Topographic cues of a novel bilayered scaffold modulate dental pulp stem cells differentiation by regulating YAP signalling through cytoskeleton adjustments

Abstract Objectives Topographic cues can modulate morphology and differentiation of mesenchymal stem cells. This study aimed to determine how topographic cues of a novel bilayered poly (lactic‐co‐glycolic acid) (PLGA) scaffold affect osteogenic/odontogenic differentiation of dental pulp stem cells (DPSCs). Methods The surface morphology of the scaffolds was visualized by scanning electron microscope, and the surface roughness was measured by profilometry. DPSCs were cultured on each side of the scaffolds. Cell morphology, expression of Yes‐associated protein (YAP) and osteogenic/odontogenic differentiation were analysed by immunohistochemistry, real‐time polymerase chain reaction, and Alizarin Red S staining. In addition, cytochalasin D (CytoD), an F‐actin disruptor, was used to examine the effects of F‐actin on intracellular YAP localisation. Verteporfin, a YAP transcriptional inhibitor, was used to explore the effects of YAP signalling on osteogenic/odontogenic differentiation of DPSCs. Results The closed side of our scaffold showed smaller pores and less roughness than the open side. On the closed side, DPSCs exhibited enhanced F‐actin stress fibre alignment, larger spreading area, more elongated appearance, predominant nuclear YAP localization and spontaneous osteogenic differentiation. Inhibition of F‐actin alignments was correlated with nuclear YAP exclusion of DPSCs. Verteporfin restricted YAP localisation to the cytoplasm, down‐regulated expression of early osteogenic/odontogenic markers and inhibited mineralization of DPSCs cultures. Conclusions The surface topographic cues changed F‐actin alignment and morphology of DPSCs, which in turn regulated YAP signalling to control osteogenic/odontogenic differentiation.

filaments and microtubules), 7 actin presents as either a free monomer (G-actin) or a linear polymer (F-actin). The F-actin cytoskeleton network plays a key role in regulating important physical processes, such as cell morphology, adhesion and proliferation. [8][9][10] Specifically, differentiation of human mesenchymal stem cells (MSCs) into chondrocytes and osteoblasts is associated with structural changes of F-actin networks. 11,12 F-actin can activate several signalling molecules, which include the Rap1, Rho family GTPases, phosphoinositide 3-kinase (PI3K) and YAP signalling. 9,13 The YAP/Tafazzin (TAZ) pathway reportedly works as a nuclear relay of mechanical signals. 9 Phosphorylated YAP is automatically excluded from the nucleus, while non-phosphorylated YAP translocates into the nucleus and actively regulates MSCs differentiation. 11,14,15 Regenerative endodontics has been recognized as an alternative treatment modality for the permanent tooth with necrotic pulp and immature root. It is a biologically based procedure designed to physiologically replace the necrotic pulp and hence promote continued formation of root and apical closure. The ultimate goal is to regenerate a functional pulp-dentin complex. 16 Regenerative endodontics applies all the principles of regenerative medicine and tissue engineering, that is, it utilizes specific cells, three-dimensional scaffolds and growth factors alone or in combination to regenerate new tissues. 17,18 The new protocol for regenerative endodontics has been introduced to clinics in 2004. 19 In most cases, the regenerative endodontic treatment failed to regenerate pulp-dentin tissue inside canals. In few cases, the 'pulp-like' tissue was found inside the treated root canal, consisting of a mixture of dentin, bone, blood vessels, and some collagen components, but it lacked the spatial organization that is most characteristic for the pulp-dentin complex. 20,21 To improve the outcomes, several biomaterials have been tested for their potential use as scaffolds in regenerative endodontics. One that has caught a lot of attention is platelet-rich plasma (PRP). Studies have shown that using PRP adjunct to blood clot promotes root lengthening and thickening. 22 Other materials, such as dentin matrix and peptide hydrogel (Puramatrix™), have also been tested in animal models and shown to have a certain level of regenerative potential. 23 In addition, growth factors, such as TGFβ1, SDF-1 and BMP, have been incorporated into scaffolds to promote stem cell differentiation and tissue regeneration. 24 Although all these new scaffold materials support three-dimensional cell cultures, their homogenous nature limits their capability to provide spatial control over cell activities, thus fail to provide the different zone for pulp (centre area) and dentin (peripheral area) regeneration. 25 Recognizing these limitations, our laboratory has generated a novel, bilayered biomimetic tissue scaffold that is designed to provide spatial control of dentin and pulp tissue regeneration. 26 The porous scaffold has two different layers, namely an open side and a closed side. On the open side, human dental pulp stem cells (DPSCs) penetrated into the scaffold through the channels, while on the closed side the cells rather proliferated on the surface and underwent spontaneous osteogenic differentiation. Importantly, the observed scaffold-guided osteogenic differentiation on the closed side occurred in basal medium, in the absence of exogenous osteogenic inducers. 26 We suppose this is likely due to the optimized topographic characteristic of our scaffold, which regulates the differentiation of DPSCs. 27 The purpose of this study was to test our hypothesis that the open and closed sides of our scaffold provide different topographic cues, which change the cytoskeleton arrangement and target the YAP signalling pathway, leading to the spontaneous differentiation of DPSCs into osteoblasts/odontoblasts on the closed side.

| Scaffold fabrication and topographic testing
Membrane-like PLGA (75:25; Evonik Industries) scaffolds were fabricated by diffusion-induced phase separation (DIPS) technique as described previously. 26 Briefly, 12% or 20% PLGA was dissolved in dimethyl sulfoxide (DMSO, Sigma) overnight. Then, the PLGA was cast on a glass plate and submerged into deionized water, independently.
After chaning the water five times, the PLGA membranes were carefully detached from the glass surface. The glass side of the 12% PLGA layer was referred as the 'open side', while the water side of the 20% PLGA layer was referred as the 'closed side'. The two layers were finally laminated under vacuum (0.2 Torr) at 4°C for 24 hours.
The morphology of the freshly assembled, dry scaffolds was examined by scanning electron microscopy (SEM, FEI Quanta 450FEG).
The surface roughness of the scaffolds was measured using a contact profilometer (Surfcorder SE 1700, Kosaka Labs) with a 2 μm radius tip, a spread of 2.5 mm and a speed of 0.5 mm/s.

| Dental pulp stem cells culture on the scaffolds
Human DPSCs were purchased from AllCells LLC. These cells were guaranteed by company through 10 population doublings, and they were positive for CD105, CD166, CD29, CD90 and CD73 and negative for CD34, CD45 and CD133. DPSCs were cultured in a basal medium composed with α-MEM (Gibco), supplemented with 10% foetal bovine serum (FBS, HyClone), penicillin (100 U/mL, HyClone) and streptomycin (100 μg/mL, HyClone) and incubated in a humidified incubator of 5% CO 2 at 37°C. The PLGA membranes were cut into squares (2.5 × 2.5 cm 2 or 3.5 × 3.5 cm 2 ) and placed into 12-well or 6-well plates (Corning), with either the open side or the closed side facing upward. DPSCs (passages 4~6) were seeded at a density of 5 × 10 3 cells/cm 2 in the following experiments. The medium was changed every 3 days.

| F-actin fibre analysis
At days 1, 7 and 14, the PLGA membranes with attached DPSCs on either the open side or the closed side were fixed in 3.7% paraformaldehyde (Thermo Fisher) at room temperature for 15 minutes. Then, the cells were permeabilized with 0.3% Triton X-100 (ThermoFisher) for 1 hour. Afterwards, the microfilaments were labelled with Alexa Fluor 546 Phalloidin (ThermoFisher). Images were captured in a Fluoview FV1000 (Olympus) laser scanning confocal microscope (LSCM). Orientation of the F-actin fibres with respect to the long axis of the cells was determined, as described in the literature. 28 Briefly, from the five microscopic fields (upper, lower, left, right and centre), six cells were randomly chosen from each field; therefore, a total of 30 individual cells were chosen from each group.
For the fibre analysis, a reference line was drawn through the long axis of the cell. Five fibres from each cell were selected to obtain the average fibre angels. ImageJ software (NIH, version 1.51) was used to measure the angle between the long axis of the cell and each fibre (value between 0 and 90°). Fibres with the angles less than 10° were considered aligned.

| Cell morphology and YAP signalling analysis
The intercellular localization of YAP in DPSCs was analysed by im- The AR is the ratio of the major axis to the minor axis. By superimposing F-actin and DAPI images with YAP images, the ratio of YAP expression in the nucleus vs. cytoplasm was calculated as (nuclear YAP/nuclear area)/(cytosolic YAP/cytoplasm area). The following human-specific validated primers (Biorad) were used: collagen 1 (COL1A1, qHsaCED0043248), YAP (qHsaCED0037679), alkaline phosphatase (ALP, qHsaCID0010031) and runt-related transcription factor 2 (RUNX2, qHsaCID0006726). All procedures were performed according to the manufacturer's protocols. The expression levels of the target genes were normalized to glyceraldehyde-3-phosphate dehydrogenase (GAPDH, qHsaCED0038674) and calculated using the ∆∆CT method. 29

| Statistical analysis
Each experiment except for the single-cell analyses was repeated independently at least three times. Student's t test was used to compare two samples within the same group. The results were analysed using SPSS 23.0 (IBM) software. All values were expressed as mean ± standard error, and statistical significance was set at a P < 0.05.

| Topographic cues regulate actin organization of DPSCs
The    Figure 2G). The YAP nuclear/cytosolic ratio was higher in DPSCs on the closed side on day 1 ( Figure 2H).

| DPSCs exhibit different morphology and YAP localization on the different sides of the scaffold
Consistently, DPSCs on the closed side on day 7 also exhibited a higher YAP nuclear/cytosolic ratio ( Figure 2I). changed ( Figure 3D), the CSI was increased ( Figure 3E) and the AR ( Figure 3F) was decreased, indicating that the cell morphology changed from spindle to circular shape.

| Scaffold-induced osteogenic differentiation of DPSCs on the closed side is attenuated by YAP inhibitor
As shown previously, YAP was predominantly expressed in the nu- between the two sides ( Figure 4D). On the closed side, verteporfin treatment inhibited the expression of ALP and COL1A1 ( Figure 4E).

| D ISCUSS I ON
As a main component of cytoskeleton, F-actin actively responds to the surrounding environment. Once a cell senses physical cues from substrate attachment, the dynamic cytoskeleton network starts to reorganize accordingly. 30 For instance, when endothelial cells were cultured on a grooved dimethylsiloxane substrate, F-actin aligned along the grooves and ridges, but it did not show any preferential orientation on a smooth surface. 31 Comparably, ridged or grooved silicon master surfaces also dramatically impacted alignment, elongation and aspect ratio of MSCs in a size-and shape-dependent manner. MSCs on both planar and porous surfaces seemed to be more rounded when compared to their parallel orientation on ridged or grooved surface. 2 We developed a bilayered PLGA scaffold in order to provide differential spatial guidance for DPSCs. promoted osteogenic differentiation of MSCs. 36 Similarly, a layered PLGA membrane with a RA of ~0.3 μm facilitated osteogenic differentiation. 37 In addition, the small pore size (1~5 μm) on the closed side seemed to be advantageous for DPSCs mineralization, while another study showed that porous PLGA microscaffolds with pore diameters of 10~30 μm supported the tri-lineage differentiation potential of DPSCs. 38 Growing evidence shows that cytoskeleton change is essential The YAP/TAZ pathway is a transcriptional co-activator of the Hippo pathway, which has been implied as a key mediator of mech- which may contribute to the enhanced osteogenic differentiation observed on the closed side. In our hands, enhanced nuclear localization of YAP was consistent with higher F-actin alignment, larger spreading area and higher AR in DPSCs. This finding is in line with previous studies. 6,9,47 To further confirm the above findings, two distinct pharmacological inhibitors of F-actin polymerization and YAP signalling were used. First, after CytoD treatment, DPSCs showed a dramatic decrease in nuclear YAP accumulation and an increase in cytosolic YAP expression, indicating that F-actin polymerization regulates YAP translocation between cytoplasm and nucleus. Our result is consistent with previous studies demonstrating that CytoD treatment promoted production of intranuclear actin and led to nuclear export/exclusion of YAP. 48 Previous study also showed that CytoD treatment decreased the cell volume and increased the cytosolic localization of YAP in MSCs cultured in hydrogels. 9 Second, we treated DPSCs with verteporfin, a transcriptional inhibitor of YAP.
Verteporfin was identified as an inhibitor of the interaction of YAP with TEAD and thus blocks transcriptional activation of YAP downstream genes. 49 Although verteporfin treatment could remarkably up-regulate cytoplasmic YAP in a series of cells, which may induce sequestration of cytosolic YAP through increasing levels of the cellcycle regulatory 14-3-3σ proteins, the studies related to osteogenic differentiation are very limited. 9,50 Verteporfin treatment in our experiment caused a significant down-regulation of the expression of COL1A1 and ALP, as well as a reduced formation of mineralization nodules in DPSCs culture, but had no effect on RUNX2 expression. This is similar to a previous report that verteporfin dose dependently reduced mRNA levels of COL1A1, COL1A2, BSP and OCN, but not RUNX2 and osterix in osteoblast-like cells. 51 Taken together, our data suggest an important role of YAP signalling in the osteogenic/odontogenic differentiation of DSPCs. Our study demonstrates that YAP signalling is activated during DPSCs differentiation upon F-actin alignment, regulated by topographic cues related to the roughness and pore sizes of our scaffold. We conclude that it is critical to understand dental stem cell behaviour on different scaffold surfaces and that cell behaviour can be manipulated by carefully controlling surface topographic properties. Our next step will investigate how this scaffold will guide the regeneration of pulp and dentin tissue inside root canals in vivo and whether YAP signalling is involved during the process of regeneration of the pulp-dentin complex.

CO N FLI C T O F I NTE R E S T S
The authors deny any conflicts of interest related to this study.