5. New Insights and Advancements Provided by This Thesis
5.4 Discussion and Conclusions
Beginning by considering the methodological advancements presented in this Thesis, we first introduced a technique for generating initial setups for membrane protein simulations. The proposed approach excels in its universality compared to other methods and has promoted the rest of the research presented in this Thesis. Next, the results demonstrating the shortcomings of the Martini model in describing protein–
protein interactions were described. We proposed a simple correction that allows studies of lipid diffusion in the absence of excessive protein aggregation. Indeed, this approach was exploited in our further research on crowded membranes.
Moving on to discuss our work on lipid monolayers, we first assessed the validity of the free area model using lipid monolayers that allow for a systematic variation of area per lipid. We showed that while the model fits diffusion coefficient data reasonably well across different monolayer compression states, the fitting parameters have somewhat nonphysical values. It is not surprising that the free area model based on thermally-activated and discrete jumps fails, considering that we observed flow-like motion of the lipids. Such flow-like motion has also been demonstrated by simulations [11, 100, 102] and QENS experiments for lipid bilayers [103, 104]. We also examined the nature of anomalous diffusion in lipid monolayers. We found that lipid packing via monolayer compression results in the decrease of both diffusion coefficients and diffusion exponents and extends the subdiffusive regime to the microsecond time scale. These findings shape our understanding of lateral search processes in the pulmonary surfactant, where lipid–protein interactions regulate the transfer of lipids between the interface and the aqueous subphase — a key mechanism for breathing [81].
Proceeding to discuss our work on membranes crowded with proteins, we first considered the applicability of the Saffman–Delbrück model to protein-crowded membranes. We found that crowding decreases the diffusion coefficients of proteins of all sizes, yet this decrease is less radical for smaller proteins. Due to this size-dependent behavior, the Saffman–Delbrück model gets replaced by a stronger Stokes-like size-dependence between diffusion coefficients and protein radii in the crowded environment. The stronger scaling relation changes the picture of lateral search processes in cellular membranes drastically: in dilute conditions where the Saffman–
Delbrück model holds true, proteins of different sizes diffuse at almost equal rates, whereas in crowded conditions the smaller proteins move around significantly faster
5.4. Discussion and Conclusions 69 compared to larger ones that are virtually immobile. Hence, the Saffman–Delbrück model has to applied with extreme care to estimate diffusion-limited reactions in the heterogeneous and crowded plasma membrane. Unfortunately, a theoretical picture of the crossover from the Saffman–Delbrück model to the Stokes-like model is currently lacking. Furthermore, a systematic experimental study of the size-dependence of protein diffusion at different levels of crowding has not been performed. Such a study is likely further complicated by other non-idealities present in cellular membranes, described in Chapter 2.
Next, our findings on anomalous diffusion in crowded membranes were described.
We observed that protein crowding has surprisingly similar effects as lipid packing:
it lowers the rate of diffusion, decreases the values of↵, and extends the subdiffusive regime. Notably, we found that substantial protein aggregation leads to confinement effects that maintain anomalous diffusion up to macroscopic time scales. Regarding the anomalous diffusion mechanism, our results convincingly suggest that protein crowding leads to deviations from the FBM/FLE concept, which has been found to be valid in dilute conditions [16]. Instead, the diffusion of both lipids and proteins becomes non-Gaussian and highly heterogeneous, while maintaining its ergodic nature. Unfortunately, the exact mechanism behind this peculiar behavior remains partially unknown. Curiously, our simulation work suggests that the findings might be explained by a combination of a fluctuating diffusion model — accounting for the spatiotemporal variations in local diffusivity induced by protein–lipid and protein–protein interactions — and non-Gaussian PDFs due to proteins acting as geometric obstacles. This finding was supported by our simulations on 2D models that reproduced the non-Gaussian behavior. However, as discussed in Section 5.1.2, our findings might be somewhat affected by excessive protein aggregation, even though we did not use the Martini model but its derivative.
All in all, our results help understand how lipid packing and protein crowding affect the dynamics of lipids and proteins in conditions matching those present in the lung surfactant and at the surface of cells. Our observations provide further evidence that diffusion in lipid layers takes place via flow-like motion with viscoelastic effects, yet the presence of proteins complicates this picture substantially to a degree where no single known mechanism captures the nature of the motion. At long times — which in crowded and packed systems likely means milliseconds — the size dependence of protein diffusion heavily relies on crowding, which alters our picture of diffusion-controlled processes in the crowded plasma membrane. These findings
70 5. New Insights and Advancements Provided by This Thesis will undoubtedly pave the way towards understanding processes that depend on the formation of functional protein oligomers [189] and the regulation of protein function by specific lipids [176]. Still, plenty of more research is required before a comprehensive model for diffusion-controlled processes in complex conditions can be formulated. Moreover, the role of anomalous dynamics and especially the importance of the underlying subdiffusion mechanism for biological functions remains an intriguing puzzle for theoretical research to solve in the future.
71