ASCs from diabetes or old donators showed inferior proliferation and migration ability, exhibited a senescent phenotype
To figure out whether diabetes and aging affected the phenotype of ASCs, a series of fundamental experiments were conducted. At first, EdU assay showed that the proliferation rate of ASCs isolated from old donators was 1.6-fold lower than that from young donators (Fig. 1A, 1B). Meanwhile, diabetic conditions were not proved to reduce the proliferative ability of ASCs (Fig. 1A, 1B). Then, transwell assays revealed that the migration abilities of ASCs were the best in young group, followed by old group, and diabetic group turns to be the worst (Fig. 1C, 1D). Lastly, β-galactosidase staining showed the rate of senescent stem cells in diabetic group were significantly higher than that in old group, while ASCs in young group exhibited the minimal senescent cells (Fig. 1E, 1F). In conclusion, diabetes and aging both have somewhat detrimental effects on the phenotype of ASCs.
Diabetic conditions and aging impaired the capacities of ASCs for modulating fibroblasts and HUVECs function
Given above results, we desperately want to know whether diabetic conditions and aging impaired the therapeutic abilities of ASCs. Here we evaluated the differences of ASCs from these three groups in optimizing HUVECs and fibroblasts function. In vitro tube formation assay showed that HUVECs treated with condition media isolated from Y-ASCs exhibited 2-fold increases of closed tubular structures than that from O-ASCs (Fig. 1G, 1H). Meanwhile, the number of tubular structures in O-ASCs group was 2-fold more than D-ASCs group, but not statistically significant (Fig. 1G, 1H). Besides, wound scratch assay revealed that the migration ability of fibroblasts treated with condition media isolated from Y-ASCs was significantly promoted compared with that from O-ASCs (Fig. 1I, 1J). In addition, the number of migrative fibroblasts in D-ASCs group was slightly less than O-ASCs group (Fig. 1I, 1J). Therefore, high age of donators has remarkably deleterious effects on ASCs capacities for modulating cellular functions, while diabetic conditions of donators may have slightly but not significantly harmful effect on ASCs function.
Differential expression analyses: Y-ASCs vs O-ASCs and D-ASCs vs O-ASCs
Having known the unwell impact of diabetes and aging on ASCs phenotype and function, next we performed the RNA-seq analysis on ASCs from these three groups in order to uncover the underlying mechanisms. The differentially expressed miRNAs, mRNAs, lncRNAs and circRNAs from three groups (n = 3) were illustrated by the hierarchical clustering heat maps (Fig. 2). A total of 34 (27 up and 7 down) differentially expressed miRNAs, 553 (291 up and 262 down) differentially expressed mRNAs, 132 (39 up and 92 down) differentially expressed lncRNAs and 196 (123 up and 73 down) differentially expressed circRNAs were detected between Y-ASCs and O-ASCs (Fig. 2A-D, Data S1-4). Up means RNAs up-regulated in Y-ASCs compared to O-ASCs. Comparing D-ASCs with O-ASCs, we found a total of 43 (24 up and 19 down) differentially expressed miRNAs, 926 (483 up and 443 down) differentially expressed mRNAs, 535 (496 up and 39 down) differentially expressed lncRNAs and 1966 (1806 up and 160 down) differentially expressed circRNAs (Fig. 2E-H, Data S5-8). Up means RNAs up-regulated in D-ASCs compared to O-ASCs. Combining above results, we found 4 miRNAs, 22 mRNAs, 8 circRNAs were up regulated in O-ASCs compared to D-ASCs, and further up regulated in Y-ASCs compared to O-ASCs (Data S9-11). Moreover, 15 mRNAs, 14 lncRNAs and 17 circRNAs were down regulated in O-ASCs compared to D-ASCs, and further down regulated in Y-ASCs compared to O-ASCs (Data S10-12).
Functional enrichment analysis of differentially expressed mRNAs
GO and KEGG analyses were conducted to investigate the biological function of dysregulated mRNAs. Comparing Y-ASCs with O-ASCs, GO biological process analysis showed that 750 GO terms were significantly enriched (Data S13). Some terms like aging, positive regulation of cell proliferation, positive regulation of cell migration and angiogenesis may be associated with the dysfunction of ASCs related to donator’s age (Fig. 3A). KEGG pathway analysis showed that 47 pathways were statistically enriched (Data S14). Among them, cellular senescence, TGF-beta signaling pathway, p53-signaling pathway and PPAR signaling pathway may be involved in the impaired function of ASCs (Fig. 3B). Comparing O-ASCs with D-ASCs, 553 GO terms were significantly enriched in GO biological process analysis (Data S15). Some terms such as oxidation-reduction process, aging, cell migration, wound healing and angiogenesis may be associated with the undesirable therapeutic capacity of ASCs exposing to diabetes (Fig. 3C). KEGG pathway analysis revealed that 37 pathways were statistically enriched (Data S16). Of them, pathways like AGE-RAGE signaling pathway, TGF-beta signaling pathway, p53-signaling pathway, mTOR signaling pathway, cellular senescence may be participated in the unsatisfactory function of ASCs isolated from diabetic donators (Fig. 3D).
PPI networks construction
PPI networks were constructed in order to identify the critical genes from those differentially expressed mRNAs. Comparing Y-ASCs with O-ASCs, the network was established and compromised of 107 nodes and 373 edges (Fig. 3E). Of them, the top 15 genes with high degree were APOE, BMP4, EGR1, GPC3, CXCL8, SPP1, FOS, HGF, TNC, IGFBP7, JUN, APOL1, GBP2, HLA-F, OAS1 (Fig. 3E). Comparing O-ASCs with D-ASCs, 131 nodes and 783 edges were included in the network (Fig. 3F). Among them, the top 15 genes with high degree were C3, CXCL8, IL6, CXCL1, CCNB1, BUB1, CXCL12, MCHR2, CXCL2, CXCL6, CXCR4, CXCL5, QSOX1, S1PR1, TAS2R31 (Fig. 3F).
RT-PCR confirmation of differentially expressed miRNAs
We randomly selected 14 miRNAs to validate the results of RNA-seq data separately. Comparing Y-ASCs with O-ASCs, miR-145-3P, miR-145-5p, miR-126-3p, miR-126-5p, miR-214-3p, miR-181a-3p, miR-210-3p were identified to be up regulated in Y-ASCs by PCR analysis, which was consistent with the RNA-seq results (Fig. 4A, 4B). Besides, PCR results showed miR-766-3p was down regulated in Y-ASCs, which was opposite with that in the RNA-seq data (Fig. 4A, 4B). Comparing D-ASCs with O-ASCs, the PCR results of miR-214-3p, miR-193a-3p, miR-3529-3p, miR-145-3p, miR-302a-3p, miR-302b-3p were verified to be statistically significant and consistent with the RNA-seq data (Fig. 4C, 4D). However, the PCR results of two miRNAs (miR-615-3p, miR-12136) were opposite with the RNA-seq data (Fig. 4C, 4D). Other 6 miRNAs which showed no significant difference between these groups by PCR analysis were illustrated in Fig. 4E-F.
Construction of circRNA or lncRNA-miRNA-mRNA network
CeRNA networks were constructed based on the differentially expressed RNAs. MiRNAs verified by PCR analysis and consistent with RNA-seq data were selected to be the core of ceRNA networks. Comparing Y-ASCs with O-ASCs, the lncRNA associated ceRNA network was compromised of 4 miRNAs, 171 mRNAs and 46 lncRNAs (Fig. 5A), and the circRNA associated ceRNA network contained 3 miRNAs, 160 mRNAs and 36 circRNAs (Fig. 5B). Comparing D-ASCs with O-ASCs, 4 miRNAs, 118 mRNAs and 97 lncRNAs were included in the lncRNA associated ceRNA network (Fig. 6A), 4 miRNAs, 119 mRNAs and 103 circRNAs were involved in the circRNA associated ceRNA network (Fig. 6B).
RT-PCR confirmation of differentially expressed mRNAs, lncRNAs and circRNAs
MRNAs, lncRNAs and circRNAs contained in above PPI and ceRNA networks were selected for further RT-PCR analyses. Comparing Y-ASCs with O-ASCs, RT-PCR results showed 14 mRNAs were significantly dysregulated (Fig. 7A). Among them, MMP3, EGR1, JUNB, BMP4 were involved in the PPI network (Fig. 3E, 7A), ANK2, PODN, NOVA1, CLEC3B, RARRES3, IGFBP6, MFAP5 were involved in the ceRNA network (Fig. 5A-B, 7A), and ITGB8, WNT11, BMPER were involved in both PPI and ceRNA networks (Fig. 3E, 5A-B, 7A). Comparing D-ASCs with O-ASCs, 11 mRNAs were verified to be differentially expressed by PCR analyses (Fig. 7B), of which CXCL8, CXCL1, CXCL6, CXCL5 were all up-regulated in O-ASCs and involved in the PPI network (Fig. 3F, 7B). Besides, SOX4, ANGPT1, SLC5A3, NDUFV3, LUC7L3 were involved in the ceRNA network (Fig. 6A-B, 7B), and COL18A1, FMOD were involved in both PPI and ceRNA networks (Fig. 3F, 6A-B, 7B). Moreover, we found 5 lncRNAs (RAET1E-AS1, LOC102723591, LOC105373230, LOC112267883, LINC02595) and 5 circRNAs (hsa_circ_0017534, hsa_circ_0092630, hsa_circ_0080906, hsa_circ_0075045, hsa_circ_0080909) were significantly up-regulated in O-ASCs comparing to Y-ASCs (Fig. 7C). Also, the PCR results revealed that 2 lncRNAs (LOC102724087, LOC101928000) and 3 circRNAs (hsa_circ_0088199, hsa_circ_0088195, hsa_circ_0058158) were significantly up-regulated in O-ASCs comparing to D-ASCs (Fig. 7D). Adversely, 2 lncRNAs (LOC105377989 and NEAT1) were significantly down-regulated in O-ASCs comparing to D-ASCs (Fig. 7D). Other RNAs without statistically significant difference between these groups were displayed in supplementary Figure S1A-B.
Establishment of ceRNAs sub-networks based on the RNAs verified by PCR analyses
Based on above PCR results, we drew the circRNAs and lncRNAs-miRNAs-mRNAs sub-networks for further investigation. Comparing Y-ASCs with O-ASCs, the network included 5 miRNAs, 10 mRNAs, 5 lncRNAs and 5 circRNAs (Fig. 7E). Comparing D-ASCs with O-ASCs, 6 miRNAs, 7 mRNAs, 4 lncRNAs and 3 circRNAs were compromised in the network (Fig. 7F). Then, we randomly selected hsa-miR-145-5p for our further study, which had ceRNA relationships with four mRNAs (ITGB8, WNT11, BMPER, RARRES3), two lncRNAs (RAET1E-AS1, LOC102723591) and one circRNA named hsa_circ_0092630 (Fig. 7E). The correlation analyses revealed that the expression levels of RAET1E-AS1, WNT11 and BMPER had significantly negative correlations with that of hsa-miR-145-5p (Fig. 7G-I), while others not (supplementary Figure S1C-D). Meanwhile, the expression level of RAET1E-AS1 had a significantly positive correlation with that of WNT11 and BMPER (Fig. 7J-K). Lastly, we found overexpression of has-miR-145-5p in ASCs could significantly down-regulated the expression levels of WNT11, BMPER, RAET1E-AS1, LOC102723591, hsa_circ_0092630 (Fig. 7L). Therefore, we speculated that the RAET1E-AS1- hsa-miR-145-5p- WNT11/BMPER network might play a vital role in the phenotype and function of ASCs.
Overexpression of hsa-miR-145-5p improved the phenotype and function of ASCs, while inhibition of it adversely.
To evaluate the character of the RAET1E-AS1- hsa-miR-145-5p- WNT11/BMPER axis, here we modulated the expression level of hsa-miR-145-5p in ASCs to investigate the changes of their phenotypes and functions. The EdU and transwell assays showed that the proliferative and migrative abilities of ASCs have been promoted more than two-fold after hsa-miR-145-5p overexpression, while inhibition of it significantly reduced these abilities (Fig. 8A-D). Also, the number of β-galactosidase stained cells in the miR-145-5p group was two-fold less than that of the mimic-NC group, while inhibition of miR-145-5p significantly increased the stained cells (Fig. 8E-F). Next, the in vitro tube formation and wound scratch assays revealed that cellular supernatants of cultured ASCs with miR-145-5p overexpression could remarkably enhance the function of HUVECs and fibroblasts as shown by increases of closed tubular structures and migrative cells, while inhibition of miR-145-5p exhibited the opposite results (Fig. 8G-J). In addition, the western blot analysis showed overexpression of miR-145-5p in ASCs could significantly improve the expression of migration associated protein FN1, proliferation associated proteins CCNA1 and CCND1, pluripotent markers NANOG and OCT4, and decrease the expression of senescence associated protein P21 (Fig. 8K). The PCR analysis showed overexpression of miR-145-5p in ASCs could significantly improve the expression of MMP9, NANOG, growth factors PDGFA and FGF, and decrease the expression of IGF1, HIF, CXCL8, and senescent markers TP53 and CDKN2A (Fig. 8L). In conclusion, overexpression of miR-145-5p in ASCs could ameliorate the unsatisfactory phenotype and function of ASCs isolated from old donators.