Introduction

Diabetes mellitus is caused by insulin secretion deficiency associated with varying degrees of damage to pancreatic beta cells, and the inability to regulate the metabolic balance of blood sugar. Therefore, replacement therapy with pancreatic beta cells has become a therapeutic option for diabetes. The clinical application of islet transplantation is limited by insufficient donors, immune rejection, immunosuppressive drug side effects, and possible beta cell toxicity [1]. Stem cells are an important alternative that could provide innumerable potential islet cell sources for transplants. Umbilical cord mesenchymal stem cells (UC-MSCs) are obtained from discarded placentas, and they are more readily available and have a higher proliferative potential than other MSCs [2, 3]. UC-MSCs are more primitive MSCs than bone marrow or adipose mesenchymal stem cells (BM-MSCs or AD-MSCs) and do not express major histocompatibility complex (MHC) class II (HLA-DR) antigens [3, 4], which makes them good candidates for potential allogeneic therapeutic applications. Previous studies have shown that UC-MSCs can differentiate into IPCs in vitro and improve the glycaemic status of diabetic mice after transplantation in vivo [5, 6]; however, the induction efficiency was low and the maintenance time of hypoglycaemia was slightly short [5,6,7]. Therefore, the differentiation efficiency and function of IPCs must be improved. Moreover, studies should focus on identifying the underlying molecular mechanisms by which UC-MSCs differentiate into IPCs. Many small molecule compounds, growth factors, activators, and inhibitors can transdifferentiate MSCs into IPCs, which improves the survival of IPCs and the ability to release insulin in vitro [8,9,10,11]. MSC transdifferentiation efficiency and insulin secretion can be improved by overexpression of Pdx1, Neurog3, MafA, and Pax4 [12,9: Table S1). Some results of the quality control of the RNA-seq data sets are shown (Additional file 5: Fig S5A-C). We aligned the filtered reads to the reference sequence (human GRch38 genome) by HISAT2, predicted lncRNAs and calculated their expression levels using Cufflinks. The number of annotated lncRNAs and novel lncRNAs of in each of the five stages was profiled. Among them, 1053 annotated lncRNAs and 48 novel lncRNAs were continuously expressed across all five stages (Fig. 3A). LncRNAs obtained by screening included lincRNAs (30.4%), antisense lncRNAs (15.8%), and sense-overlap** lncRNAs (53.8%) (Additional file 5: Fig. S5D). Most lncRNAs contained approximately 1 or 2 exons and no more than 10 exons; however, most protein-coding transcripts contained more than 10 exons (Fig. 3B). In addition, a length of 0–200 bp represented the dominant portion of the exon length of the lncRNAs and protein-coding transcripts (Fig. 3C). The length of the dominant portion of lncRNAs was around 1000 bp, while the length of protein-coding transcripts was nearly 2500 bp (Fig. 3D). Overall, lncRNAs have fewer exons than protein-coding transcripts, and the lengths of these lncRNAs are generally shorter than those of mRNAs. To understand the quality control of the experimental data, a Pearson correlation analysis was performed for all pairs of RNA-seq samples, and it demonstrated that the correlation coefficient between each repeated sample reached 0.8, indicating that the overall quality of the data was good (Fig. 3E, Additional file 6: Fig S6. A).

Fig. 3
figure 3

The profiling of differentially expressed lncRNAs during IPCs differentiation of UC-MSCs. A Venn diagram of detected known lncRNAs (left) and novel lncRNAs (right) in samples from five stages. At least two samples with RPKM >  = 0.2 was is considered to be detected in the group; B Distribution of exon count of known lncRNAs, novel lncRNAs, and protein-coding RNA; C Distribution of exon length of known lncRNAs, novel lncRNAs, and protein-coding RNA.; D Density of the length distribution of known lncRNAs, novel lncRNAs, and protein-coding RNA. The length density distribution was generated by density function in R; E Heatmap clustering analysis of sample correlation based on the normalized mapped reads on each lncRNAs

Full size image

Differentially expressed lncRNAs and functional enrichment of target genes

To identify the differentially expressed lncRNAs during IPC differentiation of UC-MSCs, the number of lncRNAs was detected in each five stages. A total of 2615 lncRNAs with differential expression were identified, of which 1101 lncRNAs were continuously expressed across all five stages (Fig. 4A). To further explore changes in the dynamic expression of differentially expressed lncRNAs across the five stages during induction, the S0 stage was compared with the S1, S2, S3, and S4 stages, the S4 stage was compared with the S1, S2, and S3 stages, the S3 stage was compared with the S2 and S1 stages, and the S2 stage was compared with the S1 stage. Up- or down-regulated lncRNAs were identified, which revealed that the number of differentially expressed lncRNAs among them decreased successively. Moreover, mRNAs levels followed the same expression trend (Fig. 4B, Additional file 6: Fig. S6B). The overlap of differentially expressed lncRNAs is shown in the S1, S2, S3, and S4 stages compared with the S0 stage. Among them, 30 lncRNAs were continuously up-regulated, and 38 lncRNAs were continuously down-regulated across the five stages. The overlap of differentially expressed mRNAs is shown (Fig. 4C, Additional file 6: Fig. S6C).

Fig. 4
figure 4

The expression of LncRNAs at each stage and functional enrichment of target genes of differentially expressed lncRNAs. A Venn diagrams of LncRNAs detected in samples of 5 stages, the p value or corrected p value (padj) was used to determine the level of significance for multiple samples; B bar plots of differential LncRNAs at different stages; C the Venn diagram showed the overlap of differentially up-regulated or down-regulated LncRNAs (S0 stage was compared with S1, S2, S3, and S4 stages, respectively); D S0 stage was compared with S1, S2, S3, and S4 stages, respectively, and GO(molecular process) enrichment analysis of the mRNA of differentially up-regulated or down-regulated LncRNAs, the colour scale shows the significance of these terms scaled by column (− log10 corrected p value); E S4 stage was compared with S1,S2,and S3 stages, respectively, and KEGG enrichment analysis of the mRNA of differentially up-regulated or down-regulated LncRNAs, the colour scale shows the significance of these terms scaled by column(− log10 corrected p value)

Full size image

To identify the function of differentially expressed lncRNAs that controlled IPC differentiation of UC-MSCs, we performed Gene Ontology (GO) analysis and pathway analysis. Since lncRNAs do not encode proteins, their regulatory effect can only be exerted by regulating the genes coexpressed with them. First, GO analysis was performed to identify enrichment in the significant functions of the genes coexpressed with differentially expressed lncRNAs, which were obtained during IPC differentiation of UC-MSCs. We obtained the top 20 enriched GO functions according to the p value and FDR (p < 0.05, FDR < 0.05) in the S1, S2, S3, and S4 stages compared with the S0 stage. Coexpressed genes of highly expressed lncRNAs in the S1, S2, S3, and S4 stages were mainly concentrated in protein translation-related pathways. In contrast, the coexpressed genes of lncRNAs with low expression were mainly enriched in the cell adhesion and cell connection pathways (Fig. 4D). The top 20 enriched GO functions of differentially expressed mRNAs are shown (Additional file 6: Fig. S6D).

We then identified enrichment in the significantly changed pathways that mediated the functions of the genes coexpressed with differentially expressed lncRNAs based on the KEGG database. The top 5 significant pathways in stage S4 compared with stages S1, S2, and S3 were related to the insulin signal transduction pathway of beta cells, and they may play key roles in the function of pancreatic beta cells. Among them, up-regulated lncRNAs were mainly enriched in the NF-KB signalling pathway and MAPK signalling pathway, and down-regulated lncRNAs were mainly enriched in the HIPPO signalling pathway, PI3K–Akt signalling pathway, p53 signalling pathway, and other signalling pathways. In summary, the KEGG analysis revealed that differentially expressed lncRNAs were related to the insulin signal transduction pathway of beta cells (Fig. 4E).

The expression patterns of lncRNAs in the S1, S2, S3, and S4 stages were demonstrated by K-means analysis

To characterize the dynamic changes in lncRNA and mRNA expression, we clustered all their expression patterns (97 lncRNAs and 769 mRNAs) by K-means analysis. We identified 4 lncRNA and mRNA clusters. The upper part of the module shows four clusters identified by K-means, and the lower part shows the lncRNA and mRNA expression profiles of four clusters (Fig. 5A, Additional file 7: Fig. S7A). The lncRNA expression patterns in the four clusters are displayed, among which the trend in expression levels of lncRNAs in Clusters 1 and 3 was consistent with that of the mRNAs (Fig. 5B, Additional file 7: Fig. S7B). The top 5 signalling pathways of genes coexpressed with lncRNAs in the four different clusters are displayed (Fig. 5C). Moreover, we found that mRNA Cluster 1 mainly showed enrichment in the cell proliferation and cell differentiation pathways, indicating that cell differentiation tended was active, while Cluster 3 was mainly concentrated in the negative regulation of cell proliferation and the cell adhesion pathway. Thus, with the development of cell differentiation, the ability to inhibit cell proliferation and the connection between cells became weaker. Namely, cell differentiation became increasingly active, which was in line with the process of cell differentiation according to the functional enrichment analysis of the four mRNA clusters (Additional file 7: Fig. S7C). The expression levels of the P53, FIGF, JARID2, and PROX1 genes, which are involved in the development and function of pancreatic beta cells in Clusters 1 and 3, were determined (Additional file 7: Fig. S7D). To explore the functions of lncRNAs, the network interactions of lncRNAs and their coexpressed mRNAs in Clusters 1–3 and the enrichment pathways related to cell differentiation are shown (Fig. 5D).

Fig. 5
figure 5

The expression pattern of LncRNAs in S1 stage to S4 stage was demonstrated by K-means analysis. A Differential LncRNAs in S1 stage to S4 stage were clustered by K-means; B LncRNAs expression profiles generated by K-means clustering; C the top 5 GO terms (molecular processes) of coexpressed mRNAs with 4 clusters of differential LncRNAs generated by K-means clustering, and the colour scale shows the significance of these terms scaled by column (− log10 corrected p value). D Coexpression network of co-regulated LncRNAs in gene clusters 1 to 3 and coexpression mRNAs participating in 9 GO terms

Full size image

Screening of pancreatic development-related lncRNAs and their coexpressed mRNAs

Since the focus of our study was to investigate lncRNAs associated with IPC differentiation, we screened the mRNAs JARID2, PROX1, and p53, which are related to the development and function of the pancreas in the functional enrichment pathway. Cervanets et al. found that JARID2 plays a role in the late differentiation of embryonic pancreatic beta cells. Pethe et al. found that the expression levels of JARID2 decreased gradually during directed differentiation and spontaneous differentiation of the pancreatic system. Studies have also found that PROX1 is a specific marker of the endoderm in early pancreatic development. Other studies have found that PROX1 is highly expressed in pancreatic endocrine progenitor cells, but is lacking in mature beta cells. PROX1 activity is essential for the formation of endocrine progenitor cells and the differentiation of α cells [35,36,37,38,39]. Paul et al. found that beta cells overexpressing Prox1 rapidly inactivate MafA after birth, and downregulating Prox1 is a prerequisite to expand the β-cell mass after birth and for proper maturation of this lineage [40]. The activity of MafA plays an important role in beta cell maturation because it governs the glucose-responsive transcription of insulin and key components of the glucose-stimulated insulin secretion mechanism [41, 42]. Kung et al. demonstrated that p53 regulates insulin secretion and pancreatic beta cell survival through multiple signalling pathways [43]. The coexpression of the lncRNAs:AC009014.3 and Gs1-72m21.1 with JARID2 and lncRNA CTBP1-AS2 with PROX1 was identified through coexpression analysis (Fig. 6B, D, E). Omidvar et al. found that lncRNA CTBP1-AS2 expression levels are associated with T2D susceptibility [44], which means that lncRNA CTBP1-AS2 may regulate pancreatic maturation by regulating PROX1 mRNA. These results indicate that the lncRNAs AC009014.3, Gs1-72m21.1 and CTBP1-AS2 may be involved in the development of pancreatic beta cells. The lncRNAs XLOC_050969, LINC00883, XLOC_050981, XLOC-050925, MAP3K14-AS1, RP11148K1.12, and CTD2020K17.3 were coexpressed with p53 and may be involved in regulating insulin secretion by pancreatic beta cells. The expression trend of corresponding lncRNAs at all stages of induction was also analysed and verified by qPCR (Fig. 6AC). The qPCR results were consistent with the RNA-seq data (Fig. 7A–J).

Fig. 6
figure 6

The expression trend of coexpressed LncRNAs in differentiation. A The panel showed the coexpression network of coexpressed LncRNAs with TP53 in cluster1; B the panel showed the coexpression network of coexpressed LncRNAs with JARID2 and RPOX1 in cluster3; C boxplot of expression levels coexpressed LncRNAs with TP53 at various stages; D boxplot of expression levels coexpressed LncRNAs with JARID2 at various stages; E boxplot of expression levels of coexpressed LncRNAs with RPOX1 at various stages

Full size image
Fig. 7
figure 7

Validation of ten differentially expressed lncRNAs. **represents P < 0.05, ****represents P < 0.001, and the difference is statistically significant

Full size image

Discussion

Long noncoding RNAs (lncRNAs) are the largest component of the mammalian noncoding transcriptome, with a length greater than 200 nt [45]. According to the genomic position of lncRNAs relative to adjacent or overlap** protein-coding genes, they can be divided into sense, antisense, intron, intergenic or enhancer lncRNAs that mediate short- and long-range interactions between transcriptional enhancers and other regulatory elements in the genome [46]. lncRNAs can control the expression of cis or trans genes by directly interacting with transcription factors or recruiting chromatin modification complexes [47,48,49]. The expression levels of lncRNAs are generally lower than those of mRNAs, although a large proportion of the identified lncRNAs have high islet specificity, with a dynamic modulated pattern observed in differentiated beta-like cells in vitro, and most detected lncRNAs have increased specificity in functional endocrine differentiated cells [50]. Akerman et al. found that the lncRNAs HI-LNC-12, 15, 30, 78, 80, and 71 significantly affected the steady-state transcription levels of transcription factors of beta cells such as PDX1 and HNF1A. HI-LNC-12, HI-LNC-78, and HI-LNC-71 knockdown (KD) inhibited insulin secretion and reduced the insulin content in Endoc cells stimulated by glucose, and the expression levels of many lncRNAs were highly correlated with transcription factors (GLIS3, HNF1A, NKX2.2, PDX1, and MAFB) in beta cells [51]. Therefore, lncRNAs play important roles in the development of pancreatic beta cells and perform regulatory functions in pancreatic beta cells.

In this study, differentially expressed lncRNAs and mRNAs were obtained during the differentiation of stem cells into insulin-producing cells. GO analysis of these differentially expressed lncRNAs showed that the up-regulated genes were mainly functionally enriched in translation-related pathways, such as the initiation, extension, and termination of translation, and down-regulated genes were mainly concentrated in cell connection and adhesion and extracellular matrix tissue pathways. Differentially expressed lncRNAs had more pathway enrichment in the S4 stage compared with the S1, S2, and S3 stages based on KEGG analysis. Many target genes were involved in the NF-KB signalling pathway, MAPK signalling pathway, HIPPO signalling pathway, PI3K–Akt signalling pathway, and p53 signalling pathway. In a study of beta cells treated with free fatty acids that mimic pancreatic dysfunction and apoptosis induced by gut obesity, activation of p53 led to the induction of miR34a, a microRNA that sensitizes beta cells to apoptosis and inhibits the insulin exocytosis pathway, thus leading to impaired insulin secretion [52]. It has been shown that in beta cells, elevated glucose levels lead to elevated Ca2+ levels and enhanced NF-KB activity, which promotes insulin release. Overexpression of IKBa inhibits intracellular NF-KB activity and leads to impaired glucose-stimulated insulin secretion [53]. Studies have shown that PI3K-Akt is one of the major signalling pathways that maintains the survival and replication of beta cells and the expression and secretion of insulin genes [53,54,55]. The above discussion indicates that these pathways are related to the insulin secretion of islet beta cells.

LncRNAs can promote the differentiation of stem cells by interacting with miRNAs. Huang et al. found that lncRNA Gm10451 could regulate the histone H3K4 methyltransferase complex PTIP to facilitate insulin+/Nkx6.1+ beta-like cell differentiation by targeting miR-338-3p as a competing endogenous RNA (ceRNA), and Gm10451 loss in beta-like cells prevented the expression of mature beta cell markers, such as insulin, Nkx6.1, and MafA, after transplantation into streptozotocin (STZ)-mice [56]. Zou et al. demonstrated that lncRNA-ROR effectively maintains Sox2 gene expression through competitive binding to miR-145, thereby improving the efficiency of human amniotic epithelial stem cell (HuAEC) differentiation into beta islet-like cells [57].

In this research, we found that lncRNA CTBP1-AS2 is coexpressed with PROX1, PROX1 regulates pancreatic beta cell maturation by affecting MafA, and lncRNA CTBP1-AS2 is associated with T2DM. Therefore, lncRNA CTBP1-AS2 may promote beta cell maturation by regulating PROX1 and then MafA, which may be the direction of our future research. This also provides a new idea for improving the efficiency of the differentiation of stem cells into insulin-producing cells in the future.

Conclusion

In conclusion, HUC-MSCs combined with small molecule compounds were successfully induced to differentiate into IPCs. Differentially expressed lncRNAs may regulate the insulin secretion of pancreatic beta cells by regulating multiple signalling pathways. The lncRNAs AC009014.3, Gs1-72m21.1 and CTBP1-AS2 may be involved in the development of pancreatic beta cells, and the lncRNAs: XLOC_050969, LINC00883, XLOC_050981, XLOC_050925, MAP3K14-AS1, RP11-148K1.12, and CTD2020K17.3 may be involved in regulating the insulin secretion of pancreatic beta cells. However, the specific mechanisms underlying the involvement of these lncRNAs in the regulation of pancreatic development and insulin secretion of pancreatic beta cells need to be further explored, and the lncRNA CTBP1-AS2 will be our focus. These findings expand the lncRNA database relating to stem cell differentiation of IPCs and can also serve as a reliable foundation for further research.