Inhibitory effects of alcohol on glucose transport across the blood–brain barrier leads to neurodegeneration: preventive role of acetyl-l-carnitine
- First Online:
- Cite this article as:
- Muneer, P.M.A., Alikunju, S., Szlachetka, A.M. et al. Psychopharmacology (2011) 214: 707. doi:10.1007/s00213-010-2076-4
- 277 Views
Evidence shows that alcohol intake causes oxidative neuronal injury and neurocognitive deficits that are distinct from the classical Wernicke-Korsakoff neuropathy. Our previous findings indicated that alcohol-elicited blood-brain barrier (BBB) damage leads to neuroinflammation and neuronal loss. The dynamic function of the BBB requires a constant supply and utilization of glucose. Here we examined whether interference of glucose uptake and transport at the endothelium by alcohol leads to BBB dysfunction and neuronal degeneration.
Material and methods
We tested the hypothesis in cell culture of human brain endothelial cells, neurons and alcohol intake in animal by immunofluorescence, Western blotting and glucose uptake assay methods.
We found that decrease in glucose uptake correlates the reduction of glucose transporter protein 1 (GLUT1) in cell culture after 50 mM ethanol exposure. Decrease in GLUT1 protein levels was regulated at the translation process. In animal, chronic alcohol intake suppresses the transport of glucose into the frontal and occipital regions of the brain. This finding is validated by a marked decrease in GLUT1 protein expression in brain microvessel (the BBB). In parallel, alcohol intake impairs the BBB tight junction proteins occludin, zonula occludens-1, and claudin-5 in the brain microvessel. Permeability of sodium fluorescein and Evans Blue confirms the leakiness of the BBB. Further, depletion of trans-endothelial electrical resistance of the cell monolayer supports the disruption of BBB integrity. Administration of acetyl-l-carnitine (a neuroprotective agent) significantly prevents the adverse effects of alcohol on glucose uptake, BBB damage and neuronal degeneration.
These findings suggest that alcohol-elicited inhibition of glucose transport at the blood-brain interface leads to BBB malfunction and neurological complications.
KeywordsBlood–brain barrierGlucose transporter proteinTrans-endothelial electrical resistanceAcetyl-l-carnitineTight junctionNeurodegeneration
Glial fibrillar acidic protein
Glucose transporter protein
Human brain endothelial cells
Trans-endothelial electrical resistance
von Willebrand factor
Alcohol is the most commonly used and abused drug that accounts for more than 100,000 deaths and 297,000 disfigured persons each year. About 20 million people are alcoholics or alcohol abusers in the USA (WHO 2007). Chronic alcohol abusers suffer from neurocognitive deficits, neuronal injury, and neuronal loss (Harper 1998; Parsons 1998; Zeigler et al. 2005). Although the central nervous system is a major target of alcohol for causing metabolic and neurological disorders (Oscar-Berman and Marinkovic 2003), the exact mechanism of such neurodegeneration remains unclear. There is strong evidence that neurological disease like Alzheimer’s, Parkinson’s, and stroke are resulted from mitochondrial oxidative damage (Lin and Beal 2006; Maracchioni et al. 2007). In alcohol consumption, the role of mitochondrial oxidative damage has been demonstrated in alcoholic liver damage (Bailey et al. 1999; Cahill et al. 2002; Kessova and Cederbaum 2007; Pastorino et al. 1999) and in neuronal degeneration (Haorah et al. 2008a; Rump et al. 2010). These findings therefore suggest the role of oxidative damage and imbalanced bioenergy homeostasis for neuronal loss and neurological complications. That is because mitochondria are the “power house” and the “oxidative phosphorylation” organelle center for operating the cellular and organ function. In chronic alcohol abuse, Wernicke–Korsakoff neuropathy is an example of energy metabolic disorder syndrome where deficiency of thiamine (a co-factor for pyruvate dehydrogenase) impairs the conversion of pyruvate to acetyl-coenzyme A, resulting in pathogenesis (Harper 1998; Harper et al. 2003).
The brain cells derive 90% of the energy requirements from glucose metabolism (Handa et al. 2000). Therefore, disruption of glucose uptake and utilization in the brain is expected to cause adverse effects on the function and survival of the brain cells. It has been reported that acute ethanol exposure significantly reduces the uptake and utilization of glucose by rat fetal astrocytes culture (Singh et al. 2006). Interestingly, acute administration of ethanol decreases the levels of glucose transporter protein 1 (GLUT1) and GLUT3 in protein extracts from cortical plasma membrane without affecting glucose uptake and utilization in rat brain (Handa et al. 2000). Recent work by Volkow et al. (2006) measures the metabolism of glucose in different regions of the brain by positron emission tomography using 2-deoxy-2-[18F]fluoro-d-glucose in alcoholics (Volkow et al. 2006). Findings from these studies indicate a decline glucose metabolism in the frontal cortex of chronic alcohol abusers but with a significant increase in glucose metabolism during alcohol withdrawal period. These studies clearly show that alcohol negatively interferes with glucose uptake and utilization in the brain.
The rate-limiting factor for glucose transport into the brain is expected to dependent on blood glucose concentration and transporter protein at the blood–brain barrier (BBB) interface. Therefore, disruption of BBB integrity may also hamper the uptake and transport of glucose into the brain. We have shown that alcohol exposure alters the permeability of biomarkers and migration of monocytes across the BBB via oxidative stress-mediated disruption of the interface (Haorah et al. 2005b, 2007b). We previously demonstrated that this alcohol-induced loss of BBB integrity is regulated by activation of myosin light chain kinase (Haorah et al. 2005a), inositol 1,4,5-triphosphate-gated intracellular Ca2+ release (Haorah et al. 2007a), and protein tyrosine kinase-mediated matrix metalloproteinases signaling pathways (Haorah et al. 2008b). The purpose of the present study is to examine whether impairment of glucose uptake and transport at the interface contributes to alcohol-elicited BBB dysfunction and subsequent neuronal loss. At the BBB, the transport of glucose from the arterial vessel circulation into the brain is facilitated by GLUT1 (Leybaert 2005), which exists in highly glycosylated 55-kDa isoform and less glycosylated 45-kDa isoform (Maher et al. 1994; Yeh et al. 2008). Our study shows that GLUT1 55-kDa isoform is highly expressed in primary human brain endothelial cells (hBECs) as well as in mice brain microvessel. We also examined the protective effect of acetyl-l-carnitine (ALC) on glucose uptake and transporter protein from the adverse effect of alcohol insults. ALC, a naturally occurring neuroprotective agent, is currently used for treatment of neurological diseases (Hagen et al. 1998, 2002; Lolic et al. 1997; Pettegrew et al. 1995).
Materials and methods
We purchased the antibodies to GLUT1, neurofilament (NF), tyrosine hydroxylase (TH), choline acetyltransferase (ChAT) from Abcam (Cambridge, MA, USA), antibodies to occludin, claudin-5 and ZO-1 from Zymed (Invitrogen, Carlsbad, CA, USA), and antibody to α-actin from Millipore (Billerica, MA, USA). All secondary Alexa Flour antibodies were purchased from Invitrogen. d-(2-3H)-Glucose (5 mCi, 185 MBq) was purchased from PerkinElmer Life and Analytical Sciences (Waltham, MA, USA). Cytochalasin B (CB), cycloheximide (Chx), actinomycin D (act-D), and ALC were purchased from Sigma-Aldrich (St. Louis, MO, USA).
Primary hBECs were obtained from Dr. Persidsky, Temple University School of Medicine, and hBECs were cultured as described previously (Haorah et al. 2007a). Briefly, all cell culture plates and glass cover slips were pre-coated with type 1 rat-tail collagen (0.09 mg/ml in double-distilled sterile water), aspirated the excess collagen, and dried the plates overnight in sterile hood. The cells were cultured on 96-well plates (20,000 cells/well) for glucose uptake and viability assays, for immunohistochemistry cells were plated on 12-well glass cover slips (40,000 cells/well), and for protein extractions cells were cultured in T 75-cm2 flasks (1 × 106 cells/flask). Dulbecco’s modified Eagle medium (DMEM)/F-12 media containing 10 mM Hepes, 13 mM sodium bicarbonate (pH 7), 10% fetal bovine serum, penicillin, and streptomycin (100 μg/ml each, Invitrogen) was used for cell culture. Cell culture media were changed every 3 days until tight monolayers were formed in about 6–8 days.
Cortical neurons were obtained from our neural tissue core facility, where routinely isolated these cells from elective abortus specimens of human fetal brain tissue. Tissues were obtained in full compliance with the ethical guidelines of both the National Institutes of Health (NIH) and the University of Nebraska. Briefly, dissociated tissues were incubated with 0.25% trypsin for 30 min, neutralized with 10% fetal bovine serum, and further dissociated by trituration. Neurons were cultured on poly-d-lysine pre-coated cover slips and flasks (BD Labware, Bedford, MA, USA) in Neurobasal™ medium containing 0.5 mM glutamine, 50 μg/ml each of penicillin, and streptomycin in combination with GIBCO™ B-27 supplement with antioxidants as described previously (Haorah et al. 2008b). Purity of neurons was assessed by MAP-2 Abs (Chemicon), resulting in 100% enrichment of neurons. Neurons were plated on 12-well glass cover slips (40,000 cells/well), and cell culture media were changed every 3 days until 12–14 days.
In vitro glucose uptake
d-(2-3H)-Glucose uptake was performed in hBECs cultured in 96-well plates following the modified method of Takakura et al. (1991). Cells were exposed to 50 mM ethanol (EtOH) for 24 h with or without 10 μM cytochalasin B, an inhibitor of glucose transporter protein (CB, 10 mM stock was dissolved in dimethyl sulfoxide), or 200 μM ALC in a CO2 incubator. Cells were incubated overnight in glucose-free DMEM/F-12 media containing 2.0 μCi of d-(2-3H)-glucose (specific activity = 20–30 Ci/mmol) and 2.0 mM of non-radiolabeled glucose. After washing off the excess 3H-glucose with phosphate saline buffer (PBS), cellular protein was precipitated with 10% trichloroacetic acid (TCA) at 4°C for 15 min. Following the manufacturer’s instruction, precipitated proteins were transferred onto a 96-well nitrocellulose filter using the Unifilter-96 well Harvester (PerkinElmer, Waltham, MA, USA). Using the Beckman 96-well plate reader, radioactivity was measured by β-top counter. The concentrations of EtOH (50 mM) and CB (10 μM, specific inhibitor of GLUT) were derived from dose- and time-dependent toxicity assays (5–200 mM EtOH and 0.5–100 μM CB for 24–72 h), in which 50 mM EtOH or 10 μM of CB had no cell toxicity effect (by 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyl-tetrazolium bromide assay). The concentration of EtOH from 65 to 100 mM showed about 20% cell death after 48-h exposure. Concentration of CB higher than 20 μM was toxic to hBECs.
Analyses of GLUT1 protein and mRNA levels
Primary human brain endothelial cells cultured in rat-tail collagen and fibronectin pre-coated 75-cm2 flasks were pre-incubated for 15 min with 100 ng/ml of act-D or 10 μg/ml of Chx (Sigma) prior to 16 h treatment with EtOH in the presence of act-D. Cell pellets were lyzed with ice-cold lysis buffer (50 mM Tris-Hcl, pH 7.5) containing 20% glycerol, 5 mM MgCl2, 0.5 mM DTT, and protease inhibitor cocktail for protein extracts that were used for Western blot analyses of GLUT1 protein levels. Total cellular RNA was extracted from cell pellets by RNAeasy kit (Qiagen, Valencia, CA, USA) for GLUT1 mRNA levels. RNA was reverse-transcribed with random hexamers (Promega, Madison, WI, USA). Real-time quantitative PCR was performed with cDNA using an ABI PRISM 7000 sequence detector (Applied Biosystems, Foster City, CA, USA). Thermocycling conditions were 95°C for 10 min, 15 s, and 60°C for 1 min. Human-specific primer pairs for amplification of GLUT1 mRNA by quantitative reverse transcription polymerase chain reaction of total RNA were forward primer 5′-GAG TGC CTG AAA CCA GAG GA-3′ and the reverse primer 5′-CTC ACA CTT GGG AGT CA-3′. Human-specific primer pairs for glyceraldehyde 3-phosphate dehydrogenase (GAPDH) were forward primer: 5′-AAG GCC ATC ACC ATC TTC CA-3′ and reverse primer: 5′-ACC CCA CTT GAT GTT GG CA-3′. Using SYBR Green I detection system, a melting temperature dissociation curve was generated for the PCR product. All PCR reagents were obtained from Applied Biosystems (Foster City, CA, USA). Gene expression was normalized to GAPDH, used as an endogenous control.
For immunocytochemistry, the hBECs were cultured on glass cover slips in 12-well flasks until 80–100% confluent. Cells were treated with 50 mM EtOH with and without CB (10 μM) or ALC (200 μM) for 24 h. Brain tissue sections (8 μm thickness) were derived from chronic EtOH, EtOH + ALC, or pair-fed control mice. Cells and tissue sections were washed with PBS; fixed in acetone–methanol (1:1 v/v) fixative; blocked the cellular antigen with 3% bovine serum albumin at room temperature for 1 h, in the presence of 0.4% Triton X-100; and incubated with respective primary antibodies such as mouse anti-GLUT1 (1:250 dilution), rabbit anti-von Willebrand factor (vWF; 1:150 dilution), rabbit anti-NF (Abcam, 1:250), sheep anti-TH (Abcam, 1:1,000 dilution), and rabbit anti-ChAT (Abcam, 1:100 dilution) for overnight at 4°C. After washing with PBS, cells and tissues were incubated for 1 h with secondary antibody: anti-mouse IgG Alexa Flour 488 for GLUT1, anti-rabbit-IgG Alexa Flour 594 for vWF, anti-rabbit IgG Alexa Flour 488 for NF, anti-sheep IgG Alexa Flour 488 for TH, and anti-rabbit IgG Alexa Flour 488 for ChAT. Cover slips were then mounted onto glass slides with immunomount containing DAPI (Invitrogen), and fluorescence microphotographs were captured by fluorescent microscopy (Eclipse TE2000-U, Nikon microscope, Melville, NY, USA) using NIS elements (Nikon, Melville, NY, USA) software. GLUT1 expression was also analyzed in brain microvessels that were surgically dissected under microscope.
The hBECs cultured in T 75-cm2 flasks were lysed with CellLytic-M buffer (Sigma) for 30 min at 4°C, centrifuged at 1,4000×g, and then total cell lysates protein concentrations were estimated by bicinchoninic acid (Thermo Scientific, Rockford, IL, USA). We loaded 20-μg protein/lane and resolved the various molecular weight proteins by sodium dodecyl sulfate polyacrylamide gel electrophoresis on gradient gels (Thermo Scientific) and then transferred the protein onto nitrocellulose membranes. After blocking, membranes were incubated for overnight with primary antibodies against mouse GLUT1 (1:1,000, Abcam, Cambridge, MA, USA), rabbit occludin (1:250), mouse claudin-5 (1:1,000), and rabbit ZO-1 (1:250) at 4°C followed by 1 h incubation with horseradish peroxidase-conjugated secondary antibodies. Immunoreactive bands were detected by West Pico chemiluminescence substrate (Thermo Scientific). Data were quantified as arbitrary densitometry intensity units using the Gelpro32 software package (Version 3.1, Media Cybernetics, Marlow, UK).
In vivo glucose transport assay
Five-week-old C57BL/6J male mice from Jackson Laboratory (Bar Harbor, ME, USA) were housed in University of Nebraska Medical Center animal facility following NIH Guide for Care and Use of Laboratory Animals. Mice were acclimated to Lieber–DeCarli control and 28% calorie (4% vol/vol) ethanol liquid diets from Dyets Inc. for 5 days prior to 8- to 9-week weight-match pair-feeding regimens. Pair feeding was based on the amount of liquid diets consumed by ethanol group animals. In mouse model, 4% ethanol is considered as moderately high alcohol concentration; thus, in this study, the effects of ethanol on glucose transport and BBB function referred to high-dose chronic effects. ALC (1.0 mg/ml) was given daily in liquid diets mixture for 8 to 9 weeks. Experimental conditions consisted of control liquid diets, EtOH liquid diets, ALC in EtOH liquid diets, and ALC in control liquid diets. After week 8 of liquid-diet feeding, d-(2-3H)-glucose (4 μCi) and 4.0 mM non-radiolabeled glucose in 500 μl of saline were cavaged orally for control, EtOH, and EtOH + ALC animals. After 1 h, mice were euthanized, microvessels were removed carefully, and tissues were dissected from different brain regions. Known tissue weights were homogenized with 100 μl of Krebs–Ringer phosphate HEPES buffer, centrifuged at 12,000 rpm for 15 min, and 20 μl of supernatants from each condition was mixed with 4 ml of scintillation fluid. The levels of 3H-glucose in the samples were detected by liquid scintillation counter (Beckman) along with a standard curve of 3H-glucose that was run in parallel. Results were extrapolated from the standard curve, and data were expressed as nanocurie per gram tissue weight.
In vivo BBB permeability assay
Using the sodium fluorescein (NaFl) and Evans Blue (EB) tracer dye mixtures (5 μM each), the effect of EtOH on BBB permeability was examined in acute and chronic animal model following an established method (Hawkins and Egleton 2006). In acute studies, mice were anesthetized, infused 100 μl of 200 μM of ALC or 650 mM of EtOH through the common carotid artery for 1 h, and followed by infusion of NaFl/EB mixtures for another 30 min. ALC was infused at 30 min prior to EtOH infusion. In chronic studies, the conditions of control, EtOH, and EtOH + ALC were same as that of glucose uptake assay. Mixtures of NaFl/EB dyes were directly infused into the right carotid artery. Animals were decapitated; brains were removed, dissected, weighed, and homogenized in 600 μl 7.5% (w/v) TCA. Resulting suspensions were divided into two aliquots (300 μl each). One aliquot was neutralized with 50 μl of 5 N NaOH and measured by flourimetry on a GENios microplate reader (excitation 485 nm, emission 535 nm) for NaFl determination. The other aliquot was centrifuged for 10 min at 10,000 rpm 4°C, and the supernatant was measured by absorbance spectroscopy at 620 nm for EB determination. Standard curve was constructed by serial dilutions of a standard EB/NaFl solution in 7.5% TCA.
To determine the integrity of BBB function, changes in trans-endothelial electrical resistance (TEER) across the BBB were analyzed by a highly sensitive 1600R ECIS system (Applied Biophysics, Troy, NY, USA). The ECIS system provides real-time monitoring of changes in TEER. In brief, hBECs at 20,000 cells/well were plated on type I rat-tail collagen-coated 8W10E electrode arrays (Applied Biophysics). Once tight cell monolayers were formed, stable TEER readings were monitored for 2 h prior to treatment of cell monolayers with 50 and 100 mM EtOH in the presence or absence of CB or ALC. We then monitored the TEER readings for another 10 h at 400 Hz and with 10-min intervals. Confluent cell monolayers demonstrated baseline TEER readings of 1,100 to 1,200 Ω/cm2.
Values are expressed as the mean ± SD. Within an individual experiment, each data point was determined from three to five replicates. Statistical analysis of the data was performed by using GraphPad Prism V5 (Sorrento Valley, CA, USA). Comparisons between samples were performed by one-way ANOVA with Dunnett’s post hoc test. Differences were considered significant at P values ≤0.05.
Ethanol inhibits glucose uptake and GLUT1 expression in hBECs
ALC protects glucose uptake in animal model
Disruption of glucose uptake impairs BBB integrity
BBB disruption promotes neuronal loss
Glucose uptake, transport, and metabolism provide 90% of the energy in the brain under normal physiological condition. Since endothelial cells (primary cell component of BBB) regulate the delivery of glucose into the brain, the BBB endothelium plays a vital role for the supply of energy and nutrients to brain cells (Lund-Andersen et al. 1976). Thus, malfunctioning of the BBB is expected to adversely affect the uptake, transport, and utilization of glucose by different types of brain cells (neurons and glial cells) and their very survival. Our finding is the first to show that inhibitory effect of alcohol on glucose uptake and transport by brain endothelial cells causes BBB dysfunction and neuronal degeneration.
In the present study, the concentration of EtOH was determined from dose- and time-dependent effect on glucose uptake. Concentration of glucose was based on cellular glucose saturation uptake and that of cytochalasin B was determined by cell viability assay. Concentration of EtOH in chronic study was based on our animal model of ethanol liquid diets and pair-fed control. The concentration of glucose used in glucose-free media was justified by our cellular glucose saturation uptake study. Note that the cells were exposed EtOH but not starved with glucose; thus, the saturation of glucose uptake was less than 1.0 mM, which was within the 3.4 mM intracellular concentration of glucose in endothelial cells (Pardridge and Boado 1993).
The present findings show that alcohol significantly inhibited the uptake and transport of glucose across the BBB, in which the in vitro finding was validated by in vivo data. Decrease in glucose uptake in cells and glucose transport across the BBB in mice correlated with reduction of GLUT1 protein expression in cell culture and in brain microvessels. Clearly, ethanol decreased the levels of highly glycosylated 55-kDa GLUT1 protein in microvessel and the levels of the less glycosylated 45-kDa isoform in the brain. Co-localization of GLUT1 protein with astrocytes marker (glial fibrillar acidic protein) suggests that 45-kDa GLUT1 appeared to localize in the perivascular end-feet of astrocytes, similar to the findings of others (Maher et al. 1994; Yeh et al. 2008). Functionally, the 55-kDa GLUT1 is actively involved at the BBB for transport of glucose into the brain, whereas the 45-kDa GLUT1 is less active glucose transporter. This is because packaging of glucose into the transporter protein happens only when GLUT1 is glycosylated by acetylglucosamine. Since glycosylation of GLUT1 regulates the functional and structural integrity (Maher et al. 1994), it is possible that once the 55-kDa GLUT1 releases the glucose at the end-feet of astrocytes, it becomes less glycosylated and remains as 45-kDa isoform.
It was concluded that reduction in GLUT1 protein level was in part due to defective translocation of mRNA for biosynthesis of GLUT1 protein. This rationale was based on the similar effect of Chx, which inhibits protein synthesis by interfering the translocation of mRNA in the ribosomal reading frame. Treatment of cells with act-D did not affect the GLUT1 protein level (Fig. 2b), which was due to stabilization of GLUT1 mRNA levels by ethanol exposure. Interestingly, Boado (1998) showed that the brain-derived peptide preparation Cerebrolysin stabilizes the levels of GLUT1 mRNA in endothelial cells. Indeed, this is very similar to our previous findings that ethanol reduces the levels of insulin-like growth factor receptor II protein without affecting the IGFR II mRNA levels (Haorah et al. 2003).
Both in acute and in chronic conditions, alcohol negatively interferes the metabolism of glucose in the brain (Handa et al. 2000; Volkow et al. 2006). The question is whether the inhibitory effect of alcohol on glucose metabolism is due to inhibition of glucose transport at the BBB or the direct inhibition of glucose uptake by neurons and glial cells without affecting GLUT1 protein at the BBB. Our data suggest that inhibition of glucose transport across BBB by ethanol contributes to deprivation and defective metabolism of glucose in the brain. This notion is further supported by the fact that impairment of glucose uptake and GLUT1 protein expression in brain endothelium by alcohol is accompanied by breakdown of BBB integrity. That is because the dynamic function of the BBB is operated by a constant supply of glucose energy source. Therefore, alcohol-induced inhibition of endothelial glucose uptake is likely to disrupt the integrity of BBB function as well as the normal deliver of glucose into the brain leading to ultimate neurotoxicity and neuronal degeneration. Cytochalasin B, an inhibitor of GLUTs, decreased the rate of glucose uptake without affecting glucose transporters protein levels as indicated by immunocytochemistry and Western blot analyses. These results show that CB inhibited glucose uptake and transport by modulating the active binding sites of GLUTs and not by degrading the actual GLUTs protein contents.
The stabilization of glucose uptake, GLUT1 protein levels, and BBB integrity by ALC demonstrates the beneficial effect for improving the deleterious health condition of alcohol abusers. The complete penetration of ALC across the BBB (Kido et al. 2001) and our recent findings on neuroprotective mechanisms of ALC from alcohol-induced oxidative damage in the brain (Rump et al. 2010) strongly advocate the attractive therapeutic application for chronic alcohol abusers. ALC would stabilize glycosylation of GLUT1 protein by donating acetyl group to glucosamine so as to enhance the activity of acetylglucosamine, which glycosylates GLUT1 protein. Taken together, our findings indicate that ALC is a potent protective agent for neurological disorders.
In conclusion, our data suggest that EtOH-induced inhibition of glucose transport leads to BBB dysfunction, which fails to deliver adequate amount of glucose into the brain that exacerbates neurotoxicity and neuronal degeneration in vitro and in vivo. Further, therapeutic application of ALC would be beneficial for improving the health of EtOH abusers.
This work was supported in part by NIH/NIAAA grant AA016403-01A2 (to JH) and by University of Nebraska Medical Center Faculty Retention Fund. Primary human endothelial cells were kindly provided by Dr. Yuri Persidsky.
Conflict of interest
The authors declare that they have no conflict of interest.