CHANGES IN THE READILY RELEASABLE POOL OF TRANSMITTER AND IN

EFFICACY OF RELEASE INDUCED BY HIGH-FREQUENCY FIRING AT APLYSIA

SENSORIMOTOR SYNAPSES IN CULTURE

Yali Zhao and Marc Klein

Department of Physiological Science, UCLA, 621 Charles E. Young Drive South, Los Angeles

CA 90095-1606

Abbreviated Title: Regulation of the RRP by high-frequency stimulation

Corresponding Author: Marc Klein, Department of Physiological Science, UCLA, 621

Charles E. Young Drive South, Los Angeles CA 90095-1606.

Telephone: (310) 825-8562; Fax: (310) 206-9184; Email: kleinm@ucla.edu

Copyright (c) 2003 by the American Physiological Society.

Articles in PresS. J Neurophysiol (November 26, 2003). 10.1152/jn.01019.2003

1

ABSTRACT

Synaptic transmission at the sensory neuron-motor neuron synapses of Aplysia, like

transmission at many synapses of both vertebrates and invertebrates, is increased after a

short burst of high-frequency stimulation (HFS), a phenomenon known as post-tetanic

potentiation (PTP). PTP is generally attributable to an increase in transmitter release

from presynaptic neurons. We investigated whether changes in the readily releasable

pool of transmitter (RRP) contribute to the potentiation that follows HFS. We compared

the changes in EPSPs evoked with action potentials to changes in the RRP as estimated

from the asynchronous transmitter release elicited by a hypertonic solution. The changes

in the EPSP were correlated with changes in the RRP, but the changes matched

quantitatively only at connections whose initial synaptic strength was greater than the

median for all experiments. At weaker connections, the increase in the RRP was

insufficient to account for PTP. Weaker connections initially released a smaller fraction

of the RRP with each EPSP than stronger ones, and this fraction increased at weaker

connections after HFS. Moreover, the initial transmitter release in response to the

hypertonic solution was accelerated after HFS, indicating that the increase in the efficacy

of release was not restricted to excitation-secretion coupling. Modulation of the RRP and

of the efficacy of release thus both contribute to the enhancement of transmitter release

by HFS.

Keywords: synaptic plasticity; synaptic vesicles; excitation-secretion coupling;

posttetanic potentiation.

2

INTRODUCTION

Synaptic transmission can be dramatically altered by short trains of high-frequency

stimulation (HFS), with depression immediately following a train (Zucker and Regehr,

2002; Zucker, 1999) being replaced at some synapses by long-term facilitation lasting

hours or days (Bliss and Collingridge, 1993). Many synapses in both vertebrates and

invertebrates exhibit an increase in transmission lasting for seconds to minutes following

HFS. This increase, known as post-tetanic potentiation (PTP), results from an increase in

transmitter release from the presynaptic terminals, and depends on presynaptic calcium

entry (Zucker, 1999; Zucker and Regehr, 2002). PTP occurs at Aplysia sensorimotor

synapses as well (Walters and Byrne, 1984; Bao et al., 1997). The induction of PTP at

these synapses depends on postsynaptic depolarization and calcium entry also (Lin and

Glanzman, 1994; Bao et al., 1997; Schaffhausen et al., 2001), but the postsynaptic

contribution is limited to connections of initially low synaptic strength (Schaffhausen et

al., 2001). In this report, we examine the underlying mechanisms of PTP at the

sensorimotor synapses of Aplysia.

Synaptic transmission is determined by the amount of neurotransmitter released from

the presynaptic neuron and by the transduction of the chemical signal into an electrical

response by the target cell. Transmitter release, in turn, depends on the size of the readily

releasable pool (RRP) of transmitter (Rosenmund and Stevens, 1996; Gillis et al., 1996),

thought to represent the release-ready synaptic vesicles docked at the active zone

(Schikorski and Stevens, 2001), and on the efficacy of the release process.

The RRP and the efficacy of release can be modulated by electrical activity and by

signaling molecules recruited by neurotransmitters. For example, the RRP is decreased

3

in long-term depression (Goda and Stevens, 1998), and is increased through the action of

protein kinase C (Stevens and Sullivan, 1998; Waters and Smith, 2000; Berglund et al.,

2002). Efficacy of release is also modulated in short-term synaptic plasticity at the

sensorimotor synapses of Aplysia (Zhao and Klein, 2002).

In a study on Aplysia sensorimotor synapses (Zhao and Klein, 2002), we examined

the underlying basis of homosynaptic depression, caused by low-frequency stimulation,

and of the facilitation of depressed synapses by serotonin, an endogenous facilitatory

transmitter. We found that both the size of the RRP and the efficacy of release were

modulated, and that the changes in the efficacy of release were limited to excitation-

secretion coupling. We now present a similar analysis of the potentiation of synaptic

transmission that follows HFS.

Changes in the EPSP at initially strong connections are matched quantitatively by

changes in the RRP. At initially weak connections, there is, in additon, an increase in the

efficacy of release. This increase can be detected even when release is elicited without

action potentials, implicating a process that is independent of excitation-secretion

coupling. Taken together with previous findings, our results suggest that an increase in

the RRP contributes to PTP at all connections, and in addition, a retrograde signal from

the motor neuron may contribute to the increase in the efficacy of transmitter release at

weaker connections.

4

MATERIALS AND METHODS

Preparation of cultures

Adult Aplysia californica (75−150 g; Marine Specimens Unlimited, Pacific

Palisades, CA and Alacrity Marine Biological Services, Redondo Beach, CA) were

anesthetized by injection of 50−100 ml of 385 mM (isotonic) MgCl2. Tail sensory

neurons (Walters et al., 1983) and siphon motor neurons (LFS neurons; (Frost and

Kandel, 1995) were isolated and maintained as previously described (Schacher and

Proshansky, 1983; Klein, 1994; Coulson and Klein, 1997; Zhao and Klein, 2002).

Both conventional sensory neuron-motor neuron co-cultures (Schacher and Proshansky,

1983) and soma-to-soma co-cultures (Klein, 1994; Coulson and Klein, 1997; Zhao and

Klein, 2002) were used for the experiments, with no differences detected between them

in any of the parameters we examined.

Electrophysiology

An Axoclamp 2A or 2B amplifier (Axon Instruments) and borosilicate glass

micropipettes (tip resistance 10−20 MΩ) filled with 2M potassium acetate (pH 7.5) were

used for intracellular recordings from the LFS neuron. Recordings were carried out in

ASW (composition, in mM: 460 NaCl, 10 KCl, 11 CaCl2, 55 MgCl2, and 10 HEPES (N-

2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid), pH 7.5). Neuron type was

confirmed by the characteristic notch on the rising phase of the voltage in LFS cells after

release from hyperpolarizing current (Eliot et al., 1994). Throughout the experiments,

the LFS motor neuron was hyperpolarized to -80 mV in current clamp mode to minimize

spiking. To elicit an EPSP in the motor neuron, an extracellular patch pipette filled with

5

ASW was put into contact with the sensory neuron membrane, and a single action

potential was evoked in the sensory neuron with a 1-5 ms pulse of current.

Voltage recording was done rather than voltage clamp because of the large

difference in the amplitudes of EPSPs and of miniature EPSPs: voltage clamp that would

be appropriate for the large synaptic currents (>10 nA) would be too noisy to resolve the

miniatures (~20 pA), while measurement of the large currents would be inaccurate when

clamping to examine the miniature synaptic currents. We did not switch between

voltage recording for the large responses and voltage clamp for the miniatures, or vice

versa, because we wanted to use the amplitudes of the miniatures to estimate the quantal

content (or “mini content”; see below) of the electrically evoked responses, which

required using the same recording method for both.

To examine PTP, the sensory neuron was fired at 20 Hz for one second, and a single

EPSP was elicited 30 sec after the train. PTP is expressed as the ratio of the EPSP

following the train to the first EPSP of the train (Figure 1).

We estimated the size of the RRP from the number of miniature synaptic potentials

(minis) triggered by application of a hypertonic solution. The hypertonic solution used to

elicit this asynchronous release consisted of ASW with sucrose added to the desired final

concentration. Either 0.5M or 1M sucrose was added to the ASW. Hypertonic solution

was applied either with a 5-10 second puff from a microperfusion system (tip diameter

approximately 50-100 µm; Warner Instruments) or by direct addition of 50 µL of the

solution from a hand-held Pipetman onto the neurons.

The number of minis in the RRP was estimated on the basis of the dose-response

relationship shown in Zhao and Klein (2002). Specifically, since 0.5M sucrose releases

6

approximately 50% of the RRP and 1M releases about 80%, the RRP was estimated to

contain twice the number of minis elicited with the 0.5M solution and 1.25 times the

number elicited with the 1M solution.

Miniature synaptic potentials elicited by the hypertonic sucrose were detected and

analyzed as described in Zhao and Klein (2002). Each mini was counted as a

single event, in spite of the likelihood that some of the largest minis probably

reflected multiquantal release events. Our conclusions are not affected by

counting all minis as single events because after high-frequency stimulation the

average mini amplitude either did not change (weak connections) or decreased

slightly (strong connections; see Figure 5), with the result that the estimated

relation between the RRP and the EPSP was affected minimally, if at all.

Because the mini frequency declined to very low levels within one minute

after sucrose application (Zhao and Klein, 2002), all minis detected during the first

55 seconds after application of the hypertonic solution were counted. All applications of

hypertonic solution within an experiment were done in the same way, i.e., either with a

puff or with an aliquot; the results were not affected by the manner of application of the

hypertonic solution. The preparation was washed with ASW for 10 minutes between

tests.

The quantal content—or the “mini content”, because the average mini amplitude is

not equal to the quantal amplitude—of EPSPs was estimated by dividing the EPSP

amplitude by the average mini amplitude for EPSPs that were 10 mV or less in

amplitude. For larger EPSPs, the peak EPSP amplitudes for the whole experiment were

first plotted against the maximal slope of the EPSPs (corresponding to the peak synaptic

7

current). We measured the maximal slope over 1 ms on the rising limb of the EPSP;

where the EPSP triggered a spike or a local active response, the measurement was taken

before the inflection point marking the active response. A partial correction for nonlinear

summation was then done as follows: The linear relation between amplitude and

maximal slope for the smaller EPSPs was extrapolated to the larger amplitude EPSPs.

The maximal slope of the larger EPSPs was then multiplied by a factor that made it fall

on the extrapolated line, and the corresponding amplitude was divided by the average

mini amplitude to estimate the mini content. The idea behind this correction is that the

maximal slope of the EPSP is less affected by nonlinear summation than the peak

because the voltage at which the maximal slope occurs is much less depolarized than the

EPSP peak. Because the maximal slope is still affected by nonlinear summation, albeit to

a lesser extent, the mini content of larger EPSPs will still be somewhat underestimated,

but this underestimate does not affect any of our conclusions.

The protocol used to examine EPSPs and asynchronous release consisted of a single

EPSP elicited by electrical stimulation of the sensory neuron followed within 1 minute by

application of the hypertonic solution (see Figure 2). In some of the experiments,

homosynaptic depression of the EPSP was first induced by stimulating the sensory

neuron 5 times at 30-sec intervals before HFS.

Experiments were recorded on a Power MacIntosh computer running Axograph 4

(Axon Instruments), and EPSPs and minis were analyzed with that program.

Data files were transferred to Microsoft Excel (version 7) and GraphPad Prism

(version 2) for statistical analysis and plotting. All data are expressed as mean ± SEM.

Two-tailed t-tests (paired or unpaired, as appropriate) were used for statistical

8

comparisons, and differences were considered significant if the p value was less than

0.05.

9

RESULTS

Posttetanic potentiation is greater at initially weak connections than at initially

strong connections

High-frequency stimulation (HFS) of the sensory neuron at sensorimotor synapses of

Aplysia is followed by an increase in synaptic transmission (Walters and Byrne, 1984),

similar to the PTP observed at many other synapses (Zucker, 1999; Zucker and Regehr,

2002). The magnitude of PTP at Aplysia synapses in intact ganglia varies inversely with

the initial synaptic strength of the connection, and the mechanisms of PTP appear to

differ at connections of different synaptic strength (Schaffhausen et al., 2001). As the

first step in our analysis, we examined the relationship between the magnitude of PTP

and the initial synaptic strength at sensorimotor connections in cell culture.

A one-second train of presynaptic action potentials at 20 Hz (Figure 1A) was

followed by PTP that varied inversely with the initial synaptic strength of the connection

(Figure 1B). When the connections were divided into strong and weak at the

approximate median initial synaptic strength (maximal slope of 14 mV/ms; see Methods)

the EPSP was potentiated about 4-fold at the initially weak connections (p=0.0003

compared to the first of the train) and about 2-fold at the initially strong connections

(p=0.012; Figure 1C), almost exactly matching the PTP measured in intact ganglia

(Schaffhausen et al., 2001).

Before proceeding with the analysis, however, it was first necessary to deal with the

possibility that the inverse relationship of Figure 1B (and Figure 3A1), as well as that

reported in Schaffhausen et al. (2001), might be due to a statistical artifact (see for

example Kim and Alger, 2001). If fluctuating synaptic responses are sorted by the

10

amplitude of the initial response, and the amplitude of a second response is compared to

the amplitude of the first response, there will often be a negative correlation between the

amplitude of the first response and the ratio of the second to the first response. This

negative correlation can be a statistical artifact that has no physiological significance.

An artifactual negative correlation can arise from the fact that a response that is near

the upper end of the amplitude distribution is more likely to be followed by a smaller

response, while a small response, which is near the bottom of the distribution, is more

likely to be followed by a larger response. If the response can vary between a very small

and an arbitrarily large value, the ratio of the second to the first response can vary up to

infinity if the first response is small, while the ratio can vary only up to 1 if the first

response is at the upper end of the amplitude distribution. Plotting the ratio of the second

to the first response against the amplitude of the first response will result in a negative

correlation similar to that shown in Figure 1B.

This artifact can occur if the synaptic responses belong to a single population with a

given mean and variance. In practice, this situation arises if the response of a single

synaptic connection is sampled repeatedly, or if the comparison is done on a population

of synaptic connections that vary around more or less the same mean and have similar

variances. If different connections have different means and variances, or, in the most

extreme case, if the different connections do not overlap at all in their amplitude

distributions, there should be no correlation between the magnitude of the first response

and the ratio of the second response to the first. Thus, for example, a large response from

a connection whose average response is large may or may not be near the upper end of

the distribution for that connection, and it is therefore equally likely that a second

11

response will be greater than the first as that it will be smaller. A similar argument holds

for a connection whose average response is small: a small response at such a connection

is equally likely to be followed by a smaller response as by a greater response. We used

this test to examine the possibility that the inverse relationship between the initial

synaptic response and the magnitude of PTP might be a statistical artifact.

If the inverse relationship of Figure 1B were the result of the artifact described

above, then sampling two successive synaptic responses without intervening HFS should

also result in a negative relationship between the magnitude of the first response and the

ratio of the second response to the first response. We therefore examined two successive

responses separated by a 1-minute interval in the absence of the HFS. In 17 experiments

looking at initial synaptic responses in the same range as those of Figure 1B there was no

correlation between the magnitude of the initial response and the ratio of the second

response to the first (Figure 1C; r2 = 0.049, p = 0.394, n = 17). This finding indicates that

the negative correlation between the extent of PTP and the magnitude of the initial

synaptic response cannot be attributed to a statistical artifact, but rather that it results

from a physiologically meaningful difference between potentiation at initially weak

connections and that at initially strong connections.

High-frequency stimulation increases the RRP at both initially strong and

initially weak synapses, but increases the efficacy of release only at initially weak

synapses

We next looked for changes in the size of the readily releasable pool of transmitter

(RRP) and in the efficacy of release that might underlie this potentiation. We estimated

the size of the RRP from the asynchronous release caused by application of a hypertonic

12

sucrose solution (see Methods), and compared EPSPs and responses to hypertonic

solution after HFS to the test values before HFS (Figure 2). A test EPSP was elicited by

stimulating the sensory neuron with one action potential, and the hypertonic bathing

solution was then applied. After 10 minutes of wash and rest, a one-second train of

action potentials at 20 Hz was evoked in the sensory neuron to induce PTP. Thirty

seconds after the end of the train, the EPSP and the response to hypertonic solution were

tested again and were compared to the test responses at the beginning of the experiment.

Note that in this comparison, unlike that of Figure 1B, the EPSP measured in the first run

of the experiment, before a 10-minute wash, was compared to the EPSP elicited after

HFS in the second run of the experiment, following the wash. These are the same

experiments as those shown in Figure 1B, but in order to compare the effects of HFS on

the EPSP and on the RRP it was necessary to measure both parameters before and after

an intervening wash to remove the hypertonic solution.

Eliciting 20 action potentials at these synapses at an interstimulus interval of 30 to 60

seconds results in depression of about 75% in the EPSP and of about 50% in the RRP,

compared to the initial test values (Zhao and Klein, 2002). In the present experiments,

the first EPSP of the high-frequency train was depressed compared to the initial test value

as a result of the prior electrical stimulation and hypertonic stimulus (to 0.644 ± 0.087 of

the initial value, n=25, p=0.0002). After HFS, the EPSP and the RRP averaged 1.711 ±

0.272 and 1.127 ± 0.075 of the initial test value, respectively (n = 29), neither of which

was significantly different from the test value (p=0.182 for the EPSP and p=0.357 for the

RRP). HFS thus reversed homosynaptic depression and led to an increase in the EPSP

and in the RRP over the initial values in many instances (Figure 3A1, B1).

13

As was true for PTP, the change in the EPSP was negatively correlated with the

maximal slope of the initial EPSP (Figure 3A1), and in addition, the change in the RRP

was also negatively correlated with the initial synaptic strength (Figure 3B1). As a

control for the statistical artifact discussed earlier, the ratio between the first Pre-HFS

EPSP and the Test EPSP was plotted against the maximal slope of the initial EPSP

(Figure 3A1, Inset). There was no correlation between the two values (r2 = 0.001, p =

0.870, n = 27), demonstrating the physiological significance of the inverse correlation

between the amount of potentiation and the initial EPSP.

When the connections were divided into weak and strong groups, as above, there was

a significant increase after HFS in the EPSP and in the size of the RRP compared to the

test values at the weak connections, while the strong connections showed no significant

change in the RRP and a small decrease in the EPSP (Figure 3A2, 3B2). Although the

EPSP at the initially strong connections did not increase over its initial value, it was

potentiated compared to the first EPSP of the train (Figure 1B). The average potentiated

EPSP at the weak connections (18.136 ± 2.704 mV/ms, n=15) did not differ from the

average potentiated EPSP at the strong connections (16.443 ± 1.966, n=14; p=0.616).

In some of the experiments, the high-frequency train was preceded by 5 action

potentials at 30-second intervals in the sensory neuron in order to examine the influence

of prior homosynaptic depression on the potentiation by HFS. Prior homosynaptic

depression had no effect on the change in the EPSP or in the RRP following HFS,

relative to the test value (Figure 3A1, 3B1, circles with asterisks).

If the effect of the high-frequency train on the EPSP were the result of an increase in

the size of the RRP, then the changes in the two parameters should be positively

14

correlated. This was in fact the case. Plotting the EPSP ratio (Post-HFS/Test) against the

corresponding RRP ratio (Figure 4A) showed that the two measures were significantly

correlated (r2 = 0.52, p<0.0001, n= 28, one outlier excluded). The fact that the changes in

the EPSP and in the RRP were correlated with each other for the strong and the weak

connections when examined separately (r2 = 0.432, p=0.008 and r2 = 0.436, p=0.014,

respectively, with one weak outlier excluded) suggests that the changes in the RRP could

be contributing to the changes in the EPSP at all connections. However, closer

examination of the plot of Figure 4A revealed that the changes in the EPSP and in the

RRP were more nearly the same for the strong connections (filled circles) than for the

weak connections (open circles). On average, the changes in the EPSP and in the RRP

were similar at strong connections (Figure 4B; ratio of Post-HFS to Test 0.852 ± 0.060

and 0.938 ± 0.077, respectively, n=14, p=0.151), while the increase in the EPSP was

significantly greater than the increase in the RRP at the weak connections (Figure 4B;

2.633 ± 0.433 and 1.329 ± 0.107, respectively, n=15, p=0.010).

If the changes in the EPSP were due entirely to changes in the RRP, then the EPSP

ratio should be equal to the RRP ratio in each experiment. It can be seen in Figure 4A

that these ratios are approximately equal for the initially strong connections (filled

circles), but not for the initially weak connections (open circles). Figure 4C shows a plot

of the EPSP ratio divided by the RRP ratio as a function of initial synaptic strength. This

quotient is approximately 1 for the initially stronger connections, but significantly greater

than 1 for the weaker connections. We excluded the possibility of a statistical artifact

(see beginning of the Results section) by plotting this quotient for control preparations in

which the HFS was omitted. The was no correlation between the quotient and the initial

15

synaptic strength (r2 = 0.027, n = 22, p > 0.25). These observations suggest that the effect

of high-frequency stimulation on the EPSP might be fully explained by an effect on the

size of the RRP at strong connections, but that additional factors must be implicated at

weak connections.

Potentiation of the EPSP is not accompanied by an increase in postsynaptic

sensitivity

The differences between the effects of HFS on synaptic transmission at initially

weak and initially strong connections support the idea that there is a qualitative difference

between the two types of synaptic connection. Schaffhausen et al. (2001) showed that

potentiation of synaptic transmission by HFS includes a postsynaptic contribution, but

only at initially weak connections. One possible postsynaptic mechanism that could

contribute to an increase in the EPSP is an increase in the postsynaptic sensitivity to the

neurotransmitter, as expressed in an increase in the quantal amplitude. If this were to

occur, it would be expected that the average amplitude of the miniature synaptic

potentials would increase.

In order to test the hypothesis that an increase in postsynaptic sensitivity amplifies

synaptic transmission after HFS, we compared the average mini amplitude before and

after HFS. At the weak connections (n=14), the average amplitude of the minis evoked

by application of hypertonic solution after the HFS did not differ from the amplitude

before HFS (Figure 5A; 0.762 ± 0.096 mV and 0.801 ± 0.089 mV, respectively,

p=0.448).

Interestingly, there was a significant decrease in the average mini amplitude after

HFS at the initially strong connections (Figure 5A; from 0.628 ± 0.071 to 0.544 ± 0.042,

16

n=15, p=0.004). In order to examine this change in the mini amplitude in more detail, we

compared the distributions of the mini amplitudes before and after HFS at the initially

strong connections. In practically all experiments, the first peak of the amplitude

frequency histogram (presumably the unitary quantal amplitude) did not shift its position,

but there was a redistribution of the amplitudes towards smaller amplitudes after HFS

(see examples in Figure 5B). Since peaks at larger amplitudes often appeared to be

integral multiples of the amplitude of the first peak, the larger minis may represent

multiquantal release events (see also Zhao and Klein, 2002). The decrease in the average

mini amplitude might therefore reflect a decrease in the relative proportion of

multiquantal events. Alternatively, the decrease in the mini amplitude could represent a

change in vesicle filling, among other possibilities. In any case, there was clearly no

increase in the mini amplitude at either weak or strong connections that could have

contributed to the increase in the EPSP.

Efficacy of release is lower at initially weak connections and increases after HFS

In order to see if there were other differences between initially weak and initially

strong connections, we compared them with respect to several other parameters. There

was no difference in the initial average mini amplitude between the two groups (Figure

6A1), nor was there any correlation between the mini amplitude and the initial synaptic

strength (Figure 6A2). The initial pool size, while not significantly different between

weak and strong connections (Figure 6B1, but see Discussion), was correlated with the

initial synaptic strength (Figure 6B2).

We next compared the fraction of the RRP that was released in the initial EPSP, a

measure of the efficacy of release. In order to compute this value, we first divided the

17

estimated amplitude of the EPSP by the average mini amplitude for each experiment (see

Methods for details). If the average mini amplitude were the same as the quantal

amplitude, this calculation would yield an estimate of the quantal content of the EPSP.

Since it is unlikely that the average mini amplitude and the quantal amplitude are the

same in our experiments (see above), this calculation gives an estimate of the number of

minis comprising the EPSP. The number of minis in the EPSP divided by the total

number of minis in the RRP gives the fraction of the RRP released with the EPSP.

Because the size of the RRP and the average mini amplitude were not significantly

different between the two groups of connections, it is to be expected that the difference in

synaptic strength was due to a difference in the fraction of the RRP released in the EPSP.

In fact, there was a highly significant difference between the RRP fractions released at

weak and at strong connections (Figure 7A1), with a single EPSP at strong connections

releasing on average (0.597±0.067, n=11) more than twice the pool fraction as an EPSP

at weak connections (0.271±0.050, n=11, p<0.001 compared to initially strong

connections). This difference was reflected in the correlation between the fraction of the

RRP released and the initial synaptic strength (Figure 7A2).

We then compared the fractions of the RRP released with a single EPSP before and

after HFS at both initially weak and initially strong connections. High-frequency

stimulation led to an increase in the fraction of the RRP released with an EPSP at initially

weak connections, but not at initially strong connections (Figure 7B; weak connections:

from 0.286 ± 0.049 before HFS to 0.508 ± 0.057 afterwards, p=0.00008, n=14; strong

connections: from 0.665 ± 0.079 to 0.726 ± 0.075, p=0.260, n=15). Thus, in addition to

a significant increase in the size of the RRP at weak connections after HFS (Figure 3B2),

18

there was also an increase in the efficacy of release, as reflected in the fraction of the

RRP released.

High-frequency stimulation increases the early phase of transmitter release in

response to hypertonic solution at initially weak connections

As another possible indicator of the efficacy of release, we examined the time course

of the asynchronous release triggered by application of the hypertonic solution. The

response to the hypertonic solution was relatively slow, peaking at about 10 seconds after

the introduction of the hypertonic solution, and declining virtually completely after about

50-60 seconds (Figure 8A). The magnitude and the time course of the asynchronous

responses were initially identical for weak and strong connections (Figure 8A).

However, after HFS, there was a marked increase in the early part of the response at

weak connections (Figure 8B), but not at strong connections (Figure 8C). At the initially

weak connections, the increase in the number of minis occurred in the first 10 seconds of

application of the hypertonic sucrose; after the first 10 seconds, the number and time

course of the minis after HFS were identical to those before HFS (Figure 8B). There was

no increase in the average number of minis at any time point at the initially strong

connections (Figure 8C).

Although there was no increase in the number of minis at the initially strong

connections, there was a modest acceleration of the time course of release at the strong

connections also. When the release at each time point was plotted as a fraction of the

total number of minis in each experiment, rather than simply as the number of minis,

there was an increase after HFS in the fraction released in the first 5 seconds at all

connections (Figure 9; weak connections: from 0.184 ± 0.021 to 0.290 ± 0.034, n=13,

19

p<0.004; strong connections: from 0.190 ± 0.015 to 0.250 ± 0.028, n=12, p<0.05). If

this change in the kinetics of the asynchronous release reflects a change in the efficacy of

release, then this change in efficacy is not due to modulation of excitation-secretion

coupling per se, because the release induced by hypertonic solution is dependent on

neither presynaptic depolarization nor presynaptic calcium entry from the outside

(Rosenmund and Stevens, 1996; Zhao and Klein, 2002).

20

DISCUSSION

Potentiation by high-frequency stimulation

In a previous study, we reported that evoking 15-20 action potentials at an

interstimulus interval of 10 to 30 seconds depressed the EPSP by approximately 75% and

reduced the RRP by about 50% (Zhao and Klein, 2002). In contrast, we report here that

triggering the same number of action potentials at 20 Hz resulted in potentiation of the

EPSP relative to the first EPSP of the high-frequency train (Figure 1B). In order to

examine the relationship between changes in the EPSP and in the RRP resulting from

HFS, we had to use a protocol in which the EPSP and the RRP after HFS were compared

to the values at the beginning of the experiment (Figure 2), which yielded a smaller

measured potentiation of the EPSP (Figure 3A1). Using this protocol, there was either no

change in the RRP or a small increase (Figure 3B2). HFS thus prevented homosynaptic

depression of both the EPSP and the RRP. It is also noteworthy that the effects of HFS

on the EPSP and on the RRP were the same whether or not homosynaptic depression was

first induced, indicating that the depression is reversed by the HFS (Figures 3A1 and

3B1, circles with asterisks). At a mammalian synapse, the accumulation of calcium

resulting from HFS increases the rate of recovery of the RRP from depletion (Wang and

Kaczmarek, 1998). Although it is not clear that homosynaptic depression at Aplysia

sensorimotor synapses results from depletion (Gover et al., 2002; Royer et al., 2000),

HFS may act in a similar way to prevent and to reverse depression.

Changes in vesicle recruitment or in the efficacy of the hypertonic stimulus as

alternative explanations

21

We have considered the changes in the number of minis to be changes in the size of

the RRP. However, there are other possible explanations for the effects of HFS on the

response to the hypertonic stimulus. First, it is possible that rather than there being a

change in the size of the RRP, the HFS leads to increased recruitment of vesicles during

the hypertonic stimulus.

Increased recruitment is unlikely to be the explanation for the increased number of

minis, for several reasons. First, our previous study (Zhao and Klein, 2002) suggested

that there is little or no recruitment during the hypertonic challenge because cross-

depletion experiments using electrical and hypertonic stimulation showed a good

quantitative agreement between the RRP that is available for each of these types of

stimulation. In addition, the fact that the asynchronous release ceases almost completely

after about a minute (see Figures 8 and 9) suggests that there is little if any rapid

replenishment of the RRP during the hypertonic challenge. Finally, if the increase in the

number of minis were due to increased replenishment of the RRP during the period of

exposure to the hypertonic solution, it would be necessary to maintain that the close

match between the changes in the number of minis and in the EPSP (at least for strong

connections—see Figure 4) is the result of a highly unlikely coincidence.

A different possible explanation for the change in the number of minis released in

response to the hypertonic sucrose after HFS is that there might be an increase in the

effectiveness of the sucrose, such that the maximal releasable number of minis (the RRP)

remains the same, but that there ocurrs a shift in the dose-response relationship between

the sucrose concentration and the asynchronous release. If this were true, we would then

have to reinterpret the change in the number of minis after HFS as a change in the

22

efficacy of release rather than a change in the size of the RRP. Here too, the fact that the

changes in the number of minis and in the EPSP are in such close agreement argues that

whatever is responsible for the increased sensitivity to sucrose is also responsible for the

increased efficacy of an action potential in triggering release. We believe that this

possibility is not very likely because the number of minis after HFS returns to close to the

pre-HFS values (as opposed to the decrease that occurs with low-frequency stimulation),

and if the restoration of the response to hypertonic solution were due to a shift in the

dose-response relationship, there is no reason why the pre-HFS and post-HFS values

should be so close to each other (see Figure 3). In any event, this hypothesis does not

address the differences between the responses of initially weak and initially strong

connections to HFS.

Distinguishing between weak and strong connections

We and others report quantitative and qualitative differences between plasticity at

initially weak and initially strong sensorimotor connections (Sugita et al., 1992; Jiang

and Abrams, 1998; Schaffhausen et al., 2001). How do we decide whether to classify a

given connection as strong or weak? All of these studies adopted the rough criterion of

designating connections stronger than the mean or the median for all experiments as

“strong”, and those weaker as “weak”. Since there is no reason a priori why the dividing

line between the two types of connections should fall exactly at the mean or the median,

it is clear that some of the connections designated as strong might actually be weak, and

vice versa, especially in the region of presumed overlap around the mean or median

value. In Figure 3A1, for example, the points in the plot fall into two distinct groups,

with a PSP ratio of 2.5-3 for connections with initial maximal slopes less than about 11

23

mV/ms, and a PSP ratio of slightly less than one for connections with initial maximal

slopes greater than 12-13 mV/ms. Connections in the intermediate region of about 11-13

mV/ms are distributed between the two groups. There may be a similar clustering of

points into two groups in Figure 7A2, although more experiments would be needed to

show this definitively.

The absence of a clear criterion for distinguishing initially weak from initially strong

connections will lead to an underestimate of the differences between the two putative

groups of connections. For example, we reported that there was no significant difference

in the size of the RRP between weak and strong connections (Figure 6B1). However, if

connections with initial maximal slopes less than 11 mV/ms are compared with those

with initial maximal slopes greater than 13.5 mV/ms (Figure 6B2), there is a significant

difference in the average RRP size (p=0.022). Thus, the lack of a statistically significant

difference between the RRP sizes of strong and weak connections using the median value

as the dividing line is probably not meaningful, and the size of the RRP is likely to be an

additional parameter in which strong and weak connections differ.

Changes in the efficacy of release at initially weak connections

We found two ways in which the efficacy of release increased after HFS at initially

weak connections: First, there was an increase in the fraction of the RRP released with a

single EPSP (Figure 7B). If the EPSP increases without a quantitatively similar increase

in the size of the RRP, and with no increase in the mini amplitude, then it must

necessarily follow that the fraction of the RRP released in an EPSP increases. Although

this would be true in any case, we observed that there was a greater than twofold

difference between initially weak and initially strong connections in the fraction of the

24

RRP released with an EPSP (Figure 7A1). These observations suggest that weak

connections are not simply scaled-down versions of strong connections, but rather that

the limiting steps in the release process differ at initially weak and initially strong

connections. Thus, at initially strong connections, where the fraction of the RRP released

is already high, potentiation is mediated through an increase in the size of the RRP. At

initially weak connections, by contrast, both parameters are up-modulated by HFS.

The second indication of an increase in efficacy of release at initially weak

connections is the large increase after HFS in the frequency of miniature synaptic

potentials in the first 10 seconds after application of hypertonic solution (Figure 8B).

Because the asynchronous release triggered by hypertonic solution depends on neither

presynaptic depolarization nor presynaptic calcium entry from the outside (Rosenmund

and Stevens, 1996; Zhao and Klein, 2002), this increase does not reflect a change in

excitation-secretion coupling per se. By contrast, in our study of homosynaptic

depression and its reversal by serotonin, we found that these forms of plasticity were

accompanied by changes in the efficacy of release, but that these changes could be

detected only when release was triggered by action potentials, thereby implying

modulation of excitation-secretion coupling (Zhao and Klein, 2002).

Postsynaptic factors in PTP

PTP at the sensorimotor synapses of Aplysia comprises a postsynaptic component in

addition to the presynaptic process that it shares with the PTP described elsewhere, but

only at initially weak connections (Schaffhausen et al., 2001). Our finding that the

amplitude of the miniature synaptic potentials did not increase after HFS (Figure 5A)

implies that this postsynaptic contribution is not expressed as an increase in postsynaptic

25

sensitivity. On the other hand, we show that a change in the size of the RRP can account

fully for the change in the EPSP only at initially strong connections, suggesting that

additional factors must play a role in the potentiation of the EPSP at weak connections

(Figure 4A-C). We therefore suggest that modulation of the size of the RRP is mediated

by an exclusively presynaptic process at all connections, whereas the changes in the

efficacy of release at initially weak connections may be caused by a retrograde signal

from the postsynaptic neuron.

Differences between strong and weak connections

What could account for the differences between initially weak and initially strong

connections? One possibility is that the populations of sensory and/or motor neurons are

not homogeneous, but that different subclasses of these cells form connections with

different properties, as has been found for cricket synapses, for example (Davis and

Murphey, 1993). Large differences in the fraction of the RRP released to a single action

potential have been reported at tonic and phasic motor neurons of the crayfish, and,

similarly to our findings, the two types of motor neuron show marked differences in their

plastic properties (Millar et al., 2002). A second possibility is that the differences in the

synaptic properties might be the result of differences in maturational state of the

synapses, as has been reported in several other systems. In mammals, LTP is more easily

induced at immature synapses (Bolshakov and Siegelbaum, 1995; Liao and Malinow,

1996), and there appears to be a switch in the biochemical signaling pathways mediating

LTP as synapses mature (Yasuda et al., 2003). Insertion of postsynaptic receptors in

Xenopus neurons decreases as synapses mature (Wu et al., 1996), and, like PTP at the

sensorimotor connections of Aplysia, the potentiation of weaker connections at

26

developing neuromuscular synapses of Xenopus is dependent on elevation of

postsynaptic calcium (Wan and Poo, 1999). In Aplysia, the processes mediating synaptic

facilitation by serotonin at the sensorimotor synapses change as the animal matures

(Marcus and Carew, 1998), and a similar change takes place as these synapses mature in

culture (Sun and Schacher, 1996). Taken together, these observations suggest that the

initially weak sensorimotor synapses of Aplysia might be immature synapses, which are

susceptible to forms of modulation that do not play as large a role in the plasticity of

strong, or mature, synapses.

Whatever the basis for the differences in plasticity at connections of different initial

synaptic strength, our study raises questions that bear on fundamental aspects of synaptic

communication. It would be of great interest to test the idea that different steps in the

release process are limiting at different subclasses of connections of the same type,

whether as a consequence of maturational stage, of prior activity, or of the identity of the

postsynaptic target cell, to take a few examples. These differences could reflect

differences in the individual protein components of the synaptic apparatus (for example,

differences in isoform, in subunit composition or in phosphorylation state), in their

relative abundance, or in their localization at the synapse. Modulation of synaptic

transmission, whether by intrinsic activity or by exogenous modulators, can thus take

different forms at different synapses depending on the relative contributions of the

various steps in the release process.

27

ACKNOWLEDGMENTS

We thank Dr. Alan Grinell and Dr. David Glanzman for their helpful comments on

an earlier version of the manuscript. This work was supported by grants OGP0138426

from the Natural Sciences and Engineering Research Council of Canada, NS 36648 from

the National Institutes of Health and 0215647 from the National Science Foundation.

28

REFERENCES

Bao,J.X., Kandel,E.R., and Hawkins,R.D. Involvement of pre- and postsynaptic mechanisms in posttetanic potentiation at Aplysia synapses. Science 275:969-973, 1997.

Berglund,K., Midorikawa,M., and Tachibana,M. Increase in the pool size of releasable synaptic vesicles by the activation of protein kinase C in goldfish retinal bipolar cells. J. Neurosci. 22:4776-85. 22:4776-85., 2002.

Bliss,T.V. and Collingridge,G.L. A synaptic model of memory: long-term potentiation in the hippocampus. Nature 361:31-39, 1993.

Bolshakov,V.Y. and Siegelbaum,S.A. Regulation of hippocampal transmitter release during development and long-term potentiation [see comments]. Science269:1730-1734, 1995.

Coulson,R.L. and Klein,M. Rapid development of synaptic connections and plasticity between sensory neurons and motor neurons of Aplysia in cell culture: implications for learning and regulation of synaptic strength. J Neurophysiol77:2316-2327, 1997.

Davis,G.W. and Murphey,R.K. A role for postsynaptic neurons in determining presynaptic release properties in the cricket CNS: evidence for retrograde control of facilitation. J. Neurosci. 13:3827-38., 1993.

Eliot,L.S., Kandel,E.R., and Hawkins,R.D. Modulation of spontaneous transmitter release during depression and posttetanic potentiation of Aplysia sensory-motor neuron synapses isolated in culture. J Neurosci 14:3280-3292, 1994.

Frost,W.N. and Kandel,E.R. Structure of the network mediating siphon-elicited siphon withdrawal in Aplysia. J. Neurophysiol. 73:2413-27., 1995.

Gillis,K.D., Mossner,R., and Neher,E. Protein kinase C enhances exocytosis from chromaffin cells by increasing the size of the readily releasable pool of secretory granules. Neuron 16:1209-1220, 1996.

Goda,Y. and Stevens,C.F. Readily releasable pool size changes associated with long term depression. Proc. Natl. Acad. Sci. U. S. A. 95:1283-8., 1998.

Gover,T.D., Jiang,X.Y., and Abrams,T.W. Persistent, exocytosis-independent silencing of release sites underlies homosynaptic depression at sensory synapses in Aplysia. J. Neurosci. 22:1942-55., 2002.

Jiang,X.Y. and Abrams,T.W. Use-dependent decline of paired- pulse facilitation at Aplysia sensory neuron synapses suggests a distinct vesicle pool or release mechanism. J. Neurosci. 18:10310-9., 1998.

29

Kim,J. and Alger,B.E. Random response fluctuations lead to spurious paired-pulse facilitation. J Neurosci 21:9608-9618, 2001.

Klein,M. Synaptic augmentation by 5HT at rested Aplysia sensorimotor synapses: Independence of action potential prolongation. Neuron 13:159-166, 1994.

Liao,D. and Malinow,R. Deficiency in induction but not expression of LTP in hippocampal slices from young rats. Learn. Mem. 3:138-49., 1996.

Lin,X.Y. and Glanzman,D.L. Long-term potentiation of Aplysia sensorimotor synapses in cell culture: regulation by postsynaptic voltage. Proc Royal Soc London - Ser B Biol Sci 255:113-118, 1994.

Marcus,E.A. and Carew,T.J. Developmental emergence of different forms of neuromodulation in Aplysia sensory neurons. Proc. Natl. Acad. Sci. U. S. A95:4726-4731, 1998.

Millar,A.G., Bradacs,H., Charlton,M.P., and Atwood,H.L. Inverse relationship between release probability and readily releasable vesicles in depressing and facilitating synapses. J Neurosci 22:9661-9667, 2002.

Rosenmund,C. and Stevens,C.F. Definition of the readily releasable pool of vesicles at hippocampal synapses. Neuron 16:1197-1207, 1996.

Royer,S., Coulson,R.L., and Klein,M. Switching off and on of synaptic sites at aplysia sensorimotor synapses. J. Neurosci. 20:626-38., 2000.

Schacher,S. and Proshansky,E. Neurite regeneration by Aplysia neurons in dissociated cell culture: modulation by Aplysia hemolymph and the presence of the initial axonal segment. J Neurosci 3:2403-2413, 1983.

Schaffhausen,J.H., Fischer,T.M., and Carew,T.J. Contribution of postsynaptic Ca2+ to the induction of post- tetanic potentiation in the neural circuit for siphon withdrawal in Aplysia. J. Neurosci. 21:1739-49., 2001.

Schikorski,T. and Stevens,C.F. Morphological correlates of functionally defined synaptic vesicle populations. Nat. Neurosci. 4:391-5., 2001.

Stevens,C.F. and Sullivan,J.M. Regulation of the readily releasable vesicle pool by protein kinase C. Neuron 21:885-93., 1998.

Sugita,S., Goldsmith,J.R., Baxter,D.A., and Byrne,J.H. Involvement of protein kinase C in serotonin-induced spike broadening and synaptic facilitation in sensorimotor connections of Aplysia. J Neurophysiol 68:643-651, 1992.

Sun,Z.Y. and Schacher,S. Development of short-term heterosynaptic facilitation at aplysia sensorimotor synapses in vitro is accompanied by changes in the

30

functional expression of presynaptic serotonin receptors. J. Neurophysiol.76:2250-2261, 1996.

Walters,E.T. and Byrne,J.H. Post-tetanic potentiation in Aplysia sensory neurons. Brain Res 293:377-380, 1984.

Walters,E.T., Byrne,J.H., Carew,T.J., and Kandel,E.R. Mechanoafferent neurons innervating tail of Aplysia. I. Response properties and synaptic connections. J Neurophysiol 50:1522-1542, 1983.

Wan,J. and Poo,M. Activity-induced potentiation of developing neuromuscular synapses. Science 285:1725-8., 1999.

Wang,L.Y. and Kaczmarek,L.K. High-frequency firing helps replenish the readily releasable pool of synaptic vesicles. Nature 394:384-8., 1998.

Waters,J. and Smith,S.J. Phorbol esters potentiate evoked and spontaneous release by different presynaptic mechanisms. J. Neurosci. 20:7863-70., 2000.

Wu,G., Malinow,R., and Cline,H.T. Maturation of a central glutamatergic synapse. Science 274:972-6., 1996.

Yasuda,H., Barth,A.L., Stellwagen,D., and Malenka,R.C. A developmental switch in the signaling cascades for LTP induction. Nat. Neurosci. 6:15-16, 2003.

Zhao,Y. and Klein,M. Modulation of the readily releasable pool of transmitter and of excitation-secretion coupling by activity and by serotonin at Aplysia sensorimotor synapses in culture. J. Neurosci. 22:10671-10679, 2002.

Zucker,R.S. Calcium- and activity-dependent synaptic plasticity. Curr. Opin. Neurobiol.9:305-13., 1999.

Zucker,R.S. and Regehr,W.G. Short-term synaptic plasticity. Annu Rev Physiol 64:355-405., 2002.

31

FIGURE LEGENDS

Figure 1. Post-tetanic potentiation varies with initial synaptic strength. A.

Experimental protocol. A train of EPSPs in a motor neuron elicited by extracellular

stimulation of a sensory neuron at 20 Hz was followed after 30 seconds by a single

stimulus to measure PTP (top traces). Potentiation of the EPSP is expressed as the ratio

of the maximal slope of the EPSP following the train (PostHFS) to that of the first EPSP

of the train (PreHFS). PreHFS and PostHFS EPSPs are shown on an expanded scale in

the bottom traces. Calibrations: Top traces, 10 mV and 100 ms; bottom, 20 mV and 20

ms. B. PTP decreases as the initial synaptic strength increases. Each point represents

one experiment; dotted line shows PTP=1, or no potentiation. C. Control experiment

showing that the inverse relationship shown in B cannot be attributed to a statistical

artifact. Two EPSPs were elicited at a 1-minute interval without intervening HFS; there

is no correlation between the magnitude of the initial EPSP and the ratio of the second

EPSP to the first EPSP (see Results for further explanation). D. Connections were

divided into weak (n=14) and strong (n=11) at the median of the initial maximal slope

(14 mV/ms), and PTP at weak and strong connections were averaged separately. The

means are significantly different (p<0.02).

Figure 2. Protocol for comparing changes in the EPSP and in the RRP after

high-frequency stimulation. A single EPSP was elicited with a stimulus to the sensory

neuron, and asynchronous transmitter release in the form of miniature synaptic potentials

was triggered by application of a hypertonic sucrose solution (Test). After a 10-minute

wash, a train of stimuli at 20 Hz was given (HFS), and the EPSP and response to

hypertonic solution were elicited again (Post). The EPSP following HFS triggered an

32

action potential, which is truncated in this record. Calibrations: 25 mV and 15 ms for the

EPSPs and 3.5 mV and 1 sec for the minis.

Figure 3. Weak and strong connections respond differently to HFS. A. The

ratio of the Post-HFS EPSP to the Test EPSP was negatively correlated with the initial

synaptic strength (A1; r 2 = 0.298, p=0.002). Circles with asterisks represent experiments

in which the test EPSP had first been depressed by stimulating the sensory neuron 5 times

(see Methods and Results). Inset shows experiment controlling for statistical artifact

(explained in Results). The ratio of the first Pre-HFS EPSP of the second run of the

experiment to the Test EPSP is plotted against the maximal slope of the initial EPSP.

Unlike the ratio of the Post-HFS EPSP to the Test EPSP (A1), there is no correlation

between the maximal slope of the initial EPSP and the Pre-HFS-to-Test EPSP ratio. A2.

After HFS, the EPSP showed a significant increase (p=0.002) at initially weak

connections (from 7.847 ± 1.019 mV/ms to 18.136 ± 2.799 mV/ms, n=14), and a

significant decrease (p=0.014) at initially strong connections (from 20.673 ± 2.770 to

16.443 ± 1.899 mV/ms, n=15). B. The ratio of the number of minis in the response to

hypertonic solution after HFS to the number of minis in the Test response was also

negatively correlated with the initial synaptic strength (B1; r2 = 0.335, p=0.001); circles

with asterisks represent depressed EPSPs, as in A1. Only the response at weak

connections changed significantly after HFS (B2). The number of minis in the RRP

increased at initially weak connections (from 113.500 ± 13.462 to 151.357 ± 21.994,

n=14, p=0.019), and was unchanged at initially strong connections (124.267 ± 12.986

before and 110.600 ± 11.556 after HFS; n=15, p=0.113).

33

Figure 4. The changes in the RRP and in the EPSP after HFS are similar at

initially strong connections, but differ at initially weak connections. A. The EPSP

ratios (Rpsp) and the mini ratios (Rminis) are correlated for all connections taken

together (r2 = 0.520, p<0.0001), but the ratios are approximately equal only at the strong

connections (filled circles). The dashed line has a slope of 1. B. The EPSP ratios and

the mini ratios are significantly different for the weak connections, but not for the strong

connections. C. Plot of the EPSP ratio divided by the mini ratio as a function of initial

synaptic strength. The ratios are approximately equal for initial EPSPs with maximal

slope greater than about 13 mV/ms, while EPSPs of initially weaker synaptic strength

show greater changes in the EPSP than in the size of the RRP.

Figure 5. Amplitudes of miniature synaptic potentials before and after HFS. A.

At initially weak connections there is no change in the mini amplitude after HFS, while at

initially strong connections there is a small decrease. B. Three examples of amplitude

frequency histograms of miniature synaptic potentials taken from experiments on initially

strong connections. The positions of the peaks are similar before and after HFS in each

experiment, but there is a decrease in the relative proportion of larger-amplitude events

and a corresponding increase in the relative proportion of smaller-amplitude events.

Figure 6. Comparison of mini amplitudes and RRP size at weak and strong

connections. A1. There is no significant difference between the mini amplitudes at

weak and strong connections (weak: 0.856 ± 0.095 mV, n=12; strong: 0.697 ± 0.081

mV, n=13, p=0.212), and (A2) there is no correlation between mini amplitude and initial

synaptic strength (r2=0.001, p=0.895). B. The size of the RRP is not significantly

different at weak and strong connections (B1; 200 ± 23 at strong connections, n=11, and

34

148 ± 17 at weak connections, n=11, p=0.088; but see Discussion), while there is a

correlation between RRP size and initial synaptic strength (B2; r2 = 0.365, p=0.003).

Figure 7. The fraction of the RRP released in the initial EPSP differs at weak

and strong connections, and HFS is followed by an increase in this fraction only at

weak connections. A. The initial EPSP at strong connections releases more than twice

the fraction of the RRP compared to weak connections (A1; p<0.001 ), and the initial

RRP fraction is correlated with the initial synaptic strength (A2; r2 = 0.293, p=0.009 ) B.

The fraction of the RRP released in the EPSP increases after HFS at weak connections

(p<0.0001 ), but not at strong connections (p=0.260 ).

Figure 8. HFS increases the early phase of release to a hypertonic stimulus only

at initially weak connections. A. The number and time course of miniature synaptic

potentials in response to a hypertonic stimulus are the same at weak and strong

connections in the initial test. B, C. The number of minis in the first 10 seconds of the

hypertonic application is greatly increased after HFS at initially weak connections (from

51.308 ± 7.774 to 87.385 ± 16.032, n=13, p=0.0075), but not at initially strong

connections (from 54.083 ± 10.155 to 49.667 ± 7.508, n=12).

Figure 9. HFS accelerates the asynchronous response to a hypertonic stimulus

at all connections. Plotting the number of minis as a fraction of the total released by the

hypertonic solution reveals an increase in the rate of release in the first 5 seconds of the

application at both weak (p<0.004 ) and strong (p<0.05) connections.

35

FIGURE 1

A

0 10 20 30 40 500

3

6

9

12

Initial PSP (mV/ms)P

TP

(Pos

tHF

SP

reH

FS

)

Weak Strong0

2

4

6

PT

P(P

ost

HF

SP

reH

FS

)

B

D

PreHFS PostHFS

PreHFS PostHFS

0 10 20 30 40 500.0

0.5

1.0

1.5

Initial PSP (mV/ms)

PS

P2/

PS

P1

C

36

FIGURE 2

Sucrose Sucrose

HFS

Test Post

10'Wash

37

FIGURE 3

0 10 20 30 40 500

1

2

3

4

Initial PSP (mV/ms)

PS

P r

atio

(Pos

t/T

est)

0 10 20 30 40 500.00.51.01.52.02.5

Initial PSP (mV/ms)M

ini r

atio

(Po

st/T

est)

0

10

20

30 TestPost

Weak Strong

PS

P(m

V/m

s)

0

100

200

300 TestPost

Weak Strong

RR

P (

# o

fm

inis

)

A1

A2 B2

B1

0 10 20 30 40 500.0

0.5

1.0

1.5

Initial PSP (mV/ms)

PS

P r

un #

2te

st P

SP

38

0.0 0.5 1.0 1.5 2.0 2.50

1

2

3

4

y=x

Rminis

Rp

sp

0

1

2

3

4 EPSPMinis

Weak Strong

Rat

io(P

ost/

Tes

t)

0 10 20 30 40 500

1

2

3

4

Initial PSP (mV/ms)

Rp

sp/R

min

is

A

CB

FIGURE 4

39

0.0 0.5 1.0 1.50

10

20

30

40 TestPost

Amplitude (mV)

# of

eve

nts

0.0 0.5 1.0 1.5 2.0 2.50

20

40

60 TestPost

Amplitude (mV)#

of e

ven

ts

0.00 0.25 0.50 0.75 1.000

10

20

30 TestPost

Amplitude (mV)

# of

eve

nts

0.00

0.25

0.50

0.75

1.00 TestPost

Weak Strong

Min

i Am

plit

(mV

)

A

B

FIGURE 5

40

0.00

0.25

0.50

0.75

1.00

Weak Strong

Min

i am

plit

(mV

)

050

100150200250

Weak StrongR

RP

(#

ofm

inis

)

0 10 20 30 40 500.0

0.5

1.0

1.5

2.0

Initial PSP (mV/ms)

Min

i am

plit

(mV

)

0 10 20 30 40 500

100

200

300

400

Initial PSP (mV/ms)

RR

P (

# of

min

is)

A1 B1

A2 B2

FIGURE 6

41

0.00

0.25

0.50

0.75

Weak Strong

Initi

al P

SP

(fra

ct R

RP

)

0.00

0.25

0.50

0.75

1.00 TestPost

Weak Strong

PS

P (

frac

tion

of R

RP

)

0 10 20 30 40 500.00

0.25

0.50

0.75

1.00

Initial PSP (mV/ms)

Initi

al P

SP

(fra

ct R

RP

)

A1

B

A2

FIGURE 7

42

0 10 20 30 40 500

10

20

30

40 StrongWeak

Time (sec)

# of

min

isA

Weak

0 10 20 30 40 500

102030405060

TestPost

Time (sec)

# of

min

is

Strong

0 10 20 30 40 500

10

20

30

40 TestPost

Time (sec)

# of

min

is

B C

FIGURE 8

43

Weak

0 10 20 30 40 500.0

0.1

0.2

0.3 TestPost

Time (sec)

Min

is(f

ract

ion

)

Strong

0 10 20 30 40 500.0

0.1

0.2

0.3 TestPost

Time (sec)

Min

is(f

ract

ion

)

FIGURE 9