Transcriptomic profiles in patients with Diabetic macrovascular complications
To determine the transcriptomic profiles, the transcriptome sequencing in the discovery cohorts composed of 6 patients diagnosed with DMC and 6 health controls. The data process flow chart is shown in Fig. 1a. The information of the DMC participants in both the discovery and the expanded cohorts are shown in Table 1, and demonstrates that most clinical characteristics were not significantly different between the two cohorts. Thus, the lncRNAs selected in the discovery cohort are reliably validated in the expanded cohort.
First, we described the transcriptomic profiles of DMC to identify the critical genes and lncRNAs in the DMC patients. A total of 14,978 lncRNAs and 17,172 mRNAs were detected in leukocytes from DMC patients. Of these, 9,235 lncRNAs have been registered in databases and defined as known. Out of the 9,235 lncRNA’s, 4,647 were upregulated, and 4,588 were downregulated. The other 5,743 lncRNAs were identified for the first time and were defined as either upregulated (3,379) or downregulated (2,364) novel lncRNAs. After optimization using an adjusted p-value (threshold of <0.05) and change fold >2, we identified 477 significantly differentially expressed (DE) lncRNAs (DMC-lncRNAs) (Table S3), among which 245 were downregulated and 232 were upregulated (Fig. 1b). The top 20 DMC-lncRNAs are shown in Table 2. We further analyzed the different biotypes of 477 DE-lncRNAs. The results indicated that 172 lncRNAs (36.1%) were antisense, 253 (53.0%) were intergenic (lincRNA), and 52 (10.9%) belonged to other types (sense-overlapping or sense-intronic) (Fig. 1c, Table S3). Further, 798 DMC-mRNAs were found by similar optimization (Table S4), with 491 downregulated and 307 upregulated (Fig. S1a). Both DMC-lncRNA and DMC-mRNA were distinguishable within DMC patients and the healthy controls by hierarchical clustering (Fig. 1d, Fig. S1b) and principal content analysis (Fig. 1e, Fig. S1c). This data clearly illustrates a distinguishable differential expression profile of the leukocytes between DMC patients and the healthy controls.
Features of DE-lncRNAs and mRNAs identified in patients with DMC
The Gene Ontology (GO) enrichment analysis was applied to classify DMC patients' DE-lncRNAs and DE-mRNAs. Under the biological process category, metabolic process, single organism process, response to stimulus, and cellular process were the top 4 DE-lncRNAs; similar items were found for the DE-mRNAs (Fig. 2a, Fig. S2). Under the molecular function category, both DE-lncRNAs and DE-mRNAs showed the highest percentages in catalytic activity and signal transducer activity items (Fig. 2a, Fig. S2). The top 20 items for DE-lncRNA demonstrated that most biological processes involved metabolic processes (17/20, 85%) (Fig. 2b). The highest molecular function is binding including RNA, DNA, and protein binding (Fig. 2c). These results imply that DE-lncRNAs identified from the DMC patients are involved in transcription regulation.
We structured co-expression networks to determine if lncRNAs are associated with lncRNAs or mRNAs. There were 509 genes involved in our co-expression networks consisting of 117 lncRNAs and 392 mRNAs (Fig. S3), with 7 lncRNAs and 18 mRNAs harboring more than 10 related genes in the CNC network. The top 10 lncRNAs and mRNAs with the number of genes to which they are related are shown in Table 3. In this CNC network, lncRNAs ENSG00000279463, MSTRG.39819, and ENSG00000228063 were the top 3 lncRNAs with the greatest number of related genes (35, 20, and 18) (Fig. 2d-f). The related genes such as TOMM5, and MYLK have been reported to have implications in diabetes [19]; PDLIM1 and CAMK2G were found to be related to atherosclerosis [20]. We also analyzed the characteristics of 9,059 transcripts of novel lncRNAs identified in DMC patients. Most novel lncRNA transcripts harbored 2 exons (7060, 77.93%) (Fig. S4a). The lengths of most of the novel lncRNAs (71.91%) were less than 2000 bp (Fig. S4b). The conservation analysis in humans indicated that more than half (61.3%) had conservation scores (CS) less than 0.1 and 6.77% less than 0.01 between humans and other species (Fig. S4c). The distribution of chromosomal transcripts demonstrated that novel lncRNAs were primarily distributed on chr1 to chr6; and that most lncRNAs with low CS in humans were from the same chromosome (Fig. S5). These results reflect that lncRNAs in leukocytes may have an important effect on DMC.
Validation Measurement of DE-lncRNAs in expanding groups
To confirm the DE-lncRNAs, we independently detected 16 lncRNAs in the validation group of DMC patients (n = 46) and healthy controls (n = 36) via real-time PCR. The expression levels of these lncRNAs in the validation cohort are shown in Fig. 3a. Of the 16 tested lncRNAs, 12 lncRNAs (75%) displayed dramatically different expression levels between the DMC and the healthy control groups; 4 lncRNAs were downregulated, and the others were upregulated in DMC. Nine novel lncRNAs and 3 known lncRNAs were positively confirmed. In addition, 8 lncRNAs belonged to the lincRNA, whereas 4 lncRNAs belonged to the antisense group (Table 4). Comparing the validation results and sequencing data showed that most lncRNAs (75%, 12/16) displayed similar trends, with 10 lncRNAs exhibiting the same significantly positive results as observed in the sequencing data (Fig.S6). Among the 12 significantly different expression levels of lncRNAs, 8 lncRNAs harbor predicted target genes (Table S5).
We sought to identify potential orthologs of our selected lncRNAs by comparing their sequences with previously identified murine lncRNAs. Of the 12 validation-positive lncRNA, 8 (66.67%) exhibited orthologous sequences in the mouse genome (Table 5). Thus, these validated lncRNAs exhibited comparable conservation in humans and mice, which could lead to the use of a mouse model to investigate lncRNA biomarkers directly.
DE-lncRNAs response to HG and IS conditions in EC
The pathological changes of DMC are primarily due to higher serum glucose levels inducing abnormal metabolism, consequently influencing the functions and homeostasis of endothelial cells [21-23]. A better understanding of lncRNAs response in EC under HG and IS is essential. Therefore, we tested the expression of 10 lncRNAs which are positively validated and exhibited the same significant trend as observed in the sequencing data within the HUVEC cell line. As shown in Fig. 3b and Fig.S7a, 3 lncRNAs exhibited increased levels after a 12 h and 24 h treatment with 25 nM, and 30 nM of glucose, respectively. LncRNA ENSG00000228063 was significantly downregulated after 24 h treated with HG in ECs (Fig. 3b, Fig.S7a).
Based on the “metabolic memory,” which is defined as a phenomenon that the vasculature can remember transient hyperglycemia for quite an extended period even after the reestablishment of normoglycemia [24], we tested if there was a metabolic memory of these 4 lncRNAs in ECs. The results illustrated that the lncRNA MSTRG.122492 and MSTRG.3528 were significantly upregulated, and ENSG000000228063 was dramatically downregulated (Fig. 3c). We also performed HO and IS treatment on ECs to mimic the environment of the pathological changes in DMC. The results showed that the expression of lncRNA MSTRG.3528 was significantly higher after a 12h HO stimulus, the others exhibited no change (Fig.S7b). The expression of lncRNA MSTRG.74858 and ENSG00000228063 markedly decreased after a 6h TNF-α treatment. MSTRG.3528 was downregulated after a 12h treatment with TNF-α. MSTRG.122492, ENSG00000269902, and MSTRG.159327 expression was downregulated after 6h and then reached normal levels after 12h (Fig. 4d). All the above results imply that ENSG00000228063 is the most sensitive in response to HG, HO, and IS.
LncRNA LYPLAL1-DT alleviates the influence of HG and IS on the proliferation, migration, autophagy, apoptosis, and inflammatory response of ECs
Based on the results of HG/IS test, and lncRNA ENSG00000228063 exhibited metabolic memory, we chose this transcript for further investigation. ENSG00000228063 is located on 1q41 LLP1 and is within 1 kb of LYPLAL1. Therefore, it was named LYPLAL1-DT (Ensembl ID: ENSG00000228063) (Fig. 4a). The major variant is 2,577 nt in length and encoded by five exons. LYPLAL1-DT is highly conserved in humans, with homologous sequences found only in chimpanzees, and other nonhuman primate animals (Fig.S7c), having a very low homology in mice. LYPLAL1-DT is expressed across diverse human tissues, including vascular tissue and ECs. As the position of the lncRNA is closely associated with its function, we first explored the location of LYPLAL1-DT in HUVEC cells. As shown in Fig. 4b, LYPLAL1-DT is expressed in both the nucleus and the cytoplasm, primarily in the cytoplasm, indicating that LYPLAL1-DT could exert cis-regulation or play a role as a competitive endogenous RNA (ceRNA).
To investigate the function of LYPLAL1-DT in ECs, we successfully constructed overexpressing HUVECs (OE) and their corresponding control cell line (OC) (Fig. 4C). The CCK8 staining results showed that LYPLAL1-DT significantly promotes the vitality of EC, especially under HG conditions (Fig. 4d). LYPLAL1-DT increases the proliferation of EC under an IS mimic via treatment with TNF-a (Fig. 4e). LYPLAL1-DT overexpression in EC alleviates migration under both HG and TNF-a treatment (Fig. 4f). These results reflect that LYPLAL1-DT overexpression effectively protects ECs under HG and IS by enhancing proliferation and migration.
Dysfunctional HUVECs promotes vascular inflammation by expressing inflammation cytokines and surface adhesion molecules involved in DMC development. We collected the media of EC-OE under HG and analyzed the concentrations of the secreted cytokines. The results illustrate that the inflammatory molecule IL-1β (Fig. 4g) and the adhesion molecule ICAM-1 decreased (Fig. 4h); conversely, anti-inflammation molecules IL-10 and IL-13 were increased (Fig. 4i) in the EC-OE group and under HG conditions. Consistent with the high level of ICAM-1 in the cell culture media, the overexpression of LYPLAL1-DT significantly suppressed the adhesion of monocytic THP-1 cells onto the HG-treated HUVEC monolayers (Fig. 4j). These results suggest that LYPLAL1-DT plays a significant role in ameliorating the inflammatory conditions in ECs treated with HG.
LYPLAL1-DT affected SIRT1 expression by acting as a ceRNA sponging miR-204-5p
To explore the molecular mechanism of LYPLAL1-DT in the protection of EC, we were able to predict the miRNAs and the corresponding target genes using the bioinformatics tools TargetScan and miRcode. Upon analysis 13 miRNAs with a higher score were detected and found that 3 were upregulated and 2 were downregulated miRNAs in EC-OE compared to the control (Fig. S8). We chose miR-204-5p, the most significantly downregulated miRNA (Fig. 5a) in EC-OE of LYPLAL1-DT to investigate its function. The miR-204-5p mimic remarkably decreases the proliferation and migration of EC-OE (Fig. 5b, 5c). Subsequently, we constructed luciferase reporters with both a wild and mutant LYPLAL1-DT site used for miR-204-5p binding (Fig. 5d). The results demonstrated that the mimic miR-204-5p significantly suppresses the luciferase activity of wild type LYPLAL1-DT, compared to either the mutant or empty vector controls (Fig. 5e). We detected 9 target genes of miR-204-5p that were predicted using STARBASE via real-time PCR (Fig. S9), then chose SIRT1 (Fig. 5f) and examined its expression level in EC-OE with and without the presence of the miR-204-5p mimic by real-time PCR and Western blot. SIRT1s expression was increased in EC-OE and conversely was decreased by the miR-204-5p mimic (Fig. 5g, 5h). Using a dual luciferase reporter test, it was demonstrated that SIRT1 is the binding targeted gene of miR-204-5p (Fig. 5i). Further results of the RIP assay showed that LYPLAL1-DT, miR-204-5p, and SIRT1 bind to Ago2 (Fig. 5j), confirming that LYPLAL1-DT functions as a competing endogenous RNA (ceRNA) regulating the expression and function of SIRT1 via inhibition with miR-204-5p. SIRT1 is an important gene related to autophagy and apoptosis, we detected the LC3B and apoptotic body under HG conditions. The data demonstrated that LYPLAL1-DT increases LC3Ⅱ/Ⅰ levels, thus decreasing the number of TUNEL positive cells under HG conditions (Fig. 5k, 5l). Therefore, our results indicated that lncRNA LYPLAL1-DT inhibits miR-204-5p, consequently upregulating the expression of SIRT1. Furthermore, promoting autophagy, attenuating apoptosis levels in order to alleviate HG injury, and exerting protective effects on ECs.
LYPLAL1-DT from leukocytes may exert protective effects on HUVEC via exosome transportation
Exosomes regulate the biological functions of recipient cells via RNA transfer. Since LYPLAL1-DT was discovered in leukocytes, we hypothesized that LYPLAL1-DT was transmitted to HUVECs via exosomes. The serum exosomes of DMC patients and healthy control were extracted and RNA-sequencing was performed to confirm this hypothesis. The purified exosomes were detected by transmission electron microscopy (TEM) (Fig. 6a), NTA (Fig. 6b), and Western blot (Fig. 6c) using the positive exosome protein markers TSG101 and Alix. The RNA sequencing data showed that DE-lncRNAs (Fig. 6d, 6e) and DE-mRNAs (Fig. S10 a and b) were found between exosomes derived from DMC and control serum, with LYPLAL1-DT having the most significantly different expression levels of the lncRNAs, harboring trends similar to the results found in leukocytes (Fig. 6f). We also detected that the leukocyte marker CD11b+ is positively expressed in exosomes (Fig. 6c), proving that exosomes may originate from leukocytes. Thus, we treated HUVEC by exosomes from DMC patients’ serum and healthy control and detected miR-204-5p level, cell viability and migration ability. As expected, miR-204-5p level was significantly lower in control than that in DMC group (Fig. 6g), while viability (Fig. 6h) and migration ability (Fig. 6i) of HUVEC were abbreviated in DMC group. This data confirmed that expression of LYPLAL1-DT in exosomes derived from the leukocytes could affected the ECs, protecting them from dysfunction.