UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Quantification and timing of processes involved in stimulus-secretion coupling at the mouse neuromuscular… Bain, Allen Ian 1993

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Item Metadata


831-ubc_1994-893214.pdf [ 6.25MB ]
JSON: 831-1.0088324.json
JSON-LD: 831-1.0088324-ld.json
RDF/XML (Pretty): 831-1.0088324-rdf.xml
RDF/JSON: 831-1.0088324-rdf.json
Turtle: 831-1.0088324-turtle.txt
N-Triples: 831-1.0088324-rdf-ntriples.txt
Original Record: 831-1.0088324-source.json
Full Text

Full Text

QUANTIFICATION AND TIMING OF PROCESSES INVOLVED IN STIMULUSSECRETION COUPLING AT THE MOUSE NEUROMUSCULAR JUNCTION. by ALLEN IAN BAIN B.Sc, The University of British Columbia, 1985  A THESIS SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY  in FACULTY OF GRADUATE STUDIES Department of Pharmacology and Therapeutics Faculty of Medicine  We accept this thesis as conforming to the required standard  THE UNIVERSITY OF BRITISH COLUMBIA September 1993  (c)  Allen Ian Bain, 1993  In presenting this  thesis  in partial fulfilment  of  the  requirements  for  an advanced  degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department  or  by  his  or  her  representatives.  It  is  understood  that  copying  or  publication of this thesis for financial gain shall not be allowed without my written permission.  (Signature)  Department of  r%***~""6.C<o  The University of British Columbia Vancouver, Canada  Date  DE-6 (2/88)  ^ g A / 9 ^ T / 7 ^  (QJ >-± <2~^f  [<(j^Cyp££<fr^  ii  Abstract At the mammalian neuromuscular junction, perhaps the most  studied  synapse, many  aspects  of  neurotransmitter  release and stimulation-induced enhancement of release are still poorly understood. release, the Ca^  Central hypotheses include:  about  hypothesis (del Castillo and Katz, 1954),  and the Ca'' -voltage hypothesis (Parnas and Parnas, 1988), and about enhancement, the residual Ca^ Miledi, 1968).  hypothesis (Katz &  In the present work, these hypotheses were  tested by analysis of the magnitude and timing of release with the technical advantages of computer-assisted analysis of data for large numbers of stimuli and responses and an emphasis on the relative magnitude of phasic and non-phasic release components. In  mouse  nerve-diaphragm  in  vitro,  phasic  neurotransmitter release evoked by action potentials grew with r«0.1 ms and decayed with r«0.3 ms, consistent for Ca 2 + , Sr 2 + and Ba 2 + .  Non-phasic release decayed, with a  polyphasic time course that varied with the divalent cation. The time course of the opening of voltage-dependent presynaptic divalent cation channels underlying the release process was assessed using "tails" of raised MEPP frequency induced by trains of "direct" pulses (TTX present) in Ba 2 + containing solution.  Pulses exceeding 50 ms duration were  nearly equi-effective (by integral) to more brief pulses,  iii  indicating  that  this  Ca  channel  undergoes  little  inactivation. In  the  presence  of  Sr^  or  Ba^ ,  short  term  stimulation-induced enhancement of release was consistent with a simple "residual ion" model, with 'cooperativity' of 4, and decay of putative intracellular ion with r«200 ms or r«3 to 5s, respectively. In Ca  , facilitation  (short  term  enhancement) was  inconsistent with a residual ion model but could be resolved into two components:  a multiplicative component seen as an  about two-fold parallel increase in m and f m  for short  trains (decay r«80 ms), dependent on intracellular Ca'' and Ca^  influx, plus an additive component  (decay r«200 ms)  consistent with the residual ion model. Potentiation  (long  term  enhancement)  was  found  to  consist primarily in parallel of a parallel multiplication of phasic and non-phasic release with r<20 s.  It was absent  when  stimulation),  tetrodotoxin  suggesting  was  dependence  present  upon Na +  ('direct' influx  and  accumulation.  With prolonged tetani, non-phasic release increased further, in a manner consistent with gradually accumulating Ca^+.  iv  Table of Contents  Abstract  ii  Table of Contents  iv  List of Tables  ix  List of Figures  x  Acknowledgements Dedication  xii xiii  I. Introduction  1  A. Quantal neurotransmitter release  2  1. Quantal nature of release  2  2. Phasic and non-phasic release  2  3. Ca 2 + hypothesis  4  4. Fourth power model  6  5. Ca  7  -voltage hypothesis  B. The divalent agonists  9  1. Additive properties  11  2. Potency  13  3. Elimination kinetics  14  C. Time course of evoked release  16  1. Phasic release  16  2. Techniques of measurement  18  3. Non-phasic release  20  V  D. Enhancement processes  21  1. Overview  21  2. Retrospective  22  3. Facilitation  23  4. Augmentation  31  5. Potentiation  32  6. Present data  34  Methods  36  A. Mouse hemidiaphragm  37  B. Solutions and chemicals  37  C. Stimulus delivery  38  D. Stimulation protocol  41  E. Time constant of the nerve terminal  46  F. Data recording  48  G. Computer programs  49  1. Ba*  tails program  49  2. On line monitoring  50  3. Quantal deconvolution - off line  51  G. Data analysis  60  1. Estimation of m  60  2. Estimation of fm  62  3. Data averaging  63  4. Derivations  64  5. Enhancement calculation  67  6. Assumptions  68  7. Definitions  VI  Results A. Time course of stimulus-secretion coupling 1. Does the presynaptic Ca channel inactivate?  73 73  a) Studies with Ba 2 +  73  b) K+ channel blockers  79  c) Membrane time constant  82  2. Timing of phasic release  82  a) Minimum latency - direct pulses  82  b) Phasic release time course  84  c) Sources of variance in timing of phasic release  86  B. Stimulation-induced enhancement of release  92  1. Residual ion: Sr  92  a) Stimulation in Sr 2+  92  b) Estimation of n and z  96  c) Estimation of Sr 2 + entry  100  d) Comparison of Sr 2 + with Ca 2 + and Ba 2 +  101  e) Effect of BAPTA/AM  105  2. Facilitation in Ca  109  a) Time components  109  b) Ca dependency  121  c) Dependency on presynaptic stimulus  127  3. Deviations from the models  129  a) Decreasing stimulus  129  b) Increasing stimulus  130  c) Ultra fast facilitation  131  Vll  4. Potentiation  131  a) Ca 2 + dependency  131  b) Na+ dependency  139  Discussion A. Phasic release time course 1. the presynaptic Ca^  channel  143 143  2. similarities for divalent agonists temperature studies  144  3. model-fitting in retrospect  145  4. time course of BAPTA effects  149  B. Models of release enhancement  150  1. Residual ion model  150  2. Variations of the residual ion model  152  3. Multiplicative model  158  4. Combined model - multiplicative and additive  160  C. Mechanisms of multiplicative facilitation  160  1. presynaptic ion channels  160  a) action potential  160  b) membrane potential  161  c) Ca^  164  entry per pulse  2. Potentiation and facilitation  165  a) Shared mechanisms?  165  b) Intracellular Na+  167  c) Presynaptic proteins  169  D. Mechanism of additive components  170  viii  E. Ultra fast facilitation  172  F. Spontaneous release  173  1. Role in the combined model  173  2. Cause of spontaneous release  174  a) Ca2+-dependent  174  b) Ca2+-independent  174  V. Conclusion - Fitting data to a model  175  References  177  ix List of Tables Table 1A  calculated  c  o/ c p  values  for  various  m, f m and n. Table IB  30  a (n=4) for various values of facilitation and co/cp.  Table 2  Effect  30 of  period  of  integration  of  phasic  release on derived values of facilitation and residual calcium. Table 3  70  Time constant of the nerve terminal measured under various conditions.  Table 4  Dependence of multiplicative extracellular Ca 2+ .  83 facilitation on 122  X  List of Figures page Fig. 1  Random train stimulation.  43  Fig. 2  Nerve terminal z determination  47  Fig. 3  Automated quantal deconvolution.  56  Fig. 4  Typical latency histogram.  58  Fig. 5  Effect of pulse duration on "Ba entry".  74  Fig. 6  Non-linear relation of Ba entry and pulse duration for brief pulses.  77  Differential effectiveness of increases in pulse duration.  78  Fig. 7 Fig. 8A  Linearity of relationship between pulse duration and Ba entry in the presence of 4AP. 81  Fig. 8B  Linearity of response to prolonged pulses  81  Fig. 9  Latency distribution of an e.p.p. and its derivative.  85  Fig. 10  Latency distributions of all quanta and of first quanta 87  Fig. 11  Autocorrelation of quantal latencies within and among e.p.p.s. 88  Fig. 12  Effect of prolonged tetanus on quantal latencies. 90  Fig. 13  Buildup and decay of m and f m during and after a train of stimuli in Sr 2+ . 93  Fig. 14  Co-modulation of phasic and non-phasic release in Sr 2+ . 94  Fig. 15  Fourth root transform of f m buildup and decay in Sr 2+ . 97  Fig. 16  Estimation of r and n for Sr 2 + by best fit. 99  Fig. 17  Demonstration of the identity in time course of phasic release in Ca 2 + as in Sr 2+ . 103  Fig. 18  Relation between Sr 2 + entry and quantal content of the e.p.p. 104  xi Fig. 19  Effect of BAPTA-AM on r and apparent entry in Sr 2+ . 107  Fig. 20  Relative effect of BAPTA-AM on phasic and nonphasic release in Sr 2+ . 108  Fig. 21  Observed and predicted facilitation of m and fm by short trains in Ca 2+ . 110  Fig. 22  Facilitation in Ca using continuous stimulation with random intervals. 114  Fig. 23  Growth of facilitation in Ca^ : multiplicative and 'residual Ca 2+ ' components. 11  Fig. 24  Decay of facilitation in Ca  : multiplicative  and 'residual Ca 2+ ' components.  117  Fig. 25  Time course of decay of residual Ca^ .  120  Fig. 26  Ca 2 + dependence of facilitation.  125  Fig. 27  Lack of facilitation in Ca^ solution after loading with BAPTA. 126 Dependency of facilitation on depolarization amplitude. 128 Potentiation: a log-linear relation between release and tetanic stimulation frequency. 133  Fig. 28 Fig. 29 Fig. 30  Relative effects on m and fm of prolonged tetanic nerve stimulation. 135  Fig. 31  Plot of phasic delta fourth root over duration of a tetanus. 136  Fig. 32  Survival of potentiation in the absence of Ca entry. 138  Fig. 33  Lack of potentiation with 'direct' pulses.  141  xii Acknowledgements  I am grateful for financial support from the H. R. MacMillan Foundation.  For a firm grounding in the applied  science of pharmacology,  I am indebted  to each of the  faculty in the Department of Pharmacology and Therapeutics.  xiii Dedication To Dr. David Quastel, whose trail-breaking led to places both curious and delightful, but whose gait I was rarely able to match; and to Maria and Jason, who waited.  1  I. Introduction  2 A. Quantal neurotransmitter release 1. Quantal nature of release Since Dale presented evidence for chemical synaptic transmission  at the  junction of nerve and muscle using  bioassay (Dale et al, 1936), much descriptive work has been carried out on the nature of chemical neurotransmission. The quantal hypothesis  of  synaptic  neurotransmitter  release (del Castillo & Katz, 1954) held that a miniature end plate potential  (m.e.p.p.) recorded post-synaptically  with an intracellular microelectrode represented the nearly simultaneous  action  molecules, a  so-called  exocytosis  of  acetylcholine.  a  of  thousands  of  neurotransmitter  quantum, presumptively  single  presynaptic  vesicle  reflecting filled with  The end-plate potential produced by a nerve  impulse, being one to two orders of magnitude larger than a m.e.p.p., was postulated release of many quanta.  to be due to the  The primary evidence for this was  the observation that with Ca lowered  to  give  an  synchronized  e.p.p.  concentration in the bath of  about  5 mV  or  less  the  distribution of e.p.p. amplitudes became multimodal, with modes  corresponding  quite  closely  to  multiples  of  the  includes  quantal  average m.e.p.p. amplitude. 2. Phasic and non-phasic release Phasic  neurotransmitter  release that occurs action potential.  release  'in phase' with, and just after, an The definition of phasic release must  always be arbitrary, since release occurs stochastically and  3 there is a time course of decay of the depolarization-evoked Ca^  (calcium ion) transient and any other process which may  be initiated by the stimulus.  Thus, arbitrary binning (in  time bins of various widths) of release following a stimulus into  phasic  and  non-phasic  components  may  become  a  significant source of inaccuracy which should be accounted for in testing of content, m, stimulus.  any particular model.  Mean quantal  is the average number of quanta released per  Phasic release can alternatively be quantified as  m or as an instantaneous release rate. Non-phasic release, or spontaneous release, is defined as that quantal release which occurs in the absence of an obvious stimulus to the terminal.  Most workers extend this  definition to include all release, after a stimulus, that is not within the time window of intensely enhanced release that closely distinction  follows a particular between  phasic  and  stimulus. non-phasic  Thus, the release  is  entirely arbitrary and has historically depended greatly on the resolution of the method of measurement and of binning in time.  Non-phasic release can be quantified as the rate  of m.e.p.p. occurrence, fm, in defined time bins. Non-phasic release rates are often not measured or reported  for  two  reasons:  firstly, m.e.p.p.s  are not  measurable in curarized preparations or without the aid of a reliable computer counting method under conditions of very low f m and, secondly, evidence that m and f m arise from the  4 same release system is still quite recent  (Guan et al,  1988) . In the present studies, it is assumed that non-phasic release  in  some  way  indicates  a  status  of  release  probability for which spatial differences within the nerve terminal are minimal and decay or growth in time is slow relative to the time course of phasic release.  Thus, fm  (or, more generally, release rate R) measured at any time after  a  stimulus  intracellular Ca events  is  an  indication  of  the  presence  of  and/or any other modulator upon which the  leading to phasic release are then superimposed.  Based on this assumption, measurement of non-phasic release and its enhancement by  stimulation  is essential to the  testing of any hypothesis regarding phasic release and its enhancement, including the Ca"6  hypothesis, residual Ca"6  hypothesis, Ca-voltage hypothesis, as well as the multiplier hypothesis presented below (see DISCUSSION). 3. Ca hypothesis The importance of Ca  to neuromuscular function was  well established when del Castillo & Katz (1954) proposed an overall  scheme  for  the  mechanism  of  neurotransmitter  release, the "calcium hypothesis". Neurotransmitter release, as observed by postsynaptic intracellular recording at the rat neuromuscular junction, was closely coupled in time to an evoked axonal action potential and was strongly dependent on extracellular Ca 2+ . Katz and co-workers (del Castillo and Katz, 1954, Katz and  5 Miledi,  19 65a) postulated  that  the  mechanism  of  Ca* -  dependent neurotransmitter release began with invasion of the nerve terminal by the axonal action potential.  Upon  depolarization of the nerve terminal, a voltage dependent increase in Ca^ Ca^  permeability allowed a sudden influx of  down its approximately 10J-fold concentration gradient Ultimately, intracellular Ca 2 + acted to  into the terminal.  promote fusion of a discrete number of neurotransmitter vesicles  with  the  neuronal  membrane  neurotransmitter into the synaptic cleft.  releasing  Katz and Miledi  (1967) showed that the same effect could be initiated with depolarizing pulses substituting for action potentials. their experiments, an lontophoretic pulse of Ca'1  In  from a  focally placed micropipette was shown to elicit release only lf extracellular Ca^  was present at the nerve terminal at  the time of the nerve terminal action potential or direct depolarization. be  a  Evoked neurotransmitter release appeared to  consequence  of  depolarization  and  presence  of  extracellular Ca^  at the same time, suggesting an absolute  dependence on Ca^  entry into the nerve terminal.  More recently, it has been emphasized that this influx occurs through a suddenly opened 'channel', a water filled transmembrane pore, as opposed to another carrier mechanism. This is indicated by ion influx rate - in the order of 10^ ions/sec/channel (Hille, 1984). of intracellular sensitive  Ca 2 +  fluorochromes  activity before  According to measurements (for example, using and  after  excitation  Ca 2 + of  6 chromaffin  cells;  presynaptic Ca  review:  Plattner,  1989),  opening  of  channels results in a 1 to 2 order increase  in intracellular Ca 2 + concentration, from about 10 rest to about 10  M.  M at  Upon channel opening, the actual  local concentration of Ca^  at the intracellular mouth of  the Ca 2 + channel, probably close to the active sites for transmitter release (Robitaille et al, 1990), could be much higher than estimated from fluorochrome studies which have limited spatial and temporal resolution. 4. Fourth power model Quantal content, release within the period, after the stimulus, defined for phasic release (see later), is found to be proportional to the 4th power of extracellular Ca^ concentration (Dodge & Rahamimoff, 1967), over a wide range of [Ca 2+ ]. 1  The  This result supports three hypotheses: relation  between  release is 4th power.  intracellular  Ca2+  and phasic  It is observed that changing  absolute entry (by addition of competitive blockers of the Ca^  channel or by large changes in extracellular  Ca 2+ ) does not appear to change the power relationship (Guan et al, 1988). 2  Intracellular Ca  does not approach saturation at its  receptor site when quantal content is low because of 7+  low extracellular [Ca  ].  The degree of saturation is  unknown for m' s larger than about 4 primarily due to the difficulties in data interpretation because of the  7 competing process of depletion, the rundown of quantal content seen at high quantal output rates. 9+  3  The entry of Ca  during an action potential does not  apparently saturate the conductance of the channel. (This  assuming  relationship  that  consistency  indicates  a  of  linear  the  4th  entry  power  process.)  Instead, entry appears to be directly proportional to 9+  extracellular Ca Ca 2 +  , over a wide range of below-normal  concentrations.  In  fact, the  apparent  power  9+  relation between Ca  entry and transmitter release is  observed to be less than 4 under conditions where the Ca  influx per channel (determined by extracellular 9+  Ca*  9+  concentration  and  thus  transmembrane  Ca  gradient) is very great, consistent with a model which accounts  for  stochastic  heterogeneity  among release  sites (Quastel et al, 1992). Under conditions in which the rate limiting step in the activity of the "divalent agonist" (Ca2+, Sr 2 + or Ba 2+ ) is one of binding intracellularly to the receptor, non-phasic release also grades with the 4th power of the extracellular concentration, as discussed later. 5. Ca-voltage hypothesis •  9+  Parnas challenged the Ca which  a  brief  nerve  hypothesis with a model in  terminal  depolarization  (action  potential or extracellular current electrode) enhances the 9+  effectiveness of intracellular Ca* or is itself directly linked to neurotransmitter release (Parnas et al, 1986;  8 review:  Parnas  &  intracellular Ca'  Parnas,  1988).  In  this  model,  plays a permissive role in a process  which is predominantly due to a mechanism other than that described by the Ca' hypothesis.  Simply stated, the Parnas  model proposes that the presynaptic depolarization caused by an action potential or current pulse leads to an e.p.p. in two ways: (1) voltage dependent Ca  channels are opened and  Ca 2 + is admitted from the extracellular medium, and (2) an intracellular  Ca' -binding  release  site  undergoes  a  transition from an inactive form which does not bind Ca' and does not promote release to a form which does. the Ca-voltage model proposes  Thus,  that active release sites .  ?+  . . .  which are capable of binding Ca' , S, are in equilibrium with inactive sites, T, with the equilibrium shifting toward S during a depolarization.  S sites are then capable of  eliciting release via binding Ca'  to form a complex which  can promote exocytosis of a vesicle.  One of the attractions  of this model is that facilitation of m and fm, measured in the presence of extracellular Ca' , can then be ascribed to residual intracellular Ca' . In the present work, the Parnas "Ca-voltage" model was tested in two 1  ways:  Release under very  low extracellular Car+  and  channel blocked conditions (Bain & Quastel, 1988);  Ca 2 + if  there were a direct effect of voltage on release which did  not  require  intracellular Ca'  transient  high  concentration  of  (or other divalent agonist), phasic  9 and  non-phasic  release  should  not  be  substantially  abolished by these manipulations; 2  Release  in  the presence  of  extracellular  absence of extracellular Ca**  Sr 2 +  and  (Bain & Quastel, 1992a);  if the Parnas voltage effect is the primary mechanism ii  underlying release in the presence of Sr* , phasic and non-phasic  release  will  follow  a  pattern  of  facilitation similar to that seen in Ca^ .  B. The divalent agonists While non-phasic release rate is promoted by a large number  of  inorganic  cations  in  the  presence  of  nerve  terminal depolarization, only three substances have been found to support phasic release, each an alkaline earth divalent cation, Ca 2 + (Katz & Miledi, 1965), Sr 2 + (Miledi, 1966) and Ba 2 +  (Blioch et al, 1968).  These  "divalent  agonists" are thus appear to be able to substitute for Ca 2+ , but with altered kinetics of one or more of the processes involved  in  coupling  of  ion  entry  to  neurotransmitter  release. l_i_  As pharmacological tools, Sr  i_i_  and Ba  may have both  pharmacokinetic and pharmacodynamic differences from Ca 2+ . The former includes non-receptor binding, including nonspecific binding to protein and lipid anionic sites, which affects the effective free ion concentration of the divalent agonist  either  Hydration  radius  intracellularly and  hydration  or free  extracellularly. energy  affect  the  10 diffusion rates through ion-selective membrane channels, as well  as  binding  receptor for Ca  affinity  at  a  putative  intracellular  .  By definition, even the characteristics of the Ca 2 + channel, binding  its voltage specificity,  dependency, as part  of  kinetics, and the  channel  apparent  activation  process (Quastel et al, 1989), are no more than complexities in the pharmacokinetics of Ca^ the hypothetical  (or surrogate) delivery to  intracellular active site for promoting  exocytosis of neurotransmitter. Also by definition, pharmacodynamic variables for each of the divalent agonists include:  (1)  the affinity of the  ion for its putative intracellular receptor,  (2)  the  intrinsic activity of the ion-receptor complex in initiating events leading to quantal neurotransmitter release, and  (3)  the number of intracellular active binding sites for Ca 2+ , each of which is perhaps made up of two parts - the vesicle and the intracellular surface of the membrane, or docking proteins thereon. Guan et al (1988) have utilised the residual Ca 2 + model (see below), generalized to include Ba"6 blockers of the putative Ca16  in a study of  channel at the neuromuscular  junction, to show a consistency between this model and the hypothesis that there is only one voltage dependent Ca 2 + channel and only one release system involved in both phasic and  non-phasic  release  supported  by  Ba 2 +  and/or  Ca 2+ ,  consequent to an action potential or to a 'direct' pulse.  11 In the present work, the residual ion model as applied to Ba 2 +  is the basis for a study of the activation and  inactivation kinetics of the putative Ca  channel involved  in neurotransmitter release at the neuromuscular junction (Bain  &  Quastel,  neurotransmitter  1988).  For  release produced  the by  enhancement  stimulation  of  in the  presence of Sr^ , the present work demonstrates that the residual ion model is sufficient to account for relative magnitude  of  the  enhancement  of  phasic  and  non-phasic  release (Bain and Quastel, 1992a), whereas this model cannot account for the relative enhancement of phasic and nonphasic release found with stimulation in the presence of Ca 2 + (Bain and Quastel, 1992b). 1. Additive properties Upon stimulation of the nerve terminal in the presence of Ba^ , f m rises with each pulse and decays for seconds after the stimulation has ended.  When trains of various  number of pulses are used, the extent of the rise in f m is related non-linearly to the amount of stimulation.  Provided  that the stimulus train is much shorter than the apparent time constant of the f m decline, it is possible to obtain a linear transform of f m vs the total amount of stimulation (number of stimuli) by taking the fourth root of the raised fm (Quastel & Saint, 1988). The  model  depolarization  that of  was  the  suggested  nerve  is  terminal  that  with  (nerve-evoked  each or  direct), presynaptic channels open and admit an amount of  12 Ba*  which is the same for constant sized pulses.  With  9+  continued stimulation, the Ba*  accumulates in the terminal  to a steady state level which depends on the frequency of stimulation and the first order decay time constant. 9+  Furthermore, it was shown that if Ca* is additionally 9+ applied in the presence of Ba* , the instantaneous release rate  measured  during  the  approximately  one  millisecond  'window' of the e.p.p. grows with repetitive stimulation to the same extent as does non-phasic release rate (f m )/ when the fourth root transforms are applied. Apparently, Ba* 9+ persists in the nerve terminal for a longer time than Ca* , and is able to cooperate with incoming and/or residual Ca^ and/or Ba 2 + (Quastel et al, 1989), in accord with a model 9+  for  stimulation-induced  alone  enhancement  of  release  (Katz & Miledi, 1968; see below).  for Ca  The resulting 9+  e.p.p. is larger than without the residual Ba  to an extent  which is predicted very closely by the addition of the fourth root of the non-phasic release rate to the fourth root  of  the  release  rate  during  phasic  release  under  9+  conditions where residual Ba*  is absent.  The prediction of  the additive model holds true for non-phasic release rates 9+  elicited in Ba*  up to as high as can be reliably measured  (up to 500/s), with or without extracellular Ca 2 + (Quastel & Saint, 1988; Quastel et al, 1989). In the present work, similar results obtained in the 9+  presence of Sr*  are reported (Bain & Quastel, 1992).  With  9+  Sr* , one has the advantage that the major component of the  13 decay time course is much faster than that for Ba non-phasic  release  after  each  stimulus  is  function of the decaying residuum of Sr^"  .  Thus,  apparently  a  in the nerve  terminal and phasic release is the same function of total Sr  present during evoked openings of presynaptic divalent  ion channels. 2. Potency Sr  is  less potent  than  Ca^  for  eliciting  epps  (Miledi, 1966) but is approximately equipotent in raising fm when the preparation is partly depolarized with elevated K + (Mellow, 1979).  Ba 2 + is the least potent of the divalent  agonists for phasic release (Blioch et al, 1968;  Silinsky,  1985) . It is not clear, for each of these 'divalent agonists', what are the relative contributions toward potency of: (a) permeability  through  intracellular potency.  the  putative  channel,  and  (b)  However, there is evidence for other  voltage sensitive Ca*" channels that the permeability of the divalent agonists does not parallel their potency and that they enter through the same channels (Augustine & Eckert, 1984).  This implies that the observed potency series is  largely dictated by differences in intracellular potency. It has been suggested that if the maximal response obtainable with the divalent agonists is related to their intracellular potency, then their different abilities to support release possibly reflects a difference among their efficacies at the putative intracellular receptor (Silinsky,  14 1985).  However, this interpretation denies the existence of  a Ca-voltage mechanism (see DISCUSSION). 3. Elimination kinetics While the primary mechanism of removal of the divalent cations Ca 2 + , Sr2_t" and Ba 2 + from the nerve terminal is not known, there is some possibility that the Na/Ca exchanger plays a role.  This facilitated diffusion system has an  apparent affinity for Ca^  in the region of concentrations  similar to that which occurs in the nerve terminal following 9+  activation under normal conditions (Philipson, 1985). 9+  and  Ba^  Sr^  9+  substitute  for  Ca''  exchanger,  although  poorly  explaining  their  prolonged  for  transport  via  the  (Philipson,  1985),  effects  fm  (non-phasic  no  method  on  perhaps  release) after each stimulus. It must  be  stressed  that  there  is  yet  available to measure non-invasively nerve terminal 9+ concentrations of the divalent agonists. The Ca -sensitive dyes  (eg.  fura, quin, aequorin, and  others)  cannot be  considered to be non-invasive since by virtue of the binding dependency of fluorescence or absorption change, they are Ca^  buffers, raising the complications of altered timing  and magnitude of the Ca''  signal in their presence.  Even  more remote is the ability to carry out such a measurement with  temporal  microseconds adequately  and  spatial  resolution  in  the  order  of  and nanometers which would be necessary to describe  distribution in the  the  pattern  of  Ca 2 +  entry  mammalian motor nerve terminal.  and  15 Inasmuch  as  non-phasic  release  rates  in  some  way  reflect the persistence of the activity of the divalent agonists in the nerve terminal, in accord with the 'residual Ca' model appear  to  (see Enhancement processes, below), these ions have  elimination  potencies or efficacies.  rates  that  parallel  their  However, the measurement of time  course of decay of the divalent agonists and estimation of a time  constant  greatly  on  (assuming  the  adopted  first model  order for  the  kinetics)  depends  effector  pathway  between intracellular divalent agonist and guantal release. In particular, the time constant  for decay of divalent  agonist activity derived from the time course of decay of non-phasic release rate should be highly dependent on the intracellular cooperativity (n) of the divalent agonist.  It  has been demonstrated that the cooperativity for Ba^  is  about 4 and that at this n apparent Ba^  the time constant  (r) for  decay is about 5 seconds (Quastel & Saint,  1988) . In the present work (Bain & Quastel, 1992a) the n and time  constant  for  intracellular  Sr^  were  determined.  Release evoked in Ca 2 + contrasts with that evoked in Sr 2 + or Ba^  in that release terminates largely upon completion of  the phasic component, with some indication of one or more small components of Ca^ varying  from  milliseconds  processes, below).  persisting with time constants to  seconds  (see  Enhancement  16 C. Time course of evoked release Evoked  release  includes  phasic  release, as defined  above, as well as that component of non-phasic release which results from stimulation.  The non-phasic component which is  evoked appears to have several underlying mechanisms and time courses. 1. Phasic release In the adult mammalian neuromuscular junction, phasic neurotransmitter release is highly synchronized, following a delay after a stimulus given to the nerve terminal.  Most of  the  (action  delay  potential  between or  a  nerve  'direct',  terminal  see  METHODS)  stimulus and  post-synaptic  potentials, the synaptic delay or latency, is of presynaptic origin  (Katz & Miledi, 1965c).  nerve,  part  of  the  latency  With stimulation of the  is  attributable  potential conduction to the terminal.  to  action  Katz & Miledi (1965c)  postulated that of the remaining presynaptic processes which are involved in stimulus-secretion coupling, the processes involved in actual exocytosis of neurotransmitter from the presynaptic minimum  cell  synaptic  are  likely  delay.  to  contribute  While  there  most  is  to the  considerable  variability, among responses, of the latency of release following  the  nerve  terminal  stimulus,  the  underlying  activation process resulting from a given stimulus appears relatively stereotyped both in time course and magnitude. Variability in both delay and magnitude of the response presumptively reflects the stochastic character of release -  17 there occurs a high probability of release for a finite brief time (Barrett & Stevens, 1972a). In the present work, measurements of latency and time of growth and of decay of the phasic release period were made for large numbers of quanta under a variety of stimulus parameters  and  superfusate  constitutions.  These  confirm and extend the finding of other workers Barrett & Stevens, 1972a; time  course  of  manipulations  phasic  (e.g.,  Parnas & Parnas, 1988) that the release  (including  data  is  substitution  independent of  other  of  most  divalent  agonists for Ca^ ) and is tightly coupled, with a fixed minimum latency, to the presynaptic depolarization. In mammals, temperature sensitivity of phasic release time course, which  appears to be parallel  for latency,  growth and decay, exhibits Qio's °f about 4 for IOC to 20C. Furthermore, that the Van't Hoff plot for f m is non-linear with a Qio °f nearly suggestive  of  a  lipid  1 from 37C down to about 20C is phase  change, according  workers (eg. Datyner & Gage, 1980).  to  some  Barrett and Stevens  (1972b) showed that, in the frog, phasic release during their "ERP" (early release period, defined by the authors as a post stimulus period of heightened release probability) had a falling phase fitted by a single exponential over 2 orders of magnitude fall, with a r at 11°C of approximately 0.5 ms. During  repetitive  nerve  stimulation,  there  are  differing reports on whether the time course of phasic  18 release is altered.  After a conditioning train of 4 pulses  at 10 Hz, 11 °C in frog, Barrett & Stevens (1972b) showed that the decay r of the ERP was slightly prolonged with each successive stimulus.  This contrasts with the result of  Datyner & Gage (1980), at 18°C in mouse, that after a high frequency train of three pulses at 65 Hz there was no change in the time course of phasic release.  Datyner and Gage  postulate that the prolongation in time course observed by Barrett and Stevens was due to prolongation in the action potential time course at the nerve terminal, since the effect could not be duplicated except under conditions which promoted failure of the action potential.  Using clamped  current pulses delivered focally to the nerve terminal, such a change in time course of decay was not found (Datyner & Gage, 1980). 2. Techniques of measurement Various types of time course measurement techniques have been utilised by other workers:  the method of first  latencies, using a computer to record the time of first threshold crossing after a stimulus (Barrett and Stevens, 1972a, 1972b; Baldo et al, 1986), the method of latency measurement of each quantum (Katz and Miledi, 1965b), and the method  of  deconvoluting  the  average  e.p.p.  by  the  average unitary e.p.p. (van der Kloot, 1988a). The  method  of  first  latencies  does  not  allow  an  accurate measure of the decay time course of phasic release, except with very  large numbers  of  stimuli at very low  19 quantal  content,  since  the  probability  of  no  quanta  appearing until late in the phasic release period is very low  (Barrett  and  Stevens, 1972b).  On the other hand,  spotting every quantum and its latency will  lead to an  underestimate of quantal release probability at any latency if e.p.p.s of multiple quanta are counted as unit quanta (Katz and Miledi, 1965b), which may occur if the quanta making up the e.p.p. are of a small height within the normal distribution of quantal heights.  Without prior knowledge of  the  latency  relative  various  contributions  known  or  postulated  to  events  in  variance  of  the  stimulus-release  coupling, this latter method is accurate only at release rates low enough that multiple quantum e.p.p.s are rare, such as at low quantal  contents or late in the phasic  release period. In the present experiments, the time parameters of phasic  release  previously. program  are  examined  in  higher  resolution  than  Latency results were obtained using a computer  (developed by Professor D. M. J. Quastel) which  spotted every quantum in time (± 0.025 ms, digitizing rate), including within multiple quantum e.p.p.s, by deconvoluting suprathreshold events by an average quantum, with several checks  in the procedure  to  avoid  deconvoluting  voltage or noise artifacts in the record.  various  In this way, both  the rising phase and falling phase time courses can be measured  accurately  and  these  time  parameters  can  be  measured over a wide range of quantal contents, from less  20 than 0.001 up to 3 or 4 (or until muscle twitching or postsynaptic summation  action of  potentials  the  e.p.p.  are did  limiting). not  Non-linear  significantly  affect  deconvolutions of e.p.p.s of quantal contents of up to 3 or 4.  This method also gave values for f m before and after  stimuli.  3. Non-phasic release Non-phasic  release  time  course was  studied  in the  present experiments only in conjunction with measurements of phasic release, since distinction between models to explain any  particular  time  component  of  increased  non-phasic  release is not appropriate unless the fitting to a model is consistent for a wide bracket of release rates, especially the  relatively  release.  high  release  rates  inherent  in  phasic  Thus, time courses of raised non-phasic release  rates during stimulation in Sr^ , while being well defined by other workers  (eg. Zengel & Magleby, 1981), are re-  examined concurrently with phasic release in terms of a model which can accommodate both (Bain & Quastel, 1992a) . The time course of non-phasic release rates elevated by stimulation in the presence of Ca* , in association with facilitation of quantal contents, was similarly re-examined (Bain & Quastel, 1992b).  21 D. Enhancement processes  1. Overview Even before the Ca was  first  shown, a  -dependency of the release process  number  of  additional  neuromuscular transmission were described. studied attributes  is the  defined  any  here  as  'enhancement'  attributes of One of the most  (a general term  stimulation-induced  neurotransmitter release, both phasic (f m ), by preceding stimulation.  (m)  increase)  of  and non-phasic  Short term (lasting in the  order of 100 ms) enhancement of e.p.p.s was observed by Feng (1940) and Eccles et al (1941), and in mammalian tissue by Liley and North (1953).  Many others since have described a  short term enhancement of m under various conditions and the phenomenon has usually been called  'facilitation'.  The  frequency of m.e.p.p.s has also been shown to facilitate with a similar time course (del Castillo and Katz, 1954; Liley,  1956;  Hubbard,  carefully described  1963).  Facilitation  has  been  in terms of the time course of its  effects on f m and m at the frog neuromuscular junction and has been  separated  exponentials  into two time components by peeling  (eg. Mallart and Martin, 1967;  Zengel and  Magleby, 1980,1981; and see review: Silinsky, 1985). Tetani appear to induce an enhancement process with a time course of seconds, called  'augmentation' (Zengel and  Magleby,  tetani  1982a,b).  Prolonged  induce  enhancement  lasting tens of seconds to tens of minutes, generally termed  22 'potentiation'  -  'frequency  facilitation'  or  'tetanic  potentiation' for observations during the tetanus, otherwise 'posttetanic potentiation'. 2. Retrospective A number of problems arise in an analysis of much of the literature, past and current, on enhancement phenomena. First, the terminology is inconsistent, with a wide variety of  terms  in  use  including  facilitation,  potentiation,  augmentation, and a number of derivations and abbreviations of these and other descriptions. of one enhancement phenomenon  Usually, the distinction  from another has depended  exclusively upon the time course of the development or the decay of the processes. enhancement  processes  However, in comparisons between of  two  tissues  or  experimental  conditions, a time course in common may not be appropriate evidence for a common mechanism since this may arise not only  by  coincidence  but  from  two  different  underlying  processes with different time courses but with transfer functions similar.  which  make  the  observed  time  courses  appear  The importance of carrying out the transform  inherent in the model used, prior to attempting to determine the time course of a process, has been shown for Ba"6 (Quastel & Saint, 1988) and for Sr 2 + (Bain & Quastel, 1992a and see DISCUSSION) .  Furthermore, in most of the literature  on enhancement phenomena, data are shown either for nonphasic or for phasic release exclusive of each other, thus precluding any analysis which requires a concurrent measure  23 of both.  Finally, in many cases, data have been presented  in normalized form (relative to unstated control levels), preventing any analysis of the reported enhancement process in terms of the relation between absolute phasic or nonphasic release rates. 3. Facilitation There are many indications in the literature and in the present data that facilitation, defined in terms of its short time course, is actually a convolution of several processes with different time courses ranging from a few milliseconds to about 200 ms.  Various proposed mechanisms  have been matched with particular time course phases of facilitation. a) Mobilisation The data of Hubbard (1963) show m and f m modulating in a  parallel  multiplicative  potentiation" potentiation.  manner,  (facilitation)  both  and  for  for  "primary  post-tetanic  It was suggested by Hubbard (1963) that since  the effect was parallel for m and fm, it could be considered an  increase  in  the  probability  of  release  due  to  mobilisation of neurotransmitter vesicles to positions of closer approximation to the intracellular surface of the nerve  terminal  membrane, resulting  in  a higher  overall  probability of exocytosis, multiplicative of both m and fm to an equal extent.  Braun et al (19 66a) further explored  the effects of various stimulation frequencies and duration of tetani.  Since both facilitation and potentiation (short  24 and longer term enhancement) affected both f m and m to an equal extent, both were multiplicative over a wide range of stimulation  parameters  and  consequent  magnitude  of  enhancement, they suggested that the enhancement process seen after one pulse (lasting for milliseconds) and that seen after many hundreds or thousands of pulses (lasting for seconds) might have a common origin. A major difficulty which is readily apparent in the study of neuromuscular facilitation is the relative weakness of data for f m as a function of time after the last stimulus in a train, compared to the data for m.  Thus, in the range  of f m seen in healthy preparations, from about 1/s resting to about 30/s during potentiation, a very large number of trials would be required in order to resolve the time course of f m and its enhancement with any accuracy.  For example,  under optimum conditions for paired pulse facilitation, fm might be 2/s falling to 1/s over a time course of a few hundred milliseconds, necessitating time bins no longer than about 20 milliseconds, which in turn means that on average only one m.e.p.p. is counted in a time bin every 25 trials. For a Poisson process, to obtain a standard deviation of 10% of the mean, 100 m.e.p.p.s must be counted in each time bin, requiring about 2,500 trials.  This is compounded by the  need to carry out the experiment with varying number and frequency of preceding pulses.  The very large number of  trials  trains)  (stimuli  or  stimulus  required  takes  sufficiently long to carry out that sufficient data for  25 accuracy are either unobtainable due to lack of ability to maintain the intracellular recording, or may be questionable due to  significant  changes  in the preparation over the  period of study. The difficulty of obtaining data for enhancement of fm, to correlate with m enhancement, has limited the ability to devise a model to account for the phenomena.  Hubbard (1963)  and Braun et al (1966) collected both m and f m data, but for a limited number of stimulation paradigms and with precision sufficient  to  establish  only  the  general  nature of facilitation and potentiation.  multiplicative Katz & Miledi  (1965a) and others have largely disregarded fm, creating the following "residual calcium model" solely on the basis of observations of e.p.p. facilitation.  In the present work  (Bain & Quastel, 1992a,b), a novel method is presented for optimizing  the  sufficient  fm  stimulation data  at  pattern  the  desired  in  order  intervals  to  acquire  after the  desired antecedent stimulation history (see METHODS, below). b) Residual Ca 2+ A few years after the work of Hubbard (1963), Katz & Miledi  (1965a)  investigated  the  Ca^  dependence  of  facilitation by iontophoresis of the cation onto the nerve They were able to show that the presence of Ca^+  terminal. was  not  necessary  excitation,  but  was  for  propagation  necessary  to  of  the  presynaptic  facilitation.  Since  facilitation had earlier been shown to occur in the absence of any change in the presynaptic spike (Hubbard & Schmidt,  26 1963) and the effect of Ca"  on the e.p.p. also occurred in  the absence of any change in the presynaptic spike, Katz & Miledi suggested that "facilitation may be due to a residual change in ionized calcium concentration at some important site of the membrane" (Katz & Miledi, 1965a).  Of the Ca2 +  which entered the nerve terminal following each stimulus, a fraction remained 'residual' in the terminal to add to the effect of Ca  entering after a subsequent stimulus.  be noted that the residual Ca*  It may  hypothesis of Katz & Miledi  (1965a, 1968) was proposed to account only for the behaviour of one of the earliest components of facilitation of m, which has an apparent r of 35 ms. Further support for the residual Ca  hypothesis arose  from its apparent success in predicting the facilitation produced by a train of pulses, from the facilitation after one pulse, in accord with a residual Ca" Thies, Younkin  1971). (1974)  model (Miledi and  Similar experiments were for  a  (apparent r of 250 ms).  later  component  carried of  out by  facilitation  Without considering the effect of  the facilitation on non-phasic release in the model, the residual calcium hypothesis of Katz and Miledi was shown to be able to explain e.p.p. facilitation observed with certain stimulation patterns. Part of the evidence of Katz & Miledi (1965a, 1968) for a residual calcium model for facilitation was its dependence upon the presence of Ca*  at the nerve terminal during the  conditioning pulse or train, as shown by timed iontophoretic  27 9+  pulsing of Ca^ .  In the present work  (Bain & Quastel,  9+  1992b), it is suggested that Ca^  presence may be required  for facilitation, but that this role is only permissive, not rate  limiting  in  terms  of  experimentally  observed  facilitation. 9+  Support for the residual Ca^ by both Parnas and  hypothesis is also given  Zucker, from different perspectives.  Zucker's predictions are based on models in which diffusion from discrete channel sites in the presynaptic membrane is taken into account, with the effective cooperativity for 9+  intracellular Ca^ the Ca  (n)  changing, depending on the source of  (Zucker and Fogelson, 1986).  Thus, predictions of 9+  facilitation magnitude are based on n = 3 for Ca^  entering 9+  during a nerve terminal stimulus, and n = 5 for Ca*  which  is already intracellular, some of which may be a residuum from a previous stimulus.  Although this model can account  for facilitation of m, it cannot account for the parallel facilitation of m and fm.  On the other hand, the Ca-voltage  hypothesis (Parnas & Parnas, 1988; see above) supports the residual  Ca 2 +  hypothesis  by  providing  a  model  which  correctly predicts the parallel facilitation of m and fm which is often observed (see DISCUSSION).  However, certain  9+  requirements of the Ca  -voltage model are not consistent  with the present data. It is possible for the residual Ca 2 +  hypothesis to  predict facilitation which would appear multiplicative for  28 phasic release, but the predictiveness is limited, according to the following: Let facilitation = Rf/R Rf1/" = k1/77 (cp + ac p + c0)  and  R1/11 = k1/77 (cp + c0)  from which: (Rf/R)1/71 -1 = a / [1 + (c0/cp)]  (la)  Rf/R = (1 + a / [1 + (Co/cp)])71  (lb)  where: R and Rf are phasic release rates, expressed as quanta per second,  for  the  control  and  facilitated  pulses,  respectively. c 0 is the amount of resting intracellular Ca^  contributing  to spontaneous release. Cp  is  the  peak  concentration  contributed by Ca*  of  intracellular  Ca*  influx during the period of maximum  phasic release. a is the fraction of the integral of Cp over the period of phasic release which remains at the time of a next pulse; le. the fraction of the Ca  (or other divalent  agonist) entry which persists in the cytoplasm from one stimulus to the next. According to equations (la) and (lb), the facilitation of m should reach a maximum equal to (a+1)73 when c 0 is very much less than c p . Ca^  This is a function of the fraction of  remaining, independent of Ca' entry itself and thus of  extracellular Ca  concentration.  One would expect that  such a condition of c p > c 0 would exist down to very low  29 quantal  contents  (very  low  extracellular  Ca2+),  as  calculated according to this model and shown in Table 1 below.  However,  facilitation  this  model  does  should be virtually  not  predict  that  identical over a large  range of quantal contents (over a range of extracellular [Ca 2+ ]), when c 0  (indicated by resting fm) is virtually  unchanged (see RESULTS ). Conditions of larger c0/cp would be observed either (a) when c 0 rises due to metabolic or other incompetence or  (b)  when Cp is small due to low extracellular Ca 2 + concentration or the presence of a competitive  Ca^  channel blocker.  Under these conditions, it can be seen in Table 1  that it  is the ratio of m to fm, not their absolute magnitudes, that leads to a given calculated c0/cp.  Thus, for a low quantal  content 0.01 (a rate of about 10/s based on phasic release in 1 ms) with a f m of 0.1, c0/cp is 0.3, as it is for a much 2higher quantal content of 1 with a f m of 10, for n=4. The higher the chosen n, combination.  the higher the c0/cp for any m and fm  At high c0/cp ratio, there is virtually no  phasic release, i.e., m as a rate approximately equals fm. Under these conditions, the above model would require a high a for any given level of facilitation (Table IB). The residual calcium hypothesis for facilitation, using a cooperativity for intracellular Ca  , is contradicted by  the results of Hubbard (1963) unless the facilitation of f m  30  Table 1A: calculated c0/cp values for various m, f m and n  n=  m=  0.01  0.1  1  0.01  0.1  1  0.01  0.1  0.1  0.1  0.3  0.01  0.3  0.2  0.1  0.4  0.3  0.2  1  0.3  0.1  0.03  0.6  0.3  0.2  0.6  0.4  0.3  10  1  0.3  0.1  1  0.6  0.3  1  0.6  0.4  fo  (a"1)  Table IB: a (using n=4) for values of facilitation and c  o/ c p*  Rf/R=  1.05  1.2  2  0  0.012  0.047  0.19  0.1  0.014  0.051  0.21  1  0.025  0.093  0.38  10  0.14  0.55  (2.8)  c  o/ c p  31  is considered to occur by a different mechanism, unlikely in view of evidence that the facilitation of f m parallels that of m in both magnitude and time course. Quastel (1974) proposed a simplifying alternative to the residual calcium hypothesis which allows for a single facilitatory  mechanism  for  both  e.p.p.s  and  m.e.p.p.s  whereby stimulation induces an increase in the probability of release, evident as a multiplier of one of the final steps in excitation-secretion  coupling.  A parallel was  drawn between the multiplicative effect of ethanol on m and fm and that of facilitation.  Other agents, such as DMSO  (McLarnon et al, 1986), also multiply both m and f m (see DISCUSSION) .  4. Augmentation Zengel exponentials  and by  Magleby best  (1982a)  fitting  found,  routines,  a  by  'peeling'  component  of  release enhancement which had a time constant of a few seconds.  Although this component was very pronounced in the  presence of Ba^ , they also found it to be present, albeit relatively small, in Ca 2 + without Ba 2 + .  They concluded that  O J.  Ba^  enhances a process that normally occurs as a result of  stimulation in Ca^ , based on the apparent match of the time course of the respective components of enhancement in Ca 2 + and Ba*  containing solutions.  An alternative view is that  the increase in f m produced by stimulation in the presence of Ba^  (Silinsky, 1978) is to be attributed to temporary  32 9+  accumulation of Ba^  within the nerve terminal (Quastel & 9+  Saint, 1988), with intracellular Ba*  acting as an agonist  on the transmitter release process.  Guan et al  (1988)  showed that the rise in m that occurs concurrently with an increase in f m conforms with the equation for a residual ion model (see below).  This result and analysis was the first  demonstration of a good fit of both m and f m data to a residual  ion  model  for  enhancement  of  neurotransmitter  release. Tanabe and Kijima  (1989) showed that at frog motor  endplates an enhancement process identified by time course as 'augmentation' was uninfluenced by BAPTA-AM, a membrane permeant agent which is converted intracellularly to a high 94-  affinity chelator of Ca* .  However, this result does not 9+  contraindicate a direct involvement of intracellular Ca^  in  augmentation (Zengel & Magleby, 1982a), since an additional intracellular buffer should only affect the time course, not the  concentration-time  integral,  of  the  buffered  intracellular cation. 5. Potentiation First described electrophysiological^ by Feng (1941), potentiation  has  been  alternatively  accumulation of intracellular Ca z  ascribed  to  an  and/or intracellular Na .  Miledi and Thies (1971) and Hurlbut et al (1971) showed that •  potentiation strenuously  •  can  occur  removed.  9 4-  even  if  extracellular  Ca  is  Misler  et  al  (1987) showed  that  potentiation of both m and f m grew in the absence of added  33 Ca^9+ ,  but  grew  even  further  after  the  cessation  of  9+  stimulation and simultaneous readmission of Ca^ . et  al  (1982)  used  various  techniques  to  Erulkar  increase  Na +  concentration in the nerve terminal and showed that Na + is involved in potentiation.  However, they concluded that the  •  9+  mechanism involves displacement of intracellular Ca^  from  Ca 2 + stores in the nerve terminal. Misler et al (1987) were led to a conclusion that although presynaptic Na + increase was  likely  involved  in  potentiation,  the  increased  •  transmitter  release  9+  (post-tetanically)  was  due  to  Ca^  entering from the extracellular solution via the Na + -Ca 2+ exchanger. A role of Na + was reaffirmed by demonstration of the effects of Na/K ATPase (the sodium pump) inhibition by Na + removal or ouabain addition (Nussinovitch and Rahamimoff, 1988).  Thus, manipulations which increase the concentration  of Na + in the nerve terminal prolong the time course of decay of potentiation, even to the point of creating a new set point for the effectiveness of the release process. However, 90% of potentiation is sensitive to the presence of 9+  extracellular Ca^ , in their experiments. •  9+  The importance of intracellular Ca"  accumulation in  the mechanism of potentiation was further indicated by work of Delaney et al (1989) in crayfish by imaging using Cer+ 9+  sensitive dye.  Intracellular Ca^  to potentiation.  rose in direct proportion  That potentiation might have a major 9+  component that is not Ca' dependent was shown by Tanabe and  34 Kijima (1989) who showed its persistence under conditions of intracellular Ca 2 + greatly reduced by intracellular BAPTA. However, although the intracellular free Ca 2 + activity is buffered low in BAPTA, the time integral of exposure of intratterminal processes to Ca*  may be identical or even  higher than without BAPTA treatment, as discussed later. As  stated  include  earlier  intracellular  for Ca^  'facilitation', models which as  the  ultimate  cause  of  potentiation are incompatible with a parallel multiplication of both m and fm, when a Hill coefficient of 4 is assumed for the effect of Ca 2+ . 6. Present data In the present work  (see Bain & Quastel, 1992a,b),  emphasis was placed upon quantification of the relative magnitude  of  each  'augmentation' in Ca  enhancement  phenomenon  solution, for which  (excluding a method of  analysis had not yet been developed) upon phasic and nonphasic release.  This approach was adhered to in recent  analyses of release evoked in the presence of Ba*  alone and  Ba 2 + and Ca 2 + together (Quastel et al, 1989; Guan et al, 1988), based upon the assumptions that: (a) both types of release arise from a common pool of presynaptic vesicles and a  single  (Guan  et  system for quantal release of neurotransmitter al,  1988) and,  (b) that  a model  of  quantal  neurotransmitter release mechanisms should account equally well for phasic and for non-phasic release.  Accordingly, it  is suggested that the magnitude of both types of release  35 will  be  enhanced  multiplicatively  by or  any  enhancement  additively,  phenomenon  according  to  the  either model  formalized in: R = k( C ± + C r )  n  (2)  On this basis, enhancement could correspond to either an increase in the multiplier k or an increase in the residual ion term c r , respectively;  a simultaneous enhancement of  both types of release that is not analyzable as purely multiplicative or additive may be consistent with a mixture of effects on k and C r which overlap in time, or with a nonconstant divalent cation influx per stimulus, C^.  The Hill  coefficient n,  often assumed to be nearly 4, may appear to  be  4 where  less than  effective  number  of  it is determined Ca'  channels,  by varying the such  as  varied  polarization of the terminal (Quastel et al, 1992). Present data 1989;  (Bain & Quastel, 1988; Quastel et al,  Bain & Quastel, 1992b) show that potentiation can  develop  in  the  apparent  absence  of  Ca"6  entry  while  facilitation cannot, confirming earlier data (potentiation Misler & Hurlbut, 1983; facilitation - Dudel, 1990). importantly  it  is demonstrated  More  that, in mouse phrenic-  diaphragm, most of both facilitation and potentiation is accounted for by a multiplicative effect, secondary to entry of Ca 2 + , for the former, and of Na + , for the latter.  36 I I . Methods  37 A. Mouse hemidiaphragm The mouse neuromuscular junction differs from the frog in its anatomy.  In the adult mouse, a single nerve terminal  innervates a single muscle fibre, and the nerve terminals usually  form a single focussed  fibre.  In  innervation elongated.  the  is  frog,  synapse with the muscle  however,  common, and  nerve  multiple terminal  convergent  synapses  are  These features make the mouse neuromuscular  junction preferable for time course studies. The  isolated,  superfused  mouse  hemidiaphragm  preparation is similar to the same preparation from rat as described by Bulbring (1946).  In the present experiments,  diaphragms were removed from  (ether) anaesthetized adult  male white CD-I mice with about 0.5 cm of each phrenic nerve left intact, rinsed immediately in cold bathing solution, trimmed of intercostal muscles and pleura, and cut in half from the sternum to the central tendon.  Hemidiaphragms were  pinned to silicone rubber (Sylgard) disks and mounted in a movable stage.  The preparation was superfused via a glass  tube with a tip about 0.5 mm diameter which was adjusted by a manipulator to provide a fast flow over the immediate area around the electrode. B. Solutions and chemicals The  superfusion  solution  usually  contained,  in mM,  5 K + , 24 HC0 3 ~, 150 Na + , and 1 H 2 P0 4 ~ and 11 D-glucose.  K+  concentration was increased to 10 mM for most experiments in which direct polarization was used.  Mg 2 + concentration was  38 varied between maintain  the  1 mM and  total  12 mM; an  concentration  attempt was made to  of  Mg^  plus  divalent  agonist greater than 2 mM since low extracellular divalent cation concentration may cause depolarization due to surface charge effects.  All solutions were bubbled continuously  with 95% O2 / 5% CO2, maintaining a pH of 7.4.  Ba 2 + , Sr 2+ ,  and Ca^ , in the form of their chloride salts, were used in the range of 0 to 2 mM. purchased  from  local  All chemicals were of reagent grade suppliers.  Chemicals  unavailable  locally were purchased from Sigma Chemical and Calbiochem. C. Stimulus delivery Excitation-secretion  coupling  at  the  neuromuscular  junction begins with the invasion of the nerve terminal by a depolarizing event.  Physiologically, depolarization of the  nerve terminal results from an action potential travelling orthodromically in the nerve.  Experimentally, in addition  to stimulation of the phrenic nerve, a motor nerve terminal in the diaphragm can be depolarized or hyperpolarized with a current  clamp  electrode  placed  focally  (direct  polarization). Nerve stimulation was achieved by inserting the nerve with gentle suction into a glass tube one end of which is tapered to slightly larger diameter than the nerve, into the other end of which is inserted a chlorided silver wire. Current clamp pulses of 0.1 to 0.3 ms duration and 2 to 3 times threshold were used.  Frequency of stimulation was  kept below about 70 Hz for prolonged tetani and below 100 Hz  39 for any train, as higher freguencies of stimulation than these were often associated with nerve conduction failure. Nerve conduction failure was obvious at m > 1, but at low m it could be detected by its characteristic intermittency (runs of failures of quantal output or intermittently low m), or by an otherwise unexplained fall in potentiated fm. An excess of failures over those predicted from the number of stimuli producing one quantum and the number producing two guanta, according to a Poisson distribution, was also used as an indication of nerve conduction failure. The direct polarization technique (Cooke and Quastel, 1973a) utilises  a  large tipped  glass electrode filled with  (approximately  30-50 jum)  3 M NaCl-agar, connected by  chlorided silver wire to a current clamp feedback circuit. The primary advantage of this technique for direct stimulus delivery  to  the  nerve  terminal  is  the  uniformity  of  polarization, due to the large size of the electrode and the distance  from  which  the  current  is  delivered  from  the  electrode to the membrane. One disadvantage of the direct polarization technique is the sensitivity of the effectiveness of the stimulus to any drift in the positioning of the electrode; this was routinely overcome by bracketing in time the test protocols with  control  protocols.  Care was  taken  to  avoid  any  movement artifact in the voltage record or visual evidence of a localized muscle fibre movement 1973a) after each stimulus.  (Cooke and Quastel,  Consistent results reguired  40 that the polarizing electrode have a tip diameter of about 50 pm, a smooth aperature rim at right angles to the shaft, and be optimally placed such that the currents used to stimulate were minimized. conventional  intracellular  These were made by breaking electrodes  with  fine forceps,  under a binocular microscope. The briefness of the direct pulse possible was limited. Since the apparent time constant of the nerve terminal is in the order of 1 ms (Quastel & Saint, 1984), both the rise time and the decay time of nerve terminal depolarizations elicited  by  square  current  pulses  pulses of duration less than 2 ms.  were  significant  for  In some experiments, in  an attempt to overcome the capacitance of the nerve terminal and elicit depolarizations much more brief than the membrane time constant, a large brief depolarizing current pulse was followed after a delay by a hyperpolarizing current pulse. Using current pulses of only 50 to 100 us duration and sizing the hyperpolarizing pulse to a magnitude sufficient to cancel the residuum of the partly decayed depolarization, the effective duration of the depolarization was close to the interpulse delay. the  magnitude  magnitude  of  repolarize  to  of the  Thus, for very brief current pulses,  the  effective  depolarization  hyperpolarizing  resting membrane  pulse  potential  d = xd R m [l-exp(6t/rm)], and  x  h = Id [l-exp(-T/rm)]  the  required  to  for  effective pulse duration, are approximated by: v  and  a  desired  41 where: V^  is the effective magnitude of depolarization (mV)  T  is the interpulse delay, and is approximately the effective duration of the depolarization (ms)  6t  is the duration of the current pulses, being much briefer than zm.  I^ and I n are the current pulse magnitudes (JUA) r m and R m are the membrane time constant and resistance (presumed constant at all membrane potentials in the presence of TTX, 4AP and TEA), respectively.  The presynaptic nerve impulse lasts for about 1 ms (Katz  &  Miledi,  depolarization  19 65)  of  and  similar  differs magnitude  from  a  in  direct  that  the  repolarization subsequent to the former is subject only to the  rate  of  activation constant  of  Na +  channel  inactivation  (Hodgkin and Huxley, the  'direct' pulses.  nerve  terminal  and  K+  channel  1952), not to the time membrane  as  it  is  for  For a direct polarization in the presence  of Na + and K + channel blockers (0.4 JUM TTX and 0.2 to 0.4 mM TEA, 0.5 to 1 mM 4AP were used; Saint et al, 1987) the presynaptic membrane time constant is about 1 ms. D. Stimulation protocol Simultaneous measurement of both phasic and non-phasic release in the same experiment has historically been avoided because experiments which are designed to give sufficient output of e.p.p.s for statistical reliability do not usually  42 result in enough m.e.p.p. output, and vice  versa.  For the  study of the magnitude and time course of facilitation in Sr  and in Ca 2 + , stimulation sequences were generated by a  computer  program,  delivered  through  a  serial  port,  amplified, and used as trigger pulses for a conventional stimulator.  These sequences were random, either in terms of  train length (primarily for work with Sr* ) or choice of interval  between  stimuli  from  a  specified  selection of  possible intervals ranging from 11 ms to 2.8 s (primarily for characterizing facilitation in Ca  ). An example of the  random length train stimulation sequence is given in Fig. 1. With  conventional  trains  of  stimuli, there is  difficulty in obtaining data on the effect of various number of stimuli in the train on the enhancement of m and of fm. Of  the  measured  parameters,  facilitated  m  is  usually  obtainable with the least number of repetitions, since a train of any number of stimuli gives data on m for each pulse number in trains of fewer stimuli, except that for pulses early in the train where m is unfacilitated, more repetitions  are  necessary  statistical  reliability.  to For  count  enough  fra growth  quanta  for  and decay, the  difficulty in obtaining data for any particular train length is that sufficient repetitions of trains of that length are necessary for statistical reliability of quantal counts in each time bin after the last stimulus of the train;  often  the long time taken to carry out these repetitions is such  43  111  :  iiiiiiiiiiiniimii  III  mini llllllllllllllllllllllll  mini  minim  IIIIIIIIII  in  mi ii  II  minimi  IIIIIIIIII  mi  ii it  iiiiiiiiiiiiiiiiii  1 sec/sweep  Fig. 1  Random train stimulation.  An example of the timing of pulses in the random train stimulation protocol used primarily for the work with Sr The  short  intervals  used  were  invariably  .  11 ms,  corresponding to a frequency in the train of about 91 Hz, while  the  long  intervals  (500 ms shown here).  most  often  used were  2000 ms  The average length of the trains was  increased by increasing the probability of selecting a short interval at any one time.  44 that long term drift in m and fra makes subsequent trials of other train lengths unreliable. length trains provides  are twofold:  for a lower  The advantages of random  first, the first algorithm  frequency of occurrence of  longer  trains, which optimises the time required to gather data because  fewer  long  trains  are  needed  for  statistical  accuracy of the results since both phasic and non-phasic release build to high levels in the longer trains; second, the  randomly  varied  train  length  means  that  for  every  stimulus in the train, there have been recorded responses to other  trains  which  have  terminated  after  fewer pulses,  resulting in the ability to accurately determine the f m at the time of each pulse in a train.  Thus, these random  trains provide data on both m and f m during the entire growth of release during a train, as well as on the decay of f m after trains of each length.  This stimulation protocol  was especially suitable for studying the effects of Sr 2+ , since the random variable, train length, covers a range of time which  encompasses  the  time  constant  of  the major  component of stimulation-induced enhancement in Sr 2+ . Random  interval  stimulation,  or  pseudo-random  stimulation, allowed measurement of the m and the f m at 11, 22,  44, 88, 176, 352, 704  preceding  stimulus.  The  and up to  highest  during the random series was 91 Hz.  1408 ms  stimulation  after a frequency  The time course of  decay of the facilitation resulting from that stimulus and all its antecedent  stimuli was then measured, at these  45 various intervals.  A measurement of f m at any particular  interval was carried out in an interval twice or more as long; for example, the determination of f m at 44 ms after a pulse would be done during 88 ms and longer intervals. measurement another  of  m at  stimulation  sequence.  a particular that  interval  formed  part  resulted  of  the  A in  overall  In the determination of the time course of  facilitation  after  any  pulse,  any  residual  effects  of  enhancement due to the particular sequence prior the that pulse could be disregarded; since the sequence of intervals was random, such residual effects were, on overall average, the same for any stimulus.  The primary advantage of this  was that a measurement of m and f m facilitation was made for many intervals with each interval being measured more than once every  second.  Variations  on the computer program  allowed increased weighting of the probability of occurrence of a short interval, resulting in an increase in the average amount of facilitation present at any one stimulus without compromising the consistency, on overall average, of the magnitude of that facilitation. Random advantages  interval of  random  stimulation train  includes  some  stimulation, but  the  of  the  random  nature of the series is addressed to study of a shorter time constant phenomenon, the major component of facilitation in the presence of Ca*" . Thus, the stimulation series does not include trains at a constant interval, unless these appear in the random series of varied intervals.  Since on average  46 a "train" containing stimuli at two different intervals, randomly occurring in the "train", can be described as a train of some intermediate interval, it was possible to obtain m and f m growth and decay over the time of these "trains", in addition to data for the various intervals from 11 ms and longer. Similar to the random train protocol, random interval stimulation has the major advantage of providing m and f m data which is affected very little by spontaneous drifts of quantal  output  which  are  normally  recordings exceeding a few minutes. the  difficulty  experienced  in  experienced  during  This is in contrast to  attempting  to  study  an  enhancement phenomenon where a two-fold effect is observed, but m or f m is low and extended recordings are necessary, sometimes  up  to  30  minutes  for  each  train  length  or  interval, during which the drift of quantal output (either m or f m or both) may be much more than two-fold. E. Time constant of the nerve terminal Two protocols  of  direct  polarization  of  the  nerve  terminal (Cooke & Quastel, 1973) were used to estimate the effective time constant of the nerve terminal, as shown in Fig. 2.  For one protocol, nerve terminal action potentials  were evoked by current pulses of 0.1 ms duration during long hyperpolarizing pulses.  As the delay between the onset of a  hyperpolarizing pulse, H, and the depolarizing pulse, D, was decreased, the depolarization required to reach threshold  47  A duration of test D pulse is held constant, 0.1ms D, depolarizing current clamp pulse amplitude, is modulated to reach threshold 1  I  H, hyperpolarizing current clamp, long 1 pulse of small amplitude  delay (of D pulse, from beginning of H pulse) is decreased from rheobasic in small steps, with a determination of threshold D at each delay.  B duration of test D pulse is changed in small steps, with a determination of threshold D at each duration of the D pulse  1 1 (long, preceding H is optional)  Fig. 2  D, depolarizing current clamp pulse amplitude, I— is modulated to reach threshold  Nerve terminal r determination  Panel A shows a stimulation protocol which depends on the time course of the voltage response to a hyperpolarizing pulse prior to a test depolarizing pulse, whereas in panel B the test pulse itself is varied in duration and compared to a rheobasic depolarizing pulse.  48 for firing the presynaptic nerve terminal decreased to a minimum, D'.  The slope of a plot of ln[(D'-D)/H] vs. delay  (see Fig. 2) was an estimate of the time constant.  An  alternate method for estimating r m was to vary the magnitude of  the  depolarizing  pulse  to  find  threshold  at  many  different durations, similar to the protocol used by Quastel &  Saint  (1986).  In  this  case,  D'  was  the  minimal  depolarization required to reach threshold at long pulse durations (rheobase). was  overlapped  by  In most cases, the depolarizing pulse a  preceding  long,  small  amplitude  hyperpolarizing pulse to remove gjja (voltage-time dependent sodium  conductance)  inactivation.  A  plot  of  ln[l -  (rheobasic current/I^) vs. d, where I^j is the threshold current at duration, d, gave a slope of 1/r.  The nerve  terminal time constants were estimated by best fit (by eye) slopes.  F. Data recording Recordings placing  of  m.e.p.p.s  conventional  3M  and  e.p.p.s  KCl-filled  were  made  by  microelectrodes  intracellularly in muscle cells near nerve endplates.  The  signal from the electrode was converted into a voltage (1:1 Picometric preamplifier) and added to an inverted signal similarly recorded from the bath nearby.  This data stream  was then amplified 100 fold and recorded digitally on VHS videotape through a Medical Systems PCM-1 analog to digital converter.  For computer analysis, the analog data stream,  49 either direct from the lOOx amplifier 'on line' (during the actual experiment) or from playback through the PCM-1, was further amplified 10 to 20 fold and filtered through a Type 3A9 Tektronix amplifier and converted with 12 bit precision into digital form by a Tecmar Labmaster A/D converter at 40 kHz. 6. Computer programs The  computer  developed in 'C  analyses  used  in  these  experiments,  language and assembler by Dr. D. M. J.  Quastel, were of three types. used for the experiments  The first type of program was  on putative  Ca^  channel time  course using the tails of high f m which follow trains of stimulation in Ba^ . The second program was used for online monitoring of both m and fm/  with f m  counting accuracy  remaining very high at fms of up to about 200/s, depending on noise level and quantal size.  The third program, or  suite of programs, allowed offline determination  of the  height and time of occurrence of every quantal component of a record, accurate in timing to about 0.05 ms. 1. Ba  tails program  In this program, the number of m.e.p.p.s was counted in the period 0.2 to 2.6 seconds following the last stimulus in a train.  The data from this period were taken at a 40kHz  sampling rate into the computer memory, then analysed in the following 10 sec or so.  From the recorded data m.e.p.p.s  were selected for their suitability for inclusion in an average for use as a template.  Moving forward in time  50 through the data array, the program identifies deflections as m.e.p.p.s according to rise time and height criteria. Upon lining up the rising phase of each m.e.p.p. thus found, the m.e.p.p. template was subtracted from the record, prior to continuing the search for other m.e.p.p.s.  The counting  was validated by inspection of outputs of 0.2 s sections of data  with  markers  indicating  where  the  program  had  identified and counted m.e.p.p.s. Accuracy of counts within time bins in the tail was assessed  by  visual  Measurements of Ba*  inspection  of  the  screen  display.  tails which included false counts or  under counts were excluded from overall averages if the counting in more than one 0.2 s time bin of the tail was miscounted.  In addition, in the case where an individual  bin had a count spoiled by noise in the record, the count in that bin was omitted from the tail measurement and replaced by a mean of the adjoining sections. 2. On line monitoring The second type of program allowed on line counting of m.e.p.p.s and estimation of m from e.p.p. height.  This  program gives a monitor display of the data stream in 0.4 second  lots,  similar  to  the  Ba*  tail  program,  using  subtraction of a quantal template (m.e.p.p. average updated every 0.4 s).  Validity of the output of this program was  established by monitoring the video display.  As a further  validity check, the ratio of the variance/mean of the fms in the 0.4 s sections was displayed every 30 s as an indicator  51 of non-Poisson  counts, which may arise  from  'giant' or  'monster' m.e.p.p.s or from erroneous counts associated with a voltage artifact in the data stream.  The program also  gave a running estimate of the average m.e.p.p. height, useful in establishing the rate of solution changeover when drugs were used which reduced or enhanced the postsynaptic effect of acetylcholine.  This program was used during  almost all experiments as a monitor and was sufficiently accurate in its m and f m determinations to allow on line studies of potentiation and other long term processes.  This  on line estimation of m and f m was accurate to about 95% as long as the signal/noise was 4 or better, the m was less than about 3 and the fra was less than about 150/s. 3. Quantal deconvolution - off line The  third,  and  most  extensive,  type  of  computer  acquisition and analysis of data on the time course of coupling  mechanisms  relied  heavily  upon  programs  which  deconvoluted the data stream by the neurotransmitter quantum (m.e.p.p.).  Validation of these programs was carried out  both manually and automatically. a) data logging To  begin  with,  the  digitised  data  stream  (Tecmar  labmaster 40kHz analog digital converter) began with on or off line logging onto the computer disk.  The data logging  program was selective for suprathreshold events, anything in the data stream greater than a threshold set manually at about one-third of the mean m.e.p.p. height.  To aid in  52 selection of threshold, the program caused an analog signal to be produced, displayed on an oscilloscope, each time an event was logged. b) templates From the data recorded on the hard disk of a computer (approximately 0.5 megabyte for each file, representing a minimum of about 6 seconds of continuous data in each file), two templates were constructed. of  an  average  quantum.  For  First, templates were made this,  individual m.e.p.p.s in the record  a  program  sought  (one file at a time,  searching more than one file if necessary to obtain a smooth average quantum) which met criteria for height, rise time, superimposed noise level, baseline stability and isolation in time from other m.e.p.p.s.  These quanta were lined up  according to the starting point of their rising phase, and averaged.  Care was taken to only make use of a particular  template for later analysis of the data from which it was averaged, unless it was clear that there was no change in quantal (m.e.p.p.) size from one group of data to another and the data were from the same muscle cell penetration. To obtain an average stimulus artifact, a program was used  which  selection  of  displayed failures  all to  stimuli be  and  included  allowed in  the  manual artifact  average. Various difficulties were experienced in obtaining a stimulus artifact template which could be subtracted from the majority of the stimuli in the data stream leaving an  53 acceptably failures  small  were  residual.  intended  to  For be  example, although  included  in  the  only  stimulus  artifact average, it was sometimes difficult to distinguish failures from single quantum e.p.p.s when the quantal size was  small  relative  to  the  artifact  transient.  This  difficulty, especially evident for the large current clamp pulses used in direct stimulation, was overcome by making two passes through the data to identify failures, the first pass  averaging  all  stimuli  and  the  second  pass,  subtraction of the average from the first pass;  after  those which  showed the deepest negativity at the estimated time of the e.p.p. were then marked as failures to be included in the stimulus artifact template.  In practice, the voltage record  of pulses of a duration short enough to be clearly separated from  the  e.p.p.  could  be  improved  even  further  by  neutralizing the capacitance at the tip of the recording electrode. These stimulus  methods artifacts  for were  making  templates  modified  widely  of  quanta  and  allow  for  to  m.e.p.p. averaging during high frequency stimulation and very high or very low fm, and stimulus artifact averaging during  high  quantal  content  output  when  the  stimulus  artifact may rarely be seen without a response. c) quantal deconvolution The deconvolution program began by locating the stimuli in the data and subtracting the average stimulus artifact for which there was null response.  Then, moving through the  54 data in the forward direction, quanta were located in the record by a threshold crossing algorithm.  In preliminary  screening of the putative quanta for voltage or movement artifacts, events which rose and fell much too quickly or too  slowly  were  not  further  analysed.  The  time  initiation of each quantum was then established  of  to the  nearest 0.05 ms or so (about 2 points of digitisation at 40 kHz) by one of two methods.  One method used was to  correlate the data from about 0.1 ms just before the quantum to about the mid rising phase of the quantum with a hockey stick shape, the time of quanta initiation being taken as the time at the angle for the closest possible correlation. The other method used depended  on extrapolation of the  baseline forward toward the quantum and of the rising phase of the quantum backward, the point of intersection taken as the time of initiation.  These methods are distinct in that  the former fixes the angle between the "handle" and "blade" of  the  hockey  stick  shape, according  to  a preliminary  estimate of m.e.p.p. rise time, then places the "hockey stick" at a position for which consequent variances of data to template are minimal, to then determine the latency of a particular  quantum.  The  latter  method,  involving  extrapolation, is less sensitive to noise which could occur at an inopportune time for the hockey stick method, for example noise occurring right at the point of rise of the quantum, but is more prone to error in counting as quanta  55 certain artifacts in the record whose rising phase is quite unlike that of a quantum. After a putative quantum was  located  in time, the  quantum template was lined up and subtracted, as exemplified in Fig. 3.  If the residuum after subtraction had greater  overall variance  during the period of the quantum than  before subtraction, the template was added back and the event was  excluded.  Multiple quantum  e.p.p.s were not  treated differently than m.e.p.p.s, in that the point of initiation of the event was found prior to each successive template  subtraction  and  variance  determination.  This  procedure was carried out progressively through the record, giving  an  output  consisting  of  the  time  of  occurrence  (relative to the time of the preceding stimulus, which was also logged, to the nearest 0.025 ms) and the height of every quantum. d) binning and statistical analysis The  standard  output  files  from  the  quantal  deconvolution program were analysed in a variety of ways. In general, the deconvoluted events were binned by latency from the latency  stimulus histogram  into  0.1 ms bins  (eg. Fig.  4).  imposed within which the number of  and displayed  as a  Latency windows  were  'phasic' events was  counted for estimates of quantal content. release,  average  m.e.p.p.  frequency  was  For non-phasic estimated  from  counts of quanta within time bins that were progressively  56  i  f*^  •HWtyi  'WW* :\  T***4\  '•s^S  ,**•*-**<  m  {***A*tf*rrv#H*''  '.Si i r JjJi^JmJmf  •/-*J+'  •/*£&**•  /X-.  Fig. 3  Automated quantal deconvolution.  An example of deconvolution by a suite of "C" and assembly language programs written by Dr. D. Quastel.  Upper trace:  raw data (digitised), the stimulus artifact adjacent to the vertical marker. artifact.  Middle trace:  after subtraction of the  Lower trace: sequential lining up and subtraction  of a average quanta, resulting in a nearly flat residuum. The  time  of  each  deconvoluted  quantum  recording in a file with its height.  is  marked  and  Bottom: the average  stimulus artifact and the average m.e.p.p.  57 longer as they were further from the stimulus, minimizing the number of trials necessary to be averaged to achieve statistical significance (greater than 100 quanta per bin, for an accuracy of ± 100 /  quanta per bin).  To observe the relative growth of f m and m during a train, m.e.p.p.s in bins straddling each stimulus and e.p.p.s  were  position  in  sequences,  averaged the  the  among  train. stimuli  trains  For were  random 11 ms  according train apart,  to  their  stimulation the  m  was  determined from the quanta located in the first 6 ms, and the f m was determined for bins after 6 ms, with separate sets of bins for each stimulus in a train.  For random  interval stimulation, sequences of stimuli were considered trains as long as there were only intervals of the shortest two types in the sequence. e) Poisson tests Throughout the analyses, it was found that only rarely did release not follow a Poisson distribution, and it was generally assumed that if this were not the case the most likely reason was error, of one kind or another.  Although  at high quantal contents variance of numbers of quanta per e.p.p. appears less than the mean, i.e. release appears to be governed by binomial statistics  (del Castillo & Katz,  1954), it has been shown that upon correction for the postsynaptic non-linear effect of neurotransmitter on voltage, e.p.p. amplitude distribution appears to be Poisson, even at  3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60  3 2 1 2 6 6 6 0 1 2 18 50 92 89 72 54 29 15 22 6 6 5 1 4 4 3 5 5 3 4 1 5 2 4 4 1 2 2 1 1 1 2 2 2 7 6 3 2 5 5 0 5 5 2 0 1 1 2  Fig. 4  2.090 1.889 1.588 1.889 2.486 2.486 2.486 0.000 1.588 1.889 3.272 4.224 4.919 4.879 4.627 4.306 3.686 3.126 3.440 2.486 2.486 2.375 1.588 2.246 2.246 2.090 2.375 2.375 2.090 2.246 1.588 2.375 1.889 2.246 2.246 1.588 1.889 1.889 1.588 1.588 1.588 1.889 1.889 1.889 2.584 2.486 2.090 1.889 2.375 2.375 0.000 2.375 2.375 1.889 0.000 1.588 1.588 1.889  Typical latency histogram.  Phasic release (e.p.p.) was calculated to include all quanta in the region of spanning the peak release phase, less the number of quanta expected from non-phasic release.  In this  typical output, the values are (from left) latency in tenths of ms, number of quanta in the 0.1 ms time bin and the fourth root of quantal frequency in the bin. 89D05A, 0.3 mM Ca 2 + 4 mM Mg 2 + , m=0.357.  Experiment  59  high m (Martin, 1955).  Furthermore, it has been shown  (Vere-Jones, 1966; Hubbard et al, 1969) that even if the final step in release is binomial in nature, the entire release  process  distribution  if  would the  appear  preceding  to  follow  step  a  involved  a  Poisson Poisson  process. From  the  output  of  the  deconvolution  program,  the  number of phasic responses of each quantal content, ie. 0, 1, 2, etc., was counted.  From the observed mean quantal  content, the predicted number of each was calculated.  In  this way, it became obvious whenever there arose a nonPoisson distribution  of quantal  contents, either due to  biological problems, such as an intermittently failing nerve terminal action potential, or a problem in the deconvolution program, as occurred for a variety of reasons, especially during early development of the algorithms.  As a quick  check of Poisson nature, the variance to mean ratio of quantal  contents  was  routinely  checked  by  the  computer  program, a ratio differing more than 10% from unity being sufficient reason to reanalyse the data. In  some  analyses,  especially  at  higher  quantal  contents, there was considerable difficulty determining the quantal size.  Scaling the quantal template would obviously  skew the distribution from a Poisson:  too small a template  skewing toward higher quantal contents, too large skewing toward lower quantal contents.  Sometimes, the best approach  60 was  to  perform  the  deconvolution  repeatedly,  each time  scaling the quantal template to improve the fit of the observed distribution to a Poisson. Goodness of fit to a Poisson was often determined by the lack of regression of quantal content calculated from the frequencies numbers  of occurrence of each pair of adjacent  of quanta  in the Poisson  distribution.  For a  Poisson, the frequency of occurrence (Fq) of an e.p.p. with a particular number of quanta (q) is given as: F q = (q e_m)/q! Thus, one can obtain several estimates of m according to: m(q,q-l) = qFq/Fq_i For failures (q=0):  m(0) = -ln(F(0)), giving a further  estimate of m. If these  estimates  of m regress  upward with number of  quanta, an overcount of quanta is indicated (possibly due to an undersized quantal template); the obverse is also true. H. Data analysis 1. Estimation of m Quantal measure  of  content of the e.p.p. phasic  release  that  (m) is an arbitrary  usually  includes  some  contribution from at least three processes which are nonphasic in nature. 1  These include:  spontaneous release of m.e.p.p.s, which are present in the absence of stimulation and may occur at the latency at which phasic release is expected,  61 2  a component of facilitation (as defined above) from previous stimuli, and  3  a fast component of non-phasic release elicited by the stimulus in question, classified as non-phasic solely on the basis of the disparity of its time course with that of phasic release, about 6 ms decay vs. about 0.5 ms decay time constants.  In the present experiments, m was estimated from the latency histogram produced from quantal deconvolution of the raw data, mentioned earlier.  The estimate of m was a sum of all  quanta between latency limits which were chosen according to the following criteria: 1  the early limit was chosen early enough to include the beginning of the e.p.p. (that is, the obvious sharp rise in probability of quantal release that begins within one or two 0.1 ms bins), but late enough to minimize counts  the which  risk of were  including  sometimes  artifactual associated  quantal  with  the  settling of the stimulus artifact; 2  the  late  limit  was  chosen  to  include  the  major  component of the decay of phasic release, estimated as the  latency  at  which  release  rates  do  not  trend  downward in more than about five 0.1 ms bins. In practice, the limits for estimating m from the latency histogram were set at 0.3 ms and 6 ms for experiments in which the phrenic nerve was stimulated.  For e.p.p.s evoked  62 by a focally placed stimulating electrode, the lower limit was set according to the duration of the stimulus artifact, striking a balance between the risk of inaccuracy due to artifactual  counts  residua  stimulus  of  of  quanta  artifacts  arising after  from  quanta-like  subtraction  of  the  average stimulus artifact and the risk of undercounting the earliest of the quanta actually released very close to the stimulus artifact.  The upper limit was adjusted, for some  data, when a late or spread out e.p.p. occurred, or when an estimate of m without its late phase (seen as high nonphasic release rate between the e.p.p. and about 10 ms, see DISCUSSION)  was required.  In the estimation of m under conditions of high fm preceding the stimulus, from the number of quanta summed for the e.p.p. was subtracted the number of quanta during the e.p.p. time window (0.3 - 6 ms) that could be attributed to fm in the absence of a stimulus.  This correction made a  significant difference to calculations only at low m or high  2. Estimation of fm Frequency of occurrence of m.e.p.p.s, called general  referring  to  any  non-phasic  release  f m in  rate, was  estimated in various time bins around a stimulus, excluding the region where phasic release occurs.  Quanta occurring in  time bins within about 0.1 ms of the beginning or end of the stimulus artifact were also excluded from fm determination.  63 In order to determine the decay time course of m and fm facilitation, the random interval stimulation protocol was used.  In this analyis, f m corresponding to each stimulus  interval was determined in a window which spanned from mid way to the next shorter interval to midway to the next longer interval.  In the binning program, a correction was  calculated to take into account that the decay of f m was curvilinear correction  within was  the  span  generally  of  very  the  fm  small.  windows; It  should  this be  emphasized that in the present data, wherever a particular m and f m are compared, the f m was determined from a section of the record for which the previous history corresponds to that preceding the stimuli leading to the m.  3. Data averaging In experiments where data for m or f m from many cells were to be averaged, an assumption was made that these data were log normally distributed among junctions and a log transform was done on the data from each cell prior to averaging. Outlying data in terms of m or f m were not included in averages and were those neuromuscular junctions at which the f m spontaneously rose to high frequencies, either gradually during the recording or in brief bursts lasting 1 to 30 s. These phenomena both correlated with a declining viability of  the  cell  since  both  preparation is made hypoxic.  become  very  common  when  the  64 For experiments in which nerve stimulation was used, the exclusion criterion was simply failure of the nerve terminal action potential, seen as a lack of any phasic release, as distinct from release failure in accord with a Poisson distribution of m.  Nerve action potential failure  often occurred in conjunction with spontaneous high fm.  For  direct stimulation experiments, junctions were excluded, in practice after preliminary recording, if the f m was more than about 10 fold greater than the average. Weighting  of  data, while  giving  the  advantage  of  avoiding bias toward junctions at which very little data was acquired, was a problem in that it created a bias toward those junctions in which the resting f m was higher.  When  means were weighted, it was according to the inverse of the variance of each mean.  4. Derivations The model which has been used for analysis of the present data is based on the Ca^ -release model (above). Those  events  which  are  generally  accepted  to  occur  intracellularly in the presynaptic nerve terminal prior to quantal  release  lead  intracellular Ca^ bind  certain  to  the  following  model.  First,  binds a site which is able to selectively  divalent  metallic  cations,  the  divalent  agonists  (as well as some trivalents, eg. Curtis et al,  1986),  with  a  cooperativity  C4B = B t C 4 / (Kc4 + C 4 ) ,  where  of C4B  4. represents  Thus, an  65 intracellular receptor for polyvalent cation agonist with four such cations bound, B-^ is the maximum number of binding sites,  C  is  the  concentration  of  intracellular  cation  agonist and K c is the apparent affinity. Release is expressed as R = k' C4B, where k' is the 'intrinsic  activity'  proportionality  of  constant,  the  C4B  complex  and  R = k" C 4 /(K C 4 +C 4 ),  and  a  where  Assuming that C 4 << K c 4 , since the guantal release  k"=k'Bt.  rates in the present experiments are about l/100th those at normal  extracellular  simplified  to  Ca^  R = k C4  concentration,  where  k = k"/K4  this  (Quastel  can  be  et al,  1992) . Thus, inherent in the k term is a dissociation constant (K c ), a factor associated with the number of release sites (B-t)/ and the average readiness of the release sites for Ca'  induced release (k'). If one or more factors of the k  term modulated during repetitive  stimulation, the effect  would  release  be  multiplicative  on  all  evoked  by  the  divalent agonist. If the amount of divalent agonist at the receptor site changes, as with an accumulation from one stimulus to the next, or from the time of phasic release to the time of nonphasic release, it is the intracellular concentration of the divalent agonist at the receptor sites averaged for the actively  exocytotic  portion  becomes the term ' C .  of  the  nerve  terminal  that  Thus, the eguations that have been  66 derived for phasic and non-phasic release and facilitation by an accumulation of divalent agonist are as follows. Spontaneous  release,  present  in  the  absence  of  stimulation, assumed to be due to resting intracellular Ca concentration, is expressed as f 0 = k C 0 4 .  Release at any  moment following a stimulus, including phasic release, is also taken as a release rate, R(t) = k (C(t)+C0) , where C(t) is the intracellular concentration of divalent agonist at the release sites due to the stimulation-evoked influx, which rapidly changes during phasic release. an average of C(t) over a short time  Defining C-^ as  (less than  1 ms)  corresponding to, but phase shifted by excitation-secretion coupling latency from, the peak of phasic release, one can write R 0 = k (C-^ + C 0 ) 4 for the phasic release rate produced by an isolated stimulus.  If there is any residuum of C-^  after a stimulus, the miniature frequency is given at any moment by f m = k C r , where C r is sum of C 0 and the residuum of C-t, that is, total residual divalent agonist at the intracellular receptor at that time.  A stimulus delivered  in the presence of a residuum gives R = k (C-t+Cr)4.  From  these equations, a value in arbitrary units can be derived for entry of divalent agonist, C, and for k: fm 1 / 4 = k1/4  Cr,  and  R 1/4  = k l/4  (C-t+Cr), therefore R / - f m l/ - k V 4 C t (3) whether C r includes a residuum from previous C^, or C r is simply C 0 in which case f m is f 0 and R is R0. This 1  4  4  67 difference is termed the phasic delta fourth root, when C r is taken just before the pulse. Another measure of entry and k can be derived from fm elevated by the presence of a residuum from previous pulses. If the elevated f m = k C r 4 and f 0 = k C 0 , then fm 1 / 4 " f o 1 / 4 = k V 4  (Cr  - Co).  (4)  This difference is termed the non-phasic delta fourth root. The usefulness of magnitude  of  C^  (Cr - C 0 ) as a indication of the  depends  on  the  assumption  that  this  difference is the fraction of C-t remaining at the time of observation, and should modulate with C^ with changes in extracellular divalent agonist concentration, nerve terminal stimulus, or other influence on entry. 5. Enhancement calculation Where enhancement processes which are multiplicative overlap in time those which are additive, assumptions must be made with regard to the sequence in which events occur leading to neurotransmitter release, and their corollaries in the model.  The approach made is to first isolate the  effects of pure multiplicative enhancement and pure additive enhancement according to the model. First, if it is supposed that facilitation is purely additive, that is, due to an addition of a residuum of C-^ from a previous stimulus to the C-^ of the next stimulus, leading to a larger m, and an addition of the same residuum to the resting C 0 for a higher fm, then it is sufficient that C r changes and k remains constant.  Alternatively, if  68 facilitation is multiplicative, either C r = C 0 , there is no significant residuum at the time of observation, or C r is unchanging in the time frame between stimuli. k  modulates  with  a  time  constant  of  the  In this case, same  magnitude as the interval between stimuli.  order  of  Multiplicative  facilitation is calculated as m/m 0 and f m /fo'  For a measure  of additive facilitation, the amount of C r is calculated, relative to C-f In order to separate the additive and multiplicative components, it is first assumed that any  changes  in the  multiplier in the model, k, correspond physiologically with actions on release at a point in the release process after those processes which determine the absolute magnitude of C^ or C r .  This assumption will hold, unless the apparently  multiplicative component of facilitation in fact reflects an increase  in C^, an unlikely  possibility  discussed  later.  The following example, typical of data from nerve stimuli given in trains in the presence of about 0.3 mM C a 2 +  and  2 mM  the  Mg^  illustrates  the  method  of  separation  of  components (subscript f denotes facilitated): 1  Let: m = 0.5  n?f = 1  2a  phasic delta fourth root  fm = 1  fmf = 3 = (m^lOOO) 1 / 4 - fm 1 / 4 = 500 1 / 4 - 1  2b  facilitated  = 3.73  = (mf.1000 J1/4 - fmf 1 / 4 = lOOO 1 /* - 3X/4 = 4.3i  3  Multiplicative component, X, is the 4th power of the ratio of 2b to 2a, and, since the C-j- cancels on the  69 assumption that entry does not change, = ( k f 1 ^ / k l/4 } 4 = 4.31/3.73 4  = 1.78  Now, fmf 1 / 4 = kf 1 / 4 C r f , and fm 1 / 4 = k1/4 C r , so fmf/fm  = X 4 (C r f /C r ) 4 ,  C rf /C r  = (fnif/fm/X)1/4  c  = (3/1.78 J1/4 = 1.13  rf/ c r  Thus, for this example, where m is 0.5 and it doubles and f^ is 1 and it triples, the multiplier k increases by 78% and residual Ca 2 + increases by 13%. It  should  be  noted  that  if  the  integration  time  ascribed to C-j- is taken as 0.5 ms instead of 1 ms, the difference is very small.  Table 2 shows X and the increment  in C r relative to control for assumed e.p.p. integration times of 0.1, 1, and 10 ms.  It is clear that the longer the  time over which the e.p.p. is integrated, the multiplicative component, X, is less and the additive component, C r , is more. 6. Assumptions In this analysis of the present studies, a number of simplifying assumptions have been made.  The assumptions  which are essential to the models proposed are  (a) that  quantal release of neurotransmitter occurs by mechanisms which are common for both evoked and spontaneous release, and  (b) that the quanta involved in both types of release  are from a single pool.  Two lines of evidence support these  70  Table 2: Effect of period of integration of phasic release on derived values of facilitation and residual calcium.  Integration time of e.p.p, (ms) 0.1  10  phasic delta fourth roots  m  -m  0.5  1  7.41  3.73  1.66  1  3  8.68  4.31  1.85  X  1.89  1.78  1.53  increase in C r  0.12  0.14  0.18  control facilitated  The time over which phasic release is integrated in order to arrive at an average phasic release rate has an effect on the interpretation of the data and fitting to a particular model in terms of the relative facilitation of m and fm. The above data are mock data which are typical of facilitation data from trains of about 5 pulses at 90 Hz in low Ca 2 + raised Mg 2 + solution.  71 assumptions:  first, the equivalence of e.p.p. quanta and  m.e.p.p. quanta, as evidenced by comparisons of evoked and spontaneous end plate potentials (del Castillo & Katz, 1954; Elmqvist & Quastel, 1965a,b) and second, the additivity of the effect of Ba^  or Sr^ , remaining in the nerve terminal  from previous stimulation events, with the effect of the influx of that ion with a subsequent stimulus.  That this  additive relation obeys four or five power kinetics has been demonstrated  (Quastel & Saint, 1988) and is part of this  thesis (Bain & Quastel, 1992).  7. Definitions As previously explained, the meaning of certain words will be closely defined, to avoid unnecessary coining of new terms.  "Enhancement" will refer to any stimulation induced  increase in neurotransmitter release.  "Facilitation" and  "potentiation" will be used as they are usually used in the literature,  that  differentiated  is  enhancement  processes  which  are  solely on the basis of their gross time  course of growth or decay, less than one second for the former and minutes for the latter.  72  I I I . Results  73 A. Time course of stimulus-secretion coupling  1. Does the presynaptic Ca channel inactivate? Prior to commencing experiments which are directed to the nature of events known to occur late in the process of stimulus-secretion coupling, experiments were conducted to establish some of the characteristics of the nerve terminal Ca 2 + channel, an early component of the coupling mechanism. In particular, the question asked in this section was: Ca 2 + channels inactivate?  do  If so, the process underlying  facilitation would be more difficult to resolve, since Ca 2 + channel opening would be likely to change upon each stimulus given in succession. 2+  a) Studies with Ba In order characteristics  to of  study  the  activation  the  Ca^  channel  and at  inactivation  the  mammalian  neuromuscular junction, the Ba*" entry paradigm (Quastel and Saint, 1988) was employed. substituted for Zor  In solution in which Ba  was  (TTX present), a train of depolarizing  current pulses delivered to a single nerve terminal elicited a buildup of f m during the train and a 'tail' of m.e.p.p.s, which  lasted  for  several  seconds  following  the  train.  Assuming the model of Quastel and Saint (1988) whereby the difference of fourth roots of f m in the tail and before the train (non-phasic delta fourth root for Ba a linear function of Ba^  , see METHODS) is  entry, variations in the duration  of direct depolarizing pulses should yield information about  74  10  Fig. 5  20 30 40 pulse duration (ms)  Effect of pulse duration on "Ba entry".  Direct depolarizing pulses (TTX present) were given in short trains. varied  Pulse duration and number of pulses per train were approximately  excessive fm.  inversely,  in  order  to  prevent  Data points are means from 11 cells, weighted  by the number of repeats for each cell, ± s.e.m.  Trains of  pulses of each duration were bracketed one minute before and after by trains of 2.4 ms pulses.  Values plotted at each  duration are for 100»('Ba entry') relative to the average of that of the bracketing trains of 2.4 ms pulses, where Ba entry is defined as (ftail V4 _  fo l/4  ) (see equation 4) . Superfusion solution was (mM) 10 KCl, 120 NaCl, 2 MgCl2 and bicarbonate buffer.  75 the  time  course  depolarization  and  of  the  effective  the  time  dependency  channel opening at the neuromuscular  nerve of  terminal  overall Ca*  junction.  Typical  experimental results with pulse durations varied between 0.15  ms  and  100 ms  are  shown  in  Fig.  5.  At  first  o J.  appearance, the apparent Ba*  entry per pulse is linearly  related to the duration of the direct depolarization pulse, at least for durations less than about 30 ms; a best fit line is drawn (by eye) to the data for durations less than 30 ms.  That is, every millisecond increment in duration of  the presynaptic apparent Ba  depolarization  appeared  to  increase the  entry a constant amount.  The slope of plots such as Fig. 5 expressed per /jA«ms of pulse duration is a measure of the average effectiveness of the direct depolarization pulse, in arbitrary units of rate of  Ba*  electrode.  entry per  charge passed  from the current  This effectiveness is very steeply dependent on  the distance of the polarizing electrode from the endplate, but for any given polarizing electrode position, gradient for  Ba*  (or  other  terminal membrane  divalent  cation)  across  the  nerve  (dictated largely by extracellular Ba^ +  concentration), and the number of channels available for activation, this effectiveness is a measure of a combination of variables which may be involved in the coupling of the stimulus to channel opening.  These variables include the  efficiency of conversion of the clamped current pulse into a voltage transient  across the membrane, dependent on the  76 resistance and capacitance of the membrane and voltage or time dependent changes in these, and the voltage sensitivity of the presynaptic Ca^  channel gate, and any voltage or  time dependent changes. In perfusates containing from 0.5 to 2 mM Ba 2 + (10 K+, TTX present), this type of plot was almost linear over a wide range of pulse widths between 5 ms and 100 ms.  That  is, the apparent effectiveness of the depolarization was constant, regardless of whether it was delivered as a small number of long pulses or a larger number of shorter pulses. This was not true, however, for pulse widths of 5 ms or less; Fig. 6 is an average from 11 junctions.  At most  junctions studied, the slope was very low for durations up to about 1 ms, after which the slope increased to a maximum at  about  2 ms,  decreased  to  nearly  zero,  subsiding  thereafter to an intermediate slope which was quite constant over  the  longer pulse widths  used.  Fig. 7 shows the  differential of the average plot. In the following two sections, the experiments were designed to dissect from this overall coupling effectiveness those components whose role is in the ultimate delivery of a depolarizing voltage transient across the Ca 2 + channels. With  these  components  removed,  or  accounted  for,  the  remaining components may be predominated by the putative voltage and time dependent Ca** terminal.  channel at the motor nerve  77  2r  c D DQ > o  2 3 4 pulse duration (ms)  Fig. 6  Non-linear relation of Ba entry and pulse duration  for brief pulses. Detail of data shown in figure 5, showing the changes in apparent effectiveness of pulse prolongation in eliciting Ba entry.  78  jn  a c a) o >+—  o c  pulse duration (ms)  Fig. 7  Differential effectiveness of increases in pulse  duration. Differential derived from data in figure 5 (s.e.m. omitted), showing the  changes  in apparent  prolongation in eliciting Ba entry.  effectiveness  of  pulse  79 b) K  channel blockers  For a given clamped current stimulus, the magnitude of the voltage transient across the membrane is made greater by decreasing membrane conductance (Cooke and Quastel, 1973a). At  the  time  of  stimulation,  the  presence  of  tetraethylammonium (TEA) ions at the junction increases the effectiveness of the current pulse  (Saint et al, 1987),  presumably by a reduction in the membrane leak current and thus an increase in the voltage transient.  Such a change in  membrane resistance with TEA could not be large, since there was no significant difference in the membrane time constant (membrane  resistance multiplied  by membrane  capacitance)  when measured in the presence or the absence of 1 mM TEA using a rheobasic comparison method (see below). very  small  change  in membrane  resistance  and  Only a thus  the  voltage excursion elicited by an applied current pulse is necessary, however, to make a large difference to divalent cation entry, inferred from the observation that the fourth root  release  rate, proportional  to  Ca  channel  opening,  appears to be very steeply graded with membrane potential. The preparation was very sensitive to the presence of 4AP.  With the addition of 0.2 to 1 mM 4AP, Ba 2 + entry  (delta fourth root release rate) was usually an order of magnitude greater than in the absence of 4AP.  The effect of  4AP was immediate in onset but very slowly and usually incompletely  reversible,  in  observations (Saint et al, 1987).  accord  with  previous  80  Fig. 8A  Linearity of relationship between pulse duration  and Ba entry in the presence of 4AP.  Fig. 8B  Linearity of response to prolonged pulses.  The relation between direct depolarization duration and nonphasic delta fourth root is used as an index of Ba*  entry.  Results from two different cells are shown in each panel, the top panel  showing the shorter pulse durations, the  bottom panel showing the longer durations. (separate  experiments), the  solution  For both cells  contained  2 Mg 2 + , with 1 mM 4AP (and TTX) present.  0.5 Ba^+,  The Panel A shows  linearity around the 1 to 4 ms region, in contrast to what was observed in the absence of 4AP; panel B shows that the same slope persists, for each durations. same  cell, even to very  long  The lines in the two panels are drawn with the  parameters  respectively.  (slope  and  intercept)  for  each  cell,  The difference in slope between the two cells  shown was characteristic of the variation seen.  81  0.5 mM B a z + / 2 mM M g z + / 0 C a / 1 mM 4AP/0.2 ^M TTX  10 t——  0  2  4  6  8  10  12  pulse duration ( m s ) 30 ,  0  .  10  20  30  40  pulse duration (ms)  50  60  70  82 When either or both of the potassium channel blockers 4AP (1 mM) or TEA (0.5 mM) was present in the superfusate during the experiments with varied duration direct pulses (Ba2+-containing solution, TTX present), the deviation from linearity of Ba^"  entry per pulse with increasing pulse  duration was abolished, as shown in the example in Fig. 8 for 4AP. c) Membrane time constant A  non-linearity  solutions  which  persisted  in  all  bathing  (shown in Table 3) consisted of a decrease in  pulse effectiveness  for durations  less than about  1 ms,  consistent with a membrane time constant in the order of 1 ms.  Values for membrane time constant, determined in a  variety  of  conditions,  using  METHODS),  are given in Table 3.  differing  protocols  (see  These data are consistent  with those of Quastel & Saint (1986). Direct pulses of durations less than the nerve terminal time  constant  depolarizing METHODS)  synthesized  followed  appeared  by  a  from  a  combination  hyperpolarizing  relatively  equi-effective  of  pulse, to  a  (see  longer  depolarizing pulses in evoking release in Ba^ , indicating that Ca^  channel activation kinetics are much faster than  the membrane time constant. 2. Timing of phasic release a) Minimum latency — direct pulses In Ca* , the average latency of peak phasic release  83  Table 3: Time constant of the nerve terminal measured under various conditions.  K+  TEA  Ca 2 +  Mg 2 +  method  r  10  .5  1  1  delay  0.7  10  .5  1  1  delay  0.7  10  .5  1  1  delay  1.2  5  8  1  duration  0.5  5  8  1  duration  0.7, 0.2  1  1  duration  0.5  5  1  1  duration  1.0, 0.5  5  1  1  duration  0.5  5  1  1  duration  1.3, 0.6  1  1  duration  1.4, 0.5  5  5  0.5  1.0  At some junctions, the transform did not yield a clear single time constant, rather an indication of two linear sections of the plot, wherefrom two slope estimates and two estimates of r were made.  84 after the beginning of a direct depolarizing pulse focally applied  to  the nerve  terminal  (TTX present) was  0.4 ms, consistent among junctions. latency  with  direct  pulses  was  about  Precision for minimum limited  by  occasional  interference with the residuum of the stimulus artifact after subtraction. artifact  was  Even with the best data in which the  minimal  and  apparently  consistent  among  stimuli, there was enough residuum after subtraction of the stimulus artifact to cluster the latency measurement of the first quanta released.  Due to this limitation, further  precision in determining minimum latency was not sought. b) Phasic release time course Phasic release consequent to a nerve action potential usually occurs in a period much less than 1 ms, 90% within about 0.2 ms. Ca 2 +  model  for  According to a fourth power intracellular release  (see  equation  2), the  latency  histogram of the fourth root of quantal release might be expected to fit first order growth and decay kinetics.  In  the example shown in Fig. 9, the onset and decay time constants  were  estimated  by  interpolating  to  find  the  latency of the peak of release, then iterating according to fitting equation derived from the sum of two exponentials: y = A (e - a t - e - b t ) , where y is the fourth root of the instantaneous release rate, a and b are the inverses of the onset and decay time  85  1 mM C a 2 + / 1 2 mM M g 2 + / W v e stim  expt 88303i  10 Hz 827stimuli  20 r  nst  15 /  y 4-  T  ^y  <5t  10 /  o  °Px \  / o  5 -  o. oo,  /  OBO  \/\/°  0  -e  3  ©—I  4  latency ( m s )  Fig. 9  Latency  distribution  of  an  e.p.p.  and  its  derivative. and 12 mM Kg*  at  Iterative best fitting of zr  and  This example was recorded in 1 mM Ca 10 Hz, 827 nerve stimuli. Z& (rise  and  according  to  decay a  time  formula  constants) (see  text).  was  done  The  by  eye,  ordinate  explained in the fiqure legend, in arbitrary units. •  is  86 constants, respectively.  The differential of y with respect  to time is dy/dt = A (-ae"at + b e " b t ) . At the time of the peak, P, dy/dt = 0, so b/a = e _ a P / e" bP , and P = ln(b/a) / (b-a) The differential curve (e.g. fig. 9, closed circles), should  approach  'undershoot', chosen.  an  when  impulse  the  function,  appropriate  time  without  any  constants  are  Upon iteration, best fit time constants were found  to be about 0.2 ms (=l/a) and 0.5 ms (=l/b), respectively. The  time  course  of  the  process  can be  determined  directly from a histogram of either the latency of every quantum released, or the latency of the first quantum to be released after each stimulus, the first latencies method (Barrett  &  Stevens,  1972a).  Fig.  10  shows  that  for  recordings in conditions of low quantal content, ie. less than about 2, the difference between the latency histograms from these two methods is very slight. c) Sources of variance in timing of phasic release Autocorrelations  of quantal release and of multiple  quanta e.p.p.s show very close coupling in time of the release process with the nerve terminal action potential, with more variability among the apparent times of arrival of nerve action potentials at the nerve terminal, as shown in Fig. 11.  87  1000  100 m  Qi  10  V<v' 0.5  1.0  1.5  2.0  2.5  3.0  latency (ms)  Fig. 10  Latency distributions of all quanta and of first  quanta. An example of the latency distribution of phasic release, expressed as instantaneous release rates in each 0.1 ms bin, with nerve stimulation.  The upper curve is the latency  histogram of all quanta; the slightly lower curve is the histogram of the latency of the first quantum after each stimulus.  Ordinate:  instantaneous release rate, measured  in 0.1 ms bins after each stimulus. time in ms after the stimulus.  Abscissa:  latency, or  88  0.001 0.5  0.0  1.5  1.0  time between quanta (ms)  Fig. 11  Autocorrelation of quantal latencies within and  among e.p.p.s. Comparison  of  the  autocorrelation  within multiple quantum e.p.p.s  of  quantal  latencies  (open circles) with the  autocorrelation of all quantal responses (closed circles). The  data  are  from  the  same  junction  as  in  Fig. 10.  89 Such autocorrelations were used to determine the viability of the nerve terminal action potential.  Those which were  viable had an autocorrelation among all responses which at 50% of the responses was no more than 0.1 to 0.2 ms wider that the autocorrelation within multiple quanta responses. Non-viable  nerve  terminal  action potentials  resulted  in  autocorrelations among responses that were spread out over 1 ms  or more, while  the  autocorrelation  remained tightly distributed.  within  e.p.p.s  The drawback to the routine  use of the autocorrelation test for nerve terminal action potential viability is that it is not useful for experiments in which the quantal content is much less than one, since there are not enough multiple quantum e.p.p.s from which to construct the autocorrelation within the e.p.p. Another arose  in  source of variance  prolonged  tetanic  in nerve evoked release  stimulation.  Tetani  were  accompanied by an increase of 0.1 to 0.4 ms in latency and spreading of the phasic release latency histogram, depending on the stimulation frequency, as exemplified in Fig. 12. In terms  of  autocorrelation,  both  the  among  and  within  correlations were broadened, indicating that the variability in timing had increased in both the nerve terminal action potential and the coupling of the action potential with release, respectively.  The latter variability in timing may  indicate a broadening of the presynaptic action potential. Another  source of e.p.p. timing variability  is one  which is endogenous to the nerve terminal itself, the effect  90  500  400  I 300 w 200  100 AAA-A  latency (ms)  Fig. 12 Prolonged  Effect of prolonged tetanus on quantal latencies. tetanus  in  low  Ca*1  raised  Mq*  results  in  increase in synaptic delay and spreading of the time course of phasic  release.  perfusate  included  bekanamycin  This example 0.4 mM  Ca 2 + ,  (to maintain low m).  is Expt. 88328a; the 2 mM  Mg 2 +  and  66 juM  Stimulation was 55 Hz;  progressively higher curves correspond to first, second and third 30 s samples of a 90 s tetanus. as in Fig. 10.  Ordinate and abscissa  91 or correlate of spontaneous 'bursts' of m.e.p.p. frequency. These bursts are characterised by increases in f m by one or two orders of magnitude which persist for seconds.  During a  burst, the latency of phasic release increases slightly, recovering to normal as the burst subsides. Bursts can also be elicited, somewhat repeatably, by stimulating a terminal with a hyperpolarizing current (Cooke and Quastel, 1973a).  In solution containing 2 mM Ca 2 + , 1 mM  Mg 2 + , 10 mM K + , and no TTX, it was possible to deliver long hyperpolarizing pulses of 5 to 15 ms in duration ( < 1 Hz) at a magnitude between 5 and 15 /JA such that about half of the pulses elicited a burst.  Under these conditions in  which  present  a  burst  was  sometimes  within  the  milliseconds of a long hyperpolarization, a brief 1 ms) depolarization  of magnitude  first (about  just at threshold for  firing a nerve terminal action potential was superimposed on the hyperpolarization at about 3 ms delay.  Whenever a burst  was present during the hyperpolarizing pulse, the latency of the  e.p.p.  response  decreased,  indicating  increased  excitability of the nerve terminal occurred concomitantly with an elicited burst. Finally, alterations thereof,  of  phasic  in the latency, or variability  release  pharmacological agents.  can  be  induced  by  various  In preliminary experiments, the K +  channel blocker 4AP (0.2-1 mM) caused a prolongation (and enlargement) histogram.  of  the  phasic  component  of  the  latency  92  B. Stimulation-induced enhancement of release 21. Residual ion: Sr 2+  a) Stimulation in Sr* Short trains of nerve stimuli at 90 Hz in the presence of Sr  (no Ca^ ) elicited phasic and non-phasic release  which grew with each additional pulse during the train, the non-phasic release subsiding within seconds after the train to pre-train levels.  In the series of experiments reported  here, the number of pulses in the train was randomized to between 1 and 35, with progressively fewer trains as train length increased, according to the weighted randomization algorithm  (see  2 seconds  without  METHODS).  stimuli  Each  train  to permit  was  followed  measurement  declining non-phasic release rate, fm.  by  of the  Fig. 13 shows an  example of the growth of m and f m during a train and the fm tail after. As with the results of Quastel and Saint (1988) with Ba  , it was observed early in the experiments with Sr  that for short trains, the increase in the f m after the train depended upon the number of pulses in the train.  As a  first approximation, the data were analysed in a similar manner to those of Quastel and Saint (1988), according to a fourth power residual ion model for enhancement of nonphasic release. non-phasic  delta  For short trains, it was clear that the fourth  root  (see  Equation  4, METHODS)  increased linearly with number of pulses, consistent with a  93  i—1N  ««**>•*! M * l ^ " l  *«M«  V"^  l»ylll 01* ••l^.  r  ^—r^N—[^—f^^^^ A  *  A  A  fc  A  A  A  A  A  A  A  A  A  A  A^J\^N^_N™  > '  Fig. 13  "  • *u\m* ' I M M ^ M ^ • •«*«*  Buildup and decay of m and f m during and after a  train of stimuli in Sr^ . An example of the m and f m response to one train of nerve impulses at 11 ms intervals in Sr 2 + containing solution. Experiments were carried out with random trains  in the  presence of 0.5 to 1 mM Sr 2 + with 2 to 8 mM Mg 2 + , this example is 1.3 mM Sr 2 + with 8 mM Mg 2 + .  Stimuli as marked.  R(t)A  94  10 r  -I 1QOCO  4000  %  R(t)  - 1300  ;. ASO n~~B  nn  a » •  9  •  aa  n-  =a  •  A  « *  ••••  ,'•' n  oo "Oo-0^0ooo0U  uu 0 < 3 ooj)  V l a t e n c y - ms F i g . 14  Co-modulation of p h a s i c and n o n - p h a s i c r e l e a s e  in  .2+ Sr" On a fourth present  in  root basis, indicating the  nerve  terminal,  .2+ the amount of Sr'  repetitive  stimulation  resulted in additive, residual ion facilitation.  Latency  histograms show transmitter release as rates determined in each 0.1 ms bin after a stimulus.  The curves, progressing  upward, correspond to the average histogram for stimulus numbers 1-3, 4-10, 11-17 and 18-35 of a train.  The lines  are a best fit to the average of the two lowest curves, vertically  displaced  comparison purposes.  to  superimpose  each  curve  for  /*  95 fourth  power  residual  ion  (Sr^ )  model,  such  that  f m = k(N«C^ + C 0 ) , where N is number of pulses, C^ is Sr 2+ influx per pulse, and C 0 is Sr"  activity in the nerve  terminal in the absence of stimulation.  For longer trains  (trains lasting longer than about 60 ms, more than about 5 stimuli at 90 Hz, for example), the non-phasic delta fourth root, or apparent Sr^  entry, appeared to increase less than  predicted by the model. of f m in Sr^  It was obvious that the rapid decay  during the delivery of the train compared to (r ~ 3 s) would necessitate correction of the  that in Ba  actual number of pulses given in a train to an 'effective' number of pulses.  The 'effective* number of stimuli in a  train would be the number of pulses that would raise f m an equivalent amount if they were delivered at one time at the beginning of the train, before any decay of the purported residual Sr occur.  giving rise to the non-phasic release could  In order to make this correction, the time constant  of the putative residual Sr" was required and was estimated based on a first order decay of non-phasic delta fourth root determined for a series of tails of raised fm.  Using the  corrected number of pulses, N* for each train, it was clear that for trains of any length, the putative Sr" phasic  delta  'effective' residual ion  fourth  number  of  root)  increased  pulses  given,  linearly  entry (nonwith  consistent  the  with  a  model (Bain and Quastel, 1992).  The apparent buildup of intracellular residual  Sr 2 +  during trains also enhances the phasic component of release  96 in strict accord with a residual ion model.  Figure 14 shows  that the peak amplitude of release histograms for successive stimuli in trains grows with position in the train by the same amount as the non-phasic component for the same stimuli grows, on a fourth root basis. b) Estimation of n and r In order to avoid building a bias toward a fourth power release model into the analysis of the Sr^  data, a method  was used which assessed the most likely power, n, same  time  as  intracellular  the  time  9 4-  .  Sr^  constant  activity.  A  extensive, analysis was previously (Quastel & Saint, 1988).  of  decay  of  at the apparent  similar, although  less  carried out for Ba^  In the present analysis, it was  observed that the decay curve of f m after the train could be linearized with a log transform, whatever the root taken  prior  to  the  transform.  Likewise,  the  (1/n) curve  describing the growth of f m could also be linearized with any root. computer  A least squares best fit was carried out by (program by Dr. D. M. J. Quastel) for each line  generated for each of several values of n greater than 1. Each plot (Figure 15 is an example) gave an estimate of z as a slope and a goodness of fit as a correlation coefficient. It was found that for various sets of data, either multiple stimulation series from the same junction or from different junctions, the n  of the particular plot which gave the  highest correlation coefficient was not consistent.  That  is, for best fit linearization of the apparently first order  97  up T = 2 0 6  ms  down r = 2 0 3 ms CO  °-o.  0  i  i  i  i  i  i  \  0.2  i  i  0.4  i  i  i  i  i  i  i  '  i  0.5  I  '  '  i  1.0  i  i  I__J  1.5  i  i  i  2.0  sec  Fig. 15  Fourth root transform of f m buildup and decay in  Sr 2+ . Non-phasic release buildup during random length trains and the decay of f m after the train.  Note the consistency of  the first order time constant for the growth and decay of the non-phasic release rate when the chosen transform is with n = 4. 891212e).  The  solution  was  1.3 Sr 2+ ,  8 Mg 2 +  (Expt.  98 process of f m growth and decay in Sr^ , one particular n did not appear to have a clear advantage.  Thus, the best fit  line for each n gave an estimate of the r for that proposed n, but the n and r combination which was consistent with any particular  model  was  not  apparent  from  considering  the  growth and the decay of f m separately. A definitive measurement of n and r was achieved upon making  the  assumption  that  the  same  first  order  decay  process occurs both during the buildup of f m during the train (putative buildup of intracellular Sr  ) and as the fm  tail subsides (putative loss of intracellular Sr^ ) after a train.  It is also assumed that the same n pertains during  buildup and decay.  Thus, the best fit z was calculated for  n between 1 and 8, for both the rising and the decay phases of fm.  For the rising phase, the larger the chosen n,  the  shorter the best fit r, whereas for the falling phase, the higher the n,  the longer the r.  In this way, it was  observed that there could be only one combination of n and r that would provide a good fit for both the rising and falling phase, and thus satisfy the above assumptions. Fig. 15 exemplifies the consistency of growth and decay r in Sr 2+ when the chosen n is 4.  Fig. 16 is an example of one such  analysis in which n and z were determined as the coordinates at the intersection of plots of the rising and falling phase best fits for n and r. means of n and z respectively,  According to this analysis, the  (± s.e.m.) were 4.23 ± 0.22 and 248 ± 10,  n = 22 junctions.  Recordings  which  were  99  o  f m rise during train  •  f m tail after train  700 600 500 400 300  °°o8 i •• ##t  200  °Oooo  o o  o  o  100 0 4  1  5  6  i  i  7  8  'n'  Fig. 16  Estimation of x. and n for Srz  by best  fit. An example of the best fit routine for n and x. on the growth of non-phasic release during a train in 1 Sr 2+ , 8 Mg 2 + (open circles) and after the train (filled circles).  The intersection was always  well defined for junctions which had a moderate to low resting fm, with n about 4 and x about 220 ms. (Expt. 90104-2505).  100 excluded from this analysis included those in which the  fm  was persistently high, defined in practice as an f m over 10/s persisting in the absence of stimulation. Thus, while  the observation  of  linearity  of  delta  fourth root with the 'effective' number of pulses clearly supports a residual ion model for the short term enhancement seen in Sr^ supports  the  containing solutions, the above analysis also residual  ion model  for  Sr 2+ ,  requiring assumption of a particular n.  but without  Instead, the only  assumption was that the process(es) with the predominant influence on the growth and on the decay of f m had a time constant in common. c) Estimation of Sr' entry For each series of data, usually consisting of 100 to 300 random trains or more for cells or conditions of low quantal content, two parameters were estimated:  entry per  pulse, and the time constant of decay, z.  There are three  variables involved in the calculation:  entry, which is  Sr^  equivalent, according to the model, to the non-phasic delta nth  root at the time of the e.p.p., Sr^  elimination, terminal.  and  the  n  for  Sr^  time constant of  action  in  the  nerve  Of these variables, an assumption is made that n  is known to be 4 (according to the independent method of determination mentioned above).  If the time constant was  assumed to be 200 ms (which was approximately the mean of the time constants as determined above), entry could be calculated from the f m in the tail.  Non-phasic delta fourth  101 root  was  extrapolated  back  to  the  time  of  the  pulse  according to a first order decay, n of 4, with time constant The initial value of Sr 2 + or amount of Sr2+  r of 200 ms.  entry, according to the model, was determined by  least  squares fitting of non-phasic delta fourth roots from the tail to: In S(t) = In S 0 - t/200 where S is the non-phasic delta fourth root at any time t and S 0 is S extrapolated to the time of peak Sr 2 + entry, time 0.  In some analyses, entry and r were estimated by  iteration of the above equation with varied r to find the entry  and  the  correlation  coefficient was highest for the decay curve.  However, in  practice  it  r  combination  was  found  for  that  the  which  error  variance, which  increased greatly toward the end of the tail where fms were relatively low, appeared to introduce an unacceptably high variability into the entry and r best fits to make them much greater or less than the mean values among series, despite very little improvement of the correlation coefficient over that determined with the assumption of r = 200 ms. d) Comparison of Sr 2+ with Ca 2 + and Ba 2 + Assuming  a  residual  impulse or direct stimulus,  ion model, for  any  one  nerve  divalent cation agonist enters  the nerve terminal in a small bolus through channels which are opened in a phasic (although not simultaneous, Quastel et  al,  1992)  neurotransmitter  manner,  resulting  followed  by  in  phasic  diffusion  of  release the  of  divalent  102 agonist  throughout  the  nerve  terminal  cytoplasm.  The  'residual ion' concentration at release sites resulting from diffusion then causes non-phasic release of neurotransmitter until the active ion concentration is reduced through a process which removes the ion from the cytoplasm. With this assumption, the difference among the divalent agonists  in  the  manner  in  which  they  support  neurotransmitter release can be quantified in terms of a "dilution  factor"  (Bain  and  Quastel,  1992a).  If  the  increment of non-phasic delta fourth root per pulse is taken as an indicator of intracellular bulk Sr^  concentration and  the increment in the phasic delta fourth root for the same pulse indicates the average peak Sr^ then the  ratio of the  at the release sites,  latter to the former could be  considered a factor by which the effect of Sr  becomes  "diluted" just after it enters the nerve terminal.  This  dilution factor does not depend on the potency (inherent in k) of the divalent agonist, since the factors involved in potency are inherent in both delta fourth roots, and thus cancel.  On the other hand, for Ca  , Sr 2 + and Ba 2 + , phasic  release (as indicated by phasic delta fourth root), dilution factor and potency correlate.  The dilution factor observed  for Sr 2 + was 24.8 ± 1.0 (s.e.m.), less than that estimated for Ca 2+ but more than for Ba 2+ . Although the apparent time course of phasic release is identical in Ba 2 + and Ca 2 + (Quastel et al, 1989) as well as in Sr 2 +  and Ca 2 +  (Fig. 17) the divalent agonists differ  103  x D  E  o- -o Sr2+ •  1.0I-  • Ca2+  (6427 stimuli) (6556. stimuli)  •*->  O CD  en o jD  0.5i-  <D i—  TJ  a; "o  E o c  0 ^-•-•-••8* 1.0  Fig. 17  •9«s§« • • $ § • • •§••<«••  2.0 3.0 latency (ms from stimulus)  4.0  Demonstration of the identity in time course  phasic release in Ca 2 + as in Sr 2+ .  104  Relation between S r 2 + entry  'm'  and quantal content of EPP 10  0.1 0.1 1 Increment in 4th root of f m (slow phase) / p u l s e  Fig. 18  Relation between Sr 2 + entry and quantal content of  the e.p.p. Direct depolarizations (0.5ms duration) in 1 Sr 2+ , 1 Mg 2 + , 0.2 TEA, 0.5 4-AP, and 0.4 jM TTX.  While the n appears to  be 4 for non-phasic release, it is nearly 2 for phasic release.  105 widely in terms of apparent absolute potency and timing of their intra-terminal distribution and disposal, reflected in the peak phasic release rate for a given  extracellular  concentration and in some of the non-phasic time courses, respectively. Using direct pulses with TTX and K + channel blockers TEA and 4AP present, the dilution factor, calculated as above, decreased as the depolarization was increased. negative  correlation  stems  from  the  same  This  mechanism  underlying the result that a log-log plot of m versus nonphasic delta fourth root per pulse (putative Ba^ gives a slope of two rather than four in Ba  entry)  (Quastel et  al, 1992) and in Sr2+ , as shown in Fig. 18, when m was modulated by changing direct polarization intensity.  This  bias toward phasic release at smaller intensities of nerve terminal depolarizations can be explained in terms of a small number of channels per release site, the presence of stochastic heterogeneity among channels and release sites, and  thus  a  decreased  dominance  of  the  intracellular  cooperativity of 4, as proposed by Quastel et al (1992). e) Effect of BAPTA/AM In consideration of the indirect manner in which the majority of the data with Sr^  supported a residual ion  model for stimulation induced enhancement in the presence of Sr 2+ , a more direct assay of residual Sr 2 + was desirable. A method  for  direct  removal  of  cytoplasm by chelation would  free  Sr 2 +  ions  from  allow determination  the  of the  106 contribution enhancement  of of  these  ions  neurotransmitter  to  stimulation  release.  induced  BAPTA-AM  is  apparently able to diffuse into the cytoplasm, where its aminomethyl  group  is enzymatically  cleaved, producing a  chelator for Ca 2 + which is trapped in the cell by its ionic nature (Tsien, 1981). After incubating a diaphragm in 500 JL/M BAPTA/AM for 515 minutes followed by return to control superfusate, both phasic  and  non-phasic  release  were  considerably  irreversibly reduced at every junction so treated. examination of the  and Upon  'tails' of raised f m after trains in  BAPTA treated cells in the presence of Sr 2+ , it was found that the reduction of magnitude of non-phasic release was accompanied by a prolongation of the apparent time constant of intracellular Sr^  (determined as the best fit slope of  log transformed non-phasic delta fourth root against time). In various cells, all of which had a time constant for nonphasic delta fourth root in Sr^  of about 200 ms prior to  BAPTA, the time constant after BAPTA varied between 200 ms to 500 - 2000 ms.  As shown in Fig. 19, after BAPTA/AM the  prolongation of the time constant was correlated with the reduction in apparent Sr*  accumulation per pulse, ie. non-  phasic delta fourth root per pulse. At the same time, while the phasic and non-phasic delta fourth roots in various cells were correlated in the absence of BAPTA, they were similarly correlated after BAPTA loading, except that the  107  line for constant r x increment  100 0.01  Fig. 19  0.1 1-0 increment in 4th root of f m / p u l s e / m M S r 2 +  Effect of BAPTA-AM on z  and apparent entry in  Sr 2+ . Modification  by  stimulation in Sr T.  BAPTA-AM  exposure  of  the  response  to  included a prolongation of the f m decay  The control rs (circles) were close to 220 ms, while  after exposure ot BAPTA-AM (squares), r was prolonged, to a degree correlated with the decrease fourth root (apparent Sr^ entry).  in non-phasic delta  108  10r  ^  5  o CD O.  C  E 2  QJ i_  O  c  1 0.01  0.1  1.0  increment in 4th root of f m / p u l s e / m M Sr^+  Fig. 20  Relative effect of BAPTA-AM on phasic and nonO J.  phasic release in Sr^ . BAPTA reduced the non-phasic delta fourth root more than it reduced  the  phasic  delta  (squares); control (circles).  fourth  root.  BAPTA-treated  109 non-phasic delta fourth root was depressed more than was the phasic delta fourth root, as seen by an inflection (at an abscissa value of about 0.05) in the correlation of the two measurements in Fig. 20.  This result gives further support  for the residual ion hypothesis for Sr"  by showing that the  presence of an intracellular buffer for Sr^ apparent magnitude of the tail of  fm, presumptively by  removing a fraction of free intracellular Sr" cytoplasm.  The  slowing  of  reduced the  overall  rate  ions from the of  removal  (prolonged time constant of non-phasic delta fourth root) would be expected on this basis, from a reduction of the ratio of free to bound Sr^ . 2. Facilitation in Ca a) Time components (1) Trains and postpulses Facilitation was observed within and after short trains of stimuli at up to 100 Hz, in superfusates containing low Ca"  raised Mg" .  In the first experiments, an attempt was  made to determine the time course of facilitation of phasic and  non-phasic  release  varied intervals.  using  post-train  test pulses  at  From these data it was obvious that m and  fm facilitation under these conditions was similar to that reported by Hubbard (1963), with m and f m rising and falling nearly  in  parallel.  The  decay  time  constant  prominent phase of facilitation was approximately Fig. 21  (lower  two  curves)  shows  that  the  of  the  80 ms. maximum  facilitation during a train appeared to be reached after 5  110  8 9 1 1 3b—dh  nerve stimulation  0.5Ca 1Mg +1mM bekanamycin  predicted f m by residual Ca model m o  observed f  m  'E CO  c o  m (quantal content) O  ioa train of 7 stimuli at 80 Hz  125  150  test pulse at 83 ms delay  time elapsed during train (ms)  Fig. 21  Observed and predicted facilitation of m and f m by  short trains in Ca  .  An example of septuplets of nerve stimuli at 80 Hz with a post-pulse at delays which were varied from series to series (83ms  delay  shown,  expt.  89113b-dh).  The  solution  contained 0.5 Ca 2 + and 1 Mg 2 + ; 1 mM bekanamycin was added in place of higher Mg 2 + to reduce quantal content to below the threshold for muscle action potential and twitching.  Ill or 6 pulses (at 80 Hz stimulation rate), corresponding to a similar time constant for growth of the process as for its decay.  However, this method was not practical for obtaining  more accurate data on facilitation time course for two reasons: 1)  at usual fms, there were not enough m.e.p.p.s in time bins brief enough to allow determination of the time course of  fm  facilitation  with  reasonable  accuracy  (±10%, s.e.m); and 2)  the  time  of  data  acquisition  in order  to  acquire  sufficient numbers either of phasic quanta, under low m conditions, or of non-phasic quanta, under the usual low f m conditions for reasonable accuracy was so long that variation in the overall m or f m as they drift in time would often far exceed the magnitude of the short term  facilitatory  effects,  and  the  relative  and  absolute facilitation of m and f m often changed with drifts to higher and lower m and fm. Despite these shortcomings, the simple train and post-pulse stimulation protocol was able to clearly demonstrate, in all cells in which facilitation was evident, that facilitation in Ca 2 + could not be accounted for by a simple residual Ca 2 + model,  in  earlier.  contrast  to  the  facilitation  in  Sr 2 +  shown  The uppermost line in Fig. 21 shows the f m that  would be predicted by such a model, such that the putative residual Ca 2 + would be (m f -lOOO) 1 / 4 - (m-1000)V4, and the facilitated f m would be the fourth power of the sum of  112 Ca2+  residual  and  fo  •  Even  with  large  standard  deviations in the low f m measured between stimuli, it was clear in most junctions that the predicted f m exceeded the observed by nearly an order of magnitude. (2) Pseudo-random stimulation Experiments conducted to determine the time course of facilitation in the presence of Ca^  were made much easier  and more accurate by the use of random interval, or 'pseudorandom* , stimulation (see  A typical example of the  METHODS).  phasic and non-phasic response to pseudo-random stimulation is shown in Fig. 22a and 22b. with  the  septuplets,  the  In this example, as above  predicted  fm  according  to  residual C a 2 + hypothesis is shown (uppermost symbols). growth and decay of m and f m , normalized  relative  to  The  shown on a log scale, are  m or  the  the  fm  measured  longest of the intervals, in this example 0.5  after and  the  1.6/s,  respectively. (3) Resolution of components Since presence  enhancement  of  Ca  does  of  release  not  conform  by  stimulation  to  the  in  residual  the Ca2+  model, a variety of alternative models can be proposed (see DISCUSSION) .  In  the  present  work,  resolution  of  various  components of facilitation was not based upon time course, but instead upon components of the model proposed for all release,  as  R = k( Ci + C r )  given n  .  in  Equation  2,  repeated  here:  113  Fig. 22  Facilitation in Ca z  using continuous stimulation  with random intervals. Using repetitive nerve stimulation with randomly selected intervals between 11 ms and 0.7 s, facilitation was observed as it developed during each "train" (see Methods) and as it subsided  during  respectively).  the  longer  The pattern  intervals of  (panels A  facilitation  and B,  under  this  continuous but random interval stimulation was virtually identical to that seen with individual trains (Fig. 21). The solution was 0.5 Ca 2 + , 8 Mg 2 + , (expt. 90317h).  114  pred f m (from 'm')  5  10 pulse number  50  100 ms after pulse  150  200  115 From series of data such as the above, two components of enhancement partially  of  non-phasic  overlap,  discussed  (see  could  release, be  whose  resolved.  Enhancement  METHODS,  time As  courses  previously  Calculations),  the  analysis involves finding the contributions of increases in k, a multiplier, and increases in C r , residual Ca^ , to the enhancement of phasic and non-phasic release in the period starting with  about  10 ms  following a stimulus. mechanisms  which  and ending with about  The assumption being made that the  would  contribute  intracellular concentration of Ca* it  was  possible  to  500 ms  begin  to  k  are  distal  to  in the effector pathway,  resolution  of  components  by  establishing values (as a function of time since previous stimulation)  for  enhancement.  Calculation of the multiplicative component X  as  the  fourth  X,  power  the  of  multiplicative  the  ratio  of  component  facilitated  of  to  unfacilitated phasic delta fourth root is unaffected by the possible simultaneous presence of a residual ion process, unlike the more conventional calculation of multiplicative facilitation  as  unfacilitated  m  simply  the  (Hubbard,  ratio  1963).  of While  facilitated  to  multiplicative  facilitation calculated according to the latter calculation is  descriptive  of  the  major  component  of  m  and  fm  facilitation often seen in Ca^ , the former, more explicit calculation is necessary to dissect out a multiplicative component where a significant present.  residual ion component is  116  m  **Y^ Ac r /c. 4  Fig. 23  5 6 7 8 9 stimulus number  10 11 12 13  Growth of facilitation in Ca 2+ :  multiplicative  and 'residual Ca2"*"' components. Growth of m, fm, residual Ca 2+ and factor X for 26 series of data. In 23 series, 2-4% DMSO was present to increase f m and m in order to improve the counting statistics.  In most  cases, the solution included 0.5 Mg 2 + and 8 Mg, although this was  altered on occasion to maintain m lower than  threshold for postsynaptic activation.  117  c  50  Fig. 24  100 ms after pulse  150  Decay of facilitation in Ca 2 + :  'residual Ca  r/co"1  multiplicative and  ' components.  Same series of experiments as in Fig. 23, showing the decay of components of facilitation.  The points are calculated at  each progressively longer (doubled) interval. residual exponential  Ca 2+,  component  does  not  fit  Decay of the a  single  118 Results  from  random  interval  stimulation  in  the  9 _i_  presence of Ca^  were analysed first by determining the  value of X from the phasic delta fourth roots (as described 9 _i_  above) for each time bin.  Then, the residual Ca^  component  was determined from the enhancement of fm, after correcting for the contribution of the multiplier X.  In this manner,  the components were resolved for all time bins. that  the  underlying  processes  responsible  Assuming for  both  components followed first order kinetics, a best fit r was determined for each:  for the multiplicative component r was  about 80 ms and for the residual ion component (additive on a fourth power basis) r was about 150 ms. Fig. 23 and Fig. 24 show the growth and decay of m, fm, putative residual Ca^ 2 6 series of data.  and the multiplicative factor X for In the plot of facilitation growth,  putative accumulation of intracellular CaA  is shown as a  rising ratio of c r / Cp (ratio of total accumulated to peak intracellular at the time of phasic release).  In the decay  phase, the plot of the expression (cr / c 0 - 1) shows that the  ratio  of  intracellular  Ca^  added  as  a  result  of  stimulation to resting intracellular Ca 2 + , does not fall with a single exponential.  On the other hand, the ratio of  the increment in m to the unfacilitated m does fall with a single  exponential  (for  which  r = 84 ms),  as  does  the  multiplicative factor X (subtract 1, the X in the absence of multiplicative facilitation).  It is noteworthy that the  time course of the X factor almost superimposes with that of  119 m, confirming that under the conditions of this series the multiplicative form of facilitation predominates, seen as a parallel facilitation of m and fm. In order to study further the time course of decay of the  additive, residual  Ca  , component  of  facilitation,  instead of considering only those f m measurements which coincide in time with specific stimulation events, the nonphasic release after all stimuli was binned according to the latency from the final stimulus of each 'train', regardless of the antecedent stimulation pattern.  Taking the fourth  root of these binned values for fm, then dividing by the calculated X factor corresponding to each particular time bin, a value was calculated for each time bin for residual Ca^ .  In Fig. 25, these values, less c 0 , are plotted as  ratios to c 0 and to Cp, on a log scale.  Apart from a very  fast decay phase (see below), the plot was linear, with a time constant for decay of residual intracellular Ca 2+ of about 150 ms. Fig. 25 shows a fast component of release, evident during and immediately after the period of phasic release and persisting until about 10 or 15 ms after the stimulus, as an excess of f m multiplicative additive  enhancement over that expected from  facilitation  facilitation.  and  This  the fast  slow  component  component  was  of more  prominent under conditions of greater Ca 2 + entry - higher Ca^ such  concentration in the superfusate. that  it  would  not  contribute  Its time course was significantly  to  120  - 0.001 50  100 150 ms after pulse  200  line drawn with 150 ms T  Fig. 25  Time course of decay of residual Ca^ .  Decay of residual Ca 2 + component of facilitation after all pulses of 26 random interval series. equivalent to (cr - c 0 ) .  The expression c r e s is  For greater resolution of the time  course, the points are calculated from fms determined at smaller increment in interval than the doubling used for the increment in interval between stimuli. with r = 150 ms.  The line is drawn  121 facilitation measured at 11 ms or later after a conditioning pulse. b) Ca dependency of facilitation In  order  extracellular relative  to  find  Ca^  and  contribution  a  possible  the of  relationship  magnitude,  time  multiplicative  between  course,  and  or  additive  (residual C a 2 + ) components of facilitation, the C a 2 + in the perfusate was varied from 0.05 mM Ca^ , at which the quantal content was barely detectable, through Ca2+  (4 mM  Mg2+  in  all  solutions).  0.1, 0.2 In Table  and 4 mM 4,  it  is  evident that facilitation of both m and f m was nearly absent (multiplicative  factor  X  could  not  be  solution containing only 0.05 mM Ca^ . of  Ca2+  evident with  0.1 mM  and  higher  the  calculated)  in  For concentrations  facilitation  is  clearly  (see also Fig. 26, same data) and appears to grow increased  Ca  concentration.  However,  the  multiplicative component X appears to be present at the same magnitude for Ca^  concentration 0.1 mM or greater.  These  data imply that: 1)  multiplicative  facilitation  of  release  and in  additive the  presence  components  of  Ca2+  are  of  dependent on the concentration of extracellular C a 2 + ; 2)  with increasing extracellular Ca  , the additive  component of facilitation is increased, seen as an increase in the departure of the curves parallel,  for f m  and m from being  122  Table 4: Dependence  of  multiplicative  facilitation  on  extracellular Ca^ . The  solution  contained  4 mM  Mg  ;  stimulation was  random interval, with an overall frequency of 40.1 Hz (see also Fig. 26). 'Total facilitation* in this Table refers to the ratio of facilitated to unfacilitated m or fm.  The r is  for the decay of m facilitation at the end of the train.  total facilitation [Ca2+]  m  fm  of m  (s-1)  (mM)  of f m  X  (s-1)  T (ms)  0.05  0.012  4.0  1.17±0.11  1.15±0.04  n. s.  48±20  0.1  0.031  3.4  1.42±0.07  1.60±0.09  1.45±0.17  62±7  0.2  0.031  3.4  1.52±0.06  1.77±0.11  1.40+0.10  49±14  0.3  0.320  2.1  1.67±0.04  2.34+0.13  1.47+0.07  53±4  0.4  0.460  2.0  1.66±0.06  2.33±0.18  1.44±0.06  48+4  123 with no indication in the data obtained that this increase saturates  in the range of Ca^  concentrations  in which  e.p.p.s can be measured concomitantly with m.e.p.p.s without disruption  of  the  recordings  from  suprathreshold  postsynaptic activation; and 3)  the  multiplicative  component  of  facilitation  appears to be steeply graded with extracellular Ca^ , nearly absent at extracellular Ca^  concentration of 50 jM and  maximal at 100 jM. In  order  facilitation  was  to  determine  dependent  on  whether the  multiplicative  presence  of  resting  intracellular Ca 2 + , which might have become depleted during a prolonged exposure to bathing solution containing very low Ca 2 + , or Ca 2+ which entered during the nerve terminal action potential, 4AP (0.2 - 1.0 mM) was added to very low Ca 2 + solution in which facilitation was absent or too small to measure.  4AP increases Ca  entry  (by enlarging and/or  prolonging the presynaptic a.p., Saint et al, 1987; review: Thesleff, 1980), while resting intracellular Ca 2 + remain  virtually  unchanged.  As  expected,  increased m with little or no effect on fm.  4AP  should greatly  The effect of  the 4AP on facilitation was to restore the multiplicative component  to  its  normal  magnitude  seen  concentrations in the absence of 4AP.  in higher  Ca  At some junctions,  multiplicative facilitation not only reappeared after 4AP but was more than normal.  124  o J.  Fig. 26  Ca*  dependence of facilitation.  Composite showing the growth and decay of m and f m and the dependency of facilitation on extracellular Ca* .  Ca*  was  increased  0.2  and  from  0.4 mM Ca* .  0.05 mM  (top)  through  0.1,  The values plotted are normalized relative to  the m or f m after a long delay.  125  2 Pft-rcV^^sQ-  ft  3 •  3.  ;8  N--^_  k 0  10  20  no. of stimuli in "troin"  30  0  50  100  150  time after lost stimulus in "train" (ms)  126  2.0  D  c  1.5  o E 1.0  0.5 0  50  Fig. 27  Lack  of  100 ms from last pulse  facilitation  in  Ca*  150  solution  200  after  loading with BAPTA. Data for m (filled circles) and f m (open circles) after a preceding pulse in random interval stimulation are plotted at  the  time  stimulus,  as  at  which  they  a  ratio  to  occur their  after  the  unfacilitated  preceding values.  Unfacilitated m and f m were 1.92 and 11.3/s (with 4% DMSO present), respectively.  127 Finally, in agreement with the results of Kijima & Tanabe (1988) for the frog neuromuscular junction and Hochner et al (1991) for crayfish neuromuscular junction, facilitation was blocked by prior loading of the nerve terminal with BAPTA. In  solution  containing  1 mM  Ca^" ,  8 mM  Mg^  and  4%  dimethylsulfoxide to increase m and fm, conditions under which  both  multiplicative  and  additive  components  of  facilitation were normal, incubation for 5 to 15 minutes in 500 /JM BAPTA-AM virtually abolished facilitation, as shown in Fig. 27. c) Dependency on presynaptic stimulus Fig. 28 shows that facilitation is dependent on the magnitude of a presynaptic 'direct' depolarization. figure  is  stimulation  shown with  the  facilitation  various  of  magnitude  m  and  'directs',  In this  fm in  after the  presence of 2 mM Ca 2 + and 1 mM Mg 2 + , TTX, K+-channel blocker TEA, and K +  elevated to 10 mM.  In this solution, the  stimulus magnitude was chosen to elicit an m (final) of only 0.2 for the lowest current (10/JA, 0.4 ms) and 3.53 for the highest current (20/JA, 0.4 ms).  For comparison, in similar  solution, a normal presynaptic action potential would elicit an e.p.p. with a quantal content of over 100 (Elmqvist and Quastel, 1965b).  128  •  •  •  •  i  •  .  .  .  i  .  .  .  i  .  50  0 50 ms from last pulse f m raised by DC 4/zA depol. to 7 / s  Fig. 28  Dependency  of  facilitation  on  .  .  .  .  i  100  depolarization  amplitude. Facilitation with direct pulse (TTX present) random interval stimulation  with  three  different  pulse  intensities,  normalized m the left three plots, normalized f m on the right.  The pulse intensities were 10 yiK - triangles, 15 JJA  - circles, and 20 JJK were 0.2, 0.66 and 3.53.  squares;  actual unfacilitated ms  The solution contained (mM) 10 K + ,  2 Ca 2 + , 1 Mg 2 + , 0.2 TEA, and 0.4 }M TTX.  129 While the quantal content for the lowest current used (10 ph,  m = 0.2) was still much greater than that for the  0.1 mM Ca2+/4 mM Mg 2 + situation shown in Fig. 26 (with nerve stimulation) where the multiplicative facilitation was fully expressed, the facilitation here was absent. intensity  was  increased,  evident.  In  experiments  facilitation where  As stimulus  became  stimulus  clearly  intensity  was  increased to give still higher quantal content, there was no indication of any increase in multiplicative facilitation greater than that shown in Fig. 28. 3. Deviations from the models a) Decreasing stimulus In principle, additive fourth power facilitation would be  mimicked,  existed, during  if the  even the train  if  only  stimulus (see  multiplicative  were  to  facilitation  decrease In  DISCUSSION).  progessively the  case  of  stimulation via action potentials, this could occur either as a result of an increasing probability of action potential failure, or by a decrease in the effectiveness of each action potential to open Ca^ potential  failure  was  found  channels. to  occur  Usually, action primarily  at  stimulation frequencies more than 100 Hz (avoided in the present experiments).  An apparent decrease in the action  potential effectiveness (as evidenced by a decrease in the phasic delta fourth root) was found to occur whenever fjj, rose spontaneously, or after prolonged stimulation, to more than 200/s or so, either in Ca 2 + , Sr 2 + or Ba 2 + .  The latter  130 indicates a possible role of raised intracellular Ca 2 + (or Sr  or Ba 2 + ) in suppressing the action potential. Apparent  decrease  in effective  stimulation  observed within a short train of direct  was not  stimuli.  b) Increasing stimulus In principle, it is possible for a graded change from stimulus to stimulus in the voltage excursion in the nerve terminal, or its effectiveness to open Ca 2 + channels, to mimic multiplicative facilitation.  If a nerve terminal were  normally to facilitate in accord with the residual ion model (as seen in Ba*  or Sr*  containing solution) but the action  potential grew or became more effective with each successive stimulus, the e.p.p. would grow more than predicted from the residual accumulated divalent agonist calculated from raised fm.  Conceivably, the growth  divalent  agonist  in f m  could be matched  due to accumulated  by the growth  in m,  resulting mainly from the increased stimulus, appearing as a parallel  or  near  parallel  multiplication  of  release.  However, if this were the case one would expect enhancement of  the  stimulus  facilitation  of  itself m,  with  sometimes less  to  evident  facilitation  predicted by a parallel multiplication. seen.  be  Moreover, the results with Sr*  of  fm  as  a  than  This was never  (e.g. Fig. 15) cannot  be reconciled to a model in which impulse effectiveness rised within a train.  131 c) Ultra fast facilitation There was  a  component  of  increased  probability  of  release which was not specifically investigated, but was evident  from  latency  histograms  of  release  in  Ca 2+ -  containing solution. This component of increased non-phasic release  appears  to  be  of  the  residual  Ca  type  of  enhancement, since the m.e.p.p. frequency at 11 ms after a previous  stimulus  often  exceeded  that  predicted  by the  multiplicative effect of facilitation of the quantal content at a corresponding time (eg. Fig. 25).  Furthermore, this  early tail of non-phasic release was always more pronounced under  conditions  of  extracellular Ca*  greater  or less Mg  Ca* .  entry,  le.  higher  A detailed analysis of  this component of enhancement could not be carried out using nerve terminal action potentials (focally evoked or by nerve trunk stimulation), since action potential generation and conduction is not reliable when the interval between nerve impulses is less than about 10 ms. 4. Potentiation 2+  a) Ca*  dependency  (1) Effect of bekanamycin With tetanic stimulation of the nerve under conditions of low quantal content in the presence of Ca2+ , m and fm continue to rise after the facilitation, as described above, is complete.  This tetanic enhancement of release, called  potentiation, is similar to the short terra enhancement of release in Ca*  (usually referred to as facilitation), in  132 that  it  appears  to  be  nearly  multiplicative  with  the  addition of a component, usually developing after persistent stimulation (usually for more than 30 s), in which f m growth exceeds m growth. purely  Suspecting that potentiation might be a  multiplicative  process  supplemented  by  gradual buildup of f m due to intracellular Ca^  an  extra  accumulation  during the tetanus, and to rule out the possibility that accumulation of intracellular Mg^  might play a part in  enhancement of release by prolonged trains, an attempt was made to inhibit the influx of Ca  using bekanamycin, an  aminoglycoside blocker of the neuromuscular junction Ca^ channel (Uchiyama et al, 1981).  A typical result is shown  in Fig. 29; the plots of log f m and log m against freguency of  tetanic  linearity  stimulation implies  underlying  process  that of  are a  linear  and  parallel.  multiplicative  potentiation  is  unit  added  This of  with  the each  stimulus, such that with p amount of potentiation adding over r pulses in a tetanus, where r is large enough that the potentiation is fully developed, the amount of potentiation present (P) is: P = p + pz + pz 2 + pz 3 + pz 4 + pz r = p[(l-zr)/(l-z)] where:  z = e-Vf* f is the frequency of stimulation (sec--*-) r is the time constant of decay of potentiation  133  'm' 1.0  0.5  0.2  0.1  0.05 0  Fig. 29  Potentiation:  20 40 60 stim. frequency (Hz)  a  log-linear  relation  80  between  release and tetanic stimulation frequency. An example in a single junction of tetanic potentiation with nerve impulses under conditions of low quantal content, with 66 juM bekanamycin present, 0.2 Ca  , 2 Mg^+.  134 This approximates to: P = p»f»t where f»r >> 1 Thus, development  the  amount  of  is directly  potentiation  proportional  to  at the  its  full  stimulation  frequency, and if it is a multiplicative factor in release, then  a  logarithmic  plot  of  release  against  stimulation  frequency should be linear. The parallel nature of the plots implies that the same multiplicative process affects both phasic and non-phasic release equally. (2) Fourth root transform In the absence of bekanamycin and especially when the tetanus was prolonged and at a high frequency ( > 40 Hz), the f m potentiation generally outstripped that of m, as seen in Fig. 30. by  fm  On the assumption that this outstripping of m  might  reflect  a  component  of  fm  supported  by  intracellular Ca 2 + (or M g 2 + ) , the phasic delta fourth root was plotted for each 2 s time bin. multiplicative  component  completely developed  of  Fig. 30 shows that the  potentiation  actually  is  rather early on in a tetanus, the  continuing creep upwards of m and more dramatic rise in fm with persistent stimulation being attributable to a gradual accumulation  of  intracellular  divalent  agonist.  The  apparent time constant r for multiplicative potentiation at the junction shown was 9.1 s.  That tetanic stimulation  135  0.3 m M C a 2 + 2 m M M g 2 + 64 Hz nerve stim.  0  10  20  30  40  50  60  70  80  time (sec from start)  Fig. 30  Relative effects on m and f m of prolonged tetanic  nerve stimulation. Stimulation was at 64 Hz under conditions of low quantal content, 0.3 Ca 2 + , 2 Mg 2 + (Expt. 88923e).  136  Effect of a prolonged tetanus on phasic release 6.0 r  0.3 mM C a 2 + 2 mM M g 2 + 64 Hz nerve stim.  a _a u ^ S  5.5  a  nuan°ua  a  a° • ^> I  q  a  D  ° n  a  5.0 T = 9.1 S  a  sP- 4.5  4.0  0  Fig. 31  10  20  30 40 50 60 Time (sec from start)  70  80  Plot of phasic delta fourth root over duration of  a tetanus. Same data as in Fig. 30, plotted as the phasic delta fourth root transform. observed was 9.1 s.  The best  fit  r for the growth phase  137 might  lead  to  increased  intracellular  Ca 2 +  has  been  suggested by a number of workers and there have been two major mechanisms proposed:  (1) simple accumulation of Ca2+  (Katz and Miledi, 1968), and (2) increase of intracellular Ca 2 + secondary to Na + accumulation (eg. Misler et al, 1987; Nussinovitch  and  Rahamimoff,  1988).  In  the  present  experiments, the stimulation protocol was not designed to allow determination of the time constant of the apparent Ca 2 +  accumulation  phase  of  potentiation,  although  in  preliminary experiments with prolonged high frequency (50100 Hz)  stimulation,  upward,  in  minutes.  excess  fm of  appeared  to  multiplicative  continually  creep  potentiation,  for  This is consistent with the time constant of about  100 s determined by Nussinovitch and Rahamimoff (1988). (3) Near absence of extracellular Ca 2+ The predominantly multiplicative nature of potentiation was best demonstrated by its apparent independence from Ca 2 + entry.  In order to eliminate any significant Ca 2 + entry,  the preparation 2 rtiM Mg 2 +  and  was  superfused  50 }M bekanamycin  with to  solution reduce  containing  Ca 2 +  influx,  extracellular Ca 2 + reduced to 10 jM (total) with 100 pK EGTA added  to  chelate  extracellular  Ca 2 + ,  and  6% dimethylsulfoxide (DMSO) added to increase f m to allow an accurate determination of release rate in 0.1ms bins.  In  Fig. 32, the latency histogram of release shows a clear but tiny e.p.p. and frequency-dependent potentiation of both  138  80 r •  • 70 Hz, 32002 stimuli  o  o 25 Hz, 22562 stimuli  60 CO  /w>%4*¥^ WA  40  20  C«0,Q<i  o^ 0  °0 1  -O OP  '  -i  L  0  flft&&A_9^&i6& o  •  latency (ms)  9+  Fig. 32  Survival of potentiation in the absence of Ca*  entry. Prolonged tetanic stimulation in the presence of very low Ca 2 + with sufficient Ca 2 + chelator present to render the extracellular Ca 2 + concentration effectively nil, also with Ca 2 + channel blocker bekanamycin present to further decrease any possible  entry  of  Ca*  or  any other  9+  solution  was  2 mM Mg* ,  cation.  .  50 }M bekanamycin,  100 jiM EGTA, and 6% DMSO (Expt. 88509d) .  The 9+  10 jM Ca* ,  139 phasic and non-phasic release.  It is noteworthy that the  potentiation was greater at 70 Hz than at 25 Hz and that fm/ which is in the expected range for 6% DMSO regardless of extracellular  Ca^  concentration,  was  potentiated  about  9-fold at 70 Hz, within the normal range for potentiation at low extracellular Ca  concentration.  The potentiation of  phasic release is present despite a maximum phasic release rate less than double the non-phasic release rate, ie. an m of less than 1/10,000 normal physiological quantal content at the neuromuscular junction. release  can  occur,  although  It is striking that phasic it  is  very  minute,  under  conditions in which it seems just as likely that the driving force for Ca^ the  inward  be in the outward direction as it would be in direction  extracellular approximately  (Rotshenker  concentration 10-8M  of  et  al,  free  197 6)  Ca^  ions  since was  (estimated from data in Fabiato and  Fabiato, 1979). b) Na + dependency Using direct, current clamp pulses of various durations between  0.1  and  0.5 ms  and  of  various  amplitudes,  potentiation was not attainable, despite the presence of clear  e.p.p.s.,  in  the  presence  of  TTX  presynaptic voltage dependent Na + channels.  to  block  the  Fig. 33 shows a  typical example of the progress of m and f m during a 20 s tetanus for which the 'direct' stimulus was just sufficient, in solution containing 1 mM Ca 2 + and 1 mM Mg  (and TTX), to  elicit an e.p.p. similar in magnitude to that commonly seen  140 with nerve stimulation in low Ca^  raised Mg^  containing  solution (eg. 0.3 mM Ca 2 + and 2-4 mM M g 2 + ) , that is, an m at the beginning of the tetanus of about 1.  Upon commencing a  tetanus with directs, m and f m appeared only to facilitate in the first few pulses (not visible in the figure, since data from 100 stimuli (1 s) are binned for clarity), then the in declined to a minimum of about one-half the m observed with a single pulse.  This lack of potentiation was not  likely due to any kind of depletion, since the quantal content was less than 1 in most experiments.  141  100 T  10  •  D  •  •  • •  •  •  •  •  •  • •  D  • m D  fm  1  0.1 10  15  20  time (s)  Fig. 33  Lack of potentiation with 'direct' pulses.  Potentiation was not obtainable when tetani were carried out using direct pulses in the presence of TTX.  In this typical  example, the stimuli were current clamped pulses of 5 piA and 0.15 ms at 100 Hz; the solution contained 1 mM Ca 2 + , 1 mM Mg 2 + , 0.4 JL/M TTX and total K + of 10 mM. averaged in 1 s bins for clarity.  The data were  142  IV. Discussion  143  A. Phasic release time course 1. the presynaptic Ca' channel According to the Ca^  hypothesis for release, the time  course for phasic release should be predominantly affected by the time course of the presynaptic voltage-dependent Ca 2 + channel.  Present evidence indicates that the type of Ca 2 +  channel present at the neuromuscular junction is one which either does not inactivate (tested with stimulus durations up to  100 ms  in  the presence  of  Ba^ ) or  inactivates  partially with a time course too fast to observe with the present technique. The activation kinetics of the neuromuscular junction Ca  channel could not be studied by the present technique  of indirect measurement using tails of raised f m in presence of Ba"6  since no delay of time course of activation was  noticeable upon stimulation with direct pulses of less than 1 ms. Thus, according to classification of CaA  channels by  their activation and inactivation kinetics and voltage and drug sensitivity (Nowycky et al, 1985), the Ca 2 + channel at the neuromuscular junction appears not to be one of the L, T, or N types which are represented in a wide variety of other electrically  active tissues, but has  some of the  pharmacological characteristics of the P channel (Uchitel et al,  1992).  While voltage  neuromuscular junction Ca^  and drug  sensitivity  of the  channel were observed in the  144 present study, they were studied only as they pertained to experiments with other objectives. 2. similarities for divalent agonists — temperature studies , Sr 2 + and Ba 2 + exhibit the same values for  Since Ca the  three  time  parameters  of  phasic  release  latency, rise time constant of probability  (minimum  function and  decay time constant), then two possibilities are as follows: 1  the forward and reverse rate constants for binding and for effect are very fast, and the observed time course of release reflects primarily the time course of phasic divalent  cation  entry  during  channel  opening  and  subsequently the diffusion limited onward rate constant at the receptor, or 2  binding  and  efficacy,  and  all  the  rate  constants  inherent therein contribute substantially to the time course of release, but  are very similar for the three  divalent agonists. The above possibilities may be further investigated using temperature as a probe (between 5°C and 35°C).  If the first  one represents reality, then Qio's f°r phasic release time course should be near unity and the time course for phasic release should be invariant with temperature changes. noteworthy  that  parameters  is  although  the  greater  than  Q^Q  for  unity,  all  It is  three  time  especially  for  temperatures lower than 15°C, the three parameters appear to be equally prolonged by temperature reduction, consistent  145 with  the  first  exocytosis  scenario  process  at  a  with  temperature  step  after  affecting the  those  divalent agonist (Datyner & Gage, 1980).  promoted  by  Datyner & Gage  suggest that the inflection in the temperature dependency toward high Qio seen as the temperature was dropped below 15 °C may reflect a lipid phase change.  These data are  consistent with a model for phasic release whereby a bolus of  divalent  intracellular  agonist active  appears  and  site with  disappears  diffusion  at  the  limited rates,  triggering an exocytotic event for which the rate limiting step involves the reorganization of vesicular and plasma membrane lipids. If the second possibility represents reality, then the above result of Datyner and Gage (1980) with Qios as large as 4 could be consistent with a rate limiting mechanism for phasic release involving protein, for which a high Q^Q i s not uncommon, but it would not explain the non-linearity of temperature  dependence  on  the  basis  of  a  single  rate  determining step. 3. model-fitting in retrospect Some  features  of  the work  of  Barrett  and  Stevens  (1972a,b) and of Datyner and Gage (1980) on phasic release time course can be explained in terms of a combined model for phasic release and facilitation.  Given in more detail  later, this combined model proposes that a transient of Ca 2 + elicits phasic release, while facilitation can be due to residual  Ca 2 + ,  or  a  multiplier  secondary  to  the  Ca 2 +  146 transient, or a combination, as suggested by Silinsky (1985) and  others.  The  combined  model  presented  here  for  neurotransmitter release simplifies the analysis of data in which  there  are  apparent  non-exponential  or  multi-  exponential time courses of phasic release, as follows. First, the time course of the e.p.p. decay as studied by  Barrett  relation  and  Stevens  between  In  (1972b) did  R(t)  and  time  not  give  where  a  R(t)  linear is  the  freguency of guanta during each small increment of time (eg. 0.1 ms) after the peak of the e.p.p.  Although it is not  possible to attempt further guantitative analysis on their data as published, it is possible that a linear relation would have been achieved through a transformation according to the fourth power residual ion model, ie. a plot of In [R(t) 1 / 4 - fro1/4].  The fourth root difference should be a  measure of the magnitude of the transient intraterminal Ca 2 + at any time during the e.p.p. and is best suited to the model for a rate limiting factor in the release process which may decay exponentially. Second,  in  frog  (with  bathing  solution  containing  O J.  Ca^ ), in which much of the previous work has been done, the apparent time constant of phasic release decay, obtained from  the  slope  of  log  facilitation develops.  release  vs  time,  increases  as  The same result would be obtained  for the time constant of phasic release decay determined in •J A.  the same way in Sr^ , according to the present data.  These  results are not in accord with a linear release process  147 decaying  exponentially  and  summing  between  successive  stimulus pulses, since this would require a further factor to explain the change in apparent time constant.  On the  other hand, a linear process to which release is related by a power  function would  give a consistent time constant  despite various amounts of residuum from previous pulses. In the data of Barrett and  Stevens  (1972b) it  is not  possible to determine which power transform would maximize the consistency of the time constant of e.p.p. decay among various stimulation rates. Third, the apparent discrepancy between the work of Datyner and Gage (1980), in which the decay time course of phasic release did not change with successive stimuli, and the work of Barrett and Stevens (1972b), in which the phasic release appeared to be prolonged with successive stimuli, might  actually  intracellular  reflect release  a  time  The  time  the  obtained  from  not  the  repetitive stimulation and is consistent in both cases. be  is  at  by  would  which  Ca 2 +  of  altered  constant  sites  course  a  plot  of  difference in fourth roots of the phasic and non-phasic release rates prior to measuring decay time course. if  residual  Ca 2 +  played  a  predominant  role  Thus, in  the  facilitation observed by Barrett and Stevens at 11°C in the frog, the increase in the amount of residual Ca 2 + present after a pulse would be seen experimentally as an increase in interpulse f m which greatly exceeds the increase in m as the train proceeds.  The slope of the log of the difference  148 between initial  and  final  (just before the next pulse)  phasic release rates plotted against time (apparent inverse rm) would  increase with  successive  stimuli, whereas the  slope of the log of the difference between initial and final fourth  roots  intracellular constant.  (according Ca^  to  during  postulated  phasic  time  release)  course  would  of  remain  From Barrett and Stevens' published data it is  not possible to determine whether the fourth root transform would eliminate the phenomenon of prolonged phasic release decay with successive stimuli in a train, but their report that test ERP (phasic release) is facilitated less than is the  tail  of  release  (fm)  indicates  that  this  is  so.  Furthermore, that the facilitation observed by Barrett and Stevens resulted from stimulus intervals of 100 ms (10 Hz), whereas that observed by Datyner and Gage and reported herein was only significant at intervals of about 20 ms or shorter,  also  component  in  points the  data  to of  a  predominant Barrett  according to the present work Ca  and  residual Stevens,  Ca^ since  has a longer decay time  than the multiplicative component of facilitation. Finally, the  jump in the temperature dependency of  release from Qios of about 1 over 15C to about 4 under 15C may not indicate a single process such as lipid phase change (as postulated by Datyner and Gage, 1980), but rather a change from one process to another process being the rate limiting step as the temperature changes.  That the Q^Q f° r  one process is near 1 may simply indicate that that process  149 is limited by diffusion. release in elevated K  At lower temperatures, a Q^Q  in the presence of Ca^  ror  of nearly 4  may not necessarily indicate a highly temperature sensitive protein involved  in release, since a temperature change  giving a 4-fold change in release may do so by changing the Car+ channels in such a way as to change intracellular Ca^+ by only studies  1.4-fold. a  suitable  It is apparent model  is  that  needed  on  in temperature which  to  base  transformation of the data prior to constructing a van't Hoff  plot.  If the model  used  is the  combined model,  concurrent data on m and f m must be obtained in order to distinguish between temperature effects. 4. time course of BAPTA effects In Sr  , after BAPTA loading, the time constant of non-  phasic release was prolonged and the magnitude of non-phasic release was reduced to a greater extent than was phasic release.  These data fit a model in which BAPTA at the  intracellular release sites chelates a fraction of the Sr as it enters through the channel, not allowing it to bind the nearby release site and promote release.  This would  explain the observed reduction in phasic release in Ca^ + and in Sr 2 + after BAPTA loading.  Subsequently, Sr 2 + which was  immediately chelated upon entry and Sr^  which was chelated  after binding the release site during phasic release is released from the BAPTA, allowing it to bind the release site and play a part in promoting non-phasic release.  The  greater the concentration of BAPTA in the nerve terminal,  150 the greater the probability that Sr^  will remain either  bound to BAPTA or to the release site.  Thus, with Sr 2 + able  to oscillate between these two bound forms intracellularly, the time constant of the non-phasic delta fourth root is increased, as follows. In the absence of BAPTA, the concentration of Sr 2 + at the release site is Sr(t) = Sr^xe-tA With the addition of BAPTA, after the rapid equilibration of Sr 2 + with BAPTA, this becomes Sr(t) = A e - t / r l + Be _ t / r 2 where A and B add to give total Sr^ entry.  This is a pharmacokinetic consideration, analogous to the prolongation of elimination time constant for a drug which is highly protein bound relative to one which is not, all other factors being equal. Assuming the above, it should be possible at low m to show that the inhibition by BAPTA of non-phasic delta fourth root disappears if one integrates over a time period much longer than the apparent time constant of Sr  removal.  B. Models of release enhancement 1. Residual ion model The residual Ca 2 + hypothesis is that Ca 2 + is a common factor residual  for  evoked  Ca^  assumes the Ca^  release  model  for  and  for  facilitation.  facilitation  (see  The  INTRODUCTION)  release model of Katz & Miledi (1965) and  151 further postulates that a fraction of the Ca-6  which enters  phasically after a stimulus remains in the nerve terminal, adding to the Ca^  which enters after a subsequent stimulus  (Katz & Miledi, 1968), according to R = k (C t +C r ) 4 . For experiments carried out in the presence of Ca 2 + , the present  data confirm  facilitation  which  can  be  that there ascribed  is a component of to  an  increase  in  residual Ca 2 + , C r/ but that most facilitation (particularly of m) is due to a multiplicative process expressed as an increase  in  k.  This  concurs  with  the  multiplicative  character of facilitation as previously described by Hubbard (1963) . For Sr 2+ , as previously shown for Ba 2 +  (Quastel and  Saint, 1988), the present data are consistent with a pure residual  ion  model  based  on  the  residual  Ca^  model,  according to a number of tests. First, the apparent entry of Sr 2 + (difference in fourth roots of non-phasic release rate just after a short high frequency  train  and of  release  rate  in the absence of  stimulation) appears to be linear with the number of pulses, as shown previously for Ba^ .  In Sr^ , however, the time  constant of apparent loss of Sr^  from the active site is  much less than that of Ba^ , requiring a correction for loss during the stimulation-induced buildup. Second, facilitations of the phasic component and of the non-phasic component of release in Sr 2+ , i.e. of m and  152 fm, are mutually predictive according to the residual ion model. Third, for stimulation in the presence of Sr^ , the results fit a residual ion model, with an n of 4 or 5.  This  fit is confirmed by the equality of development and decay time constants of f m after transformation to the fourth or fifth root. In the following section, various models which have been studied in attempts to accomodate the discrepancy of facilitation in Ca  -containing solutions from a simple,  additive fourth power residual ion model, while maintaining the underlying assumption of that model, are discussed. 2. Variations of the residual ion model a) cooperativity In order for the present results to be interpreted solely  in  terms  of  a  residual  ion  model,  modifications of the model could be entertained.  various One such  model would be that in which the apparent cooperativity of Ca^  (n)  must be very high under low m conditions, but  decreases with increased m to a minimum of 4 at high m (presumed high Ca 2 + entry and residual Ca 2+ .  However, the  data for Sr 2 + (see above) and for Ba 2 + (Quastel and Saint, 1988) show no evidence for n not always being the same. b) inhibitor of fm Another  possible  complication  of  the  residual  Ca 2 +  hypothesis that can account for apparently multiplicative facilitation  is  the  inclusion  of  an  unknown  inhibitor  153 specific for non-phasic release which is not fully expressed except  under  those  conditions  where  multiplicative component is expressed.  the  apparent  There is no evidence  for the existence of such mechanism, which requires the invocation of separate release systems for phasic and nonphasic release, contrary to previous and present findings (eg. Guan et al, 1988; see also present RESULTS and DISCUSSION re Sr 2 + ). c) Ca voltage hypothesis The hypothesis put forward by Parnas' group (Parnas et al, 1986) is sufficient to account for the observation of multiplicative facilitation and potentiation in the presence of Ca^  while being able to maintain that these processes  arise as a result of the accumulation of intracellular Ca 2 + from previous stimulation. proposes  voltage  activated  intracellular receptor  Parnas' Ca-voltage hypothesis affinity  for  Ca^  of  an  (Kc, see Derivations, in METHODS).  Thus, for conditions under which the amount of intracellular is sufficiently low that the Ca1*  Ca  receptor is far from  saturation, this model can be restated in terms of Equation 2 as: r(t) = k(Vm) ( C t + Cr+ C 0 ) 4 , where  the  multiplier  'k' is modulated  by  the membrane  in  the  potential V m . In  this  model,  Ca  would  persist  enhancing both phasic and non-phasic release.  terminal  However, the  Ca entry resulting from the nerve terminal depolarization  154 would  be  necessarily  rather  small, to  account  relatively low f m even during facilitation.  for the  For example,  one would observe two-fold multiplicative facilitation (for both m and fm) and account for it entirely by residual Ca, with  the  two-  instantaneous  to  release  three-order rate  of  magnitude  corresponding  to  m  greater than  fm  explained by a large phasic increase in 'k', as follows: eg. facilitation of m = 2. m'/m = ( C r + C t + C 0 ) 4 / (Ct + C 0 ) 4 = 2 fm'/fm = ( C r + C 0 ) 4 / C 0 4 = 2  For equal multiplication of m and fm, C^ = 0 and the multiplier k, which cancels out from the equations for m and fm  facilitation as shown above, is then the only factor  responsible for the phasic component of release, and must be a  voltage-dependent  variable  whose  magnitude  is  approximately equal to the ratio of phasic to non-phasic release rates and whose time course is mirrored in the time course of phasic release. The Ca-voltage hypothesis thus accounts for the time course of phasic release simply by postulating that this reflects predominantly the time course of k, as a function of  membrane  observations  potential.  This  is  consistent  with  the  that the time course of phasic release is  independent of the particular active divalent cation which is present to support release (Ca, Sr or Ba; Datyner and Gage,  1980;  Quastel  et  al,  1989)  and  their  apparent  155 differences in time course of sequestration in or removal from the nerve terminal (Zengel and Magleby, 1981). The problems with this model are threefold.  First, it  does not account for phasic release in Sr 2 + and in Ca 2 + without imposing large differences between the ions either for intracellular potency or for permeability through the open voltage dependent Ca^ supported.  channel, neither of which is  For example, with m = 1 and f m = 1/s, a voltage  dependent k would have to increase phasically by about 1000 fold in the absence of a large Ca'  transient, such absence  being necessary to maintain the fit of the Ca-voltage model to multiplicative facilitation.  The present evidence shows  that Sr 2 + is able to support phasic release in the same order of magnitude as Ca^ , but without such a phasic, voltage-dependent  increase  in k  (see Fig.  14).  To be  consistent with the present results, Sr 2 + would need to be much more potent than Ca^ , by a factor of about 1000. This is inconsistent with the work of Silinsky (review: 1985) in which  combinations  pretreatment with LaJ receptors  for  efficacy of Ca 0.5,  Ca 2 +  of  Ca 2 +  and  Sr 2 +  were  used  after  to reduce the number of intracellular or  Sr 2+ ;  the  data  showed  that  the  and Sr 2 + range from 9 to 20 and from 0.2 to  respectively  (Silinsky,  1981).  The  Ca-voltage  hypothesis is thus not consistent with the present  Sr 2+  results. Second, while the consistent  time course among the  active divalent cations for phasic release has been used in  156 support of the Ca-voltage hypothesis, this consistency casts doubt on the validity of the hypothesis in view of the preceding paragraph.  That is, if phasic release in the  presence of Sr 2 + is triggered only by a transient Sr the  intracellular  active  sites  secondary  to  at  voltage  dependent cation channel opening, whereas phasic release in the presence of Ca^ a  multiplier  is triggered only by a phasic change in  (k), it  is  unlikely  that  such  disparate  mechanisms for phasic release would be associated with the same minimum latency and time course of rise and decay as shown in the present results (Fig. 13). Finally, although the argument is still ongoing whether or  not  a  depolarization  in  the  absence  of  Ca  (or  surrogate) entry can evoke phasic release using novel caged calcium compounds (Hochner et al, 1989; Mulkey & Zucker, 1991), the observation that agents which can be shown to block Ca 2 +  entry are able to block phasic release (eg.  bekanamycin: Bourret & Mallart, 1989; Guan et al, 1988) suggests that if a voltage effect exists which is contingent on resting  or  slightly  elevated  intracellular  Ca 2 + ,  it  should be manifest in a minimum e.p.p. which is not further reduced by increasing the concentration of any one of these entry blockers.  In the present results, a minimum e.p.p.  was evident after stringent measures were taken to prevent Ca 2 +  entry and intracellular Ca 2 + was depleted, but the  phasic effect of the stimulation was very small.  157 On the other hand, Neher (1988) demonstrated that for mast  cells, resting  intracellular  Ca^  concentration  is  sufficient for exocytotic release to occur, provided that GTP-gamma-S is present intracellularly, but in its absence transient  increases  in intracellular  Ca^  did not evoke  release. d) saturation of Ca"  dependency  Katz and Miledi (1968), Younkin (1974), and many others have observed that facilitation is dependent on the presence 7+  of some extracellular Ca^  ions.  However, Charlton and  Bittner (1978a), Dudel (1989) and others have shown that facilitation of m is largely independent of extracellular Ca 2 + over a range of concentrations modulating m up to 12fold.  The  present  results  are  observation for facilitation of m.  consistent  with  this  Although the residual  Ca 2 + hypothesis can theoretically predict such behaviour of m facilitation (as described in the INTRODUCTION), it does not concur with the current evidence that although facilitation stays quite  constant over changes in extracellular  Ca 2 +  concentration which change m more than 10-fold, resting f^ does not modulate over a similar range as m.  That the  residual ion model was originally postulated based on m data without  including  concurrent  fm  data  explains  the  difficulties in trying to account for facilitation of both m and fm.  158 3. Multiplicative model A multiplicative model would require a mechanism which (1) would only be active in the presence of minimal Ca 2 + entry (extracellular Ca of more than about 50/JM in 4 mM Mg 2+ ) and (2) would be strictly multiplicative in nature. This  multiplier  would  act  by  a  mechanism  which  would  increase release either by affecting the affinity of an intracellular Ca-binding active  site or by  lowering the  energy barrier for fusion of the neurotransmitter vesicle with the presynaptic membrane. Mallart and Martin  The latter was alluded to by  (1967) in their speculation that the  release mechanism is rendered hyperexcitable by a preceding stimulus. In  accord  with  the  present  evidence  that  the  interaction of a putative facilitator mechanism with the rest of the release system is multiplicative, there is also the question of how the facilitator interacts with itself. That is, does additional multiplier add to or multiply with that present (from previous stimulation).  For ethanol and  DMSO,  neurotransmitter  pharmacological  multipliers  of  release (Quastel et al, 1971; McLarnon et al, 1987), all release present is multiplied, even that which is the result of  an  unrelated  multiplicative  process,  such  as  facilitation.  It is not evident from the present results  whether  hypothetical  the  facilitation  multiplies  multiplier  all  facilitated release rate.  release,  responsible or  only  the  for non-  There are some regions of the  159 crayfish  opener  multiplicatively  muscle and  where  other  the  facilitation  regions  where  it  builds builds  additively (Robitaille and Tremblay, 1991). If  multiplicative  facilitation  does  indeed  build  multiplicatively, it would be acting in a fashion consistent with a model whereby intracellular Ca* , raised for a short time  after  intracellular  each  stimulus,  release, of  evokes  the  formation,  a multiplicative  or  modulator of  release, which acts in a manner similar to ethanol or DMSO. Biochemically, there are a great many intracellular Ca^  candidates, since  activates many enzymes whose products are  potentially multiplicative modulators of release. the  neurotransmitter  Lipids, possibly one or more species involved in  phosphoinositide  vesicular  origin  candidates. activated  pathway,  neurotransmitter are  Small molecule metabolite products of  Ca 2 +  are  exocytosis  of  reuptake,  enzymes  during  or  also  and  candidates.  A  systematic  assessment of the effect of many normal metabolites on phasic  and  non-phasic  release,  and  on  multiplicative  facilitation, has not been done. A model for facilitation involving enzyme or active carrier activity  is suggested by temperature  facilitation time course  studies of  (Balnave and Gage, 1970) which  showed that the time course of decay of facilitation of high quantal content m was slowed with low temperature with a Q^Q of about 4. support  the  It is noteworthy that this result does not residual  Ca  hypothesis, since the  simple  160 9a-  addition  of  a Ca^  residuum  from  preceding  stimulation  should have no different temperature sensitivity than Ca 2 + influx-evoked release itself.  On the other hand, it is  important that the Q^Q f° r growth of facilitation is shown to parallel that for its decay for a hypothesis involving a metabolite as multiplier to be valid. 4. Combined model - multiplicative and additive In order to test the present hypothesis that distinct multiplicative  and  additive  (residual  ion)  processes  contribute to facilitation, two effects on release rate which have time course and magnitude in the same order must be  accurately  and  appropriately  separated.  Such  a  separation was clearest when experimental conditions were varied to maximally express one or the other.  With high m  and nerve stimulation both components are expressed but the additive component lowest  is usually predominant, while at the  extracellular  Ca^1  concentration  at  which  multiplicative facilitation is still fully expressed (low m) the additive component was almost completely absent. C. Mechanisms of multiplicative facilitation 1. presynaptic ion channels a) action potential Action potential prolongation is not likely to play a role in the induction of facilitation.  In the present work,  we have shown that although the presynaptic spike becomes progressively  delayed  upon  the  commencement  of  a  high  freguency train stimulation (70-90 Hz), there is no evidence  161 for spike broadening, which would presumably prolong Ca 2 + influx and thus prolong the histogram of times of quantal release, since the release histogram is virtually the same at different periods during a train except for the latency change.  Likewise, changes in the electrical excitability of  the nerve terminal or in spike amplitude are unlikely to induce facilitation since facilitation can be observed using direct pulses given in the presence of TTX and K + channel blockers. b) membrane potential It has been shown that nerve stimulation can result in time dependent changes in presynaptic membrane potential which outlast the action potential (Gage and Hubbard, 1966). Either hyperpolarization or depolarization or a convolution of both  in time after the  action potential  (or direct  stimulus) have been demonstrated in several neuronal types, usually in somata.  If a change in polarization of the  presynaptic nerve terminal were in some way dependent on presynaptic Ca* , and if this change in polarization altered the  phasic  component  of  intracellular  Ca 2 +  without  significantly altering its apparent time course, this could be a mechanism of m facilitation, in conjunction with an 9+  additive residual Ca^  component to raise fm.  Both of the  following alternative explanations for parallel facilitation involve residual Ca  while at the same time attempt to  explain the fact that observed f m facilitation is much less than that predicted by the residual Ca 2 + model, based on hte  162 observed extent of m facilitation. raised  that  perhaps  residual  Thus, the possibility is  Ca^  can  account  for  f^  facilitation, and that m facilitation depends on a process which is in turn dependent on residual Ca^ . However, it is unlikely that an enhancement of m and f m resulting from such a  mechanism  would  be  as  well  stereotyped  as  observed  multiplicative facilitation in the presence of Ca^1 . (1) endogenous presynaptic hyperpolarization Post  stimulus  hyperpolarization,  called  after  hyperpolarization (AHP), is usually attributed in neuronal somata to an increase in K + conductance (gKca) activated by Ca 2 +  whose  concentration  decays  after  the  consequent to the somatic action potential.  bolus  entry  That AHP occurs  in nerve terminals and sums during a tetanus has been shown under certain conditions (Gage and Hubbard, 1966).  It is  possible that if this occurred, it may result in a larger action potential without any change in duration, the larger voltage swing being somehow responsible for facilitation. Although  hyperpolarization  inactivation,  it  is unlikely  can that  remove  Na +  under  normal  channel or  low  quantal content conditions a large enough fraction of the nerve terminal Na  channels would be inactivated such that a  hyperpolarization-induced increase in action potential size would be evident.  Presynaptic hyperpolarization has been  shown to enhance neurotransmitter release  (Parnas et al,  1986), although it is not apparent whether the enhancement is predominantly multiplicative, nor is it obvious that an  163 endogenous hyperpolarization would have the same effect. Nevertheless, presynaptic hyperpolarization, resulting from gKca, could account for an increase in m, with f m increasing 7+  almost in parallel due to residual Ca^ , both of which are dependent on Ca 2 + entry and its residuum after the Ca 2 + channels  close, but only the f m  enhancement  due to an  7+  additive effect of residual Ca^ . (2) endogenous presynaptic depolarization Presynaptic depolarization which is initiated with each stimulus  and  decays  thereafter  could  be  involved  in  multiplicative facilitation through an increase of action potential size, although a mechanism by which a persistent presynaptic depolarization could be dependent on Ca 2 + entry and  its  residuum  depolarization  will  is  not  known.  reduce  the  A size  small of  a  presynaptic depolarizing  stimulus necessary to elicit a presynaptic action potential if the membrane potential is much more negative than the activation voltage of the Na  channels, but will increase  the size of the stimulus required if the depolarization is sufficient to inactivate some of the voltage dependent Na + channels.  Thus,  a  mechanism  by  which  presynaptic  depolarization increases the action potential size is not 7+  clear.  Furthermore,  a  Ca^ -dependent  post-stimulus  depolarization has not been shown. Depolarization, as a mechanism possibly involved in multiplicative facilitation, is attractive because of its consistency with the possibility that depolarization-induced  164 9+  raised  fm  in  Ca  containing  solution  frequent, local influxes of Ca"  nerve  terminal  were  a  result of  rather than global Ca"  accumulation in the nerve terminal. the  is  If depolarization of  sufficient  to  account  for  facilitation of the e.p.p. (perhaps by increasing the a.p. size or by priming the Ca"  channels in some way), then it  is possible that the same depolarization could account for a relatively small facilitation of f m which in Ca" -containing solution  is much  residual Ca  smaller than predicted by an additive  model.  c) Car  entry per pulse  Nerve terminal Ca"  current inconsistency from stimulus  to stimulus could result from a variety of processes known to  occur  within  various  excitable  cells,  including  stimulation induced changes in the action potential, changes in  the  transmembrane  Ca"  mactivation of the Ca"  gradient  channel.  and  priming  or  Any of these changes  during repetitive stimulation would play a major role in determining the relative facilitation of phasic and nonphasic release, and would also largely determine the time course of the buildup of facilitation as the series of stimuli progressed. If Ca 2 + entry per pulse decreased with each stimulus, facilitation  that  would  could appear additive: not  reduced  otherwise  appear  multiplicative  as long as the entry of Ca 2 + were  sufficiently  to  block  the  development  of  multiplicative facilitation of both m and fm, m would suffer  165 from the decrease in Ca^  entry while f m would not.  Thus,  the facilitation would appear similar to a residual Ca 2 + hypothesis prediction assuming consistent Ca 2 + entry.  In  the present work, this was occasionally seen to occur when the nerve terminal action potential failed late in high frequency trains. For synapses at which facilitation is predominantly multiplicative and is maximal at about 2-fold, a progressive decrease in Ca  entry might progressively  quantal  with  content  a  time  course  detract from  similar  to  the  development of facilitation, the concurrence of the two distinct processes resulting in an apparent reduction or abolition of facilitation.  It has been pointed out (Wang  and Quastel, 1991) that this might be the explanation of the apparent block of facilitation after the addition of Cdr+ (Dudel, 1990; Zengel et al, 1988), Zn 2 +  (Zengel et al,  1988), and Pb 2 + (Wang and Quastel, 1991). On the other hand, an increase in Ca^  entry per pulse  would tend to increase quantal content more than predicted by  a  residual  ion  model,  possibly  appearing  as  a  multiplicative effect on m and f m even if the facilitation were predominantly due to an underlying residual ion process (mvolving divalent cations such as Sr^  or Ba ).  2. Potentiation and facilitation a) Shared mechanisms? Both  potentiation  and  facilitation  multiplicative and additive components  apparently  have  (discussed later),  166 and are mutually multiplicative. multiplicative  components  Is it possible that all  of  neurotransmitter  release  enhancement, without regard to time course, arise from the same presynaptic process? The  presence  of  occludes facilitation. model, with  potentiation  neither  obviates  nor  This is consistent with the combined  potentiation  sharing  a distal  step  in its  mechanism with facilitation, with the following provisos: 1)  the  process  leading  to  potentiation  mvolve a further rise in intracellular Ca^  must  not  from that  which occurs for facilitation, since present findings indicate that very low intracellular Ca 2 + concentration is necessary to saturate Ca^  dependent multiplicative  facilitation; 2)  the shared step, if saturable, must be far from  saturation  under  the  conditions  of  the  present  experiments; and 3)  if  the  shared  step  is  the  final  step,  it  apparently must be capable of being activated by two different proximate effectors. If facilitation and potentiation, as described, represent phenomena which have discrete mechanisms, it is possible that they share the ultimate mechanistic step which leads to the multiplication of release.  That the two phenomena do  not share all steps in their mechanisms is evident from the differences observed experimentally.  167  Comparison of multiplicative potentiation and facilitation.  observation  potentiation  blocked by tetrodotoxin  facilitation  yes  no  blocked by botulinum toxin A ** prolonged rd e c a v in low K  yes yes  no no  occluded or obviated by the other  no  no  time course  20s to minutes  -80ms  *  P. Sun & D. M. J. Quastel (unpublished observations) G. Polyakov & D. M. J. Quastel (unpublished observations)  b) Intracellular Na (1) Cooperation with Ca 2+ While  Na +  intracellular  has  been  proposed  in  the  combined model to promote the intracellular appearance of a multiplicative factor, a direct role for Na + in release must be considered.  A simple model including Na + is as follows:  r = k[C(t) + C 0 + qN(t) n +qN 0 n ] 4 where  r  is  intracellular Ca"6  release  at  and Na  any  time  t,  C  and  N  are  concentrations at the release  site (C0 and N 0 in the absence of phasic influx of the ion), q is a constant which refers to the affinity of Na + as a fraction of that of Ca^  for the putative binding site, k  includes the affinity and intrinsic activity of the bound  168 receptor, and n is the number of Na + ions that must be bound to take the place of one Ca^  ion bound.  This model does not account  for the multiplicative  component of facilitation, since Na + is merely acting as a substitute for Ca^ . Unless the Na -dependent component was predominant and its n>>l, the overall apparent cooperativity would be close to 4, precluding prediction of parallel changes in the log of m and of fm.  By the same logic, Na +  cannot produce multiplicative potentiation via an increase in intracelluar Ca^ . (2) Indirect multiplier, F-actin A  possible  mechanism  action, presumably through  for  indirect  multiplicative  'mobilisation' (Hubbard, 1963),  Na + may act intracellularly to increase k by activating the vesicular transport mechanism, increasing the availability of quanta docked for release at the 'active zones' (Heuser et al, 1979; Smith and Augustine, 1988).  Na + has been shown  to activate F-actin function (Bernstein and Bamburg, 1989) and  botulinum  function.  toxin  has  been  shown  to  inhibit  F-actin  Both Na + (Misler et al, 1987) and botulinum toxin  type A (Molgo et al, 1987) appear to be multiplicative in their action on release; that is, both appear to affect k in Equation 2, albeit in opposite directions.  That a ouabain-  induced increase in intra-terminal Na + might be able to partially counteract the effect of botulinum A is suggested by Molgo et al (1987).  169 Much evidence (eg. Atwood and Wojtowicz, 1986; Misler et al, 1987; Nussinovitch and Rahamimoff, 1988) including present evidence suggests that the increase in phasic and non-phasic release by a factor of about 20 during 50 to 100 Hz nerve stimulation at the neuromuscular junction is due to Na +  accumulation.  neurofilaments  The  which  possibility  exists  that  guide the neurotransmitter  the  vesicles  along their journey from the endoplasmic reticulum to the active zones for exocytotic release, such as demonstrated by freeze-fracture electron micrscopy (Heuser et al, 1979), may be the site at which many agents act. that  enhancement  of  actin  function  It is conceivable would  result  in  a  multiplicative effect, a change in the k of Equation 2, by providing closer approximation of the vesicle to the active zone,  reducing  increasing  the  the  energy  requirement  probability  of  for  fusion.  fusion It  is  and also  conceivable that ethanol, DMSO and other agents which affect release multiplicatively also exert their actions through an effect on nerve terminal actin.  c) Presynaptic proteins Whether  the  proximate  step  excitation and multiplicative intracellular  Ca 2 +  or  Na + ,  between  nerve  facilitation or  a  change  terminal  is a rise in in  membrane  potential, there are a number of candidate protein targets, in  addition  alteration  to  F-actin  in function  mentioned could  above,  result  for  which  in a change  an  in k.  170 Synapsin, for example, forms tetramers upon binding Ca^+ (Thomas et al, 1988), and other proteins isolated from nerve terminals appear to be fusogenic.  A number of presynaptic  proteins appear to be activated by Ca^  in vitro with a  cooperativity of 4 and with apparent dissociation constants (cell-free systems) in the region of Ca^  concentrations  anticipated during nerve terminal activation (Crompton et al, 1988; Plattner, 1989). D. Mechanism of additive components Only one hypothesis  seems plausible to explain the  additive components of neurotransmitter release enhancement, without  respect  hypothesis. various  to  The  time  course,  distinctiveness  multiplicative  and  the  of  residual  time  additive  course  Ca^ among  components  of  enhancement makes the separation of the components into multiplicative  and  additive  very  much  easier,  and  is  important evidence in favour of the residual Ca 2 + hypothesis for the additive component.  This evidence is summarized as  follows: 1  Release rate or probability is proportional to the  fourth power of the Ca^ INTRODUCTION),  at the release site (see  and an additive component of enhancement  is defined as one which contributes equally to phasic and non-phasic release fourth roots.  Facilitation and  potentiation, as previously defined by the short and long time courses, respectively, usually include such an additive component.  171 2  The additive components are of greater magnitude  under conditions where Ca^  entry might exceed the  capacity of one or more of the processes responsible for  its removal  facilitation,  rate  this  extracellular Ca  from the release  would  arise  from  site.  For  increases  in  (see Fig. 22) and from decrease in  interval between pulses to intervals less than about Under the very brief condition of Ca 2 + loading  10 ms.  imposed by short trains, the additive component would not be expected to be very large, and would resolve itself with a very rapid decay, as observed in the present data.  For potentiation, on the other hand,  where the nerve terminal has a large metabolic demand placed on  it due to prolonged  conceivable that Ca^  tetanization, it is  loading, and thus an additive  component, might result from a growing inability of an 9+  energy consuming Ca^  removal process to keep up with  Ca 2 + entry. In  the  present  work,  additive  and  multiplicative  components of facilitation and of potentiation have been dissected  and,  with  trains  or  brief  tetani,  the  multiplicative component predominates.  With respect to the  additive  intracellular  component,  concentration  whether  actually  or  rises with  not  a magnitude  and  Ca 2 + time  course consistent with that predicted from the residual Ca 2 + hypothesis obvious.  for the observed  additive enhancement  is not  While 'direct' measurements using fluorescent Ca 2 +  172 indicators have shown that intracellular Ca* during  repetitive  stimulation  and  fall  does rise  thereafter  to a  baseline, these measurements have not yet been made at a mammalian neuromuscular junction.  In addition, a linear  correlation between the magnitude of enhancement of release and the intracellular Ca  concentration (eg. Zucker et al,  1991) is difficult to reconcile to either an additive or a multiplicative model. E. Ultra fast facilitation The stimulation induced enhancement of release which occurs as an early indicate  a  transient.  late  component of non-phasic  component  of  the  phasic  release may release  Ca*  However, any visible elevation of f m after, but  close to, the period of phasic release (ie. about 4 to 10 ms latency),  if  necessitate  it were the  intracellular Ca*  ascribed  existence decay.  of  to a  residual third  Ca  , would  component  of  That this early component of non-  phasic release cannot be simply the residuum of the e.p.p. transient Ca*  falling with a single time constant can be  shown by the following example. One must first assume that the decay of intracellular Ca 2 + follows first order kinetics, that is, that its decay is rate limited in any one component by processes other than diffusion, such as binding and unbinding reactions. transient Ca  ?+  If the  responsible for phasic release decays with a  single time constant, then given a f 0 and m, one can find the r required for a r(6 ms) which is 5% greater than f0.  173 Thus, for m=l, f0=l/s and a 5% elevation in release rate at 6 ms latency (for example) relative to the release rate measured later (say at 20-40 ms; decay of multiplicative facilitation has little effect), c(t)=cp exp(-t/r) c p « (2000m)1/4 « 7 c(6 ms) = (1.05/s)1/4 - l 1 / 4 = 0.0123, and r = 6/(ln 7 - In 0.0123) ~ 0.9ms, much longer r than that observed in the latency histograms, The z prediction gets worse for  about 0.1 to 0.3 ms. greater elevations of fm.  Thus, if this ultra fast facilitation is mediated by residual Ca 2 + , another process for the decay of transient Ca 2 + is suggested, intermediate in time course between the diffusion limited fall of transient Ca*  at the active sites  (r=0.2ms), corresponding to the predominant termination of phasic  release,  and  extrusion (r«200ms).  the  much  slower  process  of  Ca 2 +  One possible scenario is that this  ultra fast (r of milliseconds) facilitatory process might represent the rate of Ca^  unbinding from release sites  which have been activated. F. Spontaneous release 1. Role in the combined model The present analytical method depends upon a model in which all non-phasic release is dependent upon divalent agonist at specific intracellular active sites for promoting release.  Whether the source of this active divalent agonist  174 is  the  membrane  extracellular channels  intracellular  solution  or  stores  the (such  via  source as  permeation is  through  release  mitochondria)  from  makes  no  difference to the analysis, since any increase in active intracellular divalent cation is simply an increase in c r in Equation 2, as applied to that ion. 2. Cause of spontaneous release a) Ca2+-dependent Not all spontaneous quantal release, in the absence of antecedent or concurrent nerve terminal depolarization, is Ca2+-dependent. be  attributed  However, high spontaneous release rate can to  extracellular Ca^"  intracellular  Ca*  when  removal  of  reduces the rate, and also when nerve  stimulation in Ca^" -free solution containing EGTA reduces the rate (by allowing Ca*"  out of the nerve terminal down  its gradient) (Rotshenker et al, 1976). b) Ca2+-independent However, in the present experiments there were some cells in which the spontaneous f m was orders of magnitude greater than 1 and was insensitive to any measure to reduce intracellular  Ca 2 + .  It is possible  spontaneous release is supported by Mg  that this type of , since prolonged  stimulation in solutions containing high Mg^  concentration  sometimes results in raised f m with similar characteristics (Hubbard et al, 1968). An  alternative  explanation  for  apparently  Ca 2 + -  independent high spontaneous release rate is that the cell  175 has a large Na + leak current, resulting in a persistent potentiation, according to the model in which potentiation results  Na +  from  accumulation  (eg.  Nussinovitch  &  Rahamimoff, 1988 results with ouabain) during tetanus and in some way multiplies release.  The Na + explanation might  explain why nerve terminals exibiting this behaviour often do not fire action potentials. A final explanation for Ca' -independent release is simply that there exists endogenous mechanisms which produce a substance or an effect similar to that of ethanol, such that release probability is increased without the necessity of the presence of Ca^  (Quastel et al, 1971).  effect  nerve  may  indicate  a  terminal  in  which  Such an normal  metabolism is compromised.  V. Conclusion - Fitting data to a model Data from a system such as the neuromuscular junction whose mechanisms are unknown can only be analysed in a bootstrapping fashion.  That is, a preliminary model is  proposed and the experimental data is analysed according to that model.  From this analysis, the model is refined to  explain any lack of fit.  It is generally considered valid  for this process to be repeated either until the model can correctly predict the data or until the model is discarded in favour of simpler models which are equally predictive. For the neuromuscular junction, it is apparent from the literature  and  from  the  present  data  that  the  present  176 combined model has the greatest predictiveness for the time course  of  stimulation  phasic and  release  contents  of  under the  many  conditions  superfusate,  and  of for  stimulation induced enhancement of phasic and non-phasic release.  177 References Atwood, H. L., Swenarchuk, L. E. & Gruenwald, C. R. (1975) Long-term synaptic facilitation during sodium accumulation in nerve terminal. Brain Research 10, 198204 Atwood, H. L. & Wojtowicz, J. M. (1986) Review  of  Neurobiology  International  28, 275-362  Augustine, G. J. & Eckert, R. (1984). Divalent cations differentially support transmitter relase at the squid giant synapse. Journal of Physiology 346, 257-271. Augustine, G. J., Charlton, M. P., & Smith, S. J. (1987) Calcium action in synaptic transmitter release. Annual Reviews  of  Neuroscience  10, 633-693  Bain, A. I. & Quastel, D. M. J. (1988). Time course of transmitter release and calcium channels at the mouse motor nerve terminal. Society for Neuroscience Abstracts, 1988. Bain, A. I. & Quastel, D. M. J. (1992a) Quantal transmitter release mediated by strontium at the mouse motor nerve terminal. Journal  of  Physiology  450, 63-87.  Bain, A. I. & Quastel, D. M. J. (1992b) Multiplicative and additive Ca2+-dependent components of facilitation at mouse end-plates. Journal of Physiology 455, 383-405. Baldo, G. J., Cohen, I. S., & Van der Kloot, W. (1986). Estimating the time course of evoked quantal release at the frog neuromuscular junction using end-plate current latencies. Journal of Physiology 374, 503-513. Balnave, R. J. and Gage, P. W. (1970) Temperature sensitivity of the time course of facilitation of transmitter release. Brain Research 21, 297-300 Barrett, E. E. & Stevens, C. F. (1972a). Quantal independence and uniformity of presynaptic release kinetics at the frog neuromuscular junction. Journal Physiology 227, 665-689.  of  Barrett, E. E. & Stevens, C. F. (1972b). The kinetics of transmitter release at the frog neuromuscular junction.Journal of Physiology 227, 691-708. Barzilai, A., Spanier, R. & Rahamimoff, H. (1984) Isolation, purification and reconstitution of the Na + gradient dependent Ca 2 + transporter from brain synaptic plasma  178 membranes. Proceedings of the National Sciences, U. S. A. 81, 6521-6525.  Academy  of  Bernstein, B. W. and Bamburg, J. R. (1989) Cycling of actin assembly in synaptosomes and neurotransmitter release. Neuron 3, 257-265. Bittner, G. D. and Schatz, R. A. (1981) An examination of the residual calcium theory for-facilitation of transmitter release. Brain Research 210, 431-436. Blaustein (1988)  Trends in Neuroscience 11, 438.  Blioch, Z. L., Glagoleva, I. M., Liberman, E. A. & Nenashev, V. A. (1968). A study of the mechanism of quantal transmitter release at a chemical synapse. Journal of Physiology  199,11-36.  Bourret, C. & Mallart, A. (1989) Depression of calcium current at mouse motor nerve endings by polycationic antibiotics. Brain Research 478, 403-406 Braun, M., Schmidt, R. F. & Zimmerman, M. (1966a). Facilitation at the frog neuromuscular junction during and after repetitive stimulation. Pflugers Arch. ges. Physiol. 287, 41-55. Braun, M. & Schmidt, R. F. (1966b). Potential changes recorded from the frog motor nerve terminal during its activation.  Pflugers  Arch.  ges.  Physiol.  287, 56-80.  Bulbring, E. (1946). Observations on the isolated phrenic nerve diaphragm preparation of the rat. British Journal  of  Pharmacology  1, 38-61.  Charlton, M. P. & Atwood, H. L. (1977) Modulation of transmitter release by intracellular sodium in the squid giant synapse. Brain Research 134, 367-371. Charlton, M. P. & Augustine, G. J. (1990). Classification of presynaptic calcium channels at the squid giant synapse: neither T-, L- nor N-type. Brain Research 525, 133-139. Charlton, M. P. and Bittner, G. D. (1978a). Transmitter release facilitation at squid synapses. Journal of General  Physiology  72, 4??-48?  Charlton, M. P. and Bittner, G. D. (1978b). Presynaptic potentials and facilitation of transmitter release in the squid giant synapse.  72, 487-511  Journal  of  General  Physiology  179 Charlton, M. P., Smith, S. J. & Zucker, R. S. (1982) Role of presynaptic calcium ions and channels in presynaptic facilitation and depression at the squid giant synapse. Journal  of  Physiology  323, 173-193  Cohen, I. S., Van der Kloot, W. (1986) Facilitation and delayed release at single frog neuromuscular junctions. Journal  of  Neuroscience.  6, 2366-2370.  Cooke, J. D. & Quastel, D. M. J. (197 3a). Transmitter release by mammalian motor nerve terminals in response to focal polarization. Journal of Physiology 228, 377405 Cooke, J. D. & Quastel, D. M. J. (1973b)# Cumulative and persistent effects of nerve terminal depolarization on transmitter release. Journal of Physiology 228, 407-434 Cooke, J. D., Okamoto, K. & Quastel, D. M. J. (1973). The role of calcium in depolarization-secretion coupling at the motor nerve terminal. Journal of Physiology 228, 459-497 Crompton et al (1988) Diversity in the lipocortin/calpactin family. Cell 55, 1-3 Curtis, M. J., Quastel, D. M. J. & Saint, D. A. (1986) Lanthanum as a surrogate for calcium in transmitter release at mouse motor nerve terminals. Journal of Physiology 373, 243-2 60 Dale, H. H. & Feldberg, W. & Vogt, M. (1936). Release of acetylcholine at voluntary motor nerve endings. Journal  of  Physiology  86, 353-380.  Datyner, N. B. & Gage, P. W. (1980). Phasic secretion of acetylcholine at a mammalian neuromuscular junction. Journal  of  Physiology  303, 299-314.  Del Castillo, J. & Katz, B. (1954). Quantal components of the end-plate potential. Journal of Physiology 124, 560-573. Del Castillo, J. & Katz, B. (1956) Biophysical aspects of neuromuscular transmission. Progress in Biophysics and Biophysical  Chemistry  6, 121-170  Delaney. K. R., Zucker, R. S. and Tank, D. W. (1989). Calcium in motor nerve terminals associated with posttetanic potentiation.  The Journal  of  Neuroscience  9, 3558-3567. Dodge, F. A. & Rahamimoff, R. (1967). Cooperative action of calcium ions in transmitter release at the  180 neuromuscular junction. Journal 432  of Physiology  193, 419-  Douglas, W. W. (1968) Stimulus-secretion coupling: the concept and clues from chromaffin and other cells. British Journal of Pharmacology 34, 451-474 Dudel, J. (1989) Calcium and depolarization dependence of twin-pulse facilitation of synaptic release at nerve terminals of crayfish and frog muscle. Pflugers Archiv - European Journal of Physiology 415, 304-309. Dudel, J. (1990) Inhibition of Ca 2 + inflow at nerve terminals of frog muscle blocks facilitation while phasic transmitter release is still considerable. Pflugers Archiv - European Journal of Physiology 415, 566-574. Elmqvist, D. and Quastel, D. M. J. (19 65a) Presynaptic action of hemicholinium at the neuromuscular junction. Journal of Physiology 177, 463-482. Elmqvist, D. and Quastel, D. M. J. (1965b) A quantitative study of end-plate potentials in isolated human muscle. Journal of Physiology 178, 505-529. Erulkar, S. B., Meiri, H., and Rahamimoff, R. (1982) The ionic basis of high-frequency synaptic activation. In Physiology and Pharmacology of Epileptogenic Phenomena, M. R. Klee, ed. Raven Press, New York, 1982, pp 197-205. Fabiato, A. and Fabiato, F. (1979) Calculator programs for calculating the composition of the solutions containing multiple metals and ligands used for experiments in skinned muscle cells. Journal of Physiology (Paris) 75, 463-505 Feng, T.P. (1940) Studies on the neuromuscular junction XVIII. The local potentials around N-M junctions induced by single and multiple volleys. Chin. J. Physiol 15,367-404. Feng, T.P. (1941) Studies on the neuromuscular junction XXVI. The changes of the endplate potential during and after prolonged stimulation. Chin. J. Physiol 16,341372. Gage, P. W. & Hubbard, J. I. (1966) The origin of the posttetanic hyperpolarization of mammalian motor nerve terminals. Journal of Physiology 184, 335-352.  181 Guan, Y. -Y., Quastel, D. M. J., & Saint, D. A. (1988). Single Ca z + entry and transmitter release systems at the neuromuscular synapse. Synapse 2, 558-564 Heuser, J. E, Reese, T. S., Dennis, M. J., Jan, Y., Jan, L., & Evans, L. (1979). Synaptic vesicle exocytosis captured by quick freezing and correlated with quantal transmitter release. Journal of Cell Biology 81, 275300 Hille (1984)  Ion  Channels  of  Exitable  Membranes,  Sinauer  and Assoc, Sunderland, Mass., USA. Hochner, B., Parnas, H. and Parnas, I. (1991) Effects of intra-axonal injection of Ca 2 buffers on evoked release and on facilitation in the crayfish neuromuscular junction. Neuroscience Letters 125, 215218. Hochner, B., Parnas, H., & Parnas, I. (1989). Membrane depolarization evokes neurotransmitter release in the absence of calcium entry, Nature 342, 433-435. Hubbard, J. I., Llinas, R. and Quastel, D. M. J. (1969)  Electrophysiological  Analysis  of Synaptic  Transmission,  London: Edward Arnold, p.129. Hubbard, J. I. (1963). Repetitive stimulation at the mammalian neuromuscular junction and the mobilisation of transmitter. Journal of Physiology 169, 641-662 Hubbard, J. I. and Schmidt, R. F. (1963). An electrophysiological investigation of mammalian motor nerve terminals. Journal of Physiology 166, 145-165. Hubbard, J. I., Jones, S. F. & Landau, E. M. (1968a). On the mechanism by which calcium and magnesium affect the spontaneous release of transmitter from mammalian motor nerve terminals. Journal of Physiology 194, 355-380. Hubbard, J. I., Jones, S. F. & Landau, E. M. (1968b). On the mechanism by which calcium and magnesium affect the release of transmitter by nerve impulses. Journal of Physiology 196, 75-86. Katz, B. & Miledi, R. (1965a). The effect of calcium on acetylcholine release from motor nerve terminals. Proc.  R.  Soc.  Lond.  B. Biol.  Sci.  161, 496-503.  Katz, B. & Miledi, R. (1965b). The effect of temperature on the synaptic delay at the neuromuscular junction. Journal  of  Physiology  181, 656-670.  182 Katz, B. & Miledi, R. (1965c). The measurement of synaptic delay, and the time course of acetylcholine release at the neuromuscular junction. Proc. R. Soc. Lond. B. Biol. Sci. 161, 483-495. Katz, B. & Miledi, R. (1967). The release of acetylcholine from nerve endings by graded electric pulses. Proc. R. Soc.  Lond.  B. Biol.  Sci.  167, 23-28.  Katz, B. & Miledi, R. (1968). The role of calcium in neuromuscular facilitation. Journal of Physiology 481-492. Katz, B. (1969). The Release  Substances.  of  Neural  195,  Transmitter  Springfield: Charles C. Thomas  Kijima, H. & Tanabe, N. (1988). Calcium independent increase of transmitter release at frog end-plate by trinitrobenzene sulphonic acid. Journal of Physiology 403, 135-149. Kwan, C. Y. & Putney, J. W. (1990). Uptake and sequestration of divalent cations in resting and methacholinestimulated mouse lacrimal acinar cells. Dissociation by Sr 2 + and Ba of agonist-stimulated divalent cation entry from the refilling of the agonist-sensitive intracellular pool.  Journal  of  Biological  Chemistry  265, 678-684 Landau, E.M., Smolinsky, A., Lass, Y. (1973) Post-tetanic potentiation and facilitation do not share a common calcium-dependent mechanism. Nature New Biology 244, 155-157 Liley and North (1953). An electrical investigation of effects of repetitive stimulation on mammalian neuromuscular junction. Journal of Neurophysiology 509-527.  16,  Linder, T. M. (1973) Calcium and facilitation at two classes of crustacean neuromuscular synapses. Journal of  General  Physiology  61, 56-73.  Lux, H. D. & Bruns, D. (1991) Photolytically induced rapid increase of intracellular Ca 2 . Studies on Ca channels function and synaptic transmission. Abstracts, Third IBRO world congress of Neuroscience, p 428 Magleby, K. L. & Stevens, C. F. (1972) A quantitative description of end-plate currents. Journal of Physiology 223, 173-197 Magleby, K. L. (1973) The effect of tetanic and post-tetanic potentiation on facilitation of transmitter release at  183 the frog neuromuscular junction. 234, 353-371.  Journal  of  Physiology  Mallart A., Molgo J., Angaut-Petit D. & Thesleff S. (1989) Is the internal calcium regulation altered in type A botulinum toxin-poisoned motor endings?. Brain Research 479, 167-171 Mallart, A. & Martin, A. R. (1967) An analysis of transmitter release at the neuromuscular junction of the frog. Journal of Physiology 193, 679-694. Mallart, A. & Martin, A. R. (1968). The relation between quantum content and facilitation at the neuromuscular junction of the frog. Journal of Physiology 196, 593604. McLarnon, J. G., Saint, D. A. & Quastel, D. M. J. (1987) The actions of dimethylsulfoxide on neuromuscular transmission. Molecular Pharmacology 30, 631-638 Martin, A. R. (1955) A further study of the statistical composition of the end-plate potential. Journal of Physiology 130, 114-122. Mellow AM (1979) Equivalence of Ca 2 + and Sr 2 + in transmitter + release from K -depolarised nerve terminals. Nature 282, 84-85. Miledi, R. & Thies, R. (1971) Tetanic and post-tetanic rise in frequency of miniature end-plate potentials in lowcalcium solutions. Journal of Physiology 212, 245-257. Miledi, R. (1966). Strontium as a substitute for calcium in the process of transmitter release at the neuromuscular junction. Nature 212, 1233-1234 Misler, S. and Hurlbut, W. P. (1983) Post-tetanic potentiation of acetylcholine release at the frog neuromuscular junction develops after stimulation in Ca 2 -free solutions. Proceedings of the Academy of Science (USA) 80, 315-319.  National  Misler, S. & Falke, L. C. (1987) Dependence on multivalent cations of quantal release of transmitter induced by black widow spider venom. American Journal of Physiology 253, C469-476. Misler, S., Falke, L. & Martin, S. (1987). Cation dependence of posttetanic potentiation of neuromuscular transmission. American Journal of Physiology 252, c5562  184 Molgo, J., Angaut-Petit, D. & Thesleff, S. (1987) In botulinum type A-poisoned frog motor endings ouabain induces phasic transmitter release through Na+-Ca+ exchange. Brain Research 410, 385-389. Mulkey, R. M. & Zucker, R. S. (1991). Action potentials must admit calicum to evoke transmitter release. Nature 350, 153-155. Neher, E. (1988) The influence of intracellular calcium concentration on degranulation of dialysed mast cells from rat peritoneum. Journal of Physiology 395, 193214 Nowycky, M. C , Fox, A. P. and Tsien, R. W. (1985). Three types of neuronal calcium channel with different calcium agonist sensitivity. Nature 316, 440-443. Nussinovitch I. & Rahamimoff R. (1988) Ionic basis of tetanic and post-tetanic potentiation at a mammalian neuromuscular junction. Journal of Physiology 396, 435-455. Parnas H., Dudel, J., & Parnas, I. (1986). Neurotransmitter release and its facilitation in crayfish. VII. Another voltage dependent process beside Ca entry controls the time course of phasic release. Pflugers Archiv European Journal of Physiology 406, 121-130 Parnas I. & Parnas H. (1986) A new method for determining co-operativity in neurotransmitter release. Journal of Theoretical Biology 119, 481-499. Parnas I. & Parnas H. (1988). The "Ca-voltage" hypothesis for neurotransmitter release. Biological Chemistry 29, 85-93. Parnas, I., Parnas, H. and Dudel, J. (1986). Neurotransmitter release and its facilitation in crayfish. VIII. Modulation of release by hyperpolarizing pulses. Pflugers Archiv - European Journal of Physiology 406, 131-137 Philipson, K.D. (1985) Annual  Review  of Physiology  47, 561  Plattner, H. (1989) Regulation of membrane fusion during exocytosis. International Review of Cytology 119, 197286. Quastel, D. M. J. (1974) Excitation-secretion coupling at the mammalian neuromuscular junction. In: Synaptic Transmission and Neuronal Interaction, Raven Press, New York.  185 Quastel, D. M. J. Guan, Y. -Y. & Saint, D. A. (1992) The relation between transmitter release and Ca 2 + entry at the mouse motor nerve terminal: role of stochastic factors causing heterogeneity. Neuroscience 51, 657671. Quastel, D. M. J., Hackett, J. T. & Cooke, J. D. (1971) Calcium: is it required for transmitter secretion? Science 172, 1034-1036 Quastel, D. M. J. & Saint, D. A. (1986) Modification of motor nerve terminal excitability by alkanols and volatile anaesthetics. British Journal of Pharmacology 88, 747-756 Quastel, D. M. J. & Saint, D. A. (1988). Transmitter release at mouse motor nerve terminals mediated by temporary accumulation of intracellular barium. Journal of Physiology 406, 55-73 Quastel, D. M. J., Bain, A. I., Guan, Y.-Y. & Saint, D. A. (1989). Ionic cooperativity in transmitter release. In: Neuromuscular Junction, ed. Sellin, L. C , Libelius, R. & Thesleff, S. Elsevier. Quastel, D. M. J., Hackett, J. T. & Cooke, J. D. (1971). Calcium: is it required for transmitter secretion? Science 172, 1034-1036. Robitaille, R., Adler, E. M. & Charlton, M. P. (1990) Strategic location of calcium channels at transmitter release sites of frog neuromuscular synapses. Neuron 5, 773-779 Robitaille R. & Charlton, M. P. (1990) Frequency facilitation is not caused by residual ionized calcium at the frog neuromuscular junction. Society for Neuroscience Abstracts, 16, 503 Robitaille, R. and Tremblay, J. P. (1991) Non-uniform responses to calcium along the frog neuromuscular junction effects on the probability of spontaneous and evoked transmitter release. Neuroscience 40, 571-586 Rotshenker, S., Erulkar, S. D. & Rahamimoff, R. (1976). Reduction in frequency of miniature end-plate potentials by nerve stimulation in low calcium solutions. Brain Research 101, 362-365 Saint, D. A., Quastel, D. M. J. & Guan, Y. -Y. (1987) Multiple potassium conductances at the mammalian motor nerve terminal. Pflugers Archiv - European Journal of Physiology 410, 408-412.  186 Silinsky, E. M. (1978) On the role of barium in supporting the asynchronous release of acetylcholine quanta by motor motor nerve impulses. Journal of Physiology 274, 157-171. Silinsky, E. M. (1985) The biophysical pharmacology of calcium-dependent acetylcholine secretion. Pharmacological Reviews 37, 81-132. Silinsky, E. M. (1981) On the calcium receptor that mediates depolarization-secretion coupling at cholinergic motor nerve terminals. British Journal of Pharmacology 73, 413-429 Tanabe, N. & Kijima, H. (1989) Both augmentation and potentiation occur independently of internal Ca 2 + at the frog neuromuscular junction. Neuroscience Letters 99, 147-152. Thesleff, S. (1980) Aminopyridines and synaptic transmission. Neuroscience 5, 1413 Thomas, L., Hartung, K., Langosch, D., Rehm, H., Bamberg, E. Franke, W. W., Betz, H. (1985) Identification of synaptophysin as a hexameric channel protein of the synaptic vesicle membrane. Science 242, 1050-1053 Tsien, R. Y. (1981). A non-disruptive technique for loading calcium buffers and indicators into cells. Nature 290, 527-528 Uchitel, O. D., Protti, D. A., Sanchez, V., Cherksey, B. D., Sugimori, M. and Llinas, R. (1992) P-type voltagedependent calcium channel mediates presynaptic calcium influx and transmitter release in mammalian synapses. Proceedings of the National Academy of Science USA 89, 3330-3333 Uchiyama, T., Molgo, J., Lemeignan, M. (1981) Bekanamycin blockade of transmitter release at the motor nerve terminal. European Journal of Pharmacology 72, 271-280 Van der Kloot, W. (1988a). Estimating the timing of quantal releases during end-plate currents at the frog neuromuscular junction. Journal of Physiology 402, 595-603. Van der Kloot W. (1988b). The kinetics of quantal releases during end-plate currents at the frog neuromuscular junction. Journal of Physiology 402, 605-626. Vere-Jones, D. (1966) Simple stochastic models for the release of quanta of transmitter from a nerve terminal. Australian Journal of Statistics 8, 53-63  187 Wang, Y. X. and Quastel, D. M. J. (1991) Actions of lead on transmitter release at mouse motor nerve terminals. Pflugers Archiv - European Journal of Physiology 419, 274-280. Younkin, S. G. (1974). An analysis of the role of calcium in facilitation at the from neuromuscular junction. Journal of Physiology 237, 1-14 Zengel, J. E. & Magleby, K. L. (1980). Differential effects of Ba 2 + , Sr 2+ , and Ca24" on stimulation-induced changes in transmitter release at the frog neuromuscular junction. Journal of General Physiology 76, 175-211 Zengel, J. E. & Magleby, K. L. (1981). Changes in miniature endplate potential frequency during repetitive nerve stimulation in the presence of Ca 2 , Ba 2 + , and Sr at the frog neuromuscular junction. Journal of General Physiology  77, 503-529  Zengel, J. E. & Magleby, K. L. (1982a). Augmentation and facilitation of transmitter release: a quantitative description at the frog neuromuscular junction. Journal of General Physiology 80, 583-611 Zengel, J. E. & Magleby, K. L. (1982b) A quantitative description of stimulation-induced changes in transmitter release at the frog neuromuscular junction. Journal of General Physiology 80, 613-638 Zengel, J. E., Lee, D. T., Van Veelen, M. L. and Moser, D. R. (1988) Effects of divalent cations on stimulationinduced changes in transmitter release at the frog neuromuscular junction (abstract). Biophysical Journal 53, 363. Zucker, R. S., Delaney, K. R., Mulkey, R. & Tank, D. W. (1991) Presynaptic calcium in transmitter release and posttetanic potentiation. Annals of the New York Academy of Sciences 635, 191-207 Zucker, R. S. & Haydon, P. G. (1988). Membrane potential has no direct role in evoking neurotransmitter release. Nature 335, 360-362. Zucker, R. S. & Fogelson, A. L. (1986). Relationship between transmitter release and presynaptic calcium influx when calcium enters through discrete channels. Proceedings of the National Academy of Science USA 83, 3032-3036. Zucker, R. S. (1987). The calcium hypothesis and modulation of transmitter release by hyperpolarizing pulses. Biophysical Journal 52, 347-350.  


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items