Synaptic strength is determined by release probability and the size of the readily releasable pool of docked vesicles. Here we describe the effects of blocking myosin light chain kinase (MLCK), a cytoskeletal regulatory protein thought to be involved in myosin-mediated vesicle transport, on synaptic transmission at the mouse calyx of Held synapse. Application of three different MLCK inhibitors increased the amplitude of the early excitatory postsynaptic currents (EPSCs) in a stimulus train, without affecting the late steady-state EPSCs. A presynaptic locus of action for MLCK inhibitors was confirmed by an increase in the frequency of miniature EPSCs that left their average amplitude unchanged. MLCK inhibition did not affect presynaptic Ca2+ currents or action potential waveform. Moreover, Ca2+ imaging experiments showed that [Ca2+]i transients elicited by 100-Hz stimulus trains were not altered by MLCK inhibition. Studies using high-frequency stimulus trains indicated that MLCK inhibitors increase vesicle pool size, but do not significantly alter release probability. Accordingly, when AMPA-receptor desensitization was minimized, EPSC paired-pulse ratios were unaltered by MLCK inhibition, suggesting that release probability remains unaltered. MLCK inhibition potentiated EPSCs even when presynaptic Ca2+ buffering was greatly enhanced by treating slices with EGTA-AM. In addition, MLCK inhibition did not affect the rate of recovery from short-term depression. Finally, developmental studies revealed that EPSC potentiation by MLCK inhibition starts at postnatal day 5 (P5) and remains strong during synaptic maturation up to P18. Overall, our data suggest that MLCK plays a crucial role in determining the size of the pool of synaptic vesicles that undergo fast release at a CNS synapse.
Synaptic communication depends on the reliable delivery of synaptic vesicles to the active zone and their subsequent priming for exocytosis. Delivery of synaptic vesicles is speculated to occur via an actin-myosin–mediated vesicle transport system. Thus actin-based myosin motor proteins may be involved in the process that forms the readily releasable pool of docked vesicles (Doussau and Augustine 2000; Evans et al. 1998; Prekeris and Terrian 1997; Shupliakov et al. 2002; Sudhof 2004). Distinct functional compartments in neurons such as dendritic spines and presynaptic terminals seem to lack abundant microtubules, but are rich in actin, and some studies have indicated that actin participates in a regulatory mechanism that restrains fusion of vesicles at the active zone (Dillon and Goda 2005; Morales et al. 2000). At the calyx of Held, actin is involved in synaptic vesicle recruitment (Sakaba and Neher 2003).
Myosins are versatile motor proteins localized in neuronal somata, axons, dendrites, and presynaptic terminals (Bridgman 2004; Jian et al. 1996). In actin-myosin–mediated vesicle transport, myosins attach to synaptic vesicles and drag them along actin filaments. Myosin may be necessary to sustain transmission during high-frequency firing by quickly translocating reserve vesicles to the active zone (Polo-Parada et al. 2001). Among different isoforms of myosin, myosin II and V were reported in presynaptic nerve terminals and are speculated to mobilize reserve vesicle pools toward release sites (Takagishi et al. 2005).
On the other hand, myosin itself is regulated by several kinases and phosphatases (Casaletti et al. 2003; Luo 2002). Among them myosin light chain kinase (MLCK) phosphorylates the regulatory light chain of myosin and has been suggested to be involved in neurotransmitter release (Mochida et al. 1994, 1995; Polo-Parada et al. 2001; Ryan 1999), N-methyl-d-aspartate receptor activity (Lei et al. 2001), potassium channel function (Akasu et al. 1993), and calcium release from IP3-sensitive stores (Norwood et al. 2000). In presynaptic transmission, MLCK inhibition was proposed as a blocker of vesicle mobilization in cultured hippocampal neurons (Jordan et al. 2005; Ryan 1999) and to significantly reduce glutamate release from hippocampal synaptosomes (Perkeris and Terrian 1997). By contrast, Tokuoka and Goda (2006) recently demonstrated that MLCK is not a regulator of presynaptic vesicle trafficking during high-frequency stimulation in cultured hippocampal neurons. Unfortunately, MLCK knockout is lethal for mice (Somlyo et al. 2004) and therefore not useful for functional studies. Thus the role of MLCK at different CNS synapses still remains unclear.
Here we studied the involvement of MLCK in regulating glutamate release from a CNS nerve terminal, the calyx of Held (Kandler and Friauf 1993). To do so, we assessed the effects of specific membrane-permeable MLCK inhibitors on synaptic transmission in acute mouse brain stem slices. At this synapse there are two distinct pools of vesicles: a fast releasing pool and a slowly releasing, or reluctant, pool (Sakaba and Neher 2001a,b). We found that MLCK inhibitors potentiated the early excitatory postsynaptic currents (EPSCs) in a stimulus train by preferentially increasing the size of the pool of fast releasing vesicles.
All experiments were performed according to protocols approved by the Oregon Health and Science University Animal Care and Use Committee in accordance with National Institutes of Health guidelines. Acute brain stem slices were prepared from postnatal day 3 (P3) to P18 C57BL/6J mice pups (Charles River Laboratories, Wilmington, MA). After decapitation, the brain stem was quickly immersed in ice-cold low-calcium artificial cerebrospinal fluid (aCSF) containing (in mM): 125 NaCl, 2.5 KCl, 3 MgCl2, 0.1 CaCl2, 25 glucose, 25 NaHCO3, 1.25 NaH2PO4, 0.4 ascorbic acid, 3 myo-inositol, and 2 Na-pyruvate, pH 7.4–7.5 when bubbled with carbogen (95% O2-5% CO2) and osmolarity of 310–320 mOsm. Transverse slices of the auditory brain stem were cut at a thickness of 200–300 μm, using a vibratome slicer (VT1000; Leica, Bannockburn, IL), and incubated at 37°C for 1 h in normal aCSF and thereafter kept at room temperature (22–24°C) for experiments. The normal aCSF was the same as the low-calcium aCSF except that 1 mM MgCl2 and 2 mM CaCl2 were used.
After incubation, slices were transferred to a 1-ml chamber perfused with normal aCSF at the rate of 1.5–2 ml/min. Medial nucleus of the trapezoid body (MNTB) neurons were viewed using an Olympus microscope equipped with infrared light (gradient contrast optics) and a ×40 water-immersion objective. The pipette internal solution for postsynaptic recordings contained the following (in mM): 130 K-gluconate, 20 KCl, 5 Na2-phosphocreatine, 10 HEPES, 5 EGTA, 4 Mg-ATP, and 0.5 GTP (pH adjusted to 7.3 with KOH). In addition, 1 mM QX-314 was used in some P15–P18 postsynaptic recordings to block Na currents. However, the effects of MLCK inhibition were identical in both conditions. Pipette solution for presynaptic action potentials (APs) contained the following (in mM): 130 K-gluconate, 20 KCl, 5 Na2-phosphocreatine, 10 HEPES, 0.2/0.5 EGTA, 4 Mg-ATP, and 0.3/0.5 GTP (pH adjusted to 7.3 with KOH). The same internal solution was used for measuring presynaptic Na+ and K+ currents. Pipette solution for presynaptic recordings of Ca2+ currents contained the following (in mM): 130 Cs-gluconate, 15 CsCl, 5 Na2-phosphocreatine, 10 HEPES, 0.2 EGTA, 20 tetraethylammonium chloride (TEA), 4 Mg-ATP, and 0.3 GTP (pH adjusted to 7.3 with CsOH). In addition, for presynaptic Ca2+ current recordings, TEA (5 mM), 4-aminopyridine (4-AP, 0.3 mM), tetrodotoxin (TTX, 1 μM), strychnine (0.5 μM), bicuculline (10 μM), CNQX (10 μM), and 2-amino-5-phosphonovaleric acid (APV, 50 μM) were added to the bath aCSF. To isolate α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA)–receptor-mediated miniature EPSCs (mEPSCs), strychnine (0.5 μM), bicuculine (10 μM), and APV (50 μM) were added to the bath aCSF. In addition, TTX (1 μM) was used in some cases, although results were identical with or without its presence. Drugs were dissolved in extracellular solution and, if not, in DMSO such that the final concentration was <0.01%, except for EGTA-AM experiments where the final concentration is 0.1%. Results are expressed as means ± SE (except as noted). The significance of differences among data sets was evaluated by Student's paired two-tailed t-test, unless otherwise indicated.
Recording pipettes were pulled from borosilicate glass capillaries (World Precision Instruments, Sarasota, FL) with a Sutter P-97 electrode puller (Sutter Instruments, Novato, CA) and had open tip resistances of 2.0–3.0 MΩ for postsynaptic recordings and 4.0–6.0 MΩ for presynaptic recordings. Access resistance (Rs) was ≤9 MΩ for postsynaptic recordings and ≤20 MΩ for presynaptic recordings. Rs was compensated >90% for postsynaptic recordings and about 50% for presynaptic recordings. Principal cells had a holding potential of −70 mV and presynaptic terminals were held at −80 mV. Afferent fibers were stimulated with a bipolar platinum/iridium electrode (FHC, Bowdoinham, ME) placed near the midline spanning the afferent fiber tract of the MNTB. An Iso-Flex stimulator driven by a Master 8 pulse generator (A.M.P.I., Jerusalem, Israel) was used to deliver step pulses (100 μs, <15 V DC). Data were acquired at 10- to 25-μs sampling rate using an EPC-9 amplifier (HEKA Elektronik, Lambrecht, Germany) controlled by Pulse 8.4 software and filtered on-line at 2.9 kHz. Data were stored in a Power Macintosh G3 computer (Apple Computers, Cupertino, CA). Data were analyzed off-line and presented using Igor Pro (WaveMetrics, Lake Oswego, OR). All traces for kinetic analysis and display were corrected off-line for series resistance and holding potential errors (Schneggenburger et al. 1999).
Presynaptic Ca2+ imaging
Fura-2 (30–50 μM) and Fura-4F (100 μM) were added to the K-gluconate–based presynaptic internal solution to image intracellular calcium ion concentration ([Ca2+]i) transients during presynaptic whole cell recordings. The dyes were excited at 350- and 380-nm light generated by a Polychrome V monochrometer (TILL Photonics, Gräfelfing, Germany). A 410-nm short-pass excitation filter, a 410-nm long-pass dichroic mirror, and a 440-nm long-pass emission filter (TILL Photonics) were used to select the appropriate wavelengths for excitation and to collect fluorescent signals. Fluorescence was measured using an IMAGO-QE charge-coupled device camera (TILL Photonics). Pixel binning was 8 × 8 pixels (on chip binning). Pairs of images at each wavelength were taken, allowing 10-ms exposure time. Ca2+ responses were analyzed by calculating the ratio of images obtained with 350-nm excitation over images obtained with 380-nm excitation. Fluorescence signals were background subtracted and expressed in ratio values (350/380 ratio). Fluorescence signals were obtained with or without the patch pipette in whole cell mode and these data produced similar results in experiments with MLCK inhibitors, indicating that the recording mode did not influence the conclusions of our experiments.
Drugs and reagents
All salts, kynurenic acid, and 2,3-butanedione monoxime (BDM) were purchased from Sigma (St. Louis, MO). MLCK inhibitors such as 1-(5-chloronaphthalene-1-sulfonyl) homopiperazine HCl (ML-9), 1-(5-iodonaphthalene-1-sulfonyl) homopiperazine HCl (ML-7), and wortmannin were purchased from Calbiochem (EMD Biosciences, San Diego, CA). EGTA-AM, fura-2, and fura-4F were purchased from Molecular Probes (Eugene, OR). The remaining pharmacological drugs were purchased from Tocris Bioscience (Ellisville, MO).
Inhibition of MLCK potentiates early EPSCs
To study the role of myosin light chain kinase (MLCK) on fast glutamatergic synaptic transmission at the calyx of Held synapse, we used ML-9, a selective inhibitor of MLCK. ML-9 is known to permeate the cell membrane and to act on intracellular MLCK (Polo-Parada et al. 2001; Ryan 1999; Saitoh et al. 1987). We made whole cell patch-clamp recordings from mice MNTB principal cells. The MLCK inhibitor ML-9 (25 μM) was bath applied and the effects on evoked EPSCs were measured on afferent fiber stimulation. Figure 1A shows EPSCs from a postnatal day P13 MNTB neuron before (black traces) and after ML-9 application (red) with 100-Hz trains (10 stimuli) given every 30 s. Data were acquired continuously for every 30 s. During the 100-Hz train, the EPSCs strongly depressed (von Gersdorff and Borst 2002), but the amplitude of early EPSCs (i.e., the first two or three in the train) increased on 2–4 min of ML-9 application. The first average EPSC amplitude in the presence of ML-9 increased by 54 ± 10% (n = 5 cells; P11–P13; Fig. 1B) and this increase was partially reversed during washout of the drug. To check for any time-dependent run-up effect on the EPSCs amplitude, we recorded EPSCs under control conditions (n = 4 cells; P11–P13) for about 15 min with 100-Hz-stimulation trains (10 stimuli) given at every 30 s. As shown in Fig. 1B the EPSC amplitudes fluctuated but remained unchanged on average during the control recordings (without any drug). We also continuously monitored the compensated series resistance and found no significant change in this parameter before and during ML-9 application (Fig. 1C).
Similar experiments were also done with afferent fiber stimulation at a constant low rate of 0.1 Hz (data not shown). During application of ML-9 for 10 min EPSCs gradually increased in amplitude and this increase was partially reversed during washout of the drug. The mean EPSC amplitude in the presence of ML-9 for 10 min increased by 54 ± 6% of the mean control EPSC amplitude (n = 7 cells; P10–P15).
We also compared the effect of ML-9 with other distinct MLCK inhibitors such as ML-7 and wortmannin (P10–P13; Fig. 2). To do so, we measured the amplitude of 100-Hz train EPSCs (50 stimuli; 30-s sweep interval; average of 5 sweeps) before and after 5 min of MLCK-inhibitor application. Application of ML-9 significantly increased the amplitude of the 1st and 2nd EPSCs from 2.5 ± 0.4 to 4.5 ± 0.9 nA (n = 8 cells; P = 0.005) and from 2.0 ± 0.2 to 2.5 ± 0.3 nA, respectively (P = 0.007; Fig. 2, A2 and A3). On application of ML-7 (15 μM), another specific MLCK inhibitor (Lei et al. 2001; Saitoh et al. 1987), the early EPSCs (i.e., 1st, 2nd, and 3rd) significantly increased from 2.3 ± 0.7 to 3.8 ± 1.0 nA (n = 6 cells; P = 0.011), from 2.4 ± 0.7 to 3.2 ± 0.8 nA (P = 0.003), and from 1.9 ± 0.5 to 2.3 ± 0.5 nA (P = 0.01), respectively (Fig. 2, B2 and B3). Moreover, wortmannin (5 μM), another MLCK inhibitor (Mochida et al. 1994; Nakanishi et al. 1992), also increased the amplitude of the 1st and 2nd EPSCs from 4.0 ± 0.7 to 7.4 ± 1.3 nA (n = 5 cells; P = 0.007) and from 3.5 ± 0.5 to 4.4 ± 0.7 nA (P = 0.05), respectively (Fig. 2, C2 and C3). Furthermore, we checked whether there was any effect of MLCK inhibitors on the steady-state EPSCs (mean of 48th to 50th EPSCs). On application of ML-9, ML-7, and wortmannin the amplitude of the steady-state EPSCs changed from 0.3 ± 0.06 to 0.2 ± 0.06 nA (P = 0.28), 0.3 ± 0.07 to 0.3 ± 0.08 nA (P = 0.7), and 0.8 ± 0.1 to 0.6 ± 0.1 nA (P = 0.1), respectively, although this was not significant (Fig. 2, A4, B4, and C4). Taken together, these observations indicate that three distinct MLCK inhibitors potentiated the early EPSCs in a stimulus train without significantly changing the steady-state EPSCs.
We also tested the effect of H7, a nonselective inhibitor of protein kinases [i.e., at 100 μM it is reported to inhibit protein kinase A (PKA), protein kinase C (PKC), and calmodulin-dependent kinase II], that reduces EPSC amplitudes by 15% at the end bulb of Held synapse (Oleskevich and Walmsley 2000). However, at the calyx of Held synapse application of H7 (100 μM) did not significantly change the amplitude of the EPSCs. The mean amplitude of the 1st EPSC in control was 3.7 ± 1.2 nA and on H7 application 3.8 ± 1.1 nA (n = 4 cells; P = 0.64).
Action of MLCK inhibition: a presynaptic locus
To determine the locus of the MLCK-inhibition effects, we studied the effect of ML-9 (25 μM) on spontaneous miniature EPSCs (mEPSCs) from P10–P14 neurons. Application of ML-9 for 5–8 min significantly increased the mean frequency of mEPSCs by 30.5 ± 14% of the control rate (n = 11 cells; P = 0.003; Fig. 3, A and B), whereas the mean amplitude of the mEPSCs under control and drug application displayed no significant difference. The mean amplitude of the mEPSCs under control condition and drug application was 50 ± 6 pA (2,300 minis) and 52 ± 8 pA (3,000 minis), respectively (n = 11 cells; P = 0.7). Similarly, the charge of the mEPSCs was 38 ± 3.3 fC in control condition and 37 ± 3.6 fC during drug application, which was not significantly different (n = 11 cells; P = 0.8; Fig. 3, C and D). These results indicate that the locus of action of MLCK inhibition is presynaptic.
No effect of MLCK inhibition on presynaptic AP or ionic conductances
Given that MLCK inhibition modulates a presynaptic mechanism, we next examined whether it affected presynaptic action potential (AP) waveforms.
We thus performed simultaneous pre- and postsynaptic paired recordings in the calyx synapse to directly compare the MLCK inhibition on presynaptic APs and postsynaptic EPSCs. To do so, we delivered afferent fiber stimulation (n = 5 pairs) and also DC injection (5-ms pulses; 300–500 pA; n = 2 pairs) to the calyx and measured the corresponding principal cell EPSCs. The presynaptic and postsynaptic recording pipettes had 0.2 and 5 mM EGTA, respectively. In these whole cell recording conditions, the early EPSCs were also potentiated by 64 ± 15% (n = 5 pairs) on ML-9 application, irrespective of low- or high-frequency stimulation (Fig. 4, A and B). During the short current injection experiments, a similar potentiation of EPSCs on ML-9 application was observed (data not shown). Note that ML-9 did not significantly change the waveform of the presynaptic APs (Fig. 4, A and B). In addition, the effect of ML-9 is present even when the calyx is continuously dialyzed with 0.2 mM EGTA, a nearly physiological level of presynaptic Ca buffer (Borst and Sakmann 1996). Thus the whole cell recording mode does not disrupt the effects of ML-9.
We also determined whether ML-9 affects presynaptic ionic conductances. Therefore we measured the calyx K+ and Ca2+ currents before and after 5 min of ML-9 application (25 μM) in P10–P14 rats. The presynaptic calyx displayed rapidly activating inward Na+ currents, followed by outward K+ currents in response to depolarizing voltage steps, which started from a potential of −90 to +30 mV (Forsythe 1994; Leão et al. 2005; Nakamura et al. 2007). Typical examples of these currents are shown in Fig. 5A1, under control conditions and in the presence of ML-9. ML-9 application had no effect on the presynaptic voltage-dependent K+ (both transient and steady-state) currents (n = 6 calyces). Finally, we checked whether there was any effect of ML-9 on the presynaptic Ca2+ currents (evoked by 2-ms depolarizing steps to 0 mV). An example is shown in Fig. 5B, revealing that ML-9 application had no effect on the Ca2+ current (n = 5 calyces). We conclude that MLCK inhibition leaves the presynaptic ionic conductances undisturbed.
Paired-pulse ratios: effects of MLCK inhibition
To gain insight on whether the effects of MLCK inhibition are due to a change in release probability (Pr), we calculated the paired-pulse ratio (EPSC2/EPSC1; 10-ms interpulse interval) of two consecutive EPSCs. Up to P13, MLCK inhibition significantly decreased the paired-pulse ratio (PPR) (Fig. 6A). For the range of P9–P13 ML-9 application significantly reduced the PPR (n = 8 cells; P = 0.0001). Similarly for P9–P13, the MLCK inhibitors ML-7 (n = 5 cells; P = 0.04) and wortmannin (n = 5 cells; P = 0.01) also reduced the PPR.
These results thus may suggest that MLCK inhibition changes Pr. However, at these younger ages, AMPA-receptor saturation and desensitization play a major role in short-term depression at this synapse, whereas at older ages (e.g., P14–P18) these factors do not contribute greatly to depression (Renden et al. 2005; Taschenberger et al. 2002). We thus reduced release probability by using low external Ca2+ ([Ca2+]e) Ringers and we also reduced AMPA-receptor desensitization/saturation with 1 mM kynurenic acid (KYN), a low-affinity AMPA-receptor antagonist (Wong et al. 2003). For P9–P13 mice (Fig. 6B), in 1 mM Ca2+, or in 1 mM Ca2+ plus 1 mM KYN, or in 0.65 mM Ca2+ the PPRs were now not significantly different (1 mM Ca2+: n = 5 cells, P = 0.5; 1 mM Ca2+ plus 1 mM KYN: n = 4 cells, P = 0.5; 0.65 mM Ca2+: n = 5 cells, P = 0.5). The potentiation of the 1st EPSC in P9–P13 mice with ML-9 was 80 ± 36% with 2 mM Ca2+, 67 ± 23% with 1 mM Ca2+, and 74 ± 32% with 1 mM Ca2+ plus 1 mM KYN. Thus under all these conditions the potentiation of the 1st EPSC remains fairly constant without any significant differences.
In more mature mice (P15–P17; Fig. 6C), MLCK inhibition did not significantly change the PPR in normal 2 mM Ca2+ (n = 6 cells; P = 0.3). Likewise, in low [Ca2+]e or 1 mM Ca2+ plus 1 mM KYN, the PPR was not significantly different on MLCK inhibition (P15–P17; 1 mM Ca2+: n = 4 cells, P = 0.8; and with 1 mM Ca2+ plus 1 mM KYN; n = 4 cells, P = 0.7). So the observed effect of MLCK inhibitors on PPRs at younger ages is likely due to increased AMPA-receptor desensitization. In conclusion, these results suggest that the MLCK inhibition is not likely to alter Pr.
MLCK inhibition increases the vesicle pool size: low Pr conditions
To determine the effect of MLCK inhibitors on vesicle pool size we studied its effect on high-frequency trains of stimulation. To reduce AMPA-receptor desensitization/saturation effects we first exchanged the normal aCSF with 1 mM Ca2+. This decreased Pr and reduced EPSC amplitudes (Fig. 7). In these experiments, low Ca2+ was substituted by an equimolar amount of Mg2+. We then bath-applied ML-9 (25 μM) and observed the effects on evoked EPSCs. Figure 7A shows recordings from a P13 MNTB neuron before (black trace) and after (red) ML-9 application with 100-Hz train (50 stimuli) stimulation. Irrespective of the ML-9 application, during a 100-Hz train, under 1 mM [Ca2+]e the EPSCs initially show facilitation followed by a depression (Forsythe et al. 1998; Schneggenburger et al. 1999). The mean amplitude of the first EPSCs significantly increased on ML-9 application (control: 0.45 ± 0.1 nA; ML-9: 0.7 ± 0.1 nA; n = 5 cells; P = 0.009), without any significant change in the steady-state EPSCs (control: 0.3 ± 0.06 nA; ML-9: 0.3 ± 0.05 nA; P = 0.7). Note that in low [Ca2+]e ML-9 can increase the first 10 to 20 EPSCs in the train. This dramatic effect suggests that potentiation occurs until a fast initial pool of vesicles is depleted.
To check whether the potentiation of early EPSCs on MLCK inhibition is due to changes in release probability and/or the size of the readily releasable pool (RRP) of vesicles, we estimated these parameters (Fig. 7B1). The total amount of transmitter released during a train of high-frequency stimuli can represent the RRP size provided that the pool is depleted and is not too quickly replenished (Habets and Borst 2007). Assuming that the depression is primarily caused by RRP depletion (i.e., not by desensitization/saturation), and also that a constant steady-state fraction remains due to an immediate replenishment, the RRP size can be estimated from the zero time y-intercept of a line fitted to a cumulative amplitude plot of EPSCs (Kaneko and Takahashi 2004; Schneggenburger et al. 1999). Using this method we estimated the pool size under control conditions and on MLCK inhibition for 1 mM [Ca2+]e (P9–P13; n = 5 cells). The cumulative EPSC plots are shown in Fig. 7B2 and the linear fit over the last points during the train was back-extrapolated to yield an estimate for Nq (where N is the number of readily releasable vesicles and q is the mean quantal size), which is proportional to the initial pool size. Our estimated Nq values under both control conditions and under MLCK inhibition were, respectively, 3.7 ± 0.5 and 5.7 ± 0.5 nA (P = 0.005), indicating that Nq significantly increased on MLCK inhibition (Fig. 7, B2 and B3). Release probability (Pr) was estimated by dividing the amplitude of the first EPSC by the pool size (Kaneko and Takahashi 2004; Schneggenburger et al. 1999). The release probability (Pr) was not significantly changed. The value in control and on MLCK inhibition was 0.12 ± 0.02 and 0.13 ± 0.02, respectively (P = 0.6; Fig. 7B4). These results suggest that MLCK inhibition primarily increases the RRP size (Nq increased by ∼52%) rather than Pr in low release probability conditions. Similar experiments with P9–P13 mice in 2 mM [Ca2+]e showed only a 35% increase in pool size (data not shown; n = 7 cells) and this difference is probably due to AMPA-receptor desensitization/saturation masking the real amount of glutamate release (Wong et al. 2003; Wu and Borst 1999). We also did similar experiments with P9–P13 mice in 1 mM [Ca2+]e and 1 mM kynurenic acid, which showed a 46% increase in Nq (n = 4 cells, data not shown).
MLCK inhibition increases the vesicle pool size: older calyces
Our results with P9–P13 calyces suggest an increase in RRP on ML-9 application, leaving the Pr value unchanged. We next confirmed these results with older mice (P15–P16) and with standard 2 mM [Ca2+]e. The EPSC amplitudes recorded under both control and MLCK inhibition reached a plateau value as shown in Fig. 8A1 after about 200 ms. The RRP and Pr were estimated from the cumulative EPSC plot by the same high-frequency (100-Hz) stimulation method as described earlier. The Nq value (Fig. 8, A2 and A3; y-intercept of the linear fit) significantly increased from 20 ± 3 to 29 ± 5 nA (n = 5 cells; P = 0.03) on ML-9 application. In contrast, the Pr value (Fig. 8A4) did not change significantly (control: 0.2 ± 0.02; ML-9: 0.25 ± 0.03; P = 0.06). Together these results again suggest that MLCK inhibition increases the Nq (by ∼45%) but not the Pr. We did similar experiments in 1 mM [Ca2+]e and 1 mM kynurenic acid with P15–P16 mice, which showed a 66% increase in Nq (n = 4 cells, data not shown).
Estimates of RRP size (or Nq) depend on a complete and rapid depletion of the pool, as well as an absence of AMPA-receptor desensitization/saturation. A stimulation rate of 100 Hz may, however, not be sufficiently fast to deplete the entire pool (Sakaba 2006; Wu and Borst 1999). We thus reconfirmed these results by performing experiments (Fig. 8B1) at 4 mM [Ca2+]e (i.e., at higher Pr conditions) and also with 300-Hz train stimulation (50 stimuli). The RRP should now be depleted more quickly and completely. In addition, we used 2 mM gamma-d-glutamylglycine (AMPA-receptor antagonist) to guarantee a block of AMPA-receptor saturation/desensitization even though, at these older ages (P17–P18), desensitization is greatly reduced (Renden et al. 2005). The EPSC amplitudes during the 300-Hz train under both control and MLCK inhibition now reached a plateau value as shown in Fig. 8B1. The RRP (or Nq) and the Pr were estimated by the same method as described earlier. Note, however, that the slope of the linear fit is now significantly shallower than previously obtained (Fig. 8B2), suggesting a more complete and rapid depletion of the RRP. The Nq value (Fig. 8B3) significantly increased from 9.03 ± 2.4 to 11.6 ± 2.7 nA (n = 5 cells; P = 0.03) on ML-9 application. In contrast, the Pr value (Fig. 8B4) did not change significantly (control: 0.2 ± 0.02; ML-9: 0.2 ± 0.01; P = 1). This result strongly confirms our basic finding that MLCK inhibition increases RRP size (by ∼30%) but not Pr, and this occurs even for more mature P15–P18 synapses.
Effect of MLCK inhibition under enhanced Ca2+ buffering
The increase in mEPSC frequency on MLCK inhibition could be due to an elevation of residual basal Ca2+ in the presynaptic terminal. This putative increase in basal [Ca2+]i could also facilitate the EPSC peak amplitude. We thus tested the effect of MLCK inhibition in the presence of a membrane-permeant Ca2+ buffer: EGTA-AM (Korogod et al. 2005; Oleskevich and Walmsley 2002; Turecek and Trussell 2001; Wang and Kaczmarek 1998). The effect of MLCK inhibition on evoked EPSCs (100 Hz; 50 stimuli) was examined in slices (P13–P16) perfused for 15 min with 200 μM EGTA-AM, followed by wash for 5 min to stabilize the EPSC amplitudes (Supplemental Fig. S1, top).1 The pipette solution for these postsynaptic recordings contained 5 mM EGTA, so that the effects on transmission observed with EGTA-AM were probably due solely to changes in the presynaptic buffering. EGTA-AM decreased the amplitude of the EPSCs, indicating that the Ca2+ sensors responsible for transmitter release had less access to Ca2+ ions. EGTA-AM treatment decreased the amplitude of the early EPSCs (mean of 1st to 3rd) by 61 ± 5% and the steady-state EPSCs (mean of 48th to 50th) by 53 ± 10%. To get stable EPSC amplitudes after EGTA-AM application, the slices were washed for 5 min in normal aCSF before ML-9 application.
Under these enhanced Ca2+ buffering conditions, ML-9 still potentiated the early EPSCs, whereas the late EPSCs remained unchanged in the presence of ML-9. The early EPSCs (mean of 1st to 3rd EPSCs) under control conditions had an average amplitude of 6.6 ± 0.07 nA (Supplemental Fig. S1A; n = 5 cells) and decreased during EGTA-AM treatment. They then reached a baseline in wash of 4.0 ± 0.32 nA and on ML-9 application the EPSC significantly increased to 5.2 ± 0.35 nA (P = 0.034). Like early EPSCs, the amplitude of steady-state EPSCs showed a decrease on EGTA-AM treatment. The steady-state EPSCs amplitude (mean of 48th to 50th EPSCs) in control amounted to 0.2 ± 0.002 nA. However, ML-9 application revealed no significant change (P = 0.2) compared with the EPSC amplitude after EGTA-AM treatment, i.e., in wash 0.10 ± 0.02 and 0.12 ± 0.02 nA in ML-9 application (Supplemental Fig. S1B). In conclusion, increased Ca2+ buffering does not block the potentiating action of ML-9.
To test whether there is any potentiation (or fast run-up) of EPSCs during the wash period (i.e., after EGTA-AM application) and in the absence of ML-9, we did some control experiments. After EGTA-AM treatment, the slices were continuously washed for 15 min in normal aCSF. We compared the mean amplitude of EPSCs during the first 5-min wash period with that after the 10-min wash in normal aCSF (time at which we saw the peak effect of ML-9; Supplemental Fig. S1, A and B, bottom). The mean amplitude of early EPSCs during first 5-min wash amounted to 2.8 ± 0.3 nA and after 10-min wash it was 2.9 ± 0.4 nA (n = 5 cells; P = 0.2). The mean amplitude of steady-state EPSCs during the first 5-min wash was 0.075 ± 0.006 nA and after the 10-min wash it was 0.08 ± 0.007 nA (P = 0.07). Therefore there was no significant, time-dependent change in the early or in the steady-state EPSCs during the 15-min wash after EGTA-AM treatment. These results thus suggest that an enhancement of Ca2+ buffering with the slow Ca2+ buffer EGTA did not eliminate the action of MLCK inhibition.
Presynaptic [Ca2+]i imaging and ML-9
To directly observe the effects of ML-9 on presynaptic [Ca2+]i we performed Ca2+ imaging experiments. The calyx of Held was loaded with internal solution containing fura-2 (30–50 μM), a high Ca2+ affinity fluorescent dye (Kd ∼150 nM). The effect of MLCK inhibition on [Ca2+]i transients induced by a 100-Hz train (50 stimuli) was examined in P9–P13 slices. We imaged [Ca2+]i at a sampling rate of 13–15 Hz. The action potential (AP)–induced [Ca2+]i transients were not altered after 5- to 10-min application of ML-9 (Supplemental Fig. S2). We also performed the same experiment after the calyx was loaded with dye and the pipette withdrawn (i.e., after preload with fura-2). This allows the Ca2+ concentration to rise and decay without the influence of the pipette (Korogod et al. 2005). In both cases, AP-induced [Ca2+]i transients were not affected by ML-9 (n = 3 calyces). Since fura-2 is a fast BAPTA-like Ca2+ buffer, with a high Ca2+ affinity it may saturate and mask the changes in calcium transients. So we next tried the same experiments with fura-4F (100 μM), a low Ca2+ affinity dye (Kd ∼770 nM). Even with fura-4F, an AP train (100 Hz; 50 stimuli) induced [Ca2+]i transients that were not altered by ML-9 application (Fig. 9A; n = 3 calyces).
In conclusion, our results so far show that ML-9 enhances the early EPSCs in a train without increasing basal [Ca2+]i levels or augmenting the transient Ca2+ changes during stimulation of afferent fibers. Because the Ca2+ imaging results are negative one may argue that saturation of the Ca2+ indicator may be masking the changes in Ca2+ transients. Therefore we examined the Ca2+ transients with fura-4F under elevated external Ca2+. A clear increase in the fluorescence ratio was observed with higher external Ca2+ (Fig. 9B; n = 3). These results indicate that MLCK inhibition does not alter basal [Ca2+]i or transient Ca2+ changes.
Effect of ML-9 during postnatal development
During the first two postnatal weeks of development the calyx terminal undergoes extensive morphological and functional changes (Iwasaki and Takahashi 2001; Kandler and Friauf 1993; Taschenberger et al. 2002). We therefore measured the effect of MLCK inhibition as a function of age. The effect of MLCK inhibition was studied from P3 to P16. Typical examples from three different ages (P3, P5, and P16) are shown in Fig. 10A. At P3, an age when a proto-calyx structure starts to form (Hoffpauir et al. 2006), MLCK inhibition had no effect on the amplitude of EPSCs measured with 2 mM extracellular calcium ([Ca2+]e). This lack of effect was not due to saturation of AMPA receptors because a change from 2 to 4 mM [Ca2+]e (see Fig. 10B) caused a large potentiation of EPSC amplitude (n = 3 cells; see Schneggenburger et al. 1999). At P5, however, the early EPSCs were potentiated under normal 2 mM [Ca2+]e on MLCK inhibition. In summary, at P3, the first average EPSC amplitude in control was 0.9 ± 0.06 nA and on MLCK inhibition it was 0.9 ± 0.07 nA (n = 6 cells; P = 0.82), whereas at P5–P6 it significantly increased from 2.1 ± 0.4 to 4.0 ± 1.0 nA (n = 6 cells; P = 0.03). Furthermore, the amount of change of the first EPSC amplitude on MLCK inhibition remained constant from P5 to P16, as shown in Fig. 10B. The steady-state EPSCs did not change significantly on ML-9 application for all age groups, in line with our previous results (Fig. 2). ML-9 application enhanced the synaptic depression of a 100-Hz train in all three age groups (Fig. 10C). At P5–P6, the change of synaptic depression on ML-9 application from that of control was 19 ± 4 to 11 ± 3% (P = 0.001), at P10–P13 the change was 20 ± 2 to 10 ± 2% (P = 0.0001), and at P15–P16 the change was 33 ± 4 to 16 ± 2% (P = 0.01). These results suggest an early postnatal developmental regulation of MLCK.
Recovery from synaptic depression is not affected by MLCK inhibition
Given that the MLCK inhibition increases vesicle pool size, we checked whether it also affected the recovery from synaptic depression. We measured the recovery from depression by applying two trains (20 stimuli; 100 Hz) separated by a variable recovery time (ranging from 0.1 to 13 s). After each pair of trains, the calyx was allowed to recover for 20 s, a time sufficient for complete recovery from short-term depression (Billups et al. 2005; von Gersdorff et al. 1997; Wang and Kaczmarek 1998). We executed the same protocol in the presence of ML-9 and found no difference in the rate of recovery from synaptic depression. A typical example from a P13 MNTB neuron showing recovery from synaptic depression under control condition and on ML-9 application is shown in Supplemental Fig. S3, A and B (note the difference in vertical scales). Results of normalized recovery of the EPSC over time from five different cells (P11–P14) are summarized in Supplemental Fig. S3C. The time constant of recovery from synaptic depression under control conditions was 3.8 s and in the presence of ML-9 it was 3.7 s. These values are very similar to those previously reported for mouse MNTB for 100-Hz trains (Joshi and Wang 2002; Wang and Kaczmarek 1998). Thus these results rule out an involvement of MLCK in the rate of recovery from short-term depression.
We have shown a potential role for MLCK on synaptic transmission at the calyx of Held synapse. We studied the role of MLCK with three distinct inhibitors (ML-9, ML-7, and wortmannin). Application of these inhibitors potentiated the amplitude of the early EPSCs, but not the late steady-state EPSCs, suggesting that MLCK inhibition selectively increases the size of the fast releasing pool of vesicles. A presynaptic locus of action of MLCK inhibitors was suggested by an increase in the frequency of mEPSCs, but no change in their amplitude. Studies using PPRs and high-frequency stimulus trains indicated that MLCK inhibition primarily increased the size of the RRP and not the release probability. Using EGTA-AM we then showed that the effects of ML-9 occur even in the presence of enhanced presynaptic Ca2+ buffering. Finally, we showed that ML-9 does not affect recovery from short-term depression.
Roles of myosin and actin in axon terminals
Actin and myosin are localized to presynaptic terminals (Drenckhahn and Kaiser 1983). Among the different isoforms of myosin, myosin II and V were suggested to be associated with synaptic release (Takagishi et al. 2005; Watanabe et al. 2005). However, the functional role for myosin in neurotransmitter release is controversial. For example, myosin II plays a role in release at superior cervical ganglion synapses (Mochida et al. 1994) and myosin V in hippocampal synapses (Espreafico et al. 1992). But studies with dilute-lethal myosin V mutants revealed normal synaptic transmission (Schnell and Nicoll 2001) and polyclonal antibody studies have not revealed a role for myosin II in vesicle mobilization (Miller et al. 1992). However, there may be possible roles for other myosin isoforms at these synapses. In the neuromuscular junction, myosin II is likely to replenish synaptic vesicle pools to maintain high-frequency repetitive stimulation (Polo-Parada et al. 2005). In this study, we tried to explore the function of presynaptic myosin by targeting MLCK. Interestingly, myosin V isoforms are present in the calyx of Held according to recent immunohistochemistry studies (Baqui et al. 2007).
Involvement of actin and calmodulin in the recovery from vesicle pool depletion has been shown in the calyx of Held (Sakaba and Neher 2001b, 2003). Since myosin-mediated vesicle transport involves actin (Doussau and Augustine 2000; Bridgman 2004), we assessed the role of myosin with MLCK inhibitors in recovery from depression. Our findings show that MLCK does not play a role in the rate of recovery (τ ≈ 4 s) from synaptic depression (Supplemental Fig. S3). This rate of recovery may involve a putative translocation of reserve vesicles, however, the MLCK-blocker ML-7 (15 μM) has been shown to decrease the mobility of vesicles in hippocampal boutons in culture (Jordan et al. 2005). Interestingly, in another study on the auditory pathway, MLCK inhibition enhances LTP induction without affecting basal synaptic transmission (Lamprecht et al. 2006). Thus we emphasize that MLCK inhibition may have different effects at different CNS synapses.
Modulation of synaptic transmission by MLCK inhibition
Localization of MLCK in neurons has been shown by several groups (Akasu et al. 1993; Edelman et al. 1992; Jian et al. 1996; Tokuoka and Goda 2006). MLCK is involved in diverse cellular functions (Kamm and Stull 2001). In several synapses, MLCK inhibitors are known to hinder synaptic transmission and enhance short-term depression (Mochida et al. 1994; Perkeris and Terrian 1997; Polo-Parada et al. 2001; Ryan 1999; Tokuoka and Goda 2006). Our results also show an enhanced degree of short-term depression after MLCK inhibition between the first and last EPSCs in a train (Fig. 10). However, our study shows a potentiation of early EPSC with MLCK inhibitors, but only for P5 or older synapses (Fig. 10). This could be due to little or no MLCK expression at very young calyx synapses.
Previous studies of the action of MLCK inhibition on CNS synapses were done mainly in neuronal cultures. Here we report the effects of MLCK inhibition in acute brain stem slices. MLCK inhibition potentiated the early EPSCs without affecting the steady-state EPSCs, similar to the effects of forskolin in the rat calyx of Held (Kaneko and Takahashi 2004). However, unlike the forskolin effects we did not observe a developmental weakening of the ML-9 effect from P5 to P16 (Fig. 10). PKA activation by an increase in cAMP also causes a potentiation of the fast RRP (Sakaba and Neher 2001a). In our study MLCK inhibition had no effect on the amplitude of spontaneous mEPSCs, confirming that the site of its action is presynaptic (Fig. 3). Similarly, in hippocampal synapses, MLCK inhibition showed a fivefold increase in the frequency of mEPSCs, but no significant effect on mEPSC amplitudes (Tokuoka and Goda 2006).
Specificity of MLCK inhibitors
We used the drugs ML-9 (Ishikawa et al. 1988; Saitoh et al. 1987), ML-7 (Saitoh et al. 1987), and wortmannin (Nakanishi et al. 1992) at concentrations that make them fairly specific MLCK inhibitors (although wortmannin also blocks phosphoinositide 3-kinases; Yano et al. 1995). Accordingly, they all showed very similar effects on EPSCs (Fig. 2). The naphthalene sulfonamide derivatives ML-7 and ML-9 do inhibit other kinases, including PKC and PKA, but higher concentrations are needed. In vitro, these inhibitors are approximately one order of magnitude more potent against MLCK than against PKC or PKA (Saitoh et al. 1987). Therefore we used low concentrations of these inhibitors under which these drugs are reported to potently block MLCK. In fact, we were able to observe the potentiation of early EPSCs even at very low concentrations (5 μM; n = 2 cells; 15 μM; n = 4 cells) of ML-9 (data not shown) and with 15 μM ML-7 (Fig. 2; see also Tokuoka and Goda 2006), whereas H7 (100 μM; general kinase inhibitor) had no effect on EPSCs. Furthermore, if ML-9 or ML-7 were inhibiting PKC or PKA this should lead to a decrease in EPSC amplitude, the opposite of what we observed. However, one cannot rule out the possibility that ML-9, ML-7, and wortmannin are inhibiting an as yet unknown kinase. Finally, we also tested the drug BDM, which is known to inhibit MLCK (Ryan 1999; Siegman et al. 1994). However, application of BDM (10–25 mM; n = 5 cells) drastically inhibited the amplitude of the presynaptic action potential (data not shown), so we did not further pursue its effects on the postsynaptic cell.
Potentiation of EPSCs by MLCK inhibition: no need for global Ca2+ rises
MLCK inhibitors can influence intracellular Ca2+ in some cell types (Ito et al. 2004; Nakanishi et al. 1994). However, MLCK inhibitors did not induce Ca2+ changes in the presynaptic terminal of cervical ganglion neurons (Mochida et al. 1994). ML-9 did not change the waveform or amplitude of APs and Ca2+ currents at the calyx (Figs. 4 and 5). By contrast, ML-7 inhibited the amplitude of postsynaptic APs and Ca2+ currents in hippocampal cultures (Tokuoka and Goda 2006). In addition, we found that MLCK inhibition did not affect calyx K+ currents (Tokimasa et al. 1995). Moreover, alterations of the endogenous basal Ca2+ concentration do not appear to underlie the potentiation of EPSCs (see Fig. 9 and Supplemental Fig. S2), since this potentiation occurred even under enhanced presynaptic Ca2+ buffering (Supplemental Fig. S1).
MLCK-regulated mechanisms for synaptic potentiation
Actin filaments are known to form an intricate cytoskeletal network that associates closely with vesicles and active zones (Hirokawa et al. 1989; Phillips et al. 2001). Several studies have suggested a role for actin in synaptic vesicle mobilization (Bernstein and Bamberg 1989; Kuromi and Kidokoro 1998; Sakaba and Neher 2003). Since synaptic vesicles appear to be linked to actin filaments via synapsin proteins, one function of actin may be to capture vesicles and prevent them from fusing to the plasma membrane (Bahler and Greengard 1987; Llinás et al. 1985; Schiebler et al. 1986). Thus binding to actin may constitute a structural hindrance for exocytosis (Pieribone et al. 1995).
In smooth muscle, MLCK is a multifunctional regulatory protein. When myosin is unphosphorylated, MLCK phosphorylates myosin in a Ca- and calmodulin (CaM)-dependent manner to elevate myosin ATPase activity and promote muscle contraction (Gao et al. 2001). However, MLCK can also bind to the myosin head and elevate myosin ATPase activity in a Ca-independent manner. When myosin is fully phosphorylated, MLCK inhibits myosin action by directly binding to actin (Fujita et al. 1999; Okagaki et al. 1999). We thus speculate that the potentiation of EPSCs observed here could be due to the following mechanism: synaptic vesicles that bind to myosin may become free from binding to the actin cytoskeleton by MLCK inhibition and thus ready for fusion with the plasma membrane [i.e., MLCK inhibition may promote a Ca-independent (or EGTA-resistant; Supplemental Fig. S1) movement of vesicles to the active zone].
The synaptic potentiation we observed here was associated with a selective increase in the fast releasing vesicle pool size (Figs. 2, 7, and 8), which indicates a possible role for MLCK in regulating the steps before the final fusion reaction. MLCK may thus have a direct role in inhibiting biochemical steps that lead to vesicle priming, so that its inhibition would promote exocytosis. Another possibility is that MLCK may alter the interaction of docked vesicles with nearby Ca2+ channels. If MLCK inhibition were to promote a closer association between docked vesicles and Ca2+ channels this could also explain the preferential increase in the fast releasing pool of vesicles (Wadel et al. 2007). Finally, we point out that electron-microscopic studies of the calyx of Held have shown that there are about 500 to 600 active zones (Satzler et al. 2002; Taschenberger et al. 2002). So MLCK inhibition may be converting some “silent” active zones with predocked vesicles into functionally competent (or “primed”) active zones, thereby increasing the total number of docked vesicles that are primed for rapid release at the calyx terminal. Most interestingly, myosin Va has recently been shown to mediate the stable attachment of vesicles to the plasma membrane (Desnos et al. 2007).
Together, our results suggest that MLCK is likely to play a key role in controlling the size of the readily releasable pool of vesicles. Phosphorylation can thus regulate vesicle pool size and the rate of short-term depression (Silverman-Gavrila et al. 2005). However, it seems unlikely that MLCK acts during the vesicle recycling and recovery from synaptic depression stages of neurotransmission. To better understand the mechanism underlying the action of MLCK further studies of its precise localization within the nerve terminal and its protein substrates for phosphorylation will be necessary.
This work was supported by a National Institute on Deafness and Other Communication Disorders RO1 grant to H. von Gersdorff and an American Heart Association postdoctoral grant to J. H. Kim.
We thank V. Balakrishnan, R. M. Leão, R. Renden, C. Jahr, and E. Neher for discussions.
↵1 The online version of this article contains supplemental data.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 by the American Physiological Society