J. Neurochem. (2010) 113, 943–951.
Mammalian motor nerve terminals contain hundreds of thousands of synaptic vesicles, but only a fraction of these vesicles is immediately available for release, the remainder forming a reserve pool. The supply of vesicles is replenished through endocytosis, and newly formed vesicles are refilled with acetylcholine through a process that depends on the vesicular acetylcholine transporter (VAChT). During expression of short-term plasticity, quantal release can be increased, but it is unknown whether this reflects enhanced recruitment of vesicles from the reserve pool or rapid recycling. We examined spontaneous and evoked release of acetylcholine at endplates from genetically modified VAChT KDHOM mice that express approximately 30% of the normal level of VAChT to determine steps rate-limited by synaptic vesicle filling. Quantal content and quantal size were reduced in VAChT KDHOM mice compared with wild-type controls. Although high-frequency stimulation did not reduce quantal size further, the post-tetanic increase in end-plate potential amplitude or MEPP frequency was significantly smaller in VAChT KDHOM mice. This was the case even when tetanic depression was eliminated using an extracellular solution containing reduced Ca2+ and raised Mg2+. These results reveal the dependence of short-term plasticity on the level of VAChT expression and efficient synaptic vesicle filling.
miniature endplate potential
vesicular acetylcholine transporter
Neurotransmission at chemical synapses is initiated by an action potential in the pre-synaptic nerve terminal that opens Ca2+ channels and triggers exocytosis of neurotransmitter. The post-synaptic impact of each event depends on the number of quanta released, the amount of neurotransmitter per quanta, and the post-synaptic sensitivity. All three of these factors, and thus the overall efficiency of neurotransmission, depend on the recent firing activity of the neurons involved, a phenomenon referred to as synaptic short-term plasticity (Zucker and Regehr 2002). Different forms of short-term plasticity are recognized based on the pattern of activity required for their generation and the characteristic time course of their decay. At the frog neuromuscular junction, short-term increases in neurotransmitter release have been classified as facilitation, augmentation, and potentiation (Magleby and Zengel 1982). Facilitation can be generated by a single conditioning stimulus and decays with a time constant of less than 1 s. In contrast, augmentation and post-tetanic potentiation follow repetitive stimulation and last for seconds or minutes, respectively. Qualitatively similar phenomena have been described at the mammalian neuromuscular junction (Gage and Hubbard 1966; Bain and Quastel 1992), autonomic synapses (Zengel et al. 1980), as well as many central synapses (Zucker 1989; Fisher et al. 1997).
Potentiation of end-plate potential (EPP) amplitude could be caused by an increase in size of the quanta or an increase in the number of quanta that make up each EPP. At the frog neuromuscular junction, repetitive stimulation did not alter quantal size measured as the miniature endplate potential (MEPP) amplitude (Zengel and Sosa 1994). Later measurements of miniature endplate current charge integrals indicated a reduction in the amount of acetylcholine (ACh) per quanta after repetitive stimulation (Naves and Van der Kloot 2001). As quantal size evidently is not increased by tetanic stimulation, this leaves an increase in quantal content as the likely cause of EPP potentiation. Quantal content reflects release probability (p) and number of release sites or vesicles (n) both of which change during a conditioning train and in the post-tetanic period. The vertebrate motor nerve terminal contains a large number of synaptic vesicles, but at a given moment, only a small fraction participate in evoked neurotransmitter release, the remainder comprising a reserve pool (Rizzoli and Betz 2005). It is therefore evident that during periods of increased release, vesicle recycling or recruitment from the reserve pool must be augmented to avoid depletion of release-ready synaptic vesicles.
There are many steps in the pathways that prepare synaptic vesicles for release and subsequently recycle them for reuse (Sudhof 2004). Active transport of ACh into cholinergic vesicles is mediated by the vesicular ACh transporter (VAChT) (Erickson et al. 1994; Roghani et al. 1994). Previous experiments demonstrate accelerated depression of synaptic currents and loss of quantal size when nerve terminals are stimulated in the presence of vesamicol, an inhibitor of VAChT (Searl et al. 1991; Prior et al. 1992).Vesicle filling by VAChT represents an essential step for neuromuscular function, as genetically altered VAChT KDHOM mice with reduced expression of VAChT are myasthenic (Prado et al. 2006) and VAChT knock-out mice are not viable because of a lack of neuromuscular function (de Castro et al. 2009).
In the present work, we tested the consequence of reduced VAChT on quantal size and quantal content at rest and after tetanic stimulation in neuromuscular junctions from VAChT KDHOM mice that express approximately 30% of the normal levels of VAChT to investigate how deficits in vesicular transport affect quantal release of ACh.
Generation of VAChT KDHOM mice has been described in detail (Prado et al. 2006). Animals were housed in groups of 3–5 mice/cage in a temperature-controlled room with a 12 : 12 light : dark cycles and food and water were provided ad libitum. All studies were performed in accordance with protocols approved by the local institutional animal use and welfare committee (CETEA-UFMG protocol 073/03) and followed the NIH guidelines for care and use of animals in research. Experiments were performed ‘blind’ in the sense that the genotype of the animal was unknown until all experiments and analysis were complete.
Isolation of hemi-diaphragm
Mice were killed by decapitation and the diaphragm with attached nerve rapidly removed and pinned in an acrylic chamber on a bed of silicone elastomer (Sylguard, Dow Corning, Midland, MI, USA). The preparation was continuously perfused at a rate of 2–3 mL/min with physiological solution containing (in mM): NaCl (137), NaHCO3 (26), KCl (5), NaH2PO4 (1.2) glucose (10), CaCl2 (2.4), and MgCl2 (1.3), pH adjusted to 7.4 after gassing with 95% O2/5% CO2. In some experiments, quantal content was reduced by changing the concentrations of CaCl2 and MgCl2 as described in the Results. Experiments were conducted at room temperature (22–24°C). To avoid muscle contraction during nerve stimulation, (+)-tubocurarine (curare, 0.8–2.9 μM, Sigma-Aldrich) was included in the bath or the muscle fibers were cut (Barstad and Lilleheil 1968). Salts were purchased from Sigma-Aldrich or Vetec Quimica Fina (Rio de Janeiro, Brazil).
Recordings were made using an Axoclamp-2A 10× pre-amplifier (Molecular Devices, Sunnyvale, CA, USA). Microelectrodes were fabricated from borosilicate glass (model 1B150F-4, WPI, Sarasota, FL, USA) using a PN-30 puller (Narishige, Tokyo, Japan) and had resistances of 8–15 MΩ when filled with 3 M KCl. Microelectrodes were inserted into the muscle fiber near the end-plate region as judged by the presence of MEPPs with rise times < 1 ms. Signals were band-pass filtered at 0.1–5 KHz and further amplified 10–100× as necessary. In addition, the membrane potential was recorded for correction of driving force and non-linear summation of EPPs. Data were acquired on a PC computer equipped with a National Instruments A/D board and running Strathclyde Electrophysiology Software (kindly provided by John Dempster, University of Strathclyde). MEPP and EPP amplitudes were corrected to a standard resting potential of −60 or −35 mV when the cut-fiber technique was used and corrected for non-linear summation when EPP amplitude exceeded 10% of the driving force assuming a reversal potential of 0 mV for synaptic current (McLachlan and Martin 1981).
For stimulation, the nerve was drawn into a suction electrode and stimulated with suprathreshold voltage pulses (0.1 ms duration). Contractions were blocked by submaximal concentrations of d-tubocurarine (curare, 0.8–2.9 μM; Sigma-Aldrich), or by recording in a low-Ca2+/high-Mg2+ bathing solution containing 0.6 mM Ca2+ and 7 mM Mg2+, or by cutting the muscle fibers (Barstad and Lilleheil 1968). Facilitation was evaluated using paired pulses (10 ms inter-pulse interval) applied at 0.1 Hz. Stimulus trains consisting of 200–3000 stimuli delivered at 30–50 Hz were used to study augmentation and post-tetanic potentiation. When more than one neuromuscular junction (NMJ) was evaluated in the same muscle, the minimum interval between stimulus trains was 10 min which was sufficient for full recovery. Quantal content was calculated by the direct method as (EPP/MEPP) where the average MEPP amplitude was calculated as the mean of at least 40 MEPPs (usually > 50). Coefficient of variation of EPP amplitudes was determined from 20 EPPs evoked at 0.5 Hz. In recordings using very low-Ca2+/high-Mg2+, Tyrodes solution quantal content was < 1 and permitted measurement of quantal content by the statistical methods of Del-Castillo and Katz (1954) with minor modifications. We applied stimuli once per second. EPPs were recorded following each stimuli, and MEPPs were captured during the period between stimuli, excluding the 40 ms that followed the stimulus. We measured Ffailure, the fraction of stimuli not followed by an EPP, and calculated quantal content as ln(Ffailure) (Martin 1966). We used this estimate of quantal content and the quantal size of the MEPPs measured at the same synapse to fit the observed EPP distributions with no free parameters. With the same data, we calculated quantal content by the direct method for comparison.
MEPPs were detected in the intracellular recordings using the template search algorithm in Clampfit 9.2 (Molecular Devices, Sunnyvale, CA, USA). The template was constructed by averaging three or four MEPPs identified by eye. All MEPPs identified by the program were inspected, and the template match threshold parameter was adjusted to minimize false positives. Data were exported from Clampfit and analyzed using custom macros written in Igor software (Wavemetrics, Lake Oswigo, OR, USA).
Data are presented as mean ± 95% confidence intervals. The 95% confidence intervals were calculated assuming a Student’s t distribution in the means. All results were means from between three and six (usually four) animals of each genotype. In each nerve–muscle preparation, between one and five (usually three) endplates were studied. Sample sizes used to calculate 95% confidence intervals were the number of endplates studied. Where appropriate, data were analyzed either by unpaired Student′s t-test or one-way anova followed by Tukey’s post-hoc test.
Quantal parameters during low-frequency stimulation
To examine effects of reduced VAChT expression on quantal ACh release, we measured EPPs in diaphragm neuromuscular junctions from WT and from VAChT KDHOM mice. A low frequency of simulation was used (0.5 Hz) which generated little synaptic plasticity. In an extracellular solution containing 1.2 mM Ca2+ and 1.3 mM Mg2+, evoked EPPs recorded from VAChT KDHOM animals were smaller than EPPs recorded from their age-matched WT controls (Fig. 1). Using the cut-fiber technique to avoid muscle contraction, WT EPPs averaged 11.9 ± 3.5 mV (95% confidence interval) whereas VAChT KDHOM EPPs were 5.5 ± 2.0 mV (p < 0.05). To avoid possible artifacts resulting from the muscle fiber depolarization that occurs with the cut-fiber technique, we also measured EPP amplitudes when muscle contraction was blocked with 2.4 μM d-tubocurarine (curare). The use of curare has the additional benefit of reducing EPP amplitudes sufficiently so that that correction for non-linear summation was unnecessary, thus eliminating possible artifacts because of uncertainty in synaptic current reversal potential (McLachlan and Martin 1981). In these conditions, EPPs were 1.47 ± 0.30 mV (95% confidence interval) in WT diaphragms and 0.95 ± 0.15 mV for VAChT KDHOM mice (p < 0.05).
The amplitude of the EPP reflects both quantal content, which is the mean number of quanta per EPP, and quantal size, which is the unitary effect of each quantum (Del-Castillo and Katz 1954). We measured quantal size as the average amplitude of > 40 MEPPs from each synapse. The distribution in MEPP amplitudes from one WT and one VAChT KDHOM synapse are shown in Fig. 1b. MEPP amplitudes approximated a Gaussian distribution in most cases (smooth curves in Fig. 1b). We measured EPPs as the average of 20 responses to nerve stimulation applied at 0.5 Hz. On average, the difference between EPP amplitudes in VAChT KDHOM versus WT mice was larger than expected based solely on the difference in quantal size (Fig. 1c).
The coefficient of variation of EPP amplitudes during 0.5 Hz trains was significantly larger in VAChT KDHOM mice (WT: 0.10 ± 0.028 vs. KD 0.18 ± 0.04, mean ± 95% confidence intervals, p < 0.05). This is an indication of smaller quantal content in the VAChT KDHOM mice because when the mean number of quanta per EPP is smaller, the relative impact of stochastic variations in the number of quanta released is larger. We measured quantal content by the direct method at each end-plate as the ratio of EPP to MEPP amplitude (Martin 1966), and observed it was markedly reduced in EPPs from VAChT KDHOM mice (Fig. 1d). Whereas, WT synapses released 21.5 ± 3.8 quanta per EPP (95% confidence interval), in VAChT KDHOM synapses the number was 12.5 ± 2.5 quanta per EPP (p < 0.05). We conclude that the reduction of EPP amplitude in VAChT KDHOM neuromuscular junctions reflects two factors, a reduction of quantal size and a reduction of quantal content.
The observed reduction in quantal content could reflect a difference in release probability between WT and VAChT KDHOM synapses. To examine release probability, we measured paired-pulse facilitation with an interval between stimuli of 10 ms (Fig. 2). With an extracellular Mg2+ concentration of 1.3 mM, facilitation was 29 ± 5% (95% confidence interval) in WT synapses and 34 ± 6% in VAChT KDHOM (not significantly different). To increase the sensitivity of this measurement, we raised Mg2+ to 2.5 mM to reduce release probability. In this case, facilitation was increased as expected because of the lower release probability (Mallart and Martin 1968), but there remained no significant difference in facilitation between WT and knock-down synapses (41 ± 6% vs. 44 ± 8%), suggesting that basal release probability was similar for the two genotypes.
Responses to tetanic stimulation
We next examined changes in EPP amplitude during high-frequency stimulation. Muscle contractions were blocked by curare. As shown in Fig. 3a, when the stimulation frequency was increased from 0.3 to 30 Hz, EPP amplitude rapidly fell within the first seconds to about 50% of baseline, and slowly declined further over the course of the 90 s stimulus. Upon return to 0.3 Hz stimulation, EPP amplitudes rapidly recovered. WT synapses exhibited clear post-tetanic potentiation of EPPs that returned to initial values with a time constant of approximately 30 s (Fig. 3a, black solid line). In VAChT KDHOM synapses, the peak EPP after recovery from tetanic stimulation rarely exceeded its initial value. Nonetheless, it was possible to define a small post-tetanic potentiation in these synapses as an exponential decrease of EPP amplitudes in the 100 s after cessation of tetanic stimulation (Fig. 3a, gray solid line) that had a similar time course but was smaller than WT (p < 0.05).
The peak EPP amplitude following tetanic stimulation likely reflects a balance between augmentation/post-tetanic potentiation and residual tetanic depression (Wilson and Skirboll 1974) both of which may be affected in VAChT KDHOM animals. To distinguish these two forms of short-term plasticity, we varied the amount of depression that occurred during the tetanic stimulation by varying the ratio of Ca2+ to Mg2+ concentrations (Ca2+/Mg2+) in the bathing solution (Fig. 3 b and c). As the ratio decreased, steady-state depression at the end of the train decreased (Fig 3b), presumably because the reduction in release probability produces less vesicle pool depletion (Mallart and Martin 1968). In all three conditions, post-tetanic potentiation was greater in WT synapses than in VAChT KDHOM (Fig. 3c, p < 0.05).
Following a high-frequency stimulus train, the frequency of MEPPs is transiently increased and decays back to resting levels with a time course similar to that of augmentation/post-tetanic potentiation of EPP amplitudes (Nussinovitch and Rahamimoff 1988; Bain and Quastel 1992; Zengel and Sosa 1994). To determine if post-tetanic potentiation viewed in this way was also altered in the VAChT KDHOM animals, we repeated the experiments using the cut-fiber preparation without curare. This allowed us to measure MEPPs before and after the tetanic stimulation. As shown in Fig. 4, tetanic stimulation resulted in a large increase in MEPP frequency lasting approximately 30 s in WT mice. This type of potentiation was dramatically reduced in VAChT KDHOM mice compared with WT controls (240 ± 125% in WT vs. 92 ± 53% in VAChT KDHOM; mean ± 95% confidence intervals, p < 0.05). In both WT and VAChT KDHOM synapses, MEPP amplitude was unaltered by the tetanic stimulation (Fig. 4b).
Comparison of MEPPs and uniquantal EPPs
The comparison between MEPP and EPP amplitudes assumes these two event arise from the same pool of synaptic vesicles. Although this has long been a tenet of quantal analysis, recent evidence in hippocampus has appeared both for and against the existence of separate pools of vesicles underlying spontaneous and evoked release (Sara et al. 2005; Groemer and Klingauf 2007; Fredj and Burrone 2009). We therefore examined quantal content in low-Ca2+/high-Mg2+ bathing solution. These experiments allowed direct comparison between the size of MEPPs and uniquantal EPPs, permitted analysis of the distribution of EPP amplitudes, and allowed for quantal content calculations based on the statistical methods of (Del-Castillo and Katz 1954) versus the direct method (Fig. 5).
Results from a representative WT NMJ are shown in Fig. 5(a–d). MEPP amplitude distributions and time integral distributions (‘areas’) approximated a Gaussian distribution (p > 0.1, Kolmogorov–Smirnov test for the vertical distance between two cumulative probability distributions). Figure 5c shows averaged EPPs recorded at the same synapse on the same time scale as Fig. 5a. Two sets of EPPs were grouped for averaging. The smaller waveform is the average of 182 probable uniquantal EPPs and the larger waveform is the average of 60 probable diquantal EPPs. These EPPs were selected by first calculating the mean quantal content as −ln(FFailure) where Ffailure is the fraction of stimuli that failed to evoke release (Martin 1966). Events (including failures) were ranked by amplitude, and selected using the predictions of the Poisson distribution with the mean given by the measured mean quantal content. Note that the mean uniquantal EPP is essentially identical to the mean MEPP (compare Fig. 5a and c), and the mean diquantal EPP has twice the amplitude. We conclude that MEPPs and uniquantal EPPs are indistinguishable.
To test the assumption that the EPP reflects the sum of one or more MEPPs, we constructed EPP amplitude and area distributions (Fig. 5d, solid lines) and fit them with a model distribution with no free parameters (Fig 5d, broken lines). At each synapse, mean quantal content was measured as −ln(Ffailure). Then, a set of 10 000 simulated EPPs were generated with quantal content given by a Poisson distribution matched to the data. To build up the EPPs, MEPP amplitudes or areas were randomly selected with replacement from the population of MEPPs measured at the same synapse. Based on the agreement of the data and the simulation, we find no evidence for a separate pool of quantal generating MEPPs and EPPs. Similar results were obtained from NMJs from three WT animals and four VAChT KDHOM animals, except that mean quantal content was lower in the VAChT KDHOM animals. In this set of experiments, MEPP frequency was 0.42 ± 0.11 s−1 in WT synapses and 0.27 ± 0.08 s−1 in VAChT KDHOM (p < 0.05).
We next compared quantal content measured by the statistical method of counting failures versus the direct method (EPP/MEPP). If EPPs are built up of quanta from the same pool that generates MEPPs, the two estimates should give the same result. Figure 5e and f summarizes results from 11 WT synapses obtained from three animals and eight VAChT KDHOM synapses obtained from four animals. In Fig. 5e, open symbols represent data obtained by considering MEPP and EPP amplitudes whereas the filled symbols considered MEPP and EPP areas. The solid line has unity slope. The mean and 95% confidence limits for the three quantal content estimates (method of counting failures, direct method based on amplitudes, and direct method based on amplitudes) shown in Fig. 5f indicate reduced quantal content in VAChT KDHOM mice.
In this study, we report changes in quantal ACh release in VAChT KDHOM mice in which brain and spinal cord levels of VAChT (both mRNA and protein) are reduced to 30% of WT levels whereas other cholinergic markers including the synthesizing enzyme choline acetyltransferase and the high-affinity choline transporter CHT1 are normal. Although these mice are myasthenic (Prado et al. 2006) and have reduced parasympathetic tone (Lara et al. 2010), they are viable and survive to adulthood. Accordingly, at the neuromuscular junction MEPPs are readily detectable and nerve stimulation generates large synchronous EPPs. VAChT KDHOM mice thus provide a model to study the effect of steady-state reductions in VAChT, and hence synaptic vesicle filling, on quantal release of ACh, and to test for the dependence of short-term plasticity on efficient synaptic vesicle filling. Previously, we showed enhanced depression of quantal content during very high-frequency stimulation (100 Hz) and 2.4 mM Ca2+ to enhance quantal output (Prado et al. 2006). Here, we used lower frequency stimulation and lower Ca2+ concentrations that reduce quantal content and therefore depression to examine quantal content at rest and during post-tetanic potentiation.
Many of our results concern changes in quantal content, the average number of quantal released by each pre-synaptic action potential. We use this term in the original sense of the number of electrophysiologically detectable quanta that make up the EPP. A semantic issue arises when one equates the quanta with synaptic vesicles and considers the possibility of empty vesicles. In this case, the term ‘quantal content’ is ambiguous; it could mean the number of quanta of neurotransmitter released or it could mean the total number of vesicles released. We will continue to use ‘quanta’ and ‘quantal content’ in the electrophysiological sense, and will use phrases like ‘number of vesicles released’ when needed. This distinction is relevant to our results on facilitation because changes in quantal content are often associated with changes in paired-pulse facilitation. The classic example is the association between quantal content and facilitation when release probability is manipulated by changing the extracellular Ca2+/Mg2+ ratio (Mallart and Martin 1968). We observed lower quantal content in VAChT KDHOM animals without any change in facilitation. This suggests that quantal content is lower not because of reduced release probability, but rather because of a smaller supply of quanta.
Baseline quantal parameters under low-frequency stimulation
In vivo, VACh does not achieve thermodynamic equilibrium with cytoplasm ACh (Parsons 2000). The specific molecular mechanisms by which quantal size may be regulated are largely unknown, but the activity of vesicular neurotransmitter transporter is an obvious candidate, and experimental evidence exists both for and against this possibility, as reviewed by Van der Kloot (2003). Over-expression of VAChT caused increased ACh uptake by PC12 cell post-nuclear homogenates (Varoqui and Erickson 1996) and larger spontaneous excitatory post-synaptic potentials in co-cultured Xenopus spinal neurons and myocytes (Song et al. 1997). In addition, increasing the concentration of ACh also increased its uptake into PC12 cell membranes (Varoqui and Erickson 1996). Similar observations have been made for uptake of glutamate into synaptic vesicles from CNS neurons (Ishikawa et al. 2002; Wilson et al. 2005). All of these results are consistent with the idea that vesicle filling represents a dynamic steady state between uptake and passive leak. Our present results provide further support for this model, as reduced VAChT expression reduced MEPP amplitude, although we cannot rule out the possibility of differences in post-synaptic sensitivity in the VAChT KDHOM mice. To overcome the possibility of changes in post-synaptic sensitivity, data on quantal content were obtained either by measuring EPP and MEPPs at the same synapse permitting calculation by the direct method or by the statistical methods of Del-Castillo and Katz (1954). In either case, results are independent of post-synaptic sensitivity.
The reduction of basal quantal content in VAChT KDHOM neuromuscular junctions was surprising because the low stimulation frequency used produces minimal synaptic rundown and vesicle recycling therefore is able to keep up with release. Moreover, previous experiments with FM1-43 indicated the quantity of recycling vesicles was normal in NMJs from VAChT KDHOM mice (Prado et al. 2006), and experiments measuring paired-pulse facilitation or initial EPP rundown during high-frequency trains revealed no evidence for changes in release probability. One possibility to explain these data is the presence of a population of synaptic vesicles that do not contain VAChT and thus are unable to take up much ACh. Cholinergic synaptic vesicles in the electric organ of Torpedo contain on average about four copies of VAChT (Bahr and Parsons 1986). Cholinergic vesicles in mammalian NMJs have about half the diameter (Volknandt and Zimmermann 1986) and therefore one-fourth the membrane surface area of Torpedo vesicles. The distribution in the number of copies of VAChT per mammalian vesicle is unknown, but if the mean number is small it is expected that some vesicles will contain zero copies of VAChT and that this fraction of ‘silent vesicles’ will be higher in VAChT KDHOM neuromuscular junctions.
Although vesicles without VAChT may be poorly detected by electrophysiological recordings, much evidence suggests they will still recycle normally. Studies at the neuromuscular junction with the dye FM1-43 indicated that cholinergic synaptic vesicle recycling continued to occur in nerve terminals stimulated in the presence of vesamicol (Parsons et al. 1999). Similar results were obtained when choline uptake into the terminal was blocked by hemicholinium (Ceccarelli and Hurlbut 1975). In hippocampus slices, blocking the vATPase reduced the amplitude and frequency of mEPSPs and mEPSCs, but did not alter vesicle recycling as recorded by FM dye detaining, leading to the conclusion that depleted or empty vesicles recycle normally (Zhou et al. 2000). Moreover, recent experiments using transgenic mice expressing synaptopHluorin indicated continued strong exocytosis even as synaptic depression reached 100%. It was suggested that exocytosis without a post-synaptic potential reflects fusion of ‘empty’ synaptic vesicles or other acidified membranous structures (Tabares et al. 2007). Together these results indicate that empty vesicles compete with normal vesicles for the exocytosis apparatus, and that in the VAChT KDHOM mice an increase in vesicles without VAChT reduces the apparent quantal content of EPPs and the frequency of MEPPs.
Vesicle pools in the mouse neuromuscular junction
Many synapses contain multiple anatomically and functionally distinct pools of synaptic vesicles and multiple recycling pathways (Rizzoli and Betz 2005). At the mouse neuromuscular junction, about 1% of vesicle reside in a rapidly releasable pool and are preferentially released by low-frequency stimulation or during the initial phase of high-frequency stimulation (Perissinotti et al. 2008; Zefirov et al. 2009). Release during prolonged high-frequency stimulation may be because of recruitment of vesicles from a reserve or as a result of rapid recycling of recently released vesicles. These two possibilities have been distinguished by combining post-synaptic recordings to measure released quanta with FM dye destaining to measure exocytosis of a previously labeled vesicle. Comparison of exocytosis by these two techniques at the mouse NMJ indicates preferential rapid recycling back to the readily releasable pool during tetanic stimulation (Perissinotti et al. 2008; Zefirov et al. 2009). Thus, tetanic stimulation specifically increases the demand on refilling of synaptic vesicles, not only because more vesicles are released but also because they are replaced by rapid recycling.
Role of VAChT in expression of post-tetanic potentiation
Although a population of silent vesicles may explain the lower basal quantal content of EPPs and frequency of MEPPs in VAChT KDHOM NMJs, it cannot explain the loss of post-tetanic potentiation because we followed the usual practice of expressing post-tetanic transmitter release (EPP amplitudes, or MEPP frequency) in proportion to control values measured at the same synapse before the conditioning train. Silent vesicles are thus factored out of this calculation. Rather, the data suggest that the level of VAChT activity determines the ability of the synapse to increase its output during potentiation. Possible mechanisms for the loss of potentiation in VAChT KDHOM mice include reduced final loading of synaptic vesicles (Naves and Van der Kloot 2001) and longer vesicle refilling times expected if the average number of copies of VAChT per vesicle is reduced.
If VAChT KDHOM synapses show a loss of potentiation because they cannot keep up with the increased demands for vesicle refilling, one might expect a reduction in quantal size at the end of a tetanic stimulation. It was therefore surprising that we did not observe a reduction in the amplitudes of MEPPs after tetanic stimulation. It is generally assumed that MEPPs and evoked quanta come from the same vesicle pool, although in cultured hippocampal neurons there is evidence for and against a separate pool of spontaneously releasing vesicles that recycle separately from the evocable pool (Sara et al. 2005; Groemer and Klingauf 2007; Fredj and Burrone 2009). Previous studies showed vesamicol reduces post-tetanic potentiation without reducing quantal size (Maeno and Shibuya, 1988). Recently, evidence for exocytosis of empty vesicles at the neuromuscular junction has been presented after complete EPP rundown during high-frequency stimulation (Tabares et al. 2007). In this study, the authors also did not observe a reduction in the size of asynchronously or spontaneously released quanta during or after the stimulus train, as may have been expected if vesicle recycling and refilling could not keep up with demand. These data indicate that lack of change in quantal size after tetanic stimulation cannot be taken as evidence for normal vesicle recycling and refilling.
In mice NMJs, mitochondria sequester Ca2+ during high-frequency stimulation and its release in the post-tetanic period correlates with enhanced transmitter release (García-Chacón et al. 2006). This behavior suggests a mechanism to activate transmitter release for post-tetanic potentiation. However, for calcium to increase release, recycling and refilling of the pool of vesicles involved in potentiation need to be efficient. Because vesicle loading is mediated by VAChT, refilling of rapidly recycling vesicles should be preferentially affected in VAChT KDHOM mice. Experiments with fluorescent dyes at the frog neuromuscular junction have indicated the existence of two pools of synaptic vesicles, a smaller capacity, readily releasable population that recycles rapidly, and a larger reserve population that is reluctantly releasable and recycles more slowly (Richards et al. 2000, 2003). Our data suggest that post-tetanic potentiation increases the fusion of vesicles that have rapidly recycled. When VAChT activity is reduced, some of these vesicles refill with ACh and some do not.
We thank Dr William Van der Kloot and Dr Stanley M. Parsons for valuable comments on a draft of this study. Supported by FAPEMIG (CBB APQ-4536-4.01/07 and PRONEX), NIH-Fogarty grant R21 TW007800-02, CIHR (MOP-89919, VFP. and MAMP), Heart and Stroke Foundation of Ontario (NA 6656, MAMP and VFP), CNPq (Aging and Mental Health Programs, Universal 474445/2007-0), Instituto do Milenio Toxins/MCT.