Monoacylglycerol lipase inhibitor JZL184 prevents HIV-1 gp120-induced synapse loss by altering endocannabinoid signaling

Abstract

Monoacylglycerol lipase (MGL) hydrolyzes 2-arachidonoylglycerol to arachidonic acid and glycerol. Inhibition of MGL may attenuate neuroinflammation by enhancing endocannabinoid signaling and decreasing prostaglandin (PG) production. Almost half of HIV infected individuals are afflicted with HIV-associated neurocognitive disorder (HAND), a neuroinflammatory disease in which cognitive decline correlates with synapse loss. HIV infected cells shed the envelope protein gp120 which is a potent neurotoxin that induces synapse loss. Here, we tested whether inhibition of MGL, using the selective inhibitor JZL184, would prevent synapse loss induced by gp120.

The number of synapses between rat hippocampal neurons in culture was quantified by imaging clusters of a GFP-tagged antibody-like protein that selectively binds to the postsynaptic scaffolding protein, PSD95. JZL184 completely blocked gp120-induced synapse loss. Inhibition of MGL decreased gp120-induced interleukin-1β (IL-1β) production and subsequent potentiation of NMDA receptor-mediated calcium influx. JZL184-mediated protection of synapses was reversed by a selective cannabinoid type 2 receptor (CB₂R) inverse agonist/antagonist. JZL184 also reduced gp120-induced prostaglandin E2 (PGE₂) production; PG signaling was required for gp120-induced IL-1β expression and synapse loss.

Inhibition of MGL prevented gp120-induced synapse loss by activating CB₂R resulting in decreased production of the inflammatory cytokine IL-1β. Because PG signaling was required for gp120-induced synapse loss, JZL184-induced decreases in PGE₂ levels may also protect synapses. MGL presents a promising target for preventing synapse loss in neuroinflammatory conditions such as HAND.

Graphical abstract

1. Introduction

Monoacylglycerol lipase (MGL) hydrolyzes the endocannabinoid, 2-arachidonoylglycerol (2-AG), to arachidonic acid (AA) and glycerol (Blankman and Cravatt, 2013). In the brain, hydrolysis of 2-AG by MGL is the principal source of AA for the production of prostaglandins (PGs) (Nomura et al., 2011). Thus, inhibition of MGL can affect neuronal function by enhancing endocannabinoid signaling (Viader et al., 2015) and decreasing PG-mediated inflammation (Grabner et al., 2016; Nomura et al., 2011). Both of these actions might protect the CNS from excitotoxic insult. Activation of the endocannabinoid system protects neuronal function in animal models of epilepsy, stroke and Alzheimer’s disease (Chen et al., 2012; Pacher et al., 2006). Neuroinflammation impairs neuronal function (Heneka et al., 2015), and elevated PG levels contribute to neuronal dysfunction in animal models of status epilepticus (Jiang et al., 2013), pain (Zhao et al., 2007) and Alzheimer’s disease (Johansson et al., 2015). Here we examined the effects of MGL inhibition in an in vitro model of a neuroinflammatory disorder using JZL184, a potent and selective inhibitor of MGL (Grabner et al., 2016; Pan et al., 2009).

HIV-associated neurocognitive disorder (HAND) afflicts almost half of HIV-infected individuals (Ellis et al., 2007; Saylor et al., 2016). Cognitive decline in HAND correlates closely with synaptodendritic damage such as dendritic pruning and degradation of synaptic proteins (Ellis et al., 2007; Ellis et al., 2009). Because HIV does not infect neurons, HIV neurotoxicity is indirect and thought to be mediated by a neuroinflammatory response to viral proteins and inflammatory cytokines released by infected microglia and macrophages (Ellis et al., 2007; Kaul et al., 2001; Saylor et al., 2016). The HIV envelope protein, gp120, has been detected in the brain tissue of HAND patients (Jones et al., 2000; Nath, 2002), is shed by infected cells (Kaul et al., 2001), and is a potent neurotoxin (Toggas et al., 1994).

Synapse loss is the hallmark of HAND and gp120 induces significant loss of synapses in both primary neuronal cultures and transgenic mice (Iskander et al., 2004; Kim et al., 2011; Toggas et al., 1994). When applied to hippocampal cultures composed of microglia, astrocytes and neurons, gp120_IIIB binds to CXCR4 on microglia evoking the release of the inflammatory cytokine interleukin-1β (IL-1β) (Kim et al., 2011; Viviani et al., 2006). HIV gp120-induced synapse loss is blocked by an IL-1 receptor antagonist (Kim et al., 2011). Whether modulating endocannabinoid tone affects this process of neuroinflammatory synapse loss is not known. Thus, gp120-induced loss of synapses between hippocampal neurons in culture provides a model to study the potential for inhibition of MGL to protect synaptic function.

The present study demonstrates that JZL184 completely protects from gp120-induced synapse loss. The contributions of enhanced endocannabinoid tone and reduced PGE₂-mediated neuroinflammation to synapse protection were determined. The dual mechanisms of protection that result from MGL inhibition might be particularly beneficial in neurodegenerative disorders with a strong neuroinflammatory component like HAND.

2. Materials and Methods

2.1 Materials

Drugs used in this study and their pharmacological targets are summarized in supplementary Table 1. Materials were obtained from the following sources: JZL184 from the NIDA Drug Supply Program (Research Triangle Institute, Research Triangle Park, NC, USA) and Cayman Chemical (Ann Arbor, MI, USA); IL-1β and IL-1ra from R&D System (Minneapolis, MN, USA); AM630, AH6809, JZL 195, LY320135, rimonabant, and SR144528 from Tocris Bioscience (Bristol, UK); arachidonoyl-AMC from Enzo Life Sciences (Farmingdale, NY, USA); 4-nitrophenylacetate (4-NPA) from Cayman Chemical (catalog number: 705193; Ann Arbor, MI, USA); Dulbecco’s modified Eagle’s medium (DMEM), Hanks’ balanced salt solution, fetal bovine serum, horse serum, penicillin/streptomycin, fura-2 AM, and glycine from Invitrogen (Carlsbad, CA, USA). The PSD95.FingR-eGFP expression vector (pCAG-PSD95.FingR-eGFP) was generated in the laboratory of Don Arnold and obtained from Addgene (catalog number: 46295; Cambridge, MA, USA). The expression vector for tdTomato driven by the synapsin promoter was generated by excising tdTomato from pLVX-tdTomato-N1 (Clontech-Takara Bio, Mountain View, CA, USA) and inserting it into the pSyn backbone of pSyn-PSD95-GFP kindly provided by Kirill Martemyanov (Scripps Research Institute, Jupiter, FL, USA). HIV-1 gp120_IIIB was obtained through the National Institutes of Health (NIH) AIDS Reagent Program, Division of AIDS, National Institute of Allergy and Infectious Diseases, NIH (catalog number:11784).

2.2 Cell culture

All animal care and experimental procedures conformed to the Guide for the Care and Use of Laboratory Animals published by the U.S. National Institutes of Health. Ethical approval was granted by the Institutional Animal Care and Use Committee of the University of Minnesota (protocol 1612-34372A). Primary neuronal cultures were prepared from fetal tissue collected from 70 timed pregnant Sprague Dawley rats (250–300 g when mated) supplied by Charles River Laboratories (Raleigh, North Carolina, USA). Because multiple dishes of cells are produced from a single plating, this method reduced the number of animals required to complete the study. Prior to tissue collection, the animals were housed in the University of Minnesota vivarium for 2–6 d at constant temperature and humidity on a 12 h light/dark cycle, with free access to water and standard rat chow.

Primary rat hippocampal neurons were grown as described previously (Waataja et al., 2008) with minor modifications. Dams were killed by CO₂ inhalation in an institutionally approved and calibrated CO₂ chamber. Embryonic day 17 fetuses were rapidly decapitated with sharp scissors then hippocampi dissected and placed in Ca²⁺ and Mg²⁺ free Hanks’ Balanced salt solution. Hippocampi were suspended in DMEM without glutamine, supplemented with 10% fetal bovine serum and penicillin/streptomycin (100 U mL⁻¹ and 100 µg mL⁻¹, respectively) and dissociated by trituration through flame-narrowed Pasteur pipettes of decreasing aperture. Dissociated cells were then plated onto either a 25 mm round cover glass or a 25 mm round cover glass glued to cover a 19 mm diameter opening drilled in the bottom of a 35 mm Petri dish. The cover glass was pre-coated with Matrigel (150 µL, 0.2 mg mL⁻¹) (BD Biosciences, Billerica, MA, USA). Cells were grown in a humidified atmosphere of 10% CO₂ and 90% air (pH 7.4) at 37 °C. On day 1 and day 8 in vitro, cells were fed by exchanging 75% of the media with DMEM, supplemented with 10% horse serum and penicillin/ streptomycin. Cells used in this study were grown for 12 to 16 days in vitro without mitotic inhibitors, resulting in a mixed glial-neuronal culture. Using previously described immunocytochemistry methods (Kim et al., 2011), we found that these cultures are composed of 24 ± 4% neurons, 55 ± 4% astrocytes and 13 ± 6% microglia.

2.3 MGL Activity Assay

MGL activity was measured using a previously described assay (Muccioli et al., 2008) with minor modifications. Rat hippocampal cultures were untreated or pretreated with various concentrations of JZL184 for 24 h. After pre-treatment, cells were scraped and collected in buffer containing 50 mM HEPES (pH 7.4), 1 mM EDTA, 1 µM pepstatin, 100 µM leupeptin, and 0.1 mg mL⁻¹ aprotinin. The cells were sonicated on ice five times for 10 s at 15 s intervals and then centrifuged at 100,000 × g for 50 min at 4. After resuspending the pellet, the membranes were mixed with the colorimetric substrate for lipid hydrolases 4-NPA (catalog number: 705193; Cayman Chemical, Ann Arbor, MI, USA) to a final concentration of 202 µM. The mixture was transferred immediately into an Infinite M1000 PRO Microplate Reader (Tecan, Männedorf, Switzerland), and absorbance at 405 nm was monitored every 5 min for 180 min at 37 °C. Protein concentration was measured using Pierce Coomassie (Bradford) Protein Assay Kit (catalog number: 23200; ThermoFisher Scientific, Minneapolis, MN, USA) according to the manufacturer’s instructions. Enzyme activity was determined from the linear phase of the reaction (15–55 min) and normalized to protein concentration.

2.4 FAAH Activity Assay

FAAH activity was measured using a previously described assay (Ramarao et al., 2005) with minor modifications. Rat hippocampal cultures were untreated or pretreated with 1 µM JZL195, or 1 µM JZL184 for 24 h. After pre-treatment, cells were scraped and collected in buffer containing 50 mM HEPES (pH 7.4), 1 mM EDTA, 1 µM pepstatin, 100 µM leupeptin, and 0.1 mg mL⁻¹ aprotinin. The cells were then sonicated on ice five times for 10 s at 15 s intervals and then mixed with the fluorogenic substrate of FAAH, arachidonoyl-AMC (Enzo Life Science, Farmingdale, NY, USA), to a final concentration of 100 µM. The mixture was transferred immediately into an Infinite M1000 PRO Microplate Reader (Tecan, Männedorf, Switzerland), and fluorescence monitored every 5 min for 200 min at 37 °C (355 nm excitation, 460 nm emission). Enzyme activity was determined from the linear phase of the reaction (0–75 min) and normalized to protein concentration.

2.5 Transfection

Transfection of cultured rat hippocampal neurons was conducted between 11 and 12 days in vitro using a previously described protocol with minor modifications (Kim et al., 2011). Briefly, a DNA/calcium phosphate precipitate containing 0.5 µg of total plasmid DNA per well was prepared and allowed to form for 90 min at room temperature. The media (conditioned media) was exchanged with DMEM supplemented with 1 mM kynurenic acid, 10 mM MgCl₂, and 5 mM HEPES to reduce neurotoxicity. The DNA/calcium phosphate precipitate was added dropwise to the cells and allowed to incubate for 30 min. After the incubation, cells were washed twice with DMEM supplemented with 10 mM MgCl₂ and 5 mM HEPES to remove leftover precipitate. After washing, conditioned media that had been saved at the beginning of the procedure was returned to the cells. Experiments were started 48–72 h after transfection. The calcium phosphate method of transfection was chosen because of its low toxicity.

2.6 Confocal imaging and image processing

Transfected cells were imaged with an inverted laser scanning confocal microscope (Nikon A1, Melville, NY, USA) using a 60× (1.4 numerical aperture) oil-immersion objective. Petri dishes containing cells were place in a stage-top incubator (LCI, Seoul, Korea) and maintained at 37° C and 10% CO₂ during imaging. After imaging, the cells were then returned to the cell culture incubator and the coordinates of the computer controlled stage were saved to enable repeated imaging of the same neuron for 24–48 h. A custom machined jig allowed the Petri dish to be returned to the same location on the microscope stage. GFP was excited at 488 nm and emission collected from 500 to 550 nm. TdTomato was excited at 561 nm and emission collected from 570 nm to 620 nm. Optical sections spanning 8 µm in the z-dimension were collected (1 µm per step) and combined through the z-axis into a maximum z-projection. GFP puncta were counted in an unbiased manner using an algorithm written in MetaMorph 6.2 image processing software (Waataja et al., 2008). A threshold set 0.25 times 1 standard deviation above the image mean was applied to the tdTomato image. This created a 1 bit image, which was used as a mask via a logical AND function with the GFP maximum z-projection. A top-hat filter (80 pixels) was applied to the masked GFP image. A threshold set 1.5 standard deviations above the mean intensity inside the mask was then applied to the contrast enhanced image. Structures between 12 and 80 pixels (0.13–0.86 µm²) were counted as postsynaptic densities. The structures were then dilated and superimposed on the tdTomato maximum z-projection for visualization. The change in the number of PSD95.FingR puncta from 2–3 microscopic fields, each containing one tdTomato filled neuronal soma, from a single dish were averaged and defined as an individual sample (n=1).

2.7 [Ca²⁺]_i imaging and analysis

Intracellular Ca²⁺ concentration ([Ca²⁺]_i) was measured as previously described (Krogh et al., 2014). Briefly, cells were incubated in 10 µM fura-2 acetoxymethyl ester (fura-2 AM) in 0.04% pluronic acid in HEPES Hanks’ salt solution (HHSS), pH 7.45, for 30 min at 37 °C. HHSS contained the following (in mM): HEPES 20, NaCl 137, CaCl₂ 1.3, MgSO₄ 0.4, MgCl₂ 0.5, KCl 5.0, KH₂PO₄ 0.4, Na₂HPO₄ 0.6, NaHCO₃ 3.0, and glucose 5.6. After loading with indicator, cells were washed in HHSS without fura-2 AM at 37 °C for 10 min. The cover glass containing loaded and washed cells was transferred to a recording chamber, and placed on the stage of an IX71 microscope (Olympus, Melville, NY, USA). Cells were imaged using a 20× (0.75 numerical aperture) objective. [Ca²⁺]_i was measured by sequential excitation of fura-2 at 340 and 380 nm (8 nm slit width) and emission was collected from 490 to 530 nm; image pairs were collected every 1 s. Cells were superfused at a rate of 2 mL min⁻¹ with HHSS and responses evoked by exchanging the bath with Mg²⁺-free HHSS containing 10 µM NMDA and 200 µM glycine for 60 s. After background subtraction, the 340 and 380 nm image pairs were converted to [Ca²⁺]_i using the formula [Ca²⁺]_i = K_dβ(R-R_min)/(R_max –R) (Grynkiewicz et al., 1985). The dissociation constant (K_d) for fura-2 was 145 nM. β is the ratio of fluorescence intensity acquired with 380 nm excitation measured in Ca²⁺-free buffer (1 mM EGTA) and buffer containing saturating Ca²⁺ (5 mM). R is the fluorescence intensity ratio of images collected at 340 nm and 380 nm excitation. R_min, R_max, and β were determined in a series of calibration experiments on intact cells. R_min and R_max values were generated by applying 10 µM ionomycin in Ca²⁺-free buffer (1 mM EGTA) and saturating Ca²⁺ (5 mM), respectively. Values for R_min, R_max, and β were 0.37, 9.38, and 6.46, respectively. These calibration constants were applied to all experimental recordings. The neuronal cell body was selected as the region of interest for all recordings. All neurons within the imaging field were included in the analysis and no exclusions were made. An individual sample (n=1) was defined as the average change in [Ca²⁺]_i from all the neurons imaged on a single coverslip.

2.8 Treatments

In synaptic imaging experiments an initial image (t = 0 h) was collected then HIV gp120 and drug treatments were applied directly to the cell culture media. When present, JZL184, IL-1ra, and AH6809 were added 15 min prior to addition of gp120. Selective CB receptor inverse agonists/antagonists were applied 5 min prior to the addition of JZL184. The cells were then returned to the cell culture incubator for 24 h. A second image (t = 24 h) of the same field was collected and the change in the number of synapses presented as a percentage of the initial puncta count. For [Ca²⁺]_i imaging experiments HIV gp120, IL-1β and drug treatments were applied directly to the cell culture media. JZL184, IL-1ra, and AH6809 were added 15 min prior to addition of gp120 or IL-1β. [Ca²⁺]_i responses were evoked by superfusing 10 µM NMDA and 200 µM glycine for 60 s.

2.9 Immunocytochemistry

Hippocampal cultures were transfected with PSD95.FingR-eGFP after 11 days in vitro as described above. 48 h after transfection, cells were washed with PBS and then fixed with 4% paraformaldehyde for 10 min. Cells were then washed with PBS three times and permeabilized with 0.2% Triton X-100 (Sigma, St. Louis, MO, USA) for 10 min. Cells were blocked in 10% BSA and 0.2% Triton X-100 for 30 min at room temperature with slow shaking. After blocking, cells were incubated with mouse anti-Bassoon monoclonal antibody (1:200, Enzo Life Sciences, Farmingdale, NY, USA) in blocking buffer at 4 °C overnight. Cells were then washed 3 times with PBS and incubated with tetramethylrhodamine (TRITC)-conjugated goat anti-mouse antibody (1:500; Millipore, Billerica, MA, USA) in blocking solution at room temperature for 1 h. Cells were imaged after three washes. PSD95.FingR-eGFP was excited at 488 nm and emission was collected from 500 to 550 nm. TRITC was excited at 561 nm and emission was collected from 570 nm to 620 nm.

2.10 ELISA

Secreted PGE₂ levels in the culture media were determined using a commercially available PGE₂ ELISA kit (catalog number: ADI-930-001; Enzo Life Sciences, Farmingdale, NY, USA). Each sample was from a single well of a 6-well plate. Two reactions for each sample were run in parallel and averaged (n=1). The assays were performed according to the manufacturer’s instructions. Absorbance was read at 405 nm using an Infinite M1000 PRO Microplate Reader (Tecan, Männedorf, Switzerland).

2.11 Quantitative Reverse Transcription Real-Time PCR (qRT-PCR)

RNA was extracted from cultures grown in 6-well plates using a commercially available RNA extraction kit (RNeasy Plus Mini Kit; catalog number: 74134; QIAGEN, Hilden, Germany), according to the manufacturer’s instructions. Each sample was collected from 3 wells under the same treatment. For qRT-PCR, a SuperScript III Platinum SYBR Green One-Step qPCR Kit (catalog number: 11746; Invitrogen, Carlsbad, CA, USA) was used following the manufacturer’s recommendations with minor modifications. Briefly, for each reaction, 5 uL of SYBR Green Reaction Mix with ROX was mixed with 0.2 uL Taq, 1 uL extracted RNA (containing 40–50 ng RNA), 0.2 uL primers (to make a final concentration 200 nM). The reaction was performed using a StepOnePlus Real-Time PCR System (ThermoFisher Scientific, Minneapolis, MN, USA). Synthesis of cDNA (50 °C for 3 min) was immediately followed by PCR amplification (95 °C for 5 min, followed by 40 cycles of 95 °C for 15 s and 60 °C for 1 min). IL-1β cDNA was amplified using the following primers: 5’-TCCTTGTGCAAGTGTCTGAAG-3’ and 5’-GTCTGTCAGCCTCAAAGAACA-3’ (IDT Integrated DNA Technologies, Coralville, IA, USA). Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as internal reference using the following primers: 5’-AATGGTGAAGGTCGGTGTG-3’ and 5’-GTGGAGTCATACTGGAACATGTAG-3’ (IDT Integrated DNA Technologies, Coralville, IA, USA). For each sample, two IL-1β reactions and two GAPDH reactions were run in parallel and averaged (n=1). Quantitative analysis was performed using the 2^−ΔΔCt method.

2.12 Statistics

All data are presented as mean ± SEM. For all treatment groups, there were at least 6 samples. Data were first tested for unequal variance using Levene’s test (OriginPro v8.5, Northampton, MA, USA); no samples were found to be of unequal variance. For data sets with multiple comparisons a one-way or two-way ANOVA was performed. If gp120 treatment exerted a significant interaction with other treatment groups, or if only one treatment (gp120 alone) was used, statistical significance was determined using one-way ANOVA with a Tukey’s post-hoc test (Prism, GraphPad 5, La Jolla, CA 92037 USA). Time course data were analyzed with a repeated measures ANOVA using R under Rcmdr (Baier and Neuwirth, 2007).

3. Results

3.1 JZL184 inhibits rat MGL enzyme activity but not FAAH activity

JZL184 completely inhibits MGL in mouse brain cultures at a concentration of 1 µM (Grabner et al., 2016). Here we tested the effects of JZL184 on rat hippocampal cultures. Mixed cultures containing neurons, astrocytes and microglia were treated with various concentrations of JZL184 for 24 h, the cells were lysed and an enzyme assay performed on isolated membranes (Figure 1A). JZL184 inhibited the degradation of the colorimetric substrate 4-NPA with an IC₅₀ of 0.22 ± 0.06 µM (Figure 1B). A concentration of 1 µM produced maximal inhibition, consistent with previous studies (Grabner et al., 2016), and was used for subsequent experiments.

Figure 1.

JZL184 inhibits MGL enzyme activity but not FAAH activity. (A) Representative traces show MGL enzymatic activity during incubation with the colorimetric substrate, 4-NPA. Cultures were treated for 24 h with 0, 0.3 or 1 µM JZL184. Cells were lysed and incubated with 4-NPA as described in Materials and Methods. Absorbance was monitored every 5 min for 180 min and presented as relative absorbance units (RAU) normalized to µg protein. (B) Plot shows concentration dependent inhibition of MGL activity by JZL184. Enzyme activity was determined from the linear phase of the reaction (15–55 min in A). The concentration response curve was generated using a nonlinear, least-squares curve fitting program (Origin 6.0, OriginLab Corp.) to fit a logistic equation of the form % of Control = [(A₁ – A₂)/(1 + (X/IC₅₀)^p)]+A₂ where X = drug concentration, IC₅₀ = 219 ± 62 nM, A₁ = 95 ± 3 % inhibition without drug, A₂ = 65 ± 2 % inhibition at a maximally effective drug concentration and p = 3 ± 2 slope factor. Values are expressed as mean ± SEM from 5 separate enzyme assays. The incomplete inhibition produced by a maximal concentration of JZL184 is due to the hydrolysis of 4-NPA by lipid hydrolases other than MGL present in the crude membrane preparation isolated from neuronal cell cultures. (C) Representative traces show FAAH enzymatic activity during incubation with the fluorogenic substrate, arachidonoyl-AMC. Cultures were treated for 24 h with 1 µM JZL184 or 1 µM JZL 195, a potent inhibitor of FAAH and MGL. Cells were lysed and incubated with arachidonoyl-AMC as described below. Fluorescence was monitored every 5 min for 200 min and presented as relative fluorescence units (RFU) normalized to µg protein. (D) Bar graph summarizes FAAH activity relative to untreated cultures (control) during the linear phase of the reaction (0–75 min). FAAH activity in the presence of JZL184 was not significantly different from control. *p<0.05 relative to control (one-way ANOVA with Tukey’s post-test, n=6).

This concentration of JZL184 did not significantly influence the activity of fatty acid amide hydrolase (FAAH) (Figure 1C and D). Cultures were treated for 24 h with 1 µM JZL184 or 1 µM JZL195, a potent inhibitor of FAAH and MGL. Cells were lysed and FAAH enzymatic activity assessed by measuring the degradation of the fluorogenic substrate, arachidonoyl-AMC. Thus, the 1 µM concentration of JZL184 used in these studies is selective for MGL relative to FAAH.

3.2 PSD95.FingR-eGFP labels functional synapses

The HIV-1 envelope protein gp120 is a potent neurotoxin in vitro that mimics the synapse loss observed in HAND via a neuroinflammatory mechanism (Kim et al., 2011; Mishra et al., 2012). Here we examined the effects of inhibiting MGL with JZL184 on synapse loss induced by gp120 and assessed the mechanism of action. To monitor changes in the number of synapses, rat hippocampal cultures were transfected with expression plasmids for a recombinant antibody-like protein, PSD95.FingR-eGFP, and tdTomato, as previously described (Gross et al., 2013). Transfected cells were imaged using a laser scanning confocal microscope as described in Methods. The PSD95.FingR-eGFP construct expressed a GFP-tagged fibronectin intrabody that specifically bound to the endogenous postsynaptic scaffolding protein PSD95, labelling glutamatergic synapses in a punctate pattern (Figure 2A). The red tdTomato protein filled the cytoplasm enabling visualization of the transfected cell’s morphology. This method was previously shown to label functional synapses without interfering with normal synaptic function. In extensive characterization experiments Gross et al. (2013) showed that PSD95.FingR-eGFP expressed in primary neuronal cultures co-immunoprecipitated with native PSD-95, that the eGFP puncta co-localized with PSD-95 immunoreactivity, and that siRNA knock down of endogenous PSD-95 produced a comparable knockdown of PSD95.FingR-eGFP labelling. Here, we used immunocytochemistry to confirm that green fluorescent puncta labelled synaptic structures. Immunoreactivity for the presynaptic scaffolding protein Bassoon co-localized with PSD95.FingR labeled postsynaptic sites (Figure 2B). 83 ± 1 % (n=5) of the green fluorescent puncta co-localized with Bassoon immunoreactive puncta. To track individual synapses in live cells, image stacks compressed in the z dimension were processed using an algorithm that counted green fluorescent puncta that met size (area between 0.13 µm² and 0.86 µm²) and intensity criteria, and were in contact with a binary mask derived from the red tdTomato image. This method enabled the same neuron to be imaged before and after treatment to measure changes in the number of synapses.

Figure 2.

PSD95.FingR-eGFP (Intrabody) labels postsynaptic terminals at excitatory synapses. (A) Representative confocal images of a neuron expressing PSD95.FingR-eGFP and tdTomato were acquired and processed as described in Methods. PSD95.FingR-eGFP puncta were identified by filtering compressed z-stacks (8 µm) of confocal images. All puncta that met size (between 0.13–0.86 µm²) and intensity criteria, and were in contact with a binary mask derived from the tdTomato image were counted as synapses. Puncta were dilated and overlaid on the tdTomato maximum projection (Processed) for display purposes. Insets are enlarged images of the boxed regions. Scale bars represent 10 µm. (B) Representative confocal images of a neuron expressing PSD95.FingR-eGFP (green) and immunolabeled for Bassoon (red). PSD95.FingR-eGFP puncta co-localized with Bassoon immunoreactivity (yellow, Merged). Note that non-transfected cells were also present in the field, and thus not all Bassoon immunoreactive puncta (red) co-localized with PSD95.FingR-eGFP (green) puncta. Insets are enlarged images of the boxed regions. Scale bars represent 10 µm.

3.3 gp120-induced synapse loss was blocked by inhibition of MGL

Treating hippocampal neurons in culture with the HIV-1 envelope protein gp120_IIIB induces the loss of synaptic connections via a mechanism that requires microglial production of the inflammatory cytokine IL-1β (Kim et al., 2011). Treating the culture with 600 pM gp120 for 24 h produced a 15 ± 3% loss of PSD95.FingR labeled postsynaptic sites (Figure 3A). Synapse loss was maximal by 24 h and it persisted for 48 h (Figure 3A). The time course of gp120 induced synapse loss shown here is in good agreement with previous studies using a PSD95-GFP fusion protein to report glutamatergic synapses (Kim et al., 2011). Advantages of the intrabody approach relative to imaging the expressed fusion protein are that the intrabody construct incorporates a transcriptional control mechanism to reduce the expression of excess intrabody, reducing background fluorescence, and intrabody expression does not drive increased synapse formation as does the PSD95-GFP fusion protein.

Figure 3.

HIV gp120-induced loss of synapses and IL-1β expression were blocked by inhibition of MGL. (A) Graph shows time-dependent changes in the number of PSD95.FingR-eGFP puncta for untreated cells (control, circles) and cells treated with 600 pM gp120 (gp120, squares). Data are expressed as mean ± SEM. Repeated measures ANOVA revealed gp120 by time interactions [F_{3, 35} = 6.25, p<0.01]. *p<0.05, **p<0.01 relative to control at the same time-point (Tukey’s post-test, n = 7). (B) Bar graph summarizes changes in PSD95.FingR-eGFP puncta after 24 h treatment under control conditions (open bars) in the absence (n = 8) or presence (n = 6) of 1 µM JZL184 or treated with 600 pM gp120 (solid bars) in the absence (n = 7) or presence (n = 6) of JZL184. JZL184 was added 15 min prior to gp120 for all experiments. Data are expressed as mean ± SEM. *p<0.05 relative to untreated control, #p<0.05 relative to untreated gp120 (one-way ANOVA with Tukey’s post-test). (C) Representative processed images of neurons with no treatment (control), or treated with 600 pM gp120 in the absence or presence of 1 µM JZL184 for 24 h. The insets are enlarged images of the boxed region. Scale bars represent 10 µm. (D) Bar graph summarizes IL-1β mRNA expression relative to control (n = 12) after treatment with 600 pM gp120 and in the absence (n = 18) or presence of 1 µM JZL184 (n = 16) for 4 h. Data are expressed as mean ± SEM. **p<0.01 relative to control, ##p<0.01 relative to gp120 treated group (one-way ANOVA with Tukey’s post-test).

To determine whether altering endocannabinoid tone would modulate the process of gp120-induced synapse loss, we examined the effects of inhibiting the breakdown of the endocannabinoid 2-AG to AA and glycerol by MGL. Inhibition of MGL has been shown to suppress neuroinflammation induced by neurotoxins (Grabner et al., 2016; Nomura et al., 2011; Viader et al., 2015). To determine whether inhibition of MGL could attenuate gp120-induced synapse loss, the cultures were treated with the selective MGL inhibitor, JZL184, 15 min prior to and during exposure to gp120 (Figure 3B and C). JZL184 (1 µM) treatment completely blocked synapse loss induced by 24 h exposure to 600 pM gp120, suggesting that inhibition of MGL protects synapses from gp120-induced loss (Figure 3B and C). JZL184 alone did not affect the number of synapses.

3.4 JZL184 blocks gp120-induced IL-1β production

We next studied the effects of JZL184 on gp120-induced IL-1β production. In a previous study we showed that gp120 evoked the secretion of IL-1β protein, detected by ELISA, and increased the expression of IL-1β mRNA, detected by qRT-PCR (Kim et al., 2011). Here, we used the qRT-PCR assay because of its higher sensitivity. Gp120-induced increases in IL-1β mRNA levels were measured in the absence and presence of JZL184 to determine whether inhibition of MGL suppressed IL-1β production. Figure 3D shows that 4 h treatment with 600 pM gp120 increased the expression of IL-1β mRNA by 2.0 ± 0.2-fold. This increase was significant and completely blocked by treatment with JZL184 (1 µM). These results suggest that inhibition of MGL decreases IL-1β production which might contribute to the protection from gp120-induced synapse loss afforded by JZL184.

3.5 JZL184 blocks gp120-induced potentiation of NMDA-evoked Ca²⁺ influx

The HIV protein Tat and the inflammatory cytokine IL-1β potentiate NMDARs by activating Src family tyrosine kinases (Krogh et al., 2014; Viviani et al., 2003), which are known to phosphorylate NMDARs resulting in potentiated currents (Yu and Salter, 1999). Synapse loss induced by gp120 requires activation of NMDARs (Kim et al., 2011). Here we determined whether exposure to gp120 would increase the amplitude of NMDA-evoked changes in [Ca²⁺]_i. Fura-2-based [Ca²⁺]_i imaging was performed as described in Methods. Cells were superfused with HEPES Hank’s salt solution (HHSS) at a flow rate of 2 mL min⁻¹ for 60 s and then the bath was exchanged to Mg²⁺-free HHSS containing 200 µM glycine and 10 µM NMDA for 60 s. As shown in Figure 4A, NMDA evoked a transient increase in the [Ca²⁺]_i that rose from a mean basal level of 68 ± 3 nM to peak at 241 ± 25 nM in control cells (t = 0 h). Treatment with 600 pM gp120 produced a time-dependent increase in the amplitude of the NMDA-evoked response (Figure 4A and B). The gp120-induced potentiation of the NMDA-evoked response peaked after 4 h exposure to gp120 and then adapted. This time course is consistent with potentiated NMDA-mediated Ca²⁺ influx driving cellular adaptations such as synapse loss (Figure 3A) and downregulation of NMDAR-mediated synaptic currents (Green and Thayer, 2016). The net NMDA-evoked [Ca²⁺]_i response was increased by 113 % in cells treated for 4 h with 600 pM gp120 relative to control cells (Figure 4C). In the presence of IL-1ra (1 µg mL⁻¹), an IL-1 receptor antagonist, the gp120-induced potentiation was significantly reduced (Figure 4D). This result is consistent with gp120 evoking the release of IL-1β from microglia; the released IL-1β then activates IL-1 receptors on neurons leading to potentiation of NMDARs. If JZL184 inhibited synapse loss by reducing the production of IL-1β as suggested by the qRT-PCR data shown in Figure 3D, we would expect JZL184 to prevent gp120-induced potentiation of NMDARs but not necessarily potentiation induced by direct application of IL-1β. As shown in Figure 4E and F, 1 µM JZL184 completely blocked the gp120-induced potentiation of the NMDA-evoked increase in [Ca²⁺]_i. If the effects of JZL184 were mediated entirely by decreased production of IL-1β, then direct application of IL-1β to the culture should potentiate NMDA receptors in manner insensitive to inhibition of MGL. As shown in Figure 4G and H, treating the culture with IL-1β for 4 h potentiated the amplitude of the NMDA evoked [Ca²⁺]_i increase. In this cohort of cells NMDA increased the [Ca²⁺]_i in untreated cells from a mean level of 41 ± 3 nM to peak at 458 ± 28 nM. The net NMDA-evoked [Ca²⁺]_i response increased by 60 % in cells treated with 3 ng mL⁻¹ IL-1β for 4 h. As hypothesized, treatment with 1 µM JZL184 together with IL-1β for 4 h did not significantly affect the IL-1β-induced potentiation of the NMDA-evoked [Ca²⁺]_i response. Thus, JZL184 blocks gp120-induced IL-1β production preventing potentiation of NMDARs.

Figure 4.

Inhibition of MGL suppresses gp120-induced potentiation of NMDA receptors by blocking IL-1β production. (A–B) HIV-1 gp120-induced a biphasic change in NMDA-evoked [Ca²⁺]_i responses. (A) representative traces show NMDA-evoked [Ca²⁺]_i (10 µM × 60 s) increases from a control neuron (0 h) or neurons treated with 600 pM gp120 for the times indicated above the traces. (B) plot summarizes NMDA-evoked [Ca²⁺]_i responses after treatment with gp120 for 0 to 48 h. Data are expressed as mean ± SEM. **p<0.01 relative to 0 h time-point, ##p<0.01 relative to 4 h time-point (ANOVA with Tukey’s post-test, n ≥ 6). (C) Representative traces show increase in [Ca²⁺]_i evoked by application of 10 µM NMDA (60s) to control neurons and neurons treated with 600 pM gp120 for 4 h in the absence and presence of 1 µg mL⁻¹ IL-1ra. IL-1ra was added 15 min before gp120. (D) Bar graph summarizes net [Ca²⁺]_i increase (Δ[Ca²⁺]_i) evoked by 10 µM NMDA for control and neurons treated with 600 pM gp120 for 4 h in the absence and presence of 1 µg mL⁻¹ IL-1ra. ***p<0.001 relative to untreated control; ###p<0.001 relative to gp120 only treatment (one-way ANOVA with Tukey’s post-test, n=16). (E) Representative traces show increase in [Ca²⁺]_i evoked by application of 10 µM NMDA (60s) to control neurons and neurons treated with 600 pM gp120 for 4 h in the absence and presence of 1 µM JZL184. JZL184 was added 15 min before gp120. (F) Bar graph summarizes Δ [Ca²⁺]_i evoked by 10 µM NMDA for control neurons in the absence (n=10) and presence (n=6) of 1 µM JZL184 and neurons treated with 600 pM gp120 for 4 h in the absence (n=13) and presence (n=11) of 1 µM JZL184. **p<0.01 relative to untreated control; #p<0.05, relative to the group only treated with gp120 (one-way ANOVA with Tukey’s post-test). (G) Representative traces show NMDA13 evoked increase in [Ca²⁺]_i for control neurons and neurons treated with 3 ng mL⁻¹ IL-1β for 4 h in the absence or presence of 1 µM JZL184. JZL184 was added 15 min before IL-1β. (H) Bar graph shows Δ [Ca²⁺]_i evoked by 10 µM NMDA for control neurons in the absence (n=17) and presence (n=10) of 1 µM JZL184 and neurons treated with 3 ng mL⁻¹ IL-1β for 4 h in the absence (n=20) and presence of 1 µM JZL184 (n=19). ***p<0.001 relative to untreated control, ††p<0.01 relative to JZL184 treated control (one-way ANOVA with Tukey’s post-test).

3.6 JZL184 activated CB₂Rs blocks gp120-induced synapse loss

JZL184 could prevent gp120-induced synapse loss by activating cannabinoid receptors secondary to the accumulation of 2-AG and/or its neuroprotective effects could result from reduced PG production as a result of decreased conversion of 2-AG to AA (Blankman and Cravatt, 2013; Nomura et al., 2011; Ueda et al., 2011). To determine whether the protective effect of JZL184 resulted from increased 2-AG that subsequently activated CB₂R, we examined the effects of JZL184 on gp120-induced synapse loss in the absence and presence of 100 nM AM630, a selective CB₂R inverse agonist/antagonist. We have shown previously that this concentration of AM630 selectively and completely blocks CB₂R function in rat hippocampal cultures (Kim et al., 2011). AM630 (100 nM) reversed the effect of JZL184 and restored the synapse loss-induced by gp120 (Figure 5A). We confirmed that the effect of JZL184 was mediated CB₂R using another structurally distinct CB₂R inverse agonist/antagonist. SR144528 has been shown previously to selectively block CB₂R at a concentration of 100 nM (Dhopeshwarkar and Mackie, 2016). As shown in Figure 5B, SR144528 completely blocked the effects of JZL184. AM630 and SR144528 alone produced 8 ± 6 % and 3 ± 5 % change in synapses, respectively, which is not significantly different from control. In contrast, treatment with rimonabant, a CB₁R inverse agonist/antagonist that we have shown previously to selectively and completely block CB₁R function in rat hippocampal cultures (Roloff and Thayer), did not affect JZL184-mediated protection (Figure 5C). Similarly, the structurally distinct CB₁R inverse agonist/antagonist LY320135, at a concentration of 1 µM, which has been shown previously to selectively inhibit CB₁R function (Felder et al., 1998), also failed to affect JZL184-mediated protection (Figure 5D). Thus, the effects of JZL184 were not through activation of CB₁R. Rimonabant and LY320135 alone produced −2 ± 9 % and 9 ± 6 % change in synapses, respectively, which is not significantly different from control. JZL184 suppression of IL-1β mRNA expression was partially reversed by AM630 (100 nM) (Figure 5E). In the presence of both JZL184 and AM630, gp120-evoked IL-1β production was significantly less than that evoked by gp120 alone. Additionally, in cells treated with both JZL184 and AM630, gp120-evoked IL-1β production was significantly greater than that evoked by gp120 in the presence of JZL184. AM630 alone had no effect on IL-1β mRNA level (1.1 ± 0.2 -fold induction). These results suggest that JZL184 attenuates gp120-induced synapse loss, in part, through activation of CB₂R and subsequent suppression of IL-1β production.

Figure 5.

Activation of CB₂R but not CB₁R was required for JZL184 inhibition of gp120-induced neurotoxicity. (A) Bar graph shows changes in PSD95.FingR-eGFP puncta number after 24 h treatment with the indicated treatment groups: untreated (n=12), 600 pM gp120 (n=16), gp120 + 1 µM JZL184 (n=9), and gp120 + JZL184 + 100 nM AM630 (n=8). The CB₂R inverse agonist/antagonist AM630 was added 5 min prior to JZL184. ***p<0.001 relative to untreated, #p<0.05 relative to the group only treated with gp120, †p<0.05 relative the group treated with gp120 + JZL184 (one-way ANOVA with Tukey’s post-test). (B) Bar graph shows changes in PSD95.FingR-eGFP puncta number after 24 h treatment with the indicated treatment groups: untreated (n=10), 600 pM gp120 (n=10), gp120 + 1 µM JZL184 (n=10), and gp120 + JZL184 + 100 nM SR144528 (n=10). The CB₂R inverse agonist SR144528 was added 5 min prior to JZL184. **p<0.01 relative to untreated, ###p<0.001 relative to the group only treated with gp120, ††p<0.01 relative the group treated with gp120 + JZL184 (one-way ANOVA with Tukey’s post-test). (C) Bar graph shows changes in PSD95.FingR-eGFP puncta number after 24 h treatment with the indicated treatment groups: untreated (n=8), 600 pM gp120 (n=8), gp120 + 1 µM JZL184 (n=7), and gp120 + JZL184 + 100 nM rimonabant (n=9). The CB₁R inverse agonist rimonabant was added 5 min prior to JZL184. **p<0.01 relative to untreated; #p<0.05, ##p<0.01 relative to the group only treated with gp120 (one-way ANOVA with Tukey’s post-test). (D) Bar graph shows changes in PSD95.FingR-eGFP puncta number after 24 h treatment with the indicated treatment groups: untreated (n=9), 600 pM gp120 (n=8), gp120 + 1 µM JZL184 (n=9), and gp120 + JZL184 + 1 µM LY320135 (n=8). The CB₁R inverse agonist LY320135 was added 5 min prior to JZL184. *p<0.05 relative to untreated; #p<0.05, ###p<0.001 relative to the group only treated with gp120 (one-way ANOVA with Tukey’s post-test). (E) Bar graph summarizes IL-1β mRNA expression measured using qRT-PCR assay after 4 h treatment with the indicated treatments: untreated (n=10), 600 pM gp120 (n=13), gp120 + 1 µM JZL184 (n=9), and gp120 + JZL184 + 100 nM AM630 (n=9). AM630 was added 5 min prior to JZL184. ***p<0.001 relative to untreated group, #p<0.05 relative to the group only treated with gp120, †p<0.05 relative to the group treated with gp120 and JZL184 (one-way ANOVA with Tukey’s post-test).

3.6 JZL184 suppresses gp120-induced PGE₂ production

Because AM630 only partially reversed JZL184 inhibition of gp120-induced IL-1β expression, JZL184 might protect synapses through a mechanism in addition to activation of CB₂R. A potential role for reduced PG production in the protection of synapses was considered because JZL184 decreases the conversion of 2-AG to AA which is the precursor for PGs (Nomura et al., 2011). PGE₂ contributes to neuroinflammation (Johansson et al., 2013; Johansson et al., 2015); thus, we first determined whether gp120 treatment can induce the production of PGE₂. Treatment with 600 pM gp120 stimulated PGE₂ production and this effect was blocked in the presence of 1 µM JZL184 (Figure 6A). PGE₂ (56 ± 5 pg mL⁻¹) was present in media from untreated cultures. Treatment (4 h) with HIV gp120 increased PGE₂ to 80 ± 7 pg mL⁻¹. Note that PGE2 levels were measured in the relatively large volume of the cell culture well and might be higher in the cell monolayer.

Figure 6.

MGL inhibition blocks gp120-induced PGE₂ production. EP_1–2R activation is required for synapse loss and IL-1β expression. (A) Bar graph shows changes in PGE₂ levels in culture media measured using ELISA. PGE₂ levels are shown for control conditions in the absence (n=14) or presence of 1 µM JZL184 (n=9) and after treatment with 600 pM gp120 for 4 h in the absence (n=17) or presence (n=14) of 1 µM JZL184. *p<0.05 relative to untreated control, #p<0.05 relative to the group only treated with gp120 (one-way ANOVA with Tukey’s post-test). (B) Bar graph shows changes in the number of PSD95.FingR-eGFP puncta under control conditions in the absence (n=11) or presence (n=6) of 10 µM AH6809, an EP_1–2R antagonist, or 24 h following treatment with 600 pM gp120 in the absence (n=9) or presence (n=9) AH6809. **p<0.01 relative to untreated control, #p<0.05 relative to the group treated with gp120 alone (one-way ANOVA with Tukey’s post-test). (C) Bar graph summarizes changes of IL-1β mRNA expression in untreated cultures (n=10) and after 4 h treatment with 600 pM gp120 in the absence (n=11) or presence (n=10) of 10 µM AH6809. Data are expressed as fold induction relative to untreated group. *p<0.05, **p<0.01 relative to untreated group; #p<0.05 relative to the group treated with gp120 alone (one-way ANOVA with Tukey’s post-test). (D) Bar graph shows Δ[Ca²⁺]_i evoked by 10 µM NMDA for control cells in the absence (n=7) and presence (n=6) of 10 µM AH6809 and neurons treated with 3 ng mL⁻¹ IL-1β for 4 h in the absence (n=11) and presence (n=8) of AH6809. ***p<0.001 relative to untreated control, †p<0.05 relative to AH6809 treated control (one-way ANOVA with Tukey’s post-test).

To determine whether PGE₂ contributed to gp120-induced synapse loss we examined the effects of gp120 in the absence and presence of AH6809, an antagonist for prostaglandin 1 and 2 receptors (EP_1–2R) (Woodward et al., 1995). Gp120-induced synapse loss was blocked completely by 10 µM AH6809 (Figure 6B). Thus, activation of EP_1–2Rs by PGs is necessary for gp120-induced synapse loss. However, the inhibition of gp120-induced IL-1β expression by 10 µM AH6809 was incomplete (Figure 6C), suggesting that PGs might also act downstream of IL-1β to facilitate synapse loss (Figure 7B, green arrows). AH6809 alone had no effect on IL-1β mRNA levels (1.0 ± 0.08 fold change). This is consistent with a previous report showing synapse loss induced by the direct application of IL-1β requires PGE₂ production (Mishra et al., 2012). Furthermore, calcium imaging experiments showed that IL-1β-induced potentiation of NMDA receptors was not blocked by AH6809 (Figure 6D). These data indicate that the PG pathway downstream of IL-1β is separate from that mediating the potentiation of NMDA receptors (Figure 7B). Thus, activation of PG receptors is required for gp120-induced synapse loss (Figure 6B). These results suggest that, in addition to preventing synapse loss through 2-AG activation of CB₂R (Figure 5), JZL184 inhibition of PGE₂ production could also reduce IL-1β production (Figure 6C) and exert downstream actions such as an inhibition of PG-induced glutamate release.

Figure 7.

Summary scheme shows the hypothesized mechanism of the synapse protection induced by the inhibition of MGL. (A) Diagram illustrates the JZL184-induced increase in 2-AG that activates CB₂R-dependent synapse protection and the reduced AA production that decreases PGE₂-mediated synapse loss. (B) Scheme shows 2-AG and AA dependent pathways affected by JZL184 to reduce synapse loss. Solid arrows indicate flow of signaling pathways. Green lines highlight prostaglandin signaling and red lines highlight eCB signaling.

4. Discussion

Chronic neuroinflammation underlies the pathogenesis of HAND (Saylor et al., 2016; Sodhi et al., 2004). Infected immune cells release viral proteins and inflammatory factors which act on microglia, astrocytes, and neurons to produce the synaptodendritic damage that correlates with cognitive decline in HIV infected patients (Ellis et al., 2007). Here we used a simplified in vitro model to study the interaction of the eCB system with the neuroinflammatory response evoked by the HIV envelope protein. HIV gp120 induced synapse loss, a hallmark of HAND, by activating microglia to release IL-1β that then potentiated NMDA receptor function (Kim et al., 2011; Viviani et al., 2006). Inhibiting the metabolism of the eCB, 2-AG, attenuated gp120-induced synapse loss via activation of CB₂R and may have also protected synapses via reduced PG production. These pathways are summarized in Figure 7.

The suppression of gp120-induced synapse loss described here is the first report of a CB₂R-mediated effect of JZL184 on neuroinflammation induced synaptic damage. 2-AG is one of the most abundant eCBs in the brain and is a full agonist at both CB₁R and CB₂R (Gonsiorek et al., 2000). MGL is the primary enzyme responsible for 2-AG hydrolysis in brain, accounting for 85% of its metabolism (Blankman et al., 2007). Pharmacological and genetic inactivation of MGL has been shown to significantly increase brain 2-AG levels (Grabner et al., 2016; Long et al., 2009). Thus, the accumulation of 2-AG in the presence of JZL184 is expected. However, the relative contribution of increased 2-AG versus decreased AA following inhibition of MGL has varied in different models. In the mouse experimental autoimmune encephalitis model, inhibition of MGL reduced symptoms via a CB₁R/CB₂R mechanism (Brindisi et al., 2016). Blocking MGL suppresses LPS-induced neuroinflammation and loss of dopamine neurons in a model of Parkinson’s disease by decreasing PG production with no evidence for enhanced eCB signaling (Nomura et al., 2011). Similarly, in transgenic mouse models of Alzheimer’s disease JZL184 was neuroprotective but, without clear involvement of CB receptors (Chen et al., 2012; Piro et al., 2012). In mice, JZL184 increased brain 2-AG 8-fold and elicited an array of CB₁R-mediated behavioral effects (Long et al., 2009). The principal conclusion from the studies described in this report is that CB₂R activation following inhibition of MGL affords significant protection from synapse loss induced by gp120.

There are several aspects of gp120-induced synapse loss that might render it particularly susceptible to CB₂R activation. In a previous report, we showed that the CB_1/2R agonist WIN55,212-2 blocked gp120-induced IL-1β production and synapse loss through activation of CB₂Rs on microglia (Kim et al., 2011). Perhaps gp120 activation of the CXCR4 pathway to trigger IL-1β release from microglia is particularly sensitive to inhibition by CB₂R activation. CB₂R agonists have been shown to inhibit chemokine CXCL12-induced and CXCR4-mediated chemotaxis of T lymphocytes (Ghosh et al., 2006). The neuroprotective effects of JZL184 in transgenic models of Alzheimer’s disease appear to result from activation of peroxisome proliferator-activated receptor-γ and decreased PG activation of NF-κB resulting in reduced expression of β-secretase and decreased Aβ production, a pathway not expected to be regulated by CB receptors (Piro et al., 2012; Zhang et al., 2014). Responses in microglia evoked by LPS activation of the toll-like receptor 4 pathway are inhibited by CB₂R agonists (Ma et al., 2015; Malek et al., 2015; Merighi et al.; Oh et al., 2010; Romero-Sandoval et al., 2009) although, there are reports of non-receptor mediated effects of CBs (Puffenbarger et al., 2000; Tham et al., 2007) and a lack of CB₂R effects in some studies (Kouchi, 2015). The failure of CB₂R to contribute to the anti-inflammatory effect of JZL184 in vivo may result from receptor desensitization following the prolonged treatment protocols used for in vivo studies (Nomura et al., 2011). Indeed, prolonged administration of JZL184 desensitizes CB₁R (Schlosburg et al., 2010). However, desensitization of the CB₂Rs that mediate the effect described here has not been explicitly shown, in part because the expression of CB₂Rs is upregulated by inflammatory stimuli (Benito et al., 2008). In acute models of peripheral inflammatory pain, JZL184 produces analgesia via CB₁ and CB₂ receptors (Guindon et al., 2011). Thus, JZL184 may reduce neuroinflammation by PG and/or eCB mechanisms depending on the specific inflammatory stimulus and duration of treatment.

In the brain, hydrolysis of 2-AG by MGL is the primary source of AA for conversion to PGs by cyclooxygenase (Nomura et al., 2011). PGE₂ production is required for IL-1β-evoked synapse loss (Mishra et al., 2012). Here, we tested whether decreased PG levels contributed to the synapse protective effects of JZL184. Activation of EP_1–2Rs was necessary for gp120-induced synapse loss and JZL184 blocked gp120-induced PGE₂ production. Thus, it is likely that JZL184 suppression of PGE₂ production contributed to the protective effect. However, because AM630 completely blocked the synapse protection afforded by JZL184, we conclude that the reduction in PG levels produced by MGL inhibition was not the primary mechanism in this model. JZL184 did not suppress the basal level of PGE₂ present in unstimulated cultures, suggesting that sufficient PGE₂ from a source other than MGL was available to enable gp120-induced synapse loss. Furthermore, IL-1β potentiated NMDA receptors independent of EP_1–2R activation, suggesting separate IL-1β actions on PG production and NMDA receptor signaling. A partial attenuation of gp120-evoked IL-1β production with the EP_1–2R antagonist can be reconciled with its complete block of gp120-induced synapse loss by considering the PG regulation of synapse loss at two steps in the pathway activated by gp120. Activation of EP_1–2Rs facilitates glial production of IL-1β and stimulates glutamate release resulting in biochemical potentiation of NMDARs and their direct activation, respectively (Mishra et al., 2012). Thus, we cannot rule out a contribution resulting from decreased PGE₂ synthesis. The CB₂R- and EP_1–2R-depedence of JZL184 inhibition of IL-1β production were both partial effects, consistent with the idea that the inhibition of MGL has dual actions via CB₂ receptor signaling and decreased PGE₂ production.

HIV-1 gp120 evoked synapse loss via a multi-step process (Figure 7) involving multiple cell types. The primary effects of JZL184 appear to be on glia where activation of CB₂R or reduced PGE₂-dependent activation of EP_1–2Rs inhibits the release of IL-1β. In a previous study we found that IL-1β induced PGE₂ production was required for synapse loss, presumably via an EP_1–2R mediated increase in presynaptic glutamate release (Mishra et al., 2012). If we consider a role for presynaptic glutamate release in gp120-induced synapse loss, then we might expect a CB₁R-mediated component to JZL184-mediated synapse protection because 2-AG mediates a robust inhibition of excitatory synaptic transmission (Roloff et al., 2010; Straiker et al., 2009). However, neither rimonabant nor LY320135 affected JZL184-mediated protection, suggesting that the elevation in 2-AG might be localized to a microdomain with preferential access to microglia independent of presynaptic terminals.

Cannabinoid receptor agonists have beneficial effects in models of HAND (Avraham et al., 2014; Kim et al., 2011; Purohit et al., 2014). However, drugs that directly activate cannabinoid receptors might cause receptor desensitization during long-term treatment, diminishing the neuroprotective efficacy of the endocannabinoid system. Agonists with actions on CB₁ receptors or those with low selectivity may have abuse liability. Alternatively, inhibition of MGL only potentiates endogenously produced 2-AG so that receptor activation is dependent on a stimulus. High doses of JZL184 produced extensive CB₁ receptor activation (Long et al., 2009). However, prolonged, low-dose JZL184 treatment elicited an anti-inflammatory effect without producing CB₁ receptor tolerance or cannabinoid dependence (Kinsey et al., 2013). Furthermore, because brain AA production is primarily dependent on MGL, in contrast to the gut, drugs that inhibit MGL can reduce brain PG levels without the gastrointestinal side effects produced by nonsteroidal anti-inflammatory drugs (Scheiman, 2016). Thus, drugs that inhibit MGL show promise for reducing neuroinflammation in HAND.

Dendritic damage and loss of synaptic connections correlate with cognitive decline in HAND patients. Synapse loss induced by HIV proteins occurs early and via a different signaling pathway from that leading to neuronal death (Kim et al., 2008), suggesting that synapse loss might be a mechanism to reduce excitotoxicity (Hargus and Thayer, 2013). Indeed, synapse loss is reversible (Shin et al., 2012) and rescue of synapses lost following exposure to the neuroinflammatory HIV protein Tat restored cognitive function (Raybuck et al., 2017). Chronic inhibition of MGL in amyloid precursor protein transgenic mice induced recovery of synaptic spines and improvement of spatial learning and memory function (Chen et al., 2012). Perhaps the synapse loss induced by the neuroinflammatory component of neurodegenerative diseases is readily reversible. It will be interesting to determine whether, in addition to protecting synapses from loss induced by HIV proteins, MGL inhibition by drugs such as JZL184 can rescue synapses when given after loss has already occurred.

This report shows for the first time that pharmacological inhibition of MGL can afford neuroprotection via activation of CB₂ receptors on brain microglia. This mechanism differs from other reports in which the reduced neuroinflammation produced by JZL184 was primarily mediated via reduced PG signaling. The mechanism of protection afforded by MGL inhibitors may depend on the specific cell types involved in the neuroinflammatory response, the concentration and duration of drug treatment and the unique signaling pathways recruited in response to various inflammatory stimuli. This is the first report showing inhibition of MGL affords neuroprotection in a model of HAND, suggesting that the chronic neuroinflammation that underlies this disorder may be particularly susceptible to modulation of the eCB system.

Highlights.

• HIV gp120 induces synapse loss, a hallmark of HIV-associated neurocognitive disorder

• Monoacylglycerol lipase inhibitor JZL184 prevents gp120-induced synapse loss

• JZL184 protects from gp120 via CB₂R

• JZL184 blocks gp120-induced potentiation of NMDARs

• gp120 activates and JZL184 inhibits prostaglandin signaling

Abbreviations

AA

arachidonic acid

AM630

6-iodopravadoline

AH6809

9-oxo-6-propan-2-yloxyxanthene-2-carboxylic acid

2-AG

2-arachidonoylglycerol

CB

cannabinoid

CB₁R

cannabinoid type 1 receptor

CB₂R

cannabinoid type 2 receptor

DMEM

Dulbecco’s modified Eagle’s Medium

eCB

endocannabinoid

EP_1–2R

prostaglandin 1 and 2 receptors

GFP

green fluorescent protein

HAND

HIV-associated neurocognitive disorder

HIV-1

human immunodeficiency virus type 1

HHSS

HEPES Hanks’ salt solution

IL-1β

interleukin-1β

JZL184

4-nitrophenyl 4-(dibenzo[d][1,3]dioxol-5-yl(hydroxy)methyl)piperidine-1-carboxylate

LPS

lipopolysaccharide

MGL

monoacylglycerol lipase

NMDA

N-methyl-D-aspartate

PSD95

post-synaptic density protein 95

PG

prostaglandin

qRT-PCR

quantitative reverse transcription real-time polymerase chain reaction