Potential Diagnostic Biomarkers of tRNA-Derived Small RNAs in PBMCs for Nonproliferative Diabetic Retinopathy in Patients With Type 2 Diabetes Mellitus

Abstract

Purpose: This study aimed to reveal the altered expressions of transfer RNA (tRNA)-derived small RNAs (tsRNAs) in peripheral blood mononuclear cells and identify potential diagnostic biomarkers for nonproliferative diabetic retinopathy (NPDR) from patients with type 2 diabetes mellitus.

Methods: Fifty-three patients diagnosed with type 2 diabetes mellitus were enrolled, including 25 patients with NPDR and 28 patients without diabetic retinopathy (DR) as the control group. A small RNA microarray was performed to screen the differentially expressed tsRNAs. Reverse transcriptase quantitative polymerase chain reaction was used to validate the significantly altered tsRNAs in a screening cohort and a verification cohort. The target genes, their enriched functions, and signaling pathways were predicted by bioinformatics analyses.

Results: In total, 668 upregulated and 485 downregulated tsRNAs were found in the NPDR group by microarray. Eight tsRNAs were validated preliminarily to be altered significantly by reverse transcriptase quantitative polymerase chain reaction, and their target genes were enriched in cellular macromolecule metabolic process and ubiquitin-mediated proteolysis. The verification experiments confirmed the increased levels of 5′tiRNA-35-PheGAA-8, tRF3-28-PheGAA-1, and tRF3b-PheGAA-6, and the decreased levels of mt-tRF3-19-ArgTCG, mt-tRF3-20-ArgTCG, and mt-tRF3-21-ArgTCG in patients with NPDR, which may serve as potential biomarkers with clinical significance.

Conclusions: The study recognized the tsRNA expression changes in peripheral blood mononuclear cells from patients with NPDR and discovered potential diagnostic biomarkers that hold clinical significance.

Translational Relevance: The significantly altered tsRNAs identified in the study may serve as potential diagnostic biomarkers for patients with NPDR as well as possible molecular targets of the occurrence and development of DR.

Introduction

It was estimated that the global population with diabetes mellitus (DM) reached 463 million, the number continues to increase and is expected to become 700 million by 2045.¹ As a microvascular disease, diabetic retinopathy (DR) is a common complication of DM that may induce irreversible vision loss and even blindness.^2,3 DR can be mainly divided into two stages according to the clinical characteristics and pathological progression: nonproliferative DR (NPDR) at the relatively early stage and PDR at the advanced stage.⁴ Because the retinal abnormalities are asymptomatic typically at the early stages, many patients are unaware of the early abnormality of NPDR, leading to the development of DR to the late stages with irreversible damage.⁵ Therefore, the early diagnosis and treatment of NPDR are essential to prevent pathogenesis. 

The identification of DR is usually screened by fundus examinations, fundus fluorescein angiography,⁶ and noninvasive wide-field optical coherence tomography angiography.⁷ However, some early lesions are difficult to detect through fundus examination; cataracts or vitreous opacity may affect the observation of the fundus. Moreover, some patients cannot tolerate the use of fluorescent contrast agents. Thus, misdiagnosis might happen, and it is urgently needed to find safe and effective tools for the early diagnosis of NPDR. 

A large amount of the genome is transcribed into noncoding RNAs (ncRNAs), these ncRNAs are not protein-coding genes, but may play important regulatory roles in various physiological and pathological conditions.⁸ These ncRNAs, such as microRNAs, long ncRNAs, and circular RNAs, could be crucial for researchers to understand the mechanisms and explore novel molecular targets in retinal diseases including DR.⁹ As a novel group of ncRNAs, transfer RNA (tRNA)-derived small RNAs (tsRNAs) are 18 to 40 nt in length and are generated from pre-tRNA or mature tRNA through cleavage mediated by ribonuclease.¹⁰ According to the cleavage positions from the parental precursors, tsRNAs can be divided into two major classes: tRNA-derived fragments (tRFs) and tRNA-derived stress-induced RNAs (tiRNAs).^11,12 Recent studies indicated the application values of tsRNAs as biomarkers for disease diagnosis and prognosis.¹³ In a mouse model of oxygen-induced retinopathy, the expression profiles of tsRNAs were altered significantly in the retinas of mice with retinal neovascularization, indicating that tsRNAs may be associated with the pathogenesis of retinal neovascular diseases.¹⁴ Moreover, the expression profiles of tsRNAs were also significantly changed in the vitreous humor of patients with PDR.¹⁵ Nevertheless, the expression profiles and the mechanisms of tsRNAs in patients with NDPR remain unclear. 

Peripheral blood mononuclear cells (PBMCs) include multiple types of immune cells, such as T cells, B cells, monocytes, and natural killer cells, which play crucial regulatory roles in immune responses.¹⁶ Several inflammatory cytokines (such as interleukin [IL]-17A, IL-10, and IL-6) released by PBMCs were dysregulated in patients with DR.¹⁷ PBMCs were isolated to explore the transcriptomics expression profiles in cancers,^18,19 Parkinson's disease,²⁰ and retinopathy of prematurity,^21,22 and so on. 

In this study, the small RNA microarray unveiled the expression patterns of tsRNAs, and a two-round reverse transcriptase quantitative polymerase chain reaction (RT-qPCR) validation identified potential diagnostic biomarkers with clinical significance. Additionally, the target genes and their potentially involved functions and signaling pathways were also investigated. 

Methods

Included Participants and Sample Collection

In total, 53 patients diagnosed with type 2 DM (T2DM) were included in the study, including 25 patients diagnosed with NPDR and 28 without retinopathy, who were regarded as the control group. Five patients with NPDR and five control patients were included as the screening cohort for microarray analysis, and the other included patients (20 patients with NPDR and 23 controls) were regarded as the validation cohort. Table 1 displays the clinical features of the recruited participants. 

Table 1.

 

Clinical Characteristics of the Patients Included in the Study

 

Fundus examination and/or fluorescein fundus angiography were used to determine the diagnosis of NPDR. Patients with the following criteria were excluded: (1) meet the diagnostic criteria of PDR, with other vitreoretinal diseases (e.g., vitreous hemorrhage, retinal detachment and macular degeneration), uveitis, scleritis, optic neuritis, and glaucoma; (2) history of ocular surgery and treatment (e.g., laser photocoagulation and intravitreal injection with anti-vascular endothelial growth factor drugs; and (3) had severe systemic diseases (e.g., congenital metabolic diseases, hematological diseases, and malignant tumor), infectious diseases (e.g., HIV, syphilis, and hepatitis infections) and autoimmune diseases (e.g., lupus erythematosus and ankylosing spondylitis). Patients undergoing hemodialysis and coronary stent implantation. The research procedure was authorized by the Ethics Committee of the Second Xiangya Hospital of Central South University and conformed to the principles of the Declaration of Helsinki. Informed consent was obtained from the participating individuals. 

PBMC Sample Collection, Preparation, and Isolation of RNAs

The EDTA anticoagulant tube was used to collect of peripheral blood (2.0–3.0 mL) from each individual. PBMCs were isolated using Ficoll-Paque PLUS (GE Healthcare, NJ) through density gradient centrifugation. Total RNAs were obtained using TRIzol (Invitrogen, Carlsbad, CA) and preserved at −80°C. RNA quantity was measured by using NanoDrop ND-1000 spectrophotometer, and RNA quantity and integrity were evaluated through either Bioanalyzer 2100 or denaturing gel electrophoresis. 

RNA Labeling, Array Hybridization, and Microarray Analysis

Initially, total RNA (100 ng) underwent dephosphorylation using 3 units of T4 polynucleotide kinase (37°C, 40 minutes) to eliminate both (P) and (cP) chemical groups from the 3′ end of RNAs, resulting in the formation of a 3-OH end. The reaction was terminated (70°C, 5 minutes) and promptly cooled to 0°C. To denature the RNAs, 7 µL of DMSO was added and the mixture was heated to 100°C for 3 minutes, and promptly chilled to 0°C. To conduct RNA end labeling, a 28-µL reaction was prepared by adding ligase buffer, BSA, final 50 mM pCp-Cy3, and 15 units T4 RNA ligase (16°C, overnight). 

The completed labeling reaction was mixed with 2× Hybridization buffer (Agilent, Santa Clara, CA) to reach a volume of 45 µL (1:1). The mixture was heated (100°C, 5 minutes) and promptly cooled to 0°C. The labeled sample mixture was applied to a microarray and subjected to hybridization (55°C, 20 hours). The slides underwent a washing process at room temperature in 6× SSC with 0.005% Triton X-102 (10 minutes), followed by a subsequent wash in 0.1× SSC with 0.005% Triton X-102 (5 minutes). The slides were scanned on an Agilent G2505C microarray scanner. 

The acquired array images were analyzed using Agilent Feature Extraction software (version 11.0.1.1). The raw intensities were log2 transformed and quantile normalized. The RNA level was obtained by averaging and consolidating multiple probes from the same small RNA (tsRNA). 

Validation Through RT-qPCR

To validate the tsRNAs that underwent significantly changes, RT-qPCR was used following the procedure as previously outlined.¹⁵ The experiments were conducted with the following conditions: incubation at 95°C for 10 minutes, 40 cycles at 95°C for 10 seconds, and 60°C for 60 seconds. The primer sequences of the validated tsRNAs are shown in Table 2. U6 was used as the reference gene for tsRNAs. 

Table 2.

 

Primer Sequences of the tsRNAs Used for RT-qPCR

 

Target Gene Prediction, GO, and KEGG Pathway Analyses

The common algorithms of miRanda and TargetScan were performed to predict the target genes of tsRNAs. The Gene Ontology (GO) (http://www.geneontology.org/) and Kyoto Encyclopedia of Genes and Genomes (KEGG) (http://www.genome.jp/kegg/) databases were used to predict the potential functions and pathways of these target genes. 

Statistical Analyses

Student's t test or Mann–Whitney U test were used to analyze comparisons of numeric variables, and the χ² test was used to analyze categorical variables. Statistical significance is indicated by P values of < 0.05, and a threshold of P value < 0.05 with a fold change (FC) of ≥1.5 was used to screen for significant alterations in small RNA microarray analysis. 

Results

General Characteristics of the Recruited Participants

In total, 53 patients with T2DM were enrolled in this study, comprising 25 patients with NPDR and 28 patients without DR who served as controls. Five individuals in each group were included as the screening cohort, and the other recruited participants were regarded as the verification cohort. General demographic features are shown in Table 1. For the screening cohort, there was no significant difference in age, sex, diabetic duration, body mass index, fasting blood glucose, and hemoglobin A1c. However, sex and body mass index significantly differ between the NDPR and control groups in the verification cohort. The diagnosis of NPDR was determined by fundus examination and/or fluorescein fundus angiography (Fig. 1). 

Figure 1.

 

Images of the fundus of a representative case with mild NPDR. (A and C) Color fundus photographs show no obvious microaneurysms. (B and D) Fluorescein angiography images reveal multiple microaneurysms.

Changes in tsRNA Expression Profiles in PBMCs of Patients with NPDR

To identify alterations in the expression levels of tsRNAs in PBMC samples of NPDR compared with patients without DR, we identify significant differences according to the small RNA microarray analysis. As shown in the scatter plot (Fig. 2A) and the volcano plot (Fig. 2B), the altered expressions of tsRNAs were recognized. In the NPDR group, there were 668 tsRNAs that were upregulated and 485 tsRNAs that were downregulated, as compared with the controls (P < 0.05; FC ≥ 1.5) (Figs. 2B, C). Among these significantly altered tsRNAs, the top 10 upregulated and downregulated tsRNAs were listed according to the FCs (Table 3), wherein tRF3b-PheGAA-6 and mt-tRF3-21-ArgTCG were the tsRNAs with the largest FCs of upregulation and downregulation, respectively. 

Figure 2.

 

Changes of tsRNA expression profiles in PBMCs of patients with NPDR compared with those with non-DR controls. (A) Scatter plot of altered tsRNAs between NPDR and control groups. The red and blue dots represent the upregulated and downregulated RNAs (FC ≥ 1.5). (B) Volcano plot of significantly expressed tsRNAs between NPDR and control groups. The red and blue dots represent the upregulated and downregulated RNAs (FC ≥ 1.5; P < 0.05). (C) Heatmap with analysis of hierarchical clustering according to the significantly changed tsRNAs in each group. (D) Venn diagram shows the number of dysregulated tsRNAs and those have been investigated in the database of tRFdb (human). (E, F) Pie charts of the tiRNA and tRF subtype distribution in upregulated (E) and downregulated tsRNAs (F).

Table 3.

 

Upregulated and DownRegulated tsRNAs in PBMCs of Patients With NPDR Compared With Controls (Top 10 in FC)

 

Validation of the Top Five Upregulated and Downregulated tsRNAs by RT-qPCR

To confirm the dependability of the microarray initially, RT-qPCR was conducted to evaluate the expression levels of the five highest upregulated and downregulated tsRNAs according to the FCs. As shown in Fig. 3, tRF3b-PheGAA-6, 5′tiRNA-35-PheGAA-8, and tRF3-28-PheGAA-1 were overexpressed significantly in the PBMCs of the NPDR group, whereas mt-tRF3-21-ArgTCG, mt-tRF3-17-ArgTCG, mt-tRF3-20-ArgTCG, mt-tRF5-50-LysTTT, and mt-tRF3-19-ArgTCG were significantly low expressed in the NPDR group. In addition, tRF3a-PheGAA-7 and tRF3b-PheGAA-7 were increased slightly, but not significantly changed in the samples of the NPDR group compared with the controls. These results of preliminary validation indicated a comparable pattern of changes in the microarray results. 

Figure 3.

 

Initial validation of the significantly altered tsRNAs in samples of the screening cohort. Relative expression levels of 10 tsRNAs (top five upregulation and five downregulation) by RT-qPCR. Error bars represent standard error of the mean. *P < 0.05; **P < 0.01; ***P < 0.001.

Bioinformatics Analyses: Target Gene Prediction, GO Enrichment Analysis, and KEGG Pathway Analysis

To predict the interactions between the significantly altered tsRNAs and mRNAs, the target gene network was established for the eight validated tsRNAs using the shared algorithms of miRanda and TargetScan. These eight validated tsRNAs are connected to a total of 4528 target genes through 5079 edges (Supplementary Fig. 1). 

To gain a deeper understanding of the potential biological functions and signaling pathways, we conducted GO enrichment analysis and KEGG pathway analysis according to the predicted target genes of the eight validated tsRNAs. As shown in Figure 4A, the cellular macromolecule metabolic process, cytoplasm, and protein binding were the most enriched terms of GO analysis. Ubiquitin-mediated proteolysis was the most enriched KEGG pathways (Fig. 4B). 

Figure 4.

 

GO and KEGG analyses of the eight dysregulated tsRNAs’ target genes. GO enrichment analysis (A) and KEGG pathway analysis (B) of target genes according to eight validated tsRNAs with significantly changes.

Verification and Assessment of Diagnostic Values in Samples of the Enlarged Cohort

To verify further the expression changes and assess the diagnostic values of the tsRNAs, we included 20 PBMC samples obtained from patients with NPDR and the other 23 samples from the patients without DR and with T2DM as controls, and regard these participants as the verification cohort. A total of six altered tsRNAs that have been validated preliminarily were verified further with the enlarged sample size, including three upregulated and three downregulated tsRNAs. The expression levels of three tsRNAs (5′tiRNA-35-PheGAA-8, tRF3-28-PheGAA-1, and tRF3b-PheGAA-6) were increased significantly in the NPDR group (Figs. 5A–C) (P < 0.001), and those of the other three tsRNAs (mt-tRF3-19-ArgTCG, mt-tRF3-20-ArgTCG, and mt-tRF3-21-ArgTCG) were decreased significantly in the NPDR group (Figs. 5D–F) (P < 0.001). Next, we conducted the ROC curve analyses, and calculated the areas under the curve (AUC) for the verified tsRNAs. Interestingly, all of the six tsRNAs exert great sensitivity and specificity. In particular, three tsRNAs (5′tiRNA-35-PheGAA-8, tRF3-28-PheGAA-1, and mt-tRF3-21-ArgTCG) have AUC values of 1.0. The AUCs were 0.9978, 0.963, and 0.8489 for mt-tRF3-20-ArgTCG, tRF3b-PheGAA-6, and mt-tRF3-19-ArgTCG, respectively (Figs. 5G–L). The cutoff values, diagnostic sensitivities, and specificities are shown in Table 4. 

Figure 5.

 

RT-qPCR validation of (A) 5′tiRNA-35-PheGAA-8, (B) tRF3-28-PheGAA-1, (C) tRF3b-PheGAA-6, (D) mt-tRF3-19-ArgTCG, (E) mt-tRF3-20-ArgTCG, (F) mt-tRF3-21-ArgTCG in samples of the enlarged verification cohort. ROC curve analyses of (G) 5′tiRNA-35-PheGAA-8, (H) tRF3-28-PheGAA-1, (I) tRF3b-PheGAA-6, (J) mt-tRF3-19-ArgTCG, (K) mt-tRF3-20-ArgTCG, and (L) mt-tRF3-21-ArgTCG in samples of the enlarged verification cohort.

Table 4.

 

Clinical Significance of 6 tsRNAs as Diagnostic Biomarkers for Patients With NPDR

 

Discussion

Recent studies reported that tsRNAs are regarded as potential biomarkers for diagnosis in a variety of diseases,¹³ including gastric cancer,²³ acute myeloid leukaemia,²⁴ and lupus nephritis.²⁵ Compared with the other kinds of molecular indicators, there are several advantages of tsRNAs being promising biomarkers. First, tsRNAs are significantly abundant with stable characteristics in a variety of body fluids, including blood, and are related closely to the pathogenesis of many diseases.²⁶ Second, the expression profile of tsRNAs has tissue and temporal specificities, which are beneficial to the clinical assessment of tsRNAs as molecular biomarkers.^27,28 Furthermore, tsRNAs are highly conserved in many species, including humans.^27,29 These characteristics suggest that tsRNA may become an ideal source of biomarkers for the clinical diagnosis of diseases. 

It has been reported that tsRNAs were also significantly altered in a rat model of high-fat diet and streptozocin-induced DM.³⁰ Moreover, the expression profiles were changed in the adipose tissues of overweight patients with T2DM,³¹ and in patients with T2DM and with diabetic foot ulcers,³² indicating the important potential role of tsRNAs in the pathological development of T2DM. 

In clinical practice, mild NPDR can hardly be recognized by screening of normal fundus examination. As shown in Fig. 1, the fundus photograph showed no obvious microangioma on the retinas of the patient, whereas fluorescein angiography recognized multiple microaneurysms on both of the retinas, which demonstrated the diagnosis of NDPR. Therefore, general fundus examinations may result in missed diagnoses, and repeated fundus fluorescein angiography examination is not feasible for all patients with DM, and it is of great clinical significance to screen for DR preliminarily using simple and efficient methods. 

In this study, we revealed the expression profiles of tsRNA by small RNA microarray and identified 668 upregulated and 485 downregulated tsRNAs in PBMCs of patients with NPDR (Fig. 2). These results indicate that some tsRNAs could be potential biomarkers that contribute to the diagnosis of NPDR. 

The expressions of the top five upregulated and downregulated tsRNAs were validated preliminarily by RT-qPCR, and the results showed that eight tsRNAs were changed significantly (Fig. 3). The current study included a verification cohort for the second-round RT-qPCR validation and confirmed the significantly changes of six tsRNAs (Fig. 5). The ROC curve analyses indicated the potential of these molecules to be applied as diagnostic biomarkers for NPDR (Fig. 5). In particular, 5′tiRNA-35-PheGAA-8, tRF3-28-PheGAA-1, and mt-tRF3-21-ArgTCG have the AUC values of 1.0, which show great clinical application prospects. 

Evidence has accumulated that tsRNAs may have important roles in disease pathogenesis. For example, tsRNAs may regulate the mRNAs similar to the binding mechanism of miRNAs,³³ and they may also inhibit cancer progression via binding to the RNA-binding protein and displacing the oncogenic transcripts.³⁴ In this study, through bioinformatics analyses, ubiquitin-mediated proteolysis and nuclear factor κB (NF-κB) signaling pathway were listed in the top 10 enriched KEGG pathways of the tsRNAs’ target genes (Fig. 4B). The ubiquitin-mediated proteolysis is a key biological process in regulating protein homeostasis, and its dysregulation may lead to susceptibility to various diseases including cancers.³⁵ Centromere protein U inhibited ubiquitin-proteasomal degradation of cyclo-oxygenase-2, which induced the promotion of angiogenesis in triple-negative breast cancer.³⁶ Ubiquitination and degradation acted as a key driver in the mechanisms of miR-195 in promoting EMT and cell permeability of high glucose-stimulated retinal pigment epithelial cells.³⁷ Moreover, ubiquitin-mediated proteolysis also ranks as an enriched KEGG pathway with m⁶A hypomethylated mRNAs in the oxygen-induced retinopathy model in mice.³⁸ Thus, ubiquitin-mediated proteolysis may also be involved in the tsRNA-mediated regulations in DR pathogenesis. In contrast, numerous studies indicated the importance of NF-κB signaling pathway in the mechanisms of DR. For instance, Li et al.³⁹ demonstrated that miR-874 attenuated DR via modulation of NF-κB signaling pathway in a rat model. Yun et al.⁴⁰ reported that IL-1β could lead to pericyte apoptosis through the NF-κB pathway in DR. The relevance of these involved signaling pathways with tsRNAs and the mechanisms in DR need to be studied further. 

There are still several limitations in this study. First, although we have conducted two round validations by RT-qPCR, the number of recruited participants of the study was still not large enough. NPDR is classified as mild, moderate, and severe.⁴¹ Because the sample size of the current study was not large enough, it is hard to compare the differences among the severity levels of NPDR, which should also be further explored by expanding the sample size in future studies. Second, patients with advanced PDR were not included in the study, and their tsRNA levels in PBMCs also need to be assessed. It is also interesting to explore whether laser photocoagulation anti-vascular endothelial growth factor therapies affect the PBMC expression levels of these potential biomarkers and evaluate their clinical values in prognosis. Moreover, it has been reported that gender differences may influence the PBMC function and population after lipopolysaccharide stimulation or during illness.^42,43 Therefore, further study with an expansion of the enrolled cohort is still needed to verify and address the gender bias issues present in this study. 

To summarize, this study showed the significant alterations of tsRNAs in PBMC expression levels from patients with NPDR, and discovered multiple potential diagnostic biomarkers with clinical significance. Further studies are required to investigate the roles of these dysregulated tsRNAs in the development of DR. 

Acknowledgments

Supported by grants from the National Natural Science Foundation of China (No. 82271110), the Natural Science Foundation of Hunan Province (No. 2022JJ30869, 2023JJ30748), the Science and Technology Innovation Program of Hunan Province (No. 2021SK53526), Scientific Research Project of Hunan Provincial Health Commission (No. 202207022574, D202307028349), and the New Technology Incubation Funds in Ophthalmology. 

Disclosure: C. Ding, None; N. Wang, None; A. Peng, None; Z. Wang, None; B. Li, None; X. Zhang, None; J. Zeng, None; Y. Zhou, None