All‐trans retinoic acid and human salivary histatin‐1 promote the spreading and osteogenic activities of pre‐osteoblasts in vitro

Cell‐based bone tissue engineering techniques utilize both osteogenic cells and biomedical materials, and have emerged as a promising approach for large‐volume bone repair. The success of such techniques is highly dependent on cell adhesion, spreading, and osteogenic activities. In this study, we investigated the effect of co‐administration of all‐trans retinoic acid (ATRA) and human salivary peptide histatin‐1 (Hst1) on the spreading and osteogenic activities of pre‐osteoblasts on bio‐inert glass surfaces. Pre‐osteoblasts (MC3T3‐E1 cell line) were seeded onto bio‐inert glass slides in the presence and absence of ATRA and Hst1. Cell spreading was scored by measuring surface areas of cellular filopodia and lamellipodia using a point‐counting method. The distribution of fluorogenic Hst1 within osteogenic cells was also analyzed. Furthermore, specific inhibitors of retinoic acid receptors α, β, and γ, such as ER‐50891, LE‐135, and MM‐11253, were added to identify the involvement of these receptors. Cell metabolic activity, DNA content, and alkaline phosphatase (ALP) activity were assessed to monitor their effects on osteogenic activities. Short‐term (2 h) co‐administration of 10 μm ATRA and Hst1 to pre‐osteoblasts resulted in significantly higher spreading of pre‐osteoblasts compared to ATRA or Hst1 alone. ER‐50891 and LE‐135 both nullified these effects of ATRA. Co‐administration of ATRA and Hst1 was associated with significantly higher metabolic activity, DNA content, and ALP activity than either ATRA or Hst1 alone. In conclusion, co‐administration of Hst1 with ATRA additively stimulated the spreading and osteogenicity of pre‐osteoblasts on bio‐inert glass surfaces in vitro.

Large-volume bone defects (LVBD) may severely influence aesthetics and musculoskeletal functions. Due to the limited healing capacity of bone tissues, the osseous repair of LVBD can be problematic [1]. For treatment purposes, autologous bone grafts are still considered as the gold standard. However, their application is confined by limited graft supply, donor site pain and morbidity, infections, and poor cosmetic outcomes [2]. As alternative options to autologous bone grafts, allografts, xenografts, and synthetic materials have been developed and adopted as bone-defect-filling materials [3]. However, most of these materials need to be premixed with autologous bone grafts to obtain osteogenic cells. In such cases, the disadvantages of autologous bone grafts remain.
To approach these challenges, cell-based tissue engineering techniques that integrate osteogenic cells and biomedical materials have emerged as a promising approach for bone repair [4]. However, the chance of success is, however, highly dependent on the interactions of the osteogenic cells with the cell-scaffold surfaces [5]. At first, cell-substrate interactions are critical for the determination of cell fates, such as proliferation, quiescence, or apoptosis [6]. Furthermore, surface adhesion and osteogenic cell proliferation are indispensable for initiation of bone regeneration [7]. Consequently, tremendous efforts have been made to develop a large variety of techniques (e.g., immobilized RGD peptide on titanium surface [8] and femtosecond laser-induced micropattern and Ca/P deposition [9]) to modify surface chemistry and/or topography of various biomedical materials in order to improve their cell-substrate interactions [5,10]. These material-specific approaches, however, render their broad applicability limited. In comparison, the other strategiescelltargeting techniques that directly promote cellular response to materialshave become highly attractive as they do not require surface modifications of materials, thus bearing a broader applicability.
A promising candidate cell-targeting agent to promote cell-substrate interactions is histatin-1 (Hst1), a member of a large histidine-rich salivary peptide family. Our previous findings show that Hst1 can significantly promote the attachment, spreading, and migration of various cell types including epithelial, endothelial, and osteogenic cells [11][12][13][14][15]. Our recent data confirm that Hst1 can promote the spreading of osteogenic cells on both bio-inert glass and titanium surface [14][15][16][17], which suggests a promising application potential of Hst1 in the cell-based bone tissue engineering.
In a previous study, we found that a 3-day treatment of all-trans retinoic acid (ATRA), an active metabolite of vitamin A, can cause the uniform alignment and stretch of cell skeleton (Fig. S1). This finding inspired us to apply ATRA to promote cell spreading. ATRA, the active metabolite of vitamin A, is known to act as regulator of many physiologic processes [18]. It plays a role in a wide range of biological processes mediated through binding and activation of the nuclear receptors, such as the RA receptor (RAR) and retinoid X receptor (RXR). There are three subtypes of RAR (a, b, and c) and three subtypes of RXR (a, b, and c). RARs are bound and activated by ATRA, while RXRs are bound and activated by the 9cis-RA only [19]. Heterodimers of activated RAR and RXR act as ligand-dependent transcription factors. On the other hand, it was found that a 3-day treatment of ATRA also results in significantly reduced osteogenic differentiation of pre-osteoblast cells and bone marrow stromal cells [20,21]. Consequently, in the present study, we analyzed in vitro the effect of a short (2 h) co-application of ATRA and Hst1 in order to amplify the stimulating effect of Hst1 on the spreading of osteogenic cells on the one hand and to avoid the decrease in osteogenic potential on the other hand.

Study design
The effect of a short (2 h) co-administration of ATRA and Hst1 on cell spreading was evaluated. Thereafter, we used specific inhibitors of retinoic acid receptor alpha (RARa), RARb, and RARc, that is, ER-50891, LE-135, and MM-11253, respectively, to identify the involvement of RARs. Furthermore, we examined the effects of a short co-administration of ATRA and Hst1 on the osteogenic potentials of pre-osteoblast cells, such as metabolic activity, DNA content (indicator for proliferation), and alkaline phosphatase (ALP) activity (early marker of osteogenic differentiation).

Measurement of cell spreading on glass surface
Cells were treated with serum-free medium for 24 h before being detached by 0.05% trypsin (Gibco, Thermo Fisher Scientific). Growth medium contained 2% FBS was used to inactivate the effect of trypsin and to resuspend the cells. MC3T3-E1 was seeded on coverslips (20 mm in diameter; Thermo Scientific, Braunschweig, Germany) in 12-well plates at a density of 6 9 10 4 cells/well. Cells were treated either with 0, 1, 10, or 20 µM ATRA (Sigma-Aldrich) or Hst1 or co-administered 10 µM ATRA and Hst1. To investigate the role of potential signaling pathways, 10 µM RARa antagonist (ER-50891; R&D, Bio-Techne, Minneapolis, MN, USA), 10 µM RARb antagonist (LE-135; R&D, Bio-Techne), and 10 µM RARc antagonists (MM-11253; R&D, Bio-Techne) were supplemented in cell spreading assays. Cells were photographed every 20 min for 3 h using a microscope (EVOS FL; Thermo Fisher Scientific) equipped with a LPlanFL PH2 209 using the phase-contrast setting or the Cy5 light cube (628/40 and 692/40 nm, excitation and emission filters, respectively). Relative cell spreading surface area was quantified by measuring the surface area of cells' filopodia and lamellipodia using a manual point-counting method [23] (Fig. S2). Each assay was performed in triplicate and repeated twice.

Fluorescent staining of spreading cells
Cell spreading on glass surface was performed as described in the section of cell spreading assay. 1.5 h after seeding, cells were fixed, dehydrated, and stained with FITC-Phalloidin. Fluorescent micrographs were randomly taken using a fluorescent microscope (Leica Microsystems GmbH, Wetzlar, Germany) with excitation/emission wavelengths (nm) of 496/516. On the micrographs, spreading surface of each cell was estimated using the above-mentioned point-counting method. More than 20 cells per group were calculated.

Cell metabolic activity
Subconfluent growing cells were plated on glass coverslips (diameter, 10 mm; Thermo Scientific, Germany) in 48-well plate in a density of 1.5 9 10 4 cells/well. Cells were treated with either 10 µM Hst1 or ATRA, or cells were treated with premixed ATRA and Hst1 for 2 h at 37°C. After washing with 19 PBS, cells were treated with a-MEM with 10% FBS which was refreshed on a daily basis. PrestoBlue TM Cell Viability Assay was adopted to evaluate cell viability using the reducing ability of cells (Invitrogen Corporation, Carlsbad, CA, USA). In short, 1/10th volume of Presto-Blue TM reagent was added to cells in culture medium and incubated for 30 min at 37°C. Results were measured by reading fluorescence intensity with the Multiskan FC (Thermo Scientific) using a fluorescence excitation wavelength of 560 nm and an emission wavelength of 590 nm. Each assay was performed in triplicate and repeated twice.

DNA quantification
The CyQUANT Proliferation Assay Kit (Molecular Probes, Waltham, MA, USA) was employed to monitor the proliferation of pre-osteoblasts. Subconfluent growing cells were plated on glass coverslips (diameter, 10 mm; Thermo Scientific, Germany) in a 48-well plate at a density of 1.5 9 10 4 cells/well. Cells were treated with either 10 µM Hst1 or ATRA, or co-administered ATRA and Hst1 for 2 h. After washing with 19 PBS, cells were treated with a-MEM with 10% FBS which was changed every day. The cells were retrieved right after or 5 days after the short treatment. Subsequently, the freshly prepared 100 lL CyQUANT solution was added to the well to measure the optical density with excitation at 480 nm and emission at 520 nm using a plate reader (Synergy, BioTek TM , Winooski, VT, USA). Each assay was performed in triplicate and repeated twice.

Alkaline phosphatase assays
Quantitative determination of ALP activity was done using the p-nitrophenyl phosphate (pNPP) liquid substrate method. Cells were suspended in serum-free media in the presence or absence of 10 µM Hst1 or 10 µM ATRA or both and then seeded on glass coverslips (10 mm in diameter; Thermo Scientific, Germany) in 48-well plates at a density of 5 9 10 4 cells/well. Two hours after seeding, the media were changed to 10% FBS-containing a-MEM, and subsequently, cells were cultured for 1 more day. Thereafter, cells were treated with a-MEM containing 2% FBS. After 3 days, cells were lysed in distilled water using a freeze-thaw method and harvested with a cell scraper. Cell lysates were centrifuged at 250 g for 5 min at room temperature, and supernatants were incubated with 1.86 mgÁmL À1 pNPP for 1 h at 37°C in the dark. After 1 h, 100 lL 300 mM NaOH solution was added; then, absorbance at 405 nm was measured using Multiskan FC (Thermo Scientific, Rockford, IL, USA), and ALP activity was calculated according to the standard curve. The total protein was assessed using Pierce BCA Protein Assay Kit (Thermo Fisher Scientific, Rockford, IL, USA) for normalizing the ALP activity [20]. Each assay was performed in quadruplicate and repeated twice.

Statistical analysis
Data were plotted using GRAPHPAD PRISM (GraphPad Software version 6.0, La Jolla, CA, USA) and analyzed by oneway ANOVA with Bonferroni's post hoc test for multiple comparisons. For the data from different groups at different time points in Fig. 2C, we used two-way ANOVA to analyze the data with Tukey test for multiple comparisons. Results were reported as mean AE standard deviation (SD). A value of P < 0.05 was considered as statistical significance.

Results and Discussion
ATRA and Hst1 promote the spreading of osteogenic cells on bio-inert glass surface subsection At a concentration of 10 µM, ATRA significantly promoted the spreading of pre-osteoblasts on bio-inert glass surfaces in comparison with the control (no ATRA) (Fig. 1A,B). Hst1 at 10 and 20 µM significantly promoted the spreading of osteogenic cells in comparison with the control (no Hst1) (Fig. 1C,D). Thereafter, we performed a pilot experiment to check the effect of co-administered 10 µM Hst1 and ATRA of different concentrations (e.g., 0.1, 1, and 10 µM). Only 10 µM ATRA and 10 µM Hst1 resulted in a significant cell spreading area than 10 µM Hst1 effect (data not shown). Therefore, we adopted the combination of 10 µM ATRA and 10 µM Hst1. Our data showed that there was no significant difference between the promoting effects of 10 µM Hst1 and 20 µM Hst1. In this light, it was chosen to further use 10 µM ATRA and 10 µM Hst1 in the following experiments.
The antagonists of RARa and RARb suppressed the promoting effect of ATRA and Hst1 on cell spreading The antagonists of RARa (ER-50891) and RARb (LE-135) significantly suppressed the promoting effect of the co-administered ATRA and Hst1 on the spreading of pre-osteoblasts (Fig. 4A). Consistent with above-mentioned results, 10 µM ATRA significantly elevated the promoting effects of 10 µM Hst1 (P < 0.05), which could be nullified by the pretreatment of 10 µM ER-50891 (Fig. 4B) or 10 µM LE-135 (Fig. 4C).

Co-administration of ATRA and Hst1 upregulated the osteogenic activities of pre-osteoblasts
Two-hour treatment of 10 lM Hst1 significantly enhanced the metabolic activity of pre-osteoblasts already after 1 day, in contrast to ATRA (Fig. 5A). Furthermore, the co-administration of ATRA and Hst1 resulted in significantly higher metabolic activity in comparison with Hst1 alone (P < 0.05) (Fig. 5A). Directly after seeding, the DNA content in cells treated with Hst1 and ATRA was significantly higher than those stimulated by Hst1 or ATRA alone. Five days after seeding, the DNA content in the group of ATRA alone and Hst1 alone was significantly higher compared to the control group. The co-administration of ATRA and Hst1 resulted in a significantly higher DNA content than those in the groups of ATRA or Hst1 alone (Fig. 5B). Three days postseeding, the ALP activity of the cells treated with Hst1 and ATRA was about 2.6-fold higher (P < 0.001) than those in the groups of Hst1 alone, ATRA alone, or control (Fig. 5C).
Surface adhesion, spreading, proliferation, and differentiation of osteogenic cells are critical steps for their respective success within cell-based bone tissue engineering techniques [24]. Previously, we found that Hst1 promoted the spreading of osteogenic cells on both bio-inert substrates and titanium SLA surfaces in vitro. In this study, we found that the co-administration of ATRA and Hst1 significantly increased cell spreading efficiency compared to the presence of Hst1 only.
In line with previous work, cell spreading was used as a key parameter to evaluate efficacy of surface compatibility for osteogenic cells by different bioactive agents in vitro. In our previous studies, we have used the percentage of spreading cells or cell index as parameters to evaluate cell spreading [14,15]. The former parameter indicates the percentage of cells that initiate protrusion, and the latter parameter qualifies the impedance of cells that proportionally correlate to, but not directly show, cell spreading extent. In contrast, in the current study, we adopted a point-counting method [25] to directly measure the surface area of cell spreading parts, which could directly reflect the newly formed cell-substrate contact area. We further subtracted the area of nuclei to purely evaluate area of spreading part, which helped us to more precisely evaluate the spreading extent.
Most of the current methods to promote cell-substrate interaction are focused to modify the surface chemistry and/or topography [26]. However, due to the large variety of biomaterials, there is still an apparent great need for broadly applicable approaches to promote cell spreading on biomaterials. Previously, we and others showed that Hst1 promoted adhesion, spreading, and migration of various epithelial cells from different origins, such as mucosa [12,16], gingiva [14,15], cornea [13], and skin [27], endothelial cells [14,17], and osteogenic cells [11,14]. All together, these findings underline a non-cell type-specific character of Hst1 rendering a promising application potential for tissue engineering purposes. Next to Hst1, ATRA was used as agent to promote cell spreading for cell-based tissue engineering techniques. Numerous studies have demonstrated that the RA signaling pathway, which is mediated via RAR and/or RXR, can modulate the expression of genes involved in cell growth, [28] energy metabolism, [24] and immune responses [29,30]. In a  previous study, we found that a 3-day treatment of ATRA caused uniformly-directionally alignment of actin in vitro (Fig. S2). Notably, it was reported that ATRA increased the adhesion and spreading of pancreatic stellate cells via RARb-dependent signaling, thereby inhibiting cancer cell invasion [31]. In this process, ATRA-treated pancreatic stellate cells formed larger focal adhesion complexes, spread faster, attained a larger spreading area, attached stronger to the ECM (extracellular matrix), and displayed significantly larger and brighter focal adhesion complexes (both for talin and paxillin) in comparison with untreated control cells [31].
Here, we showed that RARs were potentially involved for the promoting effect of ATRA on the spreading of pre-osteoblasts. For this purpose, we adopted specific antagonists of RARa, RARb, and RARc and found that the antagonists of RARa (ER-50891) and RARb (LE135), but not RARc (MM-11253), significantly suppressed cell spreading induced by co-administered ATRA and Hst1. Furthermore, we found that the antagonists of RARa (ER-50891) and RARb (LE135) abolished the amplification by ATRA of Hst1's effects on cell spreading. These findings suggested that ATRA affects cell spreading by RARaand RARb-dependent signaling. This may be consistent with the reports that the agonists of RARa and RARb, but not RARc, activated focal adhesion kinase (FAK) and paxillin in breast cancer cells [32].
Concerns may be raised for using ATRA since it was previously shown to have negative effects on the adhesion and migration of epithelial cells [33].
Treatment of 0.1-1 lM ATRA for 1 h significantly inhibited the adhesion of retinal pigment epithelial cells. Furthermore, it was found that ATRA significantly inhibited the spreading of retinal pigment epithelial cells with suppressed FAK, suggesting that ATRA's effect is highly cell type-dependent. Consequently, caution must be taken for extrapolating these data to osteogenic cells. Another concern may be that ATRA may inhibit the osteogenic activities of pre-osteoblasts [34] and bone marrow stromal cells [35]. In our previous studies, we showed that a long-term (3-21 days) treatment with ATRA significantly reduced cell proliferation, metabolic activity, protein expression, osteocalcin expression, and extracellular matrix mineralization of osteogenic cells [20,34,35]. In our current study, we showed that the short-term (2 h) treatment of either Hst1 alone or ATRA alone did not result in significantly higher DNA content compared to the control. Surprisingly, the DNA content in the group of the co-administered ATRA and Hst1 was significantly higher than those in the groups of ATRA alone, Hst1 alone, or control, which suggested the coadministration of ATRA and Hst1 synergistically promoted cell attachment. Our data further showed that the 2-h co-administration of ATRA and Hst1 resulted in significantly enhanced metabolic activity of pre-osteoblasts within the monitoring time span (5 days) than either ATRA or Hst1 alone. Moreover, neither ATRA alone nor Hst1 alone had any effect on ALP activity, suggesting that neither of them significantly influence osteogenic differentiation of pre-osteoblasts. In contrast, the combination of ATRA and Hst1 Folds of the metabolic activities of pre-osteoblasts within 5 days after a short (2 h) treatment with either no Hst1, no ATRA (control), or 10 µM Hst1, or 10 µM ATRA, or co-administered 10 µM ATRA and 10 µM Hst1 with a-MEM containing 10% FBS during seeding (n = 6). (B) Folds of DNA content at 0 day and 5 days after response graph of DNA content after a short (2 h) treatment with either no Hst1, no ATRA (control), or 10 µM Hst1, or 10 µM ATRA, or co-administered 10 µM ATRA and 10 µM Hst1 with a-MEM containing 10% FBS during seeding (n = 6). (C) Folds of ALP activity at 3 days after a short (2 h) treatment either without Hst1 or ATRA (control), or with 10 µM Hst1, or 10 µM ATRA, or co-administered 10 µM ATRA and 10 µM Hst1 with a-MEM containing 2% FBS during seeding (n = 8). Data were plotted using GRAPHPAD PRISM (GraphPad Software version 6.0) analyzed by one-way ANOVA with Bonferroni's post hoc test for multiple comparisons. Data were shown as mean AE SD. *P < 0.05; **P < 0.01; ***P < 0.001. significantly enhanced ALP activity. These data indicate that such a treatment with ATRA and Hst1 is potentially suitable to promote both the cell-substrate interactions of pre-osteoblasts and enhance their osteogenic differentiation.
The underlying molecular mechanisms of ATRA and Hst1 co-administration on ALP activity remain to be elucidated. Possibly, the activation of p38 MAPK signaling pathway may be involved. Recently, we found that specific p38 MAPK inhibitors abolish the promoting effect of Hst1 (data not shown) in this type of pre-osteoblasts, suggesting that Hst1 could activate p38 signaling. It is well-established that p38 MAPK is a key mediator for many drugs to upregulate ALP activity in pre-osteoblasts [36][37][38]. On the other hand, ATRA is also found to transiently activate p38 signaling [39,40]. Although a short-term treatment of either Hst1 or ATRA seemed not sufficient to induce ALP activity, in combination Hst1 and ATRA significantly upregulated ALP (Fig. 5), suggesting an additive stimulating effect on p38 MAPK signaling. Further studies are needed to confirm this hypothesis. With the inspiration of the ALP result, we, thereafter, performed an experiment of extracellular matrix mineralization with osteogenic medium to check the effect of the 2-h coadministration of ATRA and hst1. We found that the 2-h co-administration of ATRA and hst1 was associated with a higher (without statistical difference) mineralization at 21 days post-treatment than the control group (data not shown). In fact, the result is not so surprising since the effect of a 2-h treatment can quickly taper during the 21-day culture period with osteogenic medium (10% FBS-containing a-MEM with beta-glycerophosphate and L-ascorbic acid-2phosphate as supplements). Consequently, the shortterm co-administration of ATRA and hst1 can show a significant effect only in the initial cellular events of osteogenic activities, such as cell adhesion, spreading, proliferation, and early differentiation. For late (e.g., osteocalcin expression) and final differentiation (extracellular matrix mineralization), osteoinductive growth factors, such as bone morphogenetic proteins, are highly needed.
Finally, in this study we used a mouse MC3T3-E1 cell line. Although MC3T3-E1 cells are widely used as a cell model for pre-osteoblasts, these effects must be replicated in systems that bear more relevance for the human physiological situation, such as primary mesenchymal stem cells and osteoblasts. Furthermore, caution should be taken to extrapolate the current in vitro results to in vivo situation. Animal studies are highly needed to confirm the promoting effect of co-administered ATRA and Hst1.
In summary, our current study showed that Hst1 and ATRA co-administration positively influenced the spreading, cellular metabolic activity, proliferation, and osteogenic differentiation of pre-osteoblasts. Based on these observations, we postulated that such combined treatment may be supportive for cell-based bone tissue engineering techniques.

Supporting information
Additional supporting information may be found online in the Supporting Information section at the end of the article. Fig. S1. Fluorescent micrographs depicting the spreading of pre-osteoblasts (stained with FITC-Phalloidin) with or without a treatment with 1 µM ATRA for 3 days. Bar = 50lm. Fig. S2. Graph depicting a point-counting method to measure the surface area of cell spreading. The grid was randomly put on the light micrographs of cells during spreading for the point-counting method. The filopodia and lamellipodia (red arrow) was included for calculating the cell spreading area with the exclusion of the relatively constant peri-nuclear area (within red dot circle). Bar = 50lm.