Direct electrical stimulation enhances osteogenesis by inducing Bmp2 and Spp1 expressions from macrophages and preosteoblasts

Abstract The capability of electrical stimulation (ES) in promoting bone regeneration has already been addressed in clinical studies. However, its mechanism is still being investigated and discussed. This study aims to investigate the responses of macrophages (J774A.1) and preosteoblasts (MC3T3‐E1) to ES and the faradic by‐products from ES. It is found that pH of the culture media was not significantly changed, whereas the average hydrogen peroxide concentration was increased by 3.6 and 5.4 µM after 1 and 2 hr of ES, respectively. The upregulation of Bmp2 and Spp1 messenger RNAs was observed after 3 days of stimulation, which is consistent among two cell types. It is also found that Spp1 expression of macrophages was partially enhanced by faradic by‐products. Osteogenic differentiation of preosteoblasts was not observed during the early stage of ES as the level of Runx2 expression remains unchanged. However, cell proliferation was impaired by the excessive current density from the electrodes, and also faradic by‐products in the case of macrophages. This study shows that macrophages could respond to ES and potentially contribute to the bone formation alongside preosteoblasts. The upregulation of Bmp2 and Spp1 expressions induced by ES could be one of the mechanisms behind the electrically stimulated osteogenesis.

The suggested mechanisms include the enhancement of osteoblastic activities at alkaline pH (Bodamyali et al., 1999;Fliefel et al., 2016;Galow et al., 2017;Ramp, Lenz, & Kaysinger, 1994), and the upregulation of vascular endothelial growth factor (Vegf) expression from macrophages by the electrically generated H 2 O 2 , which is beneficial for bone vascularization (M. Cho, Hunt, & Hussain, 2001;Griffin & Bayat, 2011). We also thought that H 2 O 2 may also induce high-mobility group box 1 (Hmgb1) protein from these inflammatory cells, which could potentially be beneficial for bone healing and regeneration (Meng et al., 2008;Tang et al., 2007). The involvement of macrophages in osteogenesis has been reported that they are requisite for in vivo bone healing and homeostasis, in which they have enhanced osteogenic differentiation of osteoprogenitor cells (Schlundt et al., 2015;Vi et al., 2015). Moreover, it has been shown recently that macrophages are also responsive to the electric field.
Their migration, orientation, and intracellular calcium could be altered by ES similar to other cell types as well as their phagocytosis activities (Hoare, Rajnicek, McCaig, Barker, & Wilson, 2016).
However, their electrically stimulated activities relating to osteogenesis are not widely studied.
Hence, it is of interest to this study to characterize the in vitro responses of macrophages to direct ES in comparison with preosteoblasts to investigate whether or not electrically stimulated macrophages could contribute to osteogenesis besides osteoblastic cells, and whether the changes in their responses are induced by faradic by-products as hypothesized. This study provides initial proof-of-concept results regarding the role of faradic by-products, which is beneficial for the future discussion regarding the mechanism of electrically induced osteogenesis.  (Ono et al., 2018;. Further details are described in the Supporting Information. Cells were maintained in Dulbecco's modified Eagle medium (4.5 g/L glucose, 2 mM L-glutamine, without sodium pyruvate) containing 10% fetal bovine serum and 1% antibiotic and antimycotic solution. Cells were incubated at 37°C, 5% CO 2, and atmospheric O 2 concentration. All reagents were purchased from Sigma-Aldrich, UK, unless stated otherwise. In the experiments, cells were seeded into six-well plates at the density of 50,000 cells per well for macrophages and 100,000 cells per well for preosteoblasts, and the media volume was kept at 3 ml. Media change was carried out on the following day after seeding before applying ES. Optical images of the cells were taken using EVOS™ XL Imaging System (Life Technologies).

| Direct ES system
The system used in this study was 0.5 mm 99.95% L-shaped platinum wire electrodes in six-well plate arrangement (Leppik et al., 2019;Mobini et al., 2016). Electrodes were wired in parallel circuit using jumper cables, as shown in Figure 1. The devices were disinfected with 70% ethanol spray and UV irradiation before and after being used. ES has been applied to the cells (Direct ES) for 1-2 hr daily at the constant direct current (DC) voltage of 2.2 V (100 mV/mm equivalent electric field) using DC generator (B&K Precision). The total current passing through each well once reaching steady state was 0.07 ± 0.01 mA (mean ± SD) measured by digital multimeter (M-830B, Sinometer). In faradic by-product studies, acellular media were placed in a separate six-well plate (3 ml per well) and incubated overnight before being stimulated for 2 hr. The electrically stimulated media (ES media) were transferred to the cells immediately after stimulation daily, and ES media were changed every time of stimulation.

| pH measurement
Benchtop pH meter (Hanna Instruments) was used for measuring the pH of culture media. pH probe was calibrated with pH buffer between pH 4.0 and pH 10.0 (Fisher Scientific, UK) and washed with deionized water before use. Acellular media were incubated overnight before starting the experiment. The data were obtained by submerging the probe into the acellular culture media right after finishing the stimulation.

| Fluorometric H 2 O 2 assay
Fluorimetric Hydrogen Peroxide Assay Kit (Sigma-Aldrich, UK) was used in this study. The reagent preparation was carried out following the supplier's instructions. Acellular media samples were incubated overnight before applying ES, and the assay was performed immediately after stimulation. The fluorescence intensity of the samples was measured after 30 min of room-temperature incubation with the working solution at excitation wavelength of 544 nm and emission wavelength of 590 nm using microplate reader (FLUOstar ® OPTIMA, BMG Labtech). The  Table 1. However, their exact sequences are proprietary.

| Lipopolysaccharide (LPS) treatment
J774A.1 cells were exposed to the complete media containing LPS from Escherichia coli O111:B4 (100 ng/ml) on the following day after seeding. Tumor necrosis factor-α (Tnfα) mRNA expression was measured from the samples collected immediately after the first 2 hr of ES, LPS treatment, and LPS treatment with simultaneous ES as this time point was shown to exhibit the maximal level of Tnfα mRNA expression induced by LPS (Huang, Fletcher, Niu, Wang, & Yu, 2012).

| Statistical analysis
Data were collected from the samples which were stimulated by different pairs of electrodes and statistically analyzed using GraphPad Prism 7 software. The preliminary test for normal distribution was conducted using Shapiro-Wilk normality test. The data that passed the normal distribution test were analyzed using parametric approach. The statistical test details are described in each figure caption. p < .05 are considered statistical significant.
F I G U R E 1 Direct electrical stimulation device with L-shaped platinum electrodes in six-well plate arrangement used in this study. DC, direct current. The design is adapted and modified from (Leppik et al., 2019;Mobini et al., 2016) [Color figure can be viewed at wileyonlinelibrary.com] 3 | RESULTS Figure 2a shows the measured pH of the acellular media after 1 and 2 hr of ES. It is found that there was no significant change in pH of the media after stimulation (p > .96). The measured H 2 O 2 concentration of the media after stimulation is shown in Figure 2b. The average concentration of H 2 O 2 has increased by 3.6 and 5.4 µM after 1 and 2 hr of ES, respectively. Although the average H 2 O 2 concentration after 2 hr of ES are around 1.5 times higher than 1 hr, the difference was not statistically significant (p = .06).

| Cell metabolic activity
The metabolic activity of macrophages and preosteoblasts measured by resazurin assay after 3 days of stimulation are shown in Figure 3.
It is found that direct ES has significantly reduced the overall metabolic activity of both cell types. However, changes in cell metabolic activity of preosteoblasts were dependent on the stimulation time, whereas those of macrophages between 1 and 2-hr daily ES were not significantly different.

| Cell morphology and distribution
The optical images of two cell types after 3 days of 2-hr daily ES are shown in Figure 4. It appears that the cell population were noticeably lower at the area within 1,500 μm from the anodes and cathodes than in the middle area of the well plate after 3 days of stimulation.
This distribution pattern is observable in both macrophages and preosteoblasts. On the contrary, we do not observe any significant changes in cell morphology in the middle area after stimulation when compared with nonstimulated cells (control).

| mRNA expression from RT-qPCR
The expression of mRNAs from macrophages and preosteoblasts, which are responsible for translating Bmp2, Hmgb1, Vegfa, and Spp1 proteins, were evaluated in this study as well as the expression of Runx2, from preosteoblasts. It is found that Spp1 expression was upregulated after 3 days of 1-hr daily ES in both cell types, as shown in Figure 5. Moreover, an increase in Bmp2 expression has become significant when the cells were stimulated for 2 hr daily, and the level of Spp1 expression was still comparable to those stimulated for 1 hr daily. Likewise, the upregulation of Bmp2 expression was also consistent between two cell types.
However, the expression of Hmgb1 and Vegfa from these two cells were not affected by ES similarly to the Runx2 expression from preosteoblasts. Furthermore, Figure 6 shows that Tnfα mRNA expression from macrophages was reduced by ES, whereas it was increased by LPS after the first 2 hr of treatment. Besides, it is also found that the LPS-induced Tnfα expression could be mitigated by simultaneous ES.

| Effect of faradic by-products
The influence of faradic by-products as a whole was investigated by treating the cells with ES media, in which the presence of electric field, potential, and current were eliminated. The cellular responses after being exposed to the ES media in terms of the overall metabolic activity are shown in Figure 7. It can be seen that the faradic byproducts generated during the 2-hr ES has significantly reduced the metabolic activity of macrophages after 3 days of treatment, whereas that of preosteoblasts was not significantly changed (p = .06). Changes in mRNA expressions induced by faradic by-products are shown in Figure 8. It is seen that the mRNA expressions from two cell types were also significantly different from direct ES group. The results show that faradic by-products have significantly upregulated the Spp1 expression of macrophages, whilst having no effect on their F I G U R E 3 Cell metabolic activity of J774A.1 macrophages and MC3T3-E1 preosteoblasts after 3 days of 1 and 2-hr daily direct electrical stimulation (ES). Error bars represent standard deviation (n = 6). * and # represent p < .05 (one-way analysis of variance with Tukey's multiple comparisons test) when compared with the control (nonstimulated) and 1-hr daily stimulation groups, respectively is one of the ROS that could either be beneficial for cell signaling or inducing cell oxidative stress (Bartosz, 2009). It has been studied earlier that these by-products were the results from the radical generations through the reduction of water and oxygen during direct ES (Bodamyali et al., 1999;Kalbacova et al., 2007). Changes in extracellular environment due to these faradic by-products could affect cellular activities in terms of proliferation and differentiation (Galow et al., 2017;Li et al., 2009). It was shown previously that cathodic stimulation with titanium electrodes has reduced the metabolic activity of osteoblasts and macrophages and increased their intracellular ROS level (Kalbacova et al., 2007). In this study, we have further discussed this finding in terms of their pro-osteogenic mRNA expressions and also demonstrated that the effects of faradic by-products on cellular activities are not dominant compared with the effects of ES. Moreover, we also found that macrophages respond to ES in a similar manner as preosteoblasts, and it is not only faradic by-products that stimulate macrophage activities during electrically induced osteogenesis.

| Reduction in cell viability after stimulation
It was shown that pH changes after direct ES is dependent on the stimulation regime and excessive increase in pH may lead to the reduction in cell viability (Balint, Cassidy, Hidalgo-Bastida, & Cartmell, 2013). Although our results show that changes in pH are not significant during stimulation, the reduction in overall cell metabolic activity indicates that the population of metabolically active cells, which are the viable cells, is still reduced after being stimulated. This could be due to the other type of faradic byproducts. However, the optical images at different areas of the stimulated cells have shown that the apparent cell density at the vicinity of the electrodes is noticeably lower than those in the middle of the well. Besides, platinum electrodes have already been tested for its biocompatibility (Geninatti et al., 2015). Therefore, we believe that the excessive current density around the electrodes during stimulation has significant effects on cell viability and proliferation in our ES system (Balint, Cassidy, Hidalgo-Bastida et al., 2013).
On the contrary, cell metabolic activity after being exposed to ES media are significantly higher than those stimulated by direct ES. This confirms the negative effects of excessive current density on cell proliferation when the electrodes are present during direct ES. Moreover, it is also found that faradic by-products could also reduce macrophage proliferation. We initially thought that H 2 O 2 might be the reason behind this; however, it has been reported that cell oxidative stress would occur at H 2 O 2 concentration above 10 μM, which is higher than those generated from our ES system (Stone & Yang, 2006;X. Y. Wang et al., 2015). Moreover, the preosteoblastic and macrophage cell lines are  (Arai, Shibata, Pugdee, Abiko, & Ogata, 2007;Tang et al., 2007). Hence, it is possible that the reduction in macrophage proliferation could be due to other type of faradic by-products, which have not been characterized yet.

| Potential electrically stimulated osteogenesis mechanism
The RT-qPCR results have shown the similar changes in mRNA expression pattern between macrophages and preosteoblast as a result of ES. Spp1 expression has increased with 1-hr daily stimulation, whereas the expression of Bmp2 requires 2 hr of stimulation daily. It is understood that 2 hr daily are the optimal stimulation time for this system as it could induce the upregulation of both Bmp2 and Spp1.
It has been reported that Bmp2 is essential for the early stage of bone healing, which subsequently enhances differentiation of osteoprogenitor cells and their migration towards the injury site as well as bone formation (Knippenberg, Helder, Doulabi, Wuisman, & Klein-Nulend, 2006;Rickard, Sullivan, Shenker, Leboy, & Kazhdan, 1994;Tsuji et al., 2006;Wozney, 1992). Likewise, Spp1 was also found in the early stage of bone healing reportedly secreted by macrophages (McKee & Nanci, 1996;McKee, Pedraza, & Kaartinen, 2011). This Spp1 could also support the osteogenic differentiation of osteoprogenitor cells (Chen et al., 2014). However, the upregulation of Vegfa and Hmgb1 from macrophages were not observed, although it is suggested earlier in the literature as well as being hypothesized in this study. This could be due to the large difference between the H 2 O 2 concentration used in the literature (>100 μM) and those generated from ES (<10 μM) (M. Cho et al., 2001;Tang et al., 2007).
The expression of Vegf from preosteoblasts may also be dependent on ES regime. It has been shown that MC3T3-E1 preosteoblastic cells expressed higher level of Vegf after being stimulated by biphasic pulses, which is not observed when stimulated with constant DC in this study (Kim et al., 2006).
Furthermore, the level of Runx2 expression was not significantly changed after 3 days of stimulation, which indicates that there was no change in cellular activities related to the maturation of preosteoblasts during this period (Komori, 2010). This suggests that direct ES may not promote osteogenesis by inducing cellular osteogenic differentiation or osteoblast maturation directly during the early stage, but rather promote pro-osteogenic mRNA expressions from macrophages and preosteoblasts. In the case that these differential mRNA changes have been translated into their associated proteins, these proteins would then enhance bone regeneration from the native tissues or cells. It is expected that the upregulation of Bmp2 and Spp1 expressions would be one of the mechanisms behind the electrically stimulated osteogenesis.
The upregulation of Bmp2 expression after direct ES is also consistent with previous in vivo study. It was reported that the fractured bone tissue expresses Bmp2 after applying direct ES, and the expression of Vegf was not increased (Fredericks et al., 2007).
Furthermore, Bmp2 expression could also be enhanced by capacitive ES technique in both in vitro and in vivo (Gan et al., 2004;Griffin et al., 2013;Z. Y. Wang et al., 2006) as well as inductive ES in vitro (Bodamyali et al., 1998). It was previously thought that Bmp2 was produced by osteoblasts as a result of in vivo ES (Griffin & Bayat, 2011). However, it is found that Bmp2 and Spp1 are also expressed from electrically stimulated mesenchymal stem cells as well as macrophages observed in this study (Leppik et al., 2018;Mobini et al., 2016;Mobini, Leppik, Parameswaran, & Barker, 2017).
Interestingly, this study shows that Bmp2 and Spp1 mRNA expressions are primarily a result of ES, whereas faradic by-products have partially contributed to an increase in Spp1 expression from macrophages, potentially involving 4EPB-1 translation or nuclear factor-κB and AP-1 transcription (Lyle et al., 2014). Apart from pH and H 2 O 2 , faradic by-products also include H 2 generation, O 2 reduction, and any other surrounding molecules generated by redox F I G U R E 8 Messenger RNA (mRNA) expression of J774A.1 macrophages and MC3T3-E1 preosteoblasts after 3 days of treatment with 2-hr electrically stimulated media (ES media). Error bars represent upper and lower 95% confidence limits (n = 6). * and # represent p < .05 (unpaired two-tailed Student's t test) when compared with the control (nonstimulated media) and 2-hr daily direct ES groups, respectively reactions during ES (Bodamyali et al., 1999;Brighton et al., 1975;Brummer, Mchardy, & Turner, 1977;Merrill, Bikson, & Jefferys, 2005). These changes could affect cellular activities, and they have not been fully characterized yet. Therefore, we are unable to confirm whether it was H 2 O 2 alone, or other faradic by-products, or both, which triggered this response based on our current results. Despite the discussion that faradic by-products could possibly be behind the upregulation of Bmp2 mRNA expression after ES, we clearly demonstrate that the increased Bmp2 mRNA expression is more likely a result of ES than the by-products (Fredericks et al., 2007;Gan & Glazer, 2006). It would also be worth investigating further in details regarding how ES interacted with cells, and whether or not it is the interaction with voltage-sensitive ion channels as discussed in the literature that triggers these mRNA expressions from macrophages and preosteoblasts Thrivikraman, Boda, & Basu, 2018).

| Responses of macrophages to direct ES
Macrophages are well known for triggering inflammatory responses; however, they could also be beneficial for wound healing and tissue regeneration depending on their phenotypes. The proinflammatory phenotype is M1 macrophages and wound-healing phenotype is M2, and the differentiation towards these two phenotypes is dependent on their microenvironment and stimuli (Murray, 2017;Rőszer, 2015; Y. Zhang et al., 2013). It is understandable that the increased Spp1 mRNA expression from preosteoblasts is the marker for their activity along the osteoblastic lineage (Rutkovskiy, Stensløkken, & Vaage, 2016). However, Spp1 is also involved in the function, migration, and differentiation of macrophages (Bruemmer et al., 2003;Lund, Giachelli, & Scatena, 2009;Nyström, Dunér, & Hultgårdh-Nilsson, 2007). Hence, the upregulation of Spp1 expression from macrophages could have diverse interpretation (Lund et al., 2009). Apart from its roles in facilitating soft tissue and bone remodeling, Spp1 expression could also be a sign of macrophage maturation and inflammation, which can lead to the adverse effects on osteogenesis (Gilbert et al., 2000;Krause et al., 1996;Liaw et al., 1998;Lund et al., 2009;Ogawa et al., 2005;Rittling et al., 1998;Saleh, Carles-Carner, & Bryant, 2018;Zhao et al., 2011).
It is found from this study that direct ES reduced Tnfα mRNA expression from macrophages, which is a proinflammatory and anti-M2 macrophage marker that could be triggered by LPS (Huang et al., 2012;Murray, 2017). In addition, it has also been shown previously that the Bmp2 mRNA expression from J774A.1 cells was not from proinflammatory activities (Champagne, Takebe, Offenbacher, & Cooper, 2002). Therefore, it is plausible that the increased Spp1 and Bmp2 mRNA expressions after direct ES were not indicative of the proinflammatory responses. Nonetheless, it could still be asked whether the status of macrophages remained unchanged or polarized towards M2 as the recent study has suggested that the increased Spp1 mRNA expression is also a characteristic of the M2c subtype of M2 macrophages, including J774A.1 cell lines (Capote et al., 2016).
In addition to the findings from this study, the recent in vivo study also reports that ES has significantly reduced the inflammation during bone injury, which is consistent with the conclusion from an analysis of clinical studies suggesting that ES is effective in reducing pain and radiographic nonunion-healing rate (Aleem et al., 2016;Fonseca et al., 2018). Moreover, the faradic by-products generated during direct ES also have potential antibacterial effects which could be capable of reducing the risk of infection (Asadi & Torkaman, 2014). Hence, it would be worth investigating further on the antiinflammatory effects of ES and to determine whether or not ES could induce macrophage polarization towards any proregenerative M2 subtypes in future.

| Further optimization of direct ES system and the limitations of this study
It is seen from this study that although the majority of cellular responses are induced by ES, the faradic by-products also have an effect on cells. Therefore, further characterization of the ES media may be requisite to explore the unidentified faradic by-products and their effects on cellular activities, such as the radicals present during the process of H 2 O 2 generation (Kalbacova et al., 2007).
The insulation of the electrodes could be implemented in future to deconstruct the effects of direct ES and faradic by-product generation on cellular activities. At the same time, the current density near the electrodes needs to be minimized. It was suggested that the optimal current density for promoting osteogenesis using platinum electrodes is between 1 and 5 A/m 2 , and exceeding this range would result in tissue or cell necrosis (Spadaro & Becker, 1979). Otherwise, novel techniques could be developed and implemented to prevent cell penetration into the invasive region, such as surface modification. The excessive electrolysis of culture media should also be avoided during the stimulation as it can cause adverse effects and cytotoxic complications (Thrivikraman et al., 2018). Moreover, it is worth investigating the electrically stimulated cells further, in terms of their protein secretion and interactions with other osteoprogenitor cells, as well as the variation in cellular activities between tumor-associated macrophage cell lines used in this study and primary bone marrow macrophages present at bone injury sites (Chamberlain, Godek, Gonzalez-Juarrero, & Grainger, 2009;Murray, 2017;Saleh et al., 2018). The incorporation of this direct ES system with conductive scaffolds could also be a promising approach as the recent studies show that it could have significant improvement in promoting cell proliferation and/or osteogenic activities (Hu, Chen, Tsao, & Cheng, 2019;J. Y. Zhang, Li, Kang, & Neoh, 2016;S. Zhu et al., 2017).

| CONCLUSION
ES could promote bone regeneration by inducing Bmp2 and Spp1 mRNA expressions from macrophages and preosteoblasts.
Preosteoblasts did not respond to faradic by-products in terms of Bmp2 and Spp1 mRNA expressions, whereas macrophages responded by increasing their Spp1 expression. We demonstrated that the roles of faradic by-products are not dominant, and the byproducts alone would be less effective in promoting bone formation without the presence of ES. The findings from this study also imply that cellular responses from preosteoblasts and macrophages to ES are predominantly triggered by the mechanism involving electric field, potential, and/or current. On the contrary, Vegfa and Hmgb1 mRNA expressions from both types of cells and Runx2 expression from preosteoblasts are not affected by the ES regime used in this study. In addition, we showed that this ES system may need further optimization to reduce the current density near the electrodes, which locally impairs cell viability and proliferation.

ACKNOWLEDGMENTS
This study is a continued work from the project that was previously funded by Biotechnology and Biological Sciences Research Council (BBSRC) grant (BB/M013545/1). It also receives funding from The Royal Thai Government scholarship (ST 4729).
The authors would like to thank Dr Elizabeth Mann for the discussion on the characterization of macrophage status and Dr Ruikang Xue for the fluorescence imaging and quantification assistance.

CONFLICT OF INTERESTS
The authors declare that there are no conflict of interests.