ABSTRACT
Accumulating data point to a key role of Ca2+-dependent gliotransmitter release as a modulator of neuronal networks. Here, we tested the hypothesis that astrocytes in response to agonist exposure also release lipid modulators through activation of Ca2+-independent phospholipase A2 (iPLA2) activity. We found that cultured rat astrocytes treated with selective ATP and glutamatergic agonists released arachidonic acid (AA) and/or its derivatives, including the endogenous cannabinoid 2-arachidonoyl-sn-glycerol (2AG) and prostaglandin E2 (PGE2). Surprisingly, buffering of cytosolic Ca2+ resulted in a sharp increase in agonist-induced astrocytic lipid release. In addition, astrocytic release of PGE2 enhanced miniature excitatory post-synaptic potentials (mEPSPs) by inhibiting the opening of neuronal Kv channels in brain slices. This study provides the first evidence for the existence of a Ca2+-independent pathway regulating the release of PGE2 from astrocytes, and furthermore demonstrates a functional role for astrocytic lipid release in the modulation of synaptic activity.
Significance Statement Until now, the majority of studies implicating astrocytes in modulating synaptic activity have focused on Ca2+-dependent release of traditional gliotransmitters such as D-serine, ATP, and glutamate. Mobilization of intracellular stores of Ca2+ occurs within a matter of seconds, but this novel Ca2+-independent lipid pathway in astrocytes could potentially occur on a still faster time scale and thus participate in the rapid signaling processes involved in synaptic potentiation, attention, and neurovascular coupling.
INTRODUCTION
Comprising the major part of the dry weight of adult brain, lipids are an essential component of the phospholipid bilayer, mainly occurring as long-chain polyunsaturated fatty acids (PUFA) such as arachidonic acid (AA) and docosahexanoic acid (DHA) (Sinclair, 1975). The liver is the major site of AA synthesis, but the brain can produce AA and DHA in situ from their precursor fatty acids, linoleic and linolenic acids (Dhopeshwarkar and Subramanian, 1976). Astrocytes play a central role in the synthesis of AA and DHA in brain. Through their vascular endfeet, astrocytes have prime access to fatty acid precursors arriving across the blood-brain barrier (BBB), and serve as the major site for processing the essential fatty acids in the central nervous system (CNS) (Moore, 1993). Astrocytes also play a key role in macroscopic distribution of lipids in the brain parenchyma via perivascular glymphatic flux (Rangroo Thrane et al., 2013a; Plog and Nedergaard, 2018).
Lipids have gained much attention for their role as bioactive mediators in the CNS (Carta et al., 2014; Ledo et al., 2019). Numerous studies have focused on lipids in relation to functional hyperemia and synaptic activity. For instance, PGE2 is a potent vasodilator and vasoconstrictor regulating CNS blood flow (Zonta et al., 2003; Takano et al., 2006; Gordon et al., 2007; Dabertrand et al., 2013; MacVicar and Newman, 2015; Czigler et al., 2019), and modulates the membrane excitability of CA1 pyramidal neurons during synaptic activity (Chen and Bazan, 2005). Furthermore, AA and its derivatives are important intracellular second messengers that can modulate the activities of various ion channels (Piomelli, 1993; Meves, 1994; Horimoto et al., 1997; Boland and Drzewiecki, 2008; Cordero-Morales and Vasquez, 2018). In addition, PGE2 can suppress the outwardly rectifying Kv current in sensory neurons (Nicol et al., 1997; Evans et al., 1999), whereas AA suppresses Kv channels in the soma or dendrites of pyramidal neurons, and consequently broadens their presynaptic action potentials (Carta et al., 2014) and enhances EPSPs (Ramakers and Storm, 2002). However, those studies focused on the effects of neuronal lipid release effects on synaptic activity and paid scant attention to receptor mediated pathways by which astrocytic lipids might influence synaptic activity. Given that the majority of AA, DHA, and other lipids present in the extracellular fluid is produced by astrocytes (Moore et al., 1991), receptor-mediated release could be a significant factor in the modulation of synaptic activity.
In culture, astrocytes can release AA in a cPLA2 Ca2+-dependent pathway upon activation of metabotropic glutamate (mGluR) and P2Y purine (ATP) receptors (Bruner and Murphy, 1990; Stella et al., 1994; Stella et al., 1997; Chen and Chen, 1998). In addition, astrocytes also express the Ca2+-independent PLA2 (iPLA2) enzyme (Sun et al., 2005). Both of these isoforms are activated by the G-protein βγ subunit (Jelsema and Axelrod, 1987; Murayama et al., 1990; van Tol-Steye et al., 1999), but iPLA2 does not require Ca2+ or PKC phosphorylation for its activation (Winstead et al., 2000). Receptor-stimulated iPLA2 activation can release AA and DHA in numerous cell types (Gross et al., 1993; Portilla et al., 1994; Akiba et al., 1998; Seegers et al., 2002; Tay and Melendez, 2004), but has not been fully explored in the case of astrocytes.
Previous work in our laboratory and elsewhere has demonstrated that astrocytes are capable of releasing gliotransmitters upon stimulation with mGluR or ATP receptor agonists, but the impact of this signaling on synaptic activity remains controversial. Recent studies with genetically encoded calcium indicators enabled us to identify localized Ca2+ signals within astrocytic fine processes and have confirmed Ca2+-dependent astrocytic effects on synaptic activity (Yu et al., 2018). However, the existence of Ca2+-independent lipid signaling in astrocytes is still not clearly established. Therefore, in this study we tested the hypothesis that astrocytes could support Ca2+-independent lipid signaling to modulate synaptic transmission. We used Ca2+ chelation to show that ATP and mGluR agonism can release lipids through iPLA2 activation, resulting in the potentiation of synaptic activity. Because we currently lack the tools to assess Ca2+-independent signaling in astrocyte fine processes adjacent to synapses, we use Ca2+ inhibition on a grand scale to unmask this phenomenon (see discussion).
MATERIALS and METHODS
Culture
Cultured neocortical astrocytes were prepared from postnatal day 1 or 2 Wistar rat pups (Taconic Farms, Inc.) of either sex, as previously described (Lin et al., 1998). In brief, cerebral cortices were dissected on ice, and the meninges removed. The tissue was washed three times in Ca2+-free Hanks’ balanced salt solution, and then triturated, filtered through a 70 µm nylon mesh, and centrifuged. The pellet was resuspended in 10 % fetal bovine serum in Dulbecco’s modified Eagle’s medium (DMEM)/F12 containing penicillin (100 IU ml1) and streptomycin (100 µg ml-1), and transferred to culture flasks. Cells were maintained at 37 °C in an incubator containing humidified air and 5% CO2. The medium was changed after 24 hrs and twice a week thereafter. More than 95% of the cells immunostained positive for GFAP. When the cells became confluent, they were rinsed two times in Ca2+-free Hanks’ balanced salt solution, suspended in 0.05 % trypsin-containing PBS for 1 min, resuspended in DMEM/F12, centrifuged to remove the trypsin-containing supernatant, and then plated in 24-well plates. Experiments were performed when the cells were 95% confluent.
Viral Vectors and Viral Transductions
Viral vectors driving GFAP cyto-GCaMP6f (Baljit S. Khakh) were obtained from the University of Pennsylvania Vector Core (AAV 2/5 serotype). Secondary rat astrocytic cultures were transduced with AAV GFAP cyto-GCaMP6f. After transduction, the cultures were incubated at 37 °C for 5 days prior to Ca2+ imaging experiments.
Ca2+ imaging
Cultured cells in 24-well plates were transduced with AAV GFAP cyto-GCaMP6f and incubated with various pharmacological agents for 30 minutes at 37 °C. Using confocal microscopy (Olympus FV500), calcium wave activity was evoked by adding an equal volume of medium containing 100 µM of ATP to each well. Relative increases in fluorescence signal evoked by P2Y receptor agonist exposure over baseline fluoresce (ΔF/F) were calculated as previously described (Nedergaard, 1994; Smith et al., 2018).
Radiolabeling and Assessment of AA Release
Confluent rat astrocytic cultures were incubated with 100 nCi [5,6,8,9,11,12,14,15-3H]-AA (PerkinElmer) overnight before the experiments. The cells were washed three times with serum-free medium, and then allowed to recover for 20 minutes. Before stimulation of P2Y receptors with 100 µM ATP, the cells were incubated with appropriate inhibitors for 10 to 12 minutes. Aliquots of medium were taken 15 minutes after agonist stimulation, and 3H-AA and/or its metabolites were measured by liquid scintillation counting.
HPLC/MS/MS analysis of lipids
Confluent rat astrocytes were washed with serum-free media and then allowed to recover for 1 hr in serum-free media. At this time, cells were incubated with vehicle or 20 µM cyclopiazonic acid (CPA) for 15 minutes and then challenged with 100 µM of ATP or vehicle for 15 minutes, as in the above [3H]-AA experiments. Media was then removed and put aside, and 2 ml of HPLC-grade methanol was added to the flask for 5 minutes, and then removed and out aside. An additional 2 ml of HPLC-grade methanol was added, cells were scraped from the sides of the flask, and the contents added to the previous two retained fractions. Deuterìum standards were added to a final concentration of 200 pM, and the samples were centrifuged at 19,000 x g and 24 °C. Lipids in the eluent were partially purified on 500 mg C18 Bond Elut solid phase extraction columns (Varian), concentrated into fractions of 40, 60, 85, and 100% methanol, then analyzed using HPLC/MS/MS (Shimadzu SCL10Avp-Wilmington, DE, USA; API 3000 Applied Biosystems/MDS SCIEX, Foster City, CA, USA) as previously described (Leishman et al., 2018). Over 50 lipids were targeted in these analyses, including 40 lipoamines, 3 acylglycerols, 2 free fatty acids, and 2 prostaglandins (namely X and Y). It was determined in pilot studies that the analyte concentrations were too low for reliable detection in the medium alone, but the extraction of medium and cells together gave sufficient lipid concentrations for reliable assay.
PGE2 Release Assessment via PGE2 Immunoassay
Confluent rat astrocytic cultures were washed 3 times with serum-free medium, and then allowed to recover for 20 minutes. Before stimulation with 100 µM ATP, the cells were incubated with appropriate P2Y inhibitors for 10 to 12 minutes. Aliquots of the medium were collected 15 minutes after stimulation and analyzed for PGE2 content uising an immunoassay kit (Cayman Chemicals) following the manufacturer’s instructions.
Western Blot
Protein in samples harvested from the 24-well plates were separated by SDS-PAGE and transferred to a nitrocellulose membrane, which was then blocked with Tris-buffered saline containing 0.05% (wt/vol) Tween 20 and 5% nonfat dry milk. The primary antibodies were anti-iPLA2 (Sigma, St. Louis, MO), anti-β-actin (Cell Signaling, Danvers, MA), at 1:1000 to 1:2000 dilutions in blocking buffer. Detection of chemiluminescence from horseradish peroxidase-linked secondary antibodies was performed using the ChemiDoc™ XRS+ System and running Image Lab™ software.
Isolation of human fetal astrocytes
Human fetal forebrain tissues were obtained from second-trimester aborted fetuses of 20 weeks gestational age. Tissues were obtained from aborted fetuses, with informed consent and tissue donation approval from the mothers, under protocols approved by the Research Subjects Review Board of the University of Rochester Medical Center. No patient identifiers were made available to or known by the investigators; no samples with known karyotypic abnormalities were included. The forebrain tissue samples were collected and washed 2-3 times with sterile Hank’s balanced salt solution containing Ca2+/Mg2+ (HBSS+/+). The cortical plate region (CTX) of the fetal forebrain was dissected and separated from the ventricular zone/subventricular zone (VZ/SVZ) portion. The CTX was then dissociated with papain as previously described (Keyoung et al., 2001). The cells were resuspended at a density of 2-4 × 106 cells/ml in DMEM/F12 supplemented with N2, 0.5 % FBS, and 10 ng/ml bFGF and plated in suspension culture dishes. The day after dissociation, cortical cells were recovered and subjected to magnetic activated cell sorting (MACS) to purify the astrocyte progenitor population. The recovered cells were briefly incubated with CD44 microbeads as per the manufacturer’s recommendations (Miltenyi Biotech). The cells were then washed, resuspended in Miltenyi Washing buffer, and bound to a magnetic column (Miltenyi Biotech). The bound CD44+ astrocyte progenitor cells were eluted, collected, and then washed with DMEM/F12. The purified human fetal astrocyte progenitors were cultured in DMEM/F12 supplemented with N2, and 5% FBS to differentiate them further. To prepare culture dishes for PGE2 immunoassays or immunocytochemistry, the fetal cortical astrocytes were dissociated with TrypLE (Invitrogen) into single cells and then plated onto poly-L-ornithine/laminin-coated 24-well plates (50,000 cells per well).
shRNA Lentiviral knockdown of iPLA2 in astrocytes
Rat astrocytes cultures were plated in a 24-well plate and grown to approximately 50% confluence. Following the manufacturer’s instructions with minor modifications, the cultures were transduced overnight by adding either group VI iPLA2 shRNA (r) lentiviral particles (sc-270117-V) or control shRNA lentiviral particles-A (sc-108080) directly to the culture medium containing polybrene (sc-134220) (all from Santa Cruz Biotechnology, Santa Cruz). At 24 hours after transfection, the culture medium was removed, and fresh culture medium without polybrene was added. Experiments and Western blot analysis were performed 7 days after transduction.
Acute Hippocampal Slice Preparation and Electrophysiology
Unless otherwise noted, 15-21 day old C57BL/6 (Charles River, Wilmington, MA), MrgA1+/- transgenic, and littermate control MrgA1-/- pups (courtesy of Dr. Ken McCarthy) (Fiacco et al., 2007) of either sex were used for the preparation of hippocampal slices as previously described (Wang et al., 2012). The pups were anesthetized in a closed chamber with isoflurane (1.5%) and decapitated. The brains were rapidly removed and immersed in an ice-cold cutting solution that contained (in mM): 230 sucrose, 2.5 KCl, 0.5 CaCl2, 10 MgCl2, 26 NaHCO3, 1.25 NaH2PO4, and 10 glucose, pH=7.2-7.4. Coronal slices (400 μm) were cut using a vibratome (Vibratome Company, St. Louis) and transferred to oxygenated artificial cerebrospinal fluid (aCSF) that contained (in mM): 126 NaCl, 4 KCl, 2 CaCl2, 1 MgCl2, 26 NaHCO3, 1.25 NaH2PO4, and 10 glucose, pH = 7.2-7.4 (osmolarity = 310 mOsm). Slices were incubated in this aCSF for 1-5 hours at room temperature before electrophysiological recording. Experiments were performed at room temperature (21-23 °C). During the recordings, the slices were placed in a perfusion chamber and superfused with aCSF gassed with 5% CO2 and 95% O2 at room temperature. Cells were visualized with a 40X water-immersion objective and differential inference contrast (DIC) optics (BX51 upright microscope, Olympus Optical, New York, NY). Patch electrodes were fabricated from filament thin-wall glass (World Precision Instruments) on a vertical puller; the pipette’s resistance was around 6-9 MΩ with addition of intracellular pipette solution. The pipette solution contained (in mM) 140 K-gluconate, 5 Na-phosphocreatine, 2 MgCl2, 10 HEPES, 4 Mg-ATP, 0.3 Na-GTP (pH adjusted to 7.2 with KOH). The current-voltage relationship (I-V curve) of voltage-gated potassium currents was recorded under voltage-clamp using an AxoPatch MultiClamp 700B amplifier (Axon Instruments, Forster City, CA). When measuring outward currents, QX314 (0.5 mM) was added to the pipette solution to block Na+ currents. For recordings of miniature excitatory post-synaptic potentials (mEPSPs), 0.5 µM TTX was added to the aCSF. The junction potential between the patch pipette and bath solution was zeroed before forming a giga-seal. Patches with seal resistances less than 1 GΩ were rejected. Data were low-pass filtered at 2 kHz and digitized at 10 kHz with a Digidata 1440 interface controlled by pClamp Software (Molecular Devices, Union City, CA).
Pharmacological agents used in cultures and slice experiments
Adenosine 5 -triphosphate (ATP, 100 µM); Cyclopiazonic acid (CPA, 20 µM); trans-(1S, 3R)-1-Amino-1, 3-cyclopentanedicarboxylic acid (t-ACPD, 100 µM); (±)-α-Amino-3-hydroxy-5-methylisoxazole-4-propionic acid hydrobromide ((±)-AMPA, 100 µM); Prostaglandin E2 (PGE2, 50 µM, Tocris); Phe-Met-Arg-Phe amide (FMRF, 15 µM); Thr-Phe-Leu-Leu-Arg-NH2 (TFLLR-NH2, 30 µM, Tocris); N-Acetyl-Asp-Glu (NAAG, 100 µM); (1R, 4R, 5S, 6R)-4-Amino-2-oxabicyclo [3.1.0] hexane-4,6-dicarboxylic acid disodium salt (LY379268, 100 µM, Tocris); calmidazolium chloride (CMZ, 2 µM, Tocris); methyl arachidonyl fluorophosphonate (MAFP, 10 µM); bromoenol lactone (Bel, 10 µM); AA (50 µM); N-(2,6-dimethylphenylcarbamoylmethyl) triethylammonium chloride (QX314, 1 mM, Tocris); 4-(4-,9-diethoxy-1,3-dihydro-1-oxo-2H-benz[f]isoindol-2-yl)-N-(phenylsulfonyl) benzeneacetamide (GW627368X 3 µM, Tocris); 6-Isopropoxy-9-xanthone-2-carboxylic acid (AH6809, 10 µM, Tocris); N-(piperidin-1-yl)-5-(4-iodophenyl)-1-(2,4-dichlorophenyl)-4-methyl-1H-pyrazole-3-carboxamide (AM251, 5 µM, Tocris); Tetrodotoxin (TTX, 0.5 µM, Tocris); 1,2-bis(2-aminophenoxy)ethane-N,N,N’,N’-tetraacetic acid (BAPTA, 50 µM, Tocris); and 1,2-Bis(2-aminophenoxy)ethane-N,N,N’,N’-tetra-acetic acid tetrakis (acetoxymethylester) (BAPTA-AM, 20 µM). All chemicals were from Sigma unless otherwise noted.
Statistical analysis of data
Statistical significance was evaluated by one-way ANOVA and post hoc Tests (Tukey and Dunn) using Prism software and deemed significant when P<0.05 for the [3H]-AA and PGE2 assay experiments. Normality of the data was evaluated by the Shapiro-Wilk test with a = 0.05. For electrophysiology experiments, significance was determined by paired or unpaired t-tests or Tukey-Kramer post hoc multiple comparison tests. HPLC/MS/MS lipidomics data were analyzed with ANOVA and Fishers LSD post-hoc test using SPSS when P<0.05 or P<0.10. All results are reported as mean ± s.e.m.
RESULTS
Ca2+-independent release of [3H]-AA and its metabolites from cultured astrocytes
We first assessed the efficiency by which preloading with an inhibitor of the endoplasmic reticulum (ER) Ca2+ pump, cyclopiazonic acid (CPA) (20 µM), or the cytosolic Ca2+ chelator 1,2-bis(o-aminophenoxy) ethane-N,N,N′,N′-tetraacetic acid (acetoxymethyl ester) (BAPTA-AM) (20 µM) was able to block ATP (100 µM) induced cytosolic Ca2+ increases in cultured rat astrocytes. Imaging cytosolic Ca2+ (AAV GFAP cyto-GCaMP6f) showed that the purine agonist ATP (100 µM) induced a prompt increase in Ca2+ that was completely blocked in CPA and BAPTA loaded cultures, whereas 10 µM methylarachidonyl fluorophosphate (MAFP), a nonspecific inhibitor of both cPLA2 and iPLA2, or 10 µM of bromoenol lactone (Bel) (Cornell-Bell et al.), a specific inhibitor of iPLA2, did not affect ATP induced Ca2+ rises in astrocytic culture (Figure 1A-C).
As a broad-based approach to AA-specific lipidomics, Ca2+-independent release of AA and/or its metabolites was performed using a [3H]-AA assay (Figure 2A). Cultured rat astrocytes were pre-incubated overnight with [3H]-AA, providing sufficient time for its incorporation into multiple biosynthetic and metabolic pathways. This would include incorporation into membrane phospholipids, which are precursors for endocannabinoids, and related lipids that are themselves precursors for AA release (Chen and Chen, 1998; Strokin et al., 2003). Therefore, this assay serves to determine if any AA precursors or metabolites derived from the [3H]-AA incorporated into the cell are being released; hereafter we refer to the composite of [3H]-labeled AA metabolites as [3H]-AA. ATP alone failed to induce a detectable release of [3H]-AA (Figure 2B); however, in cultures pretreated for 10 to 12 minutes with CPA or BAPTA-AM, ATP evoked a robust increase in the release of [3H]-AA, whereas CPA alone had no such effect (Figure 2B). Similarly, we found that upon stimulation with a combination of the non-selective mGluR agonist tACPD (100 µM) and the ionotropic glutamate receptor agonist AMPA (100 µM), we observed a significant release of [3H]-AA when cytosolic Ca2+ was blocked with CPA, but not in cultures without CPA pretreatment (Figure 2C). These observations confirm that rat astrocytes can release AA derivatives, key precursors of bioactive eicosanoids (Strokin et al., 2003; Rosenegger et al., 2015), but that the AA release is, surprisingly, inhibited by increases in cytosolic Ca2+.
To explore the mechanism of Ca2+-independent release of [3H]-AA, we next evaluated whether inhibition of the Ca2+-sensitive cPLA2 or the Ca2+-insensitive iPLA2 enzymes would reduce [3H]-AA release. Pretreatment of astrocytes with 10 µM MAFP significantly decreased the ATP-evoked release of [3H]-AA in cultures exposed to CPA (Figure 2D). Similarly, 10 µM BEL treatment suppressed the release of [3H]-AA, thus confirming a role for iPLA2 in Ca2+-independent lipid release (Figure 2D). The [3H]-AA release was only observed when agonist-induced increases in cytosolic Ca2+ were blocked (Figures 2B-C), suggesting that intracellular Ca2+ inhibits the Ca2+-independent iPLA2.
Calmodulin is a potent Ca2+-dependent inhibitor of iPLA2 (Wolf and Gross, 1996). To assess the interaction between calmodulin and iPLA2, the cells were treated with calmidazolium (CMZ), an inhibitor of Ca2+/calmodulin interactions that has been shown to remove the calmodulin block of iPLA2 (Wolf and Gross, 1996). In the presence of CMZ (2 μM), ATP treatment led to a significant release of [3H]-AA, which was comparable to the release upon blocking increases in cytosolic Ca2+ by preloading astrocytes with either CPA or BAPTA (Figure 2D). This observation suggests that Ca2+ acts primarily as a brake through calmodulin, which effectively inhibits iPLA2 activity (Wolf and Gross, 1996; Wolf et al., 1997). These findings are consistent with previous studies showing that iPLA2 is involved in receptor-mediated AA release from pancreatic islet cells (Gross et al., 1993), smooth muscle cells (Lehman et al., 1993), and endothelial cells (Seegers et al., 2002). Taken together, these data provide evidence of a new signaling mechanism in astrocytes through Ca2+-independent iPLA2, which is the major PLA2 isoform in the brain, accounting for 70% of total PLA2 activity (Yang et al., 1999).
Targeted lipidomics reveals Ca2+-independent lipid production in cultured astrocytes
Using lipid extraction and partial purification methods coupled to HPLC/MS/MS, we performed targeted lipidomics screening on cultured astrocytes that were preincubated with CPA and then challenged with ATP, versus vehicle control astrocytes. Of the 50 lipids screened, 30 were present in each of the samples and could thus be used for comparative analyses. Figure 3A lower panel lists all of the concentrations as mean ±SEM (Figure 3A). Figure 3A summarizes those lipids showing significant concentration differences, as well as the magnitudes of the differences. Of the 30 lipids detected, 16 (including AA and PGE2) increased upon ATP challenge in the presence of CPA. Figures 3B-C are representative chromatograms from the HPLC/MS/MS methods used to detect PGE2. Notably, 5 of the 8 AA-derivatives in the set were significantly increased by ATP challenge with CPA preincubation, including docosahexaenoyl ethanolamine, although the most dramatic increases were of the prostaglandin, PGE2. Figure 3D shows representative concentration differences in PGE2, DEA, and AA.
Ca2+-independent release of PGE2 from cultured astrocytes
Given that Ca2+-dependent astrocyte lipid release has been previously implicated in vasoregulation (Zonta et al., 2003; Takano et al., 2006; Gordon et al., 2008), it was surprising that PGE2 was released via a Ca2+-independent mechanism. Our HPCL/MS/MS method required large quantities of cells to detect PGE2, whereas the PGE2 ELISA assay accurately detects PGE2 release in 24-well cultures. Therefore, we used ELISA to explore and define the Ca2+ dependence of astrocytic PGE2 release. Because iPLA2 is essential for Ca2+-independent liberation of AA-derived lipids, we first assessed whether knockdown of iPLA2 via viral transduction would inhibit astrocytic release of PGE2. In the presence of CPA (20 μM), ATP (100 μM) failed to induce PGE2 release in shRNA-transduced astrocytic cultures, whereas there was a significant increase in the release of PGE2 by ATP in the presence of CPA in control shRNA cultures (Figure 4A). Knockdown of iPLA2 via shRNA viral transduction was confirmed with western blot analysis (Figure 4B).
Since cultured astrocytes express mGluR5 (Balazs et al., 1997; Silva et al., 1999; Gebremedhin et al., 2003), we next tested whether the agonists tACPD (100 μM) and AMPA (100 μM) in the presence/absence of CPA evoked PGE2 release. In the absence of CPA, these agents evoked little to no release of PGE2 (Figure 4C). However, when CPA was used to block the release of Ca2+ from internal stores, tACPD and AMPA significantly increased PGE2 release (Figure 4C).
The most abundant mGluR receptors expressed by astrocytes are mGluR5 and mGluR3 (Petralia et al., 1996; Aronica et al., 2000; Tamaru et al., 2001). However, mGluR5 is developmentally regulated and is not expressed by astrocytes in the adult brain, whereas mGluR3 is persistently expressed at high levels throughout adulthood (Sun et al., 2013). It is important to establish whether activation of mGluR3 can induce Ca2+-independent PGE2 release, since activation of this receptor has recently been shown to induce Ca2+ transients in adult hippocampal astrocytes (Haustein et al., 2014; Tang et al., 2015). In the presence of CPA, the mGluR3 agonists NAAG and LY379268 (Wroblewska et al., 1997; Bond et al., 1999) evoked a significant increase in the release of PGE2, whereas the same agonists failed to release PGE2 in the absence of CPA (Figure 4D). Notably, this result shows that activation of an astrocytic Gi-coupled receptor can release gliotransmitters by a Ca2+-independent mechanism.
To extend the observation to human astrocytes, we performed the same experiments in primary cultured astrocytes harvested from human embryonic tissue (Windrem et al., 2004; Windrem et al., 2008; Han et al., 2013). After pharmacologically assessing iPLA2 activity in rat astrocytes (Figure 2D), we evaluated iPLA2 expression in all the culture models. Western blot analysis showed that iPLA2 expression was not limited to rat astrocytic cultures, but was also expressed in human and mouse astrocytes (Figure 4E). In the presence of CPA, ATP evoked a significant PGE2 release from cultured human astrocytes, whereas little to no release was observed with ATP alone (Figure 4F). Likewise, co-application of tACPD and AMPA showed the same effect (Figure 4F). Taken together, these findings show that ATP or mGluR3 activation leads to Ca2+-independent PGE2 release and that iPLA2 is expressed in mouse, rat, and human astrocyte cultures.
Ca2+-independent astrocytic lipid release enhances mEPSPs via Kv channel blockade
Thus far, experiments have demonstrated that cultured astrocytes in response to agonist exposure can indeed release lipids in a Ca2+-independent pathway. In fact, the in vitro analysis showed that agonist induced Ca2+ increases served to impede PGE2 release, whereas CPA, BAPTA, and CMZ pretreatment potentiated PGE2 release.
Earlier studies have shown that AA and its metabolite PGE2 inhibit neuronal Kv currents (Horimoto et al., 1997; Nicol et al., 1997; Evans et al., 1999) and thereby enhance excitability (Sekiyama et al., 1995; Chen and Bazan, 2005; Sang et al., 2005). We therefore evaluated whether astrocytic lipid release also inhibits neuronal Kv current and enhances neuronal excitability in acute hippocampal slices. We performed dual patch-clamp recordings of pairs of CA1 pyramidal neurons and astrocytes in acute hippocampal brain slices prepared from 12–18-day old mice (Figure 5A). To isolate transiently active potassium currents, we used 100 ms voltage ramps, which is sufficiently fast to capture transient A-type currents (Phillips et al., 2018). It is worth noting that the I-V ramp enables one to discern what voltage deflection in outward current a drug might affect (Jackson and Bean, 2007). Kv currents in CA1 neurons are an assay for astrocytic lipid release since these currents are sensitive to AA and/or its metabolites (Villarroel and Schwarz, 1996; Carta et al., 2014). We isolated the Kv currents by adding 1 mM QX314 in the patch pipette to block sodium channels (Talbot and Sayer, 1996; Kim et al., 2010) and by imposing a voltage ramp (from -100 mV to 50 mV) every 5 seconds to continuously monitor changes in Kv channel current (Ji et al., 2000; Rangroo Thrane et al., 2013b; Carta et al., 2014) as an assay for agonist-induced astrocytic lipid release (Figure 4A). As expected, direct puffing of PGE2 (50 μM) or AA (50 μM) from a micropipette significantly reduced the Kv current, in keeping with previous studies (Evans et al., 1999; Carta et al., 2014) (Figure 5B-C).
To assess the effects of astrocytic Ca2+ signaling on neighboring neuronal Kv channels, we first stimulated astrocytes with ATP or TFLLR-NH2, an agonist of protease-activated receptor-1 (PAR1) that is primarily expressed by astrocytes (Shigetomi et al., 2008). ATP (100 μM) induced a comparable change in neuronal Kv when BAPTA (50 μM) was present in the astrocyte patch pipette, but not in its absence (Figure 5B-C). A similar transient decrease in Kv current was observed when astrocytes were activated by TFLLR-NH2 (30 μM). Again, the reduction in neuronal Kv currents was only detected when TFLLR was applied, and when BAPTA was added to the astrocytic pipette solution (Fig. 5B-C). To selectively activate astrocytic Ca2+ signaling, we next employed acute hippocampal slices from MrgA1 transgenic animals (MrgA1+/-), which express the exogenous Gq-coupled MRG receptor (MrgA1) under control of the GFAP promoter. The MrgA1 agonist Phe-Met-Arg-Phe amide (FMRF) mobilizes intracellular astrocytic Ca2+ stores, enabling assessment of the effects of astrocytic Ca2+ signaling on neighboring neuronal Kv currents. Although FMRF (15 μM) induces potent and selective increases in astrocytic cytosolic Ca2+ in hippocampal slices (Fiacco et al., 2007; Agulhon et al., 2010; Wang et al., 2012), we saw no detectable changes in neuronal Kv current when the pipette used to patch the astrocyte did not contain BAPTA (Figure 5B-C). However, when BAPTA was present in the astrocyte patch pipette, there was a marked decrease in neuronal Kv current evoked by FMRF exposure (Figure 5B-C). Thus, astrocytes can modulate neuronal Kv current via a previously undocumented Ca2+-independent lipid release.
To assess whether a decrease in neuronal Kv current is a consequence of Ca2+-independent astrocytic lipid release, we next employed specific lipid receptor antagonists for PGE2 and endocannabinoids. Since PGE2 receptors 1, 2, 3, and 4 are expressed on hippocampal pyramidal neurons (Andreasson, 2010; Maingret et al., 2017), we employed AH6809, a PGE2 EP1, 2, and 3 antagonist (Abramovitz et al., 2000; Ganesh, 2014), and GW627368, a PGE2 EP4 antagonist (Jones and Chan, 2005; Wilson et al., 2006). With AH6809 (10 μM) and GW627368X (3 μM) in the perfusion solution, the TFLLR, FMRF, and ATP induced Ca2+-independent PGE2 decrease in neuronal Kv current was abolished (Figure 5C); In contrast, the CB1 antagonist AM251 (5 μM) failed to abolish the decrease in Kv current (Figure 5C). Taken together, these data suggest that the observed decrease in neuronal Kv current is a result of astrocytic Ca2+-independent release of PGE2.
Interestingly, the presence or absence of BAPTA in the astrocytic pipette solution also affected agonist-induced changes in neuronal membrane potential (Figure 6A). Without BAPTA in astrocyte pipettes, TFLLR induced a hyperpolarization (−2.4 ± 0.26 mV) (Figure 6A), which has been linked previously to a decrease in extracellular potassium (Wang et al., 2012). However, with BAPTA present in astrocyte pipettes, TFLLR induced depolarization (2.1 ± 0.25 mV) (Figure 6A), an effect possibly attributable to the blockage of potassium current.
PGE2 has previously been shown to enhance neuronal mEPSPs (Sekiyama et al., 1995) (Sang et al., 2005). We found that TFLLR-induced activation of astrocytes triggered a decrease in mEPSP amplitude and frequency in normal mouse acute hippocampal slices (Figure 6B). The opposite effect, an increase in the amplitude and frequency of mEPSPs, was observable only when BAPTA was added to the astrocytic pipette solution (Figure 6C). This observation supports the notion that astrocytic Ca2+-independent lipid release may function as a signaling mechanism capable of modulating proximal synaptic activity by blocking Kv channels, which in turn may increase the frequency and amplitude of mEPSPs. Interestingly, TFLLR’s opposing effects on mEPSP amplitude and frequency were dependent upon astrocytic Ca2+ levels (Figure 6D-E), suggesting that Ca2+ depletion provides a pathway favoring the enhancement by astrocytic PGE2 of neuronal excitability.
DISCUSSION
Lipidomics has the potential to open exciting new avenues within the field of gliotransmission. In the present study, we utilized a series of lipidomics methodologies to show that Ca2+-chelation followed by metabotropic glutamate or purinergic receptor stimulation provokes the formation of a variety of lipids (Figures 2, 3) in astrocytes and drives the release of PGE2 in rat and human embryonic astrocyte cultures (Figure 4). In addition, we demonstrated that receptor-mediated Ca2+-independent PGE2 release serves to modulate neuronal Kv channels, resulting in enhanced synaptic activity in slices detected as increased mEPSP amplitude and frequency (Figures 5, 6). Multiple lines of evidence presented herein also support our hypothesis that this agonist-induced and Ca2+-independent lipid release from astrocytes depends on activation of iPLA2. These observations represent, to our knowledge, the first demonstration of Ca2+-independent release of lipid gliotransmitters, and thus add a novel dimension to our understanding of glial-neuronal communication.
Importance of iPLA2 pathway in Ca2+-independent astrocytic lipid production
Although astrocytes are known to constitutively release AA and other fatty acids (Moore, 2001; Bouyakdan et al., 2015), the pathways involved in constitutive AA release are currently not known, but inverse agonists of purinergic receptors (Ding et al., 2006) and mGluR (Carroll et al., 2001) might serve as useful pharmacological tools for investigating the constitutive PLA2 activity in relation to gliotransmission. To date, many studies have shown the multifaceted functions of PLA2 in the CNS, but iPLA2 accounts for 70% of the PLA2 activity in the rat brain (Yang et al., 1999). Although the effects of cPLA2 activation are best documented (Malaplate-Armand et al., 2006; Schaeffer and Gattaz, 2007; Kim et al., 2008), iPLA2 has also been shown to participate in phospholipid remodeling (Sun et al., 2004), and to regulate hippocampal AMPA receptors involved in learning and memory (Menard et al., 2005). Furthermore, iPLA2 regulates store-operated Ca2+ entry in cerebellar astrocytes (Singaravelu et al., 2006), and provides neuroprotection in an oxygen-glucose deprivation model (Strokin et al., 2006). Astrocytes express both isoforms of PLA2 (Sun et al., 2005), and numerous studies document the release of AA and its metabolites under regulation of Ca2+-dependent PLA2 (cPLA2) signaling in astrocytes (Bruner and Murphy, 1990; Stella et al., 1994; Stella et al., 1997; Chen and Chen, 1998). Previous studies have demonstrated astrocytic release of DHA via iPLA2 (Strokin et al., 2003, 2007). Here, we focused on the largely unexplored iPLA2 lipid pathway mediating PGE2 release from astrocytes. Given the ubiquitous expression of iPLA2 in astrocytes throughout the brain, we contend that various receptor-activated pathways converge on iPLA2 for the regulation of astrocyte lipid release (Figure 7).
Ca2+-independent astrocytic PGE2 release for rapid modulation of neuronal Kv channels
An important aspect of astrocytic gliotransmitter release is its timing. Agonist-induced astrocytic Ca2+ increases occur on a slow time scale of seconds (Cornell-Bell et al., 1990; Wang et al., 2006; Srinivasan et al., 2015), which allows integration of many factors on that time scale. Increased intracellular Ca2+ is a key step in the release of gliotransmitters (Parpura et al., 1994; Bezzi et al., 1998; Kang et al., 1998), such as ATP (Coco et al., 2003; Parpura and Zorec, 2010; Illes et al., 2019) and D-Serine (Mothet et al., 2000; Yang et al., 2003; Li et al., 2018; Neame et al., 2019). We have recently shown that agonist-induced astrocytic Ca2+ signaling can modulate synaptic activity by promoting K+ uptake, resulting in a transient lowering of extracellular K+ and depression of synaptic activity (Wang et al., 2012). More recently, we demonstrated that astrocytes can modulate a rapid form of synaptic activity (≥ 500ms) via 2AG release upon agonism of mGluR3 receptors (Smith et al., 2019). Because receptor-mediated astrocytic Ca2+ drives K+ uptake and release of gliotransmitters, these processes must, therefore, occur over a relatively prolonged time course (>500 ms). However, agonist-induced Ca2+-independent iPLA2 lipid release does not require the mobilization of intracellular Ca2+ stores, thus imparting the potential to signal on a much faster time scale, possibly within 10s of milliseconds. We speculate that iPLA2-mediated lipid release may act as a feedback system that enhances fast synaptic transmission in the short-term or under conditions of minimal neuronal activity, whereas subsequent activity-mediated increases in Ca2+ serve as a brake in the form of the calmodulin-dependent inhibition of iPLA2. The slower Ca2+-dependent release of gliotransmitters and stimulation of K+ uptake seems more suited toward the slow and widespread modulation of brain activity (typically inhibition) that occurs in the setting of, for example, tonic activation of the locus coeruleus and consequent norepinephrine release (Bekar et al., 2008; Ding et al., 2013).
Voltage-gated potassium channels are located on the dendrites of hippocampal pyramidal neurons (Johnston et al., 2000), where they play a major role in controlling dendritic excitability by modulating the amplitude of EPSPs. A morphological study found that the density of Kv channels in the dendrites of pyramidal neurons increased 5-fold proceeding from the soma to the most distal point measured in the apical dendrites (Hoffman et al., 1997). Inhibition of voltage-gated K+ currents will consequently enhance EPSPs, possibly explaining why PGE2 enhances synaptic transmission and LTP in the hippocampus (Sang et al., 2005).
Lipidomics models as a guide for future gliotransmitter discoveries
Lipidomics is one of the fastest-growing branches of the metabolomics field, and offers the possibility of describing the enormous diversity of lipid species throughout the body, and especially in the brain, which is largely composed of lipids (Sinclair, 1975). The complex and fine structure of astrocytic processes gives astrocytes a larger surface area to volume ratio than most other cell types, which makes them uniquely responsive to changes in the extracellular milieu. HPLC/MS/MS lipidomics techniques have provided a means to quantify specific lipid species with an accuracy and sensitivity not possible with more traditional methods employing radio-labeled fatty acids and ELISA. Here, we have combined these classical techniques with lipidomics to test the hypothesis that lipids are produced and released from astrocytes in a Ca2+-independent manner. Although our present focus was on the release of PGE2, we found upregulated release of 15 additional lipids. Notably, among the 30 lipids measured in the present targeted HPLC/MS/MS assay, 14 did not change upon activation of glial receptors, supporting the specificity of GPCR stimulated, Ca2+-independent lipid release. Interestingly, two of the lipids released in response to agonist exposure, namely N-arachidonyl taurine and N-palmitoyl tyrosine, were recently shown to activate TRPV4, a cation channel involved in osmotic sensitivity and mechanosensitivity (Raboune et al., 2014). Emerging evidence suggests that a TRPV4/AQP4 complex regulates the response to hypo-osmotic stress in astrocytes (Benfenati et al., 2011). The present lipidomics data thus casts light on unpredicted signaling mechanisms that involve astrocytic-derived lipid modulators (Figure 3).
Potential physiological relevance of calcium-independent lipid release
Astrocytes are now known to respond to single experimental stimulation events with small calcium increases in their distal fine processes (Panatier et al., 2011). This results from synchronous firing of as many as 50-1000 synapses in the astrocytes microdomain, as may occur when a large stimulation is applied as far as 500 µm away. However, the spontaneous activity of asynchronous synaptic events need not elicit astrocyte calcium events. Metabotropic GluR- and AMPA-mediated activation of astrocytes may elicit iPLA2 activation and the concerted liberation of AA metabolites and PGE2 to affect pre- and post-synaptic potassium channels on nearby neurons. When stimulation frequency and/or amplitude increase, astrocytic calcium rises would then serve to suppress over-activation of synapses. In this way, astrocytes may help maintain the strength of relatively quiescent synapses in a calcium-independent fashion, while employing multiple calcium-dependent mechanisms to suppress over-activation. Substantiation of this model requires rigorous testing of the spatial and temporal dynamics of astrocytic and individual dendritic activation, but are limited by current methodological capabilities that do not allow electrical recording of individual dendritic spines.
In conclusion, Ca2+-independent astrocytic lipid release constitutes a largely unexplored factor in the regulation of complex neuro-glial signaling interactions. Our present analysis adds a new dimension to the understanding of agonist-induced Ca2+ signaling by demonstrating that agonism of several Gq and Gi-linked astrocytic receptors can promote the release of lipid modulators, and that increases in cytosolic Ca2+ act as a brake to prevent PGE2 release.
Author contributions
F.W., H.B.B., and N.A.S. performed the experiments. F.W., H.B.B., J.X., and N.A.S. analyzed data. S.P. and B.J. performed Western Blots. S.G. and B.L. made cultures. L.B. and N.A.S. planned the experiments. D.C.M. provided Human Astrocytes. F.W., H.B.B., and N.A.S. wrote the paper.
Acknowledgements
We thank Dr. Ken McCarthy for generously sharing transgenic mice. This work was supported by the National Institutes of Health Grant K01NS110981, NSFNCS-FR 1926781, and Department of Defense Army Research Office Award W91NF2020189 to N.A.S. We thank Vittorio Gallo, Baljit Khakh, Bartosz Kula, Stefano Vicini, Alexander S. Thrane, Vinita Rangroo Thrane, Takahiro Takano, Paul Cumming, and Fernando R. Fernandez for comments and critical discussion for this manuscript. The authors declare no competing financial interest.
Footnotes
↵10 Lead Contact
This new version is revised to updated grammatical and data errors. This version also includes an updated title and figures.