Gene Expression Profiling of Peripheral Blood Mononuclear Cells in Type 2 Diabetes: An Exploratory Study

Background and Objectives: Visceral obesity is associated with chronic low-grade inflammation that predisposes to metabolic syndrome. Indeed, infiltration of adipose tissue with immune–inflammatory cells, including ‘classical’ inflammatory M1 and anti-inflammatory ‘alternative’ M2 macrophages, causes the release of a variety of bioactive molecules, resulting in the metabolic complications of obesity. This study examined the relative expression of macrophage phenotypic surface markers, cholesterol efflux proteins, scavenger receptors, and adenosine receptors in human circulating peripheral blood mononuclear cells (PBMCs), isolated from patients with type 2 diabetes mellitus (T2DM), with the aim to phenotypically characterize and identify biomarkers for these ill-defined cells. Materials and Methodology: PBMCs were isolated from four groups of adults: Normal-weight non-diabetic, obese non-diabetic, newly diagnosed with T2DM, and T2DM on metformin. The mRNA expression levels of macrophage phenotypic surface markers (interleukin-12 (IL-12), C-X-C motif chemokine ligand 10 (CXCL10), C-C motif chemokine ligand 17 (CCL17), and C-C motif receptor 7 (CCR7)), cholesterol efflux proteins (ATP-binding cassette transporter-1 (ABCA1), ATP binding cassette subfamily G member 1 (ABCG1), and sterol 27-hydroxylase (CYP27A)), scavenger receptors (scavenger receptor-A (SR-A), C-X-C motif ligand 16 (CXCL16), and lectin-like oxidized LDL receptor-1 (LOX-1)), and adenosine receptors (adenosine A2A receptor (A2AR) and adenosine A3 receptor (A3R)) were measured using qRT-PCR. Results: In PBMCs from T2DM patients, the expression of IL-12, CCR7, ABCA1, and SR-A1 was increased, whereas the expression of CXCL10, CCL17, ABCG1,27-hydroxylase, LOX-1, A2AR and A3R was decreased. On the other hand, treatment with the antidiabetic drug, metformin, reduced the expression of IL-12 and increased the expression of 27-hydroxylase, LOX-1, CXCL16 and A2AR. Conclusions: PBMCs in the circulation of patients with T2DM express phenotypic markers that are different from those typically present in adipose tissue M1 and M2 macrophages and could be representative of metabolically activated macrophages (MMe)-like cells. Our findings suggest that metformin alters phenotypic markers of MMe-like cells in circulation.


Introduction
The pathological activation of the innate immune system in metabolic syndrome is well-established. Inflammation plays a key role in the development and progression of the multifaceted metabolic disorder type 2 diabetes mellitus (T2DM), which is characterized by peripheral insulin resistance and systemic glucolipotoxicity [1,2]. Indeed, inflammation may contribute to obesity-linked metabolic dysfunctions and is associated with polarization motif chemokine ligand 10 (CXCL10), C-C motif chemokine ligand 17 (CCL17), and C-C motif receptor 7 (CCR7)) were measured in PBMCs isolated from four groups of adults: normal-weight non-diabetic, obese non-diabetic, newly diagnosed with T2DM, and T2DM on metformin. Stimuli associated with T2DM promote PBMC-distinct mechanisms and surface markers activation in circulation. Collectively, the data from this study provide important mechanistic insights into pathways that drive the metabolic-disease-specific phenotype of PBMCs in circulation.

Subjects
The study, approved by King Abdullah International Medical Research Center (KAIMRC) Institutional Review Board (reference number 101/15; date of approval 17 February 2015), was conducted on four groups of adults: 30 normal-weight non-diabetic (BMI: 23.0 ± 0.3 kg/m 2 ), 30 obese non-diabetic (BMI: 39.1 ± 1.7 kg/m 2 ), 20 newly diagnosed with T2DM (BMI: 32.5 ± 1.9 kg/m 2 ), and 30 patients with T2DM on metformin (BMI: 40.5 ± 1.5 kg/m 2 ; 9 months-15 years duration). While most patients were on daily doses of metformin ranging between 1000 and 2000 mg, 5 patients had a smaller dose of 100-500 mg, and only 3 patients had a higher dose of 3000-4500 mg. In addition, 5 patients were on combined therapy of metformin and insulin. Few subjects from the obese, obese T2DM, and obese T2DM on metformin groups were on stable doses-for the last two months of their participation in the study-of statins or other cholesterol-lowering agents, angiotensinconverting enzyme inhibitors (ACE-I) or other anti-hypertensives, NSAIDS or antioxidants. All subjects were enrolled into the study following medical screening at King Abdulaziz Medical City in Riyadh, Saudi Arabia. Anthropometric measurements were measured (blood pressure, weight, height, waist, and hip circumferences along with complete blood count (CBC), blood urea nitrogen (BUN), creatinine, fasting blood sugar, liver function tests, and lipid profile) following 12 h of fasting before their venipuncture appointment. Exclusion criteria included coronary event or procedure in the previous 3 months, liver disease, renal impairment, history of drug or alcohol abuse, and steroid therapy.

Isolation of PBMCs
A total of 10 mL of blood was obtained from all participants and collected in EDTAcontaining tubes. Blood samples were mixed with 10 mL of PBS and layered over 15 mL of Ficoll-Hypaque (50 mL Leucosep Tubes, Greiner Bio-One North America Inc, Monroe, NC, USA). Samples were centrifuged at room temperature for 30 min at 450× g. The PBMC layer was harvested with a pipette then washed with PBS. An aliquot of 50 µL Qiagen RNALater was added, and the samples were stored at −80 • C ( Figure 1).

Statistical Analysis
SigmaStat software ver. 3.0 (Jandel Scientific, San Rafael, CA, USA) was used for statistical analysis. Changes in mRNA expression levels were computed for qRT-PCR results, and analysis was conducted using one-way ANOVA on ranks for all analytes measured in this study as the Kolmogorov-Smirnov test normality distribution test failed. Dunn's test for pairwise comparisons and comparisons against the normal-weight group was used. Correlation analysis between age and expression of macrophage phenotypic markers, cholesterol efflux proteins, scavenger receptors, or adenosine receptors was performed using Spearman's rank correlation. p-value < 0.05 was used to assess significance for all statistical analyses. Results are presented as mean ± standard error of the mean (SEM).

Demographic Data of Study Participants
The demographic data of the study participants (Table 2) indicate that groups 3 and 4 were significantly older than the normal-weight and obese groups. However, there was no significant correlation (Pearson r correlation) between age and the expression levels of macrophage phenotypic markers, cholesterol efflux proteins, scavenger receptors, or adenosine receptors examined in the study. To examine the effect of gender in the 12 analyzed biomarkers (IL-12, CXCL10, CCL17, CCR7, ABCA1, ABCG1, 27-hydroxylase, SR-A1, CXCL16, LOX-1, A2AR, and A3R), a chi-square test was performed. Gender was statistically non-significantly associated with the different biomarkers tested (χ 2 (10) = 0.828, p = 1.000). However, gender was statistically significantly associated among different study groups (χ 2 (3) = 9.50, p = 0.023).
Despite being on medication, the T2DM on metformin group had significantly higher blood glucose and HbA1c compared to the normal-weight and obese groups. The T2DM group had significantly higher LDL when compared to the normal-weight disease-free group (p < 0.05). However, the T2DM on metformin group had significantly lower LDL when compared to the T2DM group, suggesting that metformin restores LDL values to normal.

Macrophage Phenotypic Markers' Expression
IL-12 expression was significantly higher in the T2DM group (p < 0.05) when compared to the lean and obese groups. This increased expression was significantly lowered by metformin treatment (p < 0.05). Moreover, the expression of CCR7 was significantly higher in the T2DM group when compared to both non-diabetic groups (p < 0.05). Metformin had no significant effect on CCR7. On the other hand, CXCL10 and CCL17 mRNA expression was lower in the T2DM group when compared to both non-diabetic groups (p < 0.05). Metformin treatment conferred no significant change in the expression of CXCL10 and CCL17 ( Figure 2).

Expression of Cholesterol Efflux Proteins
A significant increase in the expression of ABCA1 was observed in the T2DM group when compared to the non-diabetic groups (p < 0.05). This increase was not altered by metformin treatment. In contrast, ABCG1 expression was reduced in the T2DM group when compared to the disease-free groups (p < 0.05). Moreover, 27-hydroxylase expression was significantly lower in the T2DM group when compared to non-diabetic groups (p < 0.05). Metformin significantly increased the expression of sterol 27-hydroxylase in the PBMCs of the T2DM on metformin group compared to the T2DM group (p < 0.05, Figure 3). by metformin treatment (p < 0.05). Moreover, the expression of CCR7 was significantly higher in the T2DM group when compared to both non-diabetic groups (p < 0.05). Metformin had no significant effect on CCR7. On the other hand, CXCL10 and CCL17 mRNA expression was lower in the T2DM group when compared to both non-diabetic groups (p < 0.05). Metformin treatment conferred no significant change in the expression of CXCL10 and CCL17 (Figure 2). Results are presented as Mean ± SEM. One-way ANOVA on ranks, p < 0.001 for IL-12, CXCL10, CCL17, and CCR7, followed by Dunn's test for pairwise comparisons and comparisons against normal-weight group, * p < 0.05 vs. normal-weight subjects; δ p < 0.05 vs. obese; σ p < 0.05 vs. T2DM subjects.

Expression of Cholesterol Efflux Proteins
A significant increase in the expression of ABCA1 was observed in the T2DM group when compared to the non-diabetic groups (p < 0.05). This increase was not altered by metformin treatment. In contrast, ABCG1 expression was reduced in the T2DM group when compared to the disease-free groups (p < 0.05). Moreover, 27-hydroxylase expression was significantly lower in the T2DM group when compared to non-diabetic groups (p < 0.05). Metformin significantly increased the expression of sterol 27-hydroxylase in the PBMCs of the T2DM on metformin group compared to the T2DM group (p < 0.05, Figure 3).

Expression Analysis of Scavenger Receptors
SR-A1 expression was significantly higher in the T2DM group compared to the disease-free groups (p < 0.05). Although metformin treatment reduced the expression of SR-A1, the reduction in SR-A1 expression by metformin did not reach statistical significance (p > 0.05). Moreover, LOX-1 expression was not statistically significantly lower in the Results are presented as Mean ± SEM. One-way ANOVA on ranks, p < 0.001 for ABCA1, ABCG1 and Sterol 27-hydroxylase, followed by Dunn's test for pairwise comparisons and comparisons against normalweight group, * p < 0.05 vs. normal-weight subjects; δ p < 0.05 vs. obese; σ p < 0.05 vs. T2DM subjects.

Expression Analysis of Scavenger Receptors
SR-A1 expression was significantly higher in the T2DM group compared to the diseasefree groups (p < 0.05). Although metformin treatment reduced the expression of SR-A1, the reduction in SR-A1 expression by metformin did not reach statistical significance (p > 0.05). Moreover, LOX-1 expression was not statistically significantly lower in the T2DM group compared to the disease-free groups (p > 0.05), but metformin administration significantly increased LOX-1 expression to levels higher than in the T2DM group (p < 0.05). This contradicts the reported observation that LOX-1 expression is increased in THP-1 cells induced by interferon-γ and LPS to M1 [25]. CXCL16 expression was not statistically significantly higher in the T2DM group compared to the disease-free groups (p > 0.05), and metformin treatment increased CXCL16 expression to a statistically significant level (p < 0.05, Figure 4).  One-way ANOVA on ranks, p < 0.001 for SR-A1, LOX-1, and CXCL16, followed by Dunn's test for pairwise comparisons and comparisons against normal-weight group, * p < 0.05 vs. normal-weight subjects; δ p < 0.05 vs. obese; σ p < 0.05 vs. T2DM subjects.

Adenosine Receptors A2AR and A3R Expression
A significant decrease in A2A and A3 receptor expression was observed in the T2DM groups as opposed to the disease-free groups (p < 0.05). However, metformin treatment led to a considerable increase in the expression of A2AR (p < 0.05) but not in A3R expression ((p > 0.05, Figure 5).  One-way ANOVA on ranks, p < 0.001 for SR-A1, LOX-1, and CXCL16, followed by Dunn's test for pairwise comparisons and comparisons against normal-weight group, * p < 0.05 vs. normal-weight subjects; δ p < 0.05 vs. obese; σ p < 0.05 vs. T2DM subjects.

Adenosine Receptors A2AR and A3R Expression
A significant decrease in A2A and A3 receptor expression was observed in the T2DM groups as opposed to the disease-free groups (p < 0.05). However, metformin treatment led to a considerable increase in the expression of A2AR (p < 0.05) but not in A3R expression ((p > 0.05, Figure 5).

Summary of the Changes in the Relative Expression of the Studied Biomarkers
The mRNA expression of all markers examined in this study is summarized in Figure 6. In PBMCs from T2DM patients, the expression of IL-12, CCR7, ABCA1, and SR-A1 was increased, whereas the expression of CXCL10, CCL17, ABCG1, 27-hydroxylase, A2AR and A3R was decreased. On the other hand, treatment with metformin, was associated with increased expression of CCR7, ABCA1, SR-A1, LOX-1, and CXCL16 and decreased expression of CXC10, CCL17, ABCG1, and A3R. subjects; δ p < 0.05 vs. obese; σ p < 0.05 vs. T2DM subjects.

Adenosine Receptors A2AR and A3R Expression
A significant decrease in A2A and A3 receptor expression was observed in the T2DM groups as opposed to the disease-free groups (p < 0.05). However, metformin treatment led to a considerable increase in the expression of A2AR (p < 0.05) but not in A3R expression ((p > 0.05, Figure 5). Results are presented as Mean ± SEM. One-way ANOVA on ranks, p < 0.001 for A2AR, and A3R, followed by Dunn's test for pairwise comparisons and comparisons against normal-weight group, * p < 0.05 vs. normal-weight subjects; δ p < 0.05 vs. obese; σ p < 0.05 vs. T2DM subjects. Results are presented as Mean ± SEM. One-way ANOVA on ranks, p < 0.001 for A2AR, and A3R, followed by Dunn's test for pairwise comparisons and comparisons against normal-weight group, * p < 0.05 vs. normal-weight subjects; δ p < 0.05 vs. obese; σ p < 0.05 vs. T2DM subjects.

Summary of the Changes in the Relative Expression of the Studied Biomarkers
The mRNA expression of all markers examined in this study is summarized in Figur 6. In PBMCs from T2DM patients, the expression of IL-12, CCR7, ABCA1, and SR-A1 wa increased, whereas the expression of CXCL10, CCL17, ABCG1, 27-hydroxylase, A2A and A3R was decreased. On the other hand, treatment with metformin, was associate with increased expression of CCR7, ABCA1, SR-A1, LOX-1, and CXCL16 and decrease expression of CXC10, CCL17, ABCG1, and A3R.

Discussion
The circulating PBMCs of T2DM have a different pattern of phenotypic markers tha the patterns typically present in M1 and M2 macrophages that could be representative o MMe-like cells [7]. Higher expression levels of CD16, interleukin-6 (IL-6), inducible nitri oxide synthase (iNOS), tumor necrosis factor alpha or (TNFα), and CD36 have been re ported in the PBMCs of T2DM which is consistent with the M1 macrophage-like pheno type. In addition, the PBMCs of T2DM have higher expression levels of mannose recepto (CD206), an M2-specific marker [7]. A majority of the data regarding macrophage polar zation are from studies conducted on murine models despite significant differences be tween human and murine macrophages. Other macrophage surface markers (IL-12 CXCL10, CCL17, and CCR7) were analyzed in this study to further characterize huma circulating PBMC phenotypes in T2DM and compare them to macrophage phenotype since information on human macrophage polarization remains sparse. Several studie have attempted to characterize MMe in the adipose tissue of obese and diabetic subject [26][27][28]. Ex vivo treatment of monocyte-derived macrophages with glucose, insulin, an palmitate induced a different phenotype than the M1 macrophage phenotype [6]. An i vitro study utilizing human monocytic cell lines classified IL-12, CCR7, and CXCL10 a M1 macrophage markers, and CCL17 as an M2 marker [25]. The increased expression o IL-12 and CCR7 in the PBMCs of T2DM is consistent with M1-like cells. Metformin trea ment reduced IL-12 levels indicating that it reduces the M1 macrophage-like phenotype

Discussion
The circulating PBMCs of T2DM have a different pattern of phenotypic markers than the patterns typically present in M1 and M2 macrophages that could be representative of MMe-like cells [7]. Higher expression levels of CD16, interleukin-6 (IL-6), inducible nitric oxide synthase (iNOS), tumor necrosis factor alpha or (TNFα), and CD36 have been reported in the PBMCs of T2DM which is consistent with the M1 macrophage-like phenotype. In addition, the PBMCs of T2DM have higher expression levels of mannose receptor (CD206), an M2-specific marker [7]. A majority of the data regarding macrophage polarization are from studies conducted on murine models despite significant differences between human and murine macrophages. Other macrophage surface markers (IL-12, CXCL10, CCL17, and CCR7) were analyzed in this study to further characterize human circulating PBMC phenotypes in T2DM and compare them to macrophage phenotypes since information on human macrophage polarization remains sparse. Several studies have attempted to characterize MMe in the adipose tissue of obese and diabetic subjects [26][27][28]. Ex vivo treatment of monocyte-derived macrophages with glucose, insulin, and palmitate induced a different phenotype than the M1 macrophage phenotype [6]. An in vitro study utilizing human monocytic cell lines classified IL-12, CCR7, and CXCL10 as M1 macrophage markers, and CCL17 as an M2 marker [25]. The increased expression of IL-12 and CCR7 in the PBMCs of T2DM is consistent with M1-like cells. Metformin treatment reduced IL-12 levels indicating that it reduces the M1 macrophage-like phenotype. CCL17, which is released from alternatively activated macrophages, serves to prevent a generation of classically activated macrophages, and is considered an M2 marker [26], was inhibited in the PBMCs of T2DM. Thus, our previously published CD163 results [7] and current CCL17 results are consistent with an anti-inflammatory role of CD163 and CCL17 (M2 markers) as they are reduced in T2DM [7]. Metformin treatment resulted in an increase in CD163 and CCL17, suggesting that it induces a phenotype similar to the M2-like phenotype [7]. However, the expression of CXCL10, also known as interferon gamma (IFN-γ)-inducible protein 10, was inhibited in the PBMCs of T2DM [7]. CXCL10, secreted by leukocytes and tissue cells, functions as a chemoattractant, mainly for lymphocytes. A recent study reported CXCL10 and CXCL11 as potential biomarkers for the onset of adipose tissue inflammation during obesity with CXCL11 expression correlation with NF-κB expression [27]. This discrepancy could stem from the heterogeneity of PBMCs and further studies are needed to examine the role of the CXCL subfamily of chemokines (CXCL9, CXCL10, and CXCL11; angiostatic chemokines) in adipose tissue inflammation and their utilization as biomarkers for the MMe phenotype.
Inflammation and stress cause hypoxia and an accumulation of extracellular adenosine [13,14]. Adenosine receptors' (A2AR and A3R) activation inhibits NF-κB signaling pathways, enabling them to exert anti-inflammatory effects [15]. The NF-κB pathway is responsible for triggering the transcription of several pro-inflammatory genes. The NF-κB pathway plays a pivotal role in insulin resistance and ATM activation [28]. On the other hand, A2AR plays a protective role in obesity-associated adipose tissue inflammation by suppressing macrophage pro-inflammatory activation, including inhibition of the NF-κB pathway [29]. High-fat-diet (HFD) feeding of A2AR-disrupted mice increased adipose tissue inflammation and adipose tissue insulin resistance [29]. Similarly, the activation of A3R resulted in NF-κB pathway inhibition [30]. In our study, A2AR and A3R expression levels were significantly lower in the PBMCs from T2DM. These results suggest that activation of the NF-κB pathway, with the subsequent activation of inflammation, reported in T2DM could be mediated by A2AR and A3R inhibition. Thus, adenosine could be utilized as a new strategy to regulate metabolic homeostasis through the modulation of adipocyte-macrophage interaction. Interestingly, A2AR has been shown to diminish foam cell formation by increasing the expression and function of cholesterol 27-hydroxylase, an enzyme involved in the conversion of cholesterol to oxysterols [31] and enhancement of cholesterol efflux [32]. A2AR and cholesterol 27-hydroxylase expression are inhibited in this study in T2DM. Metformin increased the expression levels of A2AR and 27-hydroxylase beyond the levels expressed in T2DM, suggesting that it enhances cholesterol efflux from peripheral tissues by upregulating A2AR and 27-hydroxylase.
ABCA1 and ABCG1 play a key role in mediating the efflux of cholesterol from peripheral cells. ABCA1 and ABCG1 promote unidirectional cholesterol efflux to lipid-poor apolipoprotein A-I (apoA-I), apoE, or HDL particles [33]. The M1 phenotype has been shown to express high levels of the ABCA1 protein [25] and lower levels of SR-B1 involved in cholesterol efflux compared to M2 macrophages [34]. No changes in the expression of ABCG1 in M1 and M2 subsets were observed by Littlefield et al. [25] compared to Waldo et al. [34]. Additionally, ABCG1 expression in metabolic syndrome patients has been shown to be significantly lower [35]. ABCG1 has been reported to promote LPL-dependent triglyceride storage in adipocytes [36] and to modulate ATM cholesterol content in obesity and weight loss regimes, leading to an alteration in M1 to M2 ratio [37]. On the other hand, the ABCA1 transporter, which plays a key role in the first steps of the reverse-cholesteroltransport pathway by mediating lipid efflux from macrophages, stimulates the production of more monocytes, leading to an exacerbation of inflamed-tissue macrophages [38]. These observations could explain the increased levels of ABCA1 and reduced ABCG1 expression in the PBMCs of T2DM observed in this study.
Adipose tissue scavenger receptors (SR-A and LOX-1) are strongly associated with insulin resistance [20]. The PBMCs of T2DM displayed increased expression of the major scavenger receptor responsible for modified lipid uptake: SR-A1. Similarly, an increase in CD36 expression in T2DM has recently been demonstrated [39]. Increased CD36 expression in both macrophages and adipocytes of the adipose tissue induces inflammation in obesity [40]. CXCL16 can serve as an adhesion molecule for immune cells expressing CXCR6. It also acts as a scavenger receptor for oxLDL. Although CXCL16 and LOX-1 levels in the PBMCs from T2DM were not statistically significant, increased expression of CXCL16 in M1 macrophages [25] and increased expression of LOX-1 in M2-polarized macrophages have been observed which is consistent with the current model for increased lipid uptake in M2 macrophages [25,41]. However, CXCL16 and LOX-1 levels were significantly higher in T2DM on metformin. Several observations suggest that metformin exhibits anti-atherogenic properties and is associated with reduced cardiovascular morbidity and mortality in patients with diabetes [42][43][44][45][46]. A significant and substantial increase in CXCL16 level in T2DM on metformin demonstrated that this drug has an athero-protective role. This is in light of the finding that CXCL16−/−/LDLR−/− mice have accelerated the progression of atherosclerosis [47].
A significant shortcoming of this study is the heterogeneity of the cells utilized, and purified monocytes could be a better model for such studies. Nevertheless, monocyte purification could activate monocytes bringing about changes in the expression of proinflammatory mediators and phenotypic markers [48]. Moreover, we elected to utilize qRT-PCR to quantitate the phenotypic markers of ATM which is a common practice [49]. Contradictions concerning white blood cells' subset phenotypes and function, as a result of discrepancies in reliable gating strategies for flow cytometric characterization, antibody specificities, and cell purification protocols, have been reported [50,51]. Moreover, the lack of quantitation of the studied phenotypic markers in this study at the protein level is another limitation of our study as mRNA expression levels may not correlate with the protein expression levels. However, the semi-quantitative Western blotting technique requires large quantities of PBMNCs; hence, the quantification of proteins using tandem mass spectrometry could be a better alternative for future studies. The correlation of the phenotypic markers with plasma cytokines represents another future study as well. Another shortcoming of the study is the statistically significant differences in the age of normal-weight subjects compared with the age of obese participants with T2DM and participants with T2DM on metformin. However, there was no significant statistical correlation between the mRNA expression of the phenotypic markers measured in this study and the subjects' age.

Conclusions
Data from this exploratory study suggest the presence of a PBMC phenotype that shares a similarity between M1 and M2 phenotypes and could represent an MMe macrophagelike phenotype. Metformin modulates the phenotypic characteristics of PBMCs resulting from metabolic stress present in T2DM. Future studies will attempt to quantitate these phenotypic markers in isolated monocytes from T2DM at the mRNA and protein levels.