Dynamin 1 Restrains Vesicular Release to a Subquantal Mode In Mammalian Adrenal Chromaffin Cells

Dynamin 1 (dyn1) is required for clathrin-mediated endocytosis in most secretory (neuronal and neuroendocrine) cells. There are two modes of Ca2+-dependent catecholamine release from single dense-core vesicles: full-quantal (quantal) and subquantal in adrenal chromaffin cells, but their relative occurrences and impacts on total secretion remain unclear. To address this fundamental question in neurotransmission area using both sexes of animals, here we report the following: (1) dyn1-KO increased quantal size (QS, but not vesicle size/content) by ≥250% in dyn1-KO mice; (2) the KO-increased QS was rescued by dyn1 (but not its deficient mutant or dyn2); (3) the ratio of quantal versus subquantal events was increased by KO; (4) following a release event, more protein contents were retained in WT versus KO vesicles; and (5) the fusion pore size (dp) was increased from ≤9 to ≥9 nm by KO. Therefore, Ca2+-induced exocytosis is generally a subquantal release in sympathetic adrenal chromaffin cells, implying that neurotransmitter release is generally regulated by dynamin in neuronal cells. SIGNIFICANCE STATEMENT Ca2+-dependent neurotransmitter release from a single vesicle is the primary event in all neurotransmission, including synaptic/neuroendocrine forms. To determine whether Ca2+-dependent vesicular neurotransmitter release is “all-or-none” (quantal), we provide compelling evidence that most Ca2+-induced secretory events occur via the subquantal mode in native adrenal chromaffin cells. This subquantal release mode is promoted by dynamin 1, which is universally required for most secretory cells, including neurons and neuroendocrine cells. The present work with dyn1-KO mice further confirms that Ca2+-dependent transmitter release is mainly via subquantal mode, suggesting that subquantal release could be also important in other types of cells.


Introduction
Sixty-four years ago, Bernard Katz proposed that presynaptic transmission is mediated as "all-or-none" neurotransmitter release (Del Castillo and Katz, 1954). Until recently, this quantal hypothesis has been one of the central concepts in synaptic transmission between neurons (Matsuda et al., 2009;Zhang et al., 2009), as well as neurotransmitter release from adrenal chromaffin cells (ACCs), a cell model for neuronal secretion studies (Chow et al., 1992;Zhao et al., 2016). Similar to neurotransmitter release in neurons (Neher and Sakaba, 2008), ACCs release catecholamine from large densecore vesicles (LDCVs) following an increase in intracellular Ca 2ϩ (Crivellato et al., 2008). Transmitter release from individual LDCVs or synaptic vesicles can be recorded as quantal spike (or foot ϩ spike) by electrochemical amperometry using a micro carbon fiber electrode (CFE) (Wightman et al., 1991;Chow et al., 1992;Alvarez de Toledo et al., 1993;Zhou and Misler, 1995a;Huang et al., 2007). In nonexcitable mast cells, there are two types of vesicle fusion modes, "full-fusion-like" (FFL) and "fusion pore flickering"/"kiss-and-run" (KAR) modes (Alvarez de Toledo et al., 1993). In principle, a KAR fusion event can produce both subquantal and full-quantal modes depending on whether the fusion pore closes before all native (or false) transmitters are discharged.
Using modified low-noise CFEs to record Ca 2ϩ -dependent single vesicle release in excitable chromaffin cells, we reported two types of native neurotransmitter release modes, the "standalone foot" (subquanta) and "foot ϩ spike" (full-quanta) (Zhou et al., 1996), which have been confirmed by others (Albillos et al., 1997;Alés et al., 1999;Elhamdani et al., 2001). Subsequently, GPCR-Gi-␤␥ is found as a signaling pathway regulates the vesicle release modes between subquanta and full quanta Gerachshenko et al., 2009). In addition, dynamin 1 (dyn1), a GTPase required for clathrin-mediated endocytosis and membrane remodeling in neurons and endocrine cells (Ferguson et al., 2007), has been implicated, using pharmacology, antibodies, and phosphorylation, in the release modes in ACCs (Elhamdani et al., 2001;González-Jamett et al., 2010;Samasilp et al., 2012;Jackson et al., 2015;Ren et al., 2016). Intracellular dialysis of a peptide derived from the proline-rich domain of dynamin eliminates the regulation of quantal size by Gi-␤␥, suggesting that dyn1 directly regulates quantal size . These experiments, however, suffer from possible off-target effects of the pharmacological manipulation of dyn1. To date, most studies on the mechanisms of subquantal release (or KAR vesicle fusion) have used cell lines (Pothos et al., 1998;Larsen et al., 2006;Westerink and Ewing, 2008;Mellander et al., 2012;Ren et al., 2016), so their results require critical confirmation in native secretory cells. In a few studies, effects of dyn1 overexpression or RNAi have used native ACCs, but unfortunately they lack controls (rescue or scrambled RNAi) for possible transfection artifacts (Anantharam et al., 2011;González-Jamett et al., 2013). Therefore, to determine the extent to which dyn1 is involved in exocytosis, it is essential to use the gold standard of loss of function and rescue of dyn1 in native dyn1-KO ACCs.
To determine the role of the subquantal mode in evoked catecholamine release, we determined whether dyn1 regulates subquantal release in native mouse ACCs using several single-vesicle assays [CFEs, total internal reflection fluorescence (TIRF) microscopy, electron microscopy (EM), and single-vesicle imaging of dextran uptake]. We found that, in contrast to the WT, dyn1-KO increased the largest quantal size by 250% and this was rescued by overexpressing dyn1-WT, but not a functionally deficient dyn1 mutant or WT dynamin 2 (dyn2). Therefore, dyn1 maintains most releasing vesicles in the subquantal mode (Ͻ40% of total content) by rapid resealing of the fusion pore before complete release of catecholamine in native ACCs.

Materials and Methods
Cell culture, transfection, and plasmids. The use and care of animals were approved and directed by the Institutional Animal Care and Use Committee of Peking University and the Association for Assessment and Accreditation of Laboratory Animal Care. dyn1-KO mice were kindly provided by Dr. Pietro De Camilli. We used dyn1-KO mice on postnatal days 2 (P2) to P5 because dyn1-KO mice can survive only Յ5 d after birth (Ferguson et al., 2007;Kuo et al., 2014). The mice were killed by hypothermic anesthesia on ice for 2 min. Both males and females were used in our experiments. Mouse ACCs were cultured as described previously with minor modifications (Albillos et al., 1997). Briefly, the adrenal glands were isolated from an anesthetized animal and then incubated in papain solution for 40 min at 37°C after removing the cortex. Cells were then dissociated by trituration and transfected with 3 g of NPY-pHluorin-expressing plasmids using the Neon (10 l) electroporation system (MPK1096, Invitrogen) according to the manufacturer's instructions. The transfected cells were plated on poly-L-lysine-coated coverslips and cultured for 18 -48 h in a humidified incubator (37°C, 5% CO 2 ) before use. The WT and KO cells from littermate mice were cultured and experimented under same conditions. NPY-pHluorin plasmid was constructed as described previously (Wang et al., 2017). NPY-mCherry was constructed by substituting pHluorin with mCherry. The dynamin 2 plasmid was a kind gift from Dr. Ana M. Cárdenas (Universidad de Valparaíso, Chile). All chemicals were from Sigma-Aldrich unless otherwise indicated.
Adrenal slice preparation. Adrenal slices were prepared according to previous reports (Chen et al., 2008) with minor modifications. Briefly, adrenal glands were removed from neonatal mice and immediately placed in ice-cold "slice solution" containing the following (in mM): 125 NaCl, 2.5 KCl, 2 CaCl 2 , 1 MgCl 2 , 1.25 NaH 2 PO 4 , 26 NaHCO 3 , and 10 glucose, pH 7.4, after gassing with 95% O 2 mixed with 5% CO 2 . The slice solution was used as the extracellular medium in the CFE recordings. When 70 mM K was used as a secretagogue, it replaced an equimolar amount of NaCl in the cutting solution. After removing overlying fatty tissue, the glands were embedded in 3% agarose (type VII-A, lowmelting-point agarose) and cut into slices (150 m thick) using a vibrating-blade microtome (Leica 1200). The slices were kept for 10 min at 37°C in a holding chamber containing cutting solution and then at room temperature. Slices were used for up to 12 h after cutting.
Electrochemical recordings. We used a standard patch-clamp amplifier (Yibo) for CFE (7 m diameter, ProCFEs, Dagan) recordings as described previously (Zhou and Misler, 1995b). CFE recordings were sampled at 5 kHz, low-pass filtered at 1 kHz, and digitized at 2-5 kHz. The top 75% of the largest events in each cell were analyzed for statistics. Igor software (Wavemetrics) and a custom-made macro-software (Zhou et al., 1996) were used for all offline data analysis. All experiments were performed at room temperature unless otherwise indicated.
TIRF imaging and analysis. TIRF imaging was performed on an inverted microscope with a 100ϫ TIRF objective lens (Olympus IX-81; numerical aperture 1.45). Images were captured by an Andor EMCCD using Andor iQ software with an exposure time of 50 ms. The standard bath solution contained the following (in mM): 145 NaCl, 2.8 KCl, 2 CaCl 2 , 1 MgCl 2 .6H 2 O, 10 H-HEPES, and 10 D-glucose, pH 7.4. The temperature was kept at ϳ35°C by a laboratory-made heater throughout all TIRF experiments. To distinguish between fusion events and vesicle movement, exocytotic events were defined as abrupt fluorescence increases immediately followed by a decrease or diffusion of NPY-pHlourin or BDNF-pHluorin puncta in the vicinity. In the analysis of single release events, each event was selected and marked with 1.92 m (center) and 2.4 m (annulus) diameter circular areas, following earlier reports (Bowser and Khakh, 2007;Wang et al., 2017). This "center and annulus" method for separating KAR and FFL events has been established in astrocytes (Bowser and Khakh, 2007) and DRG neurons (Wang et al., 2017) and it also works in chromaffin cells (see Fig. 4; also see examples from Movies 1, 2, 3 and 4, where FFL events with annulus "puffs" are profound). Fluorescence intensity values were calculated and analyzed using ImageJ. The high-K ϩ stimulus contained the following (in mM): 85 NaCl, 70 KCl, 2.5 CaCl 2 , 1 MgCl 2 , 10 H-HEPES, and 10 D-glucose, pH 7.4, and was applied using a gravity perfusion system (RCP-2B, Yibo). In bafilomycin experiments, cells transfected with NPY-pHluorin were pretreated with 1 M bafilomycin for 30 min and then TIRF imaging was performed as above.
TMR-dextran uptake. Cultured mouse ACCs were washed 3 times with standard extracellular bath solution and then incubated for 1 min with 50 M dextran (10 and 40 kDa, Invitrogen, D1816 and D1842, respectively) in 70 mM K ϩ -containing external solution. Unbound dye was washed out with standard bath solution (without Ca 2ϩ ) immediately after incubation. Z-series of 1 m optical sections were scanned through the 40ϫ oil-immersion lens of a Zeiss 710 inverted confocal microscope. The consecutive optical sections were z-projected and the total numbers of dextran fluorescent puncta per cell were determined. Images were processed with ImageJ. All the above experiments were done at room temperature.
EM imaging. EM of adrenal slices was performed according to previous descriptions with minor modifications (de Wit et al., 2009a;Wang et al., 2016). The adrenal glands were extracted from adult WT and dyn1-KO mice and then the adrenal cortex was removed and the medulla was cut into 1-2 mm 3 pieces. These pieces were then immersed in 5% glutaral-dehyde and 2% PFA in PBS for 1-2 h at room temperature. Cells were then postfixed for 30 min with 1% osmium tetroxide, washed, dehydrated through an ethanol series, embedded in Epon, and polymerized for 36 h at 60°C. Ultrathin sections (ϳ80 nm) were collected on singleslot, Formvar-coated copper grids and stained with 2% uranyl acetate for 30 min and 0.5% lead citrate for 15 min. Samples were observed in a Tecnai G2 20 200 KV transmission electron microscope at 120 KV. Vesicle diameters and numbers were measured manually with the Zeiss LSM Image Browser (version 3.0) as described previously (Chen et al., 2008).
Statistics. All experiments were replicated at least three times. Offline data analysis was performed using ImageJ and IGOR software (Wavemetrics). Data are presented as box-and-whisker plots or mean Ϯ SEM. Box-and-whisker plots show medians (central line in the box), ranges between the 25th and 75th percentiles (box), and minimum-maximum ranges (whiskers). Statistical comparisons were performed using the two-tailed unpaired Student's t test and the Kolmogorov-Smirnov test for cumulative analysis. All tests were conducted using SPSS version 13.0. Significant differences were considered to be p Ͻ 0.05.

dyn1 deficiency increases quantal size of catecholamine release
To verify dyn1-KO in mice, we found that the dyn1 protein was expressed in WT and successfully knocked out in ACCs from P3 mice ( Fig. 1H; for details, see Ferguson et al., 2007). To evoke Ca 2ϩ -dependent vesicle release in cultured ACCs, either caffeine (20 mM) or high KCl (70 mM) was used to increase intracellular Ca 2ϩ ([Ca 2ϩ ] i ) via caffeine-sensitive Ca 2ϩ stores or voltagegated Ca 2ϩ channels. When caffeine was applied for 5 s, a burst of amperometric spikes representing "quantal" catecholamine release from single LDCVs was evoked in both WT and KO cells ( Fig. 1 A, B). To analyze the kinetics of a quantal event, we defined the parameters foot duration, foot charge, peak amplitude (PA), half-height duration (HHD), and quantal size (QS) of single CFE spikes ( Fig. 1C; see also . Compared with the WT, the QS was significantly higher by 200 Ϯ 38% in dyn1-KO cells ( Fig. 1 D, E). In addition, dyn1-KO also increased the PA, HHD, foot charge, and foot duration, whereas the release frequency remained unchanged (Fig. 1E). Importantly, among the top 10% of the largest quantal events in WT and KO cells, the QS was also dramatically higher by ϳ250% in dyn1-KO cells (Fig.  1E). To determine whether the vesicles are larger in dyn1-KO cells, we used EM imaging to visualize LDCVs in adrenal sections ( Fig. 1F ) and found that the vesicle size remained unchanged in dyn1-KO ACCs (Fig. 1G). Compared with previous EM images of neural vesicles, our EM results in ACCs are similar to those in the calyx of Held synapse (Lou et al., 2008), but different from those in hippocampal synapses with cluster-like vesicular structures or bulk-like structures (Raimondi et al., 2011;. In addition, the expression levels of the vesicular catecholamine content-related proteins vesicular monoamine transporter 1 (VMAT-1) and TH also remained unchanged in KO ACCs (Fig. 1I), indicating that the increased quantal size of catecholamine release was not due to changes in vesicle size or catecholamine content per LDCV. Additional recordings of membrane capacitance revealed 50% smaller capacitance increases in WT versus KO cells (data not shown), consistent with the idea that KAR is the dominant release mode in WT cells. Collectively, these data suggest that all single-vesicle release events are subquantal in native ACCs. dyn1 contains five functional domains (Fig. 1J ), including the proline-rich domain that is involved in regulating QS in ACCs . In dyn1-KO cells, the phenotypes of increased QS and HHD were fully rescued by WT dyn1, but not dyn1-K44A, a dominant-negative GTPase mutation ( Fig. 1 J, K ). In addition, the QS effect of dyn1-KO was not rescued by dyn2 (Fig. 2). dyn2 has a role in regulating slow endocytosis, but not dyn1-dependent fast endocytosis (Artalejo et al., 2002). We found that the expression of dyn2 was intact in dyn1-KO adrenal medulla (Fig. 2D) and its overexpression did not rescue QS in dyn1-KO ACCs ( Fig. 2A-C). These experiments indicated the functional specificity of dyn1 in regulating fusion modes and fusion pore dilation. Therefore, the increased QS is specific to dyn1 deficiency and requires the GT-Pase activity of dyn1.
To further investigate the effect of dyn1 on QS under more physiological conditions, we did similar experiments in ACCs from adrenal slices (Fig. 3 A, B) versus those in primary culture (Fig. 1). As in cultured ACCs, dyn1-KO also significantly increased the QS, PA, foot charge, and foot duration in adrenal slices stimulated by caffeine ( Fig. 3C-E). Similar results were obtained in adrenal slices stimulated by 70 mM high-potassium (70 K) depolarization (Fig. 3F ). The statistical data from slices ( Fig. 3 E, F ) were consistent with those from cultured cells (Fig. 1E).
Together, the above data demonstrated that dyn1-KO increases QS, but does not affect the release frequency and vesicle size in ACCs in both the adrenal slice and primary culture.

Full-fusion events are increased by dyn1-KO
The above CFE recordings established that QS, but not release frequency, was increased by dyn1-KO. This raised the possibility that the release probability of full-quantal (or FFL) events could be increased by dyn1-KO. To test this prediction using a method independent of CFE recording, we used TIRF microscopy to monitor the dynamics of single exocytosis of pHluorin-tagged vesicle cargos in transfected ACCs (Fig. 4). The fluorescence of pHluorin is quenched as a result of the acidity of the vesicle lumen, but rapidly increases during opening of the fusion pore due to the deacidification effect of the neutral extracellular solution (Miesenböck et al., 1998). The mCherry protein is insensitive to pH changes, which permits monitoring of vesicles docking near the plasma membrane. In ACCs, the docked vesicles along the plasma membrane were similar in WT and dyn1-KO cells expressing NPY-mCherry (Fig. 4A). Based on the release and diffusion of fluorescent content measured by TIRF imaging, we distinguished two vesicle fusion modes, KAR and FFL (  Bowser and Khakh, 2007;Matsuda et al., 2009;Xia et al., 2009;Zhang et al., 2009;Wang et al., 2017). As defined previously (Wang et al., 2017), KAR events showed a brief fluorescence increase in the center, but without a clear fluorescence increase in the annular area (Fig. 4C,D), representing a transient opening and reclosure of a restricted fusion pore. Conversely, FFL events showed a robust fluorescence increase occurring at   14); one-way ANOVA. C, Immunofluorescence of dyn2 in dyn1-KO chromaffin cells transfected with GFP-tagged dyn2 plasmid for 48 h. The fluorescence intensity in cells overexpressing dyn2 (dyn2-OE) was normalized to native control cells (n ϭ 17 cells for control and n ϭ 17 cells for dyn2-OE, two-tailed unpaired Student's t test). Scale bars, 10 m. D, dyn2 expression in adrenal medulla from WT and dyn1-KO mice (n ϭ 6 independent experiments, two-tailed unpaired Student's t test). Data are presented as median with interquartile range (B, C) or mean Ϯ SEM (D). NS for p Ͼ 0.05, *p Ͻ 0.05, **p Ͻ 0.01, ***p Ͻ 0.001.
both the center and the annular area of NPY-pHluorin puncta (Fig.  4E,F), implying that the fusion pore was large enough for full cargo release. The detailed parameters used to definite the KAR and FFL events are described in the Materials and Methods.
The modes of KAR and FFL determined by TIRF imaging in Figure 4 were based on following evidence: (1) the membraneto-coverslip distance was stable over a large continuous area (see Movies 1, 2, 3 and 4); (2) the events were spatially and tem-  porally random (see Movies 1, 2, 3 and 4; also see Duncan et al., 2003;Perrais et al., 2004), similar to that in astrocytes (Bowser and Khakh, 2007) and sensory neurons (Wang et al., 2017); (3) FFL events were increased in KO versus WT cells (Fig. 4I ); (4) the number of docking vesicles was similar in KO and WT cells (Fig.  4A); (5) as revealed by NH 4 Cl quench of pHluorin (data not shown), the pHluorin decay signal reflects the transient openingclosing of the fusion pore in both WT and KO cells; and (6) similar results for KAR and FFL are fully consistent with CFE amperometry (Figs. 1, 2, 3), blockade of proton pump by bafilomycin (see below), as well as dextran uptake (see below).
We also investigated whether dyn1-KO affects single-vesicle release using TIRF imaging. Strikingly, dyn1-KO increased the FFL event ratio from 25 Ϯ 6% to Ͼ70% of the total LDCV fusion events evoked by KCl depolarization, whereas the release frequency remained unchanged (Fig. 4I ). Next, we analyzed the release kinetics of single fusion events in WT and dyn1-KO ACCs (Fig. 4K-N ). Averaging all KAR events (Fig. 4K ) showed that dyn1-KO dramatically reduced the rise time (RT) and HHD (Fig.  4M ), indicating that dyn1 decreases the fusion pore size (d p ) and decelerates release, so events are faster in the KO than in the WT (Fig. 4O). In dyn1-WT cells, there was a notable difference between KAR and FFL events. Both RT and HHD of a quantal spike were substantially slower during KAR (RT ϳ100 ms, HHD ϳ470 ms) versus FFL (RT ϳ70 ms, HHD ϳ300 ms). In KO cells, surprisingly, these kinetic differences between KAR and FFL events were abolished (Fig. 4M ). The longer RT of KAR was probably due to the smaller fusion pore and the fact that the vesicular H ϩ pump balanced a higher percentage of H ϩ influx, leading to slower quenching of pHluorin at KAR in WT cells. The HHD of quantal events determined in Figure 1E by CFE and in Figure 4,M and N, by TIRF were opposite. This was probably because that HHD in Figure 1E represents the released catecholamine from the vesicles, whereas HHD in Figure 4, M and N, represents the remained contents (NPY-pHluorin) after vesicle release. Because more contents were released in dyn1-KO cells (see below), the decay was faster in dyn1-KO (Fig. 4 M, N ).
Because dyn1 slows the vesicle content release in WT cells (Fig.  4M,N), more content may remain in vesicles after fusion. To test this hypothesis, we pretreated cells with 1 M bafilomycin (a vesicular H ϩ -ATPase inhibitor; Zhu et al., 2009), for 30 min to inhibit the reacidification of pHluorin and to detect the remaining content after KAR release (Fig. 5). Compared with results such as those shown in Figure 4M without bafilomycin treatment, we found that bafilomycin treatment significantly increased the RT and HHD of WT KAR events, but had minor effects on the HHD and RT of KO events (Fig.  5B-D and Movies 3 and 4), indicating that dyn1 constricted the release kinetics of contents from the fusion pore. Therefore, more fluorescent cargo was retained in WT versus KO cells and was released in a slower pattern (Fig. 5).
Because the release of NPY-pHluorin (monomer ϳ30 kDa) was sufficiently rapid to challenge the capture rate of our TIRF microscope (20 Hz, or 50 ms per image), we also examined the larger vesicular cargo BDNF-pHluorin (dimer ϳ2 ϫ 45 kDa). Consistent with NPY-pHluorin, dyn1-KO increased the percentage of FFL release events of BDNF-pHluorin (Fig. 4J ), but with a longer discharge time than NPY-pHluorin (Fig. 4G,H ). This BDNF-pHluorin dynamics was similar to the previous report of a slower pattern in both exocytosis and endocytosis (Aoki et al., 2010). It is known that mature BDNFs dimerize (Xia et al., 2009) and interact with other luminal components (de Wit et al., 2009b), which leads to an unbinding process and probably causes the distinct biphasic decay (Fig. 4G,H ). Although the RT of NPY- PHluorin (but not BDNF-pHluorin) was close to the limit of the TIRF sample rate (20 Hz or 50 ms), the statistical averaging of RT and HHD increased the temporal resolution beyond the TIRF sample rate, as evident by the similar results using either NPY-pHluorin or BDNF-pHluorin (Fig. 4). Therefore, either of these indicators is suitable for studying dyn1-KO effects in ACCs (Fig. 4K,N).
Together, these data strongly suggest the following: (1) the WT-d p of FFL events is as large as that of KO (both KAR and FFL) events to release most (or all) of the NPY-pHluorin during an exocytotic event (Fig. 4K,M); (2) KAR events in KO cells (Ͻ30% of total events, Fig. 4I) have a larger fusion pore (Fig. 4M) and a shorter open time; and (3) dyn1 is not required for vesicle fusion, but is an essential factor to restrict d p size during fusion (Fig. 4M). These single-vesicle imaging data lead to the conclusion that dyn1 maintains vesicles in the KAR mode, most likely by reducing the d p (Fig.  4O).

dyn1-KO increases d p as determined by endocytosis assay
We used another independent fluorescence-based method to estimate the expansion of the fusion pore in WT and dyn1-KO ACCs. Fluorescence-tagged dextran is a large inert fluid-phase marker and has been used as nanoparticles to estimate d p during the exocytosis-endocytosis process (Fulop et al., 2005;Clayton et al., 2008;Raghupathi et al., 2016). ACCs were loaded with different-sized dextrans dissolved in high K ϩ and then unbound dye was immediately washed out with standard bath solution (without Ca 2ϩ ). Cells were then scanned in Z-stacks by confocal imaging. With small dextran (10 kDa, ⌽ 5 nm), dextran uptake into vesicles was similar in KO and WT cells (Fig. 6 A, C). However, the uptake of larger dextran (40 kDa, ⌽ 9 nm) was significantly greater in KO cells than in WT cells (Fig. 6 B, D) and this phenomenon also occurred in cells without K ϩ stimulation (Fig.  6 E, F ). Together, these data imply that, in addition to dextran uptake by endocytosis fission, at least part of the dextran uptake is due to influx of dextran through the KAR fusion pore, the size of which (d p ) was increased from 5 nm Ͻ d p Ͻ 9 nm to d p Ͼ 9 nm by dyn1-KO (Fig. 6G).
Independently, somatostatin, which activates GPCR-Gi-␤␥ and reduces QS (Chen et al., 2005), inhibited the uptake of large (40 kDa), but not small (10 kDa), dextran, supporting the assumption that d p can be estimated by the uptake of dextran nanoparticles into ACCs (data not shown).

Discussion
In neuroscience, the classic neurotransmission hypothesis is that Ca 2ϩ -dependent vesicular neurotransmitter release occurs via "all-or-none" quanta. This dogma has been challenged by evidence that subquantal "stand-alone-foot" (mediated by KAR vesicle fusion) release exists in neuroendocrine ACCs (Zhou et al., 1996;Albillos et al., 1997) and in synaptic terminals (Stevens and Williams, 2000;He and Wu, 2007;Alabi and Tsien, 2013). It is unclear, however, how much subquantal release affects total catecholamine release in ACCs. In this work, we demonstrated that dyn1 maintains most (if not all) vesicular catecholamine release via the subquantal KAR (Յ40% of content per exocytotic event) mode in ACCs, as supported by four distinct fusion-pore assays (CFE, TIRF-pHluorin, bafilomycin-quenching, and dextran uptake). dyn1 restricts the fusion pore during vesicle fusion with the plasma membrane (Fig. 6G).
One major finding of this work was that, following Ca 2ϩ influx, dyn1 caused most exocytotic vesicles to release catecholamine as subquantal events in WT ACCs. This is supported by the following evidence from cultured ACCs: (1) physiologically, we have previously reported that QS is reduced by endogenous ATP via GPCR-Gi-␤␥, indicating the existence of subquantal release in ACCs (Chen et al., 2005); (2-3) pharmacologically, the QS of evoked release events was increased either (2) by acute application of a protein kinase C agonist or (3) by whole-cell dialysis of the proline-rich domain of dyn1, implying that part of the previously assumed quanta might be subquantal ; (4 -5) genetically, dyn1-KO increased the largest QS of evoked events by 250% (Fig. 1A-E), whereas the (4) vesicle size and (5) vesicle contents remained unchanged (Fig. 1 F, G,I ); and (6) dyn1-KO also increased the QS in fresh adrenal slices (Fig. 3). Together, these data established that dyn1-KO increases the largest single release events by 250%. Therefore, most (if not all) Ca 2ϩ -evoked vesicle release events are subquantal.
Our second major finding was the mechanism by which dynamin produces subquantal release. We found that dyn1-KO increases QS by increasing the d p . This is supported by the following evidence from comparison of KO and WT cells: (1) the increased QS was fully rescued by transfecting dyn1-WT but not GTPase-deficient (dominant-negative mutation dyn1-K44A) ( Fig. 1 J, K ) or dyn2-WT (Fig. 2) into ACCs; (2) TIRF imaging of single-vesicle NPY-release showed that dyn1-KO increased the ratio of FFL to KAR events (Fig. 4A-I ); (3) the temporal kinetics was faster in FFL than KAR in WT cells, but not KO cells, indi-   . Cells were from the same 4 WT mice and 4 KO littermates. Scale bar, 5 m. E, F, Statistics of 10 KD TMR-dextran puncta taken up by WT (n ϭ 9 cells) and dyn1-KO cells (n ϭ 12 cells) without K ϩ stimulation (E); statistics of 40 KD TMR-dextran puncta taken up by WT (n ϭ 16 cells) and dyn1-KO cells (n ϭ 18 cells) without K ϩ stimulation (F), the large (40 KD) TMR-dextran was more selectively taken up by KO than by WT cells. Cells were from littermates (3 WT and 3 KO). Data are presented as median with interquartile range, *p Ͻ 0.05, two-tailed unpaired t test. G, Model of fusion pore and release modes regulated by dyn1. Left, Diagram depicting vesicle fusion with the plasma membrane while the fusion pore expansion is restricted by dyn1, resulting in subquantal release (KAR). Right, in the absence of dynamin 1, the fusion pore expands more, resulting in full-quantal release ("full-fusion"). Data are presented as median with interquartile range (NS for p Ͼ 0.05, *p Ͻ 0.05, two-tailed unpaired t test).
cating that KO led to a d p as large as FFL (Fig. 4 M, N ); (4) more vesicle content was retained after release in WT cells (Fig. 5); (5) dyn1-KO increased the d p from Ͻ9 nm (range 5-9 nm) to Ͼ9 nm, as revealed by the uptake of nanoparticle dextrans (Fig. 6); and (6), importantly, the EM-based vesicle size and the expression of VMAT1 and TH did not differ between KO and WT cells (Fig. 1 F, G,I ). Together, these data demonstrated that dyn1 maintains most of the Ca 2ϩ -induced exocytosis as subquantal release by restricting the d p to Ͻ9 nm in native WT ACC cells.
In a previous study, dysbindin, a schizophrenia susceptibility protein, was found to regulate QS in both ACCs and hippocampal synapses (Chen et al., 2008). However, its regulatory pathway must be different from that of dyn1 because vesicle sizes are increased in dysbindin-KO cells but not in dyn1-KO cells (Fig.  1 F, G). Although a caveat of off-target effects must always be considered with pharmacological methods, previous studies have reported evidence for the involvement of dynamin in QS González-Jamett et al., 2010Jackson et al., 2015). Anantharam et al. reported that overexpression of dyn1-GTPase mutations modulates fusion pore flickering (Anantharam et al., 2011). Only in a small number of vesicles did exocytosis colocalize with dyn1 during and after vesicle fusion. In the present work, however, even the top 10% of the largest release events was regulated by dyn1, suggesting that all release events are subquantal. In addition, dual color TIRFM imaging of dyn1 and NPY-pHluorin revealed Ͼ50% of KAR events with dyn1 accumulated at the fusion site (data not shown). One possibility for the difference could be that we used dyn-KO versus their transfection using Ca 2ϩ phosphate. Based on single vesicle exocytosis of NPY-GFP imaged by TIRF in chromaffin cells, Almers et al. have proposed that, in 20% of exocytotic events, the cargo proteins (including NPY-GFP) are retained in the vesicle after resealing the fusion pore (Perrais et al., 2004). However, they assumed that catecholamine release is full quantal in all events (Perrais et al., 2004). dyn1 is widely expressed in all neuronal cells (Ferguson et al., 2007; and increases vesicle size in EM and peak amplitude of mEPSC and mIPSC dyn1-KO synapse in physiological recordings (Ferguson et al., 2007;Lou et al., 2008;Raimondi et al., 2011). dyn1-KO revealed that probably all previously assumed "quantal release" are subquantal in chromaffin cells (Figs. 1, 2, 3). Future works should determine the following: (1) whether dyn1 produces also subquantal release in synapses via small clear vesicles and (2) (6) what is the diversity between dyn2 in pancreatic ␤ cells (Fan et al., 2015) and dyn1 in ACCs (Fig. 2).
The present study provides compelling evidence that most (if not all) Ca 2ϩ -induced exocytosis events occur via the subquantal mode to release catecholamine content (Յ40%) in ACCs. dyn1 induces subquantal KAR release mode by restricting the d p so that Ն60% of catecholamine cannot be released before the fusion pore recloses. Our work provides an example from chromaffin cells, where Bernard Katz's hypothesis of Ca 2ϩ -dependent transmitter release is revised from the "all-or-none/quantal" to the "subquantal" mode, implying that subquantal release could also exist in other secretion areas including neurotransmissions.