Bioglass could increase cell membrane fluidity with ion products to develop its bioactivity

Abstract Objectives Silicate bioactive glass (BG) has been widely demonstrated to stimulate both of the hard and soft tissue regeneration, in which ion products released from BG play important roles. However, the mechanism by which ion products act on cells on cells is unclear. Materials and methods Human umbilical vein endothelial cells and human bone marrow stromal cells were used in this study. Fluorescence recovery after photobleaching and generalized polarization was used to characterize changes in cell membrane fluidity. Migration, differentiation and apoptosis experiments were carried out. RNA and protein chip were detected. The signal cascade is simulated to evaluate the effect of increased cell membrane fluidity on signal transduction. Results We have demonstrated that ion products released from BG could effectively enhance cell membrane fluidity in a direct and physical way, and Si ions may play a major role. Bioactivities of BG ion products on cells, such as migration and differentiation, were regulated by membrane fluidity. Furthermore, we have proved that BG ion products could promote apoptosis of injured cells based on our conclusion that BG ion products increased membrane fluidity. Conclusions This study proved that BG ion products could develop its bioactivity on cells by directly enhancing cell membrane fluidity and subsequently affected cell behaviours, which may provide an explanation for the general bioactivities of silicate material.

product, has not been fully studied. It has been widely reported that Si plays critical roles in the biological effects of BG as it is the only component with different concentrations between the culture medium containing ion products of BG and the normal cell culture medium. 1,8,13,14 But how Si in solution can produce biological effects remains unclear. Although several studies attributed bioactivity of BG to signalling pathways, 13,15 these conclusions could hardly explain the general biological activities of BG in different types of cells. It is speculated that BG ion products should work in a more primitive way beyond specific signal transduction pathways.
The cell membrane, as the interface between intra-and extra-cellular environments, is the site to receive much of the external information from those including hormone, cytokine, drugs and implanted biomaterials. 16 Cell membrane fluidity is the basis of cell membrane biological activity. Cells actively change the fluidity of membrane to adapt to different environments and signal transduction needs. 17 Increased cell membrane fluidity is often coupled with enhanced cell migration, proliferation and differentiation. 18,19 Furthermore, formation of membrane microdomains, such as lipid rafts and caveolae that play the role of signalling platforms, is also highly related to membrane fluidity. 20 Membranes, especially changes of membrane fluidity, can produce a wide range of biological effects and are not limited to specific cell types. Thus, we hypothesize that the cell membrane may be one of the main sites where Si exerts its biological activities and Si can optimize the dynamic structure of membrane by affecting the cell membrane fluidity and subsequently enhance cellular signal responses, which contributes to the general biological activities of silicate biomaterials.
Based on these assumptions, herein we investigated the effects of BG ion products released from 45S5 BG on cell membrane fluidity. Fluorescence recovery after photobleaching (FRAP) technology and membrane generalized polarization assay (GP) were used to detect the effects of Si on cell membrane fluidity.
By specifically reducing cell membrane fluidity, bioactivity of BG ion products was eliminated. Meanwhile, for injured cells, BG ion products could promote phosphatidylserine flipping of cells and induce the apoptosis of the injured cells. Finally, we simulated changes in signal transduction pathways after the membrane fluidity was increased, indicating that Si had a high impact on overall signal output for cell signal transduction. Overall, we believe that the biological activity of BG ion products may be attributed to the promotion of cell membrane fluidity. This study can not only interpret existing bioactivities of BG materials, but also provide an important theory for expanding the application of BG materials in tissue engineering.

| Cell isolation and culture
Human umbilical vein endothelial cells (HUVECs) were isolated from human umbilical cord veins using the method reported by Bordenave et al 21 The isolation of HUVEC was approved by Institutional Review Only the HUVECs and HBMSCs of early passages (passage 2-7) were taken for the subsequent experiments. HBMSCs were seeded at a density of 2 × 10 5 cells per well, and HUVECs were seeded at a density of 3 × 10 5 cells in 6-well plates. The seeding density of cells was a quarter of the above-mentioned values in the FRAP experiment and one half in the GP experiment.

| BG ion extracts
BG (45S5 glass) powders with diameter of 5-20 µm were purchased from Kunshan Chinese Technology New Materials Co., Ltd (Kunshan, China). BG ion extracts were prepared according to the methods reported in literatures adapted from ISO10993-1 procedures. 22 Briefly, 1 g of BG powders was soaked in 5 mL of serum-free ECM and MSCM, respectively. After being incubated for 24 hours in a humidified incubator with 5% CO 2 at 37°C, the supernatant was then collected and sterilized through a filter (Billerica, MA, USA, Millipore, 0.22 µm). For further use, BG ion extracts were diluted with total ECM (endothelial cell basal medium + 5% foetal bovine serum [FBS] + 1% endothelial cell growth supplement + 1% penicillin-streptomycin [P/S]) and total MSCM (MSCM + 10% FBS + 1% P/S + 1% l-glutamine), respectively. The concentrations of Ca, Si and P in the diluted ion extracts were detected by inductively coupled plasma atomic emission spectroscopy (Optima 3000DV; PerkinElmer, San Francisco, CA, USA), and the experiment was repeated three times.

| Fluorescence recovery after photobleaching
After the cells were attached on confocal dishes (Shanghai Pumai Biotechnology Co., Ltd, Shanghai, China), lipid dye Laurdan (Thermo Scientific, Waltham, MA, USA) was added to cell culture medium and the cells were cultured for another 3 hours (2 µg/ mL). Before FRAP experiments, the Laurdan was washed off with warm PBS (37°C). During the FRAP experiments, an area in the cell membrane for bleaching was selected and laser at a power of 100% generated from a confocal scanning laser microscope was used to bleach the area. After that, the images of the bleached area at different time points were recorded. All the experiments applied the same FRAP setting: 2 frames for pre-bleach acquisition, 3 s for bleaching and 2 s per frame for 50 times as postbleach acquisition. Images were captured, and bleaching process was monitored with TCS SP8 STED 3X system (Leica, Wetzlar, Germany). The fluorescence intensity of bleaching area was analysed, and the data were normalized using a method reported in a previous publication. 23

| Measurement and calculation of GP based on laurdan
Fluorescence spectra of Laurdan are sensitive to the physical state of the surrounding phospholipids gel-to-liquid crystalline phase transition, which is associated with the fluidity of cell membranes.
The maximum emission peak of Laurdan normally has 50 nm red shift under the increase of membrane fluidity. 24 After the Lauran in different mixture was excited at 405 nm, the intensities of Laurdan emission at 400-460 nm (I blue ) and 470-530 nm (I red ) were measured.
Polarization is calculated according to this equation: , and GP is the mean value of total polarization. Since GP value is a ratio, it is independent of the local Laurdan concentration. On a scale ranging from −1 to +1, higher GP values correspond to lower membrane fluidity and vice versa. In one experiment, the gain set and offset were consistent such that different groups were comparable.
For the GP measurement, cells were cultured on confocal dishes.
After the cells were attached, they were cultured with experimental medium. Laurdan was added to culture medium, and cells were incubated with Laurdan for another 3 hours (2 µg/mL) and then washed off before experiment. Fluorescence images of two channels of 400-460 nm and 470-530 nm were taken using a confocal microscope (TCS SP5 II; Leica). MATLAB was used to calculate the GP value.
Briefly, the captured images were first separated into 4 × 4 PX pieces, and these pieces of I 440 and I 490 were matched by location, respectively. The mean fluorescence intensity per area was used to calculate polarization, and polarization value of each pair of pieces was calculated according to the above equations. Finally, the GP value of one measured area was the mean value of the polarization of each pair of pieces on this area.

| Liposome and fluorescence emission spectrum detection
Liposomes were purchased from Xian Ruixi Biological Technology

| Assessment of cell migration ability
Wound healing assay and Transwell assay were used to evaluate the migration ability of HUVECs as previously described. 25 Specific inhibitor of cell membrane fluidity Cholest-5-en-3-ol (3β)-, 3-(hydrogen butanedioate) (CHS) was purchased from Sigma, St. Louis, MO, USA.

| Evaluation of osteogenic potential of HBMSCs
Human bone marrow mesenchyme stem cells were seeded at 12well culture plates and grew to confluence with MSCM medium.
Then, the MSCM medium was removed, and osteogenic induc-

| Quantitative real-time polymerase chain reaction
Quantitative real-time polymerase chain reaction was conducted as previously described. 27  (reverse).

| Evaluation of apoptosis in HUVECs
Healthy HUVECs were seeded at six-well plates and grew to 70% confluence before the culture media were changed into normal are not A (absent); (b) genetic fold change was more than twice.
Selected genes were commented according to GEO, and signalling pathway cluster was done by KEGG.

| Protein phosphorylation and signalling pathway screening
The protein expression and phosphorylation data were gener-

| Simulation of the relationship between cell membrane fluidity and signal transduction
Monte Carlo method is known as the statistical simulation method and is more convenient for discussing the effects of protein movement rate on signal transduction compared with differential model.
The classical receptor activation model was chosen for simulation (see Supporting Information, Sup_code). In this model, receptors activated by ligands to form first protein complexes (T1), and the complexes recruit scaffold proteins to form second complexes (T2).
Then, the second complexes recruit functional proteins and form third functional complexes (T3) which were able to output biological signals. The third complexes can be depolymerized after collision with inhibitory proteins and each components would re-engage to signal transduction. For simulation, membrane proteins were regarded as volumetric two-dimensional particles and protein interactions were represented as particles. The motion of particles obeys the rules of two-dimensional Brownian motion. The parameters, such as particle density, motion rate, and interaction distance, were employed based on published data. 28,29 The simulation program was written on MATLAB 2016a.

| Statistical analysis
The data were expressed as means ± standard deviation. Statistical significance between groups was calculated using two-tailed analysis of variance (ANOVA) and performed with a Student's t test programme. Statistical significance between two groups was performed with a Student's t test programme, and the differences were considered significant when P < .05 (*) or P < .01 (**). Hsp70 gene expression analysis uses two-way ANOVA for difference analysis.

| Si in BG ion extracts could increase cell membrane fluidity
Based on the previous research, 30   was 64.8%, 67.4% and 75.1%, respectively, which were significantly higher than that with control medium (58.5%) (P < .01 for all concentrations  Figure 1D.
According to the polarization value density histograms, treating HBMSCs with Si 1/128 showed the similar effects on the cell membrane polarization to incubating cells at 40°C for 30 minutes, since the cell membrane polarization values in these two groups were much closer to −1 than that in the control group. The GP value of HBMSCs and HUVECs calculated from the whole polarization value is shown in Figure 1E,F, respectively. Specifically, the GP value of . Similarly, when we use pure silicate materials (CaSiO 3 ) to treat cells, we also find that the fluidity of the cell membrane is increased (see Figure S1). This result may suggest that Si is the main element that affects the fluidity of the cell membrane. We also found that the pure calcium silicate ion extracts were not as effective as the BG material ion extracts, and it may attribute to the synergistic effect of other ions, like P.

| Si in BG ion extracts directly and physically interact with the cell membrane
Now that we had confirmed Si could increase the membrane fluidity, and the new question was in which way Si could affect the membrane fluidity: altering membrane lipid component through lipid metabolism pathway or directly interacting with membrane lipids? 31 For the first hypothesis, Si needs to enter the cells and regulate the expression of metabolism-related enzymes, which will usually take more than a few hours. On the contrary, by directly interacting with   Si-containing medium for 24 hours, the medium was replaced with control medium (Time 0, Si-remove part in Figure 2A,B). For both of HUVECs and HBMSCs, after the Si was removed for 1 hour, the cell membrane GP value sharply increased and finally recovered to the same level as that in the control group. Then, after the cells rested for several hours, they were cultured with Si-containing medium again and the GP value was detected in the following 1 hour (S-add part in Figure 2A,B). Obviously, the cell membrane GP value decreased again. Si was added again, the fluorescence recovery rate was 58.4% at 0 minute, 63.4% at 30 minutes and 74.5% at 60 minutes (n = 9), respectively ( Figure 2C; Figure S2). Similarly, for HUVECs, after Si was added, the fluorescence recovery rate was 56.4% at 0 minute, 65.3% at 30 minutes and 73.4% at 60 minutes (n = 9), respectively ( Figure 2D; Figure S3).
The GP value results, together with these FRAP results, sug-

| Membrane fluidity sensor Hsp70s was highly expressed after stimulation by BG ion extracts
As we had demonstrated that BG ion extracts effectively enhanced

| Bioactivity of Si in BG ion extracts was highly related to cell membrane fluidity
To investigate whether biological activities of Si were related to the cell membrane fluidity, cholesterol monoester succinate (CHS) was used to specifically reduce the cell membrane fluidity. As ion extract of BG has been widely reported to promote HUVECs migration and  Wound healing assay and Transwell assay were used to evaluate migration ability of HUVECs cultured with medium of NC, Si, CHS or Si + CHS ( Figure 4C). The quantitative results of migration ability of HUVECs were based on wound healing assay ( Figure 4B).  Figure 4E).
ALP staining on the HBMSCs showed the same results as relative gene expression ( Figure 4F). Taken together, CHS treatment restricted the cell membrane fluidity even in the presence of Si; thus, the stimulatory effects of Si on the migration and differentiation abilities of cells were also eliminated. These results suggested that the stimulatory effects of Si on cell behaviours were highly related to cell membrane fluidity.

| Si in BG ion extracts could promote early apoptosis of injured cells
Cell membrane not only works as the site of signal transduction, but also delivers bio-signals by itself. Phosphatidylserine (PtdSer) flipping from the intracellular leaflet to the outer leaflet was an important starting signal for apoptosis, relaying "eat me" messages to macrophages. As lipid turnover is a form of lipid movement on membrane and is directly related to cell membrane fluidity, we hy-  Figure 5A,B, respectively. Apoptosis ratio was 11.8%, 14.7% and 16.0% in healthy HUVECs treated with Si 1/256, Si 1/128 and Si 1/64, respectively, while that in HUVECs treated with PBS was 9.4%.
Statistical analysis indicated that, as compared to HUVECs treated with PBS, only Si 1/64 increased apoptosis of HUVECs (P < .05) and Si 1/256, Si 1/128 had no significant influence on the apoptosis of normal cells.
Then, the effects of Si on the apoptosis of injured cells were investigated. HUVECs treated with Si (Si 1/128) or PBS (NC) were induced for apoptosis by either UV, hydroxyurea or TSZ kit (shown in Section 2.9). The flow cytometry as well as apoptosis histogram of HUVECs are shown in Figure 5C,

| Microarray data showed that Si in BG ion extracts did not activate specific signal pathways
Gene expression and protein activity of HBMSCs treated with Si from BG ion extracts were analysed by microarray chips. As shown in Figure 6A, various genes in HBMSCs had significant differential expression after stimulation with Si in BG ion extracts (represented as class-3), which was consistent with the broad bioactivities of Si and BG material. As we sorted genes based on cell location of gene products ( Figure 6B), membrane-related genes, such as ion channels, cell adhesion, bracket of signals, G protein and cytokine receptor, were greatly impacted (n = 557, 45.8%), and the transcription-related genes were also altered (n = 420, 33.4%).
Signalling pathways clustered by KEGG are shown in Figure 6C.
These signals were mostly enriched on membrane, such as G protein signal, ion signal, cell adhesion and cytokines signal. Excluding several common signalling pathways, there was no significant difference among these functional signal pathways according to the false discovery rate (FDR) value. Microarray data of protein activity were also clustered by KEGG ( Figure 6D). Several common pathways such as PI3K-Akt signalling pathway and MAPK signalling pathway were enhanced, while other functional pathways showed an average count score according to KEGG cluster. Based on these results, Si treatment could cause a variety of biological activities on HBMSCs as many common signals were activated, while this promotion did not target specific biological process since no functional signal was activated. In addition, in both gene expression and protein activity data, none of metabolism or lipid-related pathway was significantly changed, which also indicated that Si directly interacted with cell membranes rather than regulating the metabolic pathways in cells.

| Simulation of cell membrane fluidity and signal transduction
We have verified the activation of biological signals as caused by changes in cell membrane fluidity, yet it is still not enough to explain Then, the signal system responded to step signal with increased membrane fluidity was investigated ( Figure 7C).
Step signal was given at 100-200 ms and 400-500 ms After the first step when signal appeared at 100 ms, superfluid group responded quicker than normal group ( Figure 7C,D), which was proved by a rapid formation of final protein complexes (T3). Furthermore, the F I G U R E 6 Results obtained from Affymetrix gene expression chip and PEX100 protein activation chip analysis on the HBMSCs cultured with or without Si. A, The scatter plot of gene expression (X axis for Si 1/128 treatment, Y axis for NC). Differentially expressed genes were shown as class-3. B, Cellular component distributions of differentially expressed genes according to Gene Ontology. 45.8% of these genes were membrane localized and 33.4% were nucleus localized. Other component were 20.8% only. C, KEGG cluster analysis according to gene expression data. D, KEGG cluster analysis according to protein activations chip. These cluster data showed that numerous base signal pathways were activated, including G protein signal pathways, cytokine pathways, ion channels, PI3K-IP3 pathways. HBMSC, human bone marrow mesenchyme stem cell; NC, negative control F I G U R E 7 Simulations of the relationship between membrane fluidity and signal transductions. A,B, Effects of membrane fluidity on signal system with resting-state signal, representing the effects of Si on cells for a long term.
For cascade system structures, as T1, T2, T3, increased membrane fluidity had no significant influence on protein activation ratio, which were waved around. However, increased membrane fluidity significantly increased signal output strength, which indicated the long-term benefit of enhanced membrane fluidity on the signal system. C,D, System responded with step signal. Reactions of signal activations and recovery were more quickly in superfluid group than the normal group. In addition, the secondary response of superfluid group was better than that of the normal group time point of maximum growth rate of output signal strength was 187 ms for superfluid group and 217 ms for normal group, and maximum growth rate of output signal strength was 0.74 for superfluid group and 0.34 for normal group, both suggesting that superfluid group had a better signal response than normal group.
In addition, when the second step signal was input at 400 ms, superfluid group also showed a better secondary response and more completed signal waveform than normal group according to T3 wave. Based on these results, it could be concluded that Si had a low impact on protein level but had a high impact on overall signal output for cell signal transduction. The other biological effects of BG ion products on cells, such as pH, may also be considered from the perspective of cell membranes.

| D ISCUSS I ON
Increased membrane fluidity could directly accelerate the motion of membrane protein and lipids, which facilitates the enzymatic reaction on the membrane. 42 Previous studies have shown that membrane fluidity is highly related to migration and differentiation of cells. 43 Once the cell membrane was frozen by CHS, the stimulatory effects of Si on cells migration and differentiation were totally eliminated. If the bioactivity of BG ion products was not related to the cell membrane fluidity, migration and differentiation of culturing cells with a mixture of Si and CHS should not be suppressed to the same level as cells simply cultured with CHS. Thus, cell membrane fluidity played an important role in the stimulatory effects of BG ion products on the cell behaviours. In addition to stimulating the healthy cells to migrate and differentiate, during tissue repairing, removal of injured cells is also important for reducing the diffusion of "death signals" from these injured cells. 44 In this study, BG ion products could promote apoptosis of injured cells, especially the early apoptosis reflected by PtdSer eversion which is an important apoptosis-inducing signal and an enzymatic reaction on the membrane. Taken together, Si can not only stimulate the migration and differentiation of healthy cells to but also enhance the apoptosis of injured cells by increasing the cell membrane fluidity, which both are beneficial to stimulating tissue regeneration.
Although our study mainly concerns about the effects of BG ion products on cell membrane fluidity, the interactions between Si and cell membrane will inevitably lead to changes in membrane dynamic feature, 45 which can significantly affect cell signalling. In this study, many gene products and signalling pathways were up-regulated and activated by Si, including G protein, hormone and cytokine receptor, scaffold protein, ion channel together with PI3K-Akt, focal adhesion signal pathway which are all membrane proteins and functionally relate to membrane dynamic feature. For example, amplification of G protein cascade signal specificity occurs in phosphatidylethanolamine formed non-lamellar phase propensity 46 ; voltage-gated ionic channels are sensitive to the lipid environment. 31 These lipid domains, such as lipid rafts, caveolae, synaptosomes, receptor clusters, not only enrich membrane protein to particular area, but also regulate signal transduction with lipid environment. 31 The interaction between Si and membrane may greatly affect the decomposition or recombination of lipid domains and further influence microdomain-mediated signals transduction.
Simulation models show that enhanced cell membrane fluidity has great impact on cell signal response, with regard to cell signal strength and signal response capabilities. By simply regulating the overall structure of the cell membrane, including membrane microdomain structure, the pattern of signal transduction and drug sensitivity in disease can be altered. 31,42 Lipid modification could also help drug delivery to correct the abnormality of signal transduction. 42,47 As we have proved that Si could effectively enhance cell membrane fluidity and silicate biomaterials that can release Si in biological conditions have been widely used in tissue regeneration, it is very promising to apply Si and silicate biomaterials in membrane lipid therapy for chronic diseases.
In conclusion, for the first time, we have confirmed that BG ion products could effectively enhance cell membrane fluidity and the interactions between BG ion products and cell membrane was direct and physical. The stimulatory effects of BG ion products on migration and differentiation of health cells and on apoptosis of injured cells were dependent on the increased of cell membrane fluidity. As BG ion products enhance cell membrane fluidity and the subsequent cell signal transduction in a physical way, applying silicate biomaterials in clinical applications, including tissue regeneration and membrane lipid therapy for different diseases, may be a feasible and efficient strategy.

ACK N OWLED G EM ENTS
This study was supported by grants from National Natural Science Foundation of China (81773115 to W. Xia, 31771024 and 31971274 to H. Li), and SJTU funding (YG2017MS52, to W. Xia).

CO N FLI C T O F I NTE R E S T
The authors declare no competing interests.

AUTH O R S ' CO NTR I B UTI O N S
Longxin Yan, Weiliang Xia and Haiyan Li conceived the project.
Longxin Yan performed experiment and analysed the data. Longxin Yan, Haiyan Li and Weiliang Xia wrote the manuscript.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.