Measuring Cell Membrane Heterogeneity by Quantifying Membrane Protein Diffusion Using Camera-Based Fluorescence Correlation Spectroscopy

Abstract

In this dissertation, I present results from applying a novel technique developed by our group, to study dynamic membrane ultrastructures in living cells: (i) the effect of electrostatic clustering of negatively charged lipids in response to transient calcium influx; (ii) the effect of liquid general anesthetic drug on cholesterol stabilized nanodomains at physiological condition. The method is camera based fluorescence correlation spectroscopy, which we call ‘bimFCS’ for binned-imaging fluorescence correlation spectroscopy. An EMCCD camera simultaneously acquires fluorescence intensity data over multiple areas sizes by binning each camera pixel into super pixels. Subsequent fluorescence correlation spectroscopy (FCS) analysis of this data measures the diffusion over multiple length scales and allows to distinguish and study different types of membrane heterogeneity. Electrostatic clustering of negatively charged lipids by divalent ions has been observed in artificial lipid bilayers and been proposed as mechanism creating lateral order in the membrane. Here, we study the influence of calcium on the formation of PIP2 clusters and cholesterol-stabilized nano-domains in intact cells. We study changes of these domains in intact cells over time and upon calcium channel activation by analyzing the diffusion of GFP-tagged inner-leaflet membrane proteins. Using bimFCS, we measure diffusion on multiple length scales simultaneously to derive information about the domains. To study the formation of PIP2 clusters in the PM we use GFP- PH PLCδ to directly mark PIP2, and as marker for cholesterol-stabilized nano-domains, we use Lck-mGFP. We observe that opening TRPV1 channels leads to a transient rise in calcium as imaged using GCaMP5G, increases the interaction both between GFP-PH PLCδ and PIP2 domains, Lck-mGFP and cholesterol domains. The interaction between GFP-PH PLCδ and PIP2 domains decreases to base lines within one to two minutes, while the interaction of Lck-mGFP and cholesterol domains takes another two to three minutes to decrease as the cell down-regulates the intracellular calcium level after stimulation. Using an ionophore to clamp the calcium level at a fixed value, we determine the threshold for these effects. To control for large scale signaling, we image the membrane cytoskeleton using mCherry-alpha-actinin, and use GT46-GFP to mark transmembrane domains. These results suggest a concentration dependence of calcium-induced PIP2 clusters and cholesterol-stabilized nano-domains in the PM at calcium levels, which may be reached in intact cells locally by opening of ion channels. Despite many observed effects of anesthetic drugs, it is still debated whether the mechanism of general anesthesia is still a general membrane effect or caused by specific proteins. One well-studied drug used for human anesthesia, propofol, has been shown to interact with some ligand gated ion-channels. However, propofol also easily dissolves in the lipid bilayer and alters membrane fluidity. Which mechanism dominates in anesthesia or even how anesthesia arises are unclear. Here, we study the influence of propofol on plasma membrane (PM) ultrastructure in intact cells, which affect cell signaling. In the PM, transient submicroscopic nano-domains form by interactions between lipid-acyl-chains or lipid headgroups, stabilized by cholesterol. In addition, membrane cytoskeleton may further regulate these nano-domains. These domains regulate receptor interactions and signaling. We study transient propofol effect on these domains from low to clinically relevant propofol concentrations by analyzing diffusion of GFP-tagged outer leaflet/inner leaflet membrane proteins. Using bimFCS we measure diffusion on multiple length scales simultaneously. We observe that at lower propofol concentrations (up to 2 µM), the cholesterol nano-domains trap the GPI-mGFP less, which is consistent with the recent studies showing that propofol decreases the phase transition temperature of plasma membrane derived vesicles. Interestingly, at higher concentrations of propofol (20 µM to 150 ?M), the nanodomains trap the GPI-mGFP more strongly. This is only observed at physiological temperatures (37°C). By inhibiting myosin activity or actin filaments (de-)polymerization, we find that the activity of actin filaments further alters the behavior of cholesterol nano-domains due to propofol. We compare the effect of propofol and its analog confirming its specific effect as anesthetic drug. These results suggest that in intact cells, propofol induced cholesterol nano-domains changes are temperature dependent, regulated by cytoskeleton activities, and the effect is specific comparing to its analogs. The two studies provided unprecedented new quantitative details on how intact cells respond to external perturbations of their membrane structure, relevant both in health and disease. They also demonstrate the unique power of our bimFCS technique. During my thesis, I have further developed the software for the bimFCS analysis, including a new Java version to be distributed to Fiji users, and novel fitting algorithms.