Osteogenic differentiation of human mesenchymal stromal cells and fibroblasts differs depending on tissue origin and replicative senescence

The need for an autologous cell source for bone tissue engineering and medical applications has led researchers to explore multipotent mesenchymal stromal cells (MSC), which show stem cell plasticity, in various human tissues. However, MSC with different tissue origins vary in their biological properties and their capability for osteogenic differentiation. Furthermore, MSC-based therapies require large-scale ex vivo expansion, accompanied by cell type-specific replicative senescence, which affects osteogenic differentiation. To elucidate cell type-specific differences in the osteogenic differentiation potential and replicative senescence, we analysed the impact of BMP and TGF-β signaling in adipose-derived stromal cells (ASC), fibroblasts (FB), and dental pulp stromal cells (DSC). We used inhibitors of BMP and TGF-β signaling, such as SB431542, dorsomorphin and/or a supplemental addition of BMP-2. The expression of high-affinity binding receptors for BMP-2 and calcium deposition with alizarin red S were evaluated to assess osteogenic differentiation potential. Our study demonstrated that TGF-β signaling inhibits osteogenic differentiation of ASC, DSC and FB in the early cell culture passages. Moreover, DSC had the best osteogenic differentiation potential and an activation of BMP signaling with BMP-2 could further enhance this capacity. This phenomenon is likely due to an increased expression of activin receptor-like kinase-3 and -6. However, in DSC with replicative senescence (in cell culture passage 10), osteogenic differentiation sharply decreased, and the simultaneous use of BMP-2 and SB431542 did not result in further improvement of this process. In comparison, ASC retain a similar osteogenic differentiation potential regardless of whether they were in the early (cell culture passage 3) or later (cell culture passage 10) stages. Our study elucidated that ASC, DSC, and FB vary functionally in their osteogenic differentiation, depending on their tissue origin and replicative senescence. Therefore, our study provides important insights for cell-based therapies to optimize prospective bone tissue engineering strategies.

. (A) Phenotype characterization. Phenotype characterization of human DSC, ASC, and FB cultures in P3 (grey bars) and P10 (black bars) was performed by FACS analysis. Bars represent mean ± SD of three donors. *, p < 0.05 as compared to the respective sample of the culture of P10. (B) Comparison of the osteogenic differentiation +/− BMP-2 between P3 and P10. The osteogenic differentiation potentials of DSC, ASC, and FB. DSC and ASC at P3 had the best potential to differentiate osteogenically compared to FB. Only in DSC the osteogenic differentiation could be significantly improved with BMP-2. In a comparison of P3 to P10 cells, DSC showed the greatest decrease, and ASC showed the smallest decrease in osteogenic differentiation potential compared to their younger counterparts at P3. White bars demonstrate the osteogenic differentiation with standard osteogenic differentiation media (OM). Grey bars demonstrate the osteogenic differentiation media supplemented with BMP-2 (OM + BMP2). Bars represent mean ± SD of six donors. * p < 0.05 as compared to the respective sample cultured with OM. #,$ p < 0.05 as compared to the respective sample in P3.

Scientific Reports
| (2021) 11:11968 | https://doi.org/10.1038/s41598-021-91501-y www.nature.com/scientificreports/ Impact of TGF-β signaling and BMP signaling on osteogenic differentiation. At P3 of all analyzed cell types, the osteogenic differentiation potential increased over the time course, and the additional inhibition of TGF-β signaling was significantly superior to the inhibition of BMP signaling (Figs. 5,6,7). Furthermore, if in DSC and FB at P3, TGF-β signaling was inhibited (with SB431542), and BMP signaling was induced with BMP-2, the osteogenic differentiation potential was significantly superior to that of the cells with inhibition of only TGF-β signaling (Figs. 6,7). Interestingly, if BMP-2 signaling was inhibited with dorsomorphin in FB Effects of senescence. In contrast to those at P3, in culture senescent DSC (in P10), the inhibition of TGF-β signaling and the simultaneous use of BMP-2 had an inhibitory effect on osteogenic differentiation potential (Fig. 5), similar to the observed effect in ASC.
In ASC at P10 (as in P3), the inhibition of TGF-β signaling was the best method to promote osteogenic differentiation, and this approach was significantly superior to standard osteogenic differentiation with OM. At P3 and P10, the use of BMP-2 for osteogenic differentiation was always inferior to the respective controls (with or without inhibitors) (Fig. 6).
In summary, TGF-β signaling and the exogenous activation of BMP-2 signaling inhibited osteogenic differentiation in ASC. Evaluated was the ALK-3 expression in DSC at P10. The ALK-3 was expressed at a small amount at P10. (C) Evaluated was the ALK-3 expression in ASC at P3. (D) Evaluated was the ALK-3 expression in ASC at P10. ALK-3 showed a minor expression. (E) Evaluated was the ALK-3 expression in FB at P3. The expression increased slightly on day 7 and 14, but decreased again at day 21. (F) Evaluated was the ALK-3 expression in FB at P10. Grey bars represent these cells incubated with osteogenic differentiation media supplemented with BMP-2. White bars represent cells differentiated with standard osteogenic differentiation media (OM). Bars represent mean ± SD of three donors. www.nature.com/scientificreports/ Interestingly, in FB (P3 and P10), the osteogenic differentiation potential was significantly increased by the simultaneous use of BMP-2 with the BMP signaling inhibitor dorsomorphin compared to treatment with dorsomorphin alone (Fig. 7). Moreover, in FB at P10, the inhibition of BMP signaling with the simultaneous use of BMP-2 was superior to all other applications.
Evaluation of senescence associated β-galactosidase activity. DSC, ASC and FB at P3 demonstrated a comparable level of β-galactosidase activity visualized in faint blue (Fig. 8). Control cells at P3 treated with etoposide displayed comparable more β-galactosidase.
Effects of senescence. At P10 DSC and FB exhibited the strongest β-galactosidase activity visualized in blue, in contrast to ASC, which showed a noticeably weaker staining (Fig. 8). Respective β-galactosidase protein Evaluated was the ALK-6 expression in ASC in P10. ALK-6 was expressed at a small amount. (E) Evaluated was the ALK-6 expression in FB in P3. The expression increased weakly between day 7 and 14. On day 7 was a significant difference between the BMP-2 treatment and the standard osteogenic differentiation in favor to the BMP-2 supplementation. (F) Evaluated was the ALK-6 expression in FB at P10. Grey bars represent these cells incubated with osteogenic differentiation media supplemented with BMP-2 (OM + BMP2). White bars represent cells differentiated with standard osteogenic differentiation media (OM). Bars represent mean ± SD of three donors. *p < 0.05 as compared to the respective sample treated with OM.

Discussion
Forty years ago, a pioneering publication by Friedenstein et al. 33 was released describing cells with multipotent differentiation potential. His study suggested that in the near future, destroyed organs or functionless cells could be replaced by MSC. When it was discovered that MSC were available in nearly every tissue, even in the skin, this optimistic mood reached new heights. Today, it is clear that more detailed information is needed about the specific MSC source with regard to their specific applications in regenerative tissue engineering. Moreover, it is known that MSC are precommitted depending on the tissue origin 34,35 and have varying potential to differentiate osteogenically 36 , but to date, this phenomenon has been insufficiently researched. Compounding the problem is that for an appropriate in vivo application, a high cell number is needed, and therefore, many cell doublings are necessary; this process is inevitably accompanied by replicative senescence and loss of differentiation potential. In this context, TGF-β and BMP signaling has an important role because these signaling cascades regulate bone formation during mammalian development and versatile functions in the body 16 . Against this background, the  The osteogenic differentiation of DSC. In our study, DSC showed the required antigen profile as described by Dominici 37 (Fig. 1A). Furthermore, CD26, an intrinsic membrane glycoprotein, was expressed at lower levels in DSC than in FB and ASC. It has been demonstrated that deletion or inhibition of CD26 38 increases homing transplantation efficiency 31 . In this study DSC followed by ASC had the best potential to differentiate osteogenically (Figs. 2, 3). The differentiation potential was significantly improved with BMP-2 supplementation at P3 and P10 (Figs. 1B, 2).
Trivanocic et al. showed that DSC are superior in their osteogenic differentiation behavior 39 , and further working groups also observed strong effects of BMP-2 in DSC 18,40 . Another study concluded that BMP-2 treatment can induce the expression of Runx2 but only in DSC 41 . Runx2 is a transcription factor necessary for the early differentiation from MSC to osteochondroprogenitors. Moreover, we supposed that BMP-2 could induce osteoblast mineralization in human DSC through a Wnt autocrine loop, as described by Rawadi et al. in various cell lines: C3H10T1/2, C2C12, ST2 and MC3T3-E1 42 . Alternatively, Wnt and BMP signaling may cooperate and synergize to support osteoblast differentiation, as described by Mbalaviele 43 . An indication was our own findings (Suppl 2A) because incubation with BMP-2 led to increased expression of β-catenin, the central factor of Figure 6. Comparison of the ASC osteogenic differentiation potential at P3 versus P10. (A) Evaluation of osteogenic differentiation potential in P3 with alizarin red s. At day 14 the inhibition of the TGF-β signaling with SB431542 (OM + SB) was significantly superior to the inhibition of BMP signaling with dorsomorphin (OM + DM). And the inhibition of the TGF-β signaling significantly was better than the inhibition of TGF-β signaling and the simultaneous induction of BMP signaling with BMP-2 (OM + BMP + SB). (B) Evaluation of osteogenic differentiation potential at P10 with alizarin red s. Inhibiting the TGF-β signaling with SB431542 (OM + SB431542) generated the best osteogenic differentiation potential, at day 21 and was significantly superior to an additionally application of BMP-2 (OM + BMP + SB431542). And the inhibition of BMP signaling with dorsomorphin in the course of the osteogenic differentiation (OM + DM) was significantly superior to the treatment with dorsomorphin supplemented with BMP-2 (OM + BMP + DM). Black bars are ASC differentiated with osteogenic standard differentiation media (OM). Grey bars are ASC differentiated with OM and dorsomorphin (OM + DM). Stacked grey bars are ASC differentiated osteogenically and treated with SB431542 (OM + SB). White bars are ASC coincubated with OM and BMP-2 with the inhibitor dorsomorphin (OM + BMP + DM). Stacked white bars are DSC differentiated with OM, BMP-2 and the inhibitor SB431542 (OM + BMP + SB). Bars represent mean ± SD of three donors. **p < 0.01 as compared to the respective sample. ***p < 0.001 as compared to the respective sample. (C) Visualized comparison of osteogenic differentiation potential of DSC at P3 and P10. Evaluation was performed with alizarin red s on day 21. Shown is one representative illustration of at least six identical results. The image scales represent a length of 200 µm.  45,46 , the associated BMP receptor expression levels have not been elucidated in detail 47 . ALK-3 and ALK-6 are high-affinity binding receptors for BMP-2 and are substantially involved in BMP signaling and osteogenic differentiation in bone 47 . Overall, the elevated osteogenic differentiation capacity in DSC coincided with the increased ALK-3 and ALK-6 expression at P3 (Figs. 3A, 4A).
Consistent with the aforementioned results, a target-oriented inhibitor of BMP signaling, dorsomorphin, significantly reduced the DSC differentiation potential at P3 on day 21 (Fig. 5). BMP-2 signaling interacts with Smad1, Smad5, Smad8 and the common Smad4 to translocate to the nucleus and initiate the expression of osteoblastic genes 48 . Dorsomorphin is a reversible inhibitor that prevents the ligand-associated activation of ALK-2, ALK-3 and ALK-6, which are responsible for the activation of BMP signaling 49 . This signaling can induce the expression of genes indispensable for osteogenic differentiation, such as Runx2 and Osterix 50,51 . Osterix inhibits osteoblast proliferation while inducing osteoblast terminal differentiation 52 . This inhibition is partially mediated through a feedback control mechanism involved in bone formation by decreasing Wnt signaling 53,54 .
Furthermore, the simultaneous use of BMP-2 and the inhibitor of TGF-β signaling, SB431542, significantly improved DSC osteogenic differentiation potential at P3 (Fig. 5). BMP signaling and TGF-β signaling   55 . In TGF-β signaling, Smad2 and Smad3 are phosphorylated, interact with the common Smad4 and translocate to the nucleus together, where they recruit further cofactors to regulate gene transcription. SB431542 selectively inhibits the kinase activity of the TGF-β receptors ALK-4, ALK-5 and ALK-7, and thus, Smad2 and Smad3 could not be activated by TGF-β or activin. Therefore, the use of BMP-2 with the simultaneous inhibition of TGF-β-signaling most likely led to an acceleration of BMP signaling, as described by Maeda et al. 21 .
Osteogenic differentiation in DSC with replicative senescence. However, in a comparison of P3 to P10 cells, DSC showed the greatest decrease, and ASC showed the smallest decrease in osteogenic differentiation potential compared to their younger counterparts at P3 (Fig. 1B). This observation could be underpinned with senescence-associated β-galactosidase staining at P10, because in DSC β-galactosidase activity was stronger as in ASC (Fig. 8). The significantly decreased capacity for differentiation was accompanied by decreased ALK-3 and ALK-6 expression at P10. Additionally, the inhibition of TGF-β signaling did not improve osteogenic differentiation potential in DSC at P10. Similar results were described by Patel et al., who observed that at P10 in DSC, some genes related to osteogenic differentiation were clearly downregulated 56 . Furthermore, this finding could indicate that TGF-β is probably not responsible for replicative senescence in DSC, as was declared by Walenda et al. 29 .
The osteogenic differentiation of ASC. CD29, which induces cell adhesion 30 , showed decreased expression in ASC compared to FB and DSC (Fig. 1A), but antigen expression of the factors described by Dominici was as expected. Furthermore, the ASC osteogenic differentiation potential at P3 was lower than that of DSCs (Fig. 1B). However, it must be considered that in general, abdominal fat is usually provided by individuals undergoing liposuction or abdominal plastic surgery who are in their forties and fifties 57 , whereas DSCs are donated by younger adults in their twenties when their wisdom teeth are removed 13 . It is commonly accepted that the differentiation potential of MSC decreases with donor age 58 . However, the results published by D' Alimonte et al. attested that ASC have a better osteogenic differentiation potential than DSC, but in this study, the donor age difference between DSC and ASC was only ten years (from 18 to 28 years), and therefore, both groups were relatively young 59 .
Compared to that in DSC, BMP-2 supplementation in ASC significantly inhibited their osteogenic differentiation potential at P3 and P10 (Fig. 1B). Alonso et al. 20   Another study showed that BMP-2 application is cell-type specific, as BMP-2 did not accelerate osteogenic differentiation potential in mouse fibroblasts but did in myoblasts (C2C12) and in preosteoblasts (MC3T3-E1) 61 . Consistent with our results, the expression of the BMP-2 receptors ALK-3 and ALK-6 was very low in ASC (Fig. 3C,D and 4C,D). However, interestingly, the application of dorsomorphin to ASCs during osteogenic differentiation had a short-term inhibitory effect on day 14 at P3 (Fig. 6). This finding emphasizes that (endogenous) BMP signaling is important in the early phase of osteogenic differentiation in ASC, but as described by Zuk et al. 62 , it plays a subordinate role. In ASC the inhibition of TGF-β signaling significantly accelerated the osteogenic differentiation potential to similar levels at P3 and P10 (Fig. 6). These findings are supported by the fact that SB431542 induces osteogenic differentiation in C2C12 cells and promotes matrix mineralization 21 . We assume that in ASC, osteogenic differentiation is mediated by phosphorylating TAK1 and TAB1 by inducing the MAP kinase pathway (p38 and MAPK-ERK1/2) rather than by BMP signaling. Our assumption was supported by a protein analysis, which exemplary showed elevated p38 expression in ASC treated with SB431542 (Suppl 2B). MAPK signaling can also promote osteoblastic master transcription factors, such as Runx2 and Osterix 55 .
Osteogenic differentiation in ASC with replicative senescence. In a comparison of P3 and P10 cells, decreased expression of CD44 was demonstrated in all analyzed cell types, but this effect was significant only in ASC (Fig. 1A). CD44 is a glycoprotein involved in cell-cell interactions, cell adhesion and migration 32 .
The ASC osteogenic differentiation potential at P10 barely decreased compared to that at P3 (Fig. 1B) and the osteogenic differentiation pattern did not change between P3 and P10. Therefore, BMP-2 supplementation significantly inhibited osteogenic differentiation, whereas the inhibition of TGF-β signaling significantly accelerated the osteogenic differentiation potential (Fig. 6).
Additionally, a study by Beane et al. 58 similarly determined that ASC from older patients were not as affected by senescence as bone marrow-or muscle-derived stromal cells. Another working group evaluated, that in vitro and in vivo properties in ASC were mostly maintained during aging 63 . Consistent with these findings, we demonstrated that β-galactosidase activity, a known characteristic of senescent cells, in ASC was expressed at a comparable level at P3 and P10 (Fig. 8). Furthermore, in ASC at P10, Osterix protein expression was highly expressed even in untreated cells (Suppl 1B).
The osteogenic differentiation of FB. FB expressed the antigens described by Dominici 37 (Fig. 1A). The osteogenic differentiation potential of FB was comparatively low at P3 and even worse at P10. Delayed Osterix protein expression could jointly be responsible for the decreased differentiation potential of FB at P3 and P10 because Osterix expression increased only on day 14, indicating that FB needed more time for osteogenic differentiation (Suppl 1C).
Although the application of BMP-2 did not improve differentiation (Figs. 1B, 2), the simultaneous use of BMP-2 with the parallel inhibition of TGF-β was significantly superior to the sole inhibition of TGF-β-signaling at P3 and P10 (OM + BMP + SB; Fig. 7).
Furthermore, the following observations were made at P3 and P10: if BMP-2 was used and BMP-2 signaling was simultaneously inhibited with dorsomorphin (OM + BMP + DM), osteogenic differentiation was significantly improved compared to differentiation with dorsomorphin alone (OM + DM). Based on these results, we concluded that BMP signaling plays a substantial role in human FB, especially considering that the expression of the BMP receptors ALK-3 and ALK-6 was increased during osteogenic differentiation at P3 and that ALK-6 expression was significantly elevated due to the additional BMP-2 treatment on day 7 (Figs. 3E, 4E).
However, perhaps the applied BMP-2 concentration in FB at P3 was too high and initiated a feedback loop because the effect of BMP-2 is dependent on its concentration 18 , the same concentrations were used in all analyzed cell types in our study for better comparability. This assumption was underlined by the fact that SB431542, a TGF-β signaling inhibitor, reduces the nuclear accumulation of Smads 64 . This finding could explain why the simultaneous usage of BMP-2 and SB431542 could accelerate the osteogenic differentiation potential compared to BMP-2 application only.
Moreover, BMP-2 and TGF-β signaling have time-dependent preferences. BMP-2 signaling via Smad1, Smad5 and Smad8 supports early osteogenic differentiation in particular and late osteogenic differentiation. TGF-β signaling induces osteogenic differentiation in the early stages but inhibits osteogenic differentiation in the later phases 55 .
As expected and consistent with the aforementioned results in ASC and DSC the inhibition of TGF-β signaling (OM + SB) was significantly more beneficial than the inhibition of BMP-2-signaling at P3 (Fig. 7).
Osteogenic differentiation in FB with replicative senescence. Osteogenic differentiation declined significantly between P3 and P10. In FB at P10, ALK-3 and ALK-6 expression was significantly reduced, as was the capacity for osteogenic differentiation (Figs. 3B, 4B). In line with this in FB at P10 β-galactosidase protein expression as a characteristic for senescence was the highest in contrast to ASC and DSC (Fig. 8). Nevertheless, the simultaneous use of BMP-2 and the inhibition of BMP signaling tended to be the best choice at P10 (Fig. 7).
Our study has several limitations. For our study, we preferred to use primary cells, although these cells show interpersonal donor variabilities 65 . In our opinion, this approach was closer to a prospective clinical application. Although donor age affects differentiation potential 66 , we were not allowed to document the patient's age because of the limitations of our ethical approval. However, even MSC from young donors demonstrated differences in their differentiation potential and clinical usefulness 67 , and to extrapolate this fact, MSC from the same donor with the same tissue origin obtained over a six-month period exhibited differences 67  www.nature.com/scientificreports/ In our work, the cell number of individual donors was extended as much as possible to monitor donor-specific differences. Therefore, we found in individual cases that a good osteogenic differentiation performance correlated, e. g., with high p38 expression in ASC treated with SB. However, this phenomenon was accompanied by a lack of significance in the Western blot analysis. The lack of significance was also because osteogenic differentiation is a gradual process, and most involved proteins have peak time of expression.
Furthermore, for better comparability, we used the same BMP-2, dorsomorphin and SB concentrations in ASC, DSC and FB according to a cell viability assay (Suppl 3A-F). Therefore, it might be possible that the applied concentrations were not the optimal concentrations for every cell type.
TGF-β signaling plays a decisive role in osteogenic differentiation, but beyond that, it is presumed that TGF-β activates cellular senescence by inducing the p16 and p21 pathways 68 . Additionally, Kawamura et al. 28 reported that TGF-β2 is one of the candidate genes for aging in double-positive mesenchymal stromal cells (DPMSC), particularly because old DPMSCs contain significantly more TGF-β2. And the observed aging phenomena were reversed in old DPMSC using a TGF-β antibody (1D11). Interestingly, our observations regarding this matter indicate that TGF-β signaling is not responsible for replicative senescence, although we did not use SB431542 over the course of passaging from P0 to P10. However, for a proper analysis concerning TGF-β-signaling and SB431542 in the context of senescence (which was not our objective), altered pre-mRNA processing, ROS content, disturbed proteostasis, increased mitobiogenesis, etc. should be evaluated.
In contrast, our study has many strengths, it could be demonstrated that DSC have the best osteogenic differentiation potential at P3. DSC differentiate osteogenically using SMAD-dependent BMP-2 signaling, which coincides with elevated ALK-2 and ALK-3 expression. Consistent with this finding, the differentiation potential can be further accelerated with BMP-2. However, as a source for cell-based therapies DSC are likely improbable because potential patients often have their wisdom teeth removed, and an additional growth factor treatment to boost osteogenic differentiation of DSC is critically assessed. Furthermore, an appropriate cell number is required to treat large bone defects, and in DSC, this is accompanied by replicative senescence and impaired differentiation potential. A similar effect was observed by Mehrazarin et al. with stromal cells derived from different dental tissues 69 . Therefore, DSC are suggested for the treatment of immune disorders such as graft versus host disease because of their potential homing capabilities or for the treatment of minor defects in endodontics 70,71 .
ASC use in particular the MAP kinase pathway to differentiate osteogenically. From our presented data, it appears that ASC are more genetically stable, have a greater senescence ratio, and retain their differentiation potential for a longer period in cell culture. BMP-2 is not needed to accelerate ASC osteogenic differentiation potential. Thus, ASCs are favorable candidates for bone tissue engineering strategies. ASC therapeutic usefulness has already been demonstrated in wide-ranging clinical applications, such as wound healing 72 , type 1 diabetes mellitus 73 , osteoarthritis 74 , autoinflammatory diseases 75 , age-related macular degeneration and Stargardt's macular dystrophy 76 .
SMAD-dependent BMP signaling plays a functional role in FB osteogenesis, but the osteogenic differentiation potential is comparably lower and declines with replicative senescence. Either the optimal differentiation protocol has not yet been evaluated or possibly the gradual process of FB osteogenic differentiation takes much longer, as suggested by delayed Osterix expression. Nevertheless, initial and promising approaches were made to evaluate FB therapeutic potential. Therefore, gingival fibroblasts have been used for the treatment of periodontal intrabony defects 77 , and a multicenter study confirmed the usefulness of FB in treating chronic foot ulcers 78 .
These insights could provide important information for the target-oriented use of ASC, DSC and FB in bone tissue engineering strategies. In conclusion, it could be demonstrated that every cell type prefers different signaling cascades to become osteoblastic cells. Furthermore, the analyzed DSC, ASC and FB vary in their potential to differentiate osteogenically depending on their tissue origin and replicative senescence. This is also an important finding; to date, researchers have used and compared various cell types at different cell culture passages in their in vitro and in vivo experiments, and according to our results, this is not a proper approach.

Materials and methods
Donors. Due to ethical approval, it was not possible to note donor age, sex, race, body mass index or donor diseases. According to the experiences of other working groups, we assume that the DSC donors are mostly approximately twenty years old 13,79 . FB and ASC isolated from abdominal plastic surgery are usually donated by middle-aged women in their forties and fifties 57,80 . The DSC tissue was derived from impacted molars, whereas ASC and FB were mostly isolated from abdominal plastic surgery, and a rarer event was isolation from thighs. The study design was as follows: DSCs, ASCs and FBs were isolated from three donors, and the starting cell material was spread to the maximum and used for the osteogenic differentiation analysis at P3 and P10. These DSC, ASC and FB donors were used for experiments considering osteogenic differentiation ± BMP-2, treatments with respective inhibitors and Western blot analysis. For phenotypic characterization, further donors were evaluated, and for cell viability and senescence assays, another donor was evaluated. The cells were not pooled to prevent cells from one donor from overgrowing another in the course of passaging. Isolation and culture of ASC. ASC were isolated from freshly excised human subcutaneous abdominal adipose tissue 81 . Adipose tissue was cut into small pieces (5 mm 2 ) and digested with collagenase solution type I (type: CLS 255 U/mg) (0.2%) at 37 °C for 45 min with constant shaking. The ratio for tissue to enzyme was 1:1.

Materials.
After filtration (100 µm), the fat layer was removed, and the cell suspension was centrifuged at 300×g for 7 min. After resuspension, the cells were seeded in cell culture flasks and cultured in standard cell culture medium.
Isolation and culture of DSC. After the impacted molars were broken, the pulp obtained was digested with collagenase solution type I (0.2%) at 37 °C for 45 min with constant shaking 82 . After digestion, the cell suspension was collected, diluted with phosphate buffered saline (PBS) and centrifuged at 300×g for 10 min. The pellet was suspended, and the cells were seeded in cell culture flasks .
Osteogenic differentiation with or without BMP-2. Osteogenic differentiation was induced with osteogenic differentiation medium (OM) based on standard cultivation medium containing 50 µM α-ascorbate-2-phosphate, 10 mM β-glycerophosphate, and 0.1 µM dexamethasone 83 . Medium was replaced twice a week. Alternatively, OM was supplemented with 450 ng/ml BMP-2 (PeproTech, Hamburg, Germany) 17 . After 0, 7, 14 and 21 days, the osteogenic differentiation capacity was determined using the alizarin red S assay as described below. For each time point, a sextuple determination was performed, and the value at day 0 was subtracted.

Inhibition of TGF-β and BMP-2 signaling with dorsomorphin and SB431542. For inhibition of
TGF-β and BMP-2 signaling, dorsomorphin and SB431542 were used. The optimal inhibitor concentrations were evaluated using osteogenic differentiation media (OM), and the appropriate inhibitor was added in ascending concentrations. Both dorsomorphin and SB431542 were used at a concentration of 0.5 µM because these concentrations did not significantly affect the cell viability (measured with CellTiter-Blue) of FB, DSC, and ASC over the observation period of 14 days.
Osteogenic differentiation with BMP-2, dorsomorphin, and SB431542. DSC, ASC and FB were differentiated osteogenically and additionally treated with/without 450 ng/m BMP-2, 0.5 µM dorsomorphin, and 0.5 µM SB431542 (as described above). On days 0, 7, 14, and 21, alizarin red S staining was performed. For each time point, a sextuple determination was performed. The values obtained were normalized to that of day 0, which was mathematically considered "1".
Alizarin red s staining. Alizarin red S is a dye that binds selectively to calcium salts and is widely used for calcium mineral histochemistry 84 . Adherent cell monolayers cultured in 6-well plates were washed with PBS and fixed with 4% paraformaldehyde for 15 min, rinsed 2 times with PBS, covered for 20 min at 37 °C with alizarin red S (0.5% in aqua dest., pH 4.1) and washed with dH 2 O until the supernatant was colorless. Stained monolayers were visualized by phase microscopy using an inverted microscope (Zeiss Axiovert 200 microscope). This process was followed by a quantitative destaining procedure using 10% (w/v) cetylpyridinium chloride in 10 mM sodium phosphate, pH 7.0, for 25 min at room temperature. The alizarin red S concentration was determined by absorbance measurement at 600 nm 85  www.nature.com/scientificreports/ β-Galactosidase assay. Cells were seeded in 1.9 cm 2 /well and cytochemically stained at pH 6.0 for senescence-associated β-galactosidase with a Senescence-β-Gal Staining Kit (Cell Signaling Technology, Massachusetts, USA) according to the manufacturer's instructions. As a positive control, the cells were treated with etoposide (25 µM, 48 h) and allowed to recover. Respective microscopically images were performed with Zeiss Axiovert 200. Furthermore, Western blot analysis for detection of β-galactosidase were performed (description further below). In short proteins at P3 and P10 from DSC, ASC and FB were collected, 40 µg/lane were applicated and incubated with α-β-galactosidase antibody (abcam ab616, 1:2000) and respective protein expression was normalized on total protein 86 .
Cell viability test (metabolic activity). The cell number was calculated by using CellTiter-Blue (Promega, Madison, USA). CellTiter-Blue's working dilution was 1:20 in medium. CellTiter-Blue uses an indicator dye to measure the metabolic activity of cells as indirect evidence for cell viability. After 1 h, the fluorescence (540 Ex /590 Em ) was measured in a 1420 Multilabel Counter (Victor 3 , Perkin Elmer).

Statistical analysis.
Values represent the mean ± standard deviations (SD). Statistical analysis was performed using two-way ANOVA followed by an appropriate post hoc Bonferroni test. Furthermore, we used two-sided Student's paired t-test. A p < 0.05 was considered significant.

Ethics approval and consent to participate. Study approval was obtained from the Ethics Review
Board of the Medical Faculty, Heinrich Heine University Düsseldorf (Study No. 3634). All patient-related data were anonymized before analysis. The usage of human material was conducted in compliance with the Declaration of Helsinki Principles. Written informed consent was obtained from all patients. All donors were analysed separately and not pooled, the specific numbers used are indicated in figure captions.

Data availability
The data that support the findings of this study are available from the corresponding author on reasonable request.