Two gigs of Munc18 in membrane fusion

All membrane fusion in eukaryotic cells, except mitochondrial and homotypic endoplasmic reticulum fusion, is critically dependent upon evolutionarily conserved soluble N-ethylmaleimide–sensitive factor attachment protein receptors (SNAREs) and Sec1/Muc18 (SM) proteins (1, 2). Similar to the cases for SNARE proteins, all SM-null mutants, from yeast to rat, exhibit a significant to complete loss of specific membrane fusion events (3, 4), suggesting the possibility of a conserved mechanism for the SM protein family. The SNARE proteins, as a central fusogen, have been well documented (5). In stark contrast, the function of SM proteins has been elusive, despite intense investigations. In PNAS, Yu et al. (6) present unique and exciting results that provide insights into the mechanism by which SM–SNARE interactions play out in driving intracellular membrane fusion.

To allow vesicle fusion to happen, cognate SNARE partners from two membranes associate with each other to form a parallel four-helix bundle that brings about the apposition of two membranes, facilitating fusion (7, 8). Among the SM family, neuronal Munc18-1 was the first member structurally characterized in context with SNAREs in high resolution (9). From the structure, it is evident that arch-shaped Munc18-1 binds the closed form of helical syntaxin 1a (Fig. 1A), one of the two target membrane (t-) SNAREs. Such capping of syntaxin 1a inhibits otherwise spontaneous binding to a second t-SNARE protein synaptosomal-associated protein 25 (SNAP-25) and subsequent engagement with vesicle (v-) SNARE vesicle-associated membrane protein 2 (VAMP2). As such, the data appear to contradict the expected critically positive role of Munc18-1 in membrane fusion, creating more problems than solutions (2).

Fig. 1.

Two modes of Munc18 binding to SNAREs. (A) Munc18 binds to the closed form of syntaxin. Syntaxin is composed of four structurally distinct regions: a membrane-spanning helix, an ∼70 residue-long SNARE motif, a regulatory Habc domain, and an N-terminal peptide. Arch-shaped Munc18 binds the syntaxin four-helix bundle assembled of Habc and Hd, part of the SNARE motif, whereby it protects syntaxin from spontaneous free interaction with SNAP-25. SNAP-25 is attached to the target membrane with lipids; it has two SNARE motifs. Upon spontaneous interaction, two copies of syntaxin bind to one copy of SNAP-25 and a total of four SNARE motifs form a four-helix bundle that is known to be a dead-end product. The N-terminal peptide also binds Munc18. (B) Munc13 removes the Munc18 capping from syntaxin. Upon the Ca²⁺ signal, Munc13 interacts with Munc18 and removes the binding to syntaxin, allowing only one SNAP-25 molecule to bind syntaxin, whereby it prevents misfolding to the 2:1 complex. (C) Munc18 binds to the SNARE complex. Although the Munc18–SNARE complex structure is not known, mutational analysis indicates that Munc18 is likely to bind the SNARE core region.

A few years ago, Shen et al. proposed a nice alternative to Munc18-1’s syntaxin binding and showed, using an in vitro fusion assay, that Munc18-1 binds to the SNARE complex and accelerates membrane fusion substantially (10). As a follow-up, in their recent PNAS report, Yu et al. attest that a similar binding mechanism and fusion-accelerating effect remain intact in other membrane-fusion machinery (6). This time, the authors started off with membrane fusion induced by syntaxin 4, SNAP-23, VAMP2, and Munc18c, the SNARE–SM pair for glucose transporter 4 (GLUT4) translocation in adipose cells. The authors examined the fusion between t- and v-SNARE–carrying proteoliposomes in the presence and absence of Munc18c without involving any other protein factors that might obscure the true Munc18c function. It was found that Munc18c binds to the SNARE complex and micromolar Munc18c accelerates SNARE-induced lipid mixing by as much as 10-fold, nicely upholding the proposed mechanism (Fig. 1C). Unlike neuron-specific Munc18-1, Munc18c is rather ubiquitously expressed across cells and, therefore, it was tested whether the fusion-accelerating effect could be observed with other cognate SNAREs. Indeed, Yu et al. (6) found that Munc18c can bind to and activate multiple cognate SNARE pairs.

Moreover, it turns out that the SNARE complex binding is shared by other fusion machineries, such as those in yeast (11). Taking these data together, Yu et al. (6) make a compelling case that the SNARE complex binding is a conserved mechanism that defines the essential nature of the SM protein in all SNARE-dependent membrane fusion.

Going back to the conundrum for neuronal Munc18-1, are we unequivocally sure that the Munc18-1 capping of syntaxin 1a truly results in the inhibition of synaptic vesicle fusion? A nice resolution to this seemingly paradoxical dilemma has been described in a recent report by Ma et al. (12). Syntaxin 1a and SNAP-25 tend to misfold in free binding: for example, they prefer a 2:1 stoichiometric binding. However, the 2:1 complex is known to be a dead-end product and unable to bind to VAMP2 (5). At the synapse, membrane fusion happens in a time scale less than a few hundred microseconds. To achieve such fast fusion kinetics an exquisitely coordinated effort of several SNARE complexes at the fusion site would be necessary. The presence of one or two misfolded binary complexes might be enough to kill the required coordination and cause misfiring of vesicle fusion. Mysteriously, however, such t-SNARE misfolding does not happen in the native membrane. Ma et al. (12) showed that Munc18-1 protects syntaxin 1a from free binding to SNAP-25. Instead, the productive 1:1 stoichiometric binding is only allowed in the presence of Munc13 (Fig. 1B), avoiding any misfolding. This mechanism clearly explains why deletion of Munc18-1, and the deletion of Munc13 as well, abrogates fast neurotransmitter release completely in neuronal exocytosis.

SNARE complex binding may have survived evolution and still remains functional, as Shen et al. proposed in their report (10). One might wonder, though, if the in vitro analysis by Shen et al. (10) might have been oversimplified: Only bare SNAREs were included in the assay, but critical fusion effectors, such as a major Ca²⁺ sensor synaptotagmin 1 (Syt1), were left out. What if we include Syt1 in the fusion assay and measure membrane fusion triggered by Ca²⁺? In the Ca²⁺-triggered synaptic vesicle fusion, Syt1 is supposed to sense Ca²⁺ and talk directly to the SNARE complex (13). Thus, it is likely that the protein standing right next to the SNARE complex is Syt1 at the moment of membrane fusion. Would it be possible that Syt1 makes a required intimate contact with the SNARE complex while a large arch-shaped Munc18-1 of 600 amino acids wraps around it? Additionally, if the SNARE complex binding is a dominant interaction, wouldn’t some fusion activity still remain instead of observed complete abrogation for the Munc18-1–null mutant, because the SNARE activity should still be there? For trafficking of GLUT4, are we sure that a Syt isoform is not involved and membrane fusion is not regulated by Ca²⁺? Alternatively, is the trafficking system regulated by a different signaling mechanism? If so, what is it? Single-vesicle in vitro fusion assays that include Syt1 are now available (14, 15), and Shen et al.’s (10) analysis of the Munc18-1 function can be readily revisited with the improved fusion assays.

An important point of which to be reminded is that in vitro reconstitution experiments can work as double-edged swords. By

Yu et al. found that Munc18c can bind to and activate multiple cognate SNARE pairs.

stripping off peripheral effectors, one can define functions of core protein factors less ambiguously. However, the lack of complexity sometimes causes deviation from the relevant physiology. It is intriguing that a functional rescue analysis using the Munc18-1–null mutant (16) disputes the conclusion in Shen et al.’s work, although there are some supporting data (17). We anticipate that the conclusions in the report by Yu et al. (6) will be tested rigorously in cellular environments as well.

Despite many questions, the report by Yu et al. (6) makes a convincing case that among the two SM–SNARE interactions, SNARE complex binding is the prevalent and indispensable one for membrane fusion within and across the species, therefore supporting the notion that it may be the evolutionarily conserved mechanism for SM proteins. However, neuronal Munc18-1, which must be at the pinnacle of the evolutionary path, is something special, and it instead interacts with syntaxin 1a. By doing so, Munc18-1 functions as a chaperon to protect t-SNARE from the misfolding that kills membrane fusion. Either way, SM proteins play major roles in membrane fusion by interacting with SNAREs.

In summary, the PNAS article by Yu et al. (6) and the report by Ma et al. (12) lay important groundwork for sorting out the complexity related to the critical necessity of SM proteins in membrane fusion. Tentative conclusions and shortcomings discussed here will all serve as germinating seeds for the exciting research activities on SM proteins in the future.