Neuromodulator-evoked synaptic metaplasticity ... - Semantic Scholar

J Neurophysiol 108: 2846 –2856, 2012. First published August 29, 2012; doi:10.1152/jn.00586.2012.

Neuromodulator-evoked synaptic metaplasticity within a central pattern generator network Mark D. Kvarta, Ronald M. Harris-Warrick, and Bruce R. Johnson Department of Neurobiology and Behavior, S. G. Mudd Hall, Cornell University, Ithaca, New York Submitted 9 July 2012; accepted in final form 27 August 2012

Kvarta MD, Harris-Warrick RM, Johnson BR. Neuromodulator-evoked synaptic metaplasticity within a central pattern generator network. J Neurophysiol 108: 2846 –2856, 2012. First published August 29, 2012; doi:10.1152/jn.00586.2012.—Synapses show shortterm activity-dependent dynamics that alter the strength of neuronal interactions. This synaptic plasticity can be tuned by neuromodulation as a form of metaplasticity. We examined neuromodulator-induced metaplasticity at a graded chemical synapse in a model central pattern generator (CPG), the pyloric network of the spiny lobster stomatogastric ganglion. Dopamine, serotonin, and octopamine each produce a unique motor pattern from the pyloric network, partially through their modulation of synaptic strength in the network. We characterized synaptic depression and its amine modulation at the graded synapse from the pyloric dilator neuron to the lateral pyloric neuron (PD¡LP synapse), driving the PD neuron with both long square pulses and trains of realistic waveforms over a range of presynaptic voltages. We found that the three amines can differentially affect the amplitude of graded synaptic transmission independently of the synaptic dynamics. Low concentrations of dopamine had weak and variable effects on the strength of the graded inhibitory postsynaptic potentials (gIPSPs) but reliably accelerated the onset of synaptic depression and recovery from depression independently of gIPSP amplitude. Octopamine enhanced gIPSP amplitude but decreased the amount of synaptic depression; it slowed the onset of depression and accelerated its recovery during square pulse stimulation. Serotonin reduced gIPSP amplitude but increased the amount of synaptic depression and accelerated the onset of depression. These results suggest that amine-induced metaplasticity at graded chemical synapses can alter the parameters of synaptic dynamics in multiple and independent ways. graded synaptic transmission; synaptic depression; lobster pyloric neurons; dopamine; serotonin; octopamine SHORT-TERM SYNAPTIC DYNAMICS, such as facilitation and depression, alter synaptic strength in activity-dependent ways (Abbott and Regehr 2004). Synaptic dynamics have been studied most extensively at synapses using action potential (AP)-evoked transmitter release (Fioravante and Regehr 2011; Fisher et al. 1997; Thomson 2000; Zucker and Regehr 2002). Transient, activity-dependent changes in synaptic strength can alter synaptic transmission to increase the response to novel stimuli, enhance directional sensitivity, adapt to repeated sensory stimuli, and detect bursts of activity in sensory networks (Chacron et al. 2009; Fortune and Rose 2001; Klug 2011). They can also contribute to gain control and network stabilization in central networks (Abbott and Regehr 2004) and help regulate motor network activity (Gelman et al. 2011; Li et al. 2007; Nadim and Manor 2000; Parker and Grillner 2000; Sanchez and Kirk 2002).

Address for reprint requests and other correspondence: B. R. Johnson, Dept. of Neurobiology and Behavior, S. G. Mudd Hall, Cornell Univ., Ithaca, NY 14853 (e-mail: [email protected]). 2846

Chemical synapses using graded transmitter release, where release is a continuous function of presynaptic membrane potential, also show short-term dynamics (Goutman and Glowatzki 2007; Jackman et al. 2009). The roles of graded synaptic dynamics have been analyzed in the pyloric network of the lobster stomatogastric ganglion (STG). In this 14-neuron central pattern generator (CPG) network, graded chemical synapses help determine the phasing of pyloric neurons to produce coordinated motor patterns (Hartline et al. 1988). The synaptic strength in this network dynamically tracks activity levels and thus alters network connectivity as a function of cycle frequency (Mouser et al. 2008; Nadim and Manor 2000). Synaptic depression during pyloric network activity is thought to control the transition to different oscillatory modes of the network (Nadim et al. 1999) and to promote phase maintenance over different cycle frequencies (Manor et al. 2003; Nadim et al. 2003). Short-term synaptic plasticity can be regulated through neuromodulation. This is a form of metaplasticity (Philpot et al. 1999), where neuromodulators alter the characteristics of activity-dependent changes in synaptic strength (Fischer et al. 1997; Kreitzer and Regehr 2000; Parker 2001; Parker and Grillner 1999; Qian and Delaney 1997). In the pyloric network, we have previously demonstrated that monoamines such as dopamine and octopamine can alter the steady-state synaptic depression at pyloric synapses (Johnson et al. 2005, 2011). In addition, the peptide proctolin can change the sign of synaptic plasticity from depression to facilitation at a pyloric graded synapse to help stabilize the cycle period (Zhao et al. 2011). In this article we examine amine modulation of synaptic dynamics from the pyloric dilator (PD) neuron to the lateral pyloric (LP) neuron, the PD¡LP synapse. This synapse provides an important output from the pacemaker kernel, which sets the cycle frequency of its follower cells and thus helps regulate network activity. The PD¡LP graded synapse shows synaptic depression when the PD neuron is driven with presynaptic square pulses (Graubard et al. 1983; Johnson and Harris-Warrick 1990) and when the PD neuron is artificially driven with realistic voltage waveforms that resemble its normal activity (Rabbah and Nadim 2007). Graded synaptic strength at nondepressed PD¡LP synapses is reduced by dopamine and serotonin and enhanced by octopamine (Johnson and Harris-Warrick 1990), but the effects of amines on PD¡LP synaptic strength under conditions of realistic rhythmic activity are unknown. Here we show that the three amines can have independent effects on synaptic strength, the magnitude and temporal dynamics of synaptic depression, and the recovery from depression.

0022-3077/12 Copyright © 2012 the American Physiological Society

www.jn.org

METAPLASTICITY AT A GRADED CHEMICAL SYNAPSE MATERIALS AND METHODS

General Procedures California spiny lobsters (Panulirus interruptus) were supplied by Don Tomlinson Commercial Fishing (San Diego, CA) and maintained in marine aquaria at 16°C. After lobsters were anesthetized in ice, the stomatogastric nervous system (STNS) was removed and pinned in a Sylgard-coated petri dish in chilled Panulirus saline of the following composition (mM): 479 NaCl, 12.8 KCl, 13.7 CaCl2, 3.9 Na2SO4, 10.0 MgSO4, 2 glucose, and 11.1 Tris base, pH 7.35 (Mulloney and Selverston 1974). The STG was desheathed, enclosed in a 1-ml pool walled with Vaseline, and superfused at 5 ml/min with oxygenated Panulirus saline. Experiments were performed at 19°C to enhance the strength of graded synaptic transmission (Johnson et al. 1991). Dopamine (DA; 10⫺5 M), octopamine (Oct; 10⫺5 M), and serotonin (5-HT; 10⫺5 M) were dissolved in saline just before application (amine conditions). If the amine effects did not reverse after a 15- to 60-min wash, the experiment’s results were discarded. Amines were purchased from Sigma Chemical. Electrophysiological Recording and Cell Identification Pyloric neuron activity was monitored by using pin electrodes to record extracellular APs from appropriate motor roots (differential AC amplifier, model 1700; A-M Systems) and intracellular electrodes (3 M KCl, 10 –15 M⍀) to record from cell bodies in the STG. We identified pyloric neuron somata during ongoing rhythmic activity by matching the extracellularly recorded APs with intracellularly recorded APs, by the characteristic shape and amplitude of membrane potential oscillations and APs of pyloric neurons, and by the known synaptic connections between pyloric neurons (Johnson et al. 2011). Isolation of Graded PD¡LP Synapse Inhibitory glutamatergic synapses in the pyloric network were blocked by 5 ⫻ 10⫺6M picrotoxin (PTX; Sigma Chemical) to pharmacologically isolate the PD¡LP synapse (Bidaut 1980). The anterior burster (AB) neuron, which is electrically coupled to the PD neurons, was photoinactivated by intracellular iontophoresis of 5,6-carboxyflourescein and illumination with bright blue light (Miller and Selverston 1979). This eliminated any amine-induced AB activity spreading to the PD neuron through the AB¡PD electrical synapse. We waited at least 1 h after AB photoinactivation to start our experiments, to allow the preparation to recover from the acute effects of the light exposure (Flamm and HarrisWarrick 1986).

2847

example, accurate measurements of the onset of synaptic depression (see below). Second, previous studies of graded synaptic transmission between pyloric neurons have often used square pulse stimulation (for example, Graubard et al. 1983; Johnson et al. 1995; Manor et al. 1997; Zhao et al. 2011), and we wanted to compare our results with these earlier studies and determine whether postsynaptic responses differed between the two stimulus forms. Both stimulation protocols used two-electrode voltage clamp to drive the PD from a holding potential (Vhold) of ⫺55 mV in increasing amplitudes of 5-mV increments up to ⫺25 mV. Stepping the PD up to ⫺25 mV evokes near-maximum gIPSPs in the LP neuron (Johnson and Harris-Warrick 1990; Rabbah and Nadim 2007). We separated stimulation runs by a minimum of 30 s, which was sufficient to eliminate all effects of the previous stimulation. We used two-electrode current clamp to hold the postsynaptic LP membrane potential at ⫺50 mV; postsynaptic electrodes were filled with 0.6 M K2SO4 ⫹ 0.02 M KCl and presynaptic electrodes with 3M KCl. In these experiments we used Axoclamp-2A and 2B amplifiers (Molecular Devices). Under control and amine conditions (at least 5 min of amine superfusion), we measured the initial peak and steady-state amplitudes of the gIPSPs to calculate input-output curves and dynamics of synaptic depression (see Fig. 1, A and B). The steady-state amplitude was measured as the gIPSP plateau level at the end of the square pulse stimulation or as the mean amplitude of the last five gIPSPs in response to the realistic PD waveform train. A synaptic depression index (DI) was calculated as the fractional gIPSP decrease from the initial peak response at steady state: DI ⫽ (initial peak amplitude ⫺ steady-state amplitude)/initial peak amplitude. Thus a larger DI value corresponds to greater synaptic depression. The time constant (␶Dep) of the gIPSP depression was measured during presynaptic square pulse stimulation by fitting the voltage decay from the initial peak to the steady-state amplitude with a single-exponential function (see Fig. 1A). The rate of synaptic depression was rapid enough that it could not be measured accurately from the separated peaks of the realistic waveform series. In a separate measurement, we quantified the time constant of recovery from synaptic depression (␶Rec) for both realistic and square waveform stimulation under control and amine conditions. We evoked steady-state depression by stimulating the PD neuron (Vhold ⫺55 mV) as described above, followed by a single square pulse or realistic oscillation test stimulation at increasing times after the end of the last conditioning stimulus. The percent recovery of the test pulse back to the initial peak value over time was fitted with a single exponential in each experiment to calculate ␶Rec (see Fig. 2, B and C). Data Acquisition and Analysis

Dynamics of PD¡LP Graded Synaptic Transmission LP graded inhibitory postsynaptic potentials (gIPSPs) were recorded after 10⫺7 M tetrodotoxin (TTX; Tocris) was added to Panulirus saline to block APs. We studied the PD¡LP graded chemical synapse using two different presynaptic stimulation protocols. The first used single, 3-s square pulses as voltage-clamp commands to drive the PD neuron, whereas the second used a train of 12 linked realistic PD waveforms as voltage-clamp commands to drive PD oscillations (see Johnson et al. 2005, 2011 for realistic waveform construction). Averaged and filtered PD waveforms from pyloric recordings had a mean amplitude of 15.5 ⫾ 3.5 mV and a mean cycle period of 688 ⫾ 199 ms (n ⫽ 10); we used this shape and cycle period as presynaptic stimuli. Although realistic presynaptic waveforms are more physiologically relevant than square pulse waveforms, we used both stimulation regimes for two reasons. First, some basic characteristics of synaptic depression and its modulation at the PD¡LP synapse can only be measured with square pulse stimulation, for

Electrophysiological recordings were digitized at 4 kHz using a PCI-6070-E board (National Instruments) and stored on a personal computer using custom-made recording software (Scope) written in LabWindows/CVI (National Instruments). This software also controlled the injection of oscillatory and square pulse waveforms as voltage-clamp commands into the PD neuron (Johnson et al. 2005, 2011). All data were analyzed with the related custom-made software program ReadScope, also written in LabWindows/CVI; these programs were kindly provided by Dr. F. Nadim (http://www.stg.rutgers. edu/software/index.html). Data were exported to Clampfit software (Molecular Devices) to calculate ␶Dep and ␶Rec. For statistical comparisons, we used JMP software (SAS Institute) to run multivariate analysis, followed by post hoc tests to determine the statistical significance of differences between individual data groups. Statistical differences between mean values were accepted with P ⬍ 0.05 (2-tailed probability) for F and t values. Mean measured values and percentages are means ⫾ SD.

J Neurophysiol • doi:10.1152/jn.00586.2012 • www.jn.org

2848

METAPLASTICITY AT A GRADED CHEMICAL SYNAPSE

RESULTS

izations; Fig. 1, B and D; see also Rabbah and Nadim 2007). Peak and steady-state gIPSP amplitudes increased with increasing PD voltage commands from ⫺40 to ⫺25 mV under both stimulation protocols (Fig. 1, C and D). The amplitude of the initial peak LP gIPSP tended to be larger when elicited with PD square pulse stimulation than with realistic waveform stimulation (compare Fig. 1, C and D, open and filled circles), and this difference was significant during PD depolarizations to ⫺30 and ⫺35 mV (P ⫽ 0.03 and 0.05, respectively). However, the mean steady-state gIPSP amplitudes were significantly greater when elicited with oscillations compared with square steps at strong PD depolarizations (to ⫺30 and ⫺25 mV; compare Fig. 1, C and D, open and filled squares; P ⫽ 0.05 and 0.03, respectively). Since square pulse stimulation evoked larger peak initial gIPSPs and smaller steady-state amplitudes than oscillation stimulation, it also evoked greater synaptic depression at all PD stimulation voltages (Fig. 1E; P ⬍ 0.001 at all PD voltages). For example, the mean DI was twice as large at PD depolariza-

Synaptic Dynamics of the PD¡LP Graded Synapse Under Control Conditions We first characterized the baseline synaptic dynamics of the PD¡LP graded synapse under control conditions, using single 3-s square pulses (n ⫽ 10) and repeated presynaptic activation with trains of 12 realistic oscillations (n ⫽ 7). As described previously for square pulse stimulation of the PD neuron (Graubard et al. 1983; Johnson and Harris-Warrick 1990; Rabbah and Nadim 2007), above the threshold for graded PD transmitter release (approximately ⫺50 mV), the LP response is a rapid initial peak hyperpolarization that rapidly depresses to a significantly smaller steady-state hyperpolarization (P ⬍ 0.001 across PD depolarizations; Fig. 1, A and C). The LP responds to a train of repeated PD oscillations with an initial peak gIPSP followed by subsequent gIPSPs whose amplitudes depress with repetition to a significantly smaller steady-state value over several oscillations (P ⬍ 0.001 across PD depolar-

A

B PD 10 mV

-55 mV -50 mV

LP

2 mV

1s

LP gIPSP (mV)

C

D

PD Square Pulse

8

6

PD Oscillation

8

6

4

4 initial peak

2

2

steady state

0

-45

-40

-35

-30

-25

PD Potential (mV)

E

steady state 2 mV 1s

initial peak

0

0.8

*

*

*

-30

-25

0.6 PD Oscillation 0.4

0.2

0 -40

-35

-45

-40

-35

-30

PD Potential (mV)

PD Square

* Depression Index

Fig. 1. Synaptic dynamics of graded synaptic transmission at the PD¡LP chemical synapse. A: lateral pyloric (LP) graded inhibitory postsynaptic potentials (gIPSPs) in response to 3-s pyloric dilator (PD) neuron square pulse depolarizations. Initial peak and steady-state IPSP amplitudes and the time constant for synaptic depression (␶Dep) were measured for all PD depolarizations from a holding potential (Vhold) of ⫺55 mV. B: LP gIPSPs during a train of 12 realistic PD oscillations of increasing amplitude from Vhold ⫽ ⫺55 mV. The first gIPSP amplitude (initial peak) and the mean of the last 5 gIPSPs amplitudes (steady state) were measured for all PD depolarizations. C: mean input/output relationship for the amplitude of the initial peak (open circles) and steady-state (open squares) gIPSPs during square pulse PD stimulation (n ⫽ 11). D: mean input/output relationship for initial (filled circles) and steady-state (filled squares) LP gIPSP amplitudes during PD oscillation stimulation (n ⫽ 7). E: mean depression index (DI) for square pulse (open bars) and oscillation (filled bars) stimulation across PD stimulation voltages. *Significantly greater DI for PD square pulse stimulation than for PD oscillation stimulation (P ⬍ 0.001 for all PD voltages). The DI did not show voltage dependence with either stimulation protocol (P ⬎ 0.20 for both).

τDep steady state

PD Potential (mV) J Neurophysiol • doi:10.1152/jn.00586.2012 • www.jn.org

-25


A

2849

B

gIPSP τDep (ms)

800

PD

600

10 mV -55 mV 400

-50 mV 200

LP

1 mV

τRec = 516 ms

0

-40

-35

-30

-25

1s

PD Potential (mV)

C

D 20 mV

-55 mV -50 mV

LP

τRec = 2900 ms

3000

gIPSP τRec (ms)

PD

*

2500 2000 1500 1000 500 0

1 mV

PD PD Square Oscillation

1s

tion to ⫺25 mV with the use of square waves compared with oscillations (Fig. 1E). This probably results from the greater depression during continued presynaptic stimulation than during trains of oscillations, where the synapse could partially recover between oscillations (see DISCUSSION). There was no significant effect of PD voltage on the magnitude of the depression index over the voltage range of either PD stimulation protocol (Fig. 1E; P ⫽ 0.23 and 0.21 for square pulse and oscillations, respectively). We also characterized the time constants of synaptic depression and recovery from depression at the PD¡LP graded synapse using the square pulse series. The decay of the LP response to PD stimulation (Fig. 1A) was well fit by a single exponential (fit correlation r ⫽ 0.99 ⫾ 0.001, n ⫽ 11). The mean time constant of synaptic depression, ␶Dep, was 400 ⫾ 200 ms during PD square pulse steps to ⫺25 mV. There was no

D PD 10 mV

-55 mV LP

Ctl

-50 mV

Wash

B

5 mV

C -50 mV

Ctl Wash DA

2 mV

Ctl

2 mV

Ctl

DA

4 2

PD Square

PD Oscillation

PD 10 mV

-55 mV -50 mV

1s

DA

6

E

DA

Ctl

voltage dependence of ␶Dep during square step stimulation (P ⫽ 0.95; Fig. 2A). Because of the comparatively slow cycle frequency of the oscillation stimulation (cycle period 688 ms), we were unable to calculate the faster ␶Dep from the oscillation data. In a separate measurement, we characterized the time constant of recovery from depression, ␶Rec, by using both stimulation protocols (PD steps to ⫺25 mV only) and fitting the gIPSPs at increasing poststimulation intervals with a single exponential (Fig. 2, B and C; square pulse: exponential fit correlation r ⫽ 0.95 ⫾ 0.03; oscillation: fit correlation r ⫽ 0.98 ⫾ 0.01). Recovery from synaptic depression was over twice as fast with square pulse PD stimulation (n ⫽ 13) as with oscillation stimulation (n ⫽ 12; Fig. 2D; P ⫽ 0.003). This may result from increased mobilization of the recovery process during maintained presynaptic depolarization (see DISCUSSION).

8

0

DA -50 mV

10

LP gIPSP (mV)

A

Fig. 2. Synaptic depression and recovery at the PD¡LP graded synapse. A: mean ␶Dep across PD square pulse stimulation voltages (n ⫽ 11). ␶Dep did not show any voltage dependence (P ⫽ 0.95). B: measurement of the time constant of LP gIPSP recovery from depression (␶Rec) in response to PD square pulse depolarization to ⫺25 mV. C: measurement of ␶Rec in response to PD oscillation depolarization to ⫺25 mV. D: mean ␶Rec for LP gIPSPs elicited by PD square pulse (n ⫽ 13) and oscillation stimulation (n ⫽ 10). *Significantly longer ␶Rec with PD oscillation stimulation (P ⫽ 0.003).

Fig. 3. Dopamine (DA; 10⫺5 M) modulation of synaptic strength at the PD¡LP synapse. A–C: examples showing variable effects of DA to reversibly enhance (A), reduce (B), or have no effect (C) on the initial LP gIPSP amplitude with PD square pulse stimulation. D: mean effects of DA on amplitudes of initial peak and steady-state LP gIPSPs elicited by PD square pulse (open bars, initial peak; light gray bars, steady state) and oscillation stimulation (solid bars, initial peak; dark gray bars, steady state) at ⫺25-mV PD depolarization. E: example showing DA enhancement of the initial peak gIPSP during PD oscillation stimulation. Ctl, control.

1 mV

LP

DA

1s



DA, 5-HT, and Oct affect synapses in the pyloric network by a complex set of pre- and postsynaptic actions (Harris-Warrick and Johnson 2010; Johnson and Harris-Warrick 1997). In addition, they can alter synaptic strength indirectly by modifying the waveforms of the pre- or postsynaptic neuronal oscillations (Johnson et al. 2005). To focus on synaptic metaplasticity, we attempted to limit the other consequences of amine modulation. For square pulse stimulations, we held the presynaptic neuron at ⫺50 mV by voltage clamp both before and during amine application. For oscillation stimulation, we used the realistic waveforms obtained from PD oscillations during the normal pyloric rhythm for both control and amine conditions. Thus the effects of amine-evoked changes in PD oscillations were eliminated. The postsynaptic LP neuron was held at ⫺55 mV under all conditions to prevent indirect effects of neuronal membrane potential changes evoked by the amines (Flamm and Harris-Warrick 1986). Dopamine. The physiological DA concentrations normally achieved in vivo will depend on the spike frequency of the dopaminergic neurons and are not known. We demonstrated previously that 10⫺4 M DA greatly reduces or abolishes the LP gIPSP in response to square pulse PD stimulation (Johnson and Harris-Warrick 1990). This also occurs using realistic waveform PD stimulation (data not shown), so we decided instead to use a lower concentration of 10⫺5 M DA, which does not eliminate PD¡LP transmission (Fig. 3), to examine DA’s effects on PD¡LP graded synaptic dynamics. At this lower concentration, DA had weak and highly variable effects on the LP gIPSPs in different preparations. With the use of square pulse PD stimulation, DA reversibly increased initial gIPSP amplitudes in about one-half of the preparations (Fig. 3A) and reduced or had little effect on gIPSP amplitudes in the other half (Fig. 3, B and C); all of these effects reversed upon washout of DA. We suggest that this variability is caused by the known opposing effects of DA to decrease presynaptic PD transmitter release and increase postsynaptic LP input resistance (Harris-Warrick and Johnson 2010), but to differing amounts in different preparations. On average, 10⫺5 M DA did not significantly change the mean amplitude of the peak and steady-state components of the LP gIPSP during PD square pulse stimulation (Fig. 3D, n ⫽ 5, P ⬎ 0.3 for both). In contrast, during oscillation stimulation, DA significantly increased the mean first gIPSP amplitude when compared across all PD depolarizations (example shown in Fig. 3E; P ⫽ 0.02, n ⫽ 16). However, this effect was weak and only produced a trend to increase initial peak gIPSP amplitude at single PD voltage step amplitudes (Fig. 3D; P ⫽ 0.08, n ⫽ 4 each). DA did not significantly change the steady-state gIPSP amplitude across PD oscillation steps (Fig. 3D; P ⫽ 0.26). With regard to synaptic depression, there was no significant effect of DA on the amplitude of synaptic depression, as measured by the DI, for either square pulses (control DI: 0.68 ⫾ 0.07, DA DI: 0.70 ⫾ 0.10 at ⫺25 mV) or oscillations (control DI: 0.32 ⫾ 0.06, DA DI: 0.37 ⫾ 0.17 at ⫺25 mV) at any PD voltage (P ⬎ 0.10 for both). Despite these weak and variable effects on the amplitudes of LP gIPSPs, DA consistently accelerated both the onset of, and the recovery from, synaptic depression. Figure 3C shows a

typical square wave measurement, where ␶Dep was accelerated by DA compared with the control gIPSP response, with little change in gIPSP amplitude. The mean ␶Dep was significantly decreased over the entire PD voltage range (Fig. 4A; n ⫽ 5, P ⬍ 0.005), but the DA effect showed no significant voltage dependence (P ⫽ 0.78). Recovery from depression was also faster with both square pulse and realistic waveform stimulation (Fig. 4A). This can be seen in the example using oscillation stimulation in Fig. 4B, which shows more complete recovery of

A gIPSP τ (ms)

Amines Change the Synaptic Dynamics of the PD¡LP Graded Synapse

2500

τDep

2000

τRec

τRec

1500

*

1000

* *

500 0

Ctl DA PD Square

Ctl DA PD Square

Ctl DA PD Oscillation

B PD 10 mV -55 mV -50 mV

Ctl 1 mV

LP -50 mV

DA 1 mV 1s

C

25 0

% Change τ

2850

-25 -50

τDep τRecSq τRecOsc

-75 -100 -50

0

50

100

150

% Change Initial Peak gIPSP Fig. 4. Dopamine acceleration of synaptic depression and recovery at the PD¡LP synapse. A: mean effects of DA on ␶Dep (left, open bars; n ⫽ 5) and ␶Rec (middle, open bars; n ⫽ 5) during PD square pulse stimulation and on ␶Rec during PD oscillation stimulation (right, filled bars; n ⫽ 4). *Significantly shorter ␶ in DA compared with control conditions (P ⬍ 0.005 for all). B: example showing more rapid recovery of gIPSP amplitude during DA after a 1-s interval following PD oscillation stimuli and no effect of DA on the initial peak gIPSP. Gray horizontal lines mark the amplitude of the initial peak gIPSP. C: lack of correlation between DA effects on initial peak LP gIPSP and its effects on depression and recovery in individual experiments. ␶RecSq, ␶Rec for PD square pulse stimulation; ␶RecOsc, ␶Rec for PD oscillation stimulation.



the depressed gIPSP 1 s after the stimulus train during DA (89% recovery) than under control conditions (72% recovery). The mean ␶Rec at PD depolarization to ⫺25 mV was significantly faster in DA for both square pulse PD stimulation (Fig. 4A, 39 ⫾ 23% ␶Rec decrease; P ⬍ 0.001, n ⫽ 5) and realistic waveform PD stimulation (Fig. 4A, 43 ⫾ 19% ␶Rec decrease; P ⫽ 0.03, n ⫽ 4). This acceleration of onset and recovery from depression did not depend on the sign of the rather variable DA effect on gIPSP amplitudes described above. As summarized in Fig. 4C, in all but one experiment the ␶Dep and ␶Rec values were smaller while DA either increased or decreased gIPSP amplitude. Thus, although DA (10⫺5 M) had highly variable effects on gIPSP amplitude, it had reliable, significant, and independent effects to accelerate the time course of synaptic depression onset and recovery from synaptic depression. Octopamine. We previously demonstrated that 10⫺5 M Oct strengthens the PD¡LP synapse when tested with square wave PD stimulation (Johnson and Harris-Warrick 1990). In our current experiments, Oct consistently increased the amplitude of gIPSPs over the full range of presynaptic voltages using both square pulse (n ⫽ 5) and realistic (n ⫽ 4) PD waveforms (Fig. 5, A and B). The peak and steady-state responses to square pulse stimulation were significantly larger during Oct application at all PD depolarizations (P ⬍ 0.0001 and P ⫽ 0.004, respectively): during a PD step to ⫺25 mV, the mean peak gIPSP increased by 55 ⫾ 29% whereas the steady-state gIPSP increased by 157 ⫾ 32% (Fig. 5C). As shown in Fig. 5C, Oct increased the steady-state gIPSP amplitude significantly more than the initial peak amplitude across the voltage steps (P ⫽ 0.04). With the oscillation protocol, Oct also increased the first peak and steady-state gIPSP amplitudes across all PD stimulation voltages (P ⬍ 0.0001 for both); for example, at ⫺25 mV, the peak amplitudes increased by 57 ⫾ 33% whereas the steady-state amplitudes increased by 48 ⫾ 37% (Fig. 5C). Unlike the results using square pulse stimulation, during oscillation stimulation there was no significant

A

2851

difference between Oct’s enhancement of the peak and the steady-state gIPSP amplitudes at any PD voltage (P ⫽ 0.15; Fig. 5C). There was no voltage dependence of the Oct effect on gIPSP amplitude. During square pulse stimulation, the greater Oct enhancement of the steady state over the peak gIPSP amplitude resulted in a significant decrease in the DI across PD depolarizations (16 ⫾ 6% decrease at ⫺25 mV, P ⫽ 0.0001; Fig. 5D). However, there was no equivalent DI decrease during oscillation stimulation, because Oct did not differentially increase the first peak and steady-state amplitudes of the gIPSP (Fig. 5, C and D; P ⫽ 0.15). Octopamine significantly altered the onset and recovery of synaptic depression in response to square pulse PD stimulation in a way that is quite different from DA. Octopamine significantly slowed ␶Dep across PD steps (increase by 80 ⫾ 41% at ⫺25 mV; n ⫽ 5, P ⬍ 0.0001; Fig. 6A). This slowing of synaptic depression was not caused simply by the larger gIPSPs evoked during Oct. When gIPSP amplitudes were matched in the presence and absence of Oct with the use of different amplitude presynaptic PD steps in the same preparation, ␶Dep remained slower during Oct (Fig. 6B). In contrast, Oct significantly accelerated the recovery from depression during square pulse PD stimulation (Fig. 6C) by an average of 47 ⫾ 24% using ⫺25-mV steps (n ⫽ 4, P ⫽ 0.026; Fig. 6A). However, when measured with oscillation stimulation, Oct had highly variable effects on ␶Rec that could either accelerate or slow down and were not significant overall (control: 1,120 ⫾ 610 ms, Oct: 1,510 ⫾ 1,510 ms; P ⫽ 0.63). Thus the main effect of Oct was to increase PD¡LP synaptic strength during both PD stimulation protocols. With the use of square pulse PD stimulation, Oct reduced the depression amplitude and slowed the onset of synaptic depression while speeding up the rate of recovery from depression. Serotonin. We previously showed that 10⫺5 M 5-HT weakens the PD¡LP synapse when tested with square wave PD

B

PD 10 mV

-55 mV

10 mV

Ctl

-50 mV

5 mV

Oct

LP

5 mV

1s

Oct

D

C 12

Ctl

Oct

Ctl

Oct

* #

*

9

6

0.8

Ctl

Oct

Ctl

Oct

*

*

3

0

Depression Index

15

LP gIPSP (mV)

1s

0.6

*

0.4

0.2

Fig. 5. Octopamine (10⫺5 M) enhances synaptic strength and reduces synaptic depression at the PD¡LP synapse. A and B: examples showing Oct enhancement of initial peak and steady-state gIPSP amplitudes during PD square pulse (A) and oscillation stimulation (B). Black traces, control conditions; gray traces, Oct condition. C: Oct enhances mean absolute amplitudes of initial peak and steady-state LP gIPSPs elicited by PD square pulse (open bars, initial peak; light gray bars, steady state; n ⫽ 5) and oscillation stimulation (filled bars, initial peak; dark gray bars, steady state; n ⫽ 4) at ⫺25-mV PD depolarization. *Significant increase (P ⬍ 0.004 for all). #Significantly less Oct enhancement of the initial peak gIPSP amplitude than the steady-state gIPSP amplitude during PD square pulse stimulation (P ⫽ 0.04). D: Oct significantly decreases synaptic depression with PD square pulse but not with oscillation stimulation at ⫺25-mV PD depolarization. *Significant reduction (P ⬍ 0.0001).

0

PD Square

PD Oscillation

PD Square

PD Oscillation


2852


A gIPSP τ (ms)

1200

τDep

τRec

900

* 600

*

300

0

Ctl Oct PD Square

Ctl Oct PD Square

B

Ctl Oct

PD

10 mV

-55 mV -50 mV LP

Ctl Oct

Serotonin significantly accelerated the time course of synaptic depression measured with square pulse PD stimulation and tended to accelerate the time course of recovery from depression (Fig. 8A), although this effect did not reach statistical significance. The acceleration in ␶Dep was not caused by a smaller gIPSP during 5-HT. For example, the trace in Fig. 8B shows that when sized matched (using different presynaptic voltage steps), the gIPSP recorded during 5-HT depressed more quickly than in control conditions. Serotonin significantly accelerated ␶Dep with square pulses across PD voltages (P ⫽ 0.02); at ⫺25-mV PD depolarization, ␶Dep was reduced by 25 ⫾ 34% (Fig. 8A; n ⫽ 4). In 3 of 3 experiments, 5-HT decreased ␶Rec by 28 ⫾ 14%, but this did not reach statistical significance with the small sample size and variability (P ⫽ 0.10; Fig. 8A). Serotonin also decreased ␶Rec in 5 of 5 experiments by 26 ⫾ 27%, measured by oscillation stimulation to ⫺25 mV (see example in Fig. 8C), but this also did not reach our criterion for statistical significance (P ⫽ 0.07; Fig. 8A). Thus 5-HT reduces PD¡LP synaptic strength and increases the rate of onset of synaptic depression and the amount of synaptic depression while tending to weakly accelerate the recovery from depression.

5 mV

1s

DISCUSSION

C PD 10 mV

-55 mV -50 mV

1 mV LP 1s

1 mV

Fig. 6. Octopamine modulation of synaptic depression and recovery at the PD¡LP synapse. A: Oct increases ␶Dep (left, n ⫽ 5) and decreases ␶Rec (right, n ⫽ 4) at ⫺25-mV PD square pulse depolarization. *Significant effects (P ⬍ 0.03 for both). B: example using PD square pulse depolarizations of differing amplitudes to adjust the LP gIPSP to the same amplitude under control and Oct conditions, showing the Oct-induced increase in ␶Dep in size-matched LP gIPSPs (black traces, control; gray trace, Oct). C: example of normalized initial peak gIPSPs in control (black traces) and Oct (gray traces), with more rapid gIPSP recovery in Oct during PD square pulse depolarizations to ⫺25 mV.

stimulation (Johnson and Harris-Warrick 1990). In our recent experiments, we reproduced these results using both square pulse (n ⫽ 4) and oscillation (n ⫽ 6) PD waveforms (Fig. 7, A and B). With the use of square pulse stimulation, the mean initial peak and steady-state responses were consistently and significantly smaller during 5-HT application at all PD depolarizations (P ⬍ 0.0001 for initial peak and P ⫽ 0.004 for steady state). This effect did not show any voltage dependence: with ⫺25-mV steps, the peak gIPSPs fell by 28 ⫾ 17% and the steady-state gIPSPs fell by 40 ⫾ 12% (Fig. 7C). Similar results were seen during oscillation PD stimulation at all voltages (P ⫽ 0.0035 and P ⬍ 0.0001 for initial peak and steady-state responses, respectively): at ⫺25-mV PD depolarization, the peak gIPSPs fell by 20 ⫾ 12% and the steady-state gIPSPs fell by 27 ⫾ 10% (Fig. 7C). During 5-HT application, synaptic depression was significantly greater across all PD depolarizations during square pulse (12 ⫾ 8% greater at ⫺25 mV; P ⫽ 0.0015; Fig. 7D) and oscillation stimulation (32 ⫾ 38% greater at ⫺25 mV; P ⫽ 0.004; Fig. 7D).

Amine-Induced Metaplasticity at the PD¡LP Graded Synapse Graded chemical synapses of the pyloric and gastric CPG networks of the lobster STG show marked synaptic depression with prolonged voltage steps or repeated trains of presynaptic activation; all pyloric synapses are tonically depressed at normal pyloric cycle frequencies (reviewed by Hartline and Graubard 1992; see also Johnson et al. 2005, 2011; Mamiya and Nadim 2004; Manor et al. 1997). Synaptic depression is commonly seen at graded chemical synapses, such as those between retinal neurons and from hair cells to afferent fibers (Edmonds et al. 2004; Wan and Heidelberger 2011), although not all graded chemical synapses show short-term activitydependent changes in synaptic strength (Narayan et al. 2011; Simmons 1981). Aside from studies with pyloric network synapses, little previous work has examined neuromodulation of synaptic dynamics at graded chemical synapses. We have previously described amine effects on the strength of the PD¡LP graded synapse (Johnson and Harris-Warrick 1990), and here we demonstrate that the magnitude and kinetics of short-term synaptic dynamics are also modulated by amines at this synapse. The most surprising conclusion from our results is that the synaptic strength and the magnitude of synaptic depression, as well as the kinetics of its onset and recovery, can be independently modulated by each amine. We suggest that this may occur through multiple pre- and postsynaptic mechanisms of neuromodulatory actions that shape synaptic strength in a functioning network. Dopamine. We have previously demonstrated that DA (10⫺4 M) increases graded synaptic strength at the LP¡PY and LP¡PD synapses, but only changes (increases) the amount of synaptic depression at the LP¡PY synapse (Johnson et al. 2005, 2011). Thus DA can have differential effects on synaptic depression and synaptic strength depending on the postsynaptic target. We used 10⫺5M DA in the present study because



A

2853

B PD 10 mV

-55 mV

10 mV

5HT -50 mV 2 mV

LP

Ctl

5 mV

Ctl

1s

C

1s

D 8

Ctl

5HT

Ctl

5HT

0.8

Ctl

5HT

Ctl

5HT

6

*

*

4

* 2

*

0

Depression Index

LP gIPSP (mV)

* 0.6

*

0.4

Fig. 7. Serotonin (5-HT; 10⫺5 M) decreases synaptic strength and enhances synaptic depression at the PD¡LP synapse. A and B: examples showing 5-HT reduction of initial peak and steady-state gIPSP amplitudes during PD square pulse (A) and oscillation stimulation (B). Black traces, control conditions; gray traces, 5-HT. C: 5-HT reduces the amplitudes of initial peak and steady-state LP gIPSPs elicited by PD square pulse (open bars, initial peak; light gray bars, steady state; n ⫽ 4) and oscillation stimulation (filled bars, initial peak; dark gray bars, steady state; n ⫽ 6) at ⫺25-mV PD depolarization. *Significant at P ⬍ 0.005 for both stimulation protocols. D: 5-HT increases synaptic depression. *Significant at P ⬍ 0.004 for both stimulation protocols.

0.2

0

PD Square

PD Oscillation

PD Square

10⫺4M DA abolishes functional synaptic transmission at the PD¡LP graded synapse (Johnson and Harris-Warrick 1990). At the lower concentration, DA consistently accelerated ␶Dep and ␶Rec despite small and variable effects on PD¡LP synaptic strength and on the magnitude of synaptic depression. This demonstrates that the kinetics of synaptic depression can be modulated independently of synaptic strength and the magnitude of synaptic depression. At present, we cannot correlate 10⫺5 M DA’s variable actions on synaptic strength with its effects on other cellular properties. None of the known effects of DA on pre- or postsynaptic ionic currents described previously (Harris-Warrick and Johnson 2010) can account for the DA-induced changes in depression kinetics. DA may be acting postsynaptically to accelerate the kinetics of transmitter receptor desensitization and recovery from desensitization (Papke et al. 2011). Motor patterns produced by the lobster pyloric network are qualitatively different when treated with 10⫺4 and 10⫺5 M DA (Flamm and Harris-Warrick 1986). Concentrationdependent DA effects on the onset and recovery of synaptic depression may contribute to shaping distinct DA-induced network activity. Octopamine. Octopamine’s enhancement of synaptic strength was accompanied by reduced synaptic depression, slowed ␶Dep, and accelerated ␶Rec for PD square pulse stimulation but not for PD oscillation stimulation. Part of the synaptic strength increase by Oct may be due to its increase in postsynaptic LP input resistance (Johnson et al. 1993), but the change in synaptic depression suggests more dynamic effects on synaptic transmission. Action potential-evoked synaptic transmission at crustacean and vertebrate central synapses depresses more at synapses with high transmitter output than with low transmitter output and depresses less with low transmitter output (Millar and Atwood 2004; Thomson 2000). Reduced synaptic depression coupled with increased synaptic strength during Oct application does not fit this pattern, which would directly link transmitter release probability with short-term synaptic dynam-

PD Oscillation

ics. Instead, at the PD¡LP graded synapse, enhancement of presynaptic calcium currents by Oct could lead to both greater transmitter release and more rapid mobilization of vesicles into the readily releasable pool, thus leading to reduced depression and accelerated recovery from depression (Babai et al. 2010; Gomis et al. 1999; see below). However, the effects of Oct on PD ionic currents are not known. Serotonin. Serotonin reduced PD¡LP graded synaptic strength but increased synaptic depression; again, this does not fit a pattern directly linking synaptic dynamics with the amount of transmitter release. Serotonin’s effects could reflect presynaptic mechanisms such as a reduction in voltage-dependent calcium currents, which would reduce release and slow the rate of vesicle mobilization. Serotonin may have additional postsynaptic effects besides decreasing LP input resistance (Johnson et al. 1993), as described above for DA. Serotonin showed a trend to accelerate both the onset and recovery from depression, which would allow greater flexibility of the synapse as the cycle frequency changes; the accelerated recovery may limit the reduction in synaptic strength during the normal oscillating activity of the neurons in the pyloric rhythm. More work is needed to clarify the mechanisms for the metaplastic effects of amines, but it is clear that amines have complex and sometimes opposing effects on short-term dynamics of graded synaptic transmission in the pyloric network. Characteristics of Synaptic Dynamics at the PD¡LP Graded Synapse The use of both square pulse and oscillation stimulation protocols allowed us to probe characteristics of synaptic depression and its modulation at the PD¡LP graded synapse of the pyloric that might not be evident with either stimulation protocol alone. First, we found that the magnitude of the synaptic depression was greater with square pulse stimulation of PD than with trains of PD oscillations, emphasizing that


2854


A

2400

τDep

τRec

τRec

gIPSP τ (ms)

1800

1200

600

0

* Ctl 5HT PD Square

Ctl 5HT PD Square

Ctl 5HT PD Oscillation

B 5HT PD

Ctl

10 mV

-55 mV -50 mV LP

5HT 2 mV

Ctl 1s

C PD

10 mV -55 mV -50 mV

Ctl

1 mV

LP

5HT

1 mV 1s

Fig. 8. Serotonin modulation of kinetics of synaptic depression and recovery at the PD¡LP synapse. A: 5-HT reduces ␶Dep (left, open bars; n ⫽ 4) and has only weak effects on ␶Rec (middle, open bars; n ⫽ 3) during ⫺25-mV PD square pulse depolarization and on ␶Rec (right, closed bars; n ⫽ 5) during ⫺25-mV PD oscillation depolarization. *Significant effect (P ⫽ 0.002). B: example with PD square pulse depolarizations adjusted in control and 5-HT conditions to show the 5-HT-induced acceleration of ␶Dep in size-matched LP gIPSPs (black traces, control; gray traces, 5-HT). C: example showing normalized initial peak gIPSPs in control (black traces) and 5-HT (gray traces) and more rapid gIPSP recovery in 5-HT 1 s after the end of PD oscillations. Gray horizontal lines mark the amplitudes of initial peak gIPSPs.

realistic oscillation stimulation allows a more physiological characterization of pyloric synapses (Manor et al. 1997). Weaker synaptic depression with realistic oscillation stimulation may be explained simply by the partial recovery from depression that occurs between the PD oscillations. Second, we could more accurately determine the time course of onset of depression with long square pulse PD stimulation. The onset of depression, measured with square pulse PD stimulation, had a time constant of ⬃400 ms and could not be accurately measured with oscillations occurring every 688 ms. Values for ␶Dep vary from milliseconds to tens of seconds at other AP-evoked and graded chemical synapses (Cho et al. 2011; Wang and

Manis 2008; Zucker and Regehr 2002). The recovery from depression was over twice as fast with square pulse PD stimulation (␶Rec ⬃600 ms) as with PD oscillations (␶Rec ⬃1,600 ms). Our ␶Rec times are longer than those previously reported by Rabbah and Nadim (2007) at the PD¡LP graded synapse, but the stimulation protocols differed between the two studies. Time constants for recovery from depression range from tens of milliseconds at hair cell synapses to seconds at other graded synapses, again depending on experimental conditions (Rabl et al. 2006; Wan and Heidelberger 2011). Finally, amines sometimes affected synaptic depression and its onset and recovery kinetics differentially (see DISCUSSION above), suggesting more complex actions on synaptic dynamics than would be seen with PD oscillation stimulation alone. The most common mechanism of synaptic depression at both AP-evoked and graded chemical synapses is transmitter depletion during prolonged or repeated synaptic activity (von Gersdorff and Matthews 1997; Zucker and Regher 2002); other reported presynaptic mechanisms include calcium channel inactivation (Xu et al. 2007), autoreceptor activation (Davies et al. 1990), and accumulating refractoriness of the transmitter release mechanism (Waldeck et al. 2000). Our experiments were not designed to determine the site of depression, but they suggest a mixed mechanism for depression at the PD¡LP graded synapse. Changing the length of the presynaptic square pulse or the duration of the PD oscillations altered the level of depression (Rabbah and Nadim 2007). In addition, the faster ␶Rec using square pulse PD stimulation, which would accumulate more intracellular calcium, is consistent with a calciumdependent replenishment of the readily releasable transmitter pool, seen at AP-evoked (Neher and Sakaba 2008) and graded synapses (Babai et al. 2010; Gomis et al. 1999). We also found that the magnitude of PD¡LP depression and its onset kinetics were not sensitive to presynaptic voltage stimulation, suggesting that depression was independent of the amount of transmitter release. The presynaptic voltage independence of both the depression magnitude and ␶Dep is inconsistent with a simple presynaptic transmitter depletion model of synaptic depression. The lack of correlation between the amount of synaptic depression and gIPSP amplitude at the PD¡LP graded synapse is also seen at AP-evoked synapses from other preparations where synaptic depression is clearly due to a presynaptic mechanism but its magnitude is independent of PSP amplitude (Hefft et al. 2002; Wu et al. 2005). There may also be a postsynaptic contribution at our synapse, such as receptor desensitization (Jonas and Westbrook 1995; Papke et al. 2011; Trussell et al. 1993). Possible Functional Importance of Metaplasticity in the Pyloric Network Short-term synaptic depression plays an important role in organizing a flexible pyloric motor pattern. The magnitude of depression and its onset and recovery kinetics dynamically determine the synaptic strength at different pyloric cycle periods and when pyloric rhythm periods are perturbed by interactions with linked networks such as those from the cardiac sac and gastric mill motor networks (Ayali and Harris-Warrick 1998; Russell and Hartline 1981; Thurma and Hooper 2002, 2003). Synaptic depression promotes phase maintenance as the pyloric network changes its cycle frequency (Greenburg and



Manor 2005; Manor et al. 2003; Nadim et al. 2003) and contributes to stabilization of the rhythm period (Mamiya and Nadim 2004). It can enable a switch between different forms of network bistability (Manor and Nadim 2001; Nadim et al. 1999) and can even determine the sign of a synaptic interaction (Johnson et al. 2005; Mamiya et al. 2003). Neuromodulator-induced metaplasticity could help maintain or reset synaptic strengths between network neurons to adjust the postsynaptic neurons’ firing phase. For example, the 10⫺5 M DA effects to accelerate both ␶Dep and ␶Rec with no change in gIPSP amplitude or depression could stabilize PD¡LP synaptic strength as the cycle frequency changes in DA (Johnson et al. 2011). Octopamine enhances LP excitability but does not change the LP firing onset phase (Johnson et al. 2011). The greater PD¡LP synaptic strength in Oct, with reduced depression and faster onset and recovery kinetics, could balance enhanced LP excitability to maintain LP phasing at its pre-Oct values (Johnson et al. 2011). There is also no change in LP onset firing phase during 5-HT despite a reduction in LP excitability (Johnson et al. 2011). A faster ␶Dep might contribute to weakening the PD¡LP gIPSP and to enhanced depression, and this may be mitigated by the accelerated ␶Rec to maintain the LP firing onset phase. At the crab LP¡PY graded synapse, the peptide proctolin changes the sign of synaptic plasticity from depression to facilitation at certain presynaptic voltage levels and functions to help stabilize the cycle period (Zhao et al. 2011). Metaplasticity through the actions of many different neuromodulators is a common feature of many neural networks (Barriere et al. 2008; Carey et al. 2011; Kreitzer and Regehr 2000; Parker 2001; Sakurai and Katz 2009) and should be considered one of the building blocks of network operation. Although well studied at AP-evoked synapses, there is relatively little known about metamodulation at graded chemical synapses. The pyloric network of crustaceans provides an excellent model system to explore the ranges of expression of metaplasticity at graded synapses and its functional role in organizing network operation and plasticity (Johnson et al. 2005, 2011; Zhao et al. 2011). ACKNOWLEDGMENTS We thank Françoise Vermeylen from the Cornell Statistical Consulting Unit for help with the statistical analysis. Present address of M. D. Kvarta: Department of Physiology, Program in Neuroscience, Medical Scientist Training Program, University of Maryland School of Medicine, 655 W. Baltimore Street, Baltimore, MD 21201. GRANTS This research was supported by National Institute of Neurological Disorders and Stroke Grant NS17323 (to R. M. Harris-Warrick). DISCLOSURES No conflicts of interest, financial or otherwise, are declared by the authors. AUTHOR CONTRIBUTIONS M.D.K., R.M.H.-W., and B.R.J. conception and design of research; M.D.K. performed experiments; M.D.K. and B.R.J. analyzed data; M.D.K., R.M.H.W., and B.R.J. interpreted results of experiments; M.D.K. and B.R.J. prepared figures; M.D.K., R.M.H.-W., and B.R.J. drafted manuscript; M.D.K., R.M.H.W., and B.R.J. edited and revised manuscript; M.D.K., R.M.H.-W., and B.R.J. approved final version of manuscript.

2855

REFERENCES Abbott LF, Regehr WG. Synaptic computation. Nature 431: 796 – 803, 2004. Ayali A, Harris-Warrick RM. Interaction of dopamine and cardiac sac modulatory inputs on the pyloric network in the lobster stomatogastric ganglion. Brain Res 794: 155–161, 1998. Babai N, Bartoletti TM, Thoreson WB. Calcium regulates vesicle replenishment at the cone ribbon synapse. J Neurosci 30: 15866 –15877, 2010. Barriere G, Tartas M, Cazalets JR, Bertrand SS. Interplay between neuromodulator-induced switching of short-term plasticity at sensorimotor synapses in the neonatal rat spinal cord. J Physiol 586: 1903–1920, 2008. Bidaut M. Pharmacological dissection of the pyloric network of the lobster stomatogastric ganglion using picrotoxin. J Neurophysiol 44: 1089 –1101, 1980. Carey MR, Myoga MH, McDaniels KR, Marsicano G, Lutz B, Mackie K, Regehr WG. Presynaptic CB1 receptors regulate synaptic plasticity at cerebellar parallel fiber synapses. J Neurophysiol 105: 958 –963, 2011. Chacron MJ, Toporikova N, Fortune ES. Differences in the time course of short-term depression across receptive fields are correlated with directional sensitivity in electrosensory neurons. J Neurophysiol 102: 3270 –3279, 2009. Cho S, Li GL, von Gersdorff H. Recovery from short-term depression and facilitation is ultrafast and Ca2⫹ dependent at auditory hair cell synapses. J Neurosci 31: 5682–5692, 2011. Davies CH, Davies SN, Collingridge GL. Paired pulse depression of monosynaptic GABA-mediated inhibitory postsynaptic responses in rat hippocampus. J Physiol 424: 513–531, 1990. Edmonds BW, Gregory FD, Schweizer FE. Evidence that fast exocytosis can be predominantly mediated by vesicles not docked at active zones in frog saccular hair cells. J Physiol 560: 439 – 450, 2004. Fioravante D, Regehr WG. Short-term forms of presynaptic plasticity. Curr Opin Neurobiol 21: 269 –274, 2011. Fisher SA, Fischer TM, Carew TJ. Multiple overlapping processes underlying short-term synaptic enhancement. Trends Neurosci 20: 170 –177, 1997. Fischer TM, Blazis DE, Priver NA, Carew TJ. Metaplasticity at identified inhibitory synapses in Aplysia. Nature 389: 860 – 865, 1997. Flamm RE, Harris-Warrick RM. Aminergic modulation in lobster stomatogastric ganglion. II. Target neurons of dopamine, octopamine and serotonin within the pyloric circuit. J Neurophysiol 55: 847– 865, 1986. Fortune ES, Rose GJ. Short-term synaptic plasticity as a temporal filter. Trends Neurosci 24: 381–385, 2001. Gelman S, Grove CL, Faber DS. Atypical properties of release and shortterm depression at a specialized nicotinic synapse in the Mauthner cell network. J Exp Biol 214: 1560 –1570, 2011. Gomis A, Burrone J, Lagnado L. Two actions of calcium regulate the supply of releasable vesicles at the ribbon synapse of retinal bipolar cells. J Neurosci 19: 6309 – 6317, 1999. Goutman JD, Glowatzki E. Short-term facilitation modulates size and timing of the synaptic response at the inner hair cell ribbon synapse. J Neurosci 31: 7974 –7981, 2011. Graubard K, Raper JA, Hartline DA. Graded synaptic transmission between identified spiking neurons. J Neurophysiol 50: 508 –520, 1983. Greenberg I, Manor Y. Synaptic depression in conjunction with A-current channels promote phase constancy in a rhythmic network. J Neurophysiol 93: 656 – 677, 2005. Harris-Warrick RM, Johnson BR. Checks and balances in neuromodulation. Front Behav Neurosci 4: 47, 2010. Hartline DK, Graubard K. Cellular and synaptic properties in the crustacean stomatogastric nervous system. In: Dynamic Biological Networks: the Stomatogastric Nervous System, edited by Harris-Warrick RM, Marder E, Selverston AI, and Moulins M. Cambridge, MA: MIT Press, 1992, p. 31– 85. Hartline DK, Russell DF, Raper JA, Graubard K. Special cellular and synaptic mechanisms in motor pattern generation. Comp Biochem Physiol C 91: 115–131, 1988. Hefft S, Kraushaar U, Geiger JR, Jonas P. Presynaptic short term depression is maintained during regulation of transmitter release at a GABAergic synapse in rat hippocampus. J Physiol 539: 201–208, 2002. Jackman SL, Choi SY, Thoreson WB, Rabl K, Bartoletti TM, Kramer RH. Role of the synaptic ribbon in transmitting the cone light response. Nat Neurosci 12: 303–310, 2009. Johnson BR, Brown JM, Kvarta MD, Lu JYJ, Schneider LR, Nadim F, Harris-Warrick RM. Differential modulation of synaptic strength and


2856


timing regulate synaptic efficacy in a motor network. J Neurophysiol 105: 293–304, 2011. Johnson BR, Harris-Warrick RM. Amine modulation of graded synaptic transmission in the lobster stomatogastric ganglion. J Neurosci 10: 2066 – 2076, 1990. Johnson BR, Harris-Warrick RM. Amine modulation of glutamate responses from pyloric motor neurons in lobster stomatogastric ganglion. J Neurophysiol 78: 3210 –3221, 1997. Johnson BR, Peck JH, Harris-Warrick RM. Temperature sensitivity of graded synaptic transmission in the lobster stomatogastric ganglion. J Exp Biol 156: 267–285, 1991. Johnson BR, Peck JH, Harris-Warrick RM. Amine modulation of electrical coupling in the pyloric network of the lobster stomatogastric ganglion. J Comp Physiol A 172: 715–732, 1993. Johnson BR, Peck JH, Harris-Warrick RM. Distributed amine modulation of graded chemical transmission in the pyloric network of the lobster stomatogastric ganglion. J Neurophysiol 74: 437– 452, 1995. Johnson BR, Schneider LR, Nadim F, Harris-Warrick RM. Dopamine modulation of phasing of activity in a rhythmic motor network: contribution of synaptic and intrinsic modulatory actions. J Neurophysiol 94: 3101–3111, 2005. Jonas MV, Westbrook GL. Desensitized states prolong the GABAA channel responses to brief agonist pulses. Neuron 15: 181–191, 1995. Klug A. Short-term synaptic plasticity in the auditory brain stem by using in-vivo-like stimulation parameters. Hear Res 279: 51–59, 2011. Kreitzer AC, Regehr WG. Modulation of transmission during trains at a cerebellar synapse. J Neurosci 20: 1348 –1357, 2000. Li WC, Sautois B, Roberts A, Soffe SR. Reconfiguration of a vertebrate motor network: specific neuron recruitment and context-dependent synaptic plasticity. J Neurosci 27: 12267–12276, 2007. Mamiya A, Manor Y, Nadim F. Short-term dynamics of a mixed chemical and electrical synapse in a rhythmic network. J Neurosci 23: 9557–9564, 2003. Mamiya A, Nadim F. Dynamic interaction of oscillatory neurons coupled with reciprocally inhibitory synapses acts to stabilize the rhythm period. J Neurosci 24: 5140 –5150, 2004. Manor Y, Bose A, Booth V, Nadim F. Contribution of synaptic depression to phase maintenance in a model rhythmic network. J Neurophysiol 90: 3513–3528, 2003. Manor Y, Nadim F. Synaptic depression mediates bistability in neuronal networks with recurrent inhibitory connectivity. J Neurosci 21: 9460 –9470, 2001. Manor Y, Nadim F, Abbott LF, Marder E. Temporal dynamics of graded synaptic transmission in the lobster stomatogastic ganglion. J Neurosci 17: 5610 –5621, 1997. Millar AG, Atwood HL. Crustacean phasic and tonic motor neurons. Integr Comp Biol 44: 4 –13, 2004. Miller JP, Selverston AI. Rapid killing of single neurons by irradiation of intracellularly injected dye. Science 206: 702–704, 1979. Mouser C, Nadim F, Bose A. Maintaining phase of the crustacean tri-phasic pyloric rhythm. J Math Biol 57: 161–181, 2008. Mulloney B, Selverston AI. Organization of the stomatogastric ganglion of the spiny lobster. I. Neurons driving the lateral teeth. J Comp Physiol 91: 1–32, 1974. Nadim F, Booth V, Bose A, Manor Y. Short-term synaptic dynamics promote phase maintenance in multi-phase rhythms. Neurocomputing 52–54: 79 – 87, 2003. Nadim F, Manor Y. The role of short-term synaptic dynamics in motor control. Curr Opin Neurobiol 10: 683– 690, 2000. Nadim F, Manor Y, Kopell N, Marder E. Synaptic depression creates a switch that controls the frequency of an oscillatory circuit. Proc Natl Acad Sci USA 96: 8206 – 8211, 1999. Narayan A, Laurent G, Sternberg PW. Transfer characteristics of a thermosensory synapse in Caenorhabditis elegans. Proc Natl Acad Sci USA 108: 9667–9672, 2011.

Neher E, Sakaba T. Multiple roles of calcium ions in the regulation of neurotransmitter release. Neuron 59: 861– 872, 2008. Papke D, Gonzalez-Gutierrez G, Grosman C. Desensitization of neurotransmitter-gated ion channels during high-frequency stimulation: a comparative study of Cys-loop, AMPA and purinergic receptors. J Physiol 589: 1571– 1585, 2011. Parker D. Spinal-cord plasticity: independent and interactive effects of neuromodulator and activity-dependent plasticity. Mol Neurobiol 22: 55– 80, 2001. Parker D, Grillner S. Activity-dependent metaplasticity of inhibitory and excitatory synaptic transmission in the lamprey spinal cord locomotor network. J Neurosci 19: 1647–1656, 1999. Parker D, Grillner S. The activity-dependent plasticity of segmental and intersegmental synaptic connections in the lamprey spinal cord. Eur J Neurosci 12: 2135–2146, 2000. Philpot BD, Bear MF, Abraham WC. Metaplasticity: the plasticity of synaptic plasticity. In: Beyond Neurotransmission: Neuromodulation and Its Importance for Information Processing, edited by Katz PS. New York: Oxford University Press, 1999, p 160 –197. Qian SM, Delaney KR. Neuromodulation of activity-dependent synaptic enhancement at crayfish neuromuscular junction. Brain Res 771: 259 –270, 1997. Rabbah P, Nadim F. Distinct synaptic dynamics of heterogeneous pacemaker neurons in an oscillatory network. J Neurophysiol 97: 2239 –2253, 2007. Rabl K, Cadetti L, Thoreson WB. Paired-pulse depression at photoreceptor synapses. J Neurosci 26: 2555–2563, 2006. Russell DF, Hartline DK. A multiaction synapse evoking both EPSPs and enhancement of endogenous bursting. Brain Res 223: 19 –38, 1981. Sakurai A, Katz PS. State-, timing-, and pattern-dependent neuromodulation of synaptic strength by a serotonergic interneuron. J Neurosci 29: 268 –279, 2009. Sanchez JAD, Kirk MA. Ingestion motor programs of Aplysia are modulated by short-term synaptic enhancement in cerebral-buccal interneuron pathways. Invert Neurosci 4: 199 –212, 2002. Simmons PJ. Synaptic transmission between second- and third-order neurons of a locust ocellus. J Comp Physiol A 145: 265–276, 1981. Thomson AM. Molecular frequency filters at central synapses. Prog Neurobiol 62: 159 –196, 2000. Thuma JB, Hooper SL. Quantification of gastric mill network effects on a movement related parameter of pyloric network output in the lobster. J Neurophysiol 87: 2372–2384, 2002. Thuma JB, Hooper SL. Quantification of cardiac sac network effects on a movement-related parameter of pyloric network output in the lobster. J Neurophysiol 89: 745–753, 2003. Trussell LO, Zhang S, Raman IM. Desensitization of AMPA receptors upon multiquantal neurotransmitter release. Neuron 10: 1185–1196, 1993. von Gersdorff H, Matthews G. Depletion and replenishment of vesicle pools at a ribbon-type synaptic terminal. J Neurosci 17: 1919 –1927, 1997. Waldeck RF, Pereda A, Faber DS. Properties and plasticity of paired-pulse depression at a central synapse. J Neurosci 20: 5312–5320, 2000. Wan QF, Heidelberger R. Synaptic release at mammalian bipolar cell terminals. Vis Neurosci 28: 109 –119, 2011. Wang Y, Manis PB. Short-term synaptic depression and recovery at the mature mammalian endbulb of Held synapse in mice. J Neurophysiol 100: 1255–1264, 2008. Wu Y, Kawasaki F, Ordway RW. Properties of short-term synaptic depression at larval neuromuscular synapses in wild-type and temperature-sensitive paralytic mutants of Drosophila. J Neurophysiol 93: 2396 –2405, 2005. Xu J, He L, Wu LG. Role of Ca2⫹ channels in short-term synaptic plasticity. Curr Opin Neurobiol 17: 352–359, 2007. Zhao Z, Sheibanie AF, Oh M, Rabbah P, Nadim F. Peptide neuromodulation of synaptic dynamics in an oscillatory network. J Neurosci 31: 13991– 14004, 2011. Zucker RS, Regehr WG. Short-term synaptic plasticity. Annu Rev Physiol 64: 355– 405, 2002.
