HDAC inhibitor SAHA normalizes the levels of VLCFAs in human skin fibroblasts from X-ALD patients and downregulates the expression of proinflammatory cytokines in Abcd1/2-silenced mouse astrocytes.

X-adrenoleukodystrophy (X-ALD) is a peroxisomal metabolic disorder caused by mutations in the ABCD1 gene encoding the peroxisomal ABC transporter adrenoleukodystrophy protein (ALDP). The consistent metabolic abnormality in all forms of X-ALD is an inherited defect in the peroxisomal β-oxidation of very long chain FAs (VLCFAs >C22:0) and the resultant pathognomic accumulation of VLCFA. The accumulation of VLCFA leads to a neuroinflammatory disease process associated with demyelination of the cerebral white matter. The present study underlines the importance of a potent histone deacetylase (HDAC) inhibitor, suberoylanilide hydroxamic acid (SAHA) in inducing the expression of ABCD2 [adrenoleukodystrophy-related protein (ALDRP)], and normalizing the peroxisomal β-oxidation, as well as the saturated and monounsaturated VLCFAs in cultured human skin fibroblasts of X-ALD patients. The expression of ELOVL1, the single elongase catalyzing the synthesis of both saturated VLCFA (C26:0) and monounsaturated VLCFA (C26:1), was also reduced by SAHA treatment. In addition, using Abcd1/Abcd2-silenced mouse primary astrocytes, we also examined the effects of SAHA in VLCFA-induced inflammatory response. SAHA treatment decreased the inflammatory response as expression of inducible nitric oxide synthase, inflammatory cytokine, and activation of NF-κB in Abcd1/Abcd2-silenced mouse primary astrocytes was reduced. These observations indicate that SAHA corrects both the metabolic disease of VLCFA as well as secondary inflammatory disease; therefore, it may be an ideal drug candidate to be tested for X-ALD therapy in humans

matter structure and function ( 46 ) by HDAC inhibitors strongly point toward acetylation-dependent mechanisms in demyelination and neurodegeneration.
Although treatment of cancer has been the primary target for the clinical development of HDAC inhibitors, administration of HDAC inhibitors has also shown benefi cial effects in some noncancer disorders, such as sickle cell anemia, muscular dystrophy, neurodegenerative diseases, and infl ammatory disorders ( 47 ). Among the various classes tested, suberoylanilide hydroxamic acid (SAHA) was reported as the most promising therapeutic agent for treatment of spinal muscular atrophy (SMA) due to its ability to substantially increase survival motor neuron (SMN) protein levels at low micromolar concentration and completely inhibit HDAC activity.
The present study demonstrates that treatment with the potent and selective HDAC inhibitor SAHA ( 48,49 ) normalized the levels of VLCFA in skin fi broblasts from X-ALD patients by increasing the peroxisomal ␤ -oxidation activity. SAHA also inhibited the induction of proinfl ammatory cytokines and inducible nitric oxide synthase (iNOS) in astrocytes silenced for Abcd1/2. Because SAHA crosses the blood-brain barrier (BBB), increases acetylation in brain ( 39 ), and reduces neuroinfl ammation, we propose that this drug may have therapeutic potential to ameliorate the X-ALD disease process.

Cell culture
Fibroblasts. Human skin fi broblasts derived from normal (control; GM03348), X-ALD (GM04932, GM04934, and GM04904), and AMN (GM07531) patients, and were obtained from the National Institute of General Medical Sciences (NIGMS) Human Genetic Mutant Cell Repository at the Coriell Institute for Medical Research (ccr.coriell.org/). The fi broblasts were cultured in DMEM containing 10% FBS and antibiotic/antimicotic.
Astrocytes. C57BL6 mice from Jackson Laboratory (Bar Harbor, ME) were maintained at the Medical University of At present, there is no satisfactory therapy for X-ALD. Reduction of VLCFA may be an ideal approach, but then none of the reported agents signifi cantly decrease the levels of VLCFA in human brain. Other approaches, such as hematopoietic stem cell transplantation and lentiviral-vector-associated gene therapy, have shown clinical benefi ts ( 20 ). They, however, are effective only in patients with the earliest stage of cerebral adrenoleukodystrophy (cALD). Thus, fi nding a treatment for X-ALD remains a challenge. Recent studies on neuroinfl ammation using histone deacetylase (HDAC) inhibitors show potential to ameliorate metabolic defects in neurological disorders, including ( 21,22 ) compromised gene metabolic disorders, by the activation of the overlapping function of the redundant gene. Pharmacological therapy for X-ALD remains attractive because of its potential to be effective pre-as well as postnatally. Various lines of evidence suggest that pharmacological induction of ABCD2 and/or ABCD3 may represent a therapeutic strategy for X-ALD. ABCD2, the closest homolog of ABCD1 ( 23 ), and ABCD3 ( 24 ) can compensate for ␤ -oxidation defects in X-ALD fi broblasts when overexpressed ( 25 ). Furthermore, the biochemical abnormalities found in Abcd1-knockout mice can be restored by overexpression of Abcd2 ( 26 ).
A key role of protein lysine acetylation in metabolic regulation was recently shown, in which virtually every enzyme of the FA oxidation pathways was found to be acetylated ( 27 ). HDAC-inhibitory activity ( 28 ) of 4-phenyl butyrate (4-PBA) was associated with induction of ABCD2 expression in Abcd1knockout mice. This induction resulted in reduced levels of VLCFA in brain and adrenal glands ( 29 ). Surprisingly, no ABCD2 induction or VLCFA correction was observed in adrenomyeloneuropathy (AMN) patients ( 30 ) with 4-PBA therapy. Valproic acid, a nonselective HDAC inhibitor, was recently shown to induce ABCD2 expression and correction of oxidative damage in a mouse X-ALD model and in peripheral blood mononuclear cells from X-ALD patients ( 31 ). The levels of C26:0 however, remained unchanged ( 31 ). Previous studies from our laboratory have shown that lovastatin, an inhibitor of HMG-CoA reductase and sodium phenyl acetate NAPA , can enhance VLCFA ␤ -oxidation and reduce VLCFA levels in human skin fi broblasts ( 32 ) and lymphoblasts ( 33 ) from X-ALD patients. Lovastatin also lowered the plasma levels of VLCFAs in X-ALD patients ( 34 ) and decreased the production of nitric oxide in X-ALD lymphoblasts ( 33 ). A number of other compounds, including 4-PBA ( 29 ), fenofibrate ( 25 ), and testosterone metabolites ( 35 ), have been shown to have the ability to lower VLCFA levels in X-ALD fibrobroblasts. However, none of the compounds to date have shown the ability to halt neurodegenerative progression.
Perturbation in acetylation homeostasis is emerging as a central event in the pathogenesis of neurodegeneration ( 36 ). Hence, recent studies have indicated that HDAC inhibitors might prove useful in treatment of such neurodegenerative disorders as Huntington's Disease (37)(38)(39), spinal muscular atrophy ( 40 ), amyotrophic lateral sclerosis (41)(42)(43), and experimental autoimmune encephalomyelitis ( 44 ). Amelioration of neurodegenerative conditions such as oxidative stress ( 45 ), and preservation of white A pool of three siRNAs for human ABCD2 was purchased from Santa Cruz Biotechnology (Santa Cruz, CA) and was used to transfect X-ALD fi broblasts at fi nal concentration of 2 M. Forty-eight hours after silencing for ABCD2, cells were treated with SAHA for 3 days with fresh addition of medium/SAHA every day.

FA ␤ -oxidation
The peroxisomal oxidation of FAs in control, X-ALD fibroblasts, and SAHA-treated X-ALD fi broblasts was determined in 6-well plates as described previously ( 19 ). ␤ -Oxidation of FAs to acetate (water-soluble product) was determined using [1-14 C]-labeled FAs as substrate (C 24:0 , lignoceric acid, or C 16:0 , palmitic acid) (American Radiolabeled Co. ; St. Louis, MO), as described previously ( 52 ). Cells grown in parallel in the same plate were used to determine the protein present in the assays. Experiments were performed in triplicate.

Lipid extraction and FA analysis
Total lipids were extracted from control and treated cells as described previously ( 53 ). The FAMEs were analyzed by GC (Shimadzu chromatograph GC-15A attached to a Shimadzu chromatopac C-R3A integrator) using a fused silica capillary column (25 M 007 series methyl silicone, 0.25 mm internal diameter, 0.25 m fi lm thickness) from Quadrex Corporation (Woodbridge, CT) in a gas South Carolina (MUSC) animal facility. All animal procedures were approved by the MUSC Animal Review Committee, and all animals received humane care in compliance with the MUSC experimental guidelines and the National Research Council's criteria for humane care ( Guide for Care and Use of Laboratory Animals ). Primary astrocyte-enriched cultures were prepared from the whole cortex of 1 day-old C57BL/6 mice as described previously ( 51 ). All cultured cells were maintained at 37°C in 5% CO 2 .
siRNA interference of Abcd1 and/or Abcd2 in mouse primary astrocytes The Silencer siRNA (Ambion; Austin, TX) was used for Abcd1 and Abcd2 silencing in primary mouse astrocytes as described previously ( 19 ). The siRNAs were mixed and diluted in OPTI-MEM1 medium to a fi nal concentration of 30 nM/well. A negative control with sequence similarity to no known human, mouse, or rat gene was included. Cells were maintained in DMEM with reduced serum (2%). Silencing was observed with Western blot and mRNA quantifi cation. Forty-eight hours after silencing for Abcd-1/2, cells were treated with SAHA and harvested 72 h later for protein and RNA analysis. For protein analysis of the transfected cells, three wells per plate were lysed and used for protein measurements and protein levels (Western blot). Cells were maintained for 6 days in DMEM with 2% FBS before harvesting for the analysis. gct gga gc-3 ′ ; 18S FP 5 ′ -gaa aac att ctt ggc aaa tgc ttt-3 ′ , RP 5 ′ -gccgct aga ggt gaa att ctt-3 ′ . Primers for human ABCD2, ABCD3, ELOVL1, and ELOVL3 were purchased from Qiagen. Thermal cycling conditions were as follows: activation of DNA polymerase at 95°C for 10 min, followed by 40 cycles of amplifi cation at 95°C for 30 s and 58.3°C (60°C for ABCD2, ABCD3, ELOVL1, and ELOVL3) for 30 s. The normalized expression of the target gene with respect to GAPDH or 18S RNA was computed for all samples using the Microsoft Excel data spreadsheet.

Determination of TNF-␣ in culture supernatants
Cells were silenced with Abcd1/2 siRNA or ScrRNA, and concentrations of TNF-␣ were measured in culture supernatants using high-sensitivity ELISA (R and D Systems).

Statistical analysis
Using the Student Newman-Keuls test and ANOVA, P values were determined for the respective experiments from three identical experiments using GraphPad software (GraphPad Software, Inc.; San Diego, CA). The criterion for statistical signifi cance was P < 0.05.

SAHA induces the mRNA expression of ABCD2 and ABCD3 in a concentration-and time-dependent manner in control normal human skin fi broblasts
It has been reported that ABCD3/PMP70 ( 54 ) or ABCD2/ALDRP ( 23 ), two close ABCD1/ALDP homologs, can compensate for the activity of ABCD1/ALDP. Because the overexpression of ABCD2/ALDRP resulted in the complete correction of VLCFA ␤ -oxidation in X-ALD fibroblasts ( 29 ) and the prevention of clinical symptoms in Abcd1 knockout mice ( 26 ), ABCD2/ALDR is an attractive candidate gene for pharmacological gene therapy. Therefore, we fi rst studied the effect of SAHA on ABCD2 and ABCD3 expression in normal control human skin fi broblasts. The fi broblasts were treated with SAHA with varying doses and durations and the gene expressions for ABCD2 and ABCD3 were analyzed by RT-PCR. SAHA signifi cantly induced the expression of both ABCD2 ( P < 0.001) and ABCD3 ( P < 0.001) in a dose-and time-dependent manner ( Fig. 1 ). Although ABCD2 expression was induced 7.2-fold ( Fig. 1A ), there was a 5.7-fold induction of ABCD3 expression ( Fig. 1B ) at the maximum dose of 5 M SAHA. SAHA treatment for 3 days was suffi cient to induce 7.2-and 5.7fold induction of ABCD2 ( P < 0.001) and ABCD3 ( P < 0.001) gene expressions, respectively ( Fig. 1C, D ). No signifi cant increase in the expression of ABCD2 or ABCD3 was observed at higher doses of SAHA (7.5-10 M). There was an insignifi cant increase in expression of ABCD2 and chromatograph GC-17A connected with a fl ame ionization detector from Shimadzu Corporation.

Western blot analysis
Then, 40 g of total cellular protein was resolved by electrophoresis on 4-20% polyacrylamide gels. After incubation with antiserum raised against mice ALDP, ALDRP, 5-LOX, p65, Na + K + ATPase, and iNOS, the membranes were then incubated with HRP-conjugated anti-rabbit or mouse IgG for 1 h. The membranes were detected by autoradiography using ECL-plus (Amersham Biosciences) after washing with TBST buffer.

SAHA induces ABCD2 and ABCD3 mRNA expression and protein levels in X-ALD human fi broblasts
To assess the therapeutic potential of SAHA for the treatment of X-ALD patients, we cultured human X-ALD fi broblasts in the presence of SAHA for 3 days and quantifi ed its effect on ABCD2 mRNA expression by RT-PCR. Treatment with SAHA resulted in a dose-dependent increase in ABCD2 mRNA expression in four different X-ALD cell lines ( Fig. 3 ). The highest dose of SAHA (5 M) increased the ABCD2 mRNA expression by 11.8- (Fig. 3A), 6.0- (Fig. 3B), 2.8- (Fig. 3C), and 3.88- (Fig. 3D) fold in 3 days in the respective cell lines. Much like ABCD2, the overexpression of ABCD3 has also been shown to correct the defect in peroxisomal ␤ -oxidation activity in ABCD1-defi cient cells from X-ALD patients ( 57 ). Similar to ABCD2, we observed a maximum of 10-, 3.92-, 2.77-, and 3.45-fold increase in ABCD3 mRNA expression in cell lines following treatment with 5 M SAHA for 3 days ( Fig. 4 ).
Western blots using antibodies against ALDRP and PMP70 were performed on carbonate membranes (integral membrane proteins) obtained from control normal and X-ALD ABCD3 beyond 3 days of treatment; therefore, all studies were carried out with up to 5 M dose and 72 h treatment.

Effect of SAHA on FA ␤ -oxidation in control normal human fi broblasts
Dysfunction of ALDP/ABCD1 in X-ALD patients results in decreased VLCFA ␤ -oxidation activity and accumulation of VLCFAs (C26:0 and C24:0). Overexpression of ABCD2 in fi broblasts from Abcd1-defi cient mice or from X-ALD patients is shown to restore peroxisomal ␤ -oxidation and to reduce the levels of VLCFA ( 25,29,55,56 ). Because SAHA upregulated the mRNA expression of ABCD2 and ABCD3 in control human fi broblasts, we examined the effect of this induction on ␤ -oxidation of lignoceric acid (C24:0) and palmitic acid (C16:0) in control human fi broblasts. SAHA treatment increased the lignoceric acid oxidation in a dose-and time-dependent manner ( Fig. 2 ). The highest dose of SAHA (5 M) was found to stimulate lignoceric acid ␤ -oxidation by approximately 45% ( Fig.  2A ), and signifi cantly increased (maximum 55% at 5 M dose) the palmitic acid ␤ -oxidation activity ( Fig. 2A ). The maximum lignoceric acid and palmitic acid ␤ -oxidation activity was reached at day 3, with no further signifi cant increase in oxidation ( Fig. 2B ). Increase of both lignoceric and palmitic acids ␤ -oxidation activity in SAHA-treated  ( Fig. 6B ). Palmitic acid ␤ -oxidation activity was also increased in a dose-dependent manner (43.5% to 51%) following SAHA (0.5-5 M) treatment for 3 days ( Fig. 6A ). The induction of lignoceric acid and palmitic acid ␤ -oxidation by SAHA suggests that it affects both mito chondrial and peroxisomal ␤ -oxidation.
Overexpression or transfection of Abcd2 can compensate for the loss of Abcd1 and correct the VLCFA levels ( 25 ). To establish the relationship of SAHA-mediated upregulation of ABCD2 expression to increased ␤ -oxidation activity, we transiently transfected the human skin fi broblasts with ABCD2 siRNA and measured the lignoceric and palmitic acid ␤ -oxidation activities after SAHA (5 M) treatment for 3 days ( Fig. 7 ). Transient silencing with ABCD2 siRNA did not result in any further inhibition of lignoceric ( Fig. 7B ) or palmitic acid ( Fig. 7A ) ␤ -oxidation activity. However, no increase in lignoceric acid ␤ -oxidation activity was seen when the cultures were treated with SAHA. There was still a statistically signifi cant increase ( P < 0.05) in palmitic acid ␤ -oxidation activity.

SAHA reduces the levels of very long chain FAs in X-ALD human fi broblasts
Levels of total hexacosanoic acid (C26:0) and the ratios of C26:0/C22:0 or C24:0/C22:0 are widely used for fi broblasts and SAHA-treated control normal and X-ALD fibroblasts. Screening for ABCD2 ( Fig. 5A, C ) and ABCD3 ( Fig. 5B, D ) showed an appreciable dose-dependent increase in signal in SAHA-treated fi broblasts. Screening for Na + /K+-ATPase (plasma membrane protein), used as indicator of protein loading for membrane fractions, did not show an appreciable difference between the same cell samples.

SAHA increases peroxisomal ␤ -oxidation of VLCFA in X-ALD human skin fi broblasts
Because peroxisomal VLCFA ␤ -oxidation activity is determined by ALDP expression ( 58 ), ␤ -oxidation of VLCFA is impaired in X-ALD patients. ALDRP has been shown to compensate for the loss of ALDP in vitro and in vivo, leading to enhanced ␤ -oxidation activity. To assess the therapeutic potential of SAHA-mediated ALDRP stimulation for X-ALD, we examined the effect of SAHA treatment on peroxisomal and mitochondrial ␤ -oxidation activity in cultured X-ALD fi broblasts using radiolabeled FAs. Lignoceric acid ␤ -oxidation activity was signifi cantly low in all the X-ALD cell lines (approximately 47% compared with control human fi broblasts) due to the deletion/mutation of ABCD1 ( 7 ). Treatment of X-ALD fi broblasts with 0.5-5 M SAHA for 3 days showed a dose-dependent increase (40% to 55% in four different cell lines) in lignoceric acid Seven types of ELOVLs have been identifi ed in mammals, of which two (ELOVL1 and ELOVL3) have chain length specifi city toward VLCFA ( 61,62 ). There was no significant difference in the levels of ELOVL1 and ELOVL3 in control and X-ALD fi broblasts ( Fig. 9A, B ) as reported previously ( 63 ), indicating that availability of substrate is the limiting factor for their elongation. Knockdown of ELOVL1 in human X-ALD fi broblasts signifi cantly lowers the levels of C26:0 ( 63 ). SAHA treatment of control and X-ALD human fi broblasts signifi cantly decreased [81.55% ( P < 0.001) for control and 63.35% ( P < 0.001) for X-ALD fi broblasts at a 5 M dose] the ELOVL1 expression. However, the expression of ELOVL3 was signifi cantly increased after SAHA treatment. The results highlight the dual effect of SAHA on induction of ␤ -oxidation activity and inhibition of VL-CFA elongation, ultimately resulting in a net effect of lowering the levels of both saturated and unsaturated VLCFAs.

SAHA inhibits proinfl ammatory cytokines and reduces oxidative stress in Abcd1/2-silenced mouse astrocytes
Accumulation of VLCFA in X-ALD patients leads to secondary injury of infl ammatory demyelination due to a marked increase in the activation of microglia and astrocytes, and thus the production of proinfl ammatory cytokines (TNF-␣ and IL-1 ␤ ) ( 64,65 ). Expression of TNF-␣ and iNOS was higher in infl ammatory areas compared with normal areas of cALD brain ( 66 ). Therapeutic reduction of the levels of VLCFA correlates with decreased expression of proinfl ammatory cytokines ( 19 ). Our laboratory has recently reported the accumulation of VLCFA and increased production of ROS and proinfl ammatory cytokines (iNOS, TNF-␣ , and IL-1 ␤ ) in Abcd1/2silenced mouse primary astrocytes ( 19 ). Expression of iNOS and TNF-␣ was signifi cantly increased in Abcd1/2silenced mouse astrocytes. Therefore, we evaluated the effect of SAHA on ROS and proinfl ammatory cytokines in Abcd1/2-silenced mouse primary astrocytes ( Fig. 10 ). SAHA siginifi cantly ( P < 0.001) inhibited ROS generation in Abcd1/2 silenced mouse astrocytes ( Fig. 10A ). Treatment with SAHA also downregulated ( P < 0.001) the expression of iNOS and TNF-␣ (both at mRNA and protein levels) in Abcd1/2-silenced mouse astrocytes ( Fig. 10B-E ).
A growing body of evidence suggests the role of certain immunomodulatory leukotrienes ( 67 ) in the signaling cascade of infl ammatory gene expression ( 18 ). 5-LOX expression was increased in Abcd1/2-silenced mouse astrocytes, which was signifi cantly decreased by treatment with SAHA ( Fig. 10D ). Mechanisms of Abcd1/2 silencing-induced proinfl ammatory response involve activation of NF-B ( 19 ). Western blots showed increased levels of p65 protein in Abcd1/2-silenced mouse primary astrocytes ( Fig. 10D ). The level of p65 in the nucleus of Abcd1/2-silenced astrocytes was significantly reduced by SAHA. Because SAHA crosses the BBB ( 39 ), downregulates infl ammatory mediators, and corrects abnormality of VLCFA, it may have the potential for X-ALD therapy. the diagnosis of X-ALD and other peroxisomal disorders. Although the precise function of ALDP in the metabolism of VLCFA is not known at the present time, accumulation of VLCFA in X-ALD cells with loss or mutations in ALDP and their normalization following transfection of cDNA for ALDP indicate a role of ALDP in the metabolism of VLCFA. Also, overexpression of ABCD2 leads to reduced VLCFA accumulation in cultured fi broblasts from X-ALD patients ( 29 ). Therefore, we determined whether induction of ABCD2 mRNA expression and protein levels with SAHA correlates with the reduction of VLCFA levels. C26:0 levels were signifi cantly higher (9fold, P < 0.001) in human X-ALD fi broblasts as compared with control human fi broblasts ( Fig. 8A ). Treatment of X-ALD fi broblasts with SAHA (5 M) for 3 days significantly reduced (37%, P < 0.002) the levels of VLCFA (C26:0).
In addition to saturated VLCFA, X-ALD fi broblasts have signifi cantly higher levels of monounsaturated FAs (C26:1), indicating that ␤ -oxidation of C26:1 is also reduced in X-ALD fi broblasts and the plasma and brain of X-ALD patients ( 59,60 ). We detected signifi cantly higher (1.3-fold, P < 0.02) levels of C26:1 in X-ALD fi broblasts compared with control human fi broblasts ( Fig. 8B ). Treatment of human X-ALD fi broblasts with SAHA also resulted in a signifi cant decrease of 47.6% ( P < 0.001) and 71.4% ( P < 0.001) in the levels of C26:1 in a dose-and time-dependent manner.

SAHA decreases ELOVL1 expression in X-ALD fi broblasts
Elongases are responsible for the initial step in the elongation of specifi c acyl-CoA FAs with different chain lengths.  5. SAHA upregulated the levels of ABCD2 and ABCD3 proteins in control normal and X-ALD human skin fi broblasts. Human control fi broblasts (GM07576) and fi broblasts derived from an X-ALD patient (GM04932) were cultured and used to analyze protein levels of the peroxisomal integral membrane proteins transporters ALDRP (ABCD2) (A, C) and PMP70 (ABCD3) (B, D). Protein levels of the peroxisomal transporters were analyzed by Western blot in membrane fractions obtained by carbonate treatment (membrane preparation containing integral membrane proteins), as indicated in the MATERIALS AND METHODS section. Na + / K+-ATPase (plasma membrane protein) was used as indicator of protein loading for plasma membrane fractions (E, F). ABCD1 mutation/deletion and/or the metabolites on expression of other transporters in X-ALD fi broblasts. Earlier studies from our laboratory ( 7,(71)(72)(73)(74) and others ( 75 ) have reported that VLCFAs are preferentially ␤ -oxidized in peroxisomes, whereas oxidation of palmitic acid occurs largely in mitochondria. Treatment with SAHA increased both the lignoceric acid (C24:0) and palmitic acid (C16:0) ␤ -oxidation activities in X-ALD fi broblasts. However, SAHA treatment of human X-ALD fi broblasts silenced for ABCD2 led to increased C16:0 ␤ -oxidation but had no effect on C24:0 ␤ -oxidation, indicating that the restoration of the peroxisomal VLCFA ␤ -oxidation activity in SAHAtreated X-ALD fi broblasts resulted from upregulation of ABCD2 expression. Thus, SAHA-mediated induction of ABCD2 expression and reduction of C26:0 levels should be able to compensate for the lack of ABCD1 function in X-ALD patients.

DISCUSSION
Pharmacological therapy for a genetic disease is aimed at upregulating redundant genes to compensate for biochemical defects. Because the accumulation of VLCFA triggers an infl ammatory response associated with progressive demyelination in X-ALD ( 19 ), a decrease in the VLCFA content may be benefi cial for the patients. However, the other ALDP-related transporter gene, ABCD2, is not mutated in X-ALD patients, and the X-ALD phenotype is independent of the ABCD2 genotype ( 68 ). Our recent observation of an inverse correlation between the VLCFA content and the expression of Abcd2/ALDRP and a greater accumulation of VLCFAs following silencing for both Abcd1 and Abcd2 in mouse primary astrocyte cultures ( 19 ), suggests that Abcd1 and Abcd2 have overlapping functions in the metabolism of VLCFA and related compounds. Accordingly, Abcd1 and Abcd2 doubleknockout mice accumulate higher levels of VLCFA than Abcd1 knockout mice ( 69 ), and elevated expression of ALDRP increased the ␤ -oxidation activity and lowered the C26:0 and C24:0 levels ( 70 ). Treatment with SAHA increased the levels of ABCD2 and ABCD3 in control and X-ALD fi broblasts. Interestingly, the degree of increase was higher in control fi broblasts than in the X-ALD fi broblasts, possibly due to the dominant-negative effect of ELOVL3) have chain length specifi city toward VLCFA ( 31,63 ) and would be the most attractive candidate elongases to play a role in enhanced FA chain elongation in X-ALD. Treatment with SAHA decreased the expression of ELOVL1 in control as well as in X-ALD human skin fi broblasts. In contrast to ELOVL1, ELOVL3 expression was signifi cantly increased ( 31 ). A recent study reporting reduced elongation of C22:0 to C26:0 and reduced levels of C26:0 in X-ALD fi broblasts following silencing of ELOVL1 indicates a role for ELOVL1 in VLCFA elongation ( 63 ). Therefore, the effect of SAHA on lowering the C26:0 and C26:1 levels in X-ALD fi broblasts is a net effect of decreased chain elongation (decreased ELOVL1 expression), increased ABCD2 expression, and increased ␤ -oxidation activity ( Figs. 6B and 8A ).
The VLCFA-induced secondary infl ammatory response is believed to participate in the neurodegeneration in X-ALD ( 19 ). Therefore, for effective therapy, pharmacological agents need to cross the BBB to reduce the accumulated VLCFA in the brain and prevent the progression of neurological symptoms in X-ALD. SAHA has been reported to cross the BBB in different models ( 39 ), making it an attractive candidate for X-ALD therapy. Previous studies from our laboratory have documented that lowering of VLCFA leads to inhibition of the infl ammatory response ( 19 ). Abcd1/2 silencing of wild-type mouse primary astrocytes induced a proinfl ammatory response (iNOS and ABCD2 expression, signifi cantly reduced the C26:1 levels in X-ALD fi broblasts, suggesting a causal relationship between ALDRP expression and levels of monounsaturated VLCFAs. Although the differential substrate specifi cities of individual ABCD transporters remain speculative, recent studies report overlapping substrate specifi cities for ALDP and ALDRP ( 70 ), especially for saturated and monounsaturated VLCFAs ( 31,70,76 ). A signifi cant decrease in C26:1 level was observed with SAHA (1 M), at which there was a relatively small increase (2-fold) in ABCD2 expression, suggesting that higher affi nity of C26:1 for ABCD2 may be suffi cient for decrease in C26:1 with a relatively small increase in ABCD2, but not suffi cient for the transport of saturated VLCFA (C26:0). Second, increased expression of ABCD3 in 1 M SAHA-treated cells may facilitate the transport and catabolism of monounsaturated VLCFA similar to one observed for PUFA ( 77 ).
In addition to the abnormality in peroxisomal VLCFA ␤ -oxidation ( 74 ), the observed enhanced FA chain elongation activity ( 8 ) may also contribute to increased accumulation of VLCFA in X-ALD. To date, seven FA elongases (ELOVL1-7) have been identifi ed and characterized in mammals ( 78,79 ). Of these, two elongases (ELOVL1 and  cultures ( 80 ) and in mouse brain ( 81 ). The observed inhibition of NF-B in the brain ( 81 ) and protection against ischemic injury ( 46 ) in animals treated with SAHA document that SAHA is a good candidate drug for neurological disorders.
Given the recent observation of the key role of protein acetylation in FA metabolism ( 27 ), as well as the significance of altered gene expression in disease pathogenesis, a great deal of effort has been directed to the development of chemical inhibitors of HDACs as therapeutic drugs ( 82 ). Studies with phenylacetate (PA) from our laboratory ( 83 ) and with PBA from other laboratories ( 29 ) have shown that these compounds (PBA and PA) have very low potency, insofar as millimolar amounts of these proinfl ammatory cytokines) ( 19 ), which was mediated by the activation of NF-B and 5-LOX and increased levels of ROS ( 13 ). SAHA treatment of Abcd1/2-silenced mouse astrocytes reduced the expressions of iNOS and TNF-␣ (mRNA and protein), suppressed NF-B activation, and reduced ROS production. The observed increased activation of lipoxidative enzymes in X-ALD brain ( 18 ) and in Abcd1/2-silenced astrocytes ( 19 ) suggests the role of these proinfl ammatory mediators in the pathobiology of X-ALD. SAHA treatment also reduced the activation of 5-LOX in Abcd1/2-silenced mouse astrocytes. Our observation of inhibition of NF-B activation is consistent with studies from other laboratories reporting inhibition of proinfl ammatory mediators and NF-B activity by SAHA in glial cell Fig. 10. ROS and expression of proinfl ammatory cytokines (iNOS and TNF-␣ ) is decreased by SAHA in Abcd1/2-silenced mouse primary astrocytes. A: ROS generation in primary astrocytes after 4 days of silencing or SAHA-treated astrocytes. iNOS (B) and TNF-␣ (C) expression was examined using the iCycler iQ real-time PCR detection system as discussed in MATERIALS AND METHODS. The target gene expression was normalized to GAPDH expression, and the result is presented as mean normalized expression. For the detection of iNOS, 5-LOX, and p65 protein expression by immunoblot (D) in response to Abcd1/2 silencing or SAHA treatment, cell lysate from astrocytes was prepared as described in MATERIALS AND METHODS. TNF-␣ (E) was measured by ELISA in the supernatant of primary astrocytes after 4 days of silencing or SAHA treatment. * P < 0.001 Abcd1/2-silenced astrocytes compared with control or scramble-silenced astrocytes; ** P < 0.001 SAHA-treated astrocytes compared with ABCD1/2-silenced astrocytes; @@ P < 0.02 SAHA (2.5 M)-treated astrocytes compared with ABCD1/2-silenced astrocytes; NS, nonsignifi cant. nancies ( 97 ). Furthermore, SAHA was recently approved by the Food and Drug Administration for treatment of cutaneous T-cell lymphoma. Pediatric Phase 1 trials have shown that SAHA (at 230 mg/m 2 /day) is well-tolerated in children with solid tumors ( 98 ). The in vivo dose of SAHA (50 mg/kg) showing protection in animal models of infl ammation translates to 6 mg/kg (total dose of 120 mg/day or 150 mg/m 2 for a 20 kg human child) , which is much lower than the maximum dose administered safely in pediatric patients ( 98 ). In summary, ( Fig. 11 ), our observations describe upregulation of peroxisomal ␤ -oxidation activity and inhibition of chain elongation, and thus correction of VLCFA levels in X-ALD cells, as well as inhibition of infl ammatory response in ABCD1/2-silenced astrocytes by SAHA, indicating that SAHA is a potential candidate drug for correction of the metabolic defect as well as the secondary neuroinfl ammatory disease in X-ALD.
compounds were required to observe their effects on ␤ -oxidation and levels of VLCFA in X-ALD cells. Second, the effects of PBA were not long-lasting; the initially decreased VLCFA levels later returned to pretreatment levels in longterm studies ( 84 ). In a recent report, published while this manuscript was under review, a millimolar concentration of PBA was required for lowering the VLCFA levels in cultured oligodendrocytes differentiated from pluripotent stem cells of X-ALD patients ( 85 ). Furthermore, PBA/PA are nonselective inhibitors of HDAC and have been reported to inhibit the mevalonate pathway ( 86 ). Comparatively, SAHA is a potent and specifi c HDAC inhibitor (the only known inhibitor of HDAC11) and is effective at micromolar concentrations (approximately 1000-fold difference). A recent report documents the therapeutic potential of SAHA in mouse models of progressive neurodegenerative frontotemporal dementia ( 87 ). In the neuromuscular disease SMN, SAHA had a signifi cantly better outcome than butyrate or valproic acid (VPA) ( 88 ). Additionally, SAHA has been shown to provide a favorable outcome in a variety of animal models of infl ammatory disease conditions including LPS-induced endotoxemia ( 89 ), lupus ( 90 ), graft-versus-host disease ( 91 ), ischemia ( 46,81 ), TLR activation ( 92 ), septic shock ( 93 ), and infl ammatory hyperalgesia ( 94 ). Moreover, SAHA is presently in numerous clinical trials ( 95 ); more than 100 clinical trials with SAHA are in progress at the National Institutes of Health ( http:// www.clinicaltrials.gov ). An oral preparation of SAHA (200-600 mg) was safely administered chronically (up to 37 months) and has shown good bioavailability, linear pharmacokinetics, and inhibition of HDAC activity in Phase 1 trials ( 96 ) and in patients with hematologic malig- Fig. 11. A schematic representation of the effect of SAHA on stimulation of ␤ -oxidation activity and lowering of the VLCFA load in human X-ALD skin fi broblasts. Loss of ABCD1 is followed by a sequence of events (red arrows) resulting in accumulation of VLCFA and subsequent induction of ROS and neuroinfl ammation. Treatment with SAHA (green arrows) induces ABCD2/ABCD3 expression while lowering the expression of ELOVL1. The combined effect of these two events results in overall increased ␤ -oxidation activity and reduction of VLCFA levels and subsequent attenuation of oxidative stress and neuroinfl ammation. (Please see text for details).