Membrane fusion underlies numerous basic biological processes, including fertilization, viral entry, release of hormones, and rapid communication between neurons. However, the molecular mechanisms by which proteins catalyze the merger of lipid bilayers are not yet understood.
We are interested in how biological membranes fuse and, in particular, how synaptic vesicles fuse with the plasma membrane of neurons to mediate the release of neurotransmitters. We also study the botulinum neurotoxins, a group of toxins that cause botulism by blocking neuronal exocytosis. These toxins are widely used in clinical settings and, because of their extreme toxicity, they are also potential biological weapons.
The Minimal Protein Complement for Ca2+-Triggered Membrane Fusion
One approach that we have undertaken to study synaptic vesicle exocytosis is to use purified components to assemble the protein machinery that mediates regulated fusion. A reduced, defined system would make it possible to probe directly the inner workings of the fusion apparatus, allowing us to understand how it operates.
James Rothman's laboratory (Columbia University) has shown that three proteins (termed SNAREs)—synaptobrevin on the vesicle membrane and SNAP-25 and syntaxin on the target membrane—directly mediate membrane fusion in vitro. We have extended these findings and shown that in the presence of another protein, synaptotagmin (syt) I, SNARE-catalyzed membrane fusion is regulated by Ca2+. Thus, syt and SNARE proteins are the minimum protein complement for Ca2+-triggered membrane fusion in vitro. We are optimizing the minimal fusion machine, by adding back additional proteins and lipids, to make it operate on the millisecond timescale. We have combined this approach with biophysical and optical methods to monitor protein-protein and protein-lipid interactions during fusion and to resolve intermediates during the fusion reaction. These experiments are making it possible to understand the molecular rearrangements that underlie hemifusion, fusion pore opening, and complete bilayer merger.
Structure and Dynamics of Fusion Pores in Neuroendocrine Cells
During exocytosis, the first aqueous connection between the lumen of secretory vesicles and the extracellular space is called the fusion pore. Little is known regarding fusion pores; for example, it is not known whether the pore is lipidic or lined with protein. Our goals are to analyze the structure and dynamics of the pore and to relate this structure to the fusion machine discussed in the preceding section.
In a series of collaborative projects with Meyer Jackson's laboratory (University of Wisconsin–Madison), we are testing the idea that the transmembrane domains of SNARE proteins transiently line the fusion pore during secretion from neuroendocrine cells. Systematic substitutions of residues in the transmembrane domain of syntaxin revealed that three amino acids that lie along one face of this domain (when arranged into an α-helix) influence transmitter flux through the pore. There is a linear relationship between transmitter flux and the volume of the amino acid side chain placed in these three positions, but not in other positions, in the transmembrane domain. We are testing the idea that the transmembrane domains of synaptobrevin, in the vesicle membrane, form the other half of a fusion pore hemichannel.
Once opened, the pore has two choices—it can close (kiss and run) or dilate (full fusion). Using amperometry, we have shown that this decision is controlled, at least in part, by the interaction of syt with SNAREs, with different isoforms of syt giving rise to fusion pores with distinct kinetic properties. These studies are beginning to relate the minimal fusion machine that we study via reconstitution approaches to the fusion pore that we study in living cells.
Distinct Modes of Exocytosis from Hippocampal Neurons
A major challenge in the laboratory is to extend our studies of fusion pores that we carry out using amperometry and neuroendocrine cells to the fusion pores that mediate transmitter release from nerve terminals. An emerging view of presynaptic function is that neurons might employ multiple distinct release and retrieval pathways. For example, in hippocampal synapses, some fraction of release events might occur via the reversible opening of a fusion pore. Can neurons switch between different modes of release? Could such changes shape synaptic responses by precisely controlling the rate of transmitter efflux? These are some of the questions that we are addressing.
Richard Tsien's laboratory (Stanford University) discovered that in hippocampal neurons, synaptic vesicles loaded with FM dye sometimes release only a fraction of their dye—a finding that is consistent with kiss-and-run exocytosis. We also observed these phenomena and have quantified the proportion of partial dye-release events. Surprisingly, two-thirds of all events appear to be partial release events, with the balance of events involving complete release of the dye. By examining the rate of dye release, we found that complete release events can be explained by full fusion of the synaptic vesicle membrane with the plasma membrane. Partial release events, however, are very different; the kinetics of dye loss during partial release events are too slow to allow lateral diffusion of the dye into the plasma membrane. These findings suggest the predominant mode of exocytosis in hippocampal neurons does not permit lipid mixing but rather is mediated by a fixed fusion pore with a relatively small diameter.
We hypothesize that glutamate efflux through small fixed pores, which give rise to partial dye release from synaptic vesicles, will be too slow to efficiently activate AMPA receptors, the dominant excitatory receptors in the brain; this mode of release may instead result in AMPA receptor desensitization. Thus, the output of a glutamatergic bouton could be either excitatory or inhibitory, depending on the mode of release. We are working to manipulate the fusion pore in neurons and to relate changes in the mode of exocytosis to aspects of synaptic plasticity, including long-term potentiation.
Although the majority of our work concerns the molecules that mediate exocytosis and membrane fusion, we also work on a class of toxins whose biological function is to do the exact opposite: they block exocytosis. These are the botulinum neurotoxins (BoNTs), and they are the most potent toxins known. There are seven related toxins (A–G), and they cause botulism by entering neurons and cleaving SNARE proteins, resulting in paralysis and death. Although the action of BoNTs within neurons has been well established, little is known concerning the means by which the toxins gain entry into neurons in the first place. What are the cell surface receptors? What are the entry pathways that mediate internalization of the toxins? We have discovered that a number of BoNTs enter neurons by binding to synaptic vesicle proteins that are transiently exposed to the outside of neurons during exocytosis. In effect, the toxins piggyback onto recycling vesicles and enter cells via a Trojan horse strategy. This is rather clever, since entry of the toxin will shut down the uptake pathway, preventing futile uptake of additional toxin into nerve terminals that have already been poisoned; entry of additional toxin is thus limited to active nerve terminals.
We have identified some of the synaptic vesicle proteins that serve as toxin receptors, and this has made it possible to render animals resistant to these toxins. In addition, we are collaborating with Raymond Stevens (Scripps Research Institute) to cocrystallize the toxins with their receptors. This atomic resolution information is being used to create designer toxin-receptor pairs that can be used to sensitize cells to this class of molecules for therapeutic uses, and to design small molecular inhibitors that disrupt toxin-receptor recognition.