Protein-specific constraints to diffusion

Constraints to diffusion of membrane-bound proteins are thought to be mediated by plasma membrane lateral heterogeneities that arise from the interaction of certain proteins and lipids.

As the specific interaction of proteins with other proteins and/or with lipids depends on their molecular properties, it is in general expected that not all proteins ”see” the same diffusion constraints. In particular, proteins with different membrane anchoring are likely to inter-act differently with the proteins and lipids in their local environment. Here, it was tested if GFP-EGFR and GFP-GL-GPI see a different membrane environment. GFP-EGFR is a transmembrane protein with a large (≈2 nm) cytoplasmic portion, whereas GFP-GL-GPI is a GFP that is anchored to the membrane via a lipid (GPI) that is located in the outer leaflet of the plasma membrane lipid bilayer (see Fig. 4.1). Using TNIM, it was measured ”how freely”

GFP-EGFR and GFP-GL-GPI are allowed to diffuse in their submicrometer-scale membrane environment. This was achieved by computing LRO images from 0.5 s directly before and from 0.5 s directly after the binding of the marker sphere. The reason for choosing this time window was that 0.5 s are long enough to allow the protein to efficiently sample the imaged area, but the transient exclusion from dynamic membrane structures is not averaged out. The percentage area accessed was computed as the percentage of pixels that the diffusing protein visited once or more (n(x, y)≥0⇔LRO(x, y)6=−∞).

Figure 4.6: Membrane accessibility for GFP-EGFR and GFP-GL-GPI as measured with 2D-TNIM.

A,B) Example LRO images. Left: marker-sphere at cell surface before binding. Right: after binding to GFP-EGFR (A) or GFP-GL-GPI (B). The percentage of pixels that were visited once or more (accessed area,n(x, y)≥ 0⇔LRO(x, y)6=−∞) is indicated. C,D) Percentage of accessed areabefore binding to EGFR (C) or GFP-GL-GPI (D) in 22 or 37 measurements, respectively. D,F) Percentage of accessed areaafter binding to GFP-EGFR (E) or GFP-GL-GPI (F) in 22 or 37 measurements, respectively.

Results and discussion

The examples in Fig. 4.6A,B and the histograms in Fig. 4.6C,Dshow that the unbound marker sphere was free to access most of the imaged area, indicating that its diffusion was not re-stricted by cellular structures that extended out of the plane of the plasma membrane (see also section 4.2.1 and section 4.1.4). Such unrestricted lateral diffusion of the marker sphere before binding facilitates the interpretation of the diffusion after binding, because diffusion constraints can be attributed to structures within the plane of the membrane as felt by the protein. Simply put, Fig. 4.6C and Fig. 4.6D show that ”the control behaves well”. After binding to GFP-EGFR, the marker sphere did generally access less area than before bind-ing (compare Fig. 4.6C and Fig. 4.6E), indicating that lateral membrane structures constrain diffusion of GFP-EGFR on the submicrometer scale. Furthermore, the membrane area ac-cessed was different in individual measurements, indicating that the Cos7 plasma membrane is heterogeneous on the submicrometer scale. Most measurements on GFP-GL-GPI yielded a higher membrane accessibility than the measurements on GFP-EGFR (compare Fig. 4.6Eand Fig. 4.6F), indicating that diffusion of EGFR is additionally constrained by structures that are formed by lipids and/or proteins in the inner plasma membrane leaflet and/or directly below the plasma membrane, because these are structures that GFP-GL-GPI can not directly interact with. This interpretation is supported by experiments showing respectively that dis-ruption of the actin cytoskeleton or truncation of the cytoplasmic domains of transmembrane proteins increases the ”barrier free path” in SSRM measurements (Edidin et al., 1994) and decreases the fraction of proteins exhibiting transient confinement in SPT experiments (Sako et al., 1998) (see also section 1.2).

About 15% of the measurements on GFP-GL-GPI appeared as a separated peak in the his-togram of accessed areas (Fig. 4.6F) at 50 to 60 percent accessed membrane area. Although more measurements are needed to confirm the existence of a separate popuplation, this obser-vation is interesting with regard to the measurement of the binding specificity of the marker spheres, which indicated that every 6^th to 7^th sphere may not bind GFP-GL-GPI but some other membrane component. For this reason (1/6.5·100%≈15%), the separate population in Fig. 4.6Fmay report the membrane accessibility for some other plasma membrane components (measurements on GFP-EGFR probably also contained few unspecific events, however they did not appear as a distinct population). An alternative explanation is that there are different populations of GFP-GL-GPI proteins that interact differently with their plasma membrane environment. In fact, there is evidence that a part of GFP-GL-GPI resides in specific lipid domains (”lipid-rafts”, see Pralle et al., 2000 and Kenworthy et al., 2004). It would therefore be an interesting future project to study if disruption of lipid rafts by cholesterol depletion (see section 1.1 or Edidin (2001b)) changes the membrane area that is accessible to GFP-GL-GPI,

since there is still a debate if lipid domains such as ”lipid rafts” existsin vivo, and if they exist, how they change the diffusive behaviour of the embedded proteins.

In conclusion, the different membrane accessibility of GFP-EGFR and GFP-GL-GPI provide evidence that GFP-EGFR interacts with structures in the inner plasma membrane leaflet and/or directly below the plasma membrane (e.g. cytoskeletal barriers), and that TNIM de-tects specific differences regarding the submicrometer-scale plasma membrane environment of different proteins.