Effect of dietary monosodium glutamate on trans fat-induced nonalcoholic fatty liver disease.

The effects of dietary monosodium glutamate (MSG) on trans-fatty acid (TFA)-induced nonalcoholic fatty liver disease (NAFLD) are addressed in an animal model. We used Affymetrix microarray analysis to investigate hepatic gene expression and the contribution of visceral white adipose tissue (WAT) to diet-induced NAFLD. Trans-fat feeding increased serum leptin, FFA, HDL-cholesterol (HDL-C), and total cholesterol (T-CHOL) levels, while robustly elevating the expression of genes involved in hepatic lipogenesis, including the transcription factor sterol-regulatory element binding protein 1c. Histological examination revealed hepatic macrosteatosis in TFA-fed animals. Conversely, dietary MSG at doses similar to human average daily intake caused hepatic microsteatosis and the expression of &bgr;-oxidative genes. Serum triglyceride, FFA, and insulin levels were elevated in MSG-treated animals. The abdominal cavities of TFA- or MSG-treated animals had increased WAT deposition compared with controls. Microarray analysis of WAT gene expression revealed increased lipid biosynthetic gene expression, together with a 50% decrease in the key transcription factor Ppargc1a. A combination of TFA+MSG resulted in the highest levels of serum HDL-C, T-CHOL, and leptin. Microarray analysis of TFA+MSG-treated livers showed elevated expression of markers of hepatic inflammation, lipid storage, cell damage, and cell cycle impairment. TFA+MSG mice also had a high degree of WAT deposition and lipogenic gene expression. Levels of Ppargc1a were further reduced to 25% by TFA+MSG treatment. MSG exacerbates TFA-induced NAFLD.

and again at 28-30 weeks. The breeding and care of the animals were in accordance with the protocols approved by the Animal Care and Use Committee of the King Faisal Specialist Hospital and Research Centre.

Measurement of murine serum triglycerides, FFAs, cholesterols, fasting glucose, insulin, adiponectin, leptin, and RBP4 levels
Serum triglyceride (TG), total cholesterol (T-CHOL), and HDL-cholesterol (HDL-C) concentrations were measured in overnight fasted male 32-week-old mice (n = 20) using the Refl overt Plus instrument (Roche, F. Hoffmann-La Roche, Basel, Switzerland) according to the manufacturer's instructions. Overnight fasting blood glucose was measured using the Ascensia Contour glucometer (Bayer HealthCare, IN). FFAs were measured in mouse serum using the Half Micro Test (Roche Diagnostics, Mannheim, Germany). Hormone measurements were also taken at 32 weeks of age. Fasting serum insulin was measured using the ultrasensitive mouse insulin ELISA kit from Mercodia (Uppsala, Sweden) according to the manufacturer's instructions. Fasting serum leptin and adiponectin/Acrp30 were measured by ELISA using commercial assay kits (MRP300 mouse Adiponectin/Acrp30; MOB00 mouse leptin, R and D Systems, Minneapolis, MN). Fasting serum Retinol Binding Protein 4 (RBP4) was measured using the Dual mouse/rat RBP4 ELISA kit (RB0642EK; AdipoGen, Seoul, Korea).

Histology and Oil-Red-O staining of murine liver and adipose tissue
Liver and visceral adipose tissues were removed and weighed. Formalin-fi xed, paraffi n-embedded liver and adipose tissue from eight of the 32-week-old male mice were processed, and 4-mthick serial sections were cut and stained with Oil-Red-O or hematoxylin and eosin (HE) for lipid analysis according to standard pathology laboratory procedures. After mounting with glycerol gelatin, images were captured using Axio Vision Rel4.5 software (Carl Zeiss). For measurement of adipocyte cell areas, images were converted to TIFF fi les for analysis using the Discovery Series TM Quantity One® 1-D analysis software (Bio-Rad Laboratories).

Hepatic TG quantitation
Levels of mouse liver TG were quantifi ed using the Triglyceride Determination Kit TRO100 with appropriate TG standards (Sigma-Aldrich, St. Louis, MO). Frozen liver samples from 32week-old mice were fi rst powdered under liquid nitrogen and 120 mg of the frozen liver powder was weighed into 2 ml of chloroform:methanol mix (2:1, v/v) and incubated for 2 h at room temperature with occasional shaking. Following the addition of 0.2 volumes of water, vortexing, and centrifuging at 2,500 g , the lower phase containing the lipids was collected and dried under vacuum in a rotary evaporator for 5-6 h. The dried pellets were resuspended in the reaction buffer provided in the kit. Results were expressed as mean TG (mg/g tissue) ± SEM, n = 6 per diet group.

RNA isolation
To detect early changes in gene expression, total RNA was prepared from the liver and visceral adipose tissue taken from 16-week-old mice (n = 4) in the four different diet groups using the Qiagen RNeasy kit according to the manufacturer's instructions and stored at Ϫ 80 o C. RNA quality was verifi ed using a 2100 Bioanalyzer instrument and an RNA 6000 Nano LabChip assay (Agilent Technologies, Santa Clara, CA). RNA concentrations were determined by absorption at 260-nm wavelength with an ND-1000 spectrometer (Nanodrop Technologies, Wilmington, DE).
The mechanism for this is believed to involve glutamateinduced degeneration of those areas of the immature brain that are insuffi ciently protected by a mature bloodbrain barrier, including regions that regulate feeding and satiety ( 15 ). MSG consumption has increased globally in recent years, with recent estimations of the current average daily intake believed to be up to 10 g/day ( 16 ). In 1974, an acceptable daily intake for MSG was set at 0-120 mg/kg body weight (bw) ( 17 ). In Europe, intake of MSG consumption is estimated as approximately 30 mg/kg bw ( 18 ), although in some countries, intake of MSG could be as high as 143 mg/kg bw ( 16 ). In addition to obesity and dyslipidemia, recent data suggest that MSG may cause hepatic steatosis ( 19 ), infl ammation, and dysplasia ( 16 ); however, the mechanism behind this is incompletely understood. The aim of this work was to establish the effect of dietary TFA and MSG on dyslipidemia, hepatic steatosis, and gene expression profi le and to assess the role of visceral white adipose tissue (WAT) in the pathogenesis of nonalcoholic fatty liver disease using an in vivo animal model. The amount of low-dose oral MSG used in this study (91 mg/kg bw) refl ects current consumption levels ( 17,18 ) and is 30-40 times less than the level previously reported to induce neuronal damage when injected neonatally (13)(14)(15). Since it has previously been suggested that MSG excitotoxicity occurs only when the blood brain barrier is vulnerable, for example, neonatally ( 15 ), we studied C57Bl/6J mice that had been bred and weaned from animals maintained on the respective diets for a 3-week run-in period prior to mating. We analyzed hepatic and WAT microarray gene expression in order to gain insight into the commonalities and differences between the mechanism of TFA-induced steatosis versus that induced by MSG. To our knowledge, this is the fi rst article to address the effect of these common food substances on hepatic and WAT gene expression in a rodent model of NAFLD.

Animals and diets
C57BL/6J mice were from The Jackson Laboratory (Maine) and were housed/caged and fed a standard chow diet until 6 weeks of age, whereupon they were placed in one of four different dietary regimens for a period of 3 weeks prior to mating. The four diet regimens used in this study were as follows: 1 ) ad lib standard chow (control diet) with ad lib drinking water; 2 ) ad lib standard chow with ad lib drinking water containing 0.64 g/l (91.2 ± 4.6 mg/kg bw):MSG(MSG diet); 3 ) ad lib test diet Purina 5001 with 20% (w/w) partially hydrogenated vegetable shortening [Test Diet #5C4M containing 8.68% (w/w) TFAs; Test Diet Purina] and ad lib drinking water (TFA diet); and [4] Diet #5C4M and ad lib drinking water containing 0.64 g/l (97 mg/kg bw) MSG (TFA+MSG diet). See Table 1 for diet composition. Following mating, the four groups of dams were maintained on their respective diets throughout the gestation and nursing period. Male offspring used in these experiments were weighed and then weaned onto the same diets and maintained in this regimen until they reached either 16 or 32 weeks of age. Average body weight and caliper girth was assessed at 6, 16, and 32 weeks of age; water intake and food consumption were assessed at 6 weeks amount of specifi c product with melting temperature separable from primer-dimer temperature.

Relative standard curves and PCR assay conditions
Standard curves were prepared for each run using known quantities of cDNA (10-fold dilutions beginning at 15 ng/ l) and primers for the gene of interest or ␤ -actin. Light Cycler-Fast Start DNA Master SYBR Green I reaction mix containing 0.4 M forward and reverse primers, 4 M MgCl 2 , and 1 l LightCycler-Fast start DNA master SYBR Green 1 dye was used for quantitative real-time PCR analysis according to the manufacturer's protocols (Roche Molecular Biochemicals, Mannheim, Germany). A volume of 1 l from a 10-fold dilution of cDNA was added as the PCR template. A no-target control received 9 l reaction mix and 1 l water. PCR amplifi cation was performed in triplicate wells. A four-step experimental run protocol was performed: i ) denaturation program (10 min at 95°C); ii ) 45 cycles of four-segment amplifi cations consisting of denaturation at 95°C for 10 s, annealing (60-56°C) for 5 s, extension at 72°C for 10 s, and data acquisition at 83°C for 1 s. A temperature transition rate of 2°C/s with a single fl uorescence measurement was used. iii ) Melting curve program (50-95°C, with heating rate of 0.1°C/s up to 98°C with continuous fl uorescence measurement); and vi ) a cooling program down to 40°C. Relative quantitation measurements were performed using external standard curves for both target and ␤ -actin housekeeping gene. The second derivative maximum method was used for cycle threshold (ct) calculation from amplifi cation curves. The respective concentration for any given sample was calculated using crossing-cycle analysis provided by the LightCycler software.

Statistics
Microarray data analysis was performed with the Partek genomic suite software suite 6.3 (Partek, MO). Probe set data were summarized, background adjusted, and quantile normalized using robust multichip average methodology described in Irizarry et al. ( 20 ). A false discovery rate correction was applied to post hoc pairwise group comparisons of Pearson's correlation coefficients obtained by ANOVA. Heat map and clustering of the gene expression data were also generated using Partek. P р 0.05 was considered signifi cant, with an arbitrary threshold of 1.5-fold difference between the four groups analyzed simultaneously. As the mRNA expression in four animals from each of the four diet groups was separately analyzed, four comparisons were possible when comparing the effects of TFA diet animals with respect to controls receiving standard chow and the effect of the addition of MSG to both diets. Genes fulfi lling the Affymetrix quality criteria for signifi cant expression were considered to be differentially expressed between the four diet groups when the expression levels were concordantly increased or decreased in all four comparisons. Real-time RT-PCR values for each target gene were calculated as a ratio of target gene expression level to the ␤ -actin expression level in the same specimen. For concordance comparisons with the microarray data, these values were then expressed as mean percentage ± SEM of real-time RT-PCR levels in the control diet group. P values of <0.05 were considered statistically signifi cant. Pearson correlation coeffi cients ( r ) between the microarray data and the qRT-PCR were calculated. A value of 1 is indicative of a perfect correlation.

SREBP1c activity measurement
SREBP1c activity was determined in nuclear fractions of liver using the ELISA-based SREBP1 activation TransAM kit (Active Motif, CA). The Trans-AM SREBP1 kit contained a 96-well plate on which an oligonucleotide containing a sterol-responsive ele-Liver/adipose samples were pooled into two samples per diet group and applied onto two microarray chips, giving a total of n = 4 samples.

Microarray hybridization
Gene expression in these samples was analyzed using GeneChip Mouse Gene 1.0 ST array. A high-density oligonucleotide microarray (Affymetrix, CA) representing 28,853 genes, with approximately 27 probes spread across the full length of each transcript, was used. Targets were prepared and microarrays were processed as described in the Affymetrix GeneChip Whole Transcript Expression Analysis manual using the commercially available Affymetrix GeneChip WT cDNA synthesis kit, WT cDNA amplifi cation kit, and WT Terminal labeling kit as per the manufacturer's instructions. Briefl y, approximately 200 ng of total RNA was used to synthesize double-stranded DNA with random hexamers tagged with a T7 promoter sequence. The cDNA was used as a template for in vitro transcription. In the second cycle of cDNA synthesis, random primers were used in reverse transcription to convert the cRNA into single-stranded DNA, which was fragmented, labeled, and hybridized to the array for 16 h using the Fluidics 450 station. Arrays were scanned using the Affymetrix 3000 7G scanner and GeneChip Operating Software version 1.4 to produce CEL intensity fi les. This software also provided summary reports by which array QA metrics were evaluated, including average background, average signal, and 3 ′ /5 ′ expression ratios for spike-in controls, ␤ -actin, and GAPDH.

Real-time RT-PCR
Confi rmation of microarray results was performed by quantitative real-time RT-PCR (qRT-PCR) on independently derived mRNAs from the livers of a different set of four 16-week-old mice per diet group. RNA quality was verifi ed using a 2100 Bioanalyzer instrument and an RNA 6000 Nano LabChip assay (Agilent Technologies, Palo Alto, CA). Total RNA (2 g) was reverse transcribed using Promega Reverse Transcription system (Promega, Madison, WI).

Hepatic histology, lipid staining, and TG content
Increased hepatic lipid deposition as depicted with Oil-Red-O staining was apparent in 32-week MSG-and TFA-treated animals compared with control ( Fig. 1A ) . The livers of TFA-treated animals contained larger vesicular structures consistent with macrovesicular steatosis ( Fig. 1A, B ). Conversely, microvesicular steatosis was apparent in the livers of MSG-treated animals, as indicated by Oils-Red-O ( Fig. 1A ) and HE staining ( Fig. 1B ) of mouse liver tissue. Interestingly, in the TFA+MSG animals, microvesicular steatosis was predominant, and there was an attenuation of macrovesicular structures in the livers from this diet group ( Fig. 1B ). Higher magnification revealed an increase in hepatic cell volume and nuclear size in MSG-treated livers ( Fig. 1C ). While the weight of livers from control, MSG, and TFA diet group mice did not differ signifi cantly, livers from TFA+MSGtreated mice weighed an average 1.54-fold higher than control ( Fig. 1D ; P р 0.05, n = 10). Intrahepatic TG levels were quantifi ed ( Fig. 1E ). The highest level of hepatic TG occurred in the TFA diet group with a 1.43-fold increase in hepatic lipid content compared with control ( Fig. 1E ; P р 0.05, n = 6).

Hepatic gene expression
To detect early changes in gene expression that might account for the hepatic steatosis and dyslipidemia seen in 32-week-old mice, total RNA samples were isolated from liver tissue obtained from four individual 16-week-old mice from each of the diet groups and pooled (n = 2) for applying onto two individual Affymetrix Mouse Genome 1.0 ST Array Expression Set GeneChips containing 28,853 genes. Statistical analysis of these two sets of paired samples using a t -test ( P р 0.05, n = 4) identifi ed a group of 84 hepatic genes whose expression was signifi cantly changed within the four diet groups. Figure 2 shows the heat map and clustering dendogram of the differentially expressed genes. Table 3 shows 84 genes falling into 12 different functional ontologies, which were differentially expressed within the four groups, using a 1.5-fold change as a significance threshold ( P р 0.05). The expression of genes involved in lipid biosynthesis, storage, and catabolism was most signifi cantly affected by the different diets compared with control. The greatest effect on hepatic gene expression was seen in the TFA diet, where the expression of 21 genes involved in lipid biosynthesis was markedly upregulated ( Table 3; P р 0.05). Key regulatory protein SREBP1c, known to control the expression of a number of lipogenic genes ( 21 ), was also increased by the TFA diet by 2.5-fold. Other biosynthetic genes included 3-hydroxy-3-methylglytaryl-CoA synthase (3.6-fold), Fatty acid synthase (Fasn; 2.4-fold), and ATP citrate lyase (Acly; 1.6-fold). Interestingly, the expression of many TFA-induced lipogenic genes was decreased to control levels by the combination ment specifi c for SREB1c had been immobilized. SREBP1c contained in nuclear extracts (10 g liver extracts from four mice per diet group) specifi cally bound to this oligonucleotide and was detected by addition of a primary antibody that recognized an accessible epitope on SREBP1 protein upon DNA binding. The specifi city of this assay was assessed by incubating the nuclear extract with 5 pmol/well of a wild-type consensus oligonucleotide provided in the kit, which competes with the nuclear extract SREBP1 for the immobilized probe on the assay plate. Addition of a secondary horseradish peroxidase-conjugated antibody provided a colorimetric readout quantifi ed by spectrophotometry (450 nm). Results were expressed as mean OD (450 nm) ± SEM (n = 4). Table 1 lists percentages of dietary components with respect to weights of the fresh materials. Daily food and water intake was not signifi cantly different within the four diet groups. MSG was added to the drinking water at a final concentration of 0. 64 g/l so as not to infl uence food consumption. Averaged daily MSG intake was 91.21 ± 4.63 mg/kg bw. Initial and fi nal body weight were comparable between the four groups; however, the percentage weight changes between 6 and 32 weeks differed signifi cantly, with the MSG-treated groups achieving a lower growth rate and TFA diets achieving the largest increase in body weight P р 0.05, n = 20 ( Table 2 ) . The overall difference in weight change between the four diet groups was highly signifi cant ( P р 0.001). Abdominal girth was markedly increased in the TFA+MSG group ( P р 0.001), with a maximum girth increase of 52.1 ± 2.5% occurring between 6 and 32 weeks of age.

Lipid metabolism, serum glucose, insulin, and adipokine profi le
Total TG, FFA, and HDL-C were increased signifi cantly in all three diet groups compared with control ( Table 2; P = 0.0402, P р 0.0001, and P = 0.0022, respectively). The highest levels of serum TG, T-CHOL, and HDL-C were found in the TFA+MSG group, with increases of 39.3, 10.5, and 42.8%, respectively ( P р 0.05). Fasting blood glucose, insulin, leptin, adiponectin, and RBP4 were measured ( Table 2 ). There was no signifi cant difference in fasting glucose and adiponectin levels. Fasting insulin was significantly elevated in the MSG group ( P < 0.05, n = 6). Serum patic fi brosis, including ␤ -laminin (Lamb3b) and ␥ -actin (Actg1), were increased in the TFA+MSG livers by 1.9-and 1.5-fold, respectively ( Table 3; P р 0.05). The MSG diet induced an increase in the expression of a number of key transcriptional regulatory proteins, including Growth Arrest and DNA damage-inducible 45B (GADD45b; 2.7-fold) and NIPA-like domain containing 1 (2.3-fold); in each case, expression returned to control levels in animals treated with the TFA+MSG diet. The TFA+MSG diet also induced the expression of genes involved in the infl ammatory pathway, including a 1.4-fold increase in Interferon-␣ 14, a 7.5-fold increase in adipsin (complement factor D), and a 2-fold increase in defensin ␤ 1. Hepatic expression of apoptotic regulator p21 was increased by 2.4-fold in the TFA+MSG diet, further suggestive of a link between oxidative stress, infl ammation, and cell cycle impairment in this model of diet-induced NAFLD.

Confi rmation of Affymetrix microarray by real-time qRT-PCR
A set of fi ve genes was randomly selected for validation via real-time qRT-PCR using mRNA derived from mice other than those used for the microarray study (n = 4 per diet group). Figure 3A shows qRT-PCR analysis of hepatic expression of CIDEC (NM_178373), cyp7a1 (NM_ 007824), GADD45b (NM_008655), MTTP (NM_008642), and SREBP1c (NM_011480) levels relative to housekeeping ␤ -actin levels in mice from the four different diet groups (n = 4, P < 0.05). Figure 3B shows representative ethidium bromide-stained RT-PCR products electrophoresed on 2% agarose gels. Figure 3C shows concordance of qRT-PCR results versus microarray results expressed as signal intensity (percentage control diet ± SD). Pearson correlation coeffi cients ( r ) are indicated on each graph. Values of r were >0.9 in all cases, suggesting a high level of agreement between the microarray and the qRT-PCR data. Individual values were as follows: CIDEC, 0.99; Cyp7a1, 0.99; GADD45b, 0.99; MTTP, 0.99; and SREBP1c, 0.96 ( P < 0.05). of dietary TFA+MSG, including Acly, Fasn, Insulin induced gene 1, and Hydroxy-␦ -5-steroid dehydrogenase 3 ␤ (Hsd3b2 and Hsd3b3). The TFA diet also increased the expression of other genes involved in cholesterol synthesis, such as Farnesyl Diphosphate Synthase (6.7-fold) and Farnesyl Diphosphate Farnesyl Transferase 1 (3-fold); however, the addition of MSG to the TFA diet again reduced expression by approximately one-third. MSG markedly induced the expression of catabolic genes, including microsomal ␤ -oxidative Cyp2b9 and Cyp2b10 (4.4-and 5.1-fold, respectively), and arachidonic acid (AA)-metabolizing Cyp2c55 (2.3-fold). Hepatic expression of these three monooxygenases was also highly upregulated by the TFA diet (7.3-, 7.0-, and 3.5-fold), and unlike the situation with the genes involved in lipid biosynthesis, hepatic expression of these catabolic proteins remained high in the TFA+MSG diet. A second microsomal oxidative protein, fl avin-containing monooxygenase 2 (Fmo2) and its transcriptional regulator aryl hydrocarbon receptor (Ahr) were also upregulated in MSG-treated livers (2.3-and 1.6fold, respectively) but remained unaffected by the other diets. This suggests that part of the effect of MSG in the liver is to increase hepatic lipid catabolism via upregulation of oxidative genes. Additionally, MSG activated genes involved in the bile acid pathway included key regulatory enzyme cholesterol 7 ␣ -hydroxylase (CYP7A1), which was increased by 2.5-fold. This was accompanied by a 1.8-fold decrease in expression of the nuclear receptor SHP, part of the negative transcriptional regulators of classical bile acid synthesis ( 22 ).
Lipid mobilization and storage processes were most affected in livers from the TFA+MSG diet group. Fat-Specifi c Protein 27 (CIDEC), known to be associated with lipid droplet formation ( 23 ), was upregulated by both the MSG diet and the TFA+MSG diet relative to control (1.9-and 4.0-fold, respectively) but unaffected by TFA alone ( Table  3; P < 0.05). Ketogenic Fibroblast Growth Factor 21, a key mediator of hepatic lipid mobilization, was induced 2.4fold. Additionally, expression of genes associated with he- Initial and 32-week body weight, weight change (6-32 weeks), girth change (6-32 weeks), food and water intake, together with fasting serum TG in 32-week-old male C57Bl/6J mice (mean ± SEM, n = 20). FFA, HDL-C, T-CHOL fasting serum insulin, and adipokine levels in 32-week-old male C57BL/6J mice (mean ± SEM, n = 6). For statistical signifi cance, means that are not denoted with a common letter are different: a, b, c, and d, P < 0.05 by ANOVA.
MSG, TFA, and TFA+MSG diet groups compared with control. Figure 5B shows micrographs of WAT stained with HE to demonstrate increase in adipocyte size in tissue from MSG, TFA, and TFA+MSG diet groups compared with control. Figure 5C shows quantitation of averaged adipocyte area. The largest increase in averaged WAT adipocyte size was seen in the TFA group, with up to 153.5 ±

SREBP1c activity
In order to ascertain whether an increase in key transcription factor SREP1c mRNA levels could result in an increase in hepatic SREBP1c activity, we used the Trans AM SREBP1c ELISA-based kit to determine SREBP1c activity in nuclear fractions of liver protein ( Fig. 4 ) . An increase of 2.9-fold SREBP1c activity was seen in nuclear extracts from the TFA diet group compared with control. Specifi city of SREBP1c binding is indicated by the fact that nuclear SREBP1 binding could be inhibited by a wild-type consensus oligonucleotide ( Fig. 4; P р 0.01, n = 4).

Adipose tissue histology
Approximately 60% of hepatocellular lipid is derived from the circulation, particularly from adipose tissue stores ( 24 ). Since MSG diets markedly increased serum TG ( P = 0.0402), FFA ( P р 0.0001), and HDL-C levels ( P = 0.0022) in our animal model, we next investigated visceral WAT distribution and histology in response to the four diets. Figure 5A shows increased abdominal WAT deposition in  Liver expression heat map and unsupervised hierarchical clustering analysis. Red pseudocolor and blue represents upregulated and downregulated genes, respectively. Each row represents a differentially expressed gene within livers from mice in the four diet groups ( P р 0.05). Each column represents the average from four samples in each diet group. For the hierarchical tree clustering of differentially expressed genes from liver, both genes and samples are clustered using Pearson dissimilarity.  iting enzyme involved in mammalian synthesis of longchain polyunsaturated fatty acids ( 25 ), and CDPdiacylglycerol synthase 1, a rate-limiting enzyme in the synthesis of phosphatidyl inositol ( 26 ). MSG also elevated the expression of WAT glycerol kinase and decreased levels of the energy-regulating mitochondrial carrier protein UCP3. Similarly, components of the mitochondrial reduction/oxidation reaction (REDOX) machinery, for example, Ndufs1, Ndufb4, and Ndufa10, were all upregulated in both TFA diet groups by 2.1-, 1.5-, and 1.8-fold, respectively. Interestingly, the expression of many WAT lipid catabolic proteins was decreased in MSG-treated animals, for example, short, medium, long, and very-long-chain acyl-CoA dehydrogenases (Acads, Acadm, Acadl, and Acadvl). REDOX-potentiating Lipin-1 gene expression was reduced in WAT tissue from both MSG diet groups (1.5-and 1.2-fold, respectively). The TFA diet had the opposite effect on the expression of these catabolic genes, with increases of 1.7-, 1.8-, 2.2-, and 2.1-fold, respectively.
5.1% increase compared with control cells. MSG increased adipocyte volume by up to 128.3 ± 7.0% of control. A comparable increase in adipocyte size was obtained in the TFA+MSG diet group (143.4 ± 5.1% of control, P р 0.001).

Adipose tissue gene expression
Affymetrix microarray analysis of visceral WAT genes with expression levels that were signifi cantly different within the four diet groups is shown as a heat map with clustering dendogram in Fig. 6 and numerically in Table 4 ( P р 0.05). A total of 169 genes were identifi ed with signifi cant differences in expression levels at a cutoff range of 1.5-fold. Unsurprisingly, the largest effect on WAT gene expression was obtained in animals fed the TFA diet, with increases in expression of lipogenic Stearoyl-CoA desaturase 2, Fasn, and Acly of 8.2-, 4.3-, and 4.2-fold compared with control. The MSG diet raised expression of several key lipid biosynthetic genes, including Fads2, the rate-lim- Diet-induced changes in 16-week liver gene expression, according to function. Affymetrix analysis comparison within the four diet groups (control, MSG, TFA, and TFA+MSG). Only genes with signifi cant changes in expression are listed (n = 4 per diet group; P р 0.05), with an arbitrary threshold of 1.5-fold.  tion factors, such as peroxisome proliferator-activated receptor-␣ (Ppar ␣ ) and Ppar ␦ . TFA induced the expression of PPAR ␣ by 1.8-fold; however, this increase was attenuated in the TFA+MSG diet. Expression of the key energy regulator Ppargc1a was reduced by half in WAT from MSG-treated animals. A similar reduction was seen in the TFA diet group, and expression of this key energy regulator was reduced to 25% in the TFA+MSG-treated animals, making this gene a likely candidate in regulating WAT adipocyte responses to these three diets. Conversely, levels of SREBP1c were increased in both TFS diets by approximately 3-fold.
Taken together, these results indicate that MSG increases the expression of a number of genes involved in de novo lipogenesis and adipocyte differentiation, while lowering the expression of lipid-catabolizing proteins in WAT tissue. TFA, on the other hand, enhances both lipogenic and lipid catabolic (oxidative) gene expression. A combination of TFA+MSG resulted in a decrease in lipid-catabolizing gene expression and an increase in several lipid biosynthetic genes, resulting in a net increase in serum TG and T-CHOL of 39.3% and 10.5%, respectively.

DISCUSSION
The incidence of NAFLD is increasing throughout the world ( 1, 2 ), and its etiology, though not fully understood, A combination of TFA+MSG lowered the expression of these and other TFA-induced catabolic genes, including hydroxyacyl-CoA dehydrogenase, enoyl CoA hydratase, and acyl-CoA synthase. Conversely, the expression of several lipogenic TFA-induced enzymes was signifi cantly augmented by the addition of MSG, including Fatty acid desaturase 1 and Stearoyl-CoA desaturase 4. Transcriptional regulation of a number of genes affected by these diets are under the control of ligand-activated transcrip- Fig. 4. Effect of diet on hepatic SREBP1c activity: SREBP1c DNA binding activity was determined using ELISA-based SREBP1c activation kit and quantifi ed in triplicate by colorimetry. The levels of hepatic SREBP1c activity increased by 2.9-fold in the livers of TFA diet mice compared with control (dark gray bars, mean ± SEM, P < 0.01, n = 4 per diet group). Inclusion of a wild-type consensus oligonucleotide (light gray bars) attenuated the binding of nuclear SREBP1c, indicative of the specifi city of the assay. are formed by confi gurational isomerization of cis -fatty acids, which results in a prolonged shelf life of vegetable fats and any food products that contain them. However once ingested and incorporated, TFA affects tissue composition and functionality ( 27 ). TFA ingestion has been associated with cardiovascular complications and weight gain ( 28 ). Central obesity can also be induced experimentally by neonatal administration of the ubiquitous dietary fl avor enhancer MSG, at a time when the blood-brain barrier is compromised ( 15 ). Hepatic dysplasia was also recently demonstrated following neonatal MSG administration ( 16 ); however, the mechanism remains largely unknown. In this article, we fi rst examine the mechanism of TFAinduced NAFLD and the contribution of abdominal WAT lipogenesis to the liver pathology, in an animal model of diet-induced fatty liver disease. Second, we investigate whether dietary MSG could contribute to the pathology of NAFLD, either alone or in combination with TFA. A combination of 60% fat and/or neonatal administration of MSG was recently demonstrated to induce obesity in C57BL/6J and NMRI mice; however, the molecular mechanisms behind this were not fully addressed ( 14 ). An acceptable daily intake for MSG has been set at 0-120 mg/kg bw ( 17 ). The oral dose of MSG used in this study (91 mg/ kg bw) refl ects current consumption levels ( 18 ) and is 30-40 times less than the level previously reported to induce neuronal damage when injected neonatally (13)(14)(15). Because it has previously been suggested that MSG excitotoxicity occurs only when the blood-brain barrier is vulnerable, for example, neonatally ( 15 ), we studied C57Bl/6J mice that had been bred and weaned from animals maintained on the respective diets for a 3-week run-in period prior to mating. Interestingly, fasting leptin levels in mice from the MSG diet group were no different than controls. However, leptin levels were raised in the TFA diet and in the TFA+MSG diet group. Since it is known that neonatally injected MSG-lesioned mice with resultant arcuate nucleus ablation exhibit robust hyperleptinemia ( 14 ), one interpretation of our results suggests that the low-dose dietary MSG in our model was insuffi cient to completely impair MSG-sensitive areas of the brain.
Dyslipidemia was apparent in MSG, TFA, and TFA+MSG diet groups compared with control. In the absence of increased exogenous dietary lipid, MSG signifi cantly elevated serum FFA, TG, and HDL-C, while reducing weight gain. TFA feeding increased weight gain, FFA, and HDL-C; however, the largest effect on lipid homeostasis was seen in the TFA+MSG diet, which resulted in a 40% increase in TG levels together with a 10% increase in T-CHOL. Hepatic steatosis with necroinfl ammatory changes was recently observed in mice fed a diet containing 45% of calories derived from fat, together with a high-fructose corn syrup equivalent of 42 g/l ( 29 ). In this model, 30% of the dietary fat was in the form of partially hydrogenated vegetable oil, and interestingly when this was replaced by isocaloric amounts of lard, the extent of steatosis was substantially reduced, indicating greater liver injury with the consumption of synthetic TFA ( 29 ). is believed to include dietary factors. Practically unheard of prior to the 1980s, NAFLD now affects approximately 15-25% of the general population ( 1 ); thus, the cause of this progressive disease is unlikely to be due to genetic abnormalities. Among the several major changes to the global diet during the past 35 years is the introduction of trans -hydrogenated dietary oils into the food chain. TFAs   Fig. 6. WAT expression heat map and unsupervised hierarchical clustering analysis. Red pseudocolor and blue represents upregulated and downregulated genes, respectively. Each row represents a differentially expressed gene within livers from mice in the four diet groups ( P р 0.05). Each column represents the average from four samples in each diet group. For the hierarchical tree clustering of differentially expressed genes from liver, both genes and samples are clustered using Pearson dissimilarity.   Visceral adipose tissue Affymetrix microarray analysis of gene expression: list of genes with signifi cantly different expression changes ( P р 0.05) within the four diet groups (control, MSG, TFA, and TFA+MSG) with an arbitrary threshold of 1.5-fold. thesized as an inactive precursor protein bound to the endoplasmic reticulum and nuclear envelope. Once activated by sterol deprivation, the precursor protein is cleaved to release the activated N-terminal portion, which then enters the nucleus and activates transcription of genes involved in cholesterol and fatty acid synthesis. We used an ELISA-based colorimetric Trans AM SREBP1c assay to confi rm that TFA-induced increases in SREBP1c mRNA lead to an increase in SREBP1c activity, in nuclear extracts of the livers of mice from the TFA diet groups.
Accompanying the increased de novo lipogenesis, livers of TFA+MSG-fed mice had evidence of an increased infl ammatory response, with elevated adipsin expression (complement Factor D). Adipsin is the rate-limiting step in the alternative complement pathway and is activated in alcoholic fatty liver disease ( 38 ). Interferon-␣ 14, and defensin ␤ 1, part of the innate immune response ( 39 ), were also increased. Additionally, livers of TFA+MSG-treated animals had upregulated levels of Fibroblast Growth Factor 21, laminin, and actin-␥ , together with raised p21 levels. Infl ammation and fi brosis are hallmarks of NASH, the severity of each correlating well with levels of hepatic bile acids in these patients ( 40 ). Expression of p21, a universal inhibitor of cyclin-dependent kinases and replicative arrest, was markedly increased in TFA+MSG-treated livers. P21 overexpression is associated with enlarged nuclei ( 41 ) and was also present in the livers of mice in this diet group. Elevated p21 has previously been shown to correlate with periportal fi brogenesis in patients with NASH ( 42 ).Taken in composite, these results suggest that in the absence of high dietary fat, MSG induces the expression of lipid oxidizing and storage genes. TFA, however, induces genes involved in de novo lipid synthesis as well as catabolism. The addition of MSG to the TFA diet appeared to decrease the expression of genes associated with hepatic de novo lipid synthesis, while increasing fat storage genes, infl ammatory markers, profi brogenic markers, and p21 associated with replicative arrest.

Contribution of adipose tissue to diet-induced NAFLD
We next examined the contribution of diet-induced WAT gene expression to the pathology of NAFLD because it has recently been shown that visceral fat is an independent predictor of advanced steatohepatitis ( 43 ). Visceral adiposity and adipocyte size were increased in all three diet groups compared with control. Adipocyte size has recently been shown to correlate well with regional fat mass by Tchoukalova et al. ( 44 ). In WAT, expression of key regulatory protein Ppargc1a was halved by MSG or TFA-containing diets and quartered by a combination of TFA+MSG. Expression of Ppargc1a is reduced in obesity, and this is accompanied by decreased mitochondrial fatty acid oxidation due to coordinated downregulation of genes involved in this pathway ( 45 ). Additionally, genetic polymorphisms in Ppargc1a resulting in liver levels of Ppargc1a mRNA have been observed in patients with NAFLD ( 46 ). In our study, MSG reduced the expression of several WAT ␤ -oxidative genes, whereas TFA markedly enhanced their expression. Unsurprisingly, Ppar ␣ and Ppar ␦ levels were

Hepatic steatosis and gene expression
Increased hepatic TG content and microsteatosis were apparent in MSG-treated animals, in agreement with previous observations by Nakanishi et al. ( 16 ). The liver weight, a marker of steatohepatitis ( 30 ), increased by up to 1.5-fold in TFA+MSG-treated animals. We used Affymetrix microarray to investigate the possible mechanisms for the MSG-induced hepatic dyslipidemia. Most notably, we found that the expression of hepatic genes involved in ␤ -oxidation of fatty acids, bile acid synthesis, and lipid storage was upregulated in MSG-treated mice. Expression of steroid hydroxylases Cyp2b9 and Cyp2b10 was elevated by MSG treatment, together with an increase in Cyp2c55, implicated in AA and linoleic acid catabolism. Evidence for enhanced fatty acid oxidation in NAFLD is provided by Kohjima et al. ( 31 ), who found elevated expression of oxidation-related Cyp2E1 and Cyp4A11 genes among others. In our study, increased microsomal oxidative Fmo2 expression by MSG treatment was accompanied by an elevation in the transcription factor Ahr, believed to modulate the expression of numerous genes involved in cholesterol biosynthesis and fatty acid synthesis ( 32 ). Initially, research focused on the role of Ahr in the induction of AA-epoxygenative Cyp1 ( 33 ); however, it is now thought that Ahr plays a more comprehensive role in lipid homeostasis via downregulation of the nuclear receptor Ppar ␣ ( 34 ). In addition to metabolizing endogenous and exogenously derived lipids, circulating cholesterol is taken up and metabolized in the liver. Dietary MSG elevated serum HDL-C and increased the expression of Cyp7a1, a key regulatory protein in the classical pathway of bile acid formation ( 22 ), while lowering the expression of the nuclear receptor SHP, which has been found to negatively regulate FXR-dependent inhibition of Cyp7a1 transcription. In NAFLD, only roughly one-quarter of hepatic TG is formed from de novo biosynthesis, the rest being derived from dietary sources and uptake of serum FFA ( 35 ). Excess circulating lipids can be taken up and stored in hepatic lipid droplets ( 23 ), and Fat-Specifi c Protein 27 (CIDEC) is one of the main proteins implicated in this process. In our study, hepatic lipid deposition was accompanied by an increase in CIDEC expression, occurring in both the MSG and the TFA+MSG diet groups. Taken together, these results suggest that MSG diets elevate serum lipids and activate genes involved in the storage, oxidation, and clearance of exogenous lipids via activation of the bile acid pathway.
Genes involved in de novo lipogenesis and cholesterol synthesis, such as Acly, FAS, and Hsd3b3, were downregulated in MSG-treated animals. Conversely, these and many other genes were upregulated by trans -fat feeding, including the transcriptional regulatory element SREBP1c. NA-FLD patients have higher levels of SREBP1c and lipogenic gene expression, believed to be the result of impaired insulin signaling ( 36 ). Elevated SREBP1c, Acly, FAS, and other genes involved in the de novo synthesis and uptake of fatty acids have been reported in animal models of NA-FLD ( 37 ) as well as human studies ( 36 ). SREBP1c is syn-with TFA alone, and several lipogenic genes were upregulated. Therefore, MSG augments trans -fat induced hepatic steatosis by altering WAT gene expression and elevating serum lipids and markers of hepatic fi brosis, infl ammation, and DNA damage. Whether the MSG/TFA interaction is mirrored by other dietary fatty acids remains to be elucidated. Further investigation into the molecular pathology of diet-induced NAFLD should address the interaction between other dietary lipids and MSG. enhanced in the TFA diet group, as it has previously been demonstrated that these nuclear receptors regulate fatty acid degradation ( 47,48 ). Conversely, Ppar ␣ and Ppar ␦ levels were unchanged in the MSG diet groups, and Lipin-1, an inducible amplifi er of mitochondrial oxidation ( 49 ), was signifi cantly reduced. Lipin-1 can be induced by Ppargc1a and is believed to activate Ppar ␣ target pathways to increase lipid catabolism ( 49 ). Reduced Lipin-1 activity is also seen in obesity ( 50 ), adding support to the notion that dietary MSG may decrease adipocyte fatty acid oxidation. Several key WAT lipogenic genes were upregulated in MSG-treated animals, including FADS2, the rate-limiting enzyme involved in mammalian synthesis of long-chain polyunsaturated fatty acids ( 25 ), CDP-diacylglycerol synthase 1, a rate-limiting enzyme in the synthesis of phosphatidyl inositol ( 26 ), and glycerol kinase. MSG-induced enhancement of the expression of these enzymes may account for the marked elevation in serum FFA and TG observed in this study.
The trans -fat diet induced WAT expression of many lipogenic proteins, including Acly, FAS, and Stearoyl-CoA desaturase 2. SREBP1c, known to regulate lipogenic gene expression ( 51 ), was also elevated in this diet group, together with increased leptin and LDL receptor levels, indicative of an obesogenic expression profi le.
The highest levels of serum T-CHOL and HDL-C were seen in the TFA+MSG diet group. Adipocytes are capable of de novo TG and cholesterol synthesis and storage ( 52 ), with adipocyte size correlating well with cholesterol and TG content. We looked for WAT genes in the TFA+MSG diet group with the highest change in expression that might account for the accompanying elevation in serum lipids. Expression of Fatty acid desaturase 1, Stearoyl-CoA desaturase 4, Phospholipase A2, scavenger receptor-B1, and Fatty Acid Transporter were all highest in this group, suggesting that genes involved in lipogenesis and fat storage were upregulated. Additionally, genes involved in adipocyte differentiation, including leptin, Fgf-8, Fgf-11, and monocyte to macrophage differentiation-associated protein, were maximally elevated in the TFA+MSG diet group relative to the other groups.
In conclusion, we have established an animal model of diet-induced NAFLD that has allowed us to speculate on the contribution of abdominal WAT to the pathogenesis of hepatic steatosis. Dietary MSG at doses similar to the average daily intake caused hepatic microsteatosis and the expression of ␤ -oxidative genes in the offspring of mice previously established on the diet. Fasting serum TG, FFA, and insulin levels were elevated. WAT adipocytes were enlarged and showed an increase in lipid biosynthetic gene expression and a decrease in Ppargc1a levels. Trans -fat feeding increased serum leptin, HDL-C, and T-CHOL levels, while robustly elevating hepatic lipogenesis and lipid catabolism. A combination of TFA+MSG resulted in the highest levels of serum leptin, HDL-C, and T-CHOL, while reducing hepatic lipogenic gene expression and increasing markers of infl ammation, lipid storage, DNA damage, and cell cycle impairment. In the adipocytes, Ppargc1a and REDOX gene expression were reduced compared