Where do we go from here? Meeting Report on the Biophysical Society Discussion on ‘Probing Membrane Microdomains’, October 28–31, 2004, Asilomar, CA, USA

Biophysicists have long been interested in the phase behavior of lipids and the specialized roles that cholesterol plays in membranes. These questions have enjoyed renewed interest in light of the development of the lipid raft model. However, the relationship between so-called liquid-ordered domains in vitro and in vivo remains uncertain. Despite the ease of generating cholesterol-enriched microdomains in simple lipid mixtures, lipid rafts have proven remarkably difficult to identify in intact cells. The call for papers for the recent Biophysical Society Discussion meeting on ‘Probing Membrane Microdomains’ described the raft field as having reached a technical impasse. How can we best study these elusive domains? In this meeting report, I summarize the efforts of the biophysical community to increase our knowledge of raft structure and dynamics. Streaming video of the meeting is available on the Biophysical Society website at http://www.biophysics.org. It is clear that biological membranes are heterogeneous, but how this heterogeneity is regulated and its functional implications lie at the heart of many of the current controversies surrounding the lipid raft model. One such controversy concerns the size of lipid rafts. Jitu Mayor (National Centre for Biological Sciences, India) discussed recent fluorescence anisotropy (homoFRET) measurements addressing this question that show that a small fraction of glycosyl phosphatidyl inositol (GPI) anchored proteins are enriched in small clusters, the rest being dispersed as monomers. Containing a mixture of GPI-anchored proteins, these clusters are dispersed by cholesterol depletion, but sphingolipid depletion alters their susceptibility for cholesterol depletion, as might be expected for a cholesterol sensitive organization modulated by the level of sphingolipids. Surprisingly, the ratio of monomers to clustered molecules is constant across a wide range of protein expression levels, suggesting that the clustered fraction is actively maintained. The concept of small, scarce steady-state raft domains was reinforced by Aki Kusumi (Nagoya University). Using single molecule tracking techniques, he showed that a single GPI-anchored protein diffuses as rapidly as a non-raft lipid probe. However, upon crosslinking, the protein undergoes transient immobilization; remarkably, signaling molecules are recruited to these sites, suggesting these ‘receptor-cluster rafts’ define a region of localized signaling. The link between rafts and signaling was further explored by Susan Pierce (NIH), who discussed the visualization of antigen-induced B-cell receptor oligomerization and its subsequent association with rafts in living cells using Förster resonance energy transfer (FRET). A number of unanswered questions remain about the relationship between lipid rafts, caveolae, and cholesterol. Richard Anderson (University of Texas Southwestern Medical School) presented data regarding the role that caveolin plays in regulating the trafficking and subcellular distribution of cholesterol. A related point was discussed by Fred Maxfield (Weill Medical College of Cornell University), who noted that cholesterol can exchange between organelles by both vesicular and non-vesicular mechanisms. He also presented data showing the formation of large membrane domains marked by fluorescent lipid probes in cells following cholesterol depletion, a result that evokes a much different picture of plasma membrane structure than that suggested by either the fluorescence anisotropy or single molecule tracking results discussed above. The concept that liquid-ordered domains exist in cell membrane has renewed interest into the factors regulating lipid phase behavior in model membranes, especially with respect to the nature of the interaction between cholesterol and other lipids. An unusual feature of cholesterol is its condensing effect, i.e. a mixture of cholesterol and phospholipids has a smaller area per molecule than that of the sum of the two components taken individually. How does this occur, and what consequences does this have for lipid rafts? One idea, put forth by Harden McConnell (Stanford University), is that cholesterol forms complexes with phospholipids, forming a new molecular species in a manner analogous to the formation of a product by a reversible chemical reaction. He argued that complex formation regulates the ‘activity’ or chemical potential of cholesterol within the membrane. The presence of such complexes does not imply phase separation, as they could exist within a single lipid phase, be it liquid-ordered or liquid-disordered (Figure 1). A second model, discussed by Jerry Feigenson (Cornell University) in the context of his studies on the phase behavior of ternary mixtures of cholesterol and phospholipids, is an ‘umbrella’ model, in which the interactions of phospholipids and cholesterol are driven by the hydrophobic effect. Here, the need to sequester non-polar cholesterol from water causes increased order in the local packing of phospholipid acyl chains (Figure 1). The ‘cholesterol–phospholipid complex’ and ‘umbrella’ models of how cholesterol interacts with phospholipids. (A–C) Cholesterol–phospholipid complexes, shown looking down at the top of the membrane. Cholesterol is represented by the dark circles and complexed phospholipids by gray circles; uncomplexed phospholipids are shown as larger white circles. The smaller size of phospholipids in complexes represents the condensing effect of cholesterol. Note that in this model, cholesterol–phospholipid complexes do not necessarily form their own separate phase; instead, they can exist in (A) a single low-cholesterol phase, (C) a single high-cholesterol-phase, or (B) in a two-phase system consisting of a cholesterol-rich and cholesterol-poor phase. Other membrane components such as proteins can preferentially associate with either cholesterol–phospholipid complexes (white triangles) or with uncomplexed phospholipids (black triangles). Figure is reproduced from McConnell and Vrljic, Annu Rev Biophys Biomol Struct 2003, Vol. 32, pp. 469–492. (D–G) The umbrella model, shown as a side view. In this model, the interactions of phospholipids and cholesterol are driven by the need to protect the non-polar region of cholesterol from water (i.e. the hydrophobic effect). This is accomplished by shielding by the phospholipid headgroups (D). With increasing cholesterol levels, this requires a concomitant decrease in the area taken up by the phospholipids acyl chains (the condensing effect) (E). At saturating levels of cholesterol, non-polar surface is exposed, an energetically unfavorable state (F), leading to the removal of cholesterol from the membrane (G). Figure is adapted from Huang and Feigenson, Biophys J 1999, Vol. 76, pp. 2142–2157. Biophysical approaches to define the properties and phase behavior of mixtures of cholesterol and phospholipids were the focus of several talks. Sarah Keller (University of Washington) discussed how fluorescence microscopy and NMR could be used to identify immiscible liquid phases in giant unilamellar vesicles and supported membrane systems. The size and shape of domains are readily visualized by doping membranes with fluorescent probes preferring either ordered or disordered lipids, allowing for example the investigation of conditions that favor either the formation of nanoscale or micron-scale domains (Figure 2). Approaches based on fluorescence spectroscopy discussed by Erwin London (SUNY Stony Brook) either utilize fluorescent probes that exhibit environment-dependent emission spectra or monitor the short-range quenching of fluorescent probes by a spin label that prefers liquid-disordered domains. Finally, Tom McIntosh (Duke University Medical Center) discussed how X-ray diffraction could be used to study structural properties such as bilayer thickness of lipids mimicking DRMs, and the contribution of both bilayer structure and mechanical properties to the sorting of transmembrane peptides into detergent-soluble versus detergent-resistant membrane fractions. Micron-scale liquid phase separation in model membranes detected by fluorescence microscopy. Shown are giant unilamellar vesicles composed of a ternary mixture of a saturated phospholipids, DPPC, an unsaturated phospholipid, DOPC, and cholesterol at the following concentrations: (A) 2:1 DOPC/DPPC + 20% cholesterol; (B) 1:2 DOPC/DPPC + 20% cholesterol; (C) 1:2 DOPC/DPPC + 40% cholesterol; and (D) 1:9 DOPC/DPPC + 30% cholesterol. Vesicles were doped with Texas red-DPPE, which partitions into the less-ordered liquid phase (the light phase), as a contrast agent and imaged at 30 °C. Note that the fraction of the more-ordered phase (dark) increases with the relative mole fraction of cholesterol or saturated lipid. All scale bars are 20 µm. Figure is adapted from Veatch and Keller, Biophys J 2003, Vol. 85, pp. 3074–3083. The lateral diffusion of proteins and lipids in membranes is sensitive to their local environment, including the presence of membrane domains. Three major approaches have been used to monitor these effects in both cells and model membrane systems. The first, fluorescence recovery after photobleaching (FRAP), follows the exchange of bleached and non-bleached fluorescent molecules in either a diffusion-limited spot or a larger region of interest defined using a confocal microscope. As discussed by Anne Kenworthy (Vanderbilt University School of Medicine), FRAP can be used to test specific predictions of raft models, such as whether rafts diffuse as stable structures, and how protein mobility is affected by raft disruption. A second approach, single particle tracking (SPT), follows the movement of individual molecules, labeled with fluorescent tags or gold beads, to generate two-dimensional positional trajectories with very high spatial and temporal resolution, was presented by Marija Vrljic (Stanford University). In an interesting application of SPT, Vrlijic and co-workers examined the relationship between trajectories of two raft markers, finding little evidence for their co-diffusion within the same domain. A third single molecule technique sensitive to diffusional mobility is fluorescence correlation spectroscopy (FCS), described by Petra Schwille (Dresden University of Technology). Here, fluctuations in fluorescence resulting from the diffusion of individual molecules into and out of a small sampling volume defined using confocal optics are measured. Schwille described recent FCS studies from her laboratory comparing the diffusional mobilities of raft (cholera toxin B-subunit) and non-raft (DiI) lipid probes in giant unilamellar vesicles and in cell membranes in order to better understand the relationship between rafts in cells and in model systems. High-resolution microscopy techniques hold promise for revealing novel aspects of microdomain structure, a topic outlined by Linda Johnston (National Research Council Canada). Atomic force microscopy is a topographic technique with a resolution in the range of 10–100 nm. Near field scanning optical microscopy (NSOM), a technique with sub-diffraction resolution, uses a combination of both topography and fluorescence and is therefore compatible with FRET, FCS, single molecule, and co-localization studies. An exciting application of NSOM described by Johnson is its use to monitor eximer formation of BODIPY-GM1 as a tool to study its clustering. Most techniques used to study microdomains are sensitive to a limited range of dimensions and timescales. Josh Zimmerberg (NIH) argued for the importance of utilizing a combination of strategies to assess the size, structure, and function of rafts. One example is to combine a technique such as FRET that is sensitive to distances of <100 Å, with electron microscopy, which allows for measurements of protein distribution over three orders of magnitude. Because pathogens have evolved ways to exploit lipid rafts, they can be used as tools to investigate the roles of rafts in cells. As discussed by Gisou Van der Goot (University of Geneva Medical School), toxins such as cholera toxin B subunit, aerolysin, and perfringolysin O can be added exogenously and fluorescently labeled for visualization. Furthermore, many toxins undergo processing or conformational changes at intracellular compartments that can be read out biochemically, making them attractive tools for the study of lipid microdomains in cells. Electron microscopy has provided new insights into the spatial organization of cell signaling. Jan Oliver (University of New Mexico) presented data visualizing the organization of proteins involved in FcεR1 and EGF receptor signaling as detected by electron microscopy. Most of the proteins examined exhibit a clearly non-random distribution and undergo reorganization following receptor stimulation. Strikingly, however, they find little evidence for the co-localization of raft markers with these signaling complexes. The distribution of Ras on the inner leaflet of the plasma membrane, discussed by John Hancock (University of Queensland), also shows strong evidence for a non-random distribution. He summarized recent studies examining motifs responsible for targeting HRas to membrane microdomains, some of which are also important for regulating the strength of HRas binding to the plasma membrane. Immunocytochemical studies have also been important in the identification of a class of membrane domains formed by junctions of plasma membrane and ER, discussed by Mordecai Blaustein (University of Maryland School of Medicine). Containing a subset of plasma membrane and ER proteins, these junctions define a sub-plasmalemma space, which is proposed to function in Ca2 + homeostasis. Two final talks about the nature of inner leaflet domains and membrane-associated cytoskeleton focused on myristoylated alanine-rich C-kinase substrate (MARCKS), a PIP2-binding protein that contains clusters of basic and hydrophobic residues. Stuart McLaughlin (SUNY Stony Brook) suggested that MARCKS may nucleate the formation of ‘lipid shells’ by sequestering cholesterol upon PIP2 binding, while Mike Sheetz (Columbia University) described the role of MARCKS in regulating the strength of membrane-cytoskeletal interactions. Correlation spectroscopy methods have a number of potential applications that have not yet been fully explored in the study of membrane domains. Enrico Gratton (University of Illinois) discussed several applications of FCS beyond measurements of diffusional mobility. Analysis of the spectrum of the fluctuation amplitude, referred to as photon-counting histograms, yields information about brightness distribution, which can be used to assess protein self-association as well as to quantify the concentration of labeled proteins in vivo. Another variation of FCS, scanning FCS, measures multiple points as a function of time and can be used to examine spatial correlations between fluctuations. A second correlation spectroscopy method, image correlation spectroscopy (ICS), is analogous to scanning FCS in that it measures the spatial correlation of fluorescence over time. In this technique, a time series of images collected with a confocal microscope are subjected to correlation analysis. Paul Wiseman (McGill University) presented examples of how ICS can be used to measure slowly diffusing species, extract vectors of movement of fluorescent molecules, and even measure lipid diffusion from data collected at video rate. In addition to detecting clustering of proteins in rafts as discussed above, FRET is a valuable approach for studying molecular interactions in cells. Fluorescence lifetime-imaging microscopy (FLIM) can be used to monitor FRET, providing spatially and temporally resolved maps of the interactions of fluorescently labeled molecules within cells. Two talks described the application of FLIM in the study of signal transduction processes, one from Paul Van Bergen en Henegouwen (University of Utrecht) on the use of single domain Llama antibodies as probes for FLIM and a second from Banafshe Larijani (London Research Institute), which examined FRET between molecules such as EGFP-phosphoinositol transfer protein and BODIPY-labeled phospholipids. Membranes are essentially two-dimensional liquids that are ordered in space and time. John Nagle (Carnegie Mellon University) mentioned that X-ray and neutron scattering are being used to reveal features of lipid bilayer structure such as bilayer thickness and membrane-bending modulus. He proposed that the scattering, as contrasted with diffraction, approach may be useful to extract information about small-scale lateral organization within membranes. Nagle emphasized that lateral heterogeneity does not require phase separation per se and that small-scale organization may be better described by a correlation function with a characteristic decay length. The theme of correlation lengths was echoed by Ole Mouritsen (University of Southern Denmark), who presented the results of computer simulations examining the short-range order within mixtures of cholesterol and phospholipids. Computation approaches can also provide insight into atomic level interactions of molecules over very small timescales. For example, simulations presented by Larry Scott (Illinois Institute of Technology) of a ternary mixture of cholesterol with sphingomyelin and an unsaturated phospholipid, DOPC, show that cholesterol prefers to localize at the interface between sphingomyelin and the DOPC, whereas sphingomyelin packs preferentially against the smooth face of cholesterol. It will be quite interesting to learn more about the nature of the interface between liquid-ordered and liquid-disordered domains, as they may be equally biologically important as the domains themselves, an idea suggested by several of the meeting participants. Lessons from model systems have undoubtedly been informative in shaping our understanding of the properties of membrane microdomains. For example, ongoing studies examining which sterols and other lipids act as raft ‘promoters’ versus raft ‘breakers’ promise to reveal new insight into how lipid–lipid interactions contribute to domain formation and lipid phase behavior. This does not mean that a consensus on how this works has been reached, as evidenced by the lively discussion over the nature of cholesterol–phospholipid complexes versus the ‘umbrella’ model. Nevertheless, it is clear that drawing analogies between membrane models and cell membranes needs to be done with appropriate caution. As pointed out by Michael Edidin (Johns Hopkins University), lipid phases are at equilibrium and are relatively stable and static structures, while cell membranes exist in a steady state maintained by active processes. In cells, the presence of the cytoskeleton, interactions between membrane-associated proteins, and local remodeling of lipid and protein composition as the result of signaling or trafficking can all give rise to the formation of transient domains. Indeed, how to best study cholesterol-dependent domains in cells remains a difficult question to answer. A growing consensus is that both detergent-resistant membrane fractionation and cholesterol depletion, typically used to identify lipid rafts, are problematic. In contrast to the often-cited concept of raft domains existing in a sea of non-raft membrane, a large fraction of cell membranes is to with (Figure The that cholesterol depletion by cholesterol-enriched domains has also been by recent showing of protein diffusion in to cholesterol depletion that to be to changes in the organization of the However, clear single to these approaches has yet a combination of approaches may to be the best to the composition and function of membrane microdomains in cells. of detergent-resistant membrane in cells. cells on were either cholesterol by with or to with on for The cells were and labeled with toxin B subunit, a for lipid rafts. Note the large fraction of detergent-resistant membrane to cholesterol 20 Figure is adapted from Kenworthy J Vol. pp. we have not yet the biophysical tools available to this One area of promise that has not yet been fully explored in this is the use of techniques sensitive to molecular dynamics. One such be to combine positional information by single molecule tracking with fluorescence properties of probes to on local or For such the use of with probes such as in cells also be efforts between and are also to questions such as what features of a transmembrane give rise for for liquid-ordered approaches will be useful to the question of whether liquid-ordered domains are to the formation of functional domains in cells. I Harden Sarah Keller, and Jerry and their for providing and for their

Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.