Introduction

The metabolic syndrome is a complex pathological state that is associated with obesity, hypertension, atherosclerotic cardiovascular disease, and type 2 diabetes [1]. An underlying feature that is present across these common diseases is insulin resistance, which is defined as a decreased ability of insulin to perform its biological functions. Moreover, the pathophysiology of the metabolic syndrome and its associated diseases are attributable to a low grade inflammation [2]. Over the past three decades, the prevalence of the metabolic syndrome has increased, largely due to the increased prevalence observed for obesity and type 2 diabetes [3-6]. The metabolic syndrome (also referred to as insulin resistance syndrome and syndrome X), can be defined in various ways [7], with the essential components including obesity, dyslipidemia, hypertension, and glucose intolerance.

To date, the majority of our work has focused on skeletal muscle insulin resistance, where we have used global proteomics and transcriptomic approaches to demonstrate the relationship between inflammation, extracellular remodeling, cytoskeletal interactions, and mitochondrial function [2]. Moreover, many cellular, molecular and biochemical defects have been shown to contribute to the pathophysiology of insulin resistance including impaired insulin signaling, reduced insulin-stimulated glucose uptake, lower insulin-stimulated activities of enzymes such as hexokinase and glycogen synthase, increased toxic lipid metabolites and impaired mitochondrial function [8-16]. Studying skeletal muscle in relation to insulin resistance and its associated diseases such as the metabolic syndrome is particularly important, as under normal physiological conditions this tissue is the major site of insulin-stimulated glucose disposal [17]. To date, the role of peripheral whole blood in the metabolic syndrome and the underlying insulin resistance is less understood. As such, one of the main objectives of this study was to determine whether relationships that we have found in the skeletal muscle similarly exist in the peripheral whole blood cells, so that we could eventually use whole blood as a surrogate tissue for studying the metabolic syndrome.

By using global transcriptomic approaches, we set out to test the hypothesis that there would be changes in the expression of genes involved in inflammation, insulin signaling and mitochondrial function in the whole blood of the subjects classified with the metabolic syndrome. Conducting this study using global gene expression profiling allowed us to test these hypotheses and at the same time identify novel targets of metabolic syndrome and its underlying insulin resistance in whole blood, which could allow for development of new strategies for diagnosis as well as the identification of novel treatment targets.

Methods

Microarray analysis

Feature Extraction Software version 10.5.1.1 (Agilent Technologies, Palo Alto, CA), was used for analysis of the array images. The data generated from the image analysis has been deposited in the ArrayExpress database (www.ebi.ac.uk/arrayexpress) under accession number E-MEXP-12345. The data files were imported into GeneSpring version 12 (Agilent Technologies, Palo Alto, CA). Data were log transformed, quantile normalized and baseline to median of all samples. The probes below 20% expression were excluded from the analysis. A detailed overview of the analysis is shown in Figure 1. Expression values obtained were submitted to a Student’s unpaired t-test, and P values were adjusted using the Benjamini-Hochberg multiple testing correction. Significant probes with a Benjamini-Hochberg P < 0.05 and fold change ≥ 1.2 were used for exploratory KEGG pathway analysis using the database for annotation, visualization and integrated discovery (DAVID) (http://david.niaid.nih.gov), as previously described [22]. Moreover, to identify the genes that differed most robustly, we filtered the probe list further to identify probes with a Benjamini-Hochberg P < 0.05 and fold change ≥ 1.5.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Figure 1. Flow diagram of the steps used in the analysis of the microarray data.

https://doi.org/10.1371/journal.pone.0084002.g001

Results

Discussion

The present study was undertaken to decipher the role of peripheral whole blood in relation to the metabolic syndrome phenotype. We set out to test the hypothesis that genes involved in inflammation, insulin signaling and mitochondrial function would be altered in expression in the whole blood of individuals with metabolic syndrome. We further wanted to examine whether similar relationships that we have found previously in skeletal muscle exist in peripheral whole blood cells. It has been shown elsewhere that over 80% of the genes expressed in nine different human tissue types overlapped with the peripheral blood transcriptome [27]. Therefore, the peripheral blood cells express a large proportion of the genes in the human genome, suggesting that whole blood may be useful as a surrogate for studying metabolic syndrome and related conditions such as insulin resistance, diabetes and obesity.

We performed global transcriptomic analyses of peripheral whole blood as it is an easy tissue to obtain and it is therefore non-invasive. We performed genome-wide gene expression profiling on RNA obtained directly from whole blood collected in PAXgene tubes rather than use peripheral blood mononuclear cells (PBMCs). The disadvantage of using the PAXgene tubes and studying whole blood is that the results generated from this gene expression profiling study represent the contribution of RNA from all blood cell types. Whole blood is a complex tissue comprised of many cell types including neutrophils, eosinophils, basophils, B cells, T cells, and monocytes. As such, when using the PAXgene tube approach we cannot definitely state which cell type is contributing the mRNA expression signal. However, the goal for this study was to develop a transcriptome for studying metabolic syndrome in whole blood, and we chose the PAXgene tubes to minimize RNA degradation, to minimize handling and potential activation of blood cells, and to minimize loss of sample in purification of specific sub populations of leukocytes, all of which could affect gene expression levels.

In the present study, fold changes ranged from -2 to +2.7, which were much lower than our previous studies in skeletal muscle where observed changes range from -3 to +8 [22,28]. Moreover, it has been observed elsewhere that the gene expression levels from PAXgene RNA are much lower when compared to studies using PBMCs [29,30]. Ghosh et al., addressed this issue and suggested that the lower fold change ranges in PAXgene analyses is attributable to the high abundance of globin in the whole blood [30]. Although 70% of whole blood mRNA is represented by globin, we were able to validate the gene expression changes captured in the microarray with quantitative RT-PCR. This indicates that the changes observed in the present study were present, suggesting that the genes and pathways are biologically relevant and could be used for development of new strategies for diagnosis as well as the identification of novel treatment targets.

In this study, each individual was classified based on metabolic syndrome criteria. Metabolic syndrome was used to classify our sample population due to its known concordance with type 2 diabetes, obesity, hypertension, and atherosclerotic cardiovascular disease [1]. For this study, the relationship between insulin resistance and the NCEP: ATP III phenotypic criteria for metabolic syndrome were reinforced by measurements of Matsuda Index [21]. We have previously used the Matsuda index, which is considered a surrogate measure of whole body insulin sensitivity obtained from the OGTT [18,21]. However, in the present study, the Matsuda index was not solely used to determine groupings since a clear cut-off point for classification of insulin resistance versus insulin sensitivity in Hispanic populations has not yet been defined. In the present study, the metabolic syndrome criteria provides groupings that are reinforced by multiple variables, one of which being insulin sensitivity. Subjects classified with metabolic syndrome were more insulin resistant, which was fully anticipated since insulin resistance is thought to be the core defect underlying this condition [19,20]. It could be argued that the significant differences for the metabolic characteristics between the two groups could be due to age since the individuals without metabolic syndrome were slightly younger in age.

Table S1 provided sex associated differences within the Met Syn-NO and Met Syn-YES groupings. The body fat percentage was significantly higher in women than men, and it has been shown that women tend to have larger proportion of body fat mass, and a higher yield of fat storage than men [31]. Fat metabolism involves the transportation of free fatty acids for storage. An important molecule which assists with this process is HDL. The levels of HDL in our subjects were significantly higher in women than men. The observed sex difference could be due to the premenopausal state in women, where there is typically lower total cholesterol and triglycerides, and higher HDL levels than men because of sex hormone regulation differences in fat metabolism [32]. This possible explanation could be applied to the current study based on the average female ages being below the typical age of menopause, and the same trend as described for the triglycerides, cholesterol, and HDL (as shown in Table S1).

Another significant sex difference was found in the 2 hour plasma insulin concentrations, where the women were found to have higher levels than men. These results could be reflective of the study conducted by Magkos et al., which considered the effects of insulin and the suppression of endogenous glucose production between men and women [33]. When their subjects were infused with insulin to raise insulin concentrations to 15-20 mU/L, women had a prolonged suppression of endogenous glucose production than men [33]. This could infer that the proposed concentration of infused insulin is sustained longer in women. Another point of interest was the significantly higher levels of ALT in men than in women. A study performed elsewhere demonstrated that ALT (≥ 40 U/L) could be used as a predictor for coronary heart disease (CHD) in men, but not women [34]. The sex differences in ALT may also be due to the level of work performed by the liver in men (hepatic glucose production was not as easily suppressed) as opposed to women, as suggested in the aforementioned study [33]. Men were also found to present signs of CHD about ten years before women, so the average ages of the male and female subjects in our current study may add to the discrepancy [34].

In the present study, there was a higher expression of oxidative phosphorylation genes observed in the individuals classified with the metabolic syndrome phenotype. The higher expression in oxidative phosphorylation genes in whole blood is contradictory to the observed findings in skeletal muscle. We [15] and others [16] have demonstrated that nuclear encoded mitochondrial genes encoding proteins involved in electron transport and oxidative phosphorylation were decreased in insulin resistant muscle, and many of these changes could be attributed to lower peroxisome proliferator-activated receptor alpha (PPAR alpha) coactivator-1 alpha (PGC-1 alpha) gene expression [15]. We also showed that experimental insulin resistance produced by a lipid infusion in healthy volunteers decreases the mRNA expression of PGC-1 alpha and other mitochondrial and oxidative phosphorylation genes [28]. The discrepancies in the expression of genes involved in oxidative phosphorylation may be attributable to differences in tissue types and the role that each plays in the disease. A similar study using PAXgene tubes demonstrated an increased expression in oxidative phosphorylation genes in an obese Northern European White cohort; specifically cytochrome c oxidase subunits C6, B7, C7, and NADH dehydrogenase 1 alpha sub-complex 4 showed significant enrichment in both studies [30]. The increase in oxidative phosphorylation genes in whole blood may be caused by a response to increased energy demands in the metabolic syndrome individuals [30], or may be explained by an inflammation-induced increase in reactive oxygen species (ROS) generation. Although ROS was not measured in the present study, we know from other studies that in insulin resistant individuals the glucose levels in the bloodstream are elevated, which could increase the metabolic substrate input of the mitochondria leading to stress on the electron transport chain and causing an overproduction of ROS [35]. When the blood contains high amounts of ROS, a condition called oxidative stress is created. Previous studies have shown that oxidative stress plays key roles in diabetes, cardiovascular disease, and neurological disorders [36]. Although we observed similar findings to Ghosh et al., there were differences in our study design; our subjects were Latin descent, categorized based on metabolic syndrome criteria and we had much higher numbers of individuals entering into the analysis.

The ribosome genes and pathway had a strong, distinct association with metabolic syndrome. Interestingly, all of the genes within the pathway showed an increase in expression. These results suggest that in the peripheral blood of individuals with the metabolic syndrome there is an increase in the cellular content of ribosomes augmenting the capacity for protein synthesis. The increase in ribosomes suggests that there is possibly an increase in protein synthesis to account for the increased energy demands observed in individuals with the metabolic syndrome; the significantly enriched genes possibly accounting for this phenomena (also found in Ghosh et al.) being ribosomal protein L7, S7, S24, and S31 [30]. Another study had shown a direct correlation between energy demands and ribosomal RNA synthesis through AMP-activated protein kinase in mammalian cells [37]. Their findings suggest that ribosomal RNA synthesis will increase when mitogenic conditions are optimal; for example, an overabundance of available nutrients. Therefore, it is possible that the physiological conditions of our sample population could provide optimal conditions for ribosomal RNA synthesis from the elevated levels of glucose.

An increase in ribosomal genes could also be linked to inflammation. Insulin resistance causes an environment in which deoxyhypusine synthase (DHS) promotes cytokine-induced inflammation in β-cells, as well as catalyzes the conversion of lysine to hypusine that is unique to the translational elongation factor eIF5A found in the endoplasmic reticulum (ER) [38]. Therefore, when β-cell dysfunction occurs due to inflammatory stress, DHS also causes ER stress by eIF5A increased production of proteins [38]. A recent study performed by Yamaoka et al. on blood gene expression in Japanese cohorts separated by metabolic syndrome classifications also identified chronic low-grade inflammation related to ROS production; however, their study focused on data related to adipose tissue as opposed to skeletal muscle [39]. An interesting similarity between our studies is the increased expression of S100A8 related to metabolic syndrome explained by Yamaoka et al., which stated a correlation between the expression and ROS generation promoted by calprotectin (a S100A8/A9 complex) [40].

Our study demonstrated alterations in the MAPK signaling pathway genes with metabolic syndrome. Interestingly, all of the genes identified in the analysis were lower in expression in the metabolic syndrome individuals. We hypothesized that there would be alterations in genes involved in insulin signaling; however we expected to observe increased expression. Several studies have indicated that insulin resistance is connected to an escalation in activity from JNK and p38 in MAPK signaling pathway [41]. Since this was not the case in our study, the discrepancies may be explained by differences in blood tissue compared with skeletal muscle. The differences seen in blood may be related to the relationship between insulin resistance and endothelial dysfunction. Normally, insulin induces phosphatidylinositol 3-kinase dependent signaling (PI3K) to produce nitric oxide (NO) which increases blood flow in the vascular endothelium, and ultimately increases glucose uptake in skeletal muscle [42]. This process is accompanied by the MAPK-dependent insulin-signaling pathway that produces endothelin-1, which determines vascular responses to insulin [42]. Endothelial dysfunction is usually viewed as an imbalance that occurs in an insulin resistant state, where PI3K signaling is impaired and MAPK signaling is heightened [42]. The results from our study may indicate that blood is more sensitive to the effects occurring to PI3K in the vascular endothelium in relation to MAPK signaling, but not the actual enhanced effects of MAPK.

In this study, the endocytosis pathway was lower in expression in the individuals classified with metabolic syndrome. These results are reinforced from a recent study on the role of autophagy in metabolism. Autophagy is a homeostatic mechanism that is typically known for transporting items such as damaged organelles or misfolded proteins to the lysosome to be degraded [43]. This process has also been viewed as a generator for cellular energy from protein degradation into amino acids and its role in the releasing free fatty acids to be used for beta-oxidation [43]. This ability to provide energy is increased under conditions of starvation coupled by internal stimuli such as oxidative stress, protein aggregation, and damaged organelles [44]. However, in our metabolic syndrome group, there is an excess supply of nutrients. Autophagy activity is decreased in environments with elevated glucose, because glucagon causes the up regulation in autophagy [43]. These findings provide some evidence with the decreased expression seen in endocytosis for our sample population.

From our analysis a cluster of neurological pathways were found to be significant in metabolic syndrome, which included Alzheimer’s disease, Huntington’s disease, and Parkinson’s disease. The relevance of this finding is due to the growing efforts to study the late effects of metabolic syndrome type environments on the neurological system; mainly efforts to mark Alzheimer’s disease as type 3 diabetes [45]. The connection between metabolic syndrome and the nervous system is primarily due to insulin, which in the central nervous system plays a role in controlling neuronal function (memory and cognitive functioning), neurotransmitter release, and synaptic plasticity [45]. Insulin receptors are located in both neurons and glia of the brain, and are specifically prevalent in regions such as, the hypothalamus, cerebellum, olfactory bulb, hippocampus and cerebral cortex [46]. In conditions where prolonged elevation of glucose causes hyperinsulinemia (as associated with insulin resistance) in the bloodstream, a reduction of insulin crossing the blood brain barrier can occur [46]. Such an environment is thought to cause a shift to peripheral insulin use which can lead to downstream affects in memory, learning, and neuronal survival [45].

The systemic lupus erythematosus (SLE) pathway was found to be altered in our metabolic syndrome subjects. SLE has been described as a chronic inflammatory disease characterized by abnormal T and B cell function and production of antinuclear antibodies [47]. Inflammation is prominent in SLE, as well as insulin resistance and metabolic syndrome [2]. In the present study, an inflammatory marker that was increased in expression in the metabolic syndrome individuals was lactotransferrin, which has been shown to be increased in concentration during most inflammatory reactions [48]. Interestingly, in patients with SLE there is a four times greater chance of having the metabolic syndrome than without SLE [49]. Some of the genes affected in the oxidative phosphorylation pathway were also present in the SLE pathway. This could be attributed to the relationship of SLE with mitochondria. Individuals with SLE have been known to exhibit spontaneous apoptosis and defective activation-induced cell death of the T cells [47]. Since cell proliferation and apoptosis are energy-dependent processes, a disruption in the functionality of the mitochondria and alterations in ROS is suggested [47]. The most predominant finding in the SLE pathway analysis was the reduced expression in histone related genes in those individuals with metabolic syndrome. This decrease could be explained by a recent study that showed that the majority of individuals with SLE create anti-histone antibodies specifically directed towards histones within neutrophil extracellular traps (NETs) [49]. Neutrophils are one of the many cells found within whole blood, and if our population sample has similar physiology to SLE individuals it is possible that our results are showing that specific type of autoreactivity.

Collectively, the global transcriptomic analyses on metabolic syndrome in our Latino population provided data that whole blood may be a useful surrogate for studying particular pathways. Based on our findings, pathways involved in ribosomes, Alzheimer’s disease, Huntington’s disease, Parkinson’s disease, systemic lupus erythematosus, and endocytosis may logistically continue studies using whole blood in relation to outcomes of metabolic syndrome. However, the pathways specifically linked to mitochondrial genes, such as oxidative phosphorylation and MAPK signaling have shown reversed outcomes in whole blood than seen in skeletal muscle for reasons that need further investigation.

Supporting Information

Acknowledgments

We thank the participants of the study and are grateful for their participation and cooperation. We thank Veronica Vital, RN and the rest of the ASU clinical research staff for their excellent care of the participants. Dawn K. Coletta is the guarantor of this work, had full access to all the data, and takes full responsibility for the integrity of data and the accuracy of data analysis.