3.1 The results of model parameter optimization
The small intestinal epithelial cell membrane model preliminarily established in this paper is shown in Fig.3. In order to better predict the intestinal absorption of drugs, the influence of temperature, water layer thickness, and ionic strength on the small intestinal epithelial cell model was investigated in this paper to select the optimal environmental parameters.
3.1.1 Effects of temperature on membrane properties
The changes of APL of AP and BL with temperature are shown in Fig.4 and Fig. 5. It can be seen from the figure that in the temperature range of 230-390 K, the influence of temperature on both sides of the membrane is roughly the same, and the APL increases with the increase of temperature. For the heating process, there are two sharp points ( Fig.4a and Fig.5a). A smaller mutation point is between 280-290K, before which APL is small and almost unchanged, indicating that the lipid head groups are well arranged and tightly packed, and the membrane is in a complete gel state. While APL increased significantly after this mutation point, indicating that the arrangement of lipid head groups was looser than before, which was speculated to be related to the pre-phase transition of the cell membrane. The other mutation point was between 310-330K. After this mutation point, APL increased rapidly and did not appear saturation point with the increase of temperature. This indicates that with the increase of temperature, the thermal energy destroys the interaction between the lipid head groups, which increases the free space of the lipid head groups, leading to the complete transition to the liquid state. For the cooling process, APL decreases uniformly with the decrease of temperature, and there is no obvious mutation point on both sides of the membrane with the decrease of temperature. ( Fig.4b and Fig.5b).
The membrane thickness of AP and BL varies with temperature, as shown in Fig. 6. In the temperature range of 230-390 K, the influence of temperature on both sides of the membrane is roughly the same, and the membrane thickness decreases with the increase of temperature. Moreover, for the heating process, there are corresponding turning points around the two temperatures at which the phase transition occurs, which is consistent with the results of APL. For the cooling process, there is no obvious turning point except that the AP film has a big fluctuation at 250K, which is also roughly consistent with the results of APL without mutation during the cooling process. In addition, to further explain the influence of temperature on the membrane morphology, AP membrane was taken as an example to analyze the lipid molecular arrangement at three temperature points where the thickness of the membrane was significantly different during the heating process (Fig. 7). The results show that the higher the temperature is, the wider the peak shape of the density distribution map of the lipid head-based beads along the Z-axis is, which indicates that the higher the temperature, the more disordered the lipid molecules are arranged. On the contrary, the lower the temperature, the narrower the peak shape of the density distribution map along the Z-axis of the lipid head-based beads, indicating the more orderly the arrangement and the larger the peak distance between the two peaks, indicating the larger the film thickness. This is also consistent with the results of changes in APL. The increase of temperature would destroy the interaction force between the lipid head-based beads, making them move more freely and arrange more disorderly.
The morphology of AP and BL varies with temperature, as shown in Fig. 8. As can be seen from the figure, at 250K and 280K, the lipid membrane is tightly arranged and orderly, and the membrane thickness is basically unchanged, so the membrane is in a complete gel state at this time. At 310K, the arrangement of beads was obviously looser than before, indicating that a pre-phase transition occurred at some time point between 280-310K, and the film had already transited from gel state to liquid state, but the film still maintained a good film shape without severe deformation. When the temperature reached 340K, the lipid bilayer began to bend and deform, and with the increase of temperature, the film morphology became more curly until the deformation was severe and no longer existed in the form of the bilayer, indicating that a phase transition occurred at some time point between 310K-340K. The above changes in membrane morphology with temperature more intuitively prove the previous results of APL and membrane thickness.
The above results indicated that the pre-phase transformation of the small intestine epithelial cell model was between 280K and 290K, and the phase transformation was between 310K and 330K, which was consistent with the reported cell membrane phase transformation temperature near the physiological temperature[24-26]. Moreover, it is worth noting that when the temperature is maintained between 290 K and 310 K, the gelatinous state of the membrane coexists with the liquid state, and the lipid bilayer of the membrane is in good shape without serious bending deformation, which is close to the real state of the membrane.
3.1.2 Effects of water layer thickness and ionic strength on membrane properties
The results of APL and membrane thickness changing with water layer thickness and ionic strength are shown in Fig. 9. When the thickness of the water layer is 22.5Å, APL is 0.46 nm2, and membrane thickness is 4.1nm. With the thickening of the water layer, the APL increases, and the membrane thickness also becomes thicker. This indicates that increased hydration causes more water molecules to contact with the lipid head region, which reduces the compactness of lipid molecules horizontally, thus increasing APL, while the entry of excess water molecules also increases the membrane thickness when the ionic strength increases or decreases, the changes of APL and membrane thickness tend to be stable, indicating that the ionic strength has little effect on the membrane morphology in the model built in this paper.
In conclusion, when the temperature is maintained between 290 K and 310 K, the APL of the membrane constructed in this paper is stable at about 0.46-0.47 nm. At this time, the lipid bilayer of the membrane is in good shape, and the membrane is neither in absolute gel state nor will there be severe bending deformation, but close to the real membrane state, which will be conducive to the follow-up study of drug permeation membrane absorption. Therefore, the simulated temperature we selected was 310 K, which was also consistent with the temperature of the cell membrane under physiological state. The water layer thickness was 22.5Å, and the ionic strength was 0.15mol, which was consistent with the relevant literature reports.
3.2Predictive characterization of drug absorption through the membrane
In this section, three model drugs QUE, EPH, and BAI, were used to simulate the complete transmembrane process of the drug by free diffusion and umbrella sampling in the optimized model.
Free diffusion simulation, in which the drugs enter the cell membrane through free diffusion, can be used to reflect the difficulty of drug penetration preliminarily. Considering the interaction and absorption process between drugs and cell membranes in the physiological environment, the longitudinal drug diffusion along the Z-axis is closer to the real drug permeation process. While the transverse drug diffusion along the X-axis mainly indicates that the drug has different degrees of disturbance to the lateral distribution of lipid molecules. Therefore, the MSD and diffusion coefficient D of the drug along the Z-axis were calculated respectively to characterize the free diffusion velocity of the drug, and the results were shown in Fig.10 and 11. According to the figures, although the proportion of lipid molecules on both sides of the membrane was different, the longitudinal diffusion coefficient D of each model drug also had some differences, but the overall diffusion trend remained the same, and the diffusion coefficient D was ordered as EPH > QUE > BAI. The results showed that the changes of lipid membrane components in the selected lipid components in this study had little effect on the drug diffusion rate, and the order of three drugs permeating the membrane reflected by free diffusion simulation was EPH > QUE > BAI, which was consistent with the results of cell experiment.
Umbrella sampling simulation, in which the drug passes through the intact cell membrane under the action of the traction force, can be used to reflect the drug penetration capacity. In the simulation, the free energy potential surface(Fig. 12) of each model drug through the membrane was obtained, and △G of each model drug through the membrane was calculated by using PMF graph data (Table 2).
It can be seen from the table that the △G of each model drug through the membrane of both sides is different to some extent, but the same trend on both membranes of the △G reflected the same tendency of membrane permeability of each model drug, indicating that the change of lipid membrane composition in the selected lipid components in this study had little effect on the membrane permeability of the drug. In addition, it can also be seen from the table that the △G of each model drug is greatly different, and the descending order is EPH, QUE, and BAI, indicating that the membrane permeability of these three model drugs is EPH > QUE> BAI, which is also consistent with the experimental absorption order.
In this part, the free diffusion simulation, which preliminarily characterized the difficulty of drug membrane penetration, combined with the umbrella sampling simulation, which evaluated the ability of drug membrane penetration, was used to reflect the ability of drug absorption. The results showed that the free diffusion coefficient D and △G order of the three drugs in two simulations were consistent, and the order was EPH > QUE > BAI, which is also consistent with the order of absorption in cell experiments. This indicates that the model established in this paper is reliable, and the two parameters can be used to reflect the absorption rate of drugs to a certain extent jointly.