| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
* Department of Clinical Biochemistry, University of Cambridge, Addenbrooke’s Hospital, Cambridge, UK;
Ingenium Pharmaceuticals AG, Martinsried, Germany;
Diabetes Research Unit, Boston Medical Center, Boston, Massachusetts, USA;
Metabolic Research Laboratory, Oxford Centre for Diabetes, Endocrinology and Metabolism, Nuffield Department of Clinical Medicine, University of Oxford, Churchill Hospital, Oxford, UK;
|| Department of Biochemistry, University of Cambridge, Cambridge, UK; and
Cellular Regulation, MRC Dunn Human Nutrition Unit, Cambridge, UK
2 Correspondence: Department of Clinical Biochemistry, University of Cambridge, Addenbrooke’s Hospital, Hills Road, Box 232, Cambridge CB2 2QR, UK. E-mail: ajv22{at}cam.ac.uk
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Key Words: fatty acid synthase • malonyl-CoA • metabolomics • metabolic pathways
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, 2)
. NASH is a common feature of several metabolic disorders and a frequent side effect of therapeutic drugs (3
, 4)
. Conditions associated with NASH include the metabolic syndrome, obesity, type 2 diabetes (5
, 6)
, rapid weight loss in obese subjects, industrial toxins, gastrointestinal surgery, and use of a number of clinical drugs (2)
.
The pathogenic mechanisms of NASH are not completely known. The "two-hit hypothesis" identifies an initial phase of progressive deposition of fat in the liver, followed by a second phase characterized by increased oxidative stress due to additional insults (7)
. These two phases have a degree of overlap, making it extremely difficult to differentiate specific primary events from nonspecific compensatory mechanisms. Most human studies using samples obtained from patients with established NASH do not allow identification of the initial events that lead to the development of NASH. Long-term studies always show confounding factors related to reactive oxygen species (ROS) production, inflammatory changes, and fibrosis (8)
. Furthermore, production of ROS and lipid peroxidation initiates a vicious cycle leading to further impairment of the mitochondrial respiratory chain and generation of more ROS (9)
. Thus studies at this stage may not be sensitive enough to identify early pathogenic events that can provide clues about the nature of the initial insult.
In this paper, we investigate the "initial hit," or the specific processes that initiate fat deposition in liver (7)
. The free fatty acids (FFAs) that accumulate in the liver can be from different sources. They can be taken up from the plasma FFAs after being released by adipose tissue. Alternatively, fatty acids can accumulate in the liver from the hydrolysis of chylomicron remnants originating from the intestine or they can be directly synthesized de novo from carbohydrates (10)
. Accumulating evidence suggests that impaired fatty acid oxidation (FAO) may also be a cause of fatty acid accumulation in the liver (11)
. Fat accumulation in liver can result from impaired secretion of VLDL particles (12)
. Secretion of hepatic triglycerides requires correct assembling of phospholipids and apoproteins (e.g., apo B); an inappropriate supply of these elements may impair triglyceride secretion.
One such drug known to trigger NASH is the anti-tumor drug tamoxifen (TMX) (13
14
15)
. TMX is a commonly prescribed compound for the treatment of and prevention of reoccurrence of estrogen receptor (ER) -positive breast cancers (16)
. TMX was derived as a pure antioestrogenic molecule that could act against the apparent pro-tumorigenic effects of estrogen (17
, 18)
, especially important for some forms of breast cancer that express high levels of ERs. However, TMX is not a pure antioestrogen, but displays partial estrogen-like activity in tissues such as rat uterus (19)
. Thus TMX belongs to the selective estrogenic receptor modulator class of drugs. In addition, it is thought that TMX has ER-independent activity, including the induction of calcium signaling (20)
via protein kinase C activity and by influencing the properties of cellular membranes.
One of the main limitations of most studies aiming to elucidate the pathogenic mechanisms of NASH is the complication posed by the mixture of primary and secondary events. In this report we had two objectives: to 1) elucidate the initial mechanisms leading to TMX fatty liver and 2) evaluate whether our experimental strategy, a model of early stages of disease together with a combination of metabolic and transcript profiling together with appropriate bioinformatics analysis, allowed us to identify primary events in a complex process such as the development of fatty liver. Using this approach we have identified specific inhibition of fatty acid synthase (FAS) gene expression and activity as a primary target of TMX-induced fatty liver.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Blood biochemistry analysis
Serum leptin levels were measured using a mouse leptin ELISA kit (Crystal Chem Inc., Downers Grove, IL, USA), which has 90% reactivity for rat leptin (assay range 200–12,800 pg/mL, intra-assay precision CV=5.4%, interassay precision CV=6.9%). Serum insulin levels were measured using a rat insulin ELISA kit (assay range 156–10,000 pg/mL, intra-assay precision CV=3.5%, interassay precision CV=6.3%) (Crystal Chem Inc.). Kits for the measurement of glucose, cholesterol, and triglycerides were obtained from Roche Diagnostics GmbH (Mannheim, Germany). Free fatty acids were measured using the NEFA-C colorimetric kit (Wako Chemicals GmbH, Neuss, Germany).
Histological analysis
Liver sections (n=6 for each treatment group) were fixed in formalin, embedded in paraffin, and stained with hematoxylin and eosin. Steatosis was graded as none, minimal (0–10%), mild (10–33%), moderate (33–66%), or severe (>66%) based on percentage of hepatocytes showing fat accumulation in the sections (10 fields per 20x slide, concentrated around perivenular regions).
Metabolic profiles derived via 1H-NMR spectroscopy
Solution state 1H-NMR spectroscopy was performed on hepatic tissue extracts at 400.1 MHz using a Varian ANOVA spectrometer interfaced with a 9.6 Tesla superconducting magnet and high-resolution inverse geometry 1H-NMR probe. Extracts of hepatic tissue (n=4 for each treatment examined in triplicate) were prepared using a perchloric acid (6%) extraction procedure and reconstituted in D2O containing 4 mM tris-ilylpropionic acid. Solvent suppressed spectra were acquired into 32 k data points, averaging > 128 acquisitions. Spectral assignments were made with reference to published literature (21)
.
Intact liver tissue (
10 mg; n=4 for each treatment group) was examined using a high resolution magic angle spinning (HRMAS) probe interfaced with a Bruker AVANCE spectrometer and 9.6 Tesla superconducting magnet. HRMAS spectra were acquired by spinning the samples at 4000 Hz and spectra were averaged over 128 acquisitions. Two pulse sequences were used: 1) a NOESYPR1D pulse sequence involving solvent suppression using a pulse sequence based on the start of the NOESY pulse sequence and 2) a Carr Purcell Meiboom and Gill (CPMG) pulse sequence that is T2 edited to reduce the contribution of slowly moving/rotating metabolites, which have characteristic short T2 relaxation rates (such as lipids).
Spectra were processed using XWINNMR software, version 3.1 (Bruker Gmbh, Ettlingen, Germany). After multiplication with a pre-exponential factor of 1 Hz and Fourier transformation, spectra were integrated across 0.04 ppm spectral regions between 0.4 and 4.2 ppm using a program written in Matlab (Dr. Tim Ebbels, UCL, UK). Output vectors representing each spectrum were normalized across the integral regions so that each individual 0.04 ppm region was represented as a ratio to the entire integral region investigated.
Pattern recognition and data processing of 1H-NMR spectra
The 1H-NMR spectral datasets were imported into the SIMCA package (Umetrics, Umeå, Sweden) and preprocessed using three Pareto scaling ((1/sk)1/2, where sk is the standard deviation of the variable). Each dataset was examined by partial least squares discriminate analysis (PLS-DA). As this is a supervised method, correlations were tested by sequentially leaving out every seventh sample and predicting its class membership. This routine was used to determine a goodness of fit measure and examine whether a correlation was significantly better than chance (Q2>0.05).
Transcriptomic and metabolic correlations
Differentially expressed transcripts were correlated with changes in metabolites using combined datasets that were significant in a 2-way comparison as determined using a Student’s t test (P<0.05). This dataset was then examined by principal components analysis (PCA) to identify clusters.
RNA extraction protocols
RNA extractions from rat primary hepatocytes were performed using the RNeasy Mini system (Qiagen, Crawley, UK), following the manufacturer’s instructions. Rat liver RNA was extracted using the STAT-60 method (Tel-Test "B," Friendswood, TX, USA), using two rounds of purification with STAT-60/chloroform prior to isopropanol precipitation. RNA concentration and integrity were determined using a spectrophotometer at 260 nm and an ethidium bromide-stained agarose gel, respectively.
Synthesis of probes and reference RNA and solution hybridization RNase protection assay (RPA)
For specific probe templates, cDNA probes were generated from liver cDNA using PCR. Primer sequences are given in Supplementary Table 1. The PCR products were ligated into the pGEM-T Easy vector (Promega Corp., Cambridge, UK) and used to transform JM109 Supercompetent cells (Promega Corp.). Orientation of the ligation was verified by restriction digest and direct sequencing. To assess RNA loading, a ready-made template for rat cyclophilin (Cyc) was used that protects a region of 109 bases (Ambion). The template for the antisense probe was prepared by linearizing the plasmid containing the probe insert with the restriction endonuclease SpeI. The linearized vectors were then purified using a Qiaquick PCR purification column (Qiagen). Antisense probes and reference probes were transcribed from the linearized template DNA using bacteriophage T7 polymerases (Stratagene, La Jolla, CA, USA) using published protocols (22
, 23)
. Protected bands were visualized by autoradiography and quantified by PhosphorImager analysis using a Molecular Dynamics Storm 840 PhosphorImager and ImageQuant 5.2 software (Amersham Pharmacia Biotech AB, Little Chalfont, UK).
Gene expression profiling using liver from treated rats
RNA was extracted from tissue using the STAT-60 method as above. RNA was then purified using the RNA clean-up protocol from the RNeasy Mini Kit (Qiagen). RNA was quantified spectroscopically at 260 nm using a GeneQuant Nucleotide calculator (Amersham Pharmacia Biotech AB) and checked for integrity on a 1% TBE gel using ethidium bromide staining. RNA from pair-fed and TMX-treated rats liver was pooled into three mutually exclusive pools (4 RNA samples/pool).
Microarray preparation and analysis
Full details of the protocols followed can be found in the supplementary data. The data was analyzed using pathway enrichment was assessed using variations on the methods of Mootha et al. and Patti et al. (24
, 25)
. The z-score analysis was based on that described by Doniger et al. (26)
. The pathways used were obtained from the GenMAPP (27)
and GeneOntology (28)
databases. Many of these pathways contained a limited number of annotations, so we constructed our own pathways for fatty acid biosynthesis, FAO, cholesterol biosynthesis, and bile acid synthesis. The fatty acid biosynthesis pathway includes all genes from pyruvate dehydrogenase to the delta-5 and -6 desaturases as well as the glycerol-3-phosphate and diacylglycerol acyltransferases. This pathway was split into two, yielding one pathway containing all genes distal to FAS and the other all the genes proximal to FAS.
Primary rat hepatocyte culture and treatment
Rat primary hepatocytes were obtained and cultured as in Bartlett et al. (29)
. Further details can be found in the supplementary data. Either 1 µM TMX or vehicle control (ethanol, final concentration 0.1%) was added to the cells for a total of 48 h. After 24 h, the medium was replaced with fresh medium containing the supplements.
FAS activity assay
Liver samples were homogenized in ice-cold, sucrose-based buffer (0.25 M sucrose, 1 mM DTT, 1 mM EDTA, Complete Mini protease inhibitors (1 tablet/10 mL). The homogenate was then centrifuged at 105,000 x g at 0°C for 1 h. The infranatant was quantified for protein content as above. In a 96-well plate, 250 µg of protein was prepared in a volume of 100 µL homogenization buffer, added to 200 µL of NADPH/acetyl-CoA solution (33 mM K2HPO4, 67 mM KH2PO4, 200 µM NADPH, 100 µM acetyl-CoA), and incubated for 10 min at 37°C. To start the reaction, 30 µL of malonyl-CoA solution (33 mM K2HPO4, 67 mM KH2PO4, 600 µM malonyl-CoA) was added. Absorbance was measured at 340 nM over 20 min and the gradient for 
A340nm was calculated. The conversion to NADPH (in M) = A340nm/(6.22*103)*0.857, where 0.857 is the path length of the reaction in cm.
Malonyl-CoA level determination
Malonyl-CoA concentration was measured by the radioisotopic method in neutralized perchloric acid filtrates by the method from McGarry et al. (30)
with modifications as in ref 31
.
Statistical analysis
Food intake, serum determination and RPA data were analyzed using a parametric Student t test. Body weight, FAS activity, and malonyl-CoA concentration assays were analyzed with 1-way ANOVA using Bonferroni ad hoc post-test. The significance level was set at P <0.05.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
, we included a pair-fed group to allow distinction of direct TMX effects from those due to reduced food intake and body weight. We confirmed that cumulative 5 day food intake was reduced in the TMX-treated animals (Fig. 1
A: ad libitum vs. TMX; 117.8±2.0 vs. 91.1±2.5 g/rat, P<0.0001). Body weight of the TMX and pair-fed rats were reduced compared with ad libitum rats (Fig. 1B
: % body weight change; ad libitum vs. PF vs. TMX; 4.68±0.31 vs. –0.14±0.32 vs. 0.03±0.37, P<0.0001 for ad libitum vs. PF and ad libitum vs. TMX).
|
Five-day TMX-treated rat livers displayed moderate lipid accumulation
From the morphological analysis (Fig. 2
), 83% of the TMX-treated animals showed moderate, predominantly perivenular microvesicular steatosis (as small lipid droplets) (Fig. 2
, bottom panels). There was no evidence of steatohepatitis. In comparison, 100% of livers from the ad libitum (top panels) showed no fat deposition and 83% of the pair-fed group (middle panels) showed minimal steatosis. Thus, this model demonstrates predominant accumulation of lipids without microscopic inflammatory changes in the livers of TMX-treated rats.
|
Serum cholesterol levels are reduced in TMX-treated rats
Serum total cholesterol, serum LDL cholesterol, and serum HDL cholesterol were all dramatically reduced in the TMX group compared with the ad libitum and pair-fed groups (Table 1
). There was a significant reduction in the levels of serum insulin and leptin between the ad libitum and the two food intake-reduced groups (TMX and pair-fed). No significant changes in serum glucose, serum triglycerides, or serum-free fatty acids were observed between the three groups.
|
Tamoxifen induces changes in the metabolite profile of liver tissue
Once we had confirmed that our TMX model developed early stages of fatty liver, we investigated the metabolite composition of the liver using 1H-NMR. The initial analysis did not reveal drastic differences across the three groups for the three different NMR experiments performed, indicating that metabolic changes in the liver were relatively small. However, at this initial phase of NASH development, PLS-DA analysis of the dataset of spectra from aqueous extracts separated the three groups into clusters according to treatment. This pattern recognition model could be used to predict the class membership of all the spectra, (i.e., predict the treatment the animal had received from the metabolic profile detected). Liver-specific effects mediated by TMX were assessed by analyzing the spectra from aqueous extracts of liver tissue from the TMX and pair-fed group using PLS-DA. The data were clustered according to treatment (Fig. 3
A). Examining the loadings plot for this model to identify which metabolites were responsible for this clustering, TMX treatment correlated with relative increases in choline (110±60% increase in spectral resonance), lactate (95±33% increase), and glutamate (88±23% increase) as well as decreases in myo-inositol (66±22% decrease) and glucose (19±7% decrease) compared with the pair-fed group.
|
Combined PCA of transcript and metabolite changes demonstrate that the most significant changes were associated with the TMX-treated group
When we examined the spectra acquired from intact liver pieces using HRMAS, no differences in the total NMR detectable lipid resonances were found using the solvent suppression pulse sequence. This pulse sequence detects all of the NMR observable lipids in a tissue. This result suggests that the expected dramatic changes in membrane lipid composition or peroxidation typically seen during the second hit have not occurred at this early stage of NASH development. However, using a CPMG pulse sequence, sensitive to the more mobile metabolites, liver tissue from the TMX-treated animals could be clustered from control (ad libitum) and pair-fed groups (Fig. 3B
). This was caused by an increase in saturated lipids between 1.26–1.30 ppm (45±24%) and phosphatidylcholine (16±5%) compared with the pair-fed group. After a t test filter across the three pairwise comparisons for the metabolite and transcript datasets, PCA was performed on the combined transcript and metabolite dataset of significant alterations to cluster changes that varied in a similar manner across the three treatment groups. The major correlated trend across the data was the separation of the TMX exposed group from the other two groups, indicating that the largest effect in the dataset was associated with TMX treatment and not decreased food intake.
Expression analysis to identify genes changed by TMX in liver
Once we had confirmed that the liver of TMX-treated rats had an increased pool of mobile saturated fatty acids, we investigated whether these early changes were associated with specific patterns of gene expression. Using RNA from pair-fed and TMX-treated rat livers, we compared gene expression in these groups following three strategies: 1) microarray analysis of genes with the greatest differential expression, 2) gene expression analysis of lipid metabolism candidate genes, and 3) metabolic pathway analysis of microarray data.
Microarray analysis of genes with the greatest differential expression
Three mutually exclusive pools of liver RNA derived from four rats per pool were generated. Using strict filtering criteria (mean change of at least 2-fold on each of 3 arrays), we found 13 genes with increased expression and 16 genes that had decreased expression (Table 2
). The largest number of genes whose expression was changed are those contributing to lipid metabolism. We identified down-regulation of -5 and -6 fatty acid desaturases and stearoyl-CoA desaturase 2 (SCD2: a -9 desaturase) in the TMX-treated rat livers. We noted decreased expression of the gene encoding phosphatidylethanolamine methyltransferase, the key molecule in de novo phosphatidylcholine synthesis required for membranes, and very low density lipoprotein (VLDL) particle output (33)
. We found that expression of apolipoprotein A-IV, a component of high density lipoproteins (HDLs) and chylomicrons was decreased in TMX-treated liver, suggesting the TMX might be able to interfere with normal lipoprotein output from the liver. Expression of the mRNA encoding 3-hydroxy-3-methylglutaryl-CoA synthase 1 (HMG-CoA synthase 1, HCS), an important regulatory enzyme of cholesterol synthesis, was decreased in TMX-treated rat liver. Genes involved in carbohydrate metabolism were affected. Glucokinase, an enzyme and regulator of liver glucose flux, was decreased, suggesting alterations in glycolytic flux.
|
Gene expression analysis of lipid metabolism candidate genes
The initial microarray analysis identified lipid metabolism as a major target of TMX action. Following a candidate approach, we directly examined the expression of genes involved in pathways of fatty acid and cholesterol metabolism using RPAs.
From RPA-analyzed fatty acid metabolism genes (Fig. 4
A) as well as decreases in -5 and -6 fatty acid desaturases and SCD2 genes by array analysis, we identified additional decreases in genes involved in de novo lipogenesis. Specifically, FAS (pair-fed vs. TMX: 0.224±0.037 vs. 0.106±0.011, P<0.05) stearoyl-CoA desaturase 1 (pair-fed vs. TMX; 2.571±0.599 vs. 1.216±0.343, P<0.05) and acetyl-CoA carboxylase 
(ACC, pair-fed vs. TMX; 0.074±0.006 vs. 0.0439±0.008, P<0.05) were decreased in TMX-treated livers. Expression of mitochondrial glycerol-3-phosphate acyltransferase (mGPAT) remained unchanged. Expression of Sterol receptor element binding protein 1 (SREBP1) mRNA was not changed by either RPA (pair-fed vs. TMX; 0.526±0.075 vs. 0.544±0.132, P=NS) or microarray.
|
From a selection of key genes involved in fatty acid oxidation (Fig. 4B
), peroxisomal proliferator-activated receptor (PPAR) and acyl-CoA oxidase 1 (ACO1) were significantly elevated (PPAR, pair-fed vs. TMX: 0.091±0.011 vs. 0.120±0.019, P<0.05: ACO1, pair-fed vs. TMX: 1.420±0.091 vs. 2.297±0.137, P<0.05), suggesting that both mitochondrial and peroxisomal programs of fat oxidation were activated. Uncoupling protein 2 (UCP2), a PPAR target was significantly elevated when using RPAs (pair-fed vs. TMX; 0.023±0.002 vs. 0.035±0.003, P<0.01).
Analysis of the cholesterol metabolism pathway (Fig. 4C
) revealed that the expression of key enzymes in this pathway, HMG-CoA synthase (HCS, pair-fed vs. TMX; 0.409±0.06 vs. 0.116±0.032, P<0.01) and HMG-CoA reductase (HCR, pair-fed vs. TMX; 0.067±0.011 vs. 0.050±0.011, P<0.05), were decreased. Expression of SREBP2, the transcription factor responsible for activation of the cholesterol synthetic pathway, was unchanged (pair-fed vs. TMX; 0.209±0.033 vs. 0.267±0.040, P=NS). We found that the central enzyme of the cholesterol degradation and bile acid synthetic pathway, cytochrome P450 7A1 (34)
, was elevated on the microarray and by RPA (pair-fed vs. TMX; 0.476±0.015 vs. 0.601±0.022, P<0.05). These gene expression changes suggest that TMX action on the liver in vivo decreases de novo cholesterol biosynthesis while increasing cholesterol disposal and bile acid synthesis.
Since TMX had marked effects in lipoprotein metabolism, we examined changes in the expression of genes involved in lipid and lipoprotein uptake into the liver (Fig. 4D
). We found that expression of the LDL receptor (LDLr) was increased in TMX-treated liver (pair-fed vs. TMX: 0.169±0.007 vs. 0.329±0.019, P<0.0001). The expression of the HDL receptor SR-BI was unchanged (pair-fed vs. TMX; 0.094±0.015 vs. 0.083±0.011, P=NS). These data suggest that elevations of LDLr may be important for reducing circulating LDL and the accumulation of lipids into the liver. However, the levels of SR-BI expression do not account for the decrease in serum HDL.
Globally, examination of the expression of selected genes indicates that fatty acid and cholesterol biosynthesis pathways are reduced and FAO and bile acid production are increased by TMX in liver.
Metabolic pathway analysis of microarray data
Our third approach to the analysis of the microarray data was to use pathway analysis to identify pathways coordinately up- or down-regulated by TMX (summarized in Table 3
). To assemble the data, expression values for each gene from each chip was averaged. Pathway analysis using z-score and GSEA methods generally concurred with the gene-specific analysis cited above, finding that the fatty acid synthesis pathway was significantly down-regulated whereas the mitochondrial FAO pathway was up-regulated in TMX-treated rat livers. However, z-score analysis allowed us to determine that although changes in the fatty acid synthesis pathway were robust and consistent, increases in mitochondrial FAO expression were much smaller overall, since z-score analysis for mitochondrial FAO was only significant using a 1.25-fold increase cutoff. The microarray results for individual genes in these pathways are provided in the supplementary data. Globally, these different methods of gene expression analysis suggest that TMX action on the liver in vivo decreases expression of genes involved in de novo fatty acid synthesis and desaturation yet elevates genes associated with FAO. The pattern of gene expression involved in the cholesterol biosynthesis pathway was unchanged despite HMG-CoA and HMG-CoA reductase gene expression being decreased using RPA. Microarray analysis shows that HMG-CoA synthase is significantly reduced whereas three genes within the pathway are elevated [acetoacetyl-CoA thiolase (1.51), FPP synthase (1.86), and CYP51 (2.11)]. Thus, strong but opposing changes in gene expression lead to no change overall in this global pathway analysis at the mRNA level.
|
Z-score analysis indicated that expression of genes involved in glycolysis and gluconeogenesis were decreased in the TMX-treated rat livers. Examination of individual genes reveals that, as well as glucokinase, genes for the enzymes pyruvate kinase, lactate dehydrogenase, glycerol kinase, and phosphofructokinase B were significantly decreased, suggesting a coordinated suppression of this metabolic pathway (See supplementary data). Since glucokinase, phosphofructokinase, and pyruvate kinase are major points of glycolytic control, liver glucose utilization may be disturbed in addition to liver FA utilization. Other pathways involved in substrate metabolism were changed (Table 4
). Pathways for carbohydrate transporters, cysteine metabolism, and isomerase were all increased, whereas those for eicosanoid metabolism, prostaglandin metabolism, protein catabolism, steroid metabolism, and the mitochondrial inner membrane were decreased in the TMX-treated livers.
|
In vitro 48 h TMX treatment of rat primary hepatocytes replicates the effects of TMX on FAS and HMG-CoA reductase gene expression seen in vivo
The most compelling observation using our in vivo model of early steatosis development was that TMX reduced gene expression of the components of fatty acid synthesis. We used rat primary hepatocytes as an in vitro model to verify the direct effects of TMX in these cells. We reasoned that this experimental paradigm might identify early events (within 48 h of the drug administration) and avoid the problem of indirect effects induced by TMX in other tissues. Results of these assays are summarized in Table 4
. Of the fatty acid synthesis enzymes evaluated, the robust effect of TMX decreasing FAS mRNA in vivo was reproducible in vitro. However, under these experimental conditions we did not observe changes in SCD-1 gene expression. mGPAT expression was reduced in TMX-treated primary hepatocytes. Gene expression analysis of genes involved in fatty acid oxidation produced opposite changes compared with in vivo. Expression of FAO genes PPAR and ACO1 was decreased in TMX-treated hepatocytes. Expression of genes involved in cholesterol synthesis, bile acid synthesis, and lipoprotein uptake were either unchanged or regulated contrary to those seen in vivo with the exception of HMG-CoA reductase.
Thus, the only changes in gene expression replicated by in vivo and in vitro by TMX treatment are the decreases in FAS and HMG-CoA reductase mRNA, implicating these genes as direct targets of TMX action in the liver.
FAS activity is reduced and malonyl-CoA levels increased in livers from TMX-treated rats
We wanted to know whether the decrease in FAS mRNA mediated by TMX translated into a measurable effect on activity. We saw there were significant decreases of FAS activity (Fig. 5
A) in the pair-fed livers compared with ad libitum and between TMX livers and pair-fed (ad libitum vs. PF vs. TMX, nM NADPH oxidized/µg/min: 17.3±1.4 vs.12.3±0.8 vs. 5.3±0.5: ad libitum vs. PF P<0.01, PF vs. TMX P<0.001, n=6–8/group). This demonstrates that FAS activity in TMX-treated livers is significantly reduced. Since our gene expression data suggested a specific effect of TMX on FAS, we speculated that this change would be associated with accumulation of malonyl-CoA. We found a 56% elevation in malonyl-CoA levels in livers from TMX-treated rats compared with pair-fed rats (ad libitum vs. PF vs. TMX, malonyl-CoA (nmol/g): 7.0±0.3 vs. 7.6±0.3 vs. 11.9±0.6: PF vs. TMX P<0.0001, n=6–8/group) (Fig. 5B
). Thus, these results demonstrate that TMX inhibits preferentially FAS and that this results in accumulation of its substrate malonyl-CoA.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In this report we show that an early and specific effect of TMX leading to hepatic steatosis is the inhibition of lipogenesis de novo by directly inhibiting the expression and activity of FAS in the liver. Inhibition of FAS results in accumulation of malonyl-CoA, an important inhibitor of mitochondrial FAO. The target for malonyl-CoA as an inhibitor of FAO is carnitine palmitoyltransferase 1 (CPT-1), the enzyme catalyzing the first step for transporting FAs into the mitochondria. As indicated above, the complexity of the pathogenic mechanisms of NASH precludes discerning its primary pathogenic mechanisms. However, our TMX model, with the development of moderate accumulation of saturated fatty acids but without major metabolic disturbance as indicated by NMR studies, allowed us to identify the inhibition of FAS as a direct primary effect of TMX in the liver. Under this experimental protocol, TMX severely depleted HDL, LDL, and total serum cholesterol in combination with a microvesicular pattern of lipid accumulation within the liver. This pattern of lipid infiltration is consistent with previously reported pharmacological models of impaired fatty acid ß-oxidation in response to ROS mediated mitochondrial damage (11
, 35)
. We consider that the most likely mechanisms leading to the second hit of the TMX NASH model may relate to a direct accumulation of fatty acids, followed by downstream conversion into reactive FA-derived molecules (for example, phospholipids or ceramides), which could have significant effects on intracellular signaling.
Previous research has shown that TMX can directly disrupt the mitochondrial respiratory chain when administered directly to isolated mitochondria (36
, 37)
. We have analyzed the respiratory capacity of mitochondria extracted from the livers of animals treated with TMX as well as controls and confirmed that under our experimental conditions there is no difference in the metabolic profiles between the groups (Supplementary Fig. S1). We consider the concentration of TMX at the level of the mitochondria in our in vivo experiments to be much smaller than that used in vitro by Tuquet et al. (37)
. Thus, we consider the major action of TMX to be a specific and direct inhibition of FAS activity, leading to malonyl-CoA elevation and CPT-1 inhibition.
Exposure of animals to TMX produced distinct metabolic changes in liver tissue compared with ad libitum animals or pair-fed matched animals. This was caused by increases in choline, phosphocholine, phosphatidylcholine, glutamate, and mobile saturated lipids and a decrease in glucose and myoinositol. These metabolic changes are similar to those detected in the early stages of orotic acid-induced fatty liver disease (38
, 39)
. Orotic acid and TMX administration perturb choline metabolism, with exposure to orotic acid over 14 days causing a decrease in hepatic choline concentration but an increase in its products phosphocholine, phosphatidylcholine, betaine, and trimethylamine N-oxide. Both drugs decreased glycolysis as measured by transcriptional and metabolic analyses. Since our strategy was to characterize the first hit—the initial alterations leading to the development of fatty liver—it was not surprising no changes were detected in the HRMAS 1H-NMR spectra of liver tissue without a T2 filter. This suggested that the total NMR observable lipid content of the livers (e.g., lipids from membranes) was still unchanged, although we could identify increases in the quantity of saturated mobile lipids. These changes in liver lipid compartmentalization were accompanied by a paradoxical reduction of gene expression in major regulators of fatty acid synthesis such as FAS and SCD, suggesting accumulation of lipids preferentially from an extrahepatic source, probably released by the adipose tissue. Our finding that serum LDL and HDL cholesterol levels are decreased whereas LDLr mRNA is increased in livers of TMX-treated rats suggests development of lipid accumulation may be facilitated via increased LDL uptake. However, our results from in vitro hepatocytes show that this increase in LDLr in vivo is likely to be a secondary event, perhaps due to the inhibition of de novo fatty acid synthesis.
We used pathway analysis of the microarray data to take the analysis beyond simple lists of genes satisfying stringent expression and statistical criteria and look for trends among the more subtle gene expression changes. We used two different methods of assessing pathway or functional enrichment to compare their performance. Although there was good agreement between the two methods, GSEA gave a greater number of positive results than the z-score method. Combining expression information with pathway annotations allowed confirmation of predicted expression changes and trends (e.g., fatty acid synthesis) or identification of unexpected pathways for further investigation (e.g., amino acid metabolism, protein catabolism), thus enhancing the breadth of data gained from microarray-based analyses.
Pathway analysis of gene expression revealed that lipogenesis de novo was impaired by TMX. This resulted in accumulation of malonyl-CoA CoA, which would be expected to inhibit CPT-1 and thereby FAO. These inhibitory effects in FAO may have been partially compensated in response to fat accumulation by up-regulation of the genetic program of FAO. A new steady state seems to have been achieved by increasing hepatic uptake of fatty acids. In the face of a continued supply of exogenous free fatty acid, the blockade of fatty acid oxidation produced by elevated malonyl-CoA CoA is likely to be the major factor leading to steatosis.
Previous papers have suggested that inhibition of FA synthesis reduces the levels of hepatic steatosis in mouse models (40
, 41)
. However, in these cases there is a coordinated decrease of all the lipogenic steps since the modification of FA synthesis occurs at the level of key transcription factors, which control the pathway. In these cases, flux through the pathway is reduced, resulting in a reduction of the intermediates of this pathway; malonyl-CoA levels do not rise and inhibit CPT-1. Our model for the mechanism of the first hit of TMX-induced fatty liver may involve an initial specific decrease in FAS mRNA expression and activity, leading to three responses. First, specifically decreased FAS activity results in an elevation of malonyl-CoA, inhibition of FAO and retention of fatty acids. Second, a reduction in de novo fatty acid synthesis may facilitate elevated lipid uptake from the periphery (i.e., via the LDL receptor). Elevated uptake of peripheral lipoproteins would raise liver cholesterol levels, accelerate bile acid production, and reduce endogenous cholesterol synthesis by reducing expression of HMG-CoA synthase and reductase. Third, the changes in FAO genes and UCP2 expression may reflect attempts by the liver to stimulate FAO and reduce the accumulated lipids and ROS of the liver. Decreases in glycolytic gene expression may be symptomatic of concomitant alterations in carbohydrate metabolism. This could be interpreted as an attempted switch from carbohydrate to FAO metabolism or an attempt to force FAO despite inhibition at a key step. Alternatively, given elevations in tissue lactate levels, anaerobic glycolysis may be activated to provide some ATP in the face of impaired mitochondrial function. Decreased expression of glycolytic enzymes here would be an attempt to hinder lactate accumulation.
In summary, our data indicate that the generation of hepatic steatosis in rats by administration of TMX is linked to primary disruption of fatty acid metabolism. We have identified fatty acid synthase as a potential primary target of TMX, with a specific reduction in its activity leading to impaired of ß-oxidation via elevated malonyl-CoA. This provides a novel mechanism for some forms of NASH and related disorders. We also provide evidence that the study of early stages of disease models using powerful profiling technologies and pathway analysis may provide important new insights in the pathogenic mechanisms of complex diseases.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Received for publication October 12, 2004. Accepted for publication February 22, 2005.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  I. Larosche, P. Letteron, B. Fromenty, N. Vadrot, A. Abbey-Toby, G. Feldmann, D. Pessayre, and A. Mansouri Tamoxifen Inhibits Topoisomerases, Depletes Mitochondrial DNA, and Triggers Steatosis in Mouse Liver J. Pharmacol. Exp. Ther., May 1, 2007; 321(2): 526 - 535. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. C. Goodall, L. Ellis, G. S. H. Yeo, and J. S. H. Gaston Does HLA-B27 influence the monocyte inflammatory response to lipopolysaccharide? Rheumatology, February 1, 2007; 46(2): 232 - 237. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 O. A. Gudbrandsen, T. H. Rost, and R. K. Berge Causes and prevention of tamoxifen-induced accumulation of triacylglycerol in rat liver J. Lipid Res., October 1, 2006; 47(10): 2223 - 2232. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 R. Lupu and J. A. Menendez Targeting Fatty Acid Synthase in Breast and Endometrial Cancer: An Alternative to Selective Estrogen Receptor Modulators? Endocrinology, September 1, 2006; 147(9): 4056 - 4066. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. Lopez, C. J. Lelliott, S. Tovar, W. Kimber, R. Gallego, S. Virtue, M. Blount, M. J. Vazquez, N. Finer, T. J. Powles, et al. Tamoxifen-Induced Anorexia Is Associated With Fatty Acid Synthase Inhibition in the Ventromedial Nucleus of the Hypothalamus and Accumulation of Malonyl-CoA. Diabetes, May 1, 2006; 55(5): 1327 - 1336. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. J. Kauffman, B. A. Ogunnaike, and J. S. Edwards Designing experiments that aid in the identification of regulatory networks Brief Funct Genomic Proteomic, February 1, 2006; 4(4): 331 - 342. [Abstract] [Full Text] [PDF]
|
|
), pair-fed (
), and TMX-treated (•) rats. B) CPMG of hepatic tissue from ad libitum (
