In this study, we show that circulating sncRNAs are associated with prevalent DKD in people with type 2 diabetes. Importantly these sncRNAs do not only represent miRNAs but also other classes of small non-coding RNAs. In addition, these sncRNAs correlate with proteins that function in pathways known to be involved in DKD/kidney function.

MicroRNAs are involved in the post-transcriptional regulation of gene expression and are known to be involved in crosstalk between (distant) cells and tissues [2]. As such, there is an ever-growing interest to study their role in disease development and as biomarkers of disease and disease progression.

In our study one miRNA, miR-143-5p, was associated with both DKD and UACR. Levels of this miRNA are lower in both DKD and persons with high UACR. miR-143-5p remained significant in the full models, suggesting it might have biomarker potential. Previous in vitro studies showed that this miRNA is associated with vascular pathology [19], an impaired glomerular filtration barrier [20], and with mesangial cell proliferation and fibrosis [21, 22]. miR-143-5p plasma levels associate with two proteins but neither of those have a clear relationship with kidney function nor were they predicted targets for miR-143-5p (miRDB, miRTarBase, MiRBase). In the full model three additional microRNAs were significant, miR-455-3p, miR-122-3p, and miR-193b-3p. Previously it was shown that miR-455-3p was down regulated in human mesangial cells (HMC) and human proximal tubule epithelial cells (HK-2) under DKD stimulating conditions [23]. Interestingly, it was also shown that overexpression of this miR could alleviate the pathological changes such as renal fibrosis in a rat DN model [23]. In a mouse mesangial cell line miR-455-3p improved cell proliferation, inflammation and ECM accumulation induced by high glucose treatment [24]. Involvement of this miR in fibrosis in other tissues like liver and lung has also been reported. A previous report indicated that miR-122, a liver specific miR associated with liver injury, was reduced in patients with end stage renal disease but not in CKD when compared to healthy controls [25]. miR-193b-3p was identified as a potential biomarker for pre-diabetes and also for CKD after nephrectomy for renal cell carcinoma [26, 27].

In addition to the association with DKD we also found many miRNAs associated with eGFR and or UACR. A large number of which remained significant in the full models. Interestingly, of the 154 miRNAs found in the base models for eGFR and UACR only two of them were associated with both phenotypes suggesting that the associated miRNAs are specific to either defect in kidney function but not both.

PIWI-interacting RNAs are involved in silencing of transposons but they can also be involved in the regulation of gene expression in a miRNA-like manner [28]. One piRNA, piRNA-019825 (DQ597218), was significantly associated with DKD. Previously, it was shown that this piRNA is one of the most abundant piRNA species in urine and plasma [29]. In our proteomics analysis to identify potential targets or pathways we found eleven proteins whose expression was correlated with the levels of piR-019825 including cystatin C, a known marker for kidney function. Four of the negatively associated proteins are also strongly associated with eGFR. This is in line with a putative miRNA like function for piR-019825 in which lower piRNA expression would result in higher protein levels, lower eGFR, and increased risk of DKD. This is further corroborated by the fact that increased levels of two of these proteins were previously found to increase risk of CKD or end-stage renal disease (ESRD) in prospective studies in persons with diabetes (TNF sR-I (TNFRSF1A), NBL1 (DAN) [30,31,32,33]. Whereas the other negatively associated proteins we found associated with this piRNA (CST3, sCXCL16, CFD, PLA2G2A) have almost all also been implicated previously in kidney dysfunction [34,35,36]. Interestingly, we also find various other piRNAs associated with either eGFR or UACR (10-14%). Most of which remained significant in the full models. Suggesting that they may be useful early markers of DKD. Of note, eGFR-associated piRNAs often show their highest expression in urine, whereas UACR-associated piRNAs show low expression in urine, but often high levels in tissues like pancreas, liver, and colon.

Small Nucleolar RNAs (snoRNA) are predominantly localized to the nucleolus where they have a function in rRNA biogenesis, structure, and activity [37]. There are two major types of snoRNAs containing either box C/D or box H/ACA motifs. In our study, we only detected box C/D snoRNAs associated with DKD. Box C/D snoRNAs are involved in 2′-O-methylation of specific sequences in target RNA sequences, mainly rRNA, thereby affecting RNA structure, stability, and interactions. In addition, other functions have been ascribed to the snoRNAs such as mRNA splicing and editing. SnoRNA-derived fragments have also been described and it has been speculated that they exert piRNA or miRNA-like functions [38].

Although snoRNAs represent only ten percent of the total circulating sncRNA transcriptome in our study they represent seventy-one percent of the sncRNAs found associated with DKD suggesting that snoRNAs may function as early biomarkers of DKD. SnoRNA U8 was associated with all three phenotypes but has not been linked to kidney function previously. SnoRNAs SNORD118, SNORD24, SNORD107, and SNORD87 were associated with DKD and UACR but not eGFR. SnoRNAs U8 and SNORD118 both remained significant in the full models. SNORD87 and SNORD107 associate with the same set of proteins, including UFM1, NAMPT, PDGFB, and the two chemokines CCL27 and CCL28, which bind to the same receptor (CCR10). A recent report by Chen et al. showed that post-transcriptional stabilization of the CCL28 mRNA by N6-methyladenosine (m6A) modification increases protein levels of this chemokine and reduces kidney damage after ischemia-reperfusion injury [39]. Given the known role of snoRNAs in post-transcriptional RNA modification and stabilization via 2′-O-methylation this might provide a putative mechanism and explanation for the observed snoRNA-protein associations. The other proteins associated with these snoRNAs (UFM1, NAMPT, PDGFB) have also been implicated in kidney function previously [40,41,42,43]. Further studies are needed to investigate the role of these snoRNAs in DKD. In addition to the snoRNAs described above various other snoRNAs are associated with eGFR and UACR (13–31% of total number of significant sncRNA). Almost all remained significant in the full models. At present, there are no other studies that have investigated the role of snoRNAs in diabetic DKD.

Although other classes of sncRNAs, such as Y_RNA derived fragments, tRNA fragments (tRF), lncRNA fragments, and snRNAs represent roughly twenty percent of all circulating sncRNA detected we did not find any significant associations with DKD. We did find a small number of associations with eGFR and UACR though.

Our study has several strengths. It is one of the largest circulating sncRNA profiling studies studying the role of circulating sncRNAs in DKD. This improves power and reduces the chance of false positives. In addition, it uses untargeted RNA sequencing allowing not only the study of miRNAs but also several other classes of sncRNA showing that especially these other types of sncRNAs associated with DKD and its components eGFR and UACR. The use of early, mainly mild cases of DKD facilitates the identification of early markers of disease and may help to introduce early interventions aiming to preserve kidney function in those at risk.

A limitation of our study is the use of a total RNA isolation procedure and as such we were not able to investigate if the sncRNAs are carried by extracellular vesicles or are associated with carriers such as HDL and Ago-2 [2]. The use of plasma for RNA isolation is another limitation. Although from a biomarker perspective, this is not a limitation, from a functional perspective the role of the sncRNAs we identified needs to be established in the relevant cells and tissues. Another limitation of our study is its cross-sectional, observational design. As such we cannot exclude that (part of) the observed changes and associations are a reflection of the reduced kidney function in the participants.

In conclusion, in this study, we have shown that several circulating sncRNAs are associated with DKD. Interestingly, only a relatively small fraction belong to the well-studied microRNAs. This highlights the potential involvement of other classes of sncRNAs, such as snoRNAs and piRNAs, that have so far not been investigated in relation to kidney function before. Future prospective and functional studies are needed to establish the role of these novel circulating sncRNAs in subjects with diabetes and chronic kidney disease.