Introduction

Diabetes mellitus (DM) refers to a group of metabolic disorders characterized by hyperglycemia and defect in metabolism of macromolecules [1]. According to the international diabetes federation (IDF) reports, about 415 million people suffered from diabetes worldwide in 2017, and this number will rise to 645 million people by 2040 [2, 3]. The etiology of DM varies from the defect in insulin production to defect in response to insulin or both [4]. Type 2 diabetes (T2D) is the most common form of diabetes that mainly develops in adults [5, 6]. T2D starts with postprandial hyperglycemia and insulin resistance followed by the compensatory response of pancreatic β-cells for insulin production, which in turn, lead to decreased β-cell mass [4]. Therefore, it can be concluded that T2D is a combination of insulin resistance and insulin deficiency conditions. T2D is associated with several complications that mainly include: hyperosmolar coma, nephropathy, neuropathy, retinopathy, and cardiovascular disease [7]. Therefore, this metabolic disorder is one of the main factors involved in mortality and morbidity worldwide, and out of the 56.4 million deaths in 2015, about 1.6 million were due to T2D [8]. Since T2D develops as a result of the interaction between genes and environment [9], investigation of gene regulating factors is among the issues of interest to researchers.

It is documented that the majority of the mammalian genome is transcribed to the non-coding RNAs (ncRNAs) that play an important role in the regulation of gene expression [10]. Usually, ncRNAs are categorized according to their size into two major groups: long non-coding RNAs (lncRNAs, length > 200 bp) and small non-coding RNAs (small ncRNA, length < 200 bp) [11, 12]. lncRNAs can be also categorized by their structure, function and genomic location. Structurally, lncRNAs are divided into linear RNAs and circular RNAs (CircRNAs). Furthermore, small ncRNA are also composed of two groups that involved in translation, transcription and gene regulation (Fig. 1) [13, 14]. An increasing number of experimental investigations are providing evidence that altered expression of ncRNAs patterns is associated with several human diseases such as T2D [15, 16] and human malignancies [17, 18].

Fig. 1
figure 1

Schematic representation of ncRNAs classification. ncRNAs are categorized according to their size into lncRNAs and small ncRNAs. LncRNAs can be also categorized by their structure, function and genomic location. Structurally, they are divided into linear RNAs and CircRNAs. tRNA transfer RNA, piRNA piwi RNA, siRNA small interfering RNA, rRNA ribosomal RNA, cisRNA cis-acting RNA, transRNA trans-acting RNA, ceRNA competing endogenous RNA, eRNA enhancer- derived RNA, NAT natural antisense transcript, TUCRNA transcribed ultraconserved RNAs, lincRNA long intergenic noncoding RNA

Full size image

CircRNAs are a new class of RNAs that in contrast to linear RNAs have a covalently closed loop in their structure that protects them against RNA exonucleases [19]. The history of circRNAs dates back to 1976, when RNAs with covalently closed loop were discovered in some plant viroids [20]. Further investigations showed that the genome of the hepatitis delta virus is composed from circRNAs, and therefore it was thought that circRNAs are viroid [21]. Later research on human tumor suppressor genes showed transcripts with different exon arrangement compared to parent gene that were actually circRNAs [22]. However, due to the technical limitations as well as structural characteristics of circRNAs, many aspects of these molecules have remained unknown, but with the advent of new techniques such as advanced sequencing and bioinformatics tools, circRNAs have become one of the interesting research topics. CircRNAs are involved in gene expression regulation through mimicking of several roles including: microRNA (miRNA) inhibition, the sequestering of RNA-binding protein, and nuclear transcriptional regulators [23, 24].

In this narrative review, a systematically literature review was performed by Pubmed, Google Scholar, and Web of Science for English up to 4 June 2019; using the terms “T2D”, “diabetes” and “circRNAs”. Actually, we discussed the properties and molecular functions of circRNAs and their clinical relevance as potential biomarkers for T2D. Moreover, we highlighted the involvement of circRNAs in the β-cell function and T2D-related complications.

Biogenesis and classification of circRNAs

In eukaryotic systems, messenger RNA (mRNA) is first synthesized as pre-mRNA and subsequently undergoes a further process by splicing which removes its introns. In the canonical splicing process, there are three strategic positions that include: GU sequence at the 5′ end of the intron that plays the role of donor site, the AG sequence at the 3′ end of intron as an acceptor site, and a highly conserved adenosine, located 20–50 nucleotides upstream of the acceptor position [25]. In summary, the splicing procedure starts with the formation of a ribonucleoproteins complex from the five small nuclear RNAs (snRNAs) and their related proteins followed by enzymatic cleavage, thereby the intron is deleted and adjacent exons are joined together [26].

The circRNAs are transcribed by RNA polymerase II mainly from coding regions of genes [27]. In addition, circRNAs may be produced by introns, 5′ untranslated region (5′ UTR), and 3′ untranslated region (3′ UTR) or intergenic regions of genes [27, 28]. Indeed, circRNAs are formed by the process known as back-splicing. In this process, 5′ splice site (donor site in downstream) is linked to the 3′ splice site (acceptor site in upstream) [29]. Previous evidence revealed that inhibition of canonical spliceosome complex by isoginkgetin, decreases the levels of pre-mRNA and circRNAs [30]. Therefore, it can be concluded that circRNA production depends on canonical spliceosome.

There are three models that describe the circRNA formation, namely intron pairing-driven circularization [31], lariat-driven circularization (exon skipping) [31], and RNA binding proteins (RBPs) [32, 33] (Fig. 2). The intron pairing-driven circularization model is based on the pairing between complementary sequences (mainly Alu elements) within the introns that results in the correct positioning of splicing site and followed by nucleophilic attack and circRNA release [34]. There is evidence that approximately 20% of the total circRNAs loci have intron with complementary sequences [31], so other mechanisms are involved in the circularization process. Lariat-driven circularization known as the exon skipping model is proceeding by rotation of downstream exon and interaction with the upstream exon that subsequently release an exon-containing lariat. In the next step, the formed lariat undergoes back-splicing and produces circRNAs [34]. One of the key factors in circularization of RNAs is RBPs. In this model, RBPs interact with their binding sites within the flanking introns followed by intron pairing and subsequent circularization [34]. Generally, the back-splicing process can be affected by several genomic features. For example, the length of exons, especially in single-exon circRNA, as well as the presence of reverse complementary sequence, induces the circRNA formation [35]. However, the stability of intron base pairing has a negative effect on the circRNA production [36]. Furthermore, previous studies have indicated that RNA-binding proteins -as a transregulators- can stimulate or inhibit the circRNA formation [37].

Fig. 2
figure 2

Different models of circRNA biogenesis. a Lariat-driven circularization (or exon skipping model) proceed by interaction between downstream and upstream exons that result in exon-containing lariat. In the next step, the formed lariat undergoes back-splicing and circRNA was formed. b Intron pairing-driven circularization model is based on the pairing between introns that continued by nucleophilic attack in splicing site and release of circRNA. c Circularization model based on the RBPs. In this model interaction between introns are mediated by RBP and thereby back-splicing is facilitated

Full size image

Since circRNAs are produced by different regions of genes, they can be classified as: exonic circRNAs (ecircRNAs) [27, 31, 38], circular intronic RNAs (ciRNAs) [28], retained-intron or exon–intron RNAs (EIciRNAs) [39, 40] and intergenic circRNAs [27]. EcircRNAs are the most common form of circRNAs that involve approximately up to 80% of the total circRNAs and are mainly located in the cytoplasm [27, 35].

Molecular functions of circRNAs in human genome

There are several functions proposed for circRNAs that are mainly categorized into five types, including: (I) regulation of gene splicing and transcription, (II) miRNA sponging, (III) protein sponging, (IV) affecting the protein function and (V) regulation of protein translation [41].

Regulation of gene splicing and transcription

In general, the mechanism of transcription regulation by circRNAs is different from that by many linear lnRNAs, in which lncRNAs are involved in assembling chromatin-regulatory complexes in specific gene loci and also interact with several proteins simultaneously. However, these mechanisms have not been described for circRNAs [41]. An in vitro study on Arabidopsis thaliana revealed that the circRNAs bind to the sequence from which it has been transcribed, opens DNA locally and is then paired with one strand of DNA. This scenario leads to the formation of a circRNA:DNA R-loop that inhibits the transcription process in elongation phase [42].

In mammalian models, circRNAs are among the factors that regulate the transcription of linear RNAs through affecting RNA polymerase II activity in initiation and elongation phases [43]. It has been documented that intronic circRNAs, such as ciRNAs and EIciRNAs can affect the expression of their host genes. Accordingly, c-sirt7, produced by lariat mechanism, can affect the polymerase II and thereby decrease parental gene ankyrin repeat domain 52 (ANKRD52) as well as sirtuin 7 (SIRT7) [44].

miRNA sponge

miRNAs are referred to small non-coding RNAs (18-24 nucleotides) that are involved in gene expression regulation at the post-transcriptional regulation stages [45]. The miRNAs interact with the 3′ UTR of the target mRNA and subsequently leads to translation suppression, degradation or deadenylation of mRNAs [46, 47]. Recently, studies have shown that circRNAs have several binding sites for miRNAs and thereby can suppress their inhibitory effects or sponge them [48]. Accordingly, there is several pieces of evidence that overexpression of circRNAs could induce the expression of miRNAs targets and vice versa [49].

One of the well-known circRNAs that has sponging effects for mir-7 is human cerebellar degeneration related protein 1 antisense (CDR1as). This circRNA contains over 70 binding sites for mir-7 and plays an important role in neuronal development (Fig. 3) [49]. Therefore, CDR1as (has-circ-0001946) overexpression is followed by inhibition of the miR-7 activity and subsequently increased expression of miR-7 target genes [49].

Fig. 3
figure 3

Binding energy for all binding sites within the CDR1as/miR-7 sponge. Red points show the binding energy for each predicted binding site in the CDR1as, as predicted by RIsearch2 [128]. The lower the binding energy is, the stronger the binding site. The box plot shows the distribution of those predicted binding energies for miRNAs with experimentally verified targets. (Color figure online)

Full size image

Another example for miRNAs sponging is cirSRY, a testis specific circRNA. This circRNA has 16 conserved binding sites for miR-138 [24, 50, 51]. Furthermore, cir-ZNF609, mm9-circ-012559 and circ-BIC6 are other examples of circRNAs with sponging activity for miRNAs [52,53,54]. On the other hand, accumulating evidence suggests that circRNAs are more effective than linear RNAs in miRNAs sponging [55, 56].

Protein sponging

CircRNAs can participate in sponging of RBPs and thereby affect the levels of proteins. The MUSCLEBLIND protein (MBL) is encoded by mbl gene of Drosophila and is required for differentiation and development of muscle and eye photoreceptor cells [57]. Increased concentration of MBL protein promotes back-splicing in parental pre-mRNA and produces circ-MBL. Low concentrations of MBL, on the other hand, induce the canonical splicing of Mbl mRNA. Therefore, it can be concluded that MBL protein has self-regulating properties [51]. The interesting point in this regard is MBL binding sites in the structure of circMBL that enable it to sequester MBL. In other words, circ-MBL regulates its own expression through MBL sponging.

Protein interaction

CircRNAs can also interact with proteins and thereby affect several phenotypic properties. The PeBoW complex is involved in the maturation process of pre-60 ribosome [58, 59]. Circ-ANRIL interferes with this process and thereby decreases the translation potential of cells [60].

HuR protein is one of the best known of RBPs participate in RNAs splicing and nuclear export [61]. One of the main targets of HuR is polyadenylate-binding nuclear protein 1 (PABPN1) mRNA. PABPN1 protein is required for poly adenylation of 3′ terminal of mRNA in eukaryotic species and therefore is involved in mRNAs stability. There is evidence that circPABPN1 bind to the linear mRNA of PABPN1 in competition with HuR, followed by inhibition of PABPN1 translation as well as other mRNAs dependent on PABPN1 [61].

Circ-FOXO3 is encoded by Foxo3 gene and is involved in cell cycle. Indeed, cyclin-dependent kinase2 (CDK2) as a member of CDK family is essential for the transition from G1 to S phase in the cell cycle. The p21 can bind to CDK2 protein and inhibit the cell cycle in G1 stage. Circ-FOXO3 could interfere with the CDK2-p21 interaction, release CDK2, and subsequently induce cell cycle [62, 63].

Regulation of protein translation

A challenging question that could be posed at this juncture is: Are circRNAs translated? Generally, 7-methylguanosine structure in 5′ end (5′ cap) and polyA tail in 3′ end of mRNAs are required for the translation process [41]. CircRNAs do not contain these structures and their translation is beyond expectation. However, recently it has been shown that some of the circRNAs can be translated into proteins. This was suggested in a study in 1995, where internal ribosome entry site (IRES) of encephalomyocarditis virus was shown to be able to translate an artificial circRNA [64]. The human circ-ZNF609 is highly expressed in myoblasts and participates in myogenesis regulation. Circ-ZNF609 contains an open reading frame with 753nt length and can be translated into a protein [65]. Indeed, untranslated region of circ-ZNF609 mimics the IRES and promotes 5′ cap-independent translation [66].

One of the most common modifications of eukaryotic RNAs is methylation of adenosine into N6-methyladenosine (m6A) by methyltransferase complex [67, 68]. This modification has several effects on the metabolism of mRNAs, such as localization, translation, splicing and degeneration of mRNAs [69, 70]. Furthermore, previous studies have revealed that m6A in 5′ UTRs increase the potential of translation through the 5′-cap independent process [71, 72].

Molecular functions and clinical relevance of circRNAs in T2D

Recent studies suggest that the deregulation of circRNAs could be associated with human disorders such as coronary artery disease [73], neurodegenerative and autoimmune diseases [74, 75], and several types of cancer [76]. In this section, we will highlight the involvement of circRNAs in the β-cell function and T2D-related complications. We will also discuss the possibility of using circRNAs as a biomarker for T2D diagnosis.

CircRNAs and β cell function

Genome wide association studies have shown that most risk loci in type 2 are implicated in islet function [77]. circRNAs are widely expressed in human cells including both α- and β-cells of pancreas and could act as novel regulators of β-cell activities [78]. It has been reported that 10,830 high-confidence circRNAs expressed in human α, β, and exocrine cells. The most highly expressed candidates are MAN1A2 (hsa_circ_0000118), RMST, and circ-HIPK3 (hsa_circ_0000284) [79]. Lisa Stoll et al. reported that the expression of circ-HIPK3 and ciRS-7/CDR1as were reduced in the islets of diabetic db/db mice and knock-downing of circ-HIPK3 resulted in increased apoptosis and decreased β-cell proliferation [78]. Moreover, their transcriptomic analysis revealed that circ-HIPK3 acts by sequestering a group of miRNAs, including miR-124-3p and miR-338-3p, and by regulating the expression of key β-cell genes, such as Slc2a2, Akt1, and Mtpn [78]. It has been also demonstrated that both insulin content and secretion were significantly increased by overexpression of CDR1as in islet cells [80]. CDR1as was recently revealed to act as a powerful sponge for miR-7, therefore could upregulate miR-7 target genes such as Myrip in islet cells and thereby regulates insulin granule secretion [80]. Figure 4 summarizes the molecular roles of circ-HIPK3 and CDR1as circRNAs in β-cell function.

Fig. 4
figure 4

Schematic presentation of the circRNAs role in β-cell functions. Circ-HIPK3 acts as a sponge for miR-124-3p and miR-338-3p, and thereby regulating the expression of key β -cell genes including Slc2a2, Akt1, and Mtpn. In addition, Cdr1as acts as a powerful sponge for miR-7, and up-regulates miR-7 target genes such as Myrip and Pax6 in islet cells

Full size image

Association of circRNA expressions with T2D pathogenesis and its related complications

Complications of diabetes including diabetic nephropathy, diabetic retinopathy, and the risk of cardiovascular diseases are the leading causes of increased morbidity and mortality of diabetes patients and posing a huge burden on the health care system [81]. Recent data have demonstrated that circRNAs play crucial roles in vascular endothelial cell dysfunction and diabetic neuropathic pain [82, 83]. Accordingly, deregulation of circRNAs could be associated with T2D pathogenesis and related complications (Fig. 5).

Fig. 5
figure 5

The association of circRNA expression levels with T2D pathogenesis and its related complications. Deregulation of circRNA levels with sponging effects on miRNAs, induces or suppress the expression of miRNAs targets

Full size image

Cardiovascular complications

Impaired proliferation of vascular smooth muscle cells (VSMCs) plays an important role in the atherosclerotic promotion in diabetes [82]. It has been reported that the expression of circ-WDR77 in high glucose treatment VSMCs is significantly up-regulated compared to the normal group. This circRNA regulates VSMCs proliferation and migration via targeting miR-124/FGF2 axis [83].

Several investigations have shown that increased blood glucose impairs vascular endothelial cells and could develop cardiovascular complications in diabetic patients [84, 85]. Fei-Fei et al. demonstrated that high glucose level dramatically altered expression profiles of 95 circRNA including NC_000007.14:149492859_149558839 and has-circ-0004182 in endothelial cells [86]. This indicates the contribution of circRNAs in the pathogenesis of hyperglycemia-induced endothelial cells injury and diabetic cardiovascular diseases.

Interestingly, it has been reported that cZNF609 silencing decreases retinal vessel loss and suppresses pathological angiogenesis in vivo, and also protects endothelial cell against oxidative stress [87]. Furthermore, down-regulation of cZNF609 could elevate endothelial cell migration and tube formation, and also protected endothelial cell against oxidative stress. This circRNA could act as an endogenous miR-615-5p sponge, which led to the increased expression of MEF2A level [87]. MEF2A activation is strongly correlated with vascular endothelial cell function [88].

CircRNA-000203 was shown to be up-regulated in the diabetic mouse myocardium and in Angiotensin II (Ang II)-induced mouse cardiac fibroblasts [89]. Moreover, over-expression of this RNA could increase expression levels of Col1a2, Col3a1 and α-SMA in mouse cardiac fibroblasts [89]. Indeed, circRNA-000203 could block the interactions of miR-26b-5p and 3′UTRs of these genes by molecular sponge formation [89]. Taken together, that suppressing the function of miR-26b-5p contributes to the pro-fibrosis effect of circRNA-000203 in cardiac fibroblasts. Furthermore, Zhou B et al. reported that circRNA-010567 was markedly up-regulated in diabetic mice myocardium and cardiac fibroblasts treated with Ang II [90]. They revealed that circRNA-010567 silencing could up-regulate miR-141 and down-regulate TGF-β1 expression, and suppress fibrosis-associated protein resection in cardiac fibroblasts, including Col I, Col III and α-SMA [90]. These findings show that the circRNA-010567/miR-141/TGF-β1 axis plays an important regulatory role in the diabetic mice myocardial fibrosis model.

Neuropathic pain and diabetic retinopathy

Increasing evidence has verified that neuropathic pain is one of the most common diabetic complications. Its incidence is about 30% in T2D cases and is growing with the duration of diabetes [91]. It has been reported that circ-HIPK3 knockdown alleviated neuropathic pain in STZ-induced diabetic rats. Circ-HIPK3 could interact with miR-124 and negatively regulated its expression. Indeed, this circRNA regulates neuropathic pain progression through miR-124 [92]. Moreover, circ-HIPK3 silencing or overexpressing changed retinal endothelial cell viability, proliferation, migration, and tube formation in vitro. It has been revealed that circ-HIPK3 silencing in vivo alleviated retinal vascular dysfunction, as shown by decreased retinal acellular capillaries, vascular leakage, and inflammation. Circ-HIPK3 could serve as endogenous miR-30a-3p sponge retinal endothelial cells and inhibit its activity [93]. Inhibition of miR-30a-3p led to increased Vascular Endothelial Growth Factor-C (VEGF-C), FZD4, and WNT2 expression which have been shown to be involved in diabetic retinopathy [94, 95].

The circRNAs could also play a critical role in the responses of human umbilical vein endothelial cells (HUVEC) to the glucose stress [96]. It has been reported that circRNA-001175 could be involved in the protection of HUVECs against high glucose stress. Xiaoyan Pei et al. showed that the HUVECs glucose treatments gradually decreased the expressions of circRNA-00117 [97]. Moreover, they reported that the up-regulation of circRNA-001175 inhibits the HUVECs apoptosis and also increases the tubule formation ability of that under high glucose situation [97]. Furthermore, it has been recently observed that high glucose treatment could induce the expression of circ-0054633 and miR-218 levels in HUVEcs. In addition, circ-0054633 expression suppressed the injurious effects of high glucose to HUVEcs by targeting miR-218 [98]. Long Pan et al. reported that overexpression of circ-0054633 could reverse high glucose-induced cell proliferation, migration and angiogenesis inhibition by targeting the miR-218/ROBO1 and miR-218/HO-1 signals. Indeed, elevated expression of circ-0054633 increased HO-1 level by targeting miR-218, which resulted in reducing the apoptosis of HUVEcs [98].

A current investigation reported that circ-0005015 facilitated retinal endothelial angiogenic function via regulating endothelial cell proliferation, migration, and tube formation [99]. Circ-0005015 could act as miR-519d-3p sponge, leading to increased MMP-2, STAT3 and XIAP expression [99]. These genes are involved in many pathways, including regulation of vascularization [100], diabetic retinopathy [101], and neuroprotection [102].

T2D related-chronic inflammation and oxidative stress

T2D is a metabolic disorder characterized by chronic, low-grade inflammation which can lead to insulin resistance and impaired glucose tolerance [103]. The expression of circRNAs, and their association with chronic inflammation in T2D was evaluated in some investigations [98, 104]. It has been reported that circ-ANKRd36 expression level is positively correlated with IL-6 and TNF-α in diabetic patients, suggesting an association between circ-ANKRD36 and inflammation in T2D [104]. Yuan Fang et al. prediction analysis showed that circ-ANKRd36 may be involved in T2D and inflammation-associated pathways via interaction whit miR-3614-3p, miR-498 and miR-501-5p [104]. Moreover, it has been indicated that circ-0054633 expression may suppress cellular oxidative stress and inflammation [98]. This circRNA could target miR‑218 and inhibits its expression [98]. A previous study showed that miR-218 accelerated high glucose-induced podocyte apoptosis through directly downregulating HO-1 [105]. HO-1 has been demonstrated to have a protective effect against high glucose-induced cell toxicity, including oxidative stress and inflammation [106]. Furthermore, Jun-Hui Yang et al. revealed that deregulated expression of mmu-circRNA-34132, mmu-circRNA_017077 and mmu-circRNA-015216 in the Substantia Nigra and Corpus Striatum of Nrf2-Knockout Mice models could be involved with Nrf2-mediated neuroprotection against oxidative stress [107]. They reported that mmu-circRNA-34132 may be a potential regulator of Nrf2-mediated protection for diabetes mellitus and Nrf2-mediated defense against reactive oxygen species (ROS) in hearts [107]. In addition, it has been reported that the has-circ-0068087 level is up-regulated in high glucose treated-HUVECs. Down-regulation of this circRNA, suppressed the inflammation and endothelial cell dysfunction by sponging miR-197 and as a result, suppression of the TLR4/NF-κB/NLRP3 signaling pathway [108].

Diabetic nephropathy

Diabetic nephropathy is one of the most common causes of chronic renal failure in diabetic patients [109]. Previous investigations demonstrated that a large number of circulating circRNAs can be reliably detected in blood of acute kidney injury patients and in renal tissues [110, 111]. It has been validated that circRNA-15698 was up-regulated in both db/db mice and high glucose-exposed mouse mesangial cells [112]. CircRNA-15698 could act as a molecular sponge of miR‐185, and then positively regulated the transforming growth factor‐β 1 (TGF‐β 1) protein expression [112].

CircRNAs and wound healing

Impaired wound healing has been a major public health issue in patients with DM [113]. Currently, a study shows that the circ-Amotl1 can accelerate wound healing by facilitated Stat3 nuclear translocation and binding to Dnmt3a promoter, which enhanced Dnmt3a expression [114]. Zhen-Guo Yang et al. showed that the increased Dnmt3a level could methylated the promoter of miR-17 and thereby decreasing its levels. Interestingly, they reported that Stat3, similar to Dnmt3a, was a target of miR-17-5p. Therefore, decreased miR-17-5p levels would increase expression of Dnmt3a and Stat3, which led to increased cell adhesion, migration, proliferation, survival, and wound repair [114].

CircRNAs as a potential biomarkers in T2D

Most of the classical biomarkers for T2D diagnosis are useful only after the establishment of disease and fails to predict the prediabetic cases or diabetes complications at early stage of disease. Accordingly, the novel and more sensitive and complications-related biomarkers are needed [115, 116]. Due to the circular structure of circRNAs and lacking of free 5′ and 3′ ends, these molecules are highly resistant to exonuclease RNase R [117]. Indeed, circRNAs are stable molecules in body fluids with an average half-life of 48 h in plasma, much longer than 10 h of the average value of mRNAs [118]. Moreover, circRNAs are tissue and cell specific, additionally; the abundance of circRNA is higher in body fluid including blood, cerebrospinal fluid, saliva and urine [119]. Therefore, there is a growing trend in exploring the use of circulating and tissue profiling of circRNAs as a potential biomarker for the diagnosis and prognosis of a range of human disorders including diabetes [93, 119]. Table 1 represents the deregulated levels of important circRNAs in T2D.

Table 1 Different expression level of circRNAs in T2D patients
Full size table

Circ-HIPK3

It has been shown that the expression level of circ-HIPK3 is highly abundant in serum from diabetes patients who suffered from neuropathic pain compared to controls. Moreover, the circ-HIPK3 expression was up-regulated in serum and dorsal root ganglion from STZ-induced diabetes rats compared with control rats [92]. Kun Shan et al. showed that retinal level of circ-HIPK3 in diabetic mice was significantly higher than that in the non-diabetic controls [93]. Thus, circ-HIPK3 has great potential to become a diagnostic or predictive biomarker for neuropathic pain of T2D.

cZNF609

Chang Liu et al. revealed that cZNF609 expression level in the fibrovascular membranes of diabetic patients was significantly higher than that in the idiopathic epiretinal membranes of non-diabetic patients. In addition, they reported that cZNF609 expression was up-regulated in the plasma fraction of peripheral blood of diabetic patients compared with non-diabetic controls [87].

Circ-0054633

Zhenzhou Zhao et al. evaluated the expression levels of some circRNAs in the peripheral blood from control, pre-diabetes and T2D subjects [120]. They reported that the levels of circ-0124636 and circ-0139110 expression among the three groups presented no significant differences. However, the expression levels of circ-0018508 in the pre-diabetes and T2D groups were higher than that of the control group. They also revealed that, the expression levels of circ-0054633 and circ-0068087 were increased gradually from the control group to the T2D subjects. After ROC curve analysis, they demonstrated that circ-0054633 has the potential to be used as a diagnostic biomarker for pre-diabetes and T2D in clinical practice [120]. In the line with this finding, Sally et al. reported that plasma level of circ-0054633 was significantly increased gradually from controls to the T2D patients [121].

Circ-RNA11783-2

Xuejie Li et al. determined the expression profile of circRNAs in the peripheral blood of coronary artery disease and T2D patients [122]. They did not observe a significant difference for the expression levels of circ-0009036 and circ-0054129 between groups. However, they showed that the levels of circ-RNA11806-28, circ-RNA6510-1 and cir-cRNA11783-2 were lower in both the T2D and CAD groups compared with the controls. Subsequently, they validated the significance of differential circ-RNA11783-2 expression in the third cohort including 60 control, 64 T2D and 81 CAD subjects. Their result showed that circRNA11783-2 was down-regulated in both the CAD and T2D groups compared with the controls [122].

Circ-0005015

Shu-Jie Zhang et al. performed circRNAs microarray to investigate differential expression profile of that between diabetic and non-diabetic human retinas [99]. They identified 529 differentially expressed circRNAs, including 356 up-regulated and 173 down-regulated circRNAs compared with non-diabetic controls. In addition, they revealed that circ-0005015 expression was significantly up-regulated in the plasma fraction of diabetic retinopathy patients compared to cataract patients, and healthy controls [99].

Circ-ANKRD36

It has been reported that there is an association between the expression profiles of circRNAs in peripheral leucocytes of patients with T2D and inflammatory cytokines [104]. The RNAseq analysis revealed 220 differentially expressed circRNAs, including 107 up-regulated and 113 down-regulated circRNAs in diabetic subjects compared with non-diabetic controls. Among these, the expression of circ-ANKRD36 was up-regulated in patients with T2D and its expression level was positively correlated with IL-6, glucose and glycosylated hemoglobin levels [104].

Other circRNAs

Yonghao et al. reported that thirty circRNAs were significantly unregulated in the serum of T2D retinopathy patients compared with the serum from both T2D and control subjects [123]. Further, the expression of 7 circRNAs including circRNA-063981, circRNA-404457, circRNA-100750, circRNA-406918, circRNA-104387, circRNA-103410, and circRNA-100192 were significantly elevated in T2D retinopathy patients relative to the T2D or control groups [123].

Several investigations have shown that T2D is closely associated with the onset and progression of depression [124, 125]. Guangjian Jiang et al. showed that 183 circRNAs were significantly up-regulated, whereas 64 were down-regulated in the whole blood samples from T2D with depression group compared with that in the T2D group [126]. They suggested that differentially expressed circRNAs could clarify the pathogenesis of depression in patients with T2D [126]. Interestingly, it has been reported that Baduanjin, a traditional Chinese exercise therapy, could effectively ameliorate the symptoms of depression and blood glucose levels in diabetic patients with depression by regulating the expression of circRNAs [127].

Conclusions and perspectives

CircRNAs are involved in the genome regulation through mimicking of several roles including: regulation of gene splicing and transcription, miRNA and protein sponging and affecting the protein function and translation. Therefore, an increasing number of experimental investigations are providing evidence that the altered expression of circRNAs could be associated with several human diseases such as T2D. However, the molecular functions and clinical relevance of circRNAs are not yet well elucidated in T2D and there is still no available clinical relevance of circRNAs in diabetic patients.

This review provides current knowledge about the properties and functions of circRNAs and their clinical relevance as potential biomarkers for T2D. In addition, we highlight the involvement of circRNAs in the β-cell function and T2D-related complications. Further in vitro and in vivo studies are required to confirm and elucidate the functional roles of circRNAs in the establishment and development of T2D pathogenesis and the possibility using of these ncRNAs as a potential therapeutic targets and biomarkers for T2D.