INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Several lines of evidence have suggested the existence of a structural relationship between cytoskeletal elements and ligand-activated ion channels, such as the N-methyl-D-aspartate receptor (NMDA-R), -amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptor (AMPA-R), nACh receptor (nACh-R), and GABA_A receptor (GABA_A-R) (Allison et al. 1998; Shoop et al. 2000; Wang et al. 1999). Moreover, the state of cytoskeleton polymerization appears to play an important role in regulating NMDA-R and GABA_A-R channel activity (Rosenmund and Westbrook 1993; Whatley et al. 1994). Thus, these studies showed that receptor-evoked current rundown and ligand sensitivity were acutely affected by alkaloids known to disrupt selective elements of the cytoskeleton (Rosenmund and Westbrook 1993; Whatley et al. 1994). These results were interpreted as a cytoskeleton-dependent increase in receptor internalization and modifications on ligand affinity.

Previous studies have analyzed the effect of cytoskeletal disrupters on the cellular localization of gephyrin, a GlyR-associated anchoring protein that binds directly to microtubules and is also indirectly associated to actin filaments (reviewed in Kneussel and Betz 2000). Disruption of microtubules was able to alter the organization of gephyrin and GlyR clusters in spinal neurons (Kirsch and Betz 1995), although gephyrin distribution in cultured hippocampal neurons was not affected by microtubule depolymerization (Allison et al. 2000). These results suggest that microtubules could regulate the function of the GlyR in certain neurons, but no data on the physiological properties of GlyRs in spinal neurons after microtubule disruption are available. Therefore to assess the role of microtubules on GlyR function, we induced microtubule depolymerization with colchicine and analyzed the current evoked by exogenous glycine (I_glycine). In addition, we analyzed the current obtained from the spontaneous release of this neurotransmitter by recording spontaneous miniature inhibitory postsynaptic currents (mIPSCs).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Cell culture

Embryonic (13-14 days old) C57/BL mouse spinal cord neurons were cultured as previously described (Tapia and Aguayo 1998). Briefly, spinal neurons obtained from five or six embryos were plated at 300-350,000 cells/ml into 35-mm tissue culture dishes coated with poly-L-lysine (MW >350 kDa; Sigma). Serum containing media was changed every 3 days. The microtubule disrupter colchicine and its inactive analogue -lumicolchicine (Liebman 1986) were applied as previously reported (Johnson and Byerly 1993; Kirsch and Betz 1995; Rosenmund and Westbrook 1993). Colchicine (20 µM) was applied to the bath for 3 h before either fixing or recording the cells, or was internally dialized into single cells for 25 min via the internal solution of the patch pipette.

Immunostaining

Neurons were fixed in 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4) for 30 min and permeabilized with 0.25% Triton X-100 for 15 min. Microtubules were labeled with a monoclonal antibody against tubulin (1:1,000; clone DM1, Sigma) and visualized with FITC (1:50; Jackson Labs). Samples from two experiments were analyzed using an Olympus Fluoview confocal microscope.

Electrophysiology

GLYCINE-EVOKED CURRENTS. Whole cell patch-clamp recordings in 5-12 days in vitro (DIV) neurons were performed and analyzed as previously described (Tapia and Aguayo 1998). Briefly, the cell was clamped at 60 mV and rapid onset applications of glycine were used to activate GlyRs. Data were excluded if the uncompensated series resistance (<8 M) or the leak current (<75 pA) increased by >15% during the long-term recordings. Patch electrodes were filled with (in mM): 140 KCl, 10 BAPTA, 10 HEPES (pH 7.4), 4 MgCl₂, and 2 ATP-Na₂. The external solution contained (in mM): 150 NaCl, 10 KCl, 2.0 CaCl₂, 1.0 MgCl₂, 10 HEPES (pH 7.4), and 10 glucose.

MINIATURE IPSCs. Spontaneous glycinergic mIPSCs were isolated by the addition of 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX; 2 µM), bicuculline (2 µM), TTX (0.1 µM), and MgCl₂ (2 mM) to the external solution. Strychnine (750 nM) blocked all the events that remained after application of these two antagonists. Synaptic currents were recorded using Axotape 7.0 software (Axon Instruments, Union City, CA) for off-line analysis. Every identified synaptic event encountered during a 3-min period, and having an amplitude above the background noise (12-15 pA), was analyzed with MiniAnalysis 5.0 software (Synaptosoft, Leonia, NJ). Cursors were automatically set to measure the peak amplitude and rise-time and decay-time constants. To build the amplitude histogram, the amplitudes of six cells were arranged into 1-pA bins and combined to obtain a single histogram. The data at the onset of the recording (min 1) were used as an internal control (Pusch and Neher 1988). All reagents were purchased from Sigma (St. Louis, MO).

Data analysis

Unless otherwise noted, statistical analysis was performed using Student's t-test, and data are expressed as arithmetic means ± SE. mIPSC amplitude distributions were compared using Kolmogorov-Smirnov (KS) test. P < 0.05 was considered significant. The data were tested for significance at minute 25 or 60.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Effects of colchicine on microtubules and neuronal morphology

Confocal microscopy of tubulin-immunostaining confirmed the induction of cytoskeletal alterations after 20 µM colchicine was applied to the culture media before fixing and processing the neurons. Figure 1 shows that the treatment with colchicine altered the morphology of the neurons, which showed abundant blebing along their proximal dendrites. Microtubule bundles were also less evident and discrete in colchicine-treated neurons compared with control neurons (see arrows, Fig. 1). The inactive analogue -lumicolchicine (20 µM) was without effect (not shown).

View larger version (51K): [in this window] [in a new window]   Fig. 1. Effects of colchicine on microtubules. Confocal microscopy of tubulin immunostaining. A: tubulin immunostaining reveals tight bundles of microtubules within the soma and proximal neurites in control neurons. B: colchicine (20 µM for 3 h) resulted in alteration in neurite morphology and microtubule bundles (arrows). Insets show magnified views of the boxed areas. Scale bar: 10 µm.

Properties of evoked glycine current after microtubule disruption

The amplitude of the glycine-activated Cl current ran down to 73 ± 6% (n = 8) after 60 min of recording in control neurons (Fig. 2A). The current rundown was not affected by the addition of 20 µM colchicine to the internal solution (80 ± 3%, n = 5), nor after 3 h with 20 µM colchicine in the media (71 ± 9%, n = 3). Concentration-response curves in control neurons resulted in graded glycine responses with an EC₅₀ of 29 ± 1 µM (Fig. 2C, n = 10) and a Hill coefficient of 1.8 ± 0.1. Neither EC₅₀ values nor Hill coefficients were altered when colchicine was added to the patch-pipette (24 ± 2 µM, 1.6 ± 0.2, n = 4) or to the media (31 ± 1 µM, 2.0 ± 0.1, n = 6). Next, we examined whether the sensitivity of the I_glycine to the antagonist strychnine changes after colchicine treatment to analyze whether this alkaloid causes alterations on its subunit stoichiometry (Schmieden et al. 1992). However, colchicine at a concentration that clearly altered microtubule organization and cellular morphology had no effect on I_glycine properties. Strychnine (100 nM) inhibited I_glycine by 53 ± 5% (n = 5) in control neurons (Fig. 2D). The I_glycine amplitude was inhibited by 47 ± 8% (n = 5) and 57 ± 7% (n = 4) when colchicine was added to the internal solution or to the culture media, respectively.

View larger version (35K): [in this window] [in a new window]   Fig. 2. Effect of colchicine on the current evoked by 100 µM glycine. A: rundown of glycine-evoked current in control and in the presence of 20 µM colchicine added to the patch-pipette or the culture media. The current was activated every minute using 100 µM glycine. Some data points were removed for clarity. B: current traces at 60 mV were obtained from a control neuron at minute 1 and minute 60 of recording. C: concentration-response relationships from control and treated neurons was also unchanged by colchicine treatment. D: no statistically significant differences were found either in the percentage of glycine current blocked by 100 nM strychnine in control neurons or colchicine-treated neurons. Error bars indicate SE.

Disruption of microtubules reduced the quantal amplitude of spontaneous glycinergic mIPSCs

The lack of colchicine effect on the whole cell I_glycine was surprising since a structural interaction between microtubules and GlyR was previously suggested (Kirsch and Betz 1995). Therefore we analyzed the effect of colchicine on spontaneous glycinergic postsynaptic currents to determine whether microtubules are able to modify the function of synaptic GlyRs. The synaptic events recorded under our experimental conditions are assumed to arise from the spontaneous release of single vesicles and are referred to as quantal or mIPSCs. The amplitude of mIPSCs in cultured spinal neurons was 34.7 ± 7 pA (n = 6). In addition, these mIPSCs were confirmed as glycinergic because they were blocked by a low concentration (750 nM) of strychnine. It is interesting to note that very similar mIPSC properties were found in brain stem motoneurons (O'Brien and Berger 1999).

Synaptic current properties, such as peak amplitude, frequency, rise time, and decay time remained highly stable for 25 min under control whole cell recordings (Fig. 3A and Table 1). On the other hand, a time-dependent alteration of synaptic properties occurred during colchicine (20 µM) dialysis inside the cells (Fig. 3B and Table 1). After 25 min, the mean peak amplitude, frequency, and decay time were significantly decreased. Dialysis with the inactive analogue -lumicolchicine (20 µM; n = 5) was unable to significantly alter the average amplitude (93 ± 9%) or frequency (85 ± 17%). Similarly, external application of colchicine (20 µM, n = 3) for 20 min did not significantly change the frequency (94 ± 7%) or amplitude (93 ± 7%) of mIPSCs.

View larger version (37K): [in this window] [in a new window]   Fig. 3. Acute disruption of microtubules with colchicine selectively reduces pharmacologically isolated spontaneous glycinergic synaptic events. A1: combined amplitude histograms from 6 spinal neurons recorded at 1 and 25 min of dialysis with normal intracellular solution and bathed with 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) and bicuculline. A2: cumulative amplitude distributions at minute 1, 15, and 25 from a single neuron were not different. B1: the amplitude histograms from 6 neurons were combined to show the effect of intracellular application of colchicine. Note the strong reduction on glycinergic miniature inhibitory postsynaptic current (mIPSC) amplitude in the colchicine treated neurons. B2: cumulative amplitude distributions from 1 neuron in the presence of colchicine in the patch-pipette are shown. Note the time-dependent reduction on the quantal amplitude as illustrated by a shift of the distribution toward smaller amplitudes. C1: total and isolated glycinergic synaptic activity recorded 1 min after establishing the whole cell recording. C2: the trace shows that while the glycinergic activity was drastically reduced, the overall synaptic transmission remained high. The cell was continously washed with normal external solution between the recordings.

Colchicine specifically decreases spontaneous glycinergic synaptic currents

To address whether the inhibitory effect of intracellular colchicine on GlyRs was selective, we examined the effect of this alkaloid on the overall synaptic activity of 8-10 DIV neurons. Figure 3C₁, for example, shows overall spontaneous synaptic activity (left trace) and glycinergic mIPSCs pharmacologically isolated in the presence of bicuculline (2 µM) and CNQX (2 µM; right trace). Recordings after 25 min of intracellular colchicine dialysis revealed that the glycinergic activity was strongly inhibited (Fig. 3C₂, right trace), but the overall synaptic activity remained basically unchanged (left trace). Pharmacologically isolated GABAergic mIPSCs (with 1 µM strychnine and 2 µM CNQX) during colchicine dialysis showed no change in mean peak amplitudes; 61 ± 19 and 60 ± 21 pA at minutes 1 and 25, respectively (n = 3 neurons). Similarly, the corresponding mean frequencies of GABA mIPSCs were 1.3 ± 0.3 and 1.1 ± 0.3 Hz at minutes 1 and 25. These results are interesting because GlyRs and GABA_A-Rs co-localize in a large proportion of inhibitory synapses in cultured spinal neurons (Dumoulin et al. 2000) and suggest that GlyRs were preferentially affected by colchicine treatment at synaptic sites containing both receptors. Furthermore, the average peak amplitude of AMPAergic miniature excitatory postsynaptic currents (mEPSCs; isolated with 1 µM bicuculline and 2 µM CNQX) following a similar colchicine treatment was 43 ± 6 and 43 ± 8 pA at 1 and 25 min, respectively. Mean frequency of AMPA mEPSCs remained highly stable for 25 min (91 ± 10% of minute 1) in the presence of colchicine (n = 3).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Previous studies have shown that the selective disruption of cytoskeletal elements can modify several functional properties of NMDA-Rs and GABA_A-Rs (Rosenmund and Westbrook 1993; Whatley et al. 1994). Similarly, GlyRs are believed to be tightly associated to the peripheral cytoskeleton through gephyrin, a microtubule-binding protein (Kirsch and Betz 1995), but the consequences of microtubule disruption on GlyR function were unknown. Our results showed a selective effect of colchicine over synaptic transmission mediated by GlyRs, but not on AMPA- or GABA_A-mediated synaptic activity. Therefore microtubules appear more tightly associated to synaptic GlyRs than to AMPA-Rs or GABA_A-Rs. Morphological studies also suggests that the maintenance of core components of AMPA and GABA_A postsynatic receptor clusters are independent of the microtubule state (Allison et al. 2000).

We also found that microtubule disruption did not affect the glycine whole cell current. The results agree with the specific localization of gephyrin at synaptic sites in spinal neurons (Alvarez et al. 1997; Triller et al. 1985) and suggest that the GlyRs responsible for the I_glycine current are most likely extrasynaptic and do not interact with microtubules.

Recent findings have suggested a positive correlation between gephyrin cluster size and glycinergic mIPSC amplitude (Lim et al. 1999; Oleskevich et al. 1999). Therefore the reduction on glycinergic mIPSC amplitude could be due to a disorganization of the postsynaptic gephyrin cluster and parallel reduction in the number of postsynaptic receptors. Alterations in gephyrin-mediated postsynaptic clustering and an increase in the lateral mobility of GlyRs and their disappearance from postsynaptic sites has been reported in spinal neurons treated with the microtubule disrupter demecolcine (Kirsch and Betz 1995). However, others have failed to observe any alteration in gephyrin cluster structure or number after microtubule depolymerization in hippocampal neurons (Allison et al. 2000). The reduction on glycinergic inhibitory efficacy that we found agrees with a decrease in the number of postsynaptic receptors, but future experiments are needed to correlate the results described here with possible alterations in the number and/or sensitivity of postsynaptic GlyRs in these cultures.