The DLC-1 tumor suppressor is involved in regulating immunomodulation of human mesenchymal stromal /stem cells through interacting with the Notch1 protein

Cells: hUC-MSCs was gifted anonymously from TuoHua Biotech company (Siping, China), where the cells were isolated and purified from Wharton’s Jelly of a discarded umbilical cord. The expression of the featured surface markers of hMSCs, the differentiation potentials to osteocytes, chondrocytes and adipocytes, and microbiological safety were tested for hUC-MSCs in our laboratory. Peripheral blood mononuclear cells (PBMCs) were freshly isolated using a conventional Ficoll method [26] from whole blood of healthy donors provided anonymously from local Red Cross. All data analysis associated with the use of hUC-MSCs and PBMCs in this study was conducted anonymously. Antibodies: the antibodies against IDO1, NICD1, DDB-1 and phosphor-STAT1 at Y701 (pSTAT1) were from Cell Signaling (Danvers, MA); the antibodies against DLC-1, from BD Bioscience (Franklin Lakes, NJ); the antibodies against ubiquitin, Hes1, Hey1, Rock1 and Rock2, from Santa Cruz Biotechnology (Dallas, TX); the antibodies against Cullin 4A and FBXW5, from Abcam (Cambridge, MA); the antibody against β-actin, from Sigma (Milwaukee, WI); Horseradish peroxidase (HRP)-conjugated anti-mouse or anti-rabbit secondary antibodies, from GE Healthcare (Piscataway, NJ). All antibodies conjugated with different fluorescent dyes were from BD Bioscience. Constructs: pIDO1-Luc, pcDNA3.1-DLC-1, −DLC-1-Δ622, −DLC-1-662, −DLC-1-R718E containing a R718E point mutation in RhoGAP domain, and -DLC-1-RhoGAP, which is the RhoGAP domain only mutant, were constructed in previous studies [8, 23]. Chemicals: GSI-I (SCP0004) and DAPT (D5942), γ-secretase inhibitors, were purchased from Sigma Aldrich (St. Louis, MI); Bortezomib, a proteasome inhibitor, was from ChemieTek (Indianapolis, IN); Y27632, a pan-Rock inhibitor, from EMD Millipore (Darmstadt, Germany); Phorbol-12-myristate-13-acetate (PMA) was from R&D (Minneapolis, MN), Ionomycin was from Santa Cruz Biotechnology (Dallas, TX), Brefeldin A (BFA) was from Abcam (Cambridge, MA).

The hUC-MSCs surface markers, i.e. CD105, CD90 and CD73, were detected by flow cytometry using BD Stemflow hMSC Analysis Kit following the procedures described previously [8, 27].

The osteogenic differentiation of hUC-MSCs was examined by detecting the expression of Response Gene to Complement 32 protein (RGC32), an effective biomarker of osteogenesis revealed and validated in previous studies, via a real-time Polymerase chain reaction (PCR) following previously described procedures for the induction of osteogenic differentiation described previously [8].

Conventional semi-quantitative RT-PCR was employed to detect mRNA expression of Hes-1 and Hey-1 in hUC-MSCs following various treatments. The expression of Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as an internal control for the semi-quantitation. The sets of primer sequences were: AGCACAGACCCAAGTGTGCTG and GAAGGTGACACTGCGTTGGG, for HES-1; ACGAGAATGGAAACTTGAGTTCGGC and CCCAAACTCCGATAGTCCATAGCAAG, for HEY-1; ACCACAGTCCATGCCATCAC and TCCACCACCCTGTTGCTGT, for GAPDH.

Real-time PCR with SYBR Green quantification were set up using 1/20 of each complementary DNA (cDNA) preparation in Applied Biosystems® 7500 Real-Time PCR Systems (Thermo Fisher scientific, Waltham, MA). Quantitative analysis was conducted by normalizing the expression level of the testing gene to that of GAPDH. The primer sequences were: GAAGTTCTGGGTCCTTTCATC and GCATGGATCGTCTGTTCTAATA, for RGC32; GGGCTGTGGAGTTTGGTGTC and CTGCTTGGGTGGGTGGAG, for Runx2; GATGGATTCCAGTTCGAGTATG and AGTGACGCTGTAGGTGAA, for Collagen I; TGCCTTTCCTGTAACGTTGGA and CCACAATGTTCTCTTCCCAAG, for Osterix; GGTCACTGATTTTCCCACGGA and TGGATGTCAGGTCTGCGAAAC, for Osteopontin; CTGACCACATCGGCTTTC and CAGATTCCTCTTCTGGAGTTTAT, for BGLAP; AGGCTGGAGAGGCGGCTAAG and TGGAAGGTGACACTGCGTTGG, for HES1; GGATCACCTGAAAATGCTGCATAC and CCGAAATCCCAAACTCCGATAG, for HEY1; GCCGGACACCATGATCCTAAC and GAGCCTCAATGGCATCTCTGT, for DLC-1; GTGTGAACCATGAGAAGTATGA and TAGAGGCAGGGATGATGTT, for GAPDH.

The effect of hUC-MSCs on inhibiting proliferation of Th1 lymphocytes from the PIB (acronym for PMA, Ionomycin and BFA)-induced PBMCs was examinedfollowing the previously reported procedures [8]. Briefly, fresh PBMCs were co-cultured with hUC-MSCs in 5:1 ratio for 18 h in RPMI 1640 complete medium containing 10% FBS and then stimulated with PIB (25 ng/ml PMA, 1 μg/ml Ionomycin and 10 μg/ml BFA) for another 5 h. Then, the PBMCs of all testing groups were collected and stained with both PerCP-Cy5.5-conjugated CD3 and FITC-conjugated CD8 for 30 min at room temperature. The PMBCs were then fixed with 4% paraformaldehyde fix solution for 15 min at room temperature. After that, the PBMCs were incubated with permeabilization medium containing PE-conjugated IFN-γ in dark for 30 min at room temperature. Finally, the PBMCs were washed twice with PBS, and analyzed using BD FACS Calibur flow cytometer. Using flow cytometry, the CD3 positive lymphocytes were gated first, the CD8⁻IFN-γ⁺ subpopulation was then identified as Th1 lymphocytes, the CD3⁺CD8⁻IFN-γ⁺ cells.

The procedures of conventional Western blotting were followed to monitor changes in expression of relevant proteins in hUC-MSCs following various treatments [8].

The Cell lysates extracted using RIPA buffer from 1.5 × 10⁶ cells treated with GSI-I or Bortezomib were incubated with 1 μg antibody of the targeted protein for IP at 4 °C overnight, then incubated with protein A/G agarose at 4 °C for 1 h. After washing three times at room temperature, the agarose-bound cell lysates were then analyzed by Western blotting using ubiquitin antibody.

The procedures reported previously for either cDNA or siRNA transfection were followed for either over-expressing, or silencing the expression of the genes to be tested. Each plasmid DNA transfection was conducted using lipofectamine 2000-mediated transfection. To achieve equal transfection efficiency from different plasmid DNAs and to avoid biased results among different transfections in each individual experiment, all plasmid DNAs were extracted by the same laboratory personnel in the same time using endotoxin-free extraction kit following the same extraction and quality testing protocols. All plasmid DNAs used in each transfection experiment were carefully calculated to ensure all of them being in equal molar concentration. In addition, the transfection efficiency of each individual experiment was evaluated by Western blotting using β-actin as internal control to ensure equal amount of cells among different transfection groups were utilized in each individual experiment. The siRNA transfection was employed to silence the expression of Notch1, DLC-1, Hes-1 or Hey-1 in hUC-MSCs and the silencing effect for each target was determined by Western blotting or RT-PCR [8]. The siRNA sequences were CACCAGUUUGAAUGGUCAAtt for Notch1; AGAACAGCACCUCUGGGAUtt for DLC-1, CGAGGUGACCCGCUUCCUGtt (1#) [28] and AGACGAAGAGCAAGAAUAAtt (2#) [29] for Hes1, and GUGCGGACGAGAAUGGAAAtt (1#) and GACCGGAUCAAUAACAGUUtt (2#) for Hey1 [30]. The siRNA for Rock1 (sc-29,473) and Rock2 (sc-29,474) were purchased from Santa Cruz Biotechnology (Dallas, TX). The randomly scrambled siRNA was used as negative control.

The pcDNA3.1-Notch1 containing full-length Notch1 cDNA was kindly provided by Dr. Jon C. Aster [31]. The cDNA of NICD1 was amplified by RT-PCR from total RNA of hUC-MSCs and constructed into the pcDNA3.1 expression vector. The primer sequences used for amplifying NICD were: 5′-GCTCTAGAGTGCTGCTGTCCCGCAAGCG-3′ and 5′-CCCAAGCTTTTCAACTTCCCTTCTCCAACATCATTTC-3′, in which Xbal and HindIII restriction sites were added for subcloning.

The luciferase assay was employed for detecting IDO1 promoter activity. The pIDO1-Luc vector used for the assay was constructed and characterized in a previous study [8]. The procedures reported previously were followed for detecting IDO1 promoter activity in response to IFN-γ in hUC-MSCs [8].

Given the dual inhibitory activities of GSI-I, we speculated from the previous study that the effect of GSI-I on the immunomodulation of hUC-MSCs was likely the consequence of interaction between Notch1 and other protein(s) that were subjected to proteasome-mediated protein degradation [8]. To identify the candidate proteins interacting with Notch1, we examined expression of different proteins likely undergoing proteasome-mediated degradation in hUC-MSCs. Among the candidate proteins were Mcl-1, DLC-1 and other proteins, which were subjective to the proteasome-mediated protein degradation [24, 32]. Through Western blotting and immunoprecipitation, we found that the GSI-I treatment significantly elevated DLC-1 protein level with an increase also seen in polyubiquitinated form of protein (Fig. 1a & b), suggesting that DLC-1 was subjected to the proteasome-mediated degradation in hUC-MSCs.

Given that the Notch1 protein is involved in regulating the expression of the surface markers of hUC-MSCs, like CD73, CD90 and CD105, and osteogenic differentiation, which was measured in part by the transcription of RGC32, a surrogate marker of the osteogenesis of hUC-MSCs, revealed and validated in a previous study [8], to determine a possible involvement of DLC-1 in regulating surface markers and osteogenic differentiation of hUC-MSCs, we transfected hUC-MSCs with siDLC-1 in the presence of 2.5–5 μM GSI-I, then tested the expression of both the surface markers and RGC32. After confirming that the silencing effect from siDLC-1 transfection (Fig. 2a), we observed that, whereas the GSI-I treatment reduced the expression of CD73, CD90 and CD105 with the most significant reduction seen in CD105, the siDLC-1 transfection moderately reversed the GSI-I-induced reduction of all three surface markers (Fig. 2b). After compare the Alizarin red staining and osteogenic related gene marker determination using Realtime-PCR methods, we confirmed that RGC32 gene expression exhibited accumulatively increased along the entire osteogenic differentiation process. Therefore RGC32 is suitable for rapid determination during the early osteogenic differentiation step (Fig. 2c & d). The siDLC-1 transfection showed no effect on the inhibitory activity of GSI-I by RGC32 expression determination (Fig. 2e). Thus suggesting that the GSI-I-induced DLC-1 elevation contributed to the GSI-I-induced reduction in surface markers, but not in osteogenic differentiation.

Given that the immunomodulation of hMSCs can be represented in part by its inhibition of proliferation of Th1 lymphocyte subpopulation and IFN-γ-induced IDO1 expression [8], we transfected siDLC-1 or DLC-1 cDNA into hUC-MSCs and then examined the effect of the transfection on both Th1 proliferation and IDO1 expression. After confirming the effect of each transfection on DLC-1 expression (Fig. 3a & b), it was observed via flow cytometry assay that, whereas the siDLC-1 transfection further enhanced the reduction of Th1 proliferation and significantly reversed the GSI-I-induced inhibition of Th1 lymphocyte proliferation as well, the DLC-1 cDNA transfection significantly reduced the inhibition of Th1 proliferation (Fig. 3a & b). Meanwhile, through the IDO1 promoter assay, in which hUC-MSCs were co-transfected with pIDO1-Luc and siDLC-1, or pIDO1-Luc and DLC-1 cDNA for 24 h, as followed by the treatment with 10 ng/ml IFN-γ for another 24 h before measuring the luciferase activity from each transfection. It was observed that, comparing with each negative control, DLC-1 overexpression significantly reduced the IFN-γ-induced IDO1 promoter activity, whereas the DLC-1 silencing increased the promoter activity (Fig. 3c & d), all thus suggesting for the first time that DLC-1 played an inhibitory role in the immunomodulation of hUC-MSCs.

To determine which functional domain(s) were responsible for DLC-1’s activity of inhibiting the immunomodulation, we then tested the effect of different DLC-1 mutants, i.e. the mutant with the deletion of N-terminus (DLC-1-Δ662) or C-terminus (DLC-1-622), the mutant with RhoGAP domain point mutation (DLC-1-R718E), or a RhoGAP domain only mutant, in comparison with wild type DLC-1, on the IFN-γ-induced IDO1 promoter activity in hUC-MSCs. The schemes of all DLC-1 mutants were shown in Fig. 4a. It was observed in the co-transfection of DLC-1 or its mutant cDNAs with pIDO1-Luc that wild type DLC-1 (wtDLC-1) inhibited over 50% of, the DLC-1-622 mutant showed no effect, and all other mutant showed even an increase in, the IDO1 promoter activity (Fig. 4b). In addition, the wtDLC-1 and the DLC-1-622 and DLC-1-Δ662 mutants reduced, but the DLC-1-R718E mutant increased, the IDO1 protein expression (Fig. 4c), all thus suggesting that the inhibitory effect of DLC-1 on IDO1 might be both RhoGAP domain-dependent and RhoGAP domain-independent.

After associating the DLC-1 protein with the immunomodulation, we were then engaged to determine the association between DLC-1 and Notch1 protein regarding the regulation of immunomodulation of hUC-MSCs. To conduct this study, we first constructed the NICD1 expression vector according to the literature [31] and validated NICD1 expression in hUC-MSCs after transfecting the vector into the cells. Next, we tried to determine the association between DLC-1 and Notch1 in regulating IDO1 promoter activity after co-transfecting DLC-1 cDNA with either siNotch1 or NICD1 cDNA. It was found that, whereas the transfection with either siNotch1 or DLC-1 cDNA alone caused a similar reduction of the promoter activity, and siNotch1 plus DLC-1 cDNA further reduced the activity (Fig. 5a). Meanwhile, it was found that, whereas the NICD1 transfection alone caused a significant increase in the promoter activity, the effect was partially reversed by DLC-1 cDNA transfection (Fig. 5b), thus suggesting the existence of an association between DLC-1 and Notch1 at least in part in the regulation of IDO1.

To further reveal the relationship between DLC-1 and Notch1, we next examined the changes of Notch1 cleavage/activation in hUC-MSCs following the transfection with siDLC-1 [8]. Interestingly, it was found that the siDLC-1 transfection caused a significant increase in both basal NICD1 and the GSI-I-reduced NICD1 (Fig. 6a). Meanwhile, the transfection with DLC-1 cDNA alone resulted in a significant reduction of NICD1 (Fig. 6b). Furthermore, it was observed in the cDNA transfection experiments that, while the DLC-1 cDNA almost completely abolished the effect of inducing IDO1 by the NICD1 cDNA, the NICD1 cDNA blocked the effect of inhibiting IDO1 expression by the DLC-1 cDNA (Fig. 6c), all thus together strongly supporting the existence of a mutual exclusion relationship in protein expression between DLC-1 and Notch1 and on the regulation of IDO1.

Regarding the likely effect of DLC-1 on NICD1 nuclear translocation, we also revealed that, whereas the siDLC-1 transfection clearly increased -NICD1 protein level in both the cytoplasm and the nucleus, the DLC-1 cDNA transfection caused a significant reduction of NICD1 protein in both compartments with a more reduction seen in the nucleus than the cytoplasm (Fig. 6d), thus indeed suggesting that DLC-1 could also reduce NICD1 protein nuclear translocation.

Both Hey1 and Hes1 have been well characterized as two prominent downstream effectors of the Notch signaling, but may act differently in mediating different functional aspects of the Notch signaling [16, 33]. Therefore, it was of great interest to determine a possible distinction between these two effectors in mediating the Notch1-regulated immunomodulation of hUC-MSCs. By employing gene silencing with each specific siRNA followed by validation via RT-PCR, (Fig. 7a), we found that the Hes1 silencing caused a slight increase, but the Hey1 silencing resulted in a dramatic decrease, in the IFN-γ-induced IDO1 promoter activity (Fig. 7b). Meanwhile, it was also observed that it was the Hey1 silencing, but not Hes1 silencing, that was able to significantly reduce the inhibition of Th1 proliferation by hUC-MSCs (Fig. 7c), thus suggesting that it was the Notch1-Hey1 axis, but not the Notch1-Hes1 axis, that was involved in the immunomodulation of hMSCs.

Since it was reported that the Notch inhibitor DAPT exerted its inhibition more specifically on Hes1 than Hey1 [33], we then utilized DAPT as a tool in comparison with GSI-I to distinguish between Hes1 and Hey1 in the involvement in the immunomodulation of hMSCs. It was observed through RT-PCR that, while the treatment with 10 μM GSI-I equally reduced both Hes1 and Hey1 expression, the treatment with 10 μM DAPT only reduced Hes1 expression, thus confirming that DAPT was a Hes1-specific inhibitor (Fig. 7d). It was next observed that, whereas the GSI-I treatment reduced both IDO1 protein expression in hUC-MSCs and the inhibition of Th1 lymphocyte proliferation by hUC-MSCs, the DAPT treatment unexpectedly increased the IFN-γ-induced IDO1 protein expression and showed no effect on Th1 lymphocyte proliferation (Fig. 7e & f), thus supporting that it was the Notch1-Hey1 axis, but not Notch1-Hes1 axis, that promoted the immunomodulation of hUC-MSCs.

After charactering the differences between Hey1 and Hes1, we then attempted to determine whether the Hey1 protein was involved in the mutual exclusion relationship between DLC-1 and Notch1. We tested the expression of Hey1 and Hes1 genes using RT-PCR after transfection with either siDLC-1 or DLC-1 cDNA. It was found that the transfection with DLC-1 cDNA caused a slight reduction of Hes1, but a significant reduction of Hey1. In contrast, the siDLC-1 transfection resulted in a remarkable reduction of Hes1, but a significant elevation of Hey1 (Figure s1A, B & C). On the other hand, it was observed via Western blotting that, whereas the siHes1 transfection induced an apparent increase in IFN-γ-induced IDO1 expression but a clear decrease in DLC-1 expression, the siHey1 transfection induced a remarkable decrease in IDO1 expression, but a clear increase in DLC-1 expression (Figure s1D). Therefore, all these findings together clearly demonstrated that it was Hey1, but not Hes1, that served as the key signaling molecule involved in the mutual exclusion relationship between DLC-1 and Notch1.

Given that the DLC-1 protein could serve as a degradation target of the CUL4A-DDB1-FBXW5 E3 complex in tumor cells [24], it was of great interest to also determine the involvement of these E3-ligase proteins in regulating DLC-1 protein stability and in the relationship between DLC-1 and Notch1 in hUC-MSCs. For this purpose, we silenced the expression of each E3-ligase protein in hUC-MSCs, then determined the effect of each silencing on DLC-1 protein expression and on the immunomodulation of hUC-MSCs. Interestingly, after confirming the silencing effect of each E3 ligase by Western blotting (Figure s2A), we observed that the silencing of either DDB1 or FBXW5, but not CUL4A, significantly elevated DLC-1 protein level with a more significance seen in FBXW5 silencing. In addition, the silencing of either FBXW5 or DDB1 was accompanied by a significant reduction in IDO1 and p-STAT1 with a more significance seen again in FBXW5 silencing (Figure s2A). More interestingly, it was also observed that, comparing with DDB1 and CUL4A, the silencing of FBXW5 caused a significant reduction in the inhibition of Th1 lymphocyte proliferation (Figure s2B). These findings thus demonstrated that FBXW5 was the major E3 ligase for regulating DLC-1 protein stability and subsequent immunomodulation of hUC-MSCs.

To further pursue the possibility that the E3 ligase(s) for DLC-1 and Notch1-Hey1 signaling could be mutually regulated, we first silenced each E3 ligase and then tested its consequence on the expression of NICD1 and Hey1. We then found that, whereas the FBXW5 silencing led to a significant reduction in both NICD1 and Hey1, the silencing for DDB1 or CUL4A showed almost no such effect (Figure s2A). Next, we examined the effect of silencing Hes1 or Hey1 on the expression of all E3 ligases. It was then observed that, whereas the Hes1 silencing showed no effect on all E3 ligase proteins, the Hey1 silencing however caused a significant reduction only in FBXW5 protein, thus concluding that the Notch1-Hey1 signaling and FBXW5 could be mutually inhibitory for regulating DLC-1 protein stability and subsequent immunomodulation of hMSCs (Figure s2C).

Given that the Rock1/2 proteins serve as the key effectors of the DLC-1 signaling, we next attempted to determine whether they could play a role in regulating the immunomodulation of hUC-MSCs. In the new experiments, we employed Y27632, a Rock1/2 small molecule inhibitor, to treat hUC-MSCs and then tested the effect of the treatment on IFN-γ-induced IDO1 expression and IDO1 promoter activity in hUC-MSCs. Unexpectedly, we found that the treatment with Y27632 resulted in a significant dose-dependent increase in IDO1 protein expression and promoter activity (Figure s3A & B). To understand the seemingly contradictory effect of Y27632 on IDO1, we then tested via Western blotting the protein expression of Rock1 and Rock2, and the expression of DLC-1, NICD1, Hes1 and Hey1. Interestingly, it was found that the Y27632 treatment resulted in a dose-dependent reduction of both Rock1 and Rock2 with a much more significant reduction in Rock2. In addition, it also caused a dose-dependent decrease in DLC-1, but an increase in both NICD1 and Hey1 (Figure s3A). Moreover, consistent to the increase in IDO1, the hUC-MSCs pretreated with Y27632 exhibited a significantly enhanced inhibition of Th1 lymphocyte proliferation (Figure s3C), all thus suggesting that the treatment of Y27632 in fact mimicked the activity of the Notch1 signaling in enhancing the immunomodulation of hMSCs. Considering that Y27632 is an inhibitor of both Rock1 and Rock2, we next transfected either siRock1 or siRock2 before examining the expression of the relevant proteins, thus attempting to distinguish the effect between Rock1 and Rock2 in the immunomodulation of hMSCs. The results showed that the transfection with siRock2, but not siRock1, achieved the same effect as Y27632 on the expression of IDO1 and DLC-1, whereas the siRock1 transfection resulted in a clear reduction of IDO1 and unexpectedly an increase in Rock2 (Figure s3D), suggesting that the effect of Y27632 observed above on the immunomodulation of hMSCs was attributable to the inhibition of Rock2, but not Rock1, and Rock1 and Rock2 appeared exerting differently in the regulation of the immunomodulation with Rock1 seemingly being pro-immunomodulatory and Rock2 anti-immunomodulatory. It could be further suggested that Rock1 likely represented the inhibitory target downstream of the DLC-1 signaling in the inhibition of immunomodulation of hUC-MSCs, while Rock2 might serve as a negative feedback regulator of the DLC-1 signaling in this perspective.