Background
Obesity and associated metabolic pathologies are the most common risk factors for a variety of diseases leading to mortality, including type-2 diabetes [
1,
2], hypertension, cardiovascular disease [
3,
4], and cancer [
5]. Obesity is also a risk factor for diseases that cause serious morbidity such as osteoarthritis and sleep apnea [
6,
7]. Moreover, obese individuals who consume generous amounts of alcohol may be prone to the development of metabolic syndrome, since alcohol can contribute approximately 6% to 10% of the total calories consumed. However, epidemiological data show that moderate alcohol use has several beneficial effects, which include decreasing the risk of type-2 diabetes [
2,
8]. Alcohol consumption may decrease the risk of type-2 diabetes by promoting insulin sensitivity [
9]. It remains to be determined if alcohol consumption improves insulin sensitivity for all body weight phenotypes (e.g., lean, overweight, or obese). Thus, it is not clear if the effects of alcohol consumption on insulin sensitivity are modified by body weight or body fat levels.
Excess calorie consumption can lead to adipocyte hypertrophy and hyperplasia which in turn increases the production and secretion of adipokines (e.g., leptin, adiponectin) and inflammation mediators (e.g., Interleukin 6) by the fat cells [
10]. Adipokines and inflammation factors play key roles in the development of insulin resistance [
10]. Currently, there are several theories to explain how obesity causes insulin resistance. One theory is that in obesity, the excess adipose tissue produces high levels of adipokines which promote insulin resistance, on the other hand, the expression of factors that promote insulin sensitivity, such as adiponectin are decreased [
11,
12]. Adiponectin, secreted mainly by adipose tissue, is closely associated with visceral fat rather than subcutaneous fat, and is found at lower concentrations in conditions often associated with insulin resistance such as type-2 diabetes, cardiovascular disease, and metabolic syndrome [
13]. The promotion of insulin sensitivity by alcohol consumption has also been associated with an increase in systemic adiponectin levels [
12].
Insulin resistance is also shown to be associated with the accumulation of macrophages in white adipose tissue of obese individuals [
14]. Factors secreted by fat cells, such as MCP1, initiate the migration of monocytes to white adipose tissue where they develop into activated macrophages [
15]. These activated macrophages secrete factors that induce a persistent inflammatory response. The excess production of adipokines and inflammation factors by the adipocytes and macrophages can lead to the development of insulin resistance in key tissues that regulate glucose metabolism, such as skeletal muscle and adipose tissue [
16,
17]. A second proposed hypothesis suggests that excess fat leads to the accumulation of lipids in non-adipose tissue such as liver and muscle, causing lipotoxicity and eventual insulin resistance [
18]. A third hypothesis proposes obesity-dependent insulin resistance occurs through the increased production of reactive oxygen species (ROS) [
19]. ROS themselves can induce insulin resistance or adversely affect the production of insulin by the pancreas [
20,
21].
To better understand the relationship between body fat and insulin resistance, we measured the expression of genes associated with energy homeostasis and insulin regulation using a gene array containing 384 genes linked to obesity and insulin resistance. To this end, we obtained white adipose tissue from mice fed a calorie restricted diet (CR), a low fat diet (LF Control), or a high fat diet (HF), consuming either water or 20% ethanol in the drinking water. Previously, we showed CR, LF Control, and HF diets induced similar body fat phenotypes in mice to those found in individuals considered lean (BMI less than 25), overweight (BMI between 25 and 30), and obese (BMI higher than 30), respectively [
22]. Therefore, the present study design allowed us to determine the effects of alcohol consumption on genes linked to obesity and insulin resistance on lean (CR), Control (LF) and obese (HF) mice. We show that obesity induces insulin resistance, while both calorie restriction and alcohol consumption promote insulin sensitivity. Furthermore, calorie restriction and obesity have drastic opposing effects on pro- and anti-inflammatory genes as well as genes involved in adipose tissue energy metabolism such as Lpin2, Kcnj11 and Rbp4. Moreover, alcohol reduces the expression of a different spectrum of inflammatory factors than those regulated by body weight and it significantly upregulated the expression of anti-inflammatory genes. Our current study contributes to a better understanding of how diet modification and alcohol consumption affect insulin sensitivity.
Materials and methods
Study Design
Male C57BL/6 mice (The Jackson Laboratory, Bar Harbor, Maine) were housed according to NIH guidelines (National Research Council, 1996) in the Animal Resources Center at the University of Texas at Austin (UT). Animal care was provided in accordance with the procedures outlined in "Guide for the Care and Use of Laboratory Animals" (NIH Publication No. 86-23, 1985). All animal procedures were approved by UT's Institutional Animal Care and Use Committee. Lean, normal, and obese body weights were generated by manipulating caloric intake. At 6 week of age, mice were randomized to receive the following diet treatments for 20 weeks: 1) 30% calorie restricted (CR) regimen (27% protein, 54% carbohydrate, 6% fat; Research Diets, Inc. #D12492) to generate lean mice, 2) low fat diet regimen (19% protein, 67% carbohydrate, 4% fat; Research Diets, Inc. #D03020702) to generate LF normal mice, and 3) high fat diet regimen (26% protein, 26% carbohydrates, 35% fat; Research Diets Inc. #D12450B) to generate obese mice. All mice consumed either water or 20% w/v alcohol in the drinking water throughout the study. Each group was composed of 15 mice.
StellARray
Total RNA was extracted from WAT from six randomly chosen mice from each group (see qPCR method for details). Quantitative PCR data was collected using the Mouse Diabesity 384 StellARray™ qPCR Array from Bar Harbor Biotechnology (Lonza Cat. number 00188200). GeneSieve System of BHB (Bar Harbor Biotechnology, Trenton, ME) was used for selection of gene candidates included in the array analysis. Briefly, genes were selected by relevance to a biological process. First, databases containing gene-centric biological information were merged. According to this information, over 16 million abstracts were searched for genes involved in a related process. Then genes were ranked by multiple fields, such as quantity and quality of literature references. This was followed by filtering or "sieving" thousands of genes using, for example, canonical signaling pathways, microarray data, gene ontology data, downstream targets of transcription factors, etc. Finally, a list of the most pertinent genes was created and included in the array. Six biological replicates were randomly chosen from each diet group, and total RNA was extracted from epididymal fat tissue. Each StellARray qPCR Array well was loaded with 10 microliters of sample-specific, SYBR Green master mix containing a chemically modified hot-start Taq polymerase (Applied Biosystems, Inc.). The array was heat-sealed and run on a 7900HT Sequence Detection System (Applied Biosystems, Inc.) using default cycling parameters for 40 cycles (1 cycle of 50°C for 2 minutes, 1 cycle of 95°C for 10 minutes, 40 cycles of 95°C for 15 seconds, and 60°C for 1 minute). Fluorescence data was acquired during the 60°C anneal/extension plateau. Post-run data collection involved the setting of a common threshold across all arrays within an experiment, exportation and collation of the Ct values, and analysis via Global Pattern Recognition™ (GPR) software. GPR was designed to take advantage of biological replicates to extract significant changes in gene expression, thus providing a novel alternative to compare the change of expression of a gene normalized to every other gene in the array. By comparing the expression of each gene to every other gene in the array, a global pattern was established, and significant changes were identified and ranked [
23].
StellARray Data Validation by qPCR
The expression of leptin and adiponectin were examined by qRT-PCR. Immediately after removal from the animal, white adipose tissue (WAT) was snap frozen and stored in liquid nitrogen. Total RNA was extracted using RNeasy Lipid Tissue Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. The amount and quality of RNA was verified by measuring the absorbance at 260 and 280 nm. Reverse transcription of RNA was performed with the High Capacity cDNA Reverse Transcription Kit (Applied Biosystems, Foster City, CA). Primer pairs were designed using Primer 3 software [
24]: Leptin forward: 5'-TCTCCGAGACCTCCTCCATCT-3' and reverse: 5'-TTCCAGGACGCCATCCAG-3'; Adiponectin forward: 5'-AAGGACAAGGCCGTTCTCT-3' and reverse: 5'AGAGTCGTTGACGTTA TCTGCATA-3'. Real-time PCR was performed with a SYBR GreenER qPCR kit (Invitrogen, Carlsbad, CA) and a Mastercycler ep Realplex real-time PCR thermocycler (Eppendorf, Hamburg, Germany). The relative expression level of target genes was normalized to the endogenous reference gene 18 s rRNA. Amplification specificity was confirmed by melting curve analysis.
Serum leptin levels
Leptin levels were determined with a Luminex-based bead array method using a LINCOplex simultaneous multi-analyte detection system (Linco Research Inc., St. Charles, MO) according to the manufacturer's instructions.
Body fat levels
Percent body fat levels were measured with dual energy x-ray absorptiometry (DXA) using a GE Lunar Piximus II densitometer (Madison, WI).
Statistics
For glucose and insulin tolerance tests, statistical analysis of individual diet curves within water and alcohol treatment was performed with Student's unpaired t tests of areas-under-the-curve. Statistical comparisons between water and alcohol groups and for qPCR results were performed using two-way ANOVA. A p value of <0.05 was considered statistically significant. All statistical analyses were performed with SPSS software (Chicago, IL). Array data analysis and statistical significance were calculated using the Global Pattern Recognition™ (GPR) software mentioned earlier.
Discussion
We and others have previously shown that obesity induces insulin resistance. In this study, we confirmed this effect of obesity on insulin resistance and provide further evidence that CR and alcohol consumption increase insulin sensitivity. To understand the mechanism behind this phenomenon, we screened the expression of genes related to the regulation of energy homeostasis and insulin regulation in order to determine pathways by which body weight and alcohol consumption affect insulin sensitivity. Our results show that alcohol increased leptin levels in adipose tissue and increased insulin sensitivity in LF mice simultaneously. Leptin is an adipocyte hormone that acts to decrease food intake and increase energy expenditure [
44]. Alcohol increased leptin levels without affecting body weight and percent body fat, suggesting that alcohol may increase insulin sensitivity via leptin, which may increase glucose catabolism in alcohol consuming mice. Alternatively, alcohol may increase insulin sensitivity by increasing adiponectin levels, adiponectin has been shown to increase insulin sensitivity [
12], and alcohol consuming mice had higher levels of adiponectin (Fig.
3A), suggesting that alcohol consumption may promote insulin sensitivity via adiponectin.
To determine whether insulin sensitivity and glucose homeostasis are modified by alcohol, we performed the GTT and the ITT on mice. The GTT measures how quickly injected glucose is cleared from the blood. Insulin is a hormone secreted by the β cells within the pancreas as a response to blood glucose alterations. Result show that the GTT AUC was not affected by alcohol suggesting that alcohol in the drinking water did not affect insulin secretion by the β cells within the pancreas. To determine if alcohol consumption promoted insulin sensitivity, we performed the ITT on mice drinking water or 20% alcohol. The ITT determines how quickly endogenous glucose is cleared from the blood as a response to insulin administration. Results show that the ITT AUC values were significantly reduced in alcohol consuming mice suggesting that alcohol consumption affects insulin sensitivity by affecting the insulin signaling pathway.
Herein, we selectively discussed genes with the most statistically significant changes and those associated with the insulin signaling pathway, energy homeostasis, and inflammatory response. Other genes affected by body weight or alcohol consumption are listed in Tables
1,
2, and
3 for reference. Unlike most previous studies, we included lean mice in addition to the normal and obese body weights. The significance of including body weight at both ends of the spectrum, lean and obese, is to show not only the deleterious effects of obesity in related metabolic pathways but also the possible protective effects of lean body weight. We show that the CR increased the expression levels of anti-inflammatory genes Il-10 and Adrbk1 (Fig.
3B and
3D) and energy homeostasis-related genes such as Ptgds (Fig.
6A), Lpin2 (Fig.
7A), Angptl6 (Fig.
7D), Kcnj11 (Fig.
6B), and Dusp9 (Fig.
8A) but were not affected by obesity in HF mice. These results provide further insight into pathways with which to study the protective effects of these genes. Another purpose of the study was to investigate the interaction between alcohol and body weight in regulating insulin sensitivity.
We show that in white adipose tissue of obese mice, pro-inflammatory factors including Il-6, Il-1α, and macrophage marker Cd68 (Tables
1 and
2) were elevated. Conversely, CR had an opposite effect on most inflammatory factors such as Il-6 and Il-1α, except for Cd68, which was not affected by CR. This suggested that that HF diet up-regulated macrophages and macrophage-mediated secretion of inflammatory factors Il-6 and Il-1α, while CR protected against inflammation through reduced macrophage activity. We show that alcohol regulated a different profile of pro-inflammatory factors compared to body weight. These factors include Retnlb and Mif. Also, alcohol consumption increased anti-inflammatory factors Il-10 and Adrbk1, which were also increased by CR. This suggested that alcohol and lean body weight might affect the same anti-inflammatory factors to stimulate insulin sensitivity.
Diet intervention may also affect insulin sensitivity by affecting other anti-inflammatory factors such as L-Ptgds. Ragolia et al. showed that
L-Ptgds knockout mice become glucose intolerant and insulin resistant [
34]. Moreover,
H-Ptgds (Hemopoietic type prostaglandin D2 synthase) null mice display a severe inflammatory response compared to wild type mice [
45]. Our results show that L-Ptgds expression in adipose tissue was significantly up-regulated by more than three-fold in CR mice (Fig.
3A). We speculate that increased expression of L-Ptgds played a protective role in insulin sensitivity in CR mice. The anti-inflammatory effects of H-Ptgds are mediated by the cyclopentenone prostaglandins (PGs), including 15-deoxy-Delta(12,14)-PGJ2 (15d-PGJ2), a dehydrated product of PGD2 [
42]. Since the diverse effects of L-Ptgds and H-Ptgds result from tissue-specificity, instead of enzymatic function [
34,
45], we do not eliminate the possibility that in adipose tissue, L-Ptgds triggers an anti-inflammatory effect by producing cyclopentenon PGs. Further studies characterizing Ptgds deletion or over-expression in adipose tissue would be helpful in confirming the relationship between Ptgds expression and inflammation. Besides interfering with inflammatory pathways, such as the NF-κB pathway [
45], it is also speculated that L-Ptgds may suppress inflammation by inhibiting monocytes in adipose tissue of CR mice. In fact, Il-6 and Il-1α levels were reduced in CR mice. Interestingly, alcohol increased the level of Ptgds even though it did not have an additive effect on CR alcohol-consuming mice. This further indicated that CR and alcohol affect overlapping factors that regulate inflammation and immune responses.
We further explored genes associated with energy homeostasis and insulin signaling. Lipin family members play important roles in lipid and glucose metabolism, adipocyte development, and inflammation [
46].
Lpin1 was the first gene identified in this family. Lpin1 mRNA is expressed at high levels in adipose tissue and is induced during differentiation of 3T3-L1 pre-adipocytes [
46]. A second member of this family is Lpin2. Little is known about the third member of the lipin family, Lpin3. Here, we found that Lpin1 was not affected by either body weight or alcohol consumption. Interestingly, although they are family members, Lpin2 and Lpin3 had contrasting expression patterns in WAT from CR and HF mice. HF diet suppressed Lpin2 and increased Lpin3 expression. Alcohol increased Lpin2 and decreased Lpin3 expression in the LF mice compared with Control. This suggests opposing roles of Lpin2 and Lpin3 in regulating energy homeostasis and insulin sensitivity. Previous findings show a homozygous mutation of the human
Lpin2 gene leads to inflammation of the bone and skin [
46]. Combined with our findings, Lpin2 may contribute to increased insulin sensitivity by protecting against inflammatory reactions. Based on this finding, further studies on the relations of Lpin2 and Lpin3 expression levels versus insulin sensitivity regulation, as well as inflammatory factor profiles, will help better understand the role the lipin family plays in body weight and alcohol-regulated insulin sensitivity.
Another protein family that adds to the increasingly complex network of hormonal regulation of glucose and lipid homeostasis is the Angptl protein family. Among the six members, Agnptl6 has been shown to promote angiogenesis and wound closure led by keratinocyte proliferation [
47]. Angptl6 has also been shown to have metabolic functions. Others have show that loss of Angptl6 in early development is lethal and those mice that survive developed excess adiposity, adipocyte hyperplasia, and increased circulating and tissue deposition of free fatty acids [
48]. In support of the above findings, our data showed that in adipose tissue, Angptl6 mRNA was induced by CR intervention and down-regulated by obesity in HF mice. Moreover, previously we showed that CR mice healed inflicted wounds faster and obesity inhibited wound healing in mice, thus suggesting that factors like Angpt16 may promote wound healing in CR mice but in obese mice the decreased Angpt16 may impair wound healing [
49]. Similar observations as these have been noticed in humans studies [
50].
We investigated signaling pathways related to insulin function. The JNK pathway is one of the signaling pathways linked to insulin signaling [
51,
52]. JNK kinase can inhibit the signaling pathway by phosphorylating the insulin receptor substrate-1 (IRS-1) on serine residues [
53]. The phosphatase Dusp-9 is a protein that inhibits the activity of JNK [
42]. Our results show that CR increases the expression of Dusp-9 (Table
1, Fig
8). Thus, high levels of Dusp-9 may increase insulin sensitivity by inhibiting factors such as JNK. In fact, CR mice are extremely insulin sensitive as indicated by the rapid clearance of exogenous glucose in the GTT assay. Moreover, injection of exogenous insulin to CR mice leads to severe hypoglycemia which can result in death. For this reason, CR mice were not used in the ITT studies. In contrast, decreased Dusp-9 can lead to insulin resistance. In effect, Dusp-9 levels were decreased in insulin-resistant obese mice consuming the HF diet. Further studies on the phosphorylation of JNK and insulin pathway signaling factors will provide better insight into alcohol- and CR-induced insulin sensitivity and HF-associated insulin resistance.
On the other hand, phosphatases that impair the insulin pathway may be elevated in obesity. In fact, obese human adipose tissue has increased protein-tyrosine phosphatase (PTPase) activity, which can dephosphorylate and inactivate the insulin receptor kinase, leading to inhibition of IRS-1 phosphorylation [
54]. In agreement with this, we found that the mRNA of Ptpra was elevated in adipose tissue of obese mice compared to Control mice (Table
2). As expected, the levels of Ptpra were decreased in CR mice. This suggests that calorie restriction and obesity may affect insulin sensitivity by affecting the balance of phosphatases such as Dusp-9 and Ptpra. Alcohol increased Dusp-9 mRNA levels but had no obvious effect on Ptpra, indicating yet once more that alcohol may improve insulin sensitivity through overlapping but distinct mechanisms compared to body weight.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
QXP measured gene expression profiles by StellARray and qPCR and helped draft the manuscript. JH carried out the animal study and collected tissues for gene expression analyses. VBH participated in the design and helped draft the manuscript. NPN conceived the study and participated in its design and coordination. All authors read and approved the final manuscript.