrophages in a C5a chemotactic gradient with AR-C69931MX (10 μM).

Movie S6. P2ry2−/− macrophages in a C5a chemotactic gradient with 8-SPT (100 μM).

Movie S7. P2ry12−/− macrophages in a C5a chemotactic gradient with 8-SPT (100 μM).

Movie S8. Wild-type macrophages in a C5a chemotactic gradient with 8-SPT (100 μM).

Movie S9. P2ry2−/− macrophages in a C5a chemotactic gradient with “cocktail block.”

Movie S10. P2ry2−/− macrophages in a C5a chemotactic gradient with “triple block.”

Movie S11. Panx1−/− macrophages in a C5a chemotactic gradient.

Movie S12. ATP-induced lamellipodial membrane protrusions.

Movie S13. Lack of ATP-induced lamellipodial formation in P2ry2−/− macrophages.

References and Notes

View Abstract

Navigate This Article