Giant vesicles

Although the presence of giant vesicles in mossy fiber boutons has been described (Henze et al., 2002; Laatsch and Cowan, 1966; Rollenhagen et al., 2007), the origin, cargo, and functional implications of this fascinating organelle remain enigmatic. Several lines of evidence support the hypothesis that giant vesicles containing glutamate neurotransmitter cargo contribute to glutamatergic signaling: i) electron micrographs revealed giant vesicles in proximity to mossy fiber active zones (Henze et al., 2002b; Laatsch and Cowan, 1966; Rollenhagen et al., 2007);

ii) large amplitude “giant” mEPSCs recorded from postsynaptic CA3 pyramidal neurons were demonstrated to be monoquantal (Henze et al., 2002b); iii) gamma-radiation lesions ablating the DG abolished giant mEPSCs recorded from CA3 pyramidal neurons (Henze et al., 1997).

My data expands upon these studies by demonstrating that the proportion of docked giant vesicles is highly comparable to the proportion of giant mEPSCs recorded from pyramidal cells. Although direct evidence that giant vesicles contain glutamate, or possess vesicular glutamate transporters, is still lacking. However, my observation that giant vesicle docking is abolished by deletion of Munc13 priming proteins indicates that they likely harbor at least some of the vesicular molecules required for evoked fusion.

In the scenario that if giant vesicles, though capable of docking, were actually incapable of fusing at the active zone membrane, it is conceivable that they limit access of “normal”

synaptic vesicles to release sites. This hypothesis is not supported by my analyses, which failed to detect a correlation between docked vesicle numbers and the proportion of giant vesicles (data not shown).

116 Multiple questions remain to be answered. Do giant vesicles fuse? Do they have the same vesicular release probability as a “normal” synaptic vesicle? Where and how are giant vesicles generated? Are these organelles unique to mossy fiber synapses? Several forms of activity-dependent membrane retrieval operate in synapses that could generate large vesicular structures, including compound fusion (He et al., 2009), ultrafast endocytosis (Delvendahl et al., 2016; Watanabe et al., 2013b, 2013a), and bulk membrane retrieval (Cousin, 2009).

Indeed, in the calyx of Held, large neurotransmitter-filled giant vesicles have been shown to result from activity-induced compound fusion of synaptic vesicles (He et al., 2009). Ultrafast endocytosis mechanisms generate large spherical endocytic intermediates in response to action potential-evoked release from small glutamatergic synapses from cultured hippocampal neurons (Watanabe et al., 2013a). More recently, functional analyses of endocytosis kinetics indicate that ultrafast modes of endocytosis also operate in other central synapses (Brockmann and Rosenmund, 2016; Delvendahl et al., 2016). Bulk membrane retrieval has only been observed in synapses during high activity states (Cousin, 2009) and is unlikely to be the origin of giant vesicles in mossy fiber synapses. It is also possible that giant vesicles result from de-granulation DCVs in a process comparable to that described for large DCVs in chromaffin cells (Shin et al., 2018). Although I occasionally observed filamentous electron dense material in the lumen of giant vesicles in the mossy fiber synapse, giant vesicles in mossy fibers persisted following pharmacological or genetic silencing in Munc13-deletion mutants. Since both synaptic and DCV fusion at synaptic active zones is severely decreased upon loss of Munc13 priming proteins (Augustin et al., 1999; van de Bospoort et al., 2012; Varoqueaux et al., 2002), my data indicate that at least a considerable subpopulation of giant vesicles are generated via activity-independent mechanisms.

Nevertheless, further analyses of ultrastructural changes induced during defined activity states are required to establish the contribution of compensatory endocytosis to large vesicle pools in mossy fiber synapses.

Alternative options include a form of constitutive membrane retrieval, which operates preferentially in mossy fiber synapses, or anterograde axonal trafficking of giant vesicles from the soma. I found evidence of giant vesicles in granule cell axonal projections in the CA3 from acute slice preparations alongside the other vesicle types, synaptic vesicles, and DCVs. This could indicate anterograde trafficking from the soma, however the transport direction of

117 these vesicles in plastic embedded samples cannot be determined. Moreover, the purpose of these precursor giant vesicles and whether they are molecularly equipped to contribute to synaptic transmission is unclear. If a proportion of giant vesicles are morphological correlates of membrane retrieval, adaptor protein-3, an endocytosis-associated protein, would be one molecular candidate to further investigate this line of inquiry as it has a specific influence on mossy fiber endocytosis and synaptic vesicle dynamics (Evstratova et al., 2014; Scheuber et al., 2006).

The release of large quanta from giant vesicles has potential implications for mossy fiber function and several important questions remain open. For example, do giant vesicles fuse with the same release probability as other synaptic vesicles? Does quantal release from giant vesicles contribute to mossy fiber facilitation? These could be addressed by measuring mEPSCs in CA3 pyramidal neurons with increasing external calcium concentrations to determine whether the spontaneous fusion of giant vesicles changes in a calcium-dependent manner. In a previous study, the frequency of giant monoquantal events on CA3 pyramidal neurons did not increase in a calcium-dependent manner (Henze et al., 2002b). It has been further shown that vesicles with high membrane curvature, such as small synaptic vesicles, promote more lipid mixing than low-curvature vesicles, such as giant vesicles (Malinin et al., 2002). These findings indicate already that giant vesicles intrinsically have a lower fusogenicity than small synaptic vesicles.

Future directions to explore in terms of giant vesicles would include determining what proportion of giant vesicles represent endocytic intermediates generated by constitutive pathways. This question can be approached in several ways: (i) uptake assays using cell-impermeable fluorescent dyes, photoconvertable dyes, or electron dense particles; (ii) inhibition of endocytosis using pharmacological application of dynamin inhibitors; and (iii) changes in giant vesicle abundance/morphological characteristics in response to defined synaptic activity regimes.

In the case of uptake assays, ferritin (Farrant, 1954; Watanabe et al., 2013a), phluorins (Ariel and Ryan, 2010), and FM1-43 dyes (Branco et al., 2010; Rizzoli and Betz, 2004) have been used to investigate endocytosis or endocytic by-products in synapses by being captured in the membrane invagination during endocytosis. These methods are usually ideal for cell

118 monolayers as is the case in autapses and dissociated hippocampal neuron culture (Ariel and Ryan, 2010; Branco et al., 2010; Watanabe et al., 2013a).

In the case of inhibiting endocytosis, DYNAsore and DYNgo are inhibitors of dynamin, a molecule necessary for pinching endocytosed vesicles to detach them from the presynaptic membrane (Daniel et al., 2012; Mccluskey et al., 2013). The relative abundance of giant vesicle-sized invaginations could be compared to the relative abundance of giant vesicles within a defined distance from the active zone. Such an experiment could also provide the locations of endocytosis in the synapse, though previous studies in cultured hippocampal neurons have demonstrated that ultrafast endocytosis occurs at peri-synaptic sites, at the periphery of the active zone (Watanabe et al., 2013a). One caveat to dynamin inhibition in slice cultures is that dynamin has a role in other membrane trafficking events within the entire cell, and cytotoxicity of the dynamin inhibitors could partially occlude the findings (McMahon and Boucrot, 2011).