Ketogenic Diet and microRNAs Linked to Antioxidant Biochemical Homeostasis

Recently, we demonstrated the capability of the ketogenic diet (KD) to influence the microRNA (miR) expression profile. Here, we report that KD is able to normalize miR expression in obese subjects when compared with lean subjects. By applying two different bioinformatics tools, we found that, amongst the miRs returning to comparable levels in lean subjects, four of them are linked to antioxidant biochemical pathways specifically, and the others are linked to both antioxidant and anti-inflammatory biochemical pathways. Of particular interest is the upregulation of hsa-miR-30a-5p, which correlates with the decrease of catalase expression protein in red blood cells.


Introduction
The progress of the obesity pandemic is still substantially underestimated but is alarming [1]. Obese individuals have a lower quality of life and more risk of developing several clinical problems [2]. Obesity is a chronic status with sub-clinical inflammation which is associated with the abnormal synthesis of cytokine/adipokines, leading to an increase of radical oxygen species (ROS) [3]. Therefore, obesity is not per se a disease; rather, it is a status that contributes to the imbalance of anti-inflammatory and oxidative stress biochemical pathways [4]. Opportune antioxidant defenses counteract the action of ROS in different organs and are improved by diverse biomolecules [5][6][7], functional food and diet nutrients [8,9]. It is worthy of note that, recently, hyperglycemic crisis was linked to oxidative stress in diabetic patients [10]. While ketoacidosis is not safe for human health, it is well known that the ketogenic diet (KD) is safe. KD is a nutritional regimen in which the amount of carbohydrate is maintained at less than 30 g per day [11]. Although its application initially was a therapeutic regimen for refractory epilepsy, today it is often used to lose weight [12]. KD was proved to possess antioxidant and anti-inflammatory properties as well as to regulate obese subjects in stage 1 of the Edmonton Obesity Staging System (EOSS) and their microRNA (miR) expression profile [11,[13][14][15]. The 11 miRNAs analyzed previously in this subject category were normalized with KD when compared to lean subjects. Besides being regulators of the metabolic network in which ROS are always produced, these miRs are also able to counteract inflammatory and oxidative stress [16].
For the n-counter flex of NanoString Technology, 100 ng of RNA/miR was used as input. miRs were then hybridized with an nCounter Human-V3 miRNA Expression Assay CodeSet overnight at 65 • C and as previously described in [22,23]. In order to obtain robust results, the coefficient of variation (CV), expressing the ratio of the standard deviation to the mean and expressed as a percentage, was chosen as 30%. The miRs known to be linked to blood hemolysis were excluded from the analysis [24,25].

In Silico Prediction of hsa-miR Target Genes
In order to identify genes as targets of hsa-miRs from the array analysis, we performed in silico analysis. The in silico identification of the target genes was performed using miRTargetLink Human (https://ccb-web.cs.uni-saarland.de/mirtargetlink/) and DIANA Tools (http://diana.imis.athenainnovation.gr/DianaTools/index.php) databases. This latter database was used to check which miRNA target genes were already validated experimentally.

Statistical Analysis
Prism GraphPad Prism version 5.0 for Windows (GraphPad Software, San Diego, CA, USA) was used to plot the results. Differences within and between groups were evaluated by the t-test and one-way ANOVA followed by a multi-comparison Bonferroni test (* p < 0.05).

Characteristics of Subjects
A total of 43 subjects, divided into categories of obese, lean and on a ketogenic diet (KD), were selected, with numbers for obese subjects of n = 14, lean of n = 17 and KD of n = 12. The subjects' characteristics are described in Table 1. Data are presented as mean ± SD. * Using t test.

In Silico Results
Two different databases were used for the in-silico analysis. Data were compared with respect to the number of target genes experimentally validated in both databases. The results are reported in Table 2. Although similar results were found for two hsa-miRs (in bold in Table 2), the numbers of validated targets found for others hsa-miRs were consistently different. In DIANA tools, the numbers of validated target genes were higher in respect to miRtagertLink Human; therefore, DIANA tools were used for further bioinformatics analysis. , lean (n = 17) and ketogenic diet (KD) subjects (n = 12) based on the most differentially expressed microRNAs. The color and the intensity of the boxes represent changes of gene expression. In the analysis, red represents down-regulated genes and green represents up-regulated genes. Black represents an unchanged expression as evident by the color reference. n-Solver software was used.

In Silico Results
Two different databases were used for the in-silico analysis. Data were compared with respect to the number of target genes experimentally validated in both databases. The results are reported in Table 2. Although similar results were found for two hsa-miRs (in bold in Table 2), the numbers of validated targets found for others hsa-miRs were consistently different. In DIANA tools, the numbers of validated target genes were higher in respect to miRtagertLink Human; therefore, DIANA tools were used for further bioinformatics analysis.

Validated hsa-miR Interaction and Western Blot Analysis of Catalase
Predicted and validated target genes were assessed using DIANA Tools. The new hsa-miR-let7e-5p was found to regulate glutathione peroxidase 7 (GPX7), as shown by string analysis (Figure 2A), as well as in silico 3 UTR interaction ( Figure 2B). In silico 3 UTR regions of tet methylcytosine dioxygenase 3 (TET3) for hsa-miR-520h ( Figure 2D) interaction and 3 UTR regions of superoxide dismutase 2 (SOD2) for hsa-miR-548d-3p ( Figure 2F) are shown, as well as string analysis, in Figure 2C,E, respectively. String analysis showed that 10 proteins are able to physically interact with GPX7 ( Figure 2B) as well as with TET3 ( Figure 2D), SOD2 ( Figure 2F) and catalase (CAT) ( Figure 3D). This latter protein was monitored through Western blot analysis, as shown in Figure 3A. Densitometric analysis was performed and exhibited a significant decrease of CAT protein levels, as shown in Figure 3B. The hsa-miR-30a-5p was found to target 3 UTR regions of CAT, as shown in Figure 3C. However, other target genes were found to be influenced by KD and linked to antioxidant metabolism and inflammatory-related genes, as shown in Table 3. Abbreviations and gene names are described in Table 4.

Discussion
The World Health Organization states that obesity has nearly tripled since 1975. In 2016, more than 1.9 billion adults were overweight. Of these, over 650 million were obese, leading to the following report sentence: "Most of the world's population live in countries where overweight and obesity kills more people than underweight. Obesity is preventable" [26]. How can obesity be preventable? One method is represented by KD. The beneficial effects of KD in reducing the body weight and body mass index of obese subjects over both the short (6 weeks) and long (24 weeks) term has already been proven [12,27]. The administration of KD for a relatively long period is safe and can be considerate a nutritional therapy for weight reduction in obese patients [27]. Besides that, to keep KD beneficial, lifestyle change is mandatory and is the harder part of nutritional intervention during KD, with documented improvement of the endurance exercise capacity, better recovery from fatigue and exercise-induced muscle and organ damage prevention in obese subjects. Besides that, the anti-inflammatory action of physical activity was also recently reviewed [28,29]. Nutraceuticals with antioxidant properties were proposed to help in the treatment of obesity, but they are not enough when taken alone [30]. Several scientific approaches to date have tried to describe this disorder by way of genetic or environmental factors [31]. The role of epigenetics in human diseases has been well described relatively recently. Obesity and epigenetics is a consolidated union [32][33][34], and bariatric surgery induces epigenetic change in obese subjects [35]. We recently reported the influence of the ketogenic diet (KD) on the circulating microRNA (miR) expression profile [15]. The 11 miRs controlling the metabolic network identified so far in subjects on KD were almost normalized and closer to lean subjects. Besides that, new miRs targeting identified genes linked to the homeostasis of oxidant-antioxidant pathways were identified. These latter molecules act as epigenetic regulators and have the peculiarity of regulating gene expression targeting the 3 UTR mRNA region [31]. Predicted and validated target genes were assessed using DIANA Tools. The hsa-miR-let7e-5p has a role in pathogen recognition [36]. A low level of hsa-miR-520h was found in the alteration of the placenta, mediated by oxidative stress [37], and hsa-miR-548d-3p was found to be involved in the control of the homeostasis of oxidative stress damage, the metabolic network and survival pathways [38]. Here, we found an in silico interaction of that miR with glutathione peroxidase 7 (GPX7), tet methylcytosine dioxygenase 3 (TET3) and superoxide dismutase 2 (SOD2). GPX7 is a glutathione peroxidase homolog for which the exact biochemistry is not fully understood [39]. TET3, which is aberrantly expressed in acute myeloid leukemia, promotes DNA oxidation [40,41]. The mitochondria-localized manganese superoxide, SOD2, has a dichotomous role and aids in the regulation of several types of cancers [42]. All those proteins directly or indirectly exhibit a physical interaction with CAT. This latter protein was monitored through Western blot analysis and decreased after KD regime. The hsa-miR-30a-5p was found to target the 3 UTR regions of CAT. It is worth noting that the family of hsa-miR-30, to which hsa-miR-30a-5p belongs, is a promising regulator in both development and disease [41]. In the interplay of oxidative stress, pro-oxidants, and antioxidants, this is already known [43]. In particular, the regulation of antioxidant genes such as SOD, CAT, and GPX was studied in mice models kept in KD. The short time-frame of KD did not affect the SOD expression protein while it significantly decreased both GPX and CAT [44].

Conclusions
Modulating miRs linked to antioxidant and inflammatory states in obese people might be the key to the success, in particular in the long term, of a nutritional program. The reciprocal action of diet and nutrients on anti-oxidant and anti-inflammatory miRs can present tools to predict and follow the success of a nutritional programs.
Author Contributions: R.C. and E.C. designed the study and the experiments; M.C.C., C.L.T. and E.C. carried out the experiments; E.C., A.F. and M.C.C. analysed the data and revised the statistical analyses; P.P., G.L. and L.G. facilitated the experiments; E.C. prepared and edited the manuscript; and all authors approved it.
Funding: This research received no external funding.

Conflicts of Interest:
The authors declare no conflict of interest.