Summary thesis Marjolein Koopman (English)

The cell plasma membrane of eukaryotic cells is a lipid bi-layer that separates the cell cytosol from the extracellular environment. The composition and organization of proteins and lipids within this bi-layer have a direct impact on many cellular processes, since they form the senses of the cell. Technological advances, like high resolution microscopy together with the possibility to address different membrane components via specific labeling now allows researchers to investigate cell membrane organization in detail.

It is well recognized that clustering of cell surface receptors into micro-domains fulfills an important role in regulating cellular functions. Unfortunately, the domains are often too small to be resolved with conventional optical microscopy. Near-field scanning optical microscopy (NSOM) is a relatively new technique that combines ultra high optical resolution, down to 70 nm, with single molecule detection sensitivity. As such, the technique holds great potential for direct visualization of domains at the cell surface. Yet, NSOM operation under liquid condition is far from trivial. In Chapter 2 of this thesis we have shown that the performance of NSOM can be extended to measurements in liquid environments. We have presented a reliable and easy-to-use system, with a perfect analogy to a diving bell, to perform tuning fork-based near-field scanning optical microscopy on soft cells in liquid. The principle of the diving bell system is to allow vibration of the tuning fork in air, while the NSOM probe is immersed in solution. In this way Q factors of 200 and higher in liquid are routinely obtained. The force feedback is reliable and stable over hours requiring minimum adjustment of the set-point during imaging. With this system, tip-sample interaction forces are kept below 350 pN enabling imaging of soft cells in buffer solution. For the first time, individual fluorescent molecules on the membrane of cells in solution were imaged with a spatial resolution of 100 nm. As such, liquid-NSOM is capable to reveal cell membrane organization in detail, while working in conditions that allow live cell imaging.

In Chapter 3 we have used the liquid-NSOM to investigate the existence and composition of a highly debated type of membrane domain, the so called lipid rafts. These lipid rafts (domains within the membrane enriched in cholesterol and glycosphingolipids) are believed to play a key role in many membrane related processes like immune cell signaling and viral entry. Their existence is rather controversial, since evidence for the presence of lipid rafts in native cell membranes can only be obtained via indirect methods. In Chapter 3 we demonstrated the ability of NSOM to directly visualize lipid rafts, enriched in glycosphyngomyelin (GM1), via fluorescently labeled cholera toxin (CTxB) both in immature dendritic cells (imDC) and human monocytes (THP1) under liquid conditions. Remarkably, on both cell types GM1 nano-domains appeared to be smaller than 100 nm in size. Furthermore, exploiting single molecule detection we quantified the GM1 content of each individual domain. On both cell types, most domains only bind 1 to 6 CTxB molecules, while on THP1 cells, up to 25 CTxB molecules per domain were identified. These results are consistent with the most recent picture of functional raft pre-cursors as nanoscale entities containing only a few molecules. This small size of the domains in combination with a high packing density adds an extra challenge to the analysis of domain content. In fact, at high packing densities, two or more particles that have no association can coincide within the same excitation volume resulting in brighter fluorescent spots and apparent clustering. The higher the density and the lower the imaging resolution, the more apparent clustering will result. Although NSOM provides superior spatial resolution, individual fluorophores within a nanometer-sized domain cannot be directly resolved. As a consequence, domain content still needs to be quantified based on fluorescence intensity.

In Chapter 4 we demonstrated that experimentally obtained fluorescence images can be compared to simulated images of randomly distributed particles at densities related to experimental conditions in order to assess the degree of true clustering. We have used these simulations to investigate the degree of true clustering of the lipid raft marker GM1 labeled with CTxB on the membrane of wet THP1 cells. The combination of high resolution optical microscopy and computer simulations has allowed us to unequivocally demonstrate nano-scale clustering of GM1, providing direct evidence that nanometer sized lipid domains (lipid raft pre-cursors) indeed exist in the cell plasma membrane. We have also applied NSOM on wet cells to map the organization of different protein receptors on two different cell types with a spatial resolution better than 100 nm.

The experiments described in Chapter 5 were combined with simulations using experimentally obtained parameters, i.e. receptor density and fluorescence intensity, to assess the degree of clustering. From two non raft markers investigated, the transferrin receptor CD71 appears randomly organized on THP1 cells, while CD46 forms nano-domains on imDC. Remarkably, we also found that the GPI anchored protein CD55, a commonly used raft marker, does not cluster on both imDC and THP1 cells, but rather organizes in a random fashion. These results demonstrate that classification as 'lipid raft associated' does not give a priori information on surface arrangement, i.e.lipid raft partitioning does not necessarily implies clustering and clustering is not perse maintained by lipid rafts. Furthermore we have mapped the organization of the C-type lectin DC-SIGN on imDC and the integrin LFA-1 on THP1 cells. Our results on wet cells confirm clustering of these proteins at the nanometer scale, consistent with previous TEM experiments on dried cells. Our findings favor a model where not lipid raft partitioning but other mechanisms like protein-protein interactions or the cytoskeleton determine the distribution of proteins as either monomers or small clusters.

Many cellular functions depend on associations between proteins and/or lipids in the cell membrane. In Chapter 6 we therefore used the high resolution of NSOM to simultaneously investigate the nanometer scale spatial organization of different proteins and lipids on imDC and THP1 cell in solution. The extent of co-localization has been quantified using Pearsson's correlation coefficient and the results have been compared to confocal co-patching experiments. Significant association of different proteins (DC-SIGN and CD55 on imDC; and LFA-1 and CD71 on THP1) to the lipid raft marker GM1 has been observed using confocal co-patching. Strikingly, this spatial correlation has not been observed upon direct NSOM investigation, i.e., on fixed wet cells with no co-patching. The potential nano-scale spatial proximity of these proteins to the raft marker GM1 has been also investigated using interparticle nearest neighbor distance (nnd) analysis. The resultant nnd distribution for CD71-GM1 is completely randomconsistent with the fact that CD71 is a non-raft associated protein. On the contrary, the nnd distributions of CD55-GM1 and LFA1-GM1 are significantly shifted to shorter distances as to compared to random organization. These results indicate a statistically relevant preference for LFA-1 and CD55 to be in close proximity to lipid rafts. Altogether, our findings favor a model in which both proteins and lipids are pre-organized into small separate nanoscale domains, where these nanodomains might function as cell membrane organizers that facilitate and accelerate the formation of larger functional domains.

This thesis described the implementation of a diving bell concept to allow high resolution NSOM imaging in liquid conditions. Using this technique it was for the first time possible to visualize nanometer sized lipid and protein domains on both immature dendritic cells and THP1 cells in liquid. In chapter 7 the future of NSOM imaging will is discussed as well as the implications of our results for the current picture of cell membrane organization.