Vitamin D deficiency in obese rats exacerbates nonalcoholic fatty liver disease and increases hepatic resistin and toll-like receptor activation


  • Potential conflict of interest: Nothing to report.


Childhood obesity is associated with type 2 diabetes mellitus and nonalcoholic fatty liver disease (NAFLD). Recent studies have found associations between vitamin D deficiency (VDD), insulin resistance (IR), and NAFLD among overweight children. To further explore mechanisms mediating these effects, we fed young (age 25 days) Sprague-Dawley rats with a low-fat diet (LFD) alone or with vitamin D depletion (LFD+VDD). A second group of rats was exposed to a Westernized diet (WD: high-fat/high-fructose corn syrup) that is more typically consumed by overweight children, and was either replete (WD) or deficient in vitamin D (WD+VDD). Liver histology was assessed using the nonalcoholic steatohepatitis (NASH) Clinical Research Network (CRN) scoring system and expression of genes involved in inflammatory pathways were measured in liver and visceral adipose tissue after 10 weeks. In VDD groups, 25-OH-vitamin D levels were reduced to 29% (95% confidence interval [CI]: 23%-36%) compared to controls. WD+VDD animals exhibited significantly greater hepatic steatosis compared to LFD groups. Lobular inflammation as well as NAFLD Activity Score (NAS) were higher in WD+VDD versus the WD group (NAS: WD+VDD 3.2 ± 0.47 versus WD 1.50 ± 0.48, P < 0.05). Hepatic messenger RNA (mRNA) levels of Toll-like receptors (TLR)2, TLR4, and TLR9, as well as resistin, interleukins (IL)-1β, IL-4, and IL-6 and oxidative stress marker heme oxygenase (HO)-1, were higher in WD+VDD versus WD animals (P < 0.05). Logistic regression analyses showed significant associations between NAS score and liver mRNA levels of TLRs 2, 4, and 9, endotoxin receptor CD14, as well as peroxisome proliferator activated receptor (PPAR)γ, and HO-1. Conclusion: VDD exacerbates NAFLD through TLR-activation, possibly by way of endotoxin exposure in a WD rat model. In addition it causes IR, higher hepatic resistin gene expression, and up-regulation of hepatic inflammatory and oxidative stress genes. (HEPATOLOGY 2012)

Nonalcoholic fatty liver disease (NAFLD)1 is a hepatic manifestation of the metabolic syndrome (MetS) and affects about 30% of the adult population (70 million adults) in the U.S., and 8% of the population age 2-19 years.2 A subset of patients with NAFLD may develop nonalcoholic steatohepatitis (NASH), a more severe form of the disease associated with hepatic necroinflammation, fibrosis, and may progress to cirrhosis.3

It is increasingly recognized that vitamin D (VitD) plays an important role in autoimmune and inflammatory processes, and there is a growing literature that suggests vitamin D deficiency (VDD) may contribute to the development of insulin resistance (IR), MetS, and NAFLD.4 Recently, up to 55% of adolescents in the U.S. were reported to be VDD with 25(OH)D concentrations <20 ng/mL.5 Obese children are more likely to be sedentary with reduced sunlight exposure and often consume high caloric foods low in mineral and vitamin content.6 These lifestyle factors increase the risk of VDD; furthermore, higher body fat mass as well as limited bioavailability of VitD due to storage in adipose tissue may further increase the risk of VDD among obese children compared to normal weight, active children.7, 8

Recent studies of VDD in humans and animals indicate that VDD also contributes to increased oxidative stress and increased inflammation.9 Manco et al.10 found low levels of 25(OH)D correlated significantly with NAFLD Activity Score (NAS) and fibrosis in children with biopsy-proven NAFLD. However, in a recent large clinical study encompassing data from 1,630 subjects 12-19 years of age using the National Health and Nutrition Examination Survey (2001-2004), VitD status was not found to be independently associated with suspected NAFLD after adjusting for obesity.11

The goal of the current study was to further explore the relationship between diet and VDD in a rodent model of pediatric NAFLD. We mimicked dietary exposures of obese children, utilizing a Westernized diet (WD) that was both high in dietary fat and contained high-fructose corn syrup (HFCS), to generate diet-induced obesity combined with a VitD depleted condition (WD+VDD). Our central hypothesis was that VDD contributes to IR, hepatic necroinflammation, and may result in NASH in diet-induced obesity.


ALK, alkaline phosphatase; GTT, glucose tolerance test; HFCS, high-fructose corn syrup; H&E, hematoxylin-eosin; HO-1, heme oxygenase-1; IκBα, nuclear factor of kappa light polypeptide gene enhancer in B-cells inhibitor α; IKKB, inhibitor of kappa B kinase; IL, interleukin; IR, insulin resistance; ITT, insulin tolerance test; JNK, c-Jun-N-terminal kinase; LFD, low-fat diet; LPS, lipopolysaccharide; LBP, LPS binding protein; MetS, metabolic syndrome; NAFLD, nonalcoholic fatty liver disease; NAS, NAFLD Activity Score; NASH, nonalcoholic steatohepatitis; NF-κB, nuclear factor kappa B; PPARγ, peroxisome proliferator activated receptor γ; SOCS3, suppression of cytokine signaling 3; TIRAP, Toll/interleukin-1 receptor adaptor protein; T2DM, type 2 diabetes mellitus; TLR, Toll-like receptor; TNFα, tumor necrosis factor α; VDD, vitamin D deficiency; VitD, vitamin D; WD, Westernized diet.

Materials and Methods

Animal Model of NAFLD.

Weanling (25 days old at start of diets) male Sprague-Dawley rats (Charles River Laboratories, Wilmington, MA) were randomly assigned test groups and housed in pairs. Rats had a 12-hour light/dark cycle and provided ad libitum access to assigned diet and water. Additionally, WD (HFCS+HFD) groups had continuous access to a separate bottle with HFCS-55 (HFCS-55: 55% fructose, 45% glucose diluted with water to 12.5%, 0.375 kcal/mL solution). Custom rodent diets were purchased from Research Diets (New Brunswick, NJ), with 10% (LFD) or 45% (HFD) total kcal from fat (soybean oil and lard); similar amounts of maltodextrin, no sucrose, and 57% (LFD) or 22% (HFD) total kcal from cornstarch. Vitamin D3 content was either normal (1,000 IU Vitamin D3/4,057 kcal) or depleted (25 IU Vitamin D3/4,057 kcal) adjusting the latter to approximately 30% of controls to mimic VDD in children without reaching levels that may create rickets.12-14 In VDD groups the ultraviolet section of light (290-315 nm) was filtered from the room. This strategy produced a reduction of 25(OH)D levels to 26% of normal after 14 days and reached 29% of normal at 70 days in VDD animals (95% confidence interval [CI]: 23%-36%, mean 25-(OH)D levels were 4.8 ± 1.3 ng/mL in VDD groups and 13.3 ± 0.6 ng/mL in controls (LFD) (see Supporting Fig. 1). Body weight and food intake measures were recorded daily. Blood was collected from trunk (final measures) following a 12-hour fast. After euthanizing the rats, livers and epididymal fat pads were quickly excised, weighed, and an aliquot was snap-frozen. Another liver portion was placed into a 20-mL scintillation vial filled with 4% paraformaldehyde in phosphate-buffered saline (PBS) (USB, Cleveland, OH) for histologic analysis. All procedures performed were approved by the Institutional Animal Care and Use Committee at the Seattle Children's Research Institute and was in accordance with the National Institutes of Health (NIH) Guide for the Care and Use of Laboratory Animals.

Figure 1.

The photomicrographs of H&E stains show no significant steatosis, inflammation, or ballooning degeneration in LFD rats (A); mild macrovesicular steatosis in WD rats (B); macrovesicular steatosis, foci of lobular inflammation, and ballooning degeneration in WD+VDD rats (C,D). Masson trichrome stains show no significant fibrosis in WD rats (E); but incipient perivenular/pericellular fibrosis in WD+VDD rats (F).

Histological Examination.

Sections of formalin-fixed liver were stained with hematoxylin-eosin (H&E) and a second set by Masson's trichrome. Histological scoring was performed by a blinded hepatopathologist (M.Y.) for steatosis, lobular inflammation, hepatocellular ballooning, and fibrosis (score 0 to 4) using the NASH Clinical Research Network (CRN) scoring system.15, 16 Scores for steatosis (score 0 to 3), lobular inflammation (score 0 to 3), and ballooning (score 0 to 2), were also summed to produce the NAS, thus ranging from 0 to 8.

Metabolic Testing and Hormone Measurements.

During the last 8 days of dietary exposures, intraperitoneal insulin tolerance test (ITT) (1 U/kg, Humulin, Lilly) and glucose tolerance (GTT) (1.5 g glucose/kg) tests were performed by way of intraperitoneal injection after food deprivation for 12 hours. Rats were allowed to recover for at least 4 days between tests. Blood samples were obtained by way of a small tail nick at −15, 0, 15, 30, 45, and 60 minutes for glucose levels assessed using a hand-held glucometer (LifeScan OneTouch Ultra 2, Milpitas, CA) in both tests. Area under the curve (AUC) glucose 0-60 minutes (GTT) and inverse AUC % change from basal glucose 0-60 minutes (ITT) were calculated as published.17 For hormones and cytokines, blood was drawn into prechilled EDTA tubes, centrifuged immediately at 4°C, aliquoted, and stored at −80° C. Immunoreactive hormones and cytokines were determined by enzyme-linked immunosorbent assay (ELISA) (Plasma insulin: Crystal Chem, Chicago, IL, adiponectin: Millipore, Billerica, CA; serum lipopolysaccharide [LPS]-binding protein [LBP]: Cell Sciences, Canton, MA), or on a Luminex 200 instrument (Luminex, Austin, TX) by multiplex immunoassay (Plasma IL-1β, IL-6, tumor necrosis factor (TNF)-α: Millipore). For all measurements, intraassay coefficients of variation were below 8%, and interassay coefficients of variation below 12%. Levels of calcium, alkaline phosphatase (ALK), cholesterol, and triglycerides were determined in plasma on a Modular P chemistry analyzer (Roche Diagnostics, Germany) at the Northwest Lipid Research Laboratories, Seattle, WA. Furthermore, 25(OH)D levels were measured using a well-established liquid chromatography/tandem mass spectrometry (LC-MS/MS) method,18 which allows detection without crossreactivity with other VitD metabolites, in contrast to immunoassays.19 The lower limit of quantification was 1 ng/mL, intraassay coefficient of variation was below 5%, and interassay coefficients of variation below 10% for this method.18

Gene Expression Studies.

Total RNA was extracted and purified using RNeasy kit (Qiagen, Venlo, Netherlands) for liver tissue and RNeasy Lipid Tissue kit (Qiagen) for adipose tissue. RNA quality was assessed using a NanoDrop 1000 (Thermo Fisher, Waltham, MA). cDNA was synthesized using iScript complementary DNA (cDNA) synthesis kit (Bio-Rad, Hercules, CA). Real-time quantitative reverse-transcription polymerase chain reaction (qRT-PCR) reactions were performed on an ABI Prism 7300 (Applied Biosystems, Foster City, CA) using iTaq SYBR Green Supermix with ROX (Bio-Rad). Nontemplate controls were incorporated into each PCR run. Specific messenger RNA (mRNA) levels of all genes of interest including inflammatory and endotoxin pathways (cytokines, TLRs, Toll/interleukin-1 receptor adaptor protein [TIRAP], CD14, LBP), as well as genes determining insulin sensitivity (resistin, peroxisome proliferator activated receptor γ [PPARγ]), were normalized to a housekeeping gene (GAPDH) and expressed as changes normalized to controls (LFD). See Table 1 for all primers of qRT-PCR.

Table 1. Primers Used for Real-Time PCR Studies
GeneSequence (5′ - 3′)LengthGenBank (Rattus norvegicus)

Statistical Analysis.

Key outcome variables were compared between study groups using Student's t tests for continuous variables, or, for contrasts involving more than two groups, using analysis of variance (ANOVA) modeling. For analyses of different dietary effects, comparisons were made by two-way ANOVAs using the WD and the VDD factors of different groups. In addition, Student's t test was used for comparisons of single dietary interventions. Data were analyzed using Excel (Microsoft, Renton, WA) and Prism5 software (GraphPad, La Jolla, CA), and are presented as mean ± standard error of the mean (SEM), if not otherwise stated. Categorical variables including histological features like steatosis, lobular inflammation, and hepatocellular ballooning were analyzed using Fisher's exact test (STATA 9.0, College Station, TX). Ordinal logistic regression analysis was performed to determine the relationship between NAS and gene expression of different genes (STATA). In all instances, P < 0.05 was considered significant.


Effects on Body and Tissue Weights, and Food Intake.

Weight gain, total caloric intake, and Lee index, an adiposity index that highly correlates with total body fat,20 were highest in the WD+VDD group (Table 2). WD and WD+VDD rats showed higher visceral adiposity assessed by gonadal fat pad as well as significantly higher liver weights compared to LFD groups, but no significant differences were found between WD and WD+VDD rats (Table 2).

Table 2. Characteristics in Dietary Groups During And At End Of 10 Week Of Dietary Exposures
Vitamin D StatusLFD versus WDInteraction
  • Data are mean ± SEM. NS, nonsignificant.

  • *

    P < 0.05 versus LFD.

  • P < 0.05 versus WD.

  • P < 0.05 versus LFD+VDD.

Body weight gain (g, 60 d)409 ± 12415 ± 11433 ± 19447 ± 8*NS<0.05NS
Total caloric intake (kcal)4790.8 ± 127.24949.2 ± 47.05487.3 ± 58.6*6089.5 ± 249.1*,<0.01<0.0001NS
Lee index (60 d)0.309 ± 0.0010.308 ± 0.0010.310 ± 0.0020.314 ± 0.002*NSNS (P = 0.0524)NS
Gonadal fat weight (g)6.8 ± 0.67.7 ± 0.511.0 ± 0.9*10.7 ± 1.1*NS<0.0001NS
Liver weight (g)13.7 ± 0.714.1 ± 0.518.1 ± 0.7*17.9 ± 1.1*NS<0.0001NS

Metabolic Tests.

GTT showed that WD groups had higher glucose AUC than LFD animals, whereas during ITT, glucose reduction demonstrated by inverse AUC % basal glucose was stronger in VitD replete than VDD groups (Table 3), indicating IR in VDD groups.

Table 3. Biomarker and Dynamic Test Results
Vitamin D StatusLFD versus WDInteraction
  • Data are mean ± SEM. NS, nonsignificant.

  • *

    P < 0.05 versus LFD.

Cholesterol (mg/dL)73 ± 676 ± 656 ± 3*60 ± 7NS<0.01NS
Triglycerides (mg/dL)145 ± 18162 ± 20143 ± 17223 ± 28<0.05NSNS
ALK (U/L)96.8 ± 6.1110.2 ± 7.2117.4 ± 7.9*106.5 ± 10.4NSNSNS
Calcium (mg/dL)10.3 ± 0.210.3 ± 0.110.5 ± 0.110.3 ± 0.1NSNSNS
GTT AUC glucose 0-60 min11.8 ± 0.712.8 ± 0.613.9 ± 0.6*13.2 ± 0.3NS<0.05NS
ITT inverse AUC % basal (0-60 min)20.8 ± 2.214.1 ± 1.5*16.7 ± 1.112.7 ± 2.8*<0.05NSNS
Serum Adiponectin (μg/ml)31.9 ± 2.734.1 ± 3.528.5 ± 1.431.2 ± 2.7NSNSNS
Serum Leptin (ρg/ml)2783 ± 3274506 ± 603*7477 ± 1251*5113 ± 452*<0.001NS<0.01

Serum ALK was higher in WD rats, although serum calcium was comparable in all four groups without evidence of rickets. Serum alanine aminotransferase (ALT) levels were slightly higher in WD+VDD rats compared to all other groups (LFD 33.8 ± 1.2, LFD+VDD 33.4 ± 1.3, WD 33.8 ± 1.3, WD+VDD 37.7 ± 1.7 U/L; P = 0.042, one-sided WD+VDD versus WD). Serum cholesterol levels were slightly lower in WD groups, whereas higher triglyceride levels were measured in VDD groups compared to VitD replete animals. Adiponectin levels were comparable, showing no differences among the groups; however, LFD+VDD, WD and WD+VDD animals had higher leptin than LFD animals (Table 3). There were no significant differences in plasma IL-1β serum levels (LFD 14.0 ± 6.6 pg/mL, LFD+VDD 21.6 ± 11.1, WD 34.7 ± 26.2, WD+VDD 61.9 ± 47 pg/mL). IL-6 plasma levels were below the detection level of the assay (<5 pg/mL) in LFD groups but were detectable in WD (143 ± 138 pg/mL) and WD+VDD (90 ± 54 pg/mL) (t test, NS). Serum LBP levels were significantly higher in WD than in LFD groups (LFD: 878 ± 31, LFD+VDD 936 ± 21, WD: 1,373 ± 209, WD+VDD 1,493 ± 242 ng/mL, P < 0.05 WD versus LFD groups).

Liver Histology.

Hepatic steatosis and lobular inflammation were lower in LFD animals and this was unaffected by VDD. There was no significant difference in hepatic steatosis and lobular inflammation between LFD and WD animals, in part due to variability in responses. In WD+VDD animals, we found increased hepatic steatosis and lobular inflammation. Lobular inflammation and NAS were significantly higher in WD+VDD relative to all other groups. Moreover, there was a trend for higher ballooning score in WD+VDD rats compared to the other groups (Table 4, Fig. 1A-D). After the 10 weeks of dietary exposures, there was no significant fibrosis that fulfilled the NASH CRN criteria in zones 1 or 3 (Fig. 1E); however, incipient perivenular/pericellular fibrosis was seen in 3 WD+VDD rats (Fig. 1F).

Table 4. Characteristics in Different Dietary Groups
  • Data are mean ± SEM. NS, nonsignificant.

  • *

    P < 0.05 versus LFD.

  • P < 0.05 versus WD.

  • P < 0.05 versus LFD+VDD.

  • §

    P < 0.05 (1 sided) versus LFD.

Liver fat score0.25 ± 0.160.30 ± 0.151.10 ± 0.31§1.30 ± 0.21*
Lobular inflammation score0.63 ± 0.180.20 ± 0.130.30 ± 0.151.40 ± 0.26
Ballooning score0.00 ± 0.000.00 ± 0.000.10 ± 0.100.40 ± 0.26
NAFLD activity score0.88 ± 0.300.50 ± 0.221.50 ± 0.483.20 ± 0.47*

Gene Expression Studies.

Gene expression studies were executed to determine whether increased NASH was accompanied by increased inflammation. VDD had a greater effect on inflammatory genes in the liver compared to adipose tissue. Specifically, liver resistin mRNA levels were higher in VDD groups compared to VitD replete animals (Fig. 2A), but no differences between groups were seen in adipose tissue (data not shown). IL-4 mRNA levels were higher in VDD than in VitD replete groups, with greater effect exhibited in liver than adipose tissue (Fig. 2B; Supporting Fig. 2A). IL-6 mRNA levels were higher in VDD than in VitD replete groups in the liver (Fig. 2C), and higher in WD+VDD compared to all other groups in adipose tissue (Supporting Fig. 2B). Further examination of hepatic signaling pathways involved in inflammation and oxidative stress in liver revealed activation of IL-1β in both WD and VDD independently (Fig. 2D), and activation of IL-10 by WD (Fig. 2E), whereas in adipose tissue IL-1β did not differ significantly between the groups (data not shown). Liver mRNA levels for HO-1, a marker for oxidative stress, were significantly higher in WD+VDD versus WD rats (Fig. 2F). Analysis of the TLR signaling pathways showed activation of TLR4 and LBP in both WD and VDD independently (Fig. 3A,B), activation of CD14, TLR2, TIRAP, and TLR9 by WD with further increment by VDD (Fig. 3C-F), activation of IκBα by VDD and WD, and activation of TNFα in LFD+VDD versus LFD (Supporting Fig. 3A,B). Additionally, mRNA levels for PPARγ, a regulator of inflammatory responses induced by hepatic steatosis,21 were higher in WD with further increment by VDD (Supporting Fig. 3C). Nuclear factor kappa B (NF-κB) mRNA levels were found to have no significant differences between the groups, although there was a 1.54-fold increase in WD+VDD versus LFD (P = 0.17).

Figure 2.

Gene expression of resistin (A), IL-4 (B), IL-6 (C), IL-1β (D), IL-10 (E), and HO-1(F) in liver. Mean (±SEM) values of VitD replete (empty bars) and VDD (black bars) animals. Analyses were performed by two-way ANOVA for multiple groups (boxed text) as well as by pairwise comparisons for single dietary interventions. Combined and separate effects of WD and VDD are shown. (*)P < 0.05 one-sided. *P < 0.05 two-sided, **P < 0.01.

Figure 3.

TLR-dependent signaling showing results for TLR4 (A), LPS-dependent glycoproteins LBP (B) and CD14 (C), TLR2 (D) and TLR signaling molecule TIRAP (E), as well as TLR9 (F). Mean (±SEM) values of VitD replete (empty bars) and VDD (black bars) animals. Analyses were performed by two-way ANOVA for multiple groups (boxed text) as well as by pairwise comparisons. Combined and separate effects of WD and VDD are shown. (*)P < 0.05 one-sided. *P < 0.05 two-sided, **P < 0.01.

Relationship Between NAFLD and Gene Expression Profiles.

Due to some variability between NASH scores/NAS within treatment groups, regression analyses were performed using NAS as the dependent variable and liver mRNA levels as the independent variables focused on (1) TLR/TNFα/NF-κB/IκBα signaling, (2) PPARγ, a parameter induced by hepatic steatosis, and (3) HO-1, a marker of oxidative stress. NAS was significantly associated with liver mRNA expression for all three TLRs, LBP, CD14, but also with TNFα, IL-1β, IκBα, PPARγ, and HO-1, with and without adjustment for Lee adiposity index (Table 5).

Table 5. Correlation Between NAS and Expression of Different Genes in the Liver
 UnadjustedAdjusted for Lee Adiposity Index
Coef.95% CIPCoef.95% CIP
TLR23.5321.808/5.256< 0.0013.8081.852/5.764< 0.001
TLR93.4031.6695.137< 0.0013.6001.675/5.525< 0.001
TIRAP2.4261.106/3.745< 0.0012.4221.067/3.776< 0.001


In the current study, we found that IR, NAFLD, and hepatic necroinflammation were most pronounced in VDD rats fed a WD; these findings have implications for human NAFLD and also provide a novel model for experimental NASH. Although other dietary animal models for NAFLD are available,22 the current study utilizes both (1) dietary manipulations consistent with contemporary diets, i.e., HFD and HFCS which have been shown to be risk factors for the development of NAFLD,23 and (2) lifestyle trends, i.e., less time spent outdoors and therefore less solar exposure. To our knowledge, such a rodent model has not been previously utilized. Animals were subjected to the diet and VDD regimen just after weaning and continued for 10 weeks, approximately equal to adolescence and early adulthood.24 VitD levels were reduced to about 30% of normal, similar to findings in obese children.5 WD had a major effect on gonadal fat and liver weight as well as glucose tolerance, whereas VDD strongly influenced serum leptin, IR, and hepatic mRNA levels for resistin, IL-4, and IL-6. However, most other parameters were influenced by the combination of VDD and WD exposures, demonstrating that multiple environmental factors are involved in NASH pathogenesis. In the current study features of IR, NAFLD, and hepatic necroinflammation were particularly apparent in WD+VDD, despite only a slight increase in ALT levels, similar to findings in humans.25

Our results also demonstrate IR in VDD animals and IR is currently thought to play an early role in the progression from NAFLD to NASH.26 VitD has been shown to improve B-cell function,27 whereas low VitD levels are associated with IR.8 In a rat NASH model increases of VitD by way of phototherapy were shown to decrease hepatocyte apoptosis, inflammation, fibrosis, and IR.19 Furthermore, VDD may contribute to hepatic necroinflammation and fibrogenesis in patients with chronic hepatitis C and children with biopsy-proven NAFLD.10

Progression from NAFLD to NASH could be the consequence of a failure of antilipotoxic protection, leading to inflammation and fibrosis.28, 29 VitD can have diverse effects on the immune system, particularly on the function of monocytes, macrophages, and T cells.30 Calcitriol, a hormonally active form of VitD, has been observed to inhibit the production of TNFα, IL-1β, and IL-6 from isolated, LPS-stimulated peripheral mononuclear cells.31 Additionally, an increase in VitD during summer months reduced the levels of TNFα, IL-1β, and IL-6 compared to winter months.32 It is possible that lack of VitD in our VDD animals led to increases in IL-6 and IL-1β mRNA liver expression.

The proinflammatory molecules TNFα, IL-6, and IL-1β are secreted by adipocytes and infiltrated macrophages and cause systemic inflammation, which has been suggested as a major mechanism leading to IR and hepatic steatosis in obesity.29 In the current study, markers of inflammation assessed by RNA levels were more significantly stimulated in the liver than in adipose tissue by VDD, indicating that VDD might affect the liver before changes occur in adipose tissue.

The gut-liver axis is important in the development of NAFLD, as bacterial products, such as LPS, are delivered to the liver through the portal vein.33 TLRs are pattern recognition receptors that play a central role in host cell recognition and responses to bacterial and viral pathogens.34 LPS, the bacterial endotoxin, binds to two glycoproteins, LBP and endotoxin receptor CD14, which both interact with a transmembrane TLR responsible for signal transduction.35 Among 13 TLRs identified in mammals, TLR2, TLR4, and TLR9 play a role in NAFLD pathogenesis.33 TLR4 and TLR2 are involved in LPS signaling,35 and all three receptors are involved in bacterial recognition.34 Our data show that hepatic LBP, CD14, as well as TIRAP (an adaptor protein in the TLR signaling pathway36) mRNA levels were increased as a result of WD and exacerbated by VDD. Similarly, gut-derived endotoxinemia associated with dietary fructose intake has been linked to NAFLD/NASH in humans37 and animal studies, by way of activation of TLR4, NFκB, and TNFα in the liver.29 We speculate that VDD may contribute to NAFLD by increasing endoxin exposure to the liver. Furthermore, TLR9 up-regulation in the current study might also be related to VDD, as it has been shown in human monocytes that VitD down-regulates TLR9-induced IL-6 production,38 which is further supported by increased expression of IL-6 in VDD groups. Despite increased TLR4 expression by VDD, we saw only modest increases of TNFα hepatic mRNA levels in VDD animals. It is possible that high leptin levels seen in VDD animals and/or increased IL-10 expression seen in WD animals (Table 3, Fig. 2E) inhibited NFκB by way of suppression of cytokine signaling (SOCS)3, therefore also suppressing TNFα expression.39, 40

Resistin was markedly increased in the liver of both VDD groups in this study (Fig. 2A), which may contribute to signaling changes in the liver as a result of VDD. Resistin was initially identified as an adipokine released from adipose tissue and thought to play a role in IR41; however, it has recently found to be more abundant in Kupffer cells and may play a role in NAFLD.42 Resistin is known to activate c-Jun-N-terminal kinase (JNK) and NF-κB pathways; the latter by way of IKKB (inhibitor of kappa B kinase beta).43 IκBα expression, a substrate of IKKB activation, was elevated in response to diet and VDD, especially in WD+VDD animals (Supporting Fig. 3A), potentially as a compensatory mechanism. IKKB promotes IR and is a central coordinator of the inflammatory response, specifically through an IKKB/NF-κB pathway, which may also be activated through TLR4.44 This suggests that hepatic resistin is a key mediator to the development of IR and the upregulation of inflammatory markers in this study. Finally, hepatic expression of HO-1, a marker of oxidative stress that may be involved in development of fibrosis,45 was increased in WD+VDD animals (Fig. 2F), possibly by way of IL-10 and LPS.46

In summary, our results suggest that the development and progression of NAFLD by WD is exacerbated by VDD. We suggest that the mechanism is through the activation of TLR2 and TLR4 by way of CD14/LBP, and stimulation of downstream inflammatory signaling molecules leading to steatosis and inflammation. In addition, we propose that VDD leads to activation of hepatic resistin, which likely contributes to the hepatic signaling changes and development of IR. The current study has identified novel dietary factors that may contribute to the development of NAFLD in overweight children. The impact of this study is clear, as it demonstrates a role for VDD in NAFLD and IR. Additional studies are necessary to test whether VitD supplementation reduces IR and shows histologic improvement of NAFLD in clinical and experimental VDD.