Background
Extracellular vesicles (EVs) are secretory vesicles budded from the plasma membrane of a variety of cells. EVs range from 50 nm to 1 mm and differ in their origins, either from direct fusion with plasma membrane or intracellular multivesicular bodies. EVs have been detected at an elevated level in the cerebral spinal fluid in patients with mild to severe Alzheimer’s disease, Parkinson’s disease, prion disease, and amyotrophic lateral sclerosis [
1,
2]. Protein markers of EVs are present in neuritic plaques in AD brains. Furthermore, increasing EV release likely contributes to the toxicity of amyloid beta and tau phosphorylation [
3]. Therefore, EVs play a potential role in the pathogenesis of AD. Mechanisms regulating EV release remain poorly understood. Recent studies indicate that the formation and secretion of EVs are largely dependent on the proper function of ceramide, which is a type of sphingolipids catalyzed by neutral sphingomyelinase (nSMase) from sphingomyelin. More specifically, EV release can be blocked by the inhibition of the nSMase pathway [
4,
5]. GW4869, a nSMase inhibitor, significantly reduces the release of EVs and corresponding neurotoxicity in the HIV-infected, as well as immune-activated macrophages and microglia, and in AD models in vitro and in vivo [
3,
6,
7].
Glutamine is the most abundant amino acid in the plasma, and the metabolism of glutamine involves hydrolysis to glutamate by mitochondrial enzyme glutaminase 1 (GLS1). Subsequently, glutamate can be excreted or can be further metabolized to α-ketoglutarate. Our previous studies uncovered a pathogenic role of glutaminase (GLS) in neuroinflammation and neuronal injury. There are two major types of GLS identified in mammals, the “kidney-type” glutaminase (GLS1) and “liver-type” glutaminase. GLS1 is the predominant glutamine-utilizing enzyme in the central nervous system (CNS), where “liver-type” glutaminase is expressed at a lower level [
8]. GLS1 is known to be associated with cancer cell research and CNS diseases [
9‐
11]. Due to tissue-specific alternative splicing from the same gene, two isoforms of GLS1 were identified in the human brain, kidney-type glutaminase (KGA) and glutaminase C (GAC). Both KGA and GAC appear to catalyze glutamine deamination with comparable enzyme efficacies and kinetics [
12,
13]. GAC is upregulated in HIV-associated dementia brain samples and is also released in the conditioned medium from HIV-1-infected macrophages. Furthermore, mitochondrial stress during HIV infection leads to membrane destabilization and release of GLS from the mitochondrial matrix to the cytosol through the permeability transition pore [
14,
15].
GLS1 has been identified as an important metabolic factor controlling EV release from astrocytes during neuroinflammation [
16]. However, the mechanism by which GLS1 regulates EV biogenesis and release remains unknown. In this study, we determine the mechanism of EV biogenesis and release through GLS1-mediated glutamine metabolism in human macrophages. We further identify two key downstream metabolites-α-ketoglutarate and ceramide-as critical factors regulating EV release during HIV-1 infection and immune activation. These studies may help understand how glutamine metabolism regulates EV release in the context of infection and inflammation.
Discussion
Glutamine is the most abundant amino acid in the plasma, and its metabolic products provide energy and substrates for a variety of biosynthesis pathways. During viral infection and inflammation, glutamine metabolism is particularly important in that energy, biosynthesis, and antioxidative capacity are essential for a proper immune response. Our previous studies demonstrated a strong link between GLS and the neuropathogenesis of HIV-1 infection via the overproduction of neurotoxic levels of glutamate [
11,
23‐
27]. Interestingly, GLS1 has been identified as an important metabolic factor controlling EV release from astrocytes in the presence of TNF-α [
16]. However, it remains unclear how glutamine metabolism regulates EV biogenesis and release. The current studies present a major finding regarding EV release. Upregulation of GLS1 induces an increase in the release of EVs through α-KG and ceramide in HIV-1-infected macrophages and immune-activated microglia. The release of EVs is also observed to be increased in GAC-overexpressing transgenic mice. These new mechanistic regulations may help understand how glutamine metabolism shapes EV biogenesis and release during neuroinflammation.
Dysregulation of GLS1 has been reported in the pathogenesis of HIV-associated neurocognitive disorders and in cancer. In the CNS, GLS1 is a key enzyme in the glutamine metabolism, where glutamate, the main excitatory neurotransmitter, is produced to generate glutamate signaling and synaptic plasticity [
28‐
32]. The dysregulation of GLS1 could potentially lead to the aberrant release of glutamate and compromise its neurotransmission. Indeed, abnormal glutamate neurotransmission is strongly associated with the memory loss and learning deficits due to the disrupted functioning of NMDA receptors [
31‐
33]. To model upregulation of GLS1, we first constructed new adenoviruses overexpressing GLS1 isoforms. KGA- and GAC-overexpressing HeLa cells release a higher number of EVs into the supernatants. KGA- and GAC-mediated EV release can be blocked by GLS1 inhibitors BPTES and CB839, suggesting GLS1 is a critical factor regulating EV biogenesis and release. Similarly, we also used HIV-1-infected macrophages and LPS-treated microglia to model GLS1 upregulation. HIV-1-infected macrophages and LPS-treated microglia have higher GLS1 levels compared to those of controls, which are associated with higher levels of EVs. Our third approach to model GLS1 upregulation is to use newly generated GAC transgenic mice. More EVs can be collected from GAC transgenic mice compared with those from negative littermates. However, it remains to be confirmed whether GLS1 inhibitors can block the EV release in vivo and whether the blocking of EV release could have protective effects on the GAC transgenic mice.
Our investigations reveal potential mechanisms of EV release in the context of HIV-1 infection and neuroinflammation. First, the release of EVs is dependent on the presence of glutamine. Second, α-KG, a downstream product of glutamate, can rescue the inhibition of EV release by GLS1 inhibitors. Third, C
6 ceramide, a cell-permeable analog of ceramide, can rescue the inhibition of EV release by GLS1 inhibitors. In glutamine repletion experiments, changes in EV release were observed at as low as 1 mM of glutamine, the level of which is close to plasma concentration of glutamine [
34]. However, the increase of glutamine concentration from 1 to 5 mM did not show a dose-dependent increase of EVs. Further testing of lower concentrations of glutamine will help to establish a dose-dependent effect of glutamine. Another limitation of the current studies involves the experiments on α-KG and ceramide. It is unknown whether the effect of α-KG and C
6 ceramide on EV release was direct or through other downstream metabolic intermediates. Given the variety of metabolic intermediates downstream of α-KG, it is likely that other contributing factors in this pathway are required for EV biogenesis and release. Indeed, recent data suggest specific mechanisms controlled by glutaminolysis to fine-tune macrophage activities during both M2 and LPS activation involve a-KG and succinate [
35,
36]. EV biogenesis and release from macrophages may aid in the macrophage-mediated immune responses to infection. Therefore, harnessing EVs through glutamine/a-KG pathway would be an attractive strategy to regulate macrophage phenotypes in infection and inflammation.
Our characterizations of EVs involve the determination of EV markers in Western blots, EV size/concentrations in well-established NanoSight, and the morphological examination through negative staining TEM. Notably, the size of EVs remains the same throughout all the studies as determined by NanoSight analysis, indicating that the regulation of the EVs during HIV-1 infection and immune activation of macrophages/microglia are more specifically on the number and contents. EVs are known to include exosomes, microvesicles, and apoptotic bodies according to their cellular origin. In our studies, there is little evidence of apoptotic bodies in the EVs since the size of EVs was overwhelmingly smaller than 300 nm, whereas typical apoptotic bodies are more than 500 nm. Furthermore, it is known that EV isolation contains virions, and HIV-1 virions are around 145 nm [
37], which may be mistaken as EVs during data interpretation. However, based on the data from NanoSight, EV quantities (10
8 ~ 10
10/ml) greatly exceed the quantity of viral particles (~ 10
6/ml) in the supernatant. Therefore, we conclude that the particles detected by the NanoSight are predominantly EVs, including exosomes and microvesicles.
Previously, we reported that using GW4869 could effectively inhibit the release of EVs in HIV-1-infected macrophages and LPS-treated microglia [
6]. Studies have also shown that GW4869 could inhibit EV release in vitro and in vivo in AD models and prion diseases [
3,
7,
38‐
42]. It has also been reported that sphingolipid metabolism could be involved in EV release in microglia and neurological diseases [
5,
43]. Our data demonstrate a causal role of ceramide in mediating GLS1-induced EV release. However, the exact role of sphingolipid metabolism in GLS1-associated EV release remains to be fully elucidated. It is unclear whether other lipid pathways also involved in GLS-mediated EV regulation [
44,
45]. Whether sphingolipid metabolism merely provides building blocks for EVs or more refined mechanisms are involved remains to be determined.
Methods
Ethics statement
MDM were used in full compliance with the University of Nebraska Medical Center and National Institutes of Health ethical guidelines, with the Institutional Review Board (IRB) #: 162-93-FB. We have the informed written consent from all participants involved in this study. All mice were housed and bred in the Comparative Medicine Animal Facilities at the University of Nebraska Medical Center. All procedures were conducted in accordance with the protocol (11-018-04) approved by the Institutional Animal Care and Use Committee at the University of Nebraska Medical Center.
Culture, HIV-1 infection, and LPS activation of macrophages and microglia
Human peripheral blood-derived mononuclear cells were isolated through leukopheresis from healthy donors. Human macrophages were differentiated in Dulbecco’s modified Eagle’s media (DMEM) (Sigma-Aldrich, St. Louis, MO) with 10% human serum, 50 μg/ml gentamycin, 10 μg/ml ciprofloxacin (Sigma), and 1000 U/ml recombinant human macrophage colony-stimulating factor (MCSF) for 7 days. Human fetal microglial cells were obtained from fetal brain tissue-derived microglia-astrocytes mixed cultures as previously described [
46]. The HIV-1
ADA strain was used to infect the macrophages and microglia at a multiplicity of infection (MOI) of 0.1 and 0.5, respectively. After 24 h, the culture medium was changed to remove any remnant virus. Seven days after HIV-1-infection, the culture medium was changed to glutamine-free neurobasal medium for 24 h, and supernatants were collected for subsequent HPLC or Western blot analysis. HeLa and BV
2 cell lines were obtained from ATCC, and both cell lines were grown in DMEM with 10% fetal bovine serum and antibiotics. LPS was used to immune activate BV
2 cells for 24 h, and supernatants were collected for HPLC and Western blot analysis. Bis-2-(5-phenylacetamido-1,2,4-thiadiazol-2-yl)ethyl sulfide (BPTES) and CB839 (generous gifts provided by Dr. Takashi Tsukamoto from John Hopkins University and later ordered from Millipore with catalog numbers 530030 and 533717, respectively) were used in HIV-1-infected macrophages or LPS-treated microglia prior to EV isolation. α-Ketoglutarate (Sigma; 349631) and C
6 ceramide (Sigma; H6524) were also used to manipulate the metabolic intermediates in HIV-1-infected macrophages or LPS-treated microglia. All experiments involving human cell samples are approved by the Institutional Review Board at the University of Nebraska Medical Center.
Adenoviral constructs
Replication-defective adenovirus vectors expressing human KGA and GAC were generated using RAPAd® CMV Adenoviral Expression System (Cell Biolabs, Inc., San Diego, CA). Generation of the full-length human KGA and GAC constructs was described in our prior publication [
12]. Adenoviral constructs were amplified in a 293 AD cell line (Cell Biolabs) and purified by ultracentrifugation through a CsCl gradient. Viral titer was determined by the Adeno-X™ Rapid Titer Kit (Clontech Laboratories, Inc., Mountain View, CA).
Isolation of EVs from cells
EVs were isolated from the supernatants of GLS1-overexpressing cells, HIV-1-infected macrophages, and LPS-activated microglia through differential centrifugations. Briefly, the supernatants were first centrifuged at 300×g for 10 min to remove free cells, at 3000×g for 20 min to remove cellular debris, and then 10,000×g for 30 min to remove free organelles. Lastly, EVs were collected by ultracentrifugation at 100,000×g for 2 h at 4 °C. To prepare EVs for Western blotting, the EV pellets were lysed in M-PER mammalian protein extraction reagent (Thermo Scientific, Pittsburgh, PA). For negative staining, EVs were fixed in 2% glutaraldehyde and 2% paraformaldehyde. For glutaminase activity assay and neurotoxicity, the EVs were resuspended in 1 ml of glutamine-free neurobasal medium.
Isolation of EVs from mice brain
EV isolations from the brains were carried out as described previously with modifications according to the protocol [
22]. The fresh and previously frozen mice hemibrains were harvested and dissected finely. The brain samples were then treated with 20 units/ml papain (Worthington) in Hibernate E solution (BrainBits, Springfield, IL) for 15 min at 37 °C. The same volume of cold Hibernate E solution was added to the brain samples to stop the reaction of papain. The brain tissue was then gently homogenized and filtered through a 40-μm mesh filter (BD Biosciences), followed by a centrifugation at 300×
g for 10 min and 3000×
g for 20 min at 4 °C to get rid of cells, membranes, and debris. After the supernatants were filtered through 0.45-μm filter (Thermo Scientific), they were subjected to 10, 000×
g for 30 min at 4 °C to eliminate organelle contaminations. The supernatants were further centrifuged at 100,000×
g for 70 min at 4 °C to pellet EVs. The pellets were then resuspended in filtered PBS, or MPER lysate solution for NanoSight or Western blot. All the samples were ultracentrifuged in ultraclear polycarbonate tubes (Beckman Coulter) that have a volume of 13.2 ml. A Beckman Coulter ultracentrifuge (Beckman Coulter OptimaL-90K ultracentrifuge; Beckman Coulter, Fullerton, CA, USA) was used with a rotor type SW 41 Ti.
Negative staining and electron microscopy
EVs were fixed and then spread on the silicon monoxide and nitro-cellular film-coated copper grid. The droplets were removed with filter paper, air-dried at room temperature, and then subjected to transmission electron microscopy (TEM).
Nano-particle tracking analysis
A NanoSight NS 300 (Malvern) equipped with an sCMOS camera was utilized to analyze the size distribution and concentration of EVs. NanoSight utilizes NTA, which is a combination of light scattering and Brownian motion technology to measure the concentration and size and distribution of particles in the EV supernatants. After the whole process of EV isolation, the pellets were first resuspended in 100 μl of filtered PBS and then diluted 100 times. The conditions of the measurements include temperature of 25 °C; viscosity of 1 cP, 25 s per capture frame; and a measurement time of 60 s. All the conditions were kept the same among all the samples. The results indicate the mean sizes and concentration of at least three individual measurements.
Western blot
Protein concentrations were determined by Bradford protein assay. SDS PAGE separated proteins from the whole cell and EV lysates. Afterward, they were electrophoretically transferred to polyvinyldifluoridene membranes (Millipore, Billerica, MA and Bio-Rad, Hercules, CA). The membranes were incubated overnight at 4 °C with polyclonal antibodies for KGA and GAC (Dr. N. Curthoys, Colorado State University, Fort Collins, CO), tissue transglutaminase (tTG) (Lab Vision/Thermo, Fremont, CA), flotillin-2 (Cell Signaling Technology, Danvers, MA), and β-actin (Sigma), followed by horseradish peroxidase-linked secondary anti-rabbit or anti-mouse secondary antibodies (Cell Signaling Technology). Antigen-antibody complexes were visualized by Pierce ECL Western Blotting Substrate. For quantification of the data, films were scanned with a CanonScan 9950F scanner, and images were analyzed using the public domain NIH Image program (developed at the US National Institutes of Health).
Glutaminase activity assay
Highly concentrated whole cell lysates were collected from flasks and subjected to GLS activity assay using a two-step assay [
27,
47]. Briefly, protein concentrations in the lysates were tested by using BCA Protein Assay Kit (Pierce). All samples were normalized to same concentration. In the first step, 50 mg of protein were added to 100 μl of initial assay mix. The mix contains 50 mM glutamine, 0.15 M phosphate, 0.2 mM EDTA, and 50 mM Tris-acetate. The PH value of the mix was adjusted to 8.6 and incubated at 37 °C for 30 min. Ten microliters of 3 N hydrochloric acid was added to inactivate the glutaminase activity and stop the reaction. In the second step, 1 ml of the second reaction mix was added, which contained 0.4 mg of purified bovine liver glutamate dehydrogenase (Sigma-Aldrich, St. Louis, MO, USA), 0.08 M Tris-acetate at pH 9.4, 0.2 M hydrazine, 0.25 mM adenosine 5′-diphosphate sodium salt, and 2 mM β-nicotinamide adenine dinucleotide hydrate. The samples were mixed and incubated for 30 min at room temperature. One hundred microliters of the reaction was used for measurement, absorbance was determined at a wavelength of 340 nm, and glutamate concentration was determined using a standard curve of 10, 5, 2.5, 1.25, 0.625, and 0.0 mM glutamate, along with negative controls.
Analysis of glutamate concentrations
Glutamate levels were analyzed by RP-HPLC using an Agilent 1200 liquid chromatograph and fluorescence detector as previously described [
14] with a few modifications. The experiments utilized 4.6 × 75 mm, 3.5 μm ZORBAX Eclipse AAA analytical columns (Agilent). A gradient elution program was optimized for glutamate measurement with a flow rate 0.75 ml/min. The intracellular glutamate levels in the whole brain lysates of mice and whole cell lysates were determined by Amplex Red Glutamic Acid/Glutamate Oxidase Assay Kit (Invitrogen) based on the manufacturer’s instruction. The brain tissue lysates and whole cell lysates were diluted to the same protein concentration before the assay.
Statistical analysis
Data are expressed as means ± SD unless otherwise specified. Statistical analysis was performed using one-way analysis of variance (ANOVA), followed by the Bonferroni post-test for all paired observations unless otherwise specified. Significance was determined by a p value < 0.05. All experiments were performed with cells from at least three donors to account for any donor-specific differences. Assays were performed at least three times in triplicate or quadruplicate within each assay.
Acknowledgements
We kindly thank Dr. Norman Curthoys for providing the KGA and GAC antibodies. We thank Dr. Changhai Tian, Dr. Santhi Gorantla, Ms. Li Wu, and Tom Bargar for the technical support of this work. Julie Ditter, Lenal Bottoms, Myhanh Che, Johna Belling, and Robin Taylor provided outstanding administrative and secretarial support.