Decorin inhibits the insulin-like growth factor I signaling in bone marrow mesenchymal stem cells of aged humans

Age-related changes in the mitogenic activity of bmMSCs in response to IGF-I

Human bmMSCs were isolated from 6 adult donors (35~43 years), 8 middle age donors (46~57 years), and 17 aged donors (65~79 years). To examine the correlation between age and IGF-I-induced DNA synthesis in bmMSCs, we treated serum-starved bmMSCs with 5, 50, and 250 ng/ml of IGF-I, and performed BrdU incorporation assays. The correlation coefficients between DNA synthesis and age for each dose were -0.6077 (P<0.0005), -0.7781 (P<10^-6), and -0.559 (P<0.001), respectively, and the IGF-I concentration required for a 50% induction in DNA synthesis was shifted from 50 ng/ml for cells from adult donors to 250 ng/ml for cells from aged donors (Figure 1A). To examine if age decreased BrdU uptake, which might result in underestimate of the DNA synthesis in the aged group, we also examined the BrdU incorporation in these serum-starved bmMSCs without IGF-I treatment, and found that there was no significant difference between adult and middle age groups (P=0.823), and between adult and aged groups (P=0.591) (Figure 1B). These data suggested that age was unlikely to decrease BrdU uptake by these serum-starved bmMSCs.

Figure 1. Effect of age on the mitogenic response of human bmMSCs to IGF-I. BrdU incorporation analyses. Human bmMSCs were maintained in serum-free media for 24 h, and then subjected to BrdU incorporation analyses with or without concomitant treatment of 5, 50, and 250 ng/ml of IGF-I. (A) The correlation between the age and the IGF-I-triggered DNA synthesis was analyzed. (B) The average OD₄₅₀ values of bmMSCs of adult (n=6), middle aged (n=8), and aged (n=17) groups cultured without IGF-I treatment are shown. The difference between the groups was analyzed by the Student’s t-test. (C) RT-qPCR analyses. The expression of IGF-IR and IGF-I mRNAs of the aged bmMSCs (n=17) were compared to those of the adult bmMSCs (n=6) (to which a value of 1 was assigned). Data represent the mean ± S.D. from three experiments. Student’s t-test was used to analyze the differences between the groups. (D) Western blot analyses. The protein levels of IGF-IR in the Aged-1, Aged-2, Adult-1, and Adult-2 cells are shown.

To investigate the mechanism underlying the age-related impairment in the IGF-I-induced DNA synthesis, we examined the expression of IGF-I and its receptor (IGF-IR) in the adult and aged bmMSCs. RT-qPCR analyses showed that the expression of IGF-IR and IGF-I mRNAs in bmMSCs from aged human donors (n=17) was similar to those in bmMSCs from adult donors (n=6) (Figure 1C). Analyses of the IGF-IR protein levels in bmMSCs from randomly selected aged donors (Aged-1 and Aged-2) and adult subjects (Adult-1 and Adult-2) also suggested that IGF-IR expression might not decrease with aging (Figure 1D). Similar results were seen with the use of bmMSCs isolated from Fisher 344 rats with age ranging from 3 to 21 months old (Supplementary Figure 1). Together, these results indicated an age-related impairment in the IGF-I-induced mitogenic activity of bmMSCs. In addition, IGF-IR expression was not down-regulated by aging, and higher doses of IGF-I could substantially increase the DNA synthesis in bmMSCs from aged donors.

IGF-IR auto-phosphorylation in the age-related impairment of IGF-I signaling

The binding affinity of IGF-I to IGF-IR of murine bmMSCs has been shown not to change with aging [11, 17]. Given our finding that aging did not decrease IGF-IR expression, we set out to examine if aging down-regulated IGF-I-triggered IGF-IR activation. We examined the effect of AG1024, an inhibitor of IGF-IR signaling, on the IGF-I-induced DNA synthesis in Aged-1 and Aged-2 cells. Our data showed that 200 ng/ml of IGF-I caused approximately 52% and 41% increase of DNA synthesis in Aged-1 and Aged-2 bmMSCs, respectively, but the induction was inhibited by the co-treatment of 1 μM of AG1024 (Figure 2A). Similar results were also shown by analyzing rat bmMSCs (Supplementary Figure 1). Since AG1024 targets IGF-IR auto-phosphorylation for down-regulation and phosphorylation at the Ser1135 and Ser1136 sites of IGF-IR activates the receptor kinase, we examined if aging decreased the Ser1135/1136-phosphorylation of IGF-IR. We treated serum-starved Adult-1 and Aged-1 bmMSCs with 0, 50, and 250 ng/ml of IGF-I for 0, 5, 10, and 20 min. Western blot analyses showed that Ser1135/1136-phosphorylated IGF-IR was barely detected in the serum-starved, untreated Adult-1 and Aged-1 bmMSCs. Treatment with 50 ng/ml of IGF-I for 5 min caused approximately 150% and 40% increase of phosphorylation in Adult-1 and Aged-1 cells, respectively, while treatment with 250 ng/ml of IGF-I for 5 min caused approximately 130% and 150% increase of phosphorylation in Adult-1 and Aged-1 cells, respectively (Figure 2B). The IGF-IR levels decreased with increasing IGF-I doses and with time in both types of cells, which might be due to the internalization and subsequent degradation of the IGF-I-IGF-IR complex. So, IGF-IR auto-phosphorylation in response to IGF-I was impaired in Aged-1 cells, whereas this impairment was counteracted by high doses of IGF-I. These results were consistent with those shown in Figure 1A, suggesting that aging inhibited IGF-IR activation and down-regulated the mitogenic activity of bmMSCs in response to IGF-I. The aging-related impairment in IGF-IR auto-phosphorylation was also seen with rat bmMSCs (Supplementary Figure 2).

Figure 2. Effect of aging on the auto-phosphorylation of IGF-IR of bmMSCs. (A) BrdU incorporation analyses. Serum-starved Aged-1 and Aged-2 bmMSCs were examined for the IGF-I-induced DNA synthesis with or without concomitant treatment of AG1024 (1 μM). Relative DNA synthesis was calculated by compared the OD₄₅₀ readings of the treated cells to that of the untreated (U) cells. (to which a value of 1 was assigned). Data represent the mean ± S.D. from three experiments. Student’s t-test was used to analyze the differences between the groups. (B) Western blot analyses. Serum-starved Aged-1 and Adult-1 cells were treated with 0, 50, and 250 ng/ml of IGF-I for 0, 5, 10, and 20 min. The auto-phosphorylation of IGF-IR was examined. Representative blots are shown. All the signals were compared to that of the untreated cells at time 0 (to which a value of 1 was assigned). Data represent the mean ± S.D. from three experiments. A one-way ANOVA plus Scheffe’s post hoc tests were used to analyze the differences among the untreated and IGF-I-treated groups. ^*, P<0.05 versus untreated control.

Effect of DCN on the DNA synthesis and the auto-phosphorylation of IGF-IR of human bmMSCs

Based on our previous finding that the expression of DCN mRNA in human bmMSCs was increased with advancing age (r=0.57) [5], we examined if DCN played a role in the impairment of IGF-IR activation in response to IGF-I by aging. Consistent with previous findings, RT-qPCR analyses showed that the DCN mRNA levels in bmMSCs from aged donors was approximately 2.7 fold of that in bmMSCs from adult donors (Figure 3A), while the DCN protein levels in Adult-1, Adult-2, Aged-1, and Aged-2 cells also showed that DCN expression increased with aging (Figure 3B). To examine the role of DCN in the IGF-I signaling of bmMSCs, we knocked down DCN expression (approximately 52%) in Aged-1 cells, and examined the IGF-I-induced DNA synthesis. DCN knockdown did not affect IGF-IR expression (Figure 3C). BrdU incorporation assays showed that 200 ng/ml of IGF-I caused approximately 22% (P<0.05), 75% (P<0.05), and 20% (P<0.05) increase in DNA synthesis in the parental, DCN-knockdown (shDCN), and empty-vector control (shEV) Aged-1 cells, respectively (Figure 3D). So, DCN knockdown significantly increased IGF-I-induced DNA synthesis (P<0.001). Consistently, DCN knockdown enhanced IGF-I-induced DNA synthesis in Aged-2 cells (Supplementary Figure 3). Subsequently, we performed dose-response experiments to examine the effects of DCN knockdown on IGF-IR auto-phosphorylation. We treated serum-starved shEV and shDCN cells with or without increasing doses of IGF-I for 5 min, and examined the phosphorylated-IGF-IR-to-IGF-IR ratios (Figure 3E). We used linear regression to analyze the response rates of shEV and shDCN cells. The correlation coefficients between IGF-I concentrations and IGF-IR auto-phosphorylation were 0.4 (P=0.005) for shEV, and 0.7 (P<0.0001) for shDCN, indicating that the IGF-IR auto-phosphorylation in both cells increased significantly in response to increasing IGF-I doses, and that the response rate of shDCN cell was significantly higher than that of shEV cells (Figure 3E). On the other hand, IGF-I seemed not to modulate DCN expression in both groups of cells (Supplementary Figure 4A).

Figure 3. Effect of DCN knockdown on the DNA synthesis and IGF-IR auto-phosphorylation of aged bmMSCs. (A) RT-qPCR analyses. The expression of DCN mRNA of bmMSCs from 17 aged donors was compared to that of the cells from 6 adult donors (to which a value of 1 was assigned). Data represent the mean ± S.D. from a triplicate analysis. Student’s t-test was used to analyze the differences between the groups. (B) Western blot analyses of DCN levels. Representative blots of the DCN levels in the Aged-1, Aged-2, Adult-1, and Adult-2 cells are shown. (C) Western blot analysis of DCN and IGF-IR levels in cells with DCN knockdown. Aged-1 cells were infected with Lenti virus to generate DCN-knockdown (chDCN) and empty vector control (shEV) cells. The DCN and IGF-IR protein levels of the parental Aged-1, shDCN, and shEV cells are shown. (D) BrdU incorporation analyses. Serum-starved parental, shDCN, and shEV cells were treated with 200 ng/ml IGF-I for 24h, and examined for the IGF-I-induced DNA synthesis. The DNA syntheses in these cells were compared to that of the untreated parental cells (to which a value of 1 was assigned). Data represent the mean ± S.D. from three experiments. A one-way ANOVA plus Scheffe’s post hoc tests were used to analyze the differences among the untreated and IGF-1-treated groups. ^*, P<0.05 versus untreated control. Student’s t-test was used to analyze the differences between the groups. (E) Western blot analyses of IGF-IR auto-phosphorylation. Serum-starved shDCN and shEV cells were either treated with varying doses of IGF-I for 5 min or left untreated. IGF-IR auto-phosphorylation was examined and normalized to total IGF-IR expression. The difference in the response rates between shDCN and shEV cells was analyzed by linear regression analyses.

We also overexpressed DCN (approximately 68%) in Adult-1 cells (Figure 4A), and examined the IGF-I-induced DNA synthesis. DCN overexpression seemed not to affect IGF-IR expression (Figure 4A). BrdU incorporation assays showed that 200 ng/ml of IGF-I caused approximately 69% (P<0.05), 66% (P<0.05), and 72% (P<0.05) increase in DNA synthesis in the parental, DCN-overexpressing (DCN), and empty-vector control (EV) Adult-1 cells, respectively, and DCN overexpression seemed not to decrease DNA synthesis (Figure 4B, left). Considering that high dose of IGF-I might counteract the inhibitory effect of DCN overexpression (68%) on DNA synthesis, we decreased IGF-I dose. As a result, 50 ng/ml of IGF-I caused approximately 50% (P<0.05), 21% (P<0.05), and 52% (P<0.05) increase in DNA synthesis in those cells, respectively, and DCN overexpression significantly decreased the IGF-I-induced DNA synthesis (P<0.05) (Figure 4B, right). Consistently, IGF-I (50 ng/ml) induced DNA synthesis in Adult-2 cells, which was decreased by DCN overexpression (Supplementary Figure 3). To examine the rates of IGF-IR auto-phosphorylation of control (EV) and DCN-overexpressing (DCN) Adult-1 cells in response to IGF-I, we treated these serum-starved cells with IGF-I as described for the shEV and shDCN cells (Figure 3E). IGF-I seemed not to modulate DCN expression in both groups of cells (Supplementary Figure 4B). Linear regression analysis revealed no significant difference in the response rates between EV and DCN cells. However, if the data from cells treated with 200 ng/ml IGF-I were excluded from the comparison, the correlation coefficients between IGF-I concentrations and IGF-IR auto-phosphorylation were 0.35 (P<0.0001) for EV cells and -0.17 (P=0.024) for DCN cells. These data indicated that the IGF-IR auto-phosphorylation in both cells increased significantly in response to increasing IGF-I doses, and that the response rate of DCN cell was significantly lower than that of EV cells (Figure 4C). Taken together, our data indicated that DCN acted as an inhibitor of IGF-IR of human bmMSCs.

Figure 4. Effect of DCN overexpression on the DNA synthesis and IGF-IR auto-phosphorylation of adult bmMSCs. (A) Western blot analyses of DCN and IGF-IR levels in cells overexpressing DCN. Adult-1 cells were infected with Lenti virus to generate DCN-overexpressing (DCN) and empty vector control (EV) cells. The DCN and IGF-IR protein levels of the parental Aged-1, DCN, and EV cells are shown. (B) BrdU incorporation analyses. Serum-starved parental, DCN, and EV cells were examined for DNA synthesis induced by IGF-I (50 and 200 ng/ml). The DNA syntheses in these cells were compared to that of the untreated parental cells (to which a value of 1 was assigned). Data represent the mean ± S.D. from three experiments. A one-way ANOVA plus Scheffe’s post hoc tests were used to analyze the differences among the untreated and IGF-1-treated groups. ^*, P<0.05 versus untreated control. Student’s t-test was used to analyze the differences between the groups. (C) Western blot analyses of IGF-IR auto-phosphorylation in cells overexpressing DCN. Serum-starved DCN and EV cells were either treated with varying doses of IGF-I for 5 min or left untreated, and the IGF-IR auto-phosphorylation in response to IGF-I was examined by Western blot analyses. The difference in the response rates between DCN and EV cells was analyzed by linear regression.

Effect of IGF-I on the binding of DCN to IGF-IR

To examine how IGF-I counteracted the inhibitory effect of DCN on IGF-IR, we performed immunoprecipitation and Western blot analyses on serum-starved Aged-1 cells with or without IGF-I treatment to examine if DCN interacted with IGF-IR, and if IGF-I decreased the binding of DCN to IGF-IR. Results showed that a substantial amount of DCN was co-precipitated with IGF-IR, and that IGF-I add-back (250 ng/ml, 5 min) induced IGF-IR auto-phosphorylation and decreased the binding of DCN to IGF-IR, but did not increase IGF-IR levels in the input and precipitates (Figure 5). These data suggested that IGF-I and DCN might bind to the IGF-IR of human bmMSCs in a competitive manner. We also examined the co-precipitation of DCN with IGF-IR in Adult-1 and Aged-1 cells, and found that IGF-IR in Aged-1 cells bound more DCN as compared with that of Adult-1 cells (Supplementary Figure 5).

Figure 5. Effect of IGF-I on the binding of DCN to IGF-IR. Aged-1 cells were serum-starved for 16 h, and then either treated with 200 ng/ml IGF-I (IGF-I) for 5 min or left untreated (U). Cells were harvested for co-immunoprecipitation assays and Western blot analyses for DCN and IGF-IR.

DNA methylation regulated DCN expression

Epigenetic regulation is an important mechanism underlying the gene expression in aging, and has been shown to regulate DCN expression in non-small cell lung cancer cells [19]. To assess if DNA methylation plays a role in regulating DCN expression in bmMSCs, we treated adult and aged rat bmMSCs with or without 5-aza-2’-deoxycytidine (5-aza-dC, a DNA methylation inhibitor) and dimethyloxalylglycine (DMOG, an inhibitor of the ten-eleven translocation proteins (TETs) that cause DNA demethylation) for 3 and 6 days, and then examined Dcn mRNA expression. The results showed that 5-aza-dC (10 μM) decreased Dcn expression in aged bmMSCs, whereas DMOG (10 μM) increased Dcn expression in adult bmMSCs (Supplementary Figure 6). These data suggested a positive association between DNA methylation and Dcn expression. Subsequently, to examine if these results were also seen with human bmMSCs, first, we analyzed the DCN gene sequence using the online MethPrimer for the clusters of cysteine-guanine dinucleotides (CpGs), and located 4 CpG islands (CGIs) in the second and seventh introns, containing 11, 5, 6, and 12 CpG sites, respectively (Figure 6A). We treated Adult-1 and Aged-1 cells with DMOG (10 and 40 μM) and 5-aza-dC (10 μM) for 6 days, respectively, and measured the methylation status of those 34 CpG sites (Supplementary Table 1). We found that the median of methylation of Adult-1 and Aged-1 cells was 83.1% and 96.1%, respectively, in CGI-I; 92.7% and 92.9%, respectively, in CGI-II; 97.7% and 98.6%, respectively, in CGI-III; and 89.2% and 89.1%, respectively, in CGI-IV (Figure 6B). Notably, the DCN methylation of Aged-1 cells was significantly higher than that of Adult-1 cells only in CGI-I (P<0.05). Also, DMOG dose-dependently increased DNA methylation of Adult-1 cells (P<0.05), whereas 5-aza-dC decreased DNA methylation of Aged-1 cells (P<0.05) only in CGI-I. In parallel, both Adult-1 and Aged-1 cells were treated with 10 μM of 5-aza-dC for 6 days. RT-qPCR analyses showed that DCN mRNA expression in Adult-1 cells was not significantly changed (P=0.299), whereas there was approximately 53% decrease in Aged-1 cells (P<0.05) (Figure 6C). Then, both types of cells were treated with 10 μM and 40 μM of DMOG for 3 and 6 days. RT-qPCR analyses showed that DCN mRNA expression in Adult-1 cells was significantly increased with increasing DMOG dose and treatment time, whereas there was no significant change in Aged-1 cells except that 10 μM for 3 days increased DCN expression slightly (Figure 6D). These data suggested a positive association between DNA methylation and DCN expression in human bmMSCs. To examine if inhibition of DNA methylation promoted cell proliferation, we treated serum-starved Aged-1 cells (1 x 10⁵) with or without 10 μM of 5-aza-dC in the presence of IGF-I for 6 days, and compared their cell number with that of the control. The cell number of the control, IGF-I treated, and IGF-I plus 5-aza-dC-treated groups was 5 x 10⁴, 8.9 x 10⁴, and 1.3 x 10⁵, respectively (Figure 6E). These data suggested that IGF-I plus 5-aza-dC increased cell proliferation (Figure 6E).

Figure 6. Effect of 5-aza-dC and DMOG on the methylation of CpG islands in DCN gene and the expression of DCN mRNA in Adult-1 and Aged-1 bmMSCs. (A) Schematic presentation of human DCN gene. Eight exons (■) and 4 predicted CpG islands (▲) in the introns are shown. CGI-I, -II, -III, and -IV contains 11, 5, 6, and 12 CpG sites, respectively. (B) Bisulfite sequencing. Adult-1 and Aged-1 cells were treated with DMOG (10 and 40 μM) and 5-aza-dC (10 μM) for 6 days, respectively. The methylation percentage of each CpG site of the 4 CGIs was determined. A one-way ANOVA plus Scheffe’s post hoc tests were used to analyze the differences among the treated Adult-1 and Aged-1 groups versus untreated Adult-1 cells. ^*, P<0.05 versus untreated Adult-1 cells. (C) RT-qPCR analyses. Adult-1 and Aged-1 cells were either left untreated or treated with 10 μM 5-aza-dC for 6 days. DCN mRNA levels of the treated cells were compared to those of the untreated. Data represent the mean ± S.D. from three experiments. ^*, P<0.01 versus untreated control by Student’s t-test. (D) RT-qPCR analyses. Adult-1 and Aged-1 cells were treated with 0, 10, and 40 μM DMOG for 3 and 6 days, and the DCN mRNA levels were measured. The normalized DCN signals in these cells were compared to that of the untreated cells (to which a value of 1 was assigned). Data represent the mean ± S.D. from three experiments. ^*, P<0.05; ^**, P<0.005; ^#, P<10^-4; ^##, P<10^-6 versus untreated control by Student’s t-test. (E) Proliferation assay. Aged-1 cells were seeded into 12 10-cm cell culture dishes (1 x 10⁵ cells/dish) and were serum-starved for 16h. Then, 4 dishes of cells were left untreated, 4 dishes of cells were treated with 200 ng/ml IGF-I, and the other 4 dishes were treated with 200 ng/ml IGF-I plus 10 μM of 5-aza-dC. Cells were counted 6 days after. Data represent the mean ± SD. ^*, P<0.05 versus the untreated group. A one-way ANOVA plus Scheffe’s post hoc tests were used to analyze the differences among the groups. Student’s t-test was used to analyze the difference between the IGF-I-treated and the IGF-I plus 5-aza-dC-treated groups.

Cell culture

F344 rat bmMSCs were a kind gift from Dr. Chun-Chin Liang who had retired from the National Health Research Institutes in Taiwan years ago. Human bmMSCs were prepared from bone marrow samples collected from patients receiving total knee replacement surgery as described previously [5]. Informed consent was obtained from each donor. The use of human bmMSCs was approved by the Institutional Review Board of National Taiwan University Hospital, Hsin-Chu Branch (IRB No. 103-012-F). Flow cytometric analyses were performed using a Becton Dickinson FACS Calibu flow cytometer, and cells that were CD31- and CD45-negative, but CD90- and CD105-positive were recognized as bmMSCs. Both of rat and human bmMSCs were cultured in Dulbecco's modified Eagle's medium (DMEM) (low glucose) (Thermo Fisher Scientific, MA, USA) containing fetal bovine serum (15%) (HyClone Laboratories Inc., UT, USA), glutamine, penicillin and streptomycin (Thermo Fisher Scientific, MA, USA), and maintained in a humidified atmosphere containing 5% CO₂ at 37° C. Cell culture media were changed every 4 days. Cells cultured between the fourth and seventh passage were used in this study.

Plasmid construction, Lentivirus preparation, and infection

Human DCN cDNA was amplified by PCR. The 5’ and 3’ primers used were GGAATTCATGAAGGCCACTATCATCCTC and GGAATTCGAATTACTTATAGTTTCCGAGTTG. The cDNA was cloned into pLAS3w.Pneo vector to generate pLAS3w.Pneo-DCN for Lentivirus preparation. pLAS3w.Pneo and pshRNA_DCN (clone ID: TRCN0000058556, a plasmid harboring shRNA targeting human DCN mRNA with the sequence CCGGCCGTTTCAACAGAGAGGCTTACTCGAGTAAGCCTCTCTGTTGAAACGGTTTTTG) were purchased from the National RNAi Core Facility at Academia Sinica, Taiwan. For DCN knockdown or overexpression in human bmMSCs, pshRNA_DCN or pLAS3w.Pneo-DCN was co-transfected with gag-pol and VSV-G-expressing plasmids into 293T cells. Viral supernatant was harvested 2 and 3 days after transfection and filtered through 0.45-μm filters. BmMSCs were infected with virus (MOI = 40) for 3 h in the presence of polybrene (8 μg/ml).

Measurement of DNA synthesis

DNA synthesis was assessed by measuring the incorporation of 5-bromo-2-deoxyuridine (BrdU) into DNA [13]. Briefly, cells were seeded in 96-well culture plates (1200 cells/well), and maintained in medium containing varying doses of IGF-I for 24h at 37° C. BrdU or phosphate buffer saline (as background control) was added to each well during the final 8h of treatment. After fixation, cells were incubated with anti-BrdU antibody for 1h at room temperature, and with substrate solution for 15 min in the dark at room temperature. The signals were quantitated using a spectrophotometric plate reader set at wavelength of 450/540 nm. To examine if high dose IGF-I also activated the other signaling pathways to stimulate DNA synthesis, cells were treated with IGF-I and AG1024 (Sigma Aldrich, MO, USA) concomitantly.

Western blot analyses

For Western blot analysis, aliquots (40 μg) of whole-cell lysates were separated on 10% SDS-polyacrylamide gels, and electrotransferred onto polyvinylidene membranes. The membranes were incubated with anti-IGF-IR (Cell Signaling Technology, MA, USA), anti-phosphorylated IGF-IR (Cell Signaling Technology, MA, USA), anti-DCN (GTX101250, Genetex, Taiwan), and anti-β-actin (BD Biosciences, CA, USA) antibodies, and the signals were obtained by enhanced chemilluminescence (PIERCE, IL, USA).

Quantitative real-time PCR (RT-qPCR) analyses

RT-qPCR was performed as described previously [15]. The 5’ and 3’ primers used were as follows: human IGF-I, TGCTTCCGGAGCTGTGATCT and TCTGGGTCTTGGGCATGTC; human IGF-IR, TCGACATCCGCAACGACTATC and CCAGGGCGTAGTTGTAGAAGAG; human DCN, ACTGGGAGATACAGCCATCCA and GTTATAAAAATGAGGGCTTTCTTGAGA; human β-ACTIN, AAGTCCCTTGCCATCCTAAAA and ATGCTATCACCTCCCCTGTG. All RT-qPCRs were performed in triplicate on an ABI PRISM 7000 Sequence Detector System. The relative mRNA levels were calculated using the 2^−ΔΔCT method, with β-ACTIN mRNA as a normalizer.

Co-immunoprecipitation assay

To examine the association between IGF-IR and DCN, cells (1.6 x 10⁶) were lysed in RIPA buffer at 4° C for 30 min with gentle shaking. Equal aliquot of cell lysate (200 μg) was then mixed with protein A/G-agarose beads that were pre-coated with anti-IGF-IR antibody or IgG, at 4° C for 16 h with gentle shaking. After low-speed centrifugation, the supernatants were discarded, and the agarose beads were washed 4 times with RIPA buffer, and then subjected to Western blot analysis for IGF-IR and DCN.

DNA methylation analysis

Genomic DNA was isolated from bmMSCs treated with DMOG, or 5-aza-dC, or left untreated, and was sent for bisulfite conversion and methylation analyses (Mission Biotech, Taipei, Taiwan). Briefly, bisulfite treatment of DNA was performed using EpiTect 96 Bisulfite Kit (Qiagen, MD, USA). PCR amplification of the 4 CpG islands in DCN gene was performed using PyroMark PCR Kit (Qiagen, MD, USA). The 5’ and 3’ primers for the amplification of CGI-I~IV were as follows: CGI-I, TTGTTATTTAGGTTGGAGTGTAGTG and Biotin-TTCAATTCTAATTCCTCTTCTTTATCT; CGI-II, GAGGTTGGTGGATTATGAGGTTAG and Biotin-AATTTAAATCCTCACTCCAAAACTATAT; CGI-III, ATTTAGGTTGGATTGTATATAATGGTATAA and Biotin-ACTACCTCCTACCACCAACAAAATCTTAA; and CGI-IV, Biotin-TGAGGATATGAGTTTTGTAGGTTAAGAA and AACTAACATAATAAAACCCCCTCTCTA. The biotin-labeled PCR products were captured by Streptavidin-Sepharose HP (GE Healthcare Bio-Sciences Corp. NJ, USA), and were made single-stranded using a Pyrosequencing Vacuum Prep Tool. The sequencing primers were annealed to the single-stranded PCR products, and pyrosequencing was done using the PyroMark Q24 system (Qiagen, MD, USA). The sequencing primers used were as follows: CGI-I, GTTGGAGTGTAGTGG, AGTTGGGATTATAGGTATT, and AGGATGGTTTAGATTTTTTGAT; CGI-II, AGTTAGGTATTGGTGGTA, and AGATGATGTTATTGTATTTTAGG; CGI-III, TGGTATAATTTAGGTTTATTGTAAT, and AGTTTTTAGAGTAGTTGGGA; and CGI-IV, CCCTCTCTACTAAAAATACA, CCCAACTACTCCAAAAACTAAAACA, and ACACCCCACTTCAACC. Quantitation of cytosine methylation was done using the PyroMark Q24 Software.

Statistical analysis

One-way ANOVA followed by Scheffe’s post hoc tests was used to examine the significance in multiple comparison. P value less than 0.05 was considered statistically significant. For the examination of the impact of DCN overexpression and DCN knockdown on the IGF-IR auto-phosphorylation in response to IGF-I, a linear regression with three covariates, including IGF-I concentrations, cells and their interactions, was used to compare the response rates between the DCN-overexpressing and control adult bmMSCs as well as the response rates between the DCN-knockdown and control aged bmMSCs. P value less than 0.05 was considered statistically significant.