1. H19 controls matrisome of MSC-derived pre-osteoblasts
To screen for matrisome-specific proteins regulated by H19, we performed a mass spectrometry-based quantitative proteomic analysis in MSCs isolated from osteoporotic patients, following induction into the osteogenic lineage. In these cells, levels of H19 were successfully reduced by transfection with siH19, with an average efficiency of > 75% decrease (Fig. 1a). Cell fractionation shows that, contrary to other typically nuclear lncRNAs such as XIST and MALAT1, H19 transcripts are mainly located in the cytoplasm (Fig. 1b). A matrix enrichment buffer was used to facilitate solubilization of ECM proteins derived from siH19-MSCs and siCTR-MSCs. Proteomic profiling identified 4020 proteins, considering a minimum of two unique peptides. Results show that 66 proteins were significantly upregulated [Fold-change (FC) > 1.25; P < 0.05], whereas 102 were significantly downregulated (FC<-1.25; P < 0.05), in MSCs with reduced levels of H19 compared with the control (Fig. 1c).
The GO analysis on the biological processes revealed that extracellular matrix organization (GO:0030198; fold enrichment (FE) = 5.55, P = 3.52x10− 7; false discovery rate (FDR) = 3.09x10− 4) and extracellular structure organization (GO:0043062; FE = 5.54, P = 3.63x10− 7; FDR = 3.01x10− 4) are among the top 20 terms with higher significance and low FDR (Fig. 1d). Further analysis revealed that extracellular matrix structural constituent (GO:0005201) is the fourth term with the highest FE score, and within the top ten with the lowest P-value (P = 2.98x10− 6; FDR = 2.03x10− 3) (Fig. 1d). The following proteins are associated with this term: collagen alpha-1(I) chain (COL1A1; P02452); EGF-containing fibulin-like extracellular matrix protein 1 (EFEMP1; Q12805); collagen triple helix repeat containing 1 (CTHRC1; Q96CG8); fibrinogen beta chain (FGB; P02675); fibrillin-1 (FBN1; P35555); fibulin-1 (FBLN1; P23142); elastin (ELN; P15502); vitronectin (VTN; P04004) and lactadherin (MFGE8; Q08431). Among the 20 most enriched cellular components terms are collagen-containing extracellular matrix (GO:0062023; FE = 7.45, P = 1.67x10− 12; FDR = 4.75x10− 10) and extracellular matrix (GO:0031012; FE = 5.56, P = 3.09x10− 10; FDR = 6.16x10− 8) (Fig. 1d).
Venn diagram was used to compare the differently expressed proteins from the proteomic analysis and the GO cellular terms associated to ECM proteins. Results show that 38 of the differently expressed proteins in the siH19-engineered MSCs versus control MSCs are associated with the cellular functions “extracellular matrix” and “non-structural extracellular” (Fig. 1e). Out of those, 14 proteins are upregulated, while 24 are downregulated following H19 knockdown, compared with control. Specifically, the differently expressed proteins FBN1 (FC = 1.53, P = 8.06x10− 6), VTN (FC=-1.72, P = 5.09x10− 5) and COL1A1 (FC=-1.31, P = 0.03) are associated with ECM-related GO terms
(Fig. 1f), while CTHRC1 (FC=-1.374, P = 0.0085) is associated with collagen biogenesis.
2. ECM components are regulated by H19
To validate novel ECM proteins regulated by H19, the levels of FBN1, VTN and CTHRC1 were evaluated at the transcriptional level. RT-qPCR results show a 42.5% upregulation for FBN1 and a 24.1% and 42.0% downregulation for VTN and CTHRC1, respectively, in siH19-MSCs compared with control, following 7 days of osteogenic differentiation (Fig. 2a). Furthermore, immunocytochemistry was used to detect these proteins in the MSC-derived matrices. In agreement with the RT-qPCR, results show an increase for FBN1, while VTN is decreased in siH19-MSCs versus control-MSCs (Fig. 2b). Although not statistically significant, results also show a tendency for a reduction in CTHRC1 protein levels (Fig. 2b).
Being collagen type I (COL1) a major component of bone ECM, it levels were also quantified by immunocytochemistry. COL1 is reduced in siH19-MSCs versus control-MSCs (97.4%), following 7 days of osteogenic induction (Fig. 2c). In agreement, mRNA expression of COL1A1, COL1A2, and COL3A1, are reduced by siH19 (Fig. 2d).
3. H19 impairs osteogenic differentiation and mineralization
To evaluate the biological impact of H19 on MSCs, loss-of-function studies were performed. The decrease in H19 levels caused a reduction in expression levels of RUNX2 and ALP, two osteogenic markers (Fig. 2e). This was further confirmed through ALP staining, which shows a significant decrease (P < 0.05) in siH19-MSCs compared with control, after culture for 7 days in osteogenic-inducing conditions (Fig. 2f). Furthermore, H19 knockdown revealed a 79.7% decrease (P < 0.05) on the deposition of calcium nodules, after 10 days of differentiation (Fig. 2g), which demonstrates that H19 has an impact on mineralization.
Moreover, results also show a significant reduction in MSCs proliferation, assessed through measurement of the metabolic activity, after H19 depletion compared with control MSCs (Fig. 2h). To further confirm this result, cells were stained for the proliferation marker Ki-67. Results show that H19 disturbs MSCs proliferative capacity, considering the reduction in the percentage of Ki-67+/DAPI+ (P < 0.05) in the siH19-MSCs (Fig. 2h). Nevertheless, flow cytometry results show that inhibition of H19 do not affect early apoptosis or late apoptosis/cell death (Fig. 2i). Taken together, these results indicate that inhibition of H19 expression exerted an osteogenic- and proliferative- suppressive effect in MSCs in vitro.
4. ECM derived from siH19-engineered cells affect the behaviour of naïve MSCs
Considering the impact of H19 on the ECM components and on osteogenesis, we next evaluated the osteogenic behaviour of naïve MSCs on the matrices derived from siH19 cells (Fig. 3a). Firstly, MSCs were transfected with either siH19 or control, and allowed to differentiate for 10 days in osteogenic-inducing conditions to promote the deposition of an ECM network. Next, matrices were decellularized and characterized. No DAPI positive cells or Gill’s hematoxylin stained cells were found in the decellularized matrices, compared with non-decellularized MSCs, which is in agreement with the quantification of low amounts of DNA content (Supplementary Fig. 1, Additional file 1). There are no differences in GAGs content, comparing decellularized matrices derived from siH19-MSCs or control-MSCs (Fig. 3b). As expected, the collagenous content in decellularized matrices derived from siH19-MSCs, following osteogenic differentiation, is significantly reduced (63.8%, P < 0.05) compared with the controls (Fig. 3c). Furthermore, AFM microscopy images of the decellularized matrices show the presence of a continuous and thicker layer of ECM on the control matrix in opposition to a thinner and porous network of extracellular matrix fibers on the siH19-derived matrices (Fig. 3d), suggesting an impairment in the ECM production by H19 silencing.
To evaluate if the changes in the matrices secreted by siH19-engineered MSCs influence MSC lineage commitment, naïve cells were used to repopulate the decellularized matrices. Following 7 days of culture, results show that RUNX2, an osteogenic key marker, is significantly downregulated, while CEBPB, an adipogenic marker is significantly upregulated, when MSCs are cultured on top of siH19-derived matrices, compared with control-derived matrices (Fig. 3e). No significant differences in gene expression were found for ALP and CEBPA, neither for the chondrogenic markers, SOX9 and DLX5 (Fig. 3e). Furthermore, immunocytochemistry results show that osteopontin (OPN/SPP1/BSP-1), which is expressed by osteoblasts, is decreased when naïve MSC are reseed on siH19-derived matrices (Fig. 3f), suggesting the secreted matrices are detrimental for osteogenesis when H19 levels are reduced. Gene expression results confirmed the downregulation of OPN when MSCs are cultured on top of siH19-derived matrices (Fig. 3g).
To evaluate if MSC proliferation is influenced by siH19-derived matrices, naïve MSCs were cultured on top of the decellularized matrices. Results show a decrease in the metabolic activity (Fig. 3h) and in the proliferative capacity (Fig. 3i), evaluated by the percentage of Ki-67+/DAPI+ cells, in naïve MSC cultures on top of siH19-decellularived versus control-decellularized matrices. Also, levels of LDH on the conditioned-media were measured to assess cell viability/cytotoxicity. No differences were found between MSCs cultured on top of the different matrices (Fig. 3j), with LDH levels similar to those detected when cells are cultured in TCPS flasks, which excludes a cytotoxic effect. Overall, these results indicate that changes in the matrix induced by siH19 affect the lineage commitment and the proliferative capacity of naïve MSCs derived from osteoporotic patients.
5. ECM derived from siH19-engineered cells affect the behaviour of pre-adipocytes
Commitment of MSC into the osteogenic lineage instead of the adipogenic lineage is crucial for a healthy bone. To understand the impact of the matrices on the behaviour of pre-adipocytes, MSCs were differentiated for 10 days in adipogenic-inducing conditions (pre-adipocytes), and then used to reseed the decellularized matrix produced by osteogenic-induced MSCs with H19 reduced levels or control. Results show that after 7 days in culture, the expression of CEBPB was significantly increased (P < 0.01) on pre-adipocytes cultured on top of the siH19-derived matrix, compared with cells cultured on top of the control matrix, with no differences regarding osteogenic or chondrogenic markers (ALP, RUNX2; SOX9, DLX5) (Fig. 4a). Interestingly, changes in the number of cells with lipid droplets was observed with only 3 days of culture on top of the decellularized matrices. Specifically, the percentage of Oil Red O+ cells (cells with lipid droplets) was significantly increased in the siH19-derived matrices (P < 0.001) (Fig. 4b), compared with control-matrices, following normalization to the total number of cells (number of nuclei stained with hematoxylin). Similarly to the results from naïve MSC, we did not detect a cytotoxic effect of the decellularized matrices on the pre-adipocytes (Fig. 4c). Globally, the results suggest H19 knockdown promotes adipogenesis and increases the number of cells with lipid droplets. Interestingly, in bone clinical samples from osteoporosis patients, we found a significant positive correlation between the percentage of area covered by adipose tissue and the trabecular separation (Tb.Sp), which also positively correlates with age (Fig. 4d).
6. miR-29c-3p controls H19 expression
To identify potential regulators of H19, an in silico analysis on miRNAs was performed using ENCORI platform (http://starbase.sysu.edu.cn/panCancer.php), and potential binding sites for the miR-29 family, including miR-29a, miR-29b and miR-29c members, are shown in Supplementary Fig. 2, Additional file 1. To understand the relevance of those miRNAs in osteoporosis, we evaluated their expression levels in bone samples from osteoporotic fragility fractures (FF) and osteoarthritis (OA) patients. RT-qPCR results show that miR-29b-3p is poorly expressed in the bone samples or even undetected (Cq > 35). However, miR-29a-3p and miR-29c-3p were expressed and significantly decreased in FF versus OA bone samples (Fig. 5a). A multivariate analysis was performed to exclude possible associations between miRNA expression and clinically relevant variables, including age. When using miRNA expression and age as independent variables, and disease as a dependent variable, the results from the multivariate linear regression analysis with the stepwise forward method show that only miR-29c-3p, but not miR-29a-3p, is statistically significant (Fig. 5b), being a strong predictive factor of the disease.
Next, to evaluate if miR-29c-3p regulates H19, the expression levels of H19 were tested after modulation of miR-29c-3p in MSCs cultured for 3 days in osteogenic-inducing conditions (Supplementary Fig. 3, Additional file 1). Results show that miR-29c-3p overexpression significantly decreases H19 (FC = 0.58), while miR-29c-3p inhibition increases H19 levels (FC = 2.44)
(Fig. 5c).
7. miR-29c-3p reduces collagen expression and MSC proliferation but do not impact ALP and mineralization
To understand if miR-29c-3p impacts H19-mediated processes, collagen secretion, ALP level, mineralization, proliferation and apoptosis in MSCs was investigated. Immunostaining against COL1, after 7 days of culture in osteogenic-inducing conditions, shows that COL1 proteins are reduced by 94.4 % when MSCs are transfected with miR-29c-3p mimics, while COL1 protein levels are highly increased following inhibition of miR-29c-3p, compared with the respective controls (Fig. 6a). At the transcription level, a reduction was found for COL1A1, COL1A2 and COL3A1 in miR-29c-3p-overexpressing-MSCs, whereas the opposite was observed miR-29c-3p-inhibition-MSCs (Fig. 6b). Interestingly, in osteoporotic bone samples, there is a negative correlation between miR-29c-3p expression levels and COL1A1 (Fig. 6c). To test if miR-29c-3p regulates osteogenesis, ALP staining and mineralization were analyzed. However, no differences were found, suggesting that the modulation of miR-29c-3p does not impair ALP or calcium deposition (Fig. 6d, 6e and 6f). On the other hand, differences were found regarding proliferation. The metabolic activity is significantly increased when miR-29c-3p levels are reduced and, although not statistically significant, there is a tendency for a reduction in the metabolic activity following miR-29c-3p overexpression (Fig. 6g). These results were confirmed when analyzing the proliferation marker Ki-67. The percentage of Ki-67+/DAPI+ is significantly decreased by miR-29c-3p and increased following miR-29c-3p inhibition (Fig. 6g). Furthermore, analysis of Annexin V/PI staining by flow cytometry showed that the miR-29c-3p overexpression significantly decreases the percentage of viable cells, while increasing the percentage of cells in late apoptosis/cell death (Fig. 6h). Taken together, these results suggest that
miR-29c-3p regulates levels of collagen proteins and MSCs proliferation, but do not impair ALP or mineralization.