Visualization of transmitter release with zinc fluorescence detection at the mouse hippocampal mossy fibre synapse

Authors


Corresponding author J. L. Noebels: Department of Neurology, Baylor College of Medicine, One Baylor Plaza, Houston, TX 77030, USA. Email: jnoebels@bcm.tmc.edu

Abstract

Exocytosis of synaptic vesicle contents defines the quantal nature of neurotransmitter release. Here we developed a technique to directly assess exocytosis by measuring vesicular zinc release with the zinc-sensitive dye FluoZin-3 at the hippocampal mossy fibre (MF) synapse. Using a photodiode, we were able to clearly resolve the zinc fluorescence transient ([Zn2+]t) with a train of five action potentials in mouse hippocampal brain slices. The vesicular origin of [Zn2+]t was verified by the lack of zinc signal in vesicular zinc transporter Znt3-deficient mice. Manipulating release probability with the application of neuromodulators such as DCG IV, 4-aminopyridine and forskolin as well as a paired train stimulation protocol altered both the [Zn2+]t and the field excitatory postsynaptic potential (fEPSP) coordinately, strongly indicating that zinc is co-released with glutamate during exocytosis. Since zinc ions colocalize with glutamate in small clear vesicles and modulate postsynaptic excitability at NMDA and GABA receptors, the findings establish zinc as a cotransmitter during physiological signalling at the mossy fibre synapse. The ability to directly visualize release dynamics with zinc imaging will facilitate the exploration of the molecular pharmacology and plasticity of exocytosis at MF synapses.

The ability to separate presynaptic changes in neurotransmitter release from use-dependent modifications in postsynaptic sensitivity is essential to the understanding of the molecular basis of synaptic signalling and plasticity. Several elegant technical approaches have been developed to explore this problem. The measurement of presynaptic membrane capacitance was first used to quantify vesicle exocytosis at certain giant synapses, where monitoring membrane capacitance is feasible, more than a decade ago (Von Gersdorff & Matthews, 1994). At other synapses, where membrane capacitance is not accessible, the cell membrane marker styryl dyes have been widely employed for investigating activity-dependent vesicle recyling (Betz & Bewick, 1992). More recently, by incorporating molecular biosensors into the vesicular membrane, vesicle recycling has been examined on the scale of individual quanta (Gandhi & Stevens, 2003). However, these methods have only been employed in large terminals or at presynaptic terminals of cultured neurons, and currently, at most central synapses studied in acute brain slices, the output of this multistep process can only be inferred indirectly by measuring the response of postsynaptic receptors.

At the mossy fibre (MF) synapse, a major afferent input to the hippocampal formation important for learning and memory, zinc coexists with glutamate inside synaptic terminals (Haug, 1967) where it can be detected by selective histochemical stains. Zinc is accumulated by the vesicular zinc transporter Znt3, and antibodies to Znt3 label the entire pool of small clear glutamatergic vesicles in MF terminals (Wenzel et al. 1997; Cole et al. 1999). The packaging of zinc with glutamate inside vesicles and its multiple effects on synaptic receptors and ion channels (Peters et al. 1987; Westbrook & Mayer, 1987; Mayer & Vyklicky, 1989; Christine & Choi, 1990; Draguhn et al. 1990; Smart et al. 1991; Vogt et al. 2000) suggest the possibility that zinc is a neurotransmitter and co-released with glutamate during exocytosis. This hypothesis was supported by early studies demonstrating a general correlation between raised extracellular zinc concentration and neuronal activity induced by intense electrical stimulation or high potassium (Assaf & Chung, 1984; Howell et al. 1984). There was little further progress in clarifying the issue until recent efforts were made to visualize zinc release using zinc-sensitive fluorescent indicators (Thompson et al. 2000; Li et al. 2001; Ueno et al. 2002; Kay, 2003). These studies clearly demonstrated a rise of extracellular zinc concentration along the MF pathway after MF synapses were activated by high frequency stimulation (HFS) lasting many seconds; however, the stimulation protocols were far beyond the physiological range of low tonic firing frequency or short high frequency bursts at the mossy synapse that are relevant to spatial learning and memory in the hippocampal formation (Jung & McNaughton, 1993; Gothard et al. 2001). The strong stimulation required to allow detection raised the question of whether the elevation actually reflected a massive accumulation of extracellular zinc emanating from pre- or even postsynaptic sites, rather than quantal exocytosis per se, and might relate to cell toxicity and death rather than synaptic signalling (Manzerra et al. 2001). The failure to detect zinc release under physiological conditions has raised a persistent question of whether vesicular zinc is released at synapses during normal synaptic transmission, a matter of importance due to its wide variety of molecular targets in the perisynaptic space (Gilly & Armstrong, 1982; Seguela et al. 1996; Laube et al. 2000; Mitrovic et al. 2001; Baron et al. 2001), at intracellular sites regulating gene expression (Atar et al. 1995), and an important role in synaptic network behaviour in development and disease (Xie & Smart, 1991; Buhl et al. 1996; Koh, 2001). Recently, a novel zinc-sensitive membrane-impermeable fluorescent indicator, FluoZin-3 has become commercially available (Gee et al. 2002). In this study with zinc imaging, we demonstrate that vesicular zinc is in fact co-released with glutamate during exocytosis evoked by action potentials.

Methods

Preparation of brain slices

Transverse brain slices (200 μm thickness) were prepared from hippocampi of wild-type and Znt3−/− mice (3–4 weeks old) in accordance with the guidelines of the National Institutes of Health, as approved by the animal care and use commitee of Baylor College of Medicine. Mice were anaesthetised with Avertin (1.25% tribromoethanal/amyl alcohol solution) using a dose of 0.02 ml g−1. Znt3−/− mutant mice were generously provided by R. D. Palmiter (University of Washington, Seattle, WA, USA). After cervical dislocation, the mouse brain was carefully removed, cooled in 4°C saline saturated with 95% O2–5% CO2, and sliced using a vibrotome. The slice-cutting solution contained (mm): 110 sucrose, 30 NaCl, 3 KCl, 0.5 CaCl2, 28 NaHCO3, 7 MgCl2, 1.4 NaH2PO4, 11 d-glucose. Brain slices were then incubated in artificial cerebrospinal fluid (ACSF) at 32°C for 1 h and rested at room temperature for 30 min before being transferred into a submerged recording chamber mounted on an inverted microscope (Axiovert 100, Zeiss). The ACSF contained (mm): 119 NaCl, 2.5 KCl, 2.5 CaCl2, 26 NaHCO3, 2 MgSO4, 1 NaH2PO4, 11 d-glucose, gassed with 95% O2–5% CO2 to maintain a constant pH of 7.4. The temperature of the recording chamber was controlled at 30°C.

Identifying mossy fibre fEPSP responses

We used several criteria to identify mossy fibre responses. First, field recording electrodes were placed in the stratum lucidum where mossy fibre boutons synapse with CA3 pyramidal cells. Mossy fibre synapses exhibit significant paired-pulse facilitation (Salin et al. 1996), and only such recordings were accepted for analysis. Finally, at the end of each experiment, 1 or 2 μm of the type II metabotropic glutamate receptor (mGluR) agonist DCG IV was applied to test for blockade. Mossy fibre responses are blocked by DCG IV while commissural–associational responses are not.

Zinc fluorescence imaging at the mossy fibre synapse

A bipolar tungsten electrode was positioned in the stratum lucidum near the hilus of the dentate gyrus to stimulate mossy fibres. The stimulation pulse intensity was adjusted between 5 and 10 μA/0.1 ms to elicit a submaximal response. The zinc-sensitive indicator FluoZin-3 was present in the extracellular solution at a concentration of 1 μm. The slow zinc chelator Ca-EDTA (200 μm) was included in most experiments except for experiments in Figs 1 and 2 to reduce basal fluorescence. A small recording area (150 μm in diameter) in the stratum lucidum about 400 μm away from the stimulation electrode was excited at a wavelength of 488/20 nm; the emitted fluorescence was filtered by a bandpass filter of 535/25 nm and converted into electrical signals with a single photodiode (Newport Green DCF, ex: 495/30 nm, em: 545/50 nm). For field recording, patch recording electrodes (filled with ACSF) were positioned in the centre of the optical recording area. A train of 5, 10 or 20 pulses at 33 Hz or 5 Hz was used to evoke the transient zinc fluorescence signal (ΔF or [Zn2+]t) and field excitatory postsynaptic potential (fEPSP) simultaneously. When synapses were activated by a larger train of 10 or 20 pulses, 2 μm muscimol was added to ACSF to reduce the burst firing of action potentials. Signals were filtered through low-pass filters with a corner frequency of 5 kHz and sampled at 10 kHz. Sample traces shown are an average of 2–5 successive traces during steady state to improve the signal-to-noise ratio. The initial slope of the fEPSP is taken as the measure of synaptic transmission. A protocol with stimulation or without stimulation was alternatively applied every 60 s. The fluorescence time course obtained using the protocol without stimulation was used as a reference to correct for photo-bleaching of FluoZin-3 during the data sampling period. The final trace was obtained by subtracting the trace without stimulation from the consecutive trace with stimulation. Due to possible accumulation of extracellular zinc or to an internalization of zinc indicator, basal fluorescence slowly up-shifted in most experiments even though the stimulus-evoked ΔF remained stable. Therefore, for the experiments in Fig. 4, ΔF was quantified to represent the amount of transmitter release, and ΔF had to remain stable for at least 20 min before a protocol was applied. The fluorescence intensity in all figures was labelled in arbitrary units of photodiode output. The autofluorescence of slices was measured and subtracted from the raw fluorescence before data analysis in most experiments except for Figs 1 and 2. Data in each experiment were normalized to the baseline before drug application, then pooled and expressed as a mean ± standard deviation. Two-tail t tests were used to determine statistical significance.

Figure 1.

Fluorescence signal of zinc-sensitive dye FluoZin-3 evoked by a train of stimuli
Fluorescence signal of FluoZin-3 recorded in the stratum lucidum in response to a train of 20 stimulation pulses (33 Hz). A, schematic diagram of a hippocampal slice showing stimulating and recording sites (Inset: sample traces of raw fluorescence signal of FluZin-3 in response to experimental protocol with or without stimulation of mossy fibres. B, time course of basal fluorescence intensity and peak amplitude of the transient fluorescence change ([Zn2+]t) in response to application of FluoZin-3, CNQX +d-APV, and DCG IV. Inset shows sample traces of [Zn2+]t and the corresponding field excitatory postsynaptic potential (fEPSP) which exhibits a typical mossy fibre response with a steep short-term facilitation of release. The [Zn2+]t was unrelated to postsynaptic activity since the glutamate receptor antagonists CNQX (10 μm) and d-APV (25 μm) did not affect the [Zn2+]t. In contrast, [Zn2+]t was sensitive to the presynaptic type II mGluR receptor agonist DCG IV (1 μm). This indicates that the [Zn2+]t was not an artifact of light scattering as a result of postsynaptic activity. The blockade of [Zn2+]t with DCG IV is consistent with the hypothesis that zinc is released during vesicular exocytosis from presynaptic terminals.

Figure 2.

[Zn2+]t is not an artifact of light scattering but reflects accumulation of extracellular zinc
A, time course of basal fluorescence intensity in response to application of FluoZin-2 and FluoZin-3. Sample traces show [Zn2+]t obtained with low affinity zinc indicator FluoZin-2 marked as ‘a’ and high affinity zinc indicator FluoZin-3 marked as ‘b’. Absence of [Zn2+]t signal using the low affinity indicator in ‘a’ excluded the possibility that [Zn2+]t in ‘b’ was contributed by light scattering. B, left panels, time course of basal FluoZin-3 fluorescence in response to different concentrations of slow zinc chelator Ca-EDTA (upper panel) and fast chelator EDDA (lower panel). Reduction of basal FluoZin-3 fluorescence by zinc chelators indicates the existence of free extracellular zinc. Insets (right) show sample traces (a–d) taken in control and in the presence of varying concentrations of zinc chelators: Ca-EDTA (upper panel) and EDDA (lower panel). Both altered the peak amplitude as well as the decay phase of [Zn2+]t in a concentration-dependent manner. C, sample traces of [Zn2+]t and fEPSP evoked by a train of 5 stimuli at 33 Hz in the presence of different strengths of zinc chelation. Increasing concentrations of EDDA decreased the amplitude of [Zn2+]t, and shortened the time to reach peak amplitude. The latency of the [Zn2+]t peak response was measured from the last fEPSP as marked by the dashed line. D, summary data of peak amplitude and delay of [Zn2+]t in response to application of 200 μm Ca-EDTA, and 5, 25 and 125 μm EDDA.

Figure 4.

[Zn2+]t reflects glutamate release at the mossy fibre synapse
A, time course of the peak amplitude of [Zn2+]t and summation of the fEPSP slope in response to application of 5 μm 4-AP (n= 5). Insets show sample traces of [Zn2+]t and fEPSP taken before and during steady state of 4-AP application. B, time course of the peak amplitude of [Zn2+]t and summation of fEPSP slope in response to application of 50 μm forskolin (n= 5). Insets show sample traces of [Zn2+]t and fEPSP taken in control and at the peak of forskolin action. C, left: superimposed [Zn2+]t evoked by a pair of trains separated by an interval of 0.1 s, 1 s and 2 s, respectively; right: the 1st fEPSP (top trace) and the corresponding 2nd fEPSP (lower traces) within a pair of trains obtained at the different intervals. [Zn2+]t and fEPSP obtained with application of DCG IV in the same experiment are also shown. D, summary of data showing correlation of [Zn2+]t and fEPSP facilitation versus the time interval between paired trains. E, summary plot of [Zn2+]tversus fEPSP from experiments with paired train protocol and application of the release-modulating agents 4-AP, DCG IV and forskolin, indicating a good correlation between [Zn2+]t and glutamate release over an extended range of release probability.

Pharmacological reagents

Zinc indicators FluoZin-3, FluoZin-2 and Newport Green DCF were obtained from Molecular Probes. (2S,2′R,3′R)-2-(2′,3′-Dicarboxycyclopropyl)glycine (DCG IV), muscimol, 6-cyano-7-nitroquinoxaline-2,3-dione (CNQX) and d-aminophosphonvalerate) (d-APV) were purchased from Tocris. 4-AP, forskolin, Ca-EDTA and EDDA (Ethylenediamine-N, N′-diacetic acid) were from Sigma.

Results

Zinc release evoked by action potentials at the mouse mossy fibre synapse

Figure 1 shows a typical experiment measuring extracellular zinc accumulation ([Zn2+]o) evoked by the stimulation of MF synapses with FluoZin-3, a high affinity, membrane-impermeable fluorescent zinc indicator with a dissociation constant (Kd) of 15 nm. The inset shows superimposed raw fluorescence traces with and without stimulation. Over each sampling period of 10 s, there was a small decrement of about 5% of the signal due to photo-bleaching of FluoZin-3. Therefore, the fluorescence time course obtained without stimulation was used as a reference to correct for bleaching of the zinc indicator during the data sampling period. Figure 1B shows the time course of the basal FluoZin-3 fluorescence intensity (F) and fluorescence change (ΔF) evoked by a stimulation protocol consisting of 20 pulses at 33 Hz. A transient change of fluorescence intensity ([Zn2+]t) corresponding to the stimulation train was detected when 1 μm FluoZin-3 was washed into the slice. As shown by the inset sample trace, following termination of the stimulus train, the elevated fluorescence intensity decayed slowly over the entire 10 s data sampling period. The [Zn2+]t was not related to postsynaptic activity, since the glutamate receptor antagonists CNQX and d-APV had no measurable effects on the evoked [Zn2+]t signal (n= 4). However, [Zn2+]t was sensitive to DCG IV, an agonist for type II mGluRs which reduces neurotransmitter release at the MF synapse but not at the neighbouring commissural–associational pathway (Kamiya et al. 1996). This experiment ruled out the possibility that the [Zn2+]t was an artifact of light scattering as a result of swelling due to postsynaptic activity (MacVicar & Hochman, 1991) or activity-dependent change of autofluorescence that arises mainly from NAD(P)H and flavoproteins (Kunz et al. 1999; Shibuki et al. 2003). The blockade of [Zn2+]t with DCG IV is consistent with the hypothesis that zinc is released from presynaptic terminals during vesicular exocytosis.

[Zn2+]t reflects accumulation of extracellular zinc

Although postsynaptic sources were excluded, the apparent [Zn2+]t signal might still be an artifactual contribution of light scattering caused by activity in presynaptic terminals (Tasaki & Byrne, 1992), and the sensitivity of [Zn2+]t to DCG IV could be explained by the fact that DCG IV inhibits presynaptic activity, and therefore prevents the swelling of axon terminals. To evaluate this possibility, we examined the fluorescence response of two zinc indicators from the same preparation as shown in Fig. 2A. If the [Zn2+]t were indeed a result of light scattering related to presynaptic activity, then FluoZin-2, a zinc indicator with a much lower affinity for zinc (Kd= 3 μm), would yield a transient signal similar to the high affinity zinc indicator FluoZin-3. As shown in Fig. 2A, no transient signal was detected with FluoZin-2 in spite of a robust signal with FluoZin-3, indicating that the [Zn2+]t obtained with FluoZin-3 is not related to light scattering. Furthermore, as shown in Fig. 2B, the FluoZin-3 [Zn2+]t was altered by the zinc chelators Ca-EDTA and EDDA. Both types of zinc chelator affected not only the peak but also the decay phase of [Zn2+]t in a concentration-dependent manner without affecting the fEPSP (n= 3 for Ca-EDTA, n= 4 for EDDA). Since zinc chelators greatly reduce basal fluorescence, and to avoid a false increase of ΔF/F in their presence, sample traces of [Zn2+]t were normalized to the value for F obtained before the application of chelators. The fact that zinc chelators alter the kinetics of the [Zn2+]t indicates that the [Zn2+]t was not an artifact of synaptic activity-induced light scattering, but reflects the release-related accumulation of extracellular zinc. FluoZin-3 is membrane-impermeable. Since we did not observe enhanced fluorescence of the mossy fibre boutons after slices were incubated in zinc indicators and stimulated for more than 1 h, significant endocytosis of zinc dye into vesicles and a contribution of fluorescence increase due to possible dye dequenching during exocytosis is very unlikely. For all of the subsequent experiments, the zinc chelator Ca-EDTA (200 μm) was routinely present in the perfusate to buffer the extracellular zinc concentration and reduce background fluorescence.

Figure 2C shows the zinc signal evoked by a train of 5 stimuli at 33 Hz. Similar to control experiments in Fig. 2B, the addition of the strong zinc chelator EDDA altered the kinetics of [Zn2+]t. Increasing the EDDA concentration decreased the amplitude of [Zn2+]t and shortened the time to reach peak amplitude. In the presence of 200 μm Ca-EDTA only, the average latency as measured between the peak of the last fEPSP and the peak of [Zn2+]t was 163 ± 8 ms (n= 4). For the application of 125 μm EDDA, the latency was 46 ± 5 ms (n= 4).

Lack of evoked zinc release in Znt3-deficient mutants

Zinc is accumulated by the vesicular zinc transporter Znt3 (Slc30a3) (Cole et al. 1999). To further confirm the vesicular origin of [Zn2+]t, we measured the [Zn2+]t in zinc transporter Znt3-deficient mice (Znt3−/−). In contrast to the robust signal from wild-type mice as shown in Fig. 3, the [Zn2+]t signal was reduced 3-fold and could be barely detected in Znt3−/− synapses. The small fluorescence transient in response to stimulation at the highest intensities was probably due to light scattering. For a stimulus of 20 impulses, the ΔF/F was 0.36 ± 0.2% in Znt3−/− mice (n= 4) in contrast to 10.4 ± 2.7% in Znt3+/+ mice (n= 6). This result provides the first functional evidence for the role of Znt3 in sequestering zinc into synaptic vesicles for exocytotic release.

Figure 3.

Lack of vesicular zinc release in zinc transporter Znt3-deficient mice
A and B, sample traces of [Zn2+]t evoked by a train of 5, 10 and 20 stimulation pulses (33 Hz) in the presence of 200 μm Ca-EDTA from Znt3−/− mice (A, n= 4) are compared with +/+ mice (B, n= 6) shown at 1/10th scale, revealing dependence of [Zn2+]t on zinc content of mossy fibre vesicles. C, summary data of [Zn2+]t evoked by a train of 20 APs.

[Zn2+]t is highly correlated with glutamate release

If vesicular zinc is co-released with glutamate, then any manipulation of release probability should consistently alter [Zn2+]t. Our experiments revealed a good correlation between the [Zn2+]t and neurotransmitter release which was measured by the summation of all five fEPSPs within the train. As shown in Fig. 4A, application of 5 μm of the potassium channel blocker 4-AP, a concentration that prolongs repolarization of action potentials and thereby promotes presynaptic Ca2+ entry and transmitter release, greatly enhanced both the fEPSP (231 ± 50%, n= 5) as well as [Zn2+]t (280 ± 30%, n= 5). The increased [Zn2+]t was stable during the period of 4-AP application, indicating that zinc replenishment of synaptic vesicles outpaces zinc release under these experimental conditions.

Forskolin, a diterpene that activates adenylyl cyclases and enhances neurotransmitter release at the MF synapse, was also used to examine the correlation between the [Zn2+]t and neurotransmitter release. Figure 4B shows grouped data of five experiments in which a 10 min application of 50 μm forskolin resulted in a great enhancement of neurotransmitter release and [Zn2+]t. At the peak of forskolin action, fEPSP and [Zn2+]t were 272 ± 57% and 289 ± 51% of baseline values, respectively.

It is well established that an increase in release probability underlies the short-term facilitation of synaptic transmission at many synapses. We therefore next examined whether [Zn2+]t was sensitive to short-term facilitation of neurotransmitter release. Figure 4C shows the response of [Zn2+]t and fEPSPs evoked by a pair of stimulus trains each consisting of 5 pulses. The second train evoked a greater [Zn2+]t (the amplitude of the second [Zn2+]t was measured from the baseline directly preceding the second train). The facilitation of both [Zn2+]t and fEPSP was dependent on the interval between the two trains, as shown in Fig. 4D. At an interval of 0.1 s, the second [Zn2+]t was 154 ± 29% (n= 6) larger than the first [Zn2+]t. At an interval of 2 s, the facilitation of [Zn2+]t decreased to 80 ± 18% (n= 6). Over the entire interval range, the [Zn2+]t was more sensitive than the fEPSP in revealing enhancement of neurotransmitter release. Figure 4E summarizes the data from experiments with paired-train protocols as well as with the application of 4-AP, forskolin, DCG IV and 4 mm[Ca2+]o. On average, DCG IV (1 μm) reduced [Zn2+]t and fEPSP by 84 ± 3% and 87 ± 10% (n= 4), respectively. With the application of 4 mm[Ca2+]o, only those experiments without alternation of presynaptic fibre volley after the increase in [Ca2+]o were analysed. The mean [Zn2+]t and fEPSP were enhanced by 36 ± 10% and 33 ± 8% (n= 4), respectively.

To further quantify the co-release of zinc and glutamate, we measured the [Zn2+]t evoked by individual action potentials. Figure 5A shows the average of 20 [Zn2+]t traces evoked by a train of five action potentials at 5 Hz. The interval of 200 ms between consecutive stimuli enables the isolation of the zinc release corresponding to each individual action potential. Figure 5B and C demonstrates a linear relationship between the fEPSP for each action potential and [Zn2+]t, which were measured from the baseline directly preceding the stimulus. Both [Zn2+]t and fEPSP were normalized to the value obtained with the 5th stimulus. On average, the [Zn2+]t values corresponding to the 1st to the 4th action potentials were 24 ± 7%, 39 ± 6%, 60 ± 9% and 76 ± 6% of the 5th value (n= 8), respectively. The corresponding fEPSPs were 19 ± 6%, 35 ± 10%, 57 ± 10% and 78 ± 11% of the 5th value (n= 8). Our recording indicates that FluoZin-3 molecules are able to bind zinc molecules released even by the first action potential. This finding is significant because it demonstrates that the slight dissociation of the facilitation ratio (Fig. 4D) cannot be attributed to less zinc binding of the dye during the first few action potentials due to competition from extracellular zinc binding sites. Other factors, such as desensitization of glutamate receptors during the train of high frequency stimuli, or saturation of the fEPSP may therefore contribute to the result. Due to the significant short-term facilitation that occurs when the 33 Hz train is applied, the 4th and 5th fEPSPs may approach saturation, leading to a fEPSP measurement that may less accurately reflect the change in neurotransmitter release. Taken together, these data show a good correlation between fEPSP and [Zn2+]t over quite a wide range of glutamate release and that zinc is co-released with glutamate. Therefore, the [Zn2+]t obtained with FluoZin-3 can be used as a sensitive surrogate indicator of exocytotic glutamate release by action potentials at the MF synapse.

Figure 5.

Zinc release evoked by individual action potentials
A, averaged trace (n= 20) of [Zn2+]t evoked by a train of 5 action potentials at 5 Hz. Inset shows superimposed fEPSPs corresponding to each stimulus. B, summary data of the [Zn2+]t and fEPSP for each action potential within the stimulation train. Both [Zn2+]t and fEPSP were normalized to the value obtained with the 5th stimulus, respectively. C, summary plot of [Zn2+]tversus fEPSP for each action potential, indicating a linear correlation between [Zn2+]t and glutamate release.

Discussion

Zinc accompanies glutamate release at the mossy fibre synapse

Many previous studies in hippocampal slices have suggested that zinc is co-released with glutamate when a strong stimulation is applied (Assaf & Chung, 1984; Howell et al. 1984; Thompson et al. 2000; Li et al. 2001; Ueno et al. 2002; Kay, 2003). However, the stimulation protocols in those studies were typically far beyond the physiological range and the dynamic range of the zinc signal obtained was too slow to determine whether this signal reflected release from pre- or postsynaptic sites, or to determine the degree of extracellular zinc accumulation. For the same reason, it has remained unclear whether the actions ascribed to zinc release, including a range of effects on the kinetics of ion channels and transmitter receptors, are ever attained during normal synaptic signalling. More recently, strong fluorescent zinc signals evoked by single action potentials were reported (Quinta-Ferreira et al. 2004). In these studies, the membrane-permeable dye TSQ and the zinc chelator TPEN used may not exclusively reflect zinc in the extracellular compartment. Here we used a zinc indicator and chelators that are membrane impermeable as well as mice deficient in Znt3 to demonstrate that zinc from the vesicular compartment is released from mossy fibre terminals by action potentials with a stimulation protocol similar to physiological synaptic activity, and thus may be considered as a cotransmitter at these hippocampal synapses.

In a previous study using the same zinc indicator, only minimal presynaptic release of zinc was detected despite the delivery of strong and prolonged stimulation (100 Hz stimulation for 10 s, or 50 mm KCl; Kay, 2003). To account for the minimal zinc signal, a hypothesis was raised that little zinc was actually released into the synaptic cleft during exocytosis, and that most zinc ions remained bound to the externalized presynaptic membrane after release (Kay, 2003). In our experiments we observed large, well-defined zinc signals with a stimulus train as low as five action potentials. Moreover, the effects of zinc chelators on the kinetics of [Zn2+]t (Fig. 2), and in particular the delayed time to peak, are evidence in support of the free diffusion of zinc after exocytotic release. If zinc ions remained membrane bound after vesicle fusion and were even still detectable by dye molecules, then the peak of the [Zn2+]t signal should immediately follow the peak of the fEPSP, and the application of zinc chelators should only alter its peak amplitude and decay phase, but not the peak latency. In contrast, as shown in Fig. 2C and D, the peak latency ranges from 45 to 160 ms and is dependent on the zinc-chelating strength. Such a delayed response is unlikely to be due to a slow on–off reaction time of the dye to zinc. The structure of FluoZin-3 resembles that of the Ca2+ indicator fluo-4, a derivative of the fast Ca2+ dye fluo-3 (Eberhard & Erne, 1989). Although precise on–off rates have not yet been quantified, FluoZin-3 is believed to react to zinc with an extremely fast on-time (personal communication, Molecular Probes, Inc.). One explanation for the sharp contrast between the results may be the different experimental approaches used in these two studies. We stimulated mossy fibre synapses (as confirmed by simultaneous fEPSP recordings) and collected the zinc signal from the synaptically rich stratum lucidum, while in Kay's study, zinc release was investigated in the hilus of the dentate gyrus where the density of MF synaptic connections is far lower than that in the mossy fibre bundle forming the stratum lucidum. Along with the more effective stimulation and recording at this site, the relatively large size of the signal offered the additional advantage of obtaining data by stimulating at less than maximal strength, as evidenced by the control experiments which excluded artifacts related to autofluorescence and tissue light scattering encountered when strong stimulation is applied. In any case, although the zinc signals obtained varied considerably, both studies confirm the existence of extracellular chelatable zinc.

Kinetics of the evoked zinc fluorescence signal

The sensitivity of our method allows us to examine the dynamics of zinc release in much more detail. As shown in Fig. 1B, in the absence of any zinc chelator, the elevated [Zn2+]t evoked by a train of 20 stimuli lasted over 10 s. The decay exhibits two components, a fast and a slow component. The former has a mean time constant of about 1 s. The latter lasts much longer. The interpretation of this biphasic decay depends upon several still unknown parameters and requires consideration of the anatomy of MF terminals. The main body of the presynaptic MF bouton envelops a large multiheaded postsynaptic spine (Henze et al. 2000). Thus [Zn2+]t is a summation of the zinc signal from multiple compartments including the synaptic cleft beneath the site of vesicle exocytosis, the extra-synaptic cleft away from the release site, and the extracellular space surrounding the MF bouton. The slow component of the [Zn2+]t may represent saturation of the zinc signal (from compartments close to the release site) where FluoZin-3 is saturated because of its high zinc affinity. According this model, the free zinc concentration after a train of 20 action potentials remains higher than the dye concentration (1 μm) over the 10 s period. Alternatively, the zinc dye may not be saturated and the biphasic decay of [Zn2+]t may indicate different rates of zinc clearance among the compartments. Finally, if FluoZin-3 is able to detect zinc ions bound to high affinity fixed extracellular zinc binding sites on the pre- or postsynaptic membrane, then it is possible that a fraction of [Zn2+]t is contributed by the exposure of dye to this form of released zinc. For example, NMDA receptors can be potential targets. It has been shown that the IC50 of voltage-independent zinc inhibition of NMDA receptors could be as low as the nanomolar range (Williams, 1996; Chen et al. 1997; Paoletti et al. 1997; Zheng et al. 2001). In this case, our experimental data are consistent with the existence of a strong immobile extracellular zinc buffer at these synapses, since about 90% of the total [Zn2+]t is contributed by the slow component.

Since FluoZin-3 will be saturated when the zinc concentration is higher than the dye concentration applied, we tested the low affinity zinc indicators FluoZin-2 and Newport Green DCF (Kd= 1 μm). In both cases, we were not able to detect the zinc transient signal. The failure to reveal a zinc signal with low affinity zinc dyes should not simply be interpreted as a low extracellular zinc concentration after the release event. Two factors may contribute to the failure in revealing a zinc signal with low affinity zinc dyes. First, the quantum yield of FluoZin-2 and Newport Green DCF to zinc in the presence of Ca2+ and Mg2+ is much lower than FluoZin-3. As measured in ACSF, the Fmax/Fmin ratio for FluoZin-3 was about 54 while the value for Newport Green DCF was only 1.5 (Fmax was measured by adding a saturating concentration of ZnSO4 to the ACSF containing 1 μm dye; Fmin was measured by adding 125 μm EDDA to the dye–ACSF mixture; data not shown). In other words, FluoZin-3 generates a ΔF/F that is 35 (i.e. 54/1.5) times larger than Newport Green DCF once zinc is bound to the dye molecule in ACSF. Secondly, because of its high affinity to zinc, many more FluoZin-3 molecules bind to zinc as compared to Newport Green DCF. Therefore, in response to the same stimulus, the ΔF/F obtained with Newport Green DCF is at least 35 times smaller than FluoZin-3, well below the noise level of our optical detection system. Due to the uncertainties listed above, we are unable to use these lower affinity dyes to pinpoint a narrow range of the amount of zinc released per impulse in this study. Further experiments using zinc dyes with intermediate affinity combined with simulation of the zinc–dye interaction in a multicompartment model may be helpful in quantifying the dynamics of zinc release and clearance.

Zinc functions as a coneurotransmitter at the MF synapse

By establishing its exocytotic release by physiological stimuli, we add to the extensive evidence that zinc shares many of the essential properties of conventional neurotransmitters such as glutamate in the modulation of neuronal network signalling under normal and pathological conditions. The effect of zinc will vary according to the concentration reached in the various compartments of the MF synapses. Among a long list of zinc-sensitive molecules, NMDA and GABA receptors are the most prominent targets. Application of exogenous zinc potently inhibits the NMDA receptor (Peters et al. 1987; Westbrook & Mayer, 1987; Christine & Choi, 1990) by promoting its desensitization (Zheng et al. 2001). The modulation of NMDA receptors by endogenously released zinc has also been demonstrated in in vitro electrophysiology experiments at the MF synapse (Vogt et al. 2000). Similar to its action on NMDA receptors, zinc at micromolar concentrations greatly inhibits GABAA receptors as well (Westbrook & Mayer, 1987). Zinc and GABA have been shown to be contained within the same mossy fibre synaptic varicosities (Ruiz et al. 2004). These authors also demonstrated that endogenous zinc modulates GABAA receptors in the CA3 area. The kinetics of voltage-gated ion channels in hippocampal neurones are also modulated by zinc (Mayer & Vyklicky, 1989). In addition to the modulation of inhibitory and excitatory ionic channels at extracellular sites, zinc, as a catalytic and/or structural cofactor for numerous zinc-finger proteins, is capable of regulating gene expression and, indirectly, cellular signal transmission or cell viability (Matthews & Sunde, 2002; Weiss et al. 2000). Some studies report that the translocation of zinc into postsynaptic neurones during HFS may even trigger the induction of long-term potentiation (LTP) (Li et al. 2001). The effect of zinc will also vary depending on the molecular component of the receptor (Draguhn et al. 1990; Smart et al. 1991). In temporal lobe epilepsy, an alteration in the GABAA receptor subunit composition, which increases its zinc sensitivity, may be responsible for the zinc-mediated collapse of gating by GABAergic inhibition in the dentate gyrus pathway, favouring aberrant network synchronization (Buhl et al. 1996; Coulter, 2000; Lin et al. 2001).

An alternative method for measuring the dynamics of presynaptic exocytosis

In this study we have validated a technique using zinc-sensitive dyes to directly assess exocytotic neurotransmitter release and extend our ability to explore the presynaptic component of neurotransmitter release in acutely prepared brain slices. Measurement of evoked zinc release as an alternative method in determining presynaptic exocytosis has been previously explored by Quinta-Ferreira et al. (2004) where the membrane-permeable zinc dye TSQ was used. As interpreted by the authors, the transient fluorescence decrease corresponding to a single action potential is due to the dissociation of the TSQ–zinc complex following vesicle fusion, caused by the large difference between the intravesicular and extracellular free zinc concentration. This may explain why a paired-pulse decrease in the TSQ signal instead of a paired-pulse facilitation was observed in their study. Since the degree of dissociation of TSQ will depend on the build-up of extracellular zinc concentration during stimulation, it can be predicted that when repetitive stimulation is applied, the increased extracellular zinc concentration may reduce the dissociation of the TSQ–zinc complex and thereby yield a smaller signal for the next release event. In contrast, using the membrane-impermeable zinc dye FluoZin-3 with an extremely high affinity for zinc, FluoZin-3 molecules will bind to any free extracellular zinc available. As demonstrated in Figs 4 and 5, the zinc signal linearly reflects the amount of glutamate released (fEPSP) over quite a wide range. This correlation indicates that the method may prove useful in the future to probe the mechanism underlying short-term synaptic plasticity at the MF synapse.

Appendix

Acknowledgements

This work was supported by the National Institute of Neurological Disorders and Stroke (NINDS) NS97209 (J.L.N.).

Ancillary