|Home | About | Journals | Submit | Contact Us | Français|
We have been using polarized, hepatic WIF-B cells to examine ethanol-induced liver injury. These cells polarize in culture and maintain numerous liver-specific activities including the ability to metabolize alcohol. Previously, we found that microtubules were more highly acetylated and more stable in ethanol-treated WIF-B cells and that increased microtubule acetylation required ethanol metabolism and was likely mediated by acetaldehyde. This study was aimed at identifying the mechanism responsible for increased microtubule acetylation. We examined the expression of two known microtubule deacetylases, histone deacetylase 6 (HDAC6) and Sirtuin T2 (SirT2), in WIF-B cells. Immunoblotting, immunofluorescence microscopy, and assays using the SirT2 inhibitor nicotinamide revealed that WIF-B cells do not express SirT2. In contrast, HDAC6 was highly expressed in WIF-B cells. Addition of trichostatin A (TSA), an HDAC6 inhibitor, induced microtubule acetylation to the same extent as in ethanol-treated cells (approximately threefold). Although immunofluorescence labeling revealed that HDAC6 distribution did not change in ethanol-treated cells, immunoblotting showed HDAC6 protein levels slightly decreased. HDAC6 solubility was increased in nocodazole-treated cells, suggesting impaired microtubule binding. Direct microtubule binding assays confirmed this hypothesis. The decreased microtubule binding was partially prevented by 4-methyl pyrazole, indicating the effect was in part mediated by acetaldehyde. Interestingly, HDAC6 from ethanol-treated cells was able to bind and deacetylate exogenous tubulin to the same extent as control, suggesting that ethanol-induced tubulin modifications prevented HDAC6 binding to endogenous microtubules.
We propose that lower HDAC6 levels combined with decreased microtubule binding lead to increased tubulin acetylation in ethanol-treated cells.
The liver is the major site of ethanol metabolism and thus sustains the most injury from chronic alcohol consumption. Although the progression of alcoholic liver disease is well described, little is known about the molecular mechanisms that promote alcohol-induced hepatotoxicity. Our studies have been performed in WIF-B cells, an emerging model system for studying liver injury. These cells polarize in culture and exhibit many liver-specific activities.1 Importantly, WIF-B cells metabolize alcohol using endogenous alcohol dehydrogenase, acetaldehyde dehydrogenase, and cytochrome P450 2E1.2 Recently, we determined that microtubules were more stable and hyperacetylated in ethanol-treated WIF-B cells.3 This phenotype required alcohol metabolism and was likely mediated by acetaldehyde. We also determined that microtubules were acetylated to the same extent in livers from ethanol-fed rats, indicating the effect has physiological relevance.3 However, the mechanism responsible for the ethanol-induced tubulin hyperacetylation and increased stability is not known.
Protein acetylation results from the coordinated activities of acetyltransferases and deacetylases.4,5 Although α–tubulin is known to be acetylated on lysine 40, the identity of the acetyltransferase is not known.6 However, there are two known microtubule deacetylases, histone deacetylase 6 (HDAC6) and Sirtuin T2 (SirT2).7–10 These enzymes are part of a larger family of at least 18 members known collectively as histone deacetylases.11 The family is divided into three classes based on sequence identity and co-factor/co-enzyme dependence. Although most function in the nucleus to deacetylate histones and other transcriptional machinery,11,12 HDAC6 (class II) and SirT2 (class III) are cytosolic deacetylases that colocalize with microtubules in some cell types.7–10 Overexpression of either enzyme led to a loss of acetylated microtubules and decreased microtubule stability.7–10 Conversely, when the deacetylases were inactivated or their expression knocked down, microtubules were hyperacetylated and more stable.7–10 Because HDAC6 is enriched in liver13,14 (its expression in hepatic cell lines and isolated hepatocytes has not been examined) and because class III enzymes, including SirT2, require the oxidized form of reduced nicotinamide adenine dinucleotide for activity (the oxidized form of reduced nicotinamide adenine dinucleotide is a coenzyme for alcohol dehydrogenase and acetaldehyde dehydrogenase),9 both HDAC6 and SirT2 are good candidates for mediating alcohol-induced microtubule hyperacetylation.
We found that WIF-B cells do not express SirT2, consistent with reports that SirT2 expression is enriched in brain.15,16 However, HDAC6 was abundantly expressed in WIF-B cells, and its inhibition with trichostatin A (TSA) led to increased microtubule acetylation comparable to that observed in ethanol-treated cells. Although alcohol or TSA did not alter HDAC6 subcellular distribution, it led to decreased HDAC6 protein levels. Furthermore, HDAC6 binding to microtubules was significantly impaired in ethanol-treated cells. HDAC6 from ethanol-treated cells bound to and deacetylated exogenous tubulin to the same extent as HDAC6 from control cells. Thus, impaired binding of HDAC6 to microtubules is likely attributable to ethanol-induced tubulin modifications that prevented associations. Together these results indicate that the alcohol-induced increases in microtubule acetylation and stability in WIF-B cells were attributable, in part, to decreased HDAC6 protein levels and decreased microtubule binding.
F12 (Coon’s modification) medium, 4-methylpyrazole, TSA, nocodazole, nicotinamide (NTA), sodium butyrate, and taxol were purchased from Sigma-Aldrich (St. Louis, MO). Bovine brain tubulin was from Cytoskeleton (Denver, CO) and was stored as a 10 mg/mL stock in PEM (100 mM Pipes, 1 mM ethylene glycol tetraacetic acid, 1 mM MgSO4, pH 6.6) at −70°C. Fetal bovine serum was purchased from Gemini Bio-Products (Woodland, CA), and 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid was from HyClone (Logan, UT). Alexa488-conjugated and 568-conjugated secondary antibodies were purchased from Invitrogen (Carlsbad, CA). Horseradish peroxidase– conjugated secondary antibodies and monoclonal antibodies against α-tubulin or acetylated α-tubulin were from Sigma-Aldrich. Polyclonal antibodies against the HDAC6 C-terminus (H-300) or the SirT2 N-terminus were from Santa Cruz Biotechnologies (Santa Cruz, CA). The polyclonal antibody against the SirT2 C-terminus (PA3-200) was from Affinity Bio Reagents (Golden, CO). HDAC colorimetric activity assay kits were purchased from BioMol International (Plymouth Meeting, PA).
HeLa cells were grown in a humidified incubator in 5% CO2 at 37°C in Dulbecco’s modified Eagle’s medium supplemented with 10% fetal bovine serum and were seeded at 1 × 106 cells/10-cm dish and cultured for 4 to 5 days. WIF-B cells were grown in 7% CO2 in Coon’s-modified F-12, pH 7.0, supplemented with 5% fetal bovine serum, 10 μM hypoxanthine, 40 nM aminopterin, and 1.6 μM thymidine.17 Cells were seeded onto glass coverslips at 1.3 × 104 cells/cm2 or at 1 × 106 cells/10-cm dish and cultured for 8 to 12 days until they reached maximum density and polarity. Cells grown on dishes or coverslips were treated on days 3 or 7, respectively, with 50 mM ethanol buffered with 10 mM 4-(2-hydroxyethyl)-1-piperazine ethanesulfonic acid, pH 7.0, at 37°C for 72 hours as described.2
WIF-B cells were fixed for 5 minutes with methanol at −20°C. Cells were processed for indirect immunofluorescence as described,17 using anti-HDAC6 polyclonal (1:100) and anti-α-tubulin monoclonal (1:400) antibodies and Alexa488-conjugated or 568-conjugated secondary antibodies (5 μg/mL). Epifluorescence was visualized using an Olympus BX60 Microscope (OPELCO, Dulles, VA). Images were acquired using a Coolsnap HQ2 camera (Photometrics, Tucson, AZ) and IPLabs image analysis software (Biovision, Exton, PA). Images were processed and figures compiled using Adobe Photoshop (Adobe Systems Inc., Mountain View, CA).
In general, cells were lysed directly into Laemmli sample buffer18 and boiled for 3 minutes. Proteins were electrophoretically separated using sodium dodecyl sulfate polyacrylamide gel electrophoresis and transferred to nitrocellulose. Samples were immunoblotted with antibodies specific to α-tubulin (1:7500), acetylated-α-tubulin (1:4000), HDAC6 (1:2000), or the SirT2 N-terminus (1:4000). Immunoreactivity was detected using enhanced chemiluminescence (PerkinElmer, Crofton, MD). Relative levels of tubulin or HDAC6 were determined by densitometric analysis of immunoreactive bands.
Cells grown on 10-cm dishes were rinsed with phosphate-buffered saline, detached with trypsin for 2 minutes at 37°C, and pelleted by centrifugation. Cells from two dishes were pooled, resuspended in 5 mL ice-cold swelling buffer (1 mM MgCl2, 1 mM dithiothreitol, 1 mM ethylene glycol tetraacetic acid), and incubated 5 minutes on ice. Cells were pelleted by centrifugation and resuspended in 500 μL 0.25 M sucrose with added protease inhibitors [1 μg/mL each of leupeptin, antipain, phenylmethanesulfonylfluoride, and benzamidine] and Dounce-homogenized with a tight-fitting pestle for 20 strokes. Homogenates were centrifuged at 900g at 4°C for 5 minutes, and the pellet was washed and recentrifuged at 14,200g at 4°C for 10 minutes to collect nuclei. The postnuclear supernatant was centrifuged at 150,000g at 4°C for 60 minutes to prepare the cytosolic fraction. The colorimetric activity assay was performed according to the manufacturer’s specifications. Total protein was determined using bicinchoninic acid reagent (Pierce, Rockford, IL), and activity was calculated as nanomoles per minute per microgram total protein.
To determine tubulin deacetylase activity directly, cytosolic extracts were incubated with 1.8 μM bovine brain tubulin for the indicated times at 37°C. Reactions were stopped by the addition of sample buffer and boiling for 3 minutes. Samples were immunoblotted for acetylated-α-tubulin, and the relative amounts were determined using densitometry. Zero-minute values were set to 100%, from which the percentage deacetylation (decreased immunoreactivity) was determined.
WIF-B cells were pretreated with 50 mM ethanol for 72 hours or 50 nM TSA for 15 minutes. In the continued absence or presence of either agent, cells were treated with 33 μM nocodazole for the indicated times then extracted in prewarmed PEM containing 0.15% Triton X-100 for 1 minute at 37°C. Supernatants containing the released soluble tubulin and cytosolic proteins were collected. The permeabilized cells containing intact microtubule polymers and microtubule-associated proteins were washed in PEM without Triton X-100 and lysed directly into sample buffer. Samples were immunoblotted for HDAC6, and the percentage soluble HDAC6 was determined by densitometry.
Microtubules were purified based on two previously published methods.10,19 Briefly, cells grown on 10-cm dishes were resuspended in 500 μL ice-cold PEM and homogenized as described. The homogenate was centrifuged for 60 minutes at 4°C at 150,000g to prepare a cytosolic extract. Taxol (10 μM) was added to the supernatant for 15 minutes at 37°C to polymerize microtubules. The microtubules were pelleted through a 10% sucrose cushion at 150,000g at 20°C for 30 minutes. The resultant supernatant and pelleted fraction containing the microtubules and associated proteins were immunoblotted for HDAC6 and α-tubulin. The distributions of HDAC6 in the soluble or pelleted fractions were determined by densitometry.
Results are expressed as the mean ± standard error of the mean (SEM). The Student t test was performed for paired data to assess statistical significance. P ≤ 0.05 were considered significant.
We first determined HDAC6 and SirT2 protein expression by immunoblotting. We began with HeLa cells, a cell line known to be positive for both proteins. A 160-kDa immunoreactive doublet and a 43-kDa band were detected (Fig. 1A), corresponding to the predicted molecular weights of HDAC6 and SirT2, respectively. However, only the 160-kDa HDAC6 doublet was detected in WIF-B cells (Fig. 1A). To further confirm that SirT2 was absent in WIF-B cells, we assayed microtubules for increased acetylation in the presence of the SirT2 inhibitor NTA. No increased acetylation was observed with increased time of NTA incubation, and in fact, a slight (but not statistically significant) decrease was observed (Fig. 1B). This result was also observed with 50 mM NTA (data not shown). Furthermore, when cells were immunolabeled for SirT2, no specific signal was detected (data not shown). We failed to detect SirT2 using five fixation methods, antibodies to the SirT2 N- or C-terminus, or increased antibody concentrations (data not shown). From these results, we conclude that WIF-B cells lack SirT2, a finding that is consistent with reports that SirT2 expression is enriched in brain.15,16
Because HDAC6 inactivation has been shown to increase microtubule acetylation,7,8,10 we examined whether its inactivation with TSA in WIF-B cells led to the same effect. As shown in Fig. 1C, 50 nM TSA for 30 minutes led to a 2.3-fold increase in acetylated tubulin, similar to the increased acetylation observed in ethanol-treated cells (2.62 ± 0.48).3,20 Importantly, total tubulin levels did not change, indicating that the hyperacetylation was not attributable to increased protein levels. Thus, we conclude that HDAC6 is an abundant tubulin deacetylase in WIF-B cells.
One simple explanation for increased microtubule acetylation in ethanol-treated cells is decreased HDAC6 protein levels. To test this, we immunoblotted control and ethanol-treated cell lysates (Fig. 2A). After incubation for 24 hours with ethanol, HDAC6 protein levels decreased to 92.0% ± 7.4% (Fig. 2B). Levels continually declined, and after 72 hours, there was 75.0% ± 6.1% of HDAC6 control levels (P < 0.05). This decrease in HDAC6 was not the result of overall decreased total protein levels, because the cytosolic protein concentration in ethanol-treated cells was 96.5% ± 3.14% of control, far less than the decrease in HDAC6 protein.
Another possible mechanism for alcohol-induced microtubule hyperacetylation is changes in HDAC6 microtubule binding. To test this possibility, we first immunolabeled control and treated cells for HDAC6 and tubulin. Previously, we observed that ethanol promoted the formation of microtubules that appeared thicker and more gnarled, features of stable microtubules,3 whereas in TSA-treated cells, the change in morphology was much less pronounced.20 Similar results were obtained here (Fig. 3). Although HDAC6 has been shown to colocalize with microtubules in other cell types,7,8,10 we observed only diffuse cytosolic staining in WIF-B cells (Fig. 3A and B; unlabeled BCs are marked with asterisks). In cells treated with TSA (Fig. 3A) or ethanol (Fig. 3B), no overt changes in HDAC6 cytosolic distributions were observed.
Because the high level of cytosolic HDAC6 may have obscured its microtubule association, we examined HDAC6 –microtubule binding by two methods. First, we used an indirect method in which microtubules in intact cells were depolymerized with nocodazole and cells assayed for released HDAC6. Cells were pretreated for 72 hours with 50 mM ethanol or 15 minutes with 50 nM TSA. In the continued absence or presence of either agent, 33 μM nocodazole was added to depolymerize microtubules. WIF-B cells were lysed in a microtubule stabilizing buffer containing 0.15% TX-100 releasing soluble tubulin (and other soluble proteins), leaving the stable polymeric tubulin and associated proteins in the intact cells. In control cells, approximately 45% of HDAC6 was soluble, and in the presence of nocodazole, solubility was slightly increased to approximately 55% after 60 minutes (Fig. 4). In ethanol-treated cells, significantly more HDAC6 was soluble, and after 60 minutes in nocodazole, more than 80% of HDAC6 was released (P < 0.02). Similarly, in TSA-treated cells, approximately 90% of HDAC6 was soluble after 60 minutes (P < 0.05), suggesting impaired microtubule binding.
We next measured direct microtubule binding. We prepared cytosolic fractions from control and treated cells and added taxol to polymerize microtubules. The stabilized microtubules and associated proteins were pelleted through a sucrose cushion and samples were immunoblotted for HDAC6 and α-tubulin. In both control and ethanol-treated cells, all of the tubulin was pelleted, indicating complete polymerization (Fig. 5A). In contrast, much less HDAC6 was detected in the microtubule pellet from ethanol-treated cells. In control cells, 60.7% ± 11.5% of HDAC6 was detected in the pellet, whereas less than half that amount (21.9% ± 4.1%, P < 0.03) bound microtubules in ethanol-treated cells. When cells were treated with ethanol and 4-methylpyrazole (an alcohol dehydrogenase inhibitor), HDAC6 microtubule association remained closer to control levels (approximately 40%) (Fig. 5B), indicating that HDAC6’s altered microtubule binding partially required ethanol metabolism and may be mediated by acetaldehyde.
Another possible mechanism for increased microtubule acetylation in ethanol-treated cells is that HDAC6 activity was impaired. To test this, we assayed total deacetylase activity using a colorimetric assay with acetylated lysine as the substrate. Because this assay cannot discriminate between nuclear and cytoplasmic HDAC isoforms, we prepared both fractions from control and ethanol-treated WIF-B cells. Consistent with our hypothesis, nuclear HDAC activity was decreased (67.8% ± 5.5% of control, P < 0.003) in ethanol-treated cells. In contrast, cytosolic HDAC activity was increased (125.6% ± 8.3% of control, P < 0.007) (Fig. 6A).
To characterize this somewhat surprising finding, we first confirmed that the cytosolic activity was predominantly attributable to HDAC6. Although HDAC6 is exclusively cytosolic,7,10,21,22 some of the HDAC family members shuttle between the nucleus and cytosol.11 Because all of the HDAC family members are sensitive to TSA and sodium butyrate whereas HDAC6 is sensitive only to TSA,7 we measured HDAC activity in the presence of these inhibitors. As predicted, both TSA and sodium butyrate inhibited nuclear activity by approximately 50% (Fig. 6B). In contrast, only TSA inhibited the cytosolic activities; sodium butyrate had no effect (Fig. 6B). Thus, the cytosolic deacetylase activity we were measuring was mainly attributable to HDAC6.
Our next step to explain increased cytosolic activity in ethanol-treated cells was to consider the reaction mixtures. Both the acetylated lysine (supplied by the manufacturer) and tubulin (copurified in the cytosolic fraction) were present, both of which are substrates for HDAC6. As shown in Figs. 4 and and5,5, ethanol-treatment led to decreased HDAC6 binding to microtubules. Thus, one possibility is that in the ethanol-treated samples, more HDAC6 was dissociated from tubulin and therefore more available for binding to the acetylated lysine substrate, leading to an apparent increase in activity. To test this hypothesis, we added increasing concentrations of bovine brain tubulin (which was acetylated; data not shown) to the assay mixtures to compete for HDAC6 binding to acetylated lysine. Importantly, the exogenous tubulin concentrations (0.9–4.5 μM) were far lower than the concentrations of acetylated lysine (1.5 mM) or endogenous tubulin (10 μM) (approximately 350-fold to 700-fold and approximately 2.5-fold to 10-fold, respectively). Tubulin addition had no significant effect on HDAC6 activity in control cells (Fig. 7A). In contrast, there was a dose-dependent decrease in activity in ethanol-treated samples. Thus, HDAC6 preferentially associated with the low levels of exogenous tubulin, preventing interactions with the acetylated lysine substrate leading to decreased apparent activity.
These results further suggested that HDAC6’s ability to bind and deacetylate exogenous tubulin activity was not altered by ethanol treatment. To test this directly, cytosolic fractions from control and ethanol-treated WIF-B cells were incubated with 1.8 μM exogenous tubulin for the indicated times and immunoblotted for acetylated tubulin. In both control and ethanol-treated samples, there was a time-dependent increase in tubulin deacetylation (Fig. 7B). When quantitated (Fig. 7C), no change in the tubulin deacetylase activity in the ethanol-treated samples was observed; the plots were nearly superimposable. These results indicate that HDAC6 binding and deacetylation of exogenous tubulin were not changed by ethanol treatment. We conclude that ethanol treatment was not impairing HDAC6 deacetylase or microtubule-binding activities directly. Rather, we propose that tubulin from ethanol-treated cells was modified (see Discussion), preventing HDAC6 microtubule binding.
Previously, we determined that microtubules were more highly acetylated and more stable in ethanol-treated WIF-B cells and in livers from ethanol-fed rats.3 In this study, we sought to identify the mechanism by which microtubules become hyperacetylated. Because a tubulin acetyltransferase has not been identified, we focused on two known tubulin deacetylases, HDAC6 and SirT2. We found that only HDAC6 was expressed in WIF-B cells, and that its protein levels decreased by 25% in ethanol-treated cells. We also found that HDAC6 binding to endogenous microtubules was significantly impaired in ethanol-treated cells and that this impairment partially required ethanol metabolism. Measuring HDAC6 tubulin deacetylase activity by two methods further revealed that ethanol did not impair HDAC6’s ability to bind or deacetylate exogenous tubulin. This suggests that tubulin from ethanol-treated cells was modified, thereby preventing HDAC6 binding.
Our studies showed that HDAC6 protein levels were decreased by ethanol treatment, and its ability to bind microtubules was impaired. These two findings may explain the increased microtubule acetylation and stability observed in ethanol-treated WIF-B cells. Although decreased HDAC6 protein levels are a simple explanation for increased tubulin acetylation, we propose that the impaired microtubule binding has more impact. HDAC6 is abundant in the liver13,23 such that a 25% decrease in levels may not likely have profound effects on tubulin acetylation. Rather, the 70% impairment in HDAC6 binding to microtubules may have a more dramatic effect; much less of the available enzyme can bind its substrate, leading to decreased deacetylation. However, our studies cannot rule out the possibility that increased acetyltranferase activity is also contributing to increased tubulin acetylation. We are currently examining this possibility.
Based on the results from our efforts to explain the somewhat surprising finding that deactylase activity was enhanced in ethanol cytosolic extracts, we propose that alcohol-induced tubulin modifications prevented HDAC6 binding to microtubules. We determined that addition of exogenous tubulin to our reaction mixtures impaired deacetylase activity in ethanol-treated samples only. Importantly, the exogenous tubulin concentrations were approximately 350-fold to 1700-fold lower than the concentrations of acetylated lysine or approximately 2.5-fold to 10-fold lower than endogenous tubulin. Thus, the decreased activity in the ethanol-treated samples strongly argues that HDAC6 preferentially bound the exogenous tubulin, displacing it from the acetylated lysine substrate. In control samples, where HDAC6 microtubule binding properties were not changed, the low amounts of exogenous tubulin did not further displace HDAC6 from the acetylated lysine (or the unmodified endogenous tubulin) such that activity remained the same. When assayed directly, there was no difference between control or treated HDAC6 activity using exogenous tubulin, indicating that ethanol was not altering HDAC6 microtubule binding or catalytic activity directly. Rather, we conclude that endogenous tubulin from the ethanol-treated samples was modified, thereby preventing HDAC6 binding.
What is the nature of the alcohol-induced tubulin modification? In this study, we determined that impaired HDAC6 binding to microtubules was partially mediated by acetaldehyde, and from our previous work, we determined that alcohol-induced tubulin hyperacetylation was likely mediated by acetaldehyde. Because this highly reactive ethanol metabolite can readily, covalently modify a highly reactive lysine in α-tubulin in vitro,24 one provocative possibility is that tubulin acetaldehyde adducts impede HDAC6 binding. Because decreased HDAC6 binding to microtubules was only partially prevented by 4-methylpyrazole, we cannot rule out the possibility that other reactive ethanol metabolites formed detrimental tubulin adducts. Future studies are needed to sort this out.
We determined that ethanol treatment decreased nuclear HDAC activity to 67.8% ± 5.5% of control. This reduction may have significant effects on hepatic gene expression. Although traditionally it has been thought that histone acetylation promotes DNA relaxation, which better allows for transcription factor binding and increased transcription,25 an emerging hypothesis is that the dynamic turnover of histone modifications, rather than continued acetylation, leads to increased transcription.26,27 In fact, histone hyperacetylation has been shown to lead to nucleosome instability that allows transcription at cryptic promoters, resulting in aberrant transcription.26,27 Thus, altered histone acetylation may not only explain the decreased levels of HDAC6 in ethanol-treated cells due to changes in transcriptional regulation, but also may have profound effects on general hepatic gene expression.
Many proteins are known to be reversibly acetylated, including histones, transcription factors, nuclear import factors, p53, and α-tubulin.4,5 Large families of acetyltransferases and deacetylases have also been identified4,5 whose substrate specificities are not yet all defined. With this growing number of acetylated proteins and large number of modifying enzymes, it is likely that many hepatic proteins are hyperacetylated in ethanol-treated cells. Consistent with this hypothesis is the finding that histone H3 was hyperacetylated in acutely treated isolated hepatocytes.28 Furthermore, the expression levels of p300, a histone acetyltransferase, increased twofold in ethanol-treated hepatocytes.29 Thus, we predict many other hepatic proteins are acetylated in ethanol-treated cells, and we are currently using a proteomics approach to identify them.
Supported by the National Institute of Alcohol Abuse and Alcoholism (R21 AA015683) awarded to P.L.T.
Potential conflict of interest: Nothing to report.