UDP acting at P2Y6 receptors is a mediator of microglial phagocytosis (original) (raw)

. Author manuscript; available in PMC: 2012 Oct 5.

Published in final edited form as: Nature. 2007 Apr 4;446(7139):1091–1095. doi: 10.1038/nature05704

Abstract

Microglia, brain immune cells, engage in the clearance of dead cells or dangerous debris, which is crucial to the maintenance of brain functions. When a neighbouring cell is injured, microglia move rapidly towards it or extend a process to engulf the injured cell. Because cells release or leak ATP when they are stimulated1,2 or injured3,4, extracellular nucleotides are thought to be involved in these events. In fact, ATP triggers a dynamic change in the motility of microglia in vitro5,6 and in vivo3,4, a previously unrecognized mechanism underlying microglial chemotaxis5,6; in contrast, microglial phagocytosis has received only limited attention. Here we show that microglia express the metabotropic P2Y6 receptor whose activation by endogenous agonist UDP triggers microglial phagocytosis. UDP facilitated the uptake of microspheres in a P2Y6-receptor-dependent manner, which was mimicked by the leakage of endogenous UDP when hippocampal neurons were damaged by kainic acid in vivo and in vitro. In addition, systemic administration of kainic acid in rats resulted in neuronal cell death in the hippocampal CA1 and CA3 regions, where increases in messenger RNA encoding P2Y6 receptors that colocalized with activated microglia were observed. Thus, the P2Y6 receptor is upregulated when neurons are damaged, and could function as a sensor for phagocytosis by sensing diffusible UDP signals, which is a previously unknown pathophysiological function of P2 receptors in microglia.

Microglia express several functional P2 receptors, and their P2X4, P2X7 and P2Y12 receptors have already been described in relation to their physiological and pathophysiological consequences5–9. To investigate the expression of mRNAs for P2 receptors that are at a higher concentration in cultured rat microglia, we conducted reverse-transcriptase-mediated polymerase chain reaction (RT–PCR) analysis withcomplementary DNA coding forP2Y and P2X receptors(Fig. 1a). In accordance with previous reports5–9, microglia expressed mRNAs encoding P2X4, P2X7 and P2Y12 receptors. However, we found unexpectedly that cultured rat microglia expressed a large amount of mRNA coding for P2Y6 receptors, which was also confirmed by western blotting for the expression of P2Y6 receptor protein (Fig. 1b). The P2Y6 receptor is coupled to the activation of phospholipase C (PLC), leading to the production of inositol 1,4,5-trisphosphate (InsP3) and the release of Ca2+ from InsP3-receptor-sensitive stores10,11. We therefore examined changes in the intracellular Ca2+ concentration ([Ca2+]i) in microglia and found that the P2Y6 receptor agonist UDP evoked increases in [Ca2+]i in a concentration-dependent manner, and it also increased the fraction of the UDP-responsive cells (Supplementary Fig. 1a). The elevations in [Ca2+]i induced by 100 μM UDP were significantly inhibited by the PLC inhibitor U73122, the Ca2+-ATPase inhibitor in sarcoplasmic/endoplasmic reticulum thapsigargin, and the membrane-permeable InsP3 receptor inhibitor xestospongin C, but were little affected by pertussis toxin (Supplementary Fig. 1b). The UDP-evoked [Ca2+]i increases in microglia were significantly inhibited by reactive blue 2 (RB2), known as a potent P2Y6 antagonist11, suramin, which inhibits P2Y6 receptor at higher concentrations, the diisothiocyanate derivative MRS2578, which is a selective antagonist of the P2Y6 receptor12, and an antisense oligonucleotide (AS) for P2Y6 receptors, but not by a random-sequence oligonucleotide (R-oligo) (Fig. 1c). All these data show that rat microglia express functional P2Y6 receptors by which UDP mobilizes Ca2+.

Figure 1. Expression of P2Y6 receptor and UDP-evoked increase in [Ca2+]i in cultured microglia.

Figure 1

a, RT–PCR analysis of the expression of mRNAs for P2Y6, P2Y12, P2X4 and P2X7 receptors in microglial cells. b, Expression of P2Y6 receptor protein confirmed by western blotting analysis. c, Effects of various chemicals on the increase in [Ca2+]i (measured as the change in ratio of fluorescence at 340 nm to that at 380 nm) evoked by 100 μM UDP in microglia. The maximum increase in Fura-2 fluorescence evoked by 100 μM UDP was considered as the control response, and values are expressed as a percentage of control. Data show means and s.e.m. for 24–36 cells obtained from at least three independent experiments. Significant differences from the response to UDP alone: asterisk, P < 0.05; two asterisks, P < 0.01 (Student's _t_-test).

Morphogenesis, cell movement and phagocytosis are driven by dynamic reorganization of the actin cytoskeleton13,14. We showed previously that activation of P2Y12/13 receptors, another microglial G-protein-coupled receptor, resulted in membrane ruffling and chemotaxis in microglia5,6, and therefore we sought first to determine whether the P2Y6-receptor-mediated signals affect the cell movement of microglia. Membrane ruffles are structures that are found primarily at the front edges of migrating cells15. To determine whether P2Y6 activation stimulates microglial chemotaxis, cells were stimulated with either UDP or ATP. Neither lamellipodia-like membrane ruffles (Fig. 2a left) nor chemotaxis (Fig. 2b left) were observed when stimulated with UDP, whereas ATP produced both responses (Fig. 2a right and Fig. 2b right). However, UDP caused actin reorganization and formed aggregates of F-actin in the interior of the cells (Fig. 2a left, arrows). On stimulation with UDP (100 μM), microglia rapidly changed their morphology (Supplementary Fig. 2a); namely, to microglial processes with filopodia-like protrusions (arrows) and phagosome-like vacuoles (arrowheads). A crown-like circular structure rich in F-actins, termed the ‘phagocytotic cup’16, was also observed around the zymosan particles (Supplementary Fig. 2b, red). We speculated that UDP somehow regulates the morphogenesis of microglia, which may be involved in microglial endocytotic activities such as pinocytosis, macropinocytosis and phagocytosis. Phagocytosis is one of the most important physiological functions of microglia17 and is the process activating the uptake of larger particles (more than 0.5 μm) by actin-based mechanisms. We investigated the UDP-evoked phagocytosis process by time-lapse videomicroscopy and flow cytometry (fluorescence-activated cell sorting; FACS)-based assay. When stimulated with 100 μM UDP, microglia rapidly phagocytosed fluorescent zymosan particles (green) (Fig. 2c, see also Supplementary Video). A quantitative phagocytosis assay by FACS shows that UDP induced the phagocytosis of latex beads in a concentration-dependent fashion (5–1,000 μM) in a 20-min incubation period (Fig. 2d). GDP (100–1,000 μM), a weak agonist of the P2Y6 receptor, caused a slight uptake of microspheres (at 100 μM this was 49.7 ± 8.6% of UDP alone; n = 4) but ADP, also known as a weak partial agonist of the mouse P2Y6 receptor, failed to stimulate the uptake (at 100 μM it was 0.3 ± 2.3% of UDP alone; n = 4). This is in good agreement with the previous finding that ADP does not activate rat P2Y6 receptors18. The phagocytosis induced by 100 μM UDP was significantly inhibited by 30–100 μM RB2, a higher concentration of suramin (300 μM) and MRS2578 (0.01–3 μM), and was nearly abolished by P2Y6 AS (Fig. 2e; see also Supplementary Fig. 2c, d). Recent reports indicate the existence of functional cross-talk between the nucleotides and cysteinyl leukotrienes (CysLTs, for example LTD4) in orchestrating inflammatory responses19, indicating that some nucleotides may reveal their functions by means of a CysLT receptor (CysLTR). Microglia express a functional CysLT1R, whose activation by LTD4 resulted in an increase in [Ca2+]i in microglia (Supplementary Fig. 3a). Thus, UDP acting on CysLT1R may reveal various microglial responses. However, MRS2578, a selective P2Y6 receptor antagonist, did not block the LTD4-evoked Ca2+ responses in CysLT1R-transfected Chinese hamster ovary cells (Supplementary Fig. 3b) at a dose that inhibited the UDP-evoked increase in [Ca2+]i and phagocytosis in microglia (Figs 1c and 2e). In addition, 1 μM LTD4 did not induce phagocytosis in microglia (Supplementary Fig. 3c, 4.8 ± 4.2% of that with 100 μM UDP alone; n = 3). All these findings suggest that the contribution of the CysLT1R to the UDP-evoked phagocytosis in microglia is negligible. Taken together, these data strongly suggest that rat microglial P2Y6 receptors are coupled with phagocytic functions. The UDP-evoked phagocytosis was inhibited by 1 μM thapsigargin, the protein kinase C inhibitor staurosporin at 5 μM, and 10 μM U73122 (see Supplementary Fig. 4), indicating that activation of the P2Y6 receptor seems to trigger phagocytosis through the pathway(s) mediated by PLC-linked Ca2+ and protein kinase C.

Figure 2. Changes in cell motilities of microglia.

Figure 2

a, UDP- and ATP-evoked membrane ruffles. Cultured microglia were stimulated for 5 min with 100 μM UDP (left) and 10 μM ATP (right), fixed, permeabilized, and then stained with anti-phalloidin. Scale bar, 10 μm. b, Typical chemotactic responses of microglia towards 100 μM UDP (left) and 100 μM ATP (right) assessed by the Dunn chemotaxis chamber (see Methods). c, Time-lapse images showing the effect of UDP on the microglial morphogenic changes and the uptake of fluorescent zymosan particles (green). The time after addition of UDP is shown in seconds in each picture. d, The UDP-evoked uptake of microspheres was assessed quantitatively as a phagocytosis index by using FACS. Data are mean and s.e.m. for three experiments (asterisk, P < 0.05; two asterisks, P < 0.01 compared with basal). e, Effects of the P2 receptor antagonists reactive blue 2 and suramin, the P2Y6 receptor antagonist MRS2578, and P2Y6 AS or MM on the UDP-evoked phagocytosis. Data are means and s.e.m. for three or four experiments (asterisk, P < 0.05; two asterisks, P < 0.01 compared with UDP alone).

Because phagocytes remove dead or damaged cells, debris and invading pathogens, recognition is the first step in phagocytosis. It is initiated by activation of the phagocytosis-promoting receptors such as Fc receptors and complement receptors20. In the central nervous system, microglia possess these receptors and remove amyloid-β, a key molecule in Alzheimer's disease, and attenuate Alzheimer's disease-like pathology21. With regard to apoptotic cells, microglia may also remove such cells by recognizing so-called ‘eat-me’ signals20. However, in the present study we used non-opsonized zymosan (Fig. 2c) and latex beads (Fig. 2d, e), which were not recognized by opsonin-dependent receptors such as Fc receptors, complement receptors or vitronectin receptors. Phagocytosis-promoting receptors also include opsonin-independent ones such as β1-integrins, mannose receptors, scavenger receptors and phosphatidylserine receptors13; in fact, microglia expressed all these receptors (Supplementary Fig. 5a–e, cell lysates). Among these receptors, β1-integrin was detected as a bead-associated protein that was slightly increased on stimulation with UDP (Supplementary Fig. 5a, bead-associated) and localized at membrane ruffle-like or phagocytic cup-like structures (see also Supplementary Fig. 2b), to which fluorescent microspheres were attached (Supplementary Fig. 5f). However, we do not know whether β1-integrin itself binds or recognizes the microspheres. β1-Integrin might be involved in some way in the machinery of phagocytosis or in the uptake processes of the microspheres in response to UDP, but the precise target molecule or molecules that bind or recognize microspheres to be phagocytosed remains to be identified. The microglial phagocytosis seen in the present study is a new type that is promoted by the diffusible extracellular molecule UDP. However, we cannot deny the possibility that the UDP may simply facilitate the machinery of phagocytosis and that UDP-evoked phagocytosis observed in this study may even include macropinocytosis.

To determine the expression and function of microglial P2Y6 receptors in vivo, the excitotoxicity of brain injury was induced by kainic acid (KA) (Fig. 3). KA is an excitatory amino acid that is often used to cause limbic motor epilepsy or excitatory neuronal cell death in vivo and in vitro. KA acts on non-NMDA glutamate receptors to facilitate excess excitability, thereby leading to necrosis and even apoptosis of neurons. The hippocampal CA1 and CA3 regions are susceptible to neuronal death in response to KA22. When KA was injected intraperitoneally into rats (10 mg kg–1), it produced typical limbic seizure within 60 min. At 72 h after the administration of KA, the brains were removed and were used for western blotting, immunohistochemical assays and in situ hybridization (ISH). Western blotting analysis showed that KA increased the expression of P2Y6 receptors in comparison with the saline-injected control groups (Fig. 3A, B). Double staining of microglia and neurons by anti-Iba1 (green) and anti-neuronal nuclei (NeuN, red) antibodies, respectively, showed that KA induced severe neuronal loss in the hippocampal CA1 and CA3 regions, where intense Iba1-positive signals—indicative of microglia—were observed. KA increased the number of microglia appearing in the activated form with poorly ramified, short and thick processes (Fig. 3C, f–h). Small NeuN signals seemed to be incorporated in some microglia (see g and h in Fig. 3C), suggesting that microglia phagocytose damaged or dead neurons. These findings suggest that microglia might migrate or proliferate, probably as a result of KA-induced neuronal damage.

Figure 3. Increase in P2Y6 receptors in the hippocampus after kainic acid (KA)-treatment.

Figure 3

A, Western blot analysis, showing increase in P2Y6 receptor protein in rats treated intraperitoneally with 10 mg kg–1 KA, 72 h after treatment. B, Summary of quantitative data; KA was applied at 10 mg kg–1. Results are means and s.e.m. for 8 (control) and 7 (KA) experiments (asterisk, P < 0.05 compared with control). C, Immunohistochemical analysis in naive control (a–d) and KA-treated (e–h) rats; red, anti-NeuN antibody; green, anti-Iba1 antibody. Rectangles in a and e are expanded in b and f, respectively. Rectangles in b and f also correspond to c, d and g, h, respectively. D, Anti-P2Y6 antibody signals (green) were increased by KA (a, control; b, KA), which was colocalized with microglia (red in c, anti-OX42) but not with astrocytes (red in d, anti-GFAP) or neurons (red in e, anti-NeuN). E, ISH analysis. a, b, mRNA coding for P2Y6 receptor in naive rats was very low but was increased at the hippocampal CA3 region by KA (3 days later) (blue dots and arrowheads, KA). c, Double staining with P2Y6 antisense RNA probe (blue dots) and anti-Iba-1 antibody (brown signals, white (naive) or black (KA) arrows). In KA-treated rats there was an increased number of microglia, which was associated with P2Y6 receptor mRNA (blue signals, inset at higher magnification in KA).

We further examined the cell types that produced increases in P2Y6 receptor protein in response to the administration of KA, and found that P2Y6 immunoreactivities (green in Fig. 3D) were associated with the microglia (OX-42, red in Fig. 3D, c) but not with astrocytes (glial fibrillary acidic protein (GFAP), red in Fig. 3D, d) or neurons (NeuN, red in Fig. 3D, e). Furthermore, we performed ISH to characterize the expression of mRNA coding for P2Y6 receptors with the use of digoxigenin-labelled antisense RNA probe. Signals for P2Y6 receptor mRNA were very low in the naive animals but were upregulated three days after treatment with KA (Fig. 3E, b; blue dots indicated by arrowheads). At this time, the number of microglia increased markedly, especially at the hippocampal CA3 and CA1 regions (Fig. 3C). After ISH, the sections were stained with anti-Iba1 antibody to characterize P2Y6 receptor mRNA signals. In the hippocampal CA3 regions of naive rats, there were very few anti-Iba1-positive microglia that did not show P2Y6 receptor mRNA. In contrast, in the hippocampal CA3 of KA-injected rats, there was an increased number of anti-Iba1-positive microglia, in which P2Y6 receptor signals were colocalized with microglia (Fig. 3E, c; KA, black arrows, see also inset at higher magnification).

There is a growing literature about ‘eat-me’ signals that are expressed on the cell surface of apoptotic or dying cells. However, diffusible signals that trigger phagocytosis have received only limited attention. When neurons or cells are exposed to traumatic injury such as ischaemia, they swell and subsequently shrink as a result of increased permeability. This is followed by leakage of cytoplasmic molecules, leading to necrotic cell death. Thus, cytoplasmic nucleotides could be diffusible messengers that signal the crisis state to adjacent cells including microglia. In fact, the diffusible messenger ATP promotes microglial chemotaxis and/or migration3–6. Diffusible molecules might be insufficiently precise to cause phagocytes to recognize and eat cells. However, released or leaked nucleotides are immediately degraded by the extracellular nucleotide-degrading enzymes. In this respect, UDP might be a localized and transient marker of traumatized or necrotic cells.

Cell injury results in a leakage of ATP that affects the motility of adjacent cells, including microglia3,4. In addition, cells release or leak uridine nucleotides23 and nucleotide sugars24 in response to various stimuli or ischaemic injury25. We therefore next investigated whether KA increases the release of extracellular UDP from neurons to induce microglial phagocytosis. Cultured hippocampal neurons were stimulated with and without 100 μM KA for 1 h; the supernatant was then collected for nucleotide assay by high-performance liquid chromatography (HPLC) or for phagocytosis assay by FACS (Fig. 4). Because released or leaked UTP is rapidly degraded into UDP, UMP and uridine by ARL67156-sensitive ectonucleotidases, we monitored the amount of UTP rather than UDP, and collected the supernatant and the microdialysates in the presence of 20 μM ARL67156 throughout experiments. There was a close relationship between the HPLC peak corresponding to UTP and the concentration of the standard UTP (_R_2 = 0.9947). The amount of UTP in the KA-treated supernatant was significantly larger than that in the KA-untreated control supernatant (Fig. 4b; control, 2.3 ± 1.1 μM; KA treated, 10.5 ± 3.9 μM, P < 0.05). We also tested whether the KA-treated super-natant obtained from cultured hippocampal neurons facilitated microglial phagocytosis. Hippocampal neurons were treated with and without 100 μM KA for 1 h; each supernatant was collected and added to microglia; this was followed by a phagocytosis assay. As shown in Fig. 4c, when microglia were incubated with the KA-treated supernatant for 20 min, there was a significant increase in phagocytosis, which was blocked by the P2Y6 receptor antagonist MRS2578 (1 μM). KA alone did not stimulate phagocytosis.

Figure 4. KA-evoked increases in extracellular uridine nucleotides and P2Y6-receptor-mediated microglial phagocytosis in vitro and in vivo.

Figure 4

a, Schematic diagram of the experiments in vitro. Sup., supernatant. b, Summary of the UTP concentration in the KA-treated and control supernatants. Data show means and s.e.m. for at least five independent experiments (asterisk, P < 0.05 compared with control). c, Effects of the KA-treated and control supernatant on microglial uptake of fluorescent latex beads. Data show means and s.e.m. for at least four independent experiments (asterisk, P < 0.05 compared with control; hash sign, P < 0.05 compared with KA-treated supernatant). d, Schematic diagram of the experiments in vivo. KA was applied intraperitoneally at 10 mg kg–1. e, Time course of changes in [UTP]o in baseline dialysates (before treatment with KA (Pre), and 1, 2, 3 and 5 days afterwards). Inset, fold increase at day 3 (compared with before treatment). f, Typical pictures of fluorescent microspheres (green) attached or taken up by microglia (red, anti-Iba1) in the KA-treated (left) and KA + MRS2578-treated (right) hippocampal CA3 regions. Scale bar, 20 μm. g, Quantitative analysis of phagocytosis in vivo (details are provided in Supplementary Methods). Changes in P2Y6 receptor protein by P2Y6 AS or MM are shown at the top of corresponding columns. Values are means and s.e.m. (asterisk, P < 0.01 compared with control (–KA); hash sign, P < 0.05; two hash signs, P < 0.01 compared with KA-treated group). Statistical analyses were performed by ANOVA with Scheffe's multiple comparison. At least three sections containing the injection sites were analysed per animal, and at least three animals were used in each group for analysis.

Finally, we tested whether KA induces the release of UDP and P2Y6-receptor-mediated phagocytosis in vivo. An increase in extracellular UTP concentration ([UTP]o) was observed soon after injection of KA (from 1 to 4 h after injection), which reached 2–3-fold higher than the KA-untreated control (data not shown). At 1 day after KA injection, [UTP]o was about 1.5–2.0-fold higher than the KA-untreated control (Fig. 4e). Then, at day 3, [UTP]o reached almost 10-fold higher levels (9.4 ± 1.2-fold; Fig. 4e and inset), which decreased slightly at day 5. A higher (5–10-fold) [UTP]o was observed 2–3 days after the injection of KA, which lasted at least another couple of days. It should be noted that loss of neurons (removal of neurons) also became obvious 2–3 days after the KA injection. We further injected fluorescent microspheres into the hippocampal CA3 regions of KA-treated rats, and then counted the numbers of the microspheres phagocytosed or attached by microglia. The P2Y6 receptor antagonist MRS2578 was injected into the hippocampal CA3 region, and P2Y6 AS or MM (mismatch oligonucleotide) was injected into the third ventricle. The number of microspheres taken or attached by microglia was markedly increased by KA treatment, which is significantly inhibited by MRS2578 or P2Y6 AS but not by MM (Fig. 4g; see also Supplementary Fig. 6). These findings all suggest that UDP/P2Y6-receptor-mediated signals are important for microglial phagocytosis even in vivo.

A recent review described that dying cells use both ‘find-me’ and ‘eat-me’ signals for phagocyte attraction and recognition26. Nucleotides could be both ‘find-me’ and ‘eat-me’ signals. The intracellular ATP concentration is estimated to be high (more than 5 mM) and the UTP concentration is reported to be one-third that of ATP23. Cells release ATP, and here we showed that KA caused an increase in extracellular UTP or UDP. Microglia might therefore be attracted by ATP or ADP5,6 and subsequently recognize UDP, leading to the removal of the dying cells or their debris. It is interesting that ATP/ADP is not able to efficiently activate P2Y6 receptors; neither can UDP act on P2Y12/13 receptors. Thus, even if these nucleotides were leaked or released simultaneously, adenine and uridine nucleotides would regulate microglial motilities, namely chemotaxis and phagocytosis, in a mutually exclusive but coordinated fashion.

So far we have not shown quantitative data indicating that individual microglia upregulate the expression of P2Y6 receptors. A significant, but not drastic, increase in P2Y6 receptor protein in the hippocampus was observed after injection of KA (Fig. 3A, B). ISH data show that expression of mRNA coding for P2Y6 receptors in microglia was very low in naive animals but became obvious in an increased number of microglia after KA injection (Fig. 3E), suggesting that the increase in P2Y6 receptor protein is not due simply to an increased number of microglia but is upregulated in individual microglia. We also emphasize that even if the extent of P2Y6 receptor upregulation is not drastic, an increase in extracellular UDP, a ligand for P2Y6 receptor, is markedly increased after treatment with KA (detected as UTP, almost 10-fold; Fig. 4e) and therefore that the UDP/P2Y6 receptor system would be sufficiently activated to cause microglial phagocytosis after treatment with KA. In comparison with the extensive knowledge of the molecular events involved in the regulation of apoptosis or necrosis, relatively little is known about the processes responsible for the clearance of dead cells and the degradation of waste materials. Considering the present findings that injured neurons leak diffusible UTP/UDP and cause the upregulation of P2Y6 receptors in microglia, the UDP/P2Y6 receptor system might function as a critical device covering the phagocytosis of both apoptotic and necrotic cells if they release or leak UDP by sensing diffusible UDP signals.

Thus we have shown that microglia express P2Y6 receptors that function as a sensor of phagocytosis. The P2Y6 receptor agonist UDP is released (as UTP) when neurons are damaged by KA. Thus, the activation of P2Y6 receptors by UDP would be a key event in initiating the clearance of dying cells or debris in the central nervous system.

METHODS

Detailed methods are provided in Supplementary Information.

Microglia culture

The protocol was reviewed and approved by the Committee for Institutional Laboratory Animal Care of the National Institute of Health Sciences. Rat primary cultured microglia were prepared in accordance with the method described previously27.

Phagocytosis assay in vitro and in vivo

In vitro microglial phagocytosis was assessed by either FACScan analysis or imaging analysis with fluorescently labelled microspheres. For the in vivo phagocytosis assay, fluorescently labelled microspheres were injected into the hippocampal CA3 region after injection of KA, and then the number of fluorescent microspheres associated with microglia was analysed by confocal microscopy (LSM 5 Pascal; Carl Zeiss).

Microdialysis

A microdialysis probe (A-I type probe; Eicom) was inserted into the hippocampal CA3 region by means of a guide cannula, and was perfused continuously at a flow rate of 3.0 μl min–1 (collected for 60 min) supplemented with the ectonuclease inhibitor ARL67156 (20 μM) (Sigma).

Quantification of UTP by HPLC

The concentration of nucleotides in the supernatant of the hippocampal cultures was analysed with an HPLC system (Jasco) combined with a C18 column (4.6 × 250 mm, Shodex) as described28, with minor modifications.

Data analysis and statistics

All results are expressed as means ± s.e.m. A statistical analysis was performed with Student's _t_-test or analysis of variance, followed by Scheffe's multiple comparison test. Differences were considered to be significant at P < 0.05.

Supplementary Material

movie

supporting information

Acknowledgements

We thank T. Shimizu and Dr. S. Ishii for providing CysLT1 receptor-expressed Chinese hamster ovary cells; K. Sakemi for technical assistance; Y. Sasaki for helpful suggestions; K. Suzuki and R. Adachi for allowing us to use the Pascal confocal microscope system; and T. Nishimaki-Mogami, Y. Ohno and T. Nagao for continuous encouragement. This study was supported in part by a grant from The National Institute of Biomedical Innovation, a grant from Uehara Memorial Foundation, a Grant-in-Aid for Scientific Research on Priority Areas, for Creative Scientific Research, Scientific Research (A and B), and for Young Scientists (A) from the Ministry of Education, Science, Sports and Culture of Japan.

Footnotes

Author Contributions S.K. designed most experiments, performed Ca2+ imaging and in vivo experiments and wrote the paper. Y.S.M. conducted major parts of the experiments. K.N.T. and Y.S. carried out the FACS assay and the HPLC analysis, respectively. K.O. and S.K. performed the chemotaxis analysis. B.V.J. and K.A.J. made the P2Y6 receptor antagonist MRS2578. M.T. analysed the data. K.I. analysed the data and coordinated the project. K.I. also designed the project. All authors discussed the results and commented on the manuscript.

The authors declare no competing financial interests.

Supplementary Information is linked to the online version of the paper at www.nature.com/nature.

References

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

movie

supporting information