Endogenous zinc depresses GABAergic transmission via T-type Ca2+ channels and broadens the time window for integration of glutamatergic inputs in dentate granule cells

Abstract Zinc actions on synaptic transmission span the modulation of neurotransmitter receptors, transporters, activation of intracellular cascades and alterations in gene expression. Whether and how zinc affects inhibitory synaptic signalling in the dentate gyrus remains largely unexplored. We found that mono- and di-synaptic GABAergic inputs onto dentate granule cells were reversibly depressed by exogenous zinc application and enhanced by zinc chelation. Blocking T-type Ca2+ channels prevented the effect of zinc chelation. When recording from dentate fast-spiking interneurones, zinc chelation facilitated T-type Ca2+ currents, increased action potential half-width and decreased spike threshold. It also increased the offset of the input–output relation in a manner consistent with enhanced excitability. In granule cells, chelation of zinc reduced the time window for the integration of glutamatergic inputs originating from perforant path synapses, resulting in reduced spike transfer. Thus, zinc-mediated modulation of dentate interneurone excitability and GABA release regulates information flow to local targets and hippocampal networks.

J Physiol 592.1 for endogenous zinc are the triggering of apoptotic pathways (Aizenman et al. 2000), regulation of gene expression (Tsuda et al. 1997) and epileptogenesis (Buhl et al. 1996), progression of Alzheimer's disease (Duce et al. 2010) and altered pain signalling (Nozaki et al. 2011). Zinc directly inhibits a number of ionotropic receptors commonly found at central synapses including NMDA (Paoletti et al. 2000;Gielen et al. 2009), GABA A (Barberis et al. 2000;Hosie et al. 2003;Ruiz et al. 2004), glycine (Mayer & Vyklicky, 1989) and kainate receptors (Mott et al. 2008;Veran et al. 2012). It also inhibits glutamate transporters (Spiridon et al. 1998) and the neuronal K + -Cl − co-transporter KCC2 (Hershfinkel et al. 2009), as well as a wide variety of voltage-activated channels that are important for the fine-tuning of cellular excitability and neurotransmitter release. For example, zinc efficiently inhibits T-type Ca 2+ channels containing the Ca v 3.2 subunit (Sun et al. 2007;Traboulsie et al. 2007) or K + channels containing the K v 3.1 subunit (Gu et al. 2013) or the K v 1.1 subunit (Imbrici et al. 2007). Finally, it potently inhibits GABA A receptors lacking the γ-subunit (Smart et al. 1991). Despite this wealth of evidence, and the demonstration that endogenous zinc can affect glutamatergic transmission in the hippocampus (Vogt et al. 2000;Molnar & Nadler, 2001a), the amygdala (Kodirov et al. 2006) and the retina (Wu et al. 1993), it remains unclear how it may alter the functionality of GABAergic networks. Hitherto, studies undertaken in the hippocampal formation have mainly focused on monosynaptic inputs (Vogt et al. 2000;Ruiz et al. 2004;Huang et al. 2008) and there has as yet been little attempt to dissect how zinc modulates di-synaptic events, particularly those mediated by GABA A receptors. Furthermore, very few studies have provided a detailed characterization of the actions of zinc in dentate interneurones (Koh et al. 1995;Berger et al. 1998) and none have investigated how zinc signals might modulate the window of integration of excitatory synaptic inputs and spike transfer in principal cells.
Zinc is extremely abundant in the axon of dentate granule cells (or mossy fibres), which ramify extensively and make contact 'en passant' or via filopodia to a large variety of interneurones in both the hilus and stratum lucidum (Freund & Buzsaki, 1996;Szabadics & Soltesz, 2009). Morphological and electrophysiological analysis of GABAergic interneurones in the dentate gyrus revealed fast-spiking, parvalbumin-expressing cells that preferentially innervate the somatic and perisomatic regions of granule cells, and are essential for the generation of gamma oscillations (Bartos et al. 2007(Bartos et al. , 2011. Other classes of interneurones in the dentate gyrus with axons that are confined more distally to the termination zone of perforant path inputs include molecular layer-associated interneurones (MOPP) and hilar perforant path-associated interneurones (HICAP). Mossy fibre-associated interneurones that are found in stratum lucidum largely contribute to feed-forward inhibition in CA3 pyramidal neurones (Buzsaki, 1984;Acsady et al. 1998), whereas hilar and dentate interneurones innervated by recurrent mossy fibre collaterals provide a feedback input that regulates dentate granule activity (Penttonen et al. 1997;Doherty et al. 2004;Ewell & Jones, 2010;Sambandan et al. 2010). Electrical stimuli designed to release zinc at this recurrent mossy fibre pathway did not affect GABA A receptor-mediated currents evoked in granule cells by photo-uncaging GABA (Molnar & Nadler, 2001b). This result does not, however, exclude the possibility that zinc may alter GABAergic signalling in granule cells by modulating interneurone activity and thus di-synaptic inhibition.
We demonstrate here that zinc depresses GABAergic transmission to granule cells and that this modulation occurs via T-type Ca 2+ channels. We also show that zinc chelation selectively broadens the action potential waveform in dentate fast-spiking interneurones and that it narrows the window for the integration of glutamatergic inputs originating from perforant path synapses. Our results unravel a phenomenon whereby zinc depresses interneurone excitability and modulates spike routing to the hippocampus proper.

Hippocampal slice preparation
All animal procedures strictly followed University College London (UCL) Research Ethics Committee regulations. Sprague-Dawley rats (Harlan Laboratories Ltd, Oxon, UK) aged 20-40 days were killed by overdose of sodium pentobarbital injected intraperitoneally (100 mg kg −1 ) and rapidly decapitated in accordance with the UK Animals (Scientific Procedures) Act, 1986. Transverse 250-300 μm thick slices were obtained from both hippocampi using a vibratome [Leica VT-1200S; Leica Biosystems (UK) Ltd, Milton Keynes, UK]. Slices were kept at 35 • C for 30 min after slicing and then at room temperature (22 • C). For the dissection and storage of slices, the solution contained NaCl (87 mM), NaHCO 3 (25 mM), D-glucose (10 mM), sucrose (75 mM), KCl (2.5 mM), NaH 2 PO 4 (1.25 mM), CaCl 2 (0.5 mM) and MgCl 2 (7 mM). For experiments, the slices were superfused with physiological saline containing NaCl (125 mM), NaHCO 3 (25 mM), D-glucose (25 mM), KCl (2.5 mM), NaH 2 PO 4 (1.25 mM), CaCl 2 (2 mM) and MgCl 2 (1 mM), equilibrated with 95% O 2 /5% CO 2 . All solutions were prepared using water purchased from Fisher Scientific UK Ltd (Loughborough, UK). recordings were obtained from granule cells and dentate interneurones under infrared differential interference contrast imaging at 22 • C. We placed bipolar stainless steel stimulating electrodes in stratum lucidum of CA3b and stratum granulosum of the dentate gyrus. Synaptic currents were recorded with an Axopatch 200-B amplifier (Molecular Devices LLC, Sunnyvale, CA, USA), filtered at 2 kHz (internal 4-pole low-pass Bessel filter), and sampled at 10 kHz. Data were acquired and analysed offline using the Labview software environment (National Instruments Corp., Austin, TX, USA). Access resistance was monitored throughout the experiments and was <20 M ; results were discarded if it changed by >20%. Junction potentials were not corrected. The pipette solution used for IPSCs recorded at a holding potential (V holding ) of −70 mV contained CsCl (120 mM), QX314 Br (5 mM), NaCl (8 mM), MgCl2 (0.2 mM), Hepes (10 mM), EGTA (2 mM), MgATP (2 mM), Na 3 GTP (0.3 mM) (pH 7.2, osmolarity 310 mOsm l −1 ). IPSCs evoked by stratum lucidum stimulation (20 μs, 20-100 μV) was analysed only if currents were reversibly depressed by >40% by (2S,2 R,3 R)-2-(2 ,3 -dicarboxycyclopropyl)glycine (DCG-IV; 1 μM) consistent with the selective sensitivity of mossy fibre synapses to group II metabotropic receptor agonists. To measure IPSCs recorded near the reversal potential for glutamate receptors in granule cells (V holding = 0 mV), CsCl was substituted with Cs-gluconate. To record T-type Ca 2+ currents, the pipette solution contained K-gluconate (130 mM), Hepes (20 mM), NaCl (8 mM), EGTA (0.2 mM), QX-314 Br (5 mM), MgATP (2 mM) and Na 3 GTP (0.3 mM). The perfusion medium contained TEA (5 mM), 4-AP (2 mM) and TTX (1 μM) to block K + and Na + channels and nifedipine (20 μM) to block L-type voltage-gated Ca 2+ channels. Slices were maintained within an interface chamber on a Petri dish containing ω-agatoxin IVA (0.5 μM), ω-conotoxin GVIA (2 μM) and SNX-482 (0.5 μM) to block P/Q-, N-and R-type voltage-gated Ca 2+ channels, respectively, for at least 2 h prior to recording. GABA A and glutamate receptors were blocked by the addition of bicuculline methiodide (10 μM) and kynurenic acid (2 mM). V holding was set to −70 mV and a 500 ms pre-pulse to −90 mV followed by a 1 s depolarizing step was applied from −90 mV to +40 mV, with a 10 mV increment. Series resistance was compensated (range: 11-23 M ; voltage clamp 60-70% correction, time lag 10 μs). The solution for recording from granule cell somata and fast-spiking interneurones in current-clamp mode contained K-methanesulfonate (145 mM), EGTA (2 mM), Hepes (10 mM), NaCl (5 mM), MgCl 2 (2 mM), Na 2 ATP (2 mM), Na 3 GTP (0.5 mM) and Na-phosphocreatine (5 mM). For each neurone, the I-V relation was determined by measuring the amplitude of steady state voltage deflections elicited by a series of hyperpolarizing and depolarizing current steps. The input resistance was obtained by fitting the linear portion of the I-V relation. 'Sag' ratios were determined as the ratio between the steady state voltage and peak voltage in response to a current injection that resulted in a membrane potential negative to −120 mV. To determine the membrane time constant (τ m ), averaged deflections of hyperpolarizing potentials (less than −10 mV) were fitted by a mono-exponential function. Maximum firing rates were determined by the interval between the first and second action potentials at a current step that did not inactivate Na + channels. Action potential amplitudes were measured from V m preceding the current pulse. The half-spike width was determined as the duration at half-spike amplitude and was calculated offline. To determine the action potential threshold, phase plots were constructed by plotting the first derivative of the membrane potential versus the membrane voltage (Bean, 2007). Local pressure application of L-glutamic acid monosodium salt (Glu; 100 μM in control perfusion solution) was delivered via a patch pipette connected to a Picospritzer (10-50 ms, 5-30 p.s.i.; General Valve Corp., Fairfield, NJ, USA). The pipette was first positioned in stratum lucidum in the vicinity of the stimulating electrode; after a series of puffs, it was moved to stratum granulosum at <300 μm from the other stimulating electrode. The bath perfusion was arranged to keep the granule cell body recorded upstream of this position. To calculate the frequency of Glu-evoked IPSCs, inter-event intervals were measured just after the onset of the puff and the resulting frequency subtracted from that of spontaneously occurring IPSCs without glutamate puff. Spontaneous IPSCs were identified as events exceeding the threshold of three times the standard deviation of the noise level for 80 s, in the absence of glutamate puff. To analyse the offset and gain in dentate neurones, the mean firing frequency was plotted against the injected current, thus yielding an input-output (I-O) relation for each cell. The I-O relation was then fitted over a range of firing frequencies with equations describing a logarithmic function represented by the slope of the I-O relation over the whole range of curve fitting (gain) and the x-offset (e.g. the amount of current required to bring the cell to fire). Control I-O relationships were subject to non-linear fitting using: where f = firing frequency (Hz), x = injected current (pA), k = gain (Hz pA −1 ) and exp A/k = (x-offset). Fitting parameters k and A were then used to constrain the fitting of the I-O relationship during chelation of zinc, using: where the gain is equal to m * k, and C is the shift in offset ( offset ) (pA). For spike-timing experiments, one stimulus J Physiol 592.1 electrode was positioned in stratum lucidum to activate mossy fibres and the other in the outer molecular layer >300 μm away from the recorded granule cell soma to activate the lateral perforant path. Both pathways were activated at different time intervals (from −30 ms to +30 ms, in 10 ms increments) for 10 cycles, in a control condition and after chelation of zinc. Spike probability and the amplitude of subthreshold potentials evoked by single shocks were analysed offline in responses acquired in separate channels.

Statistical tests
The results are presented as the mean ± standard error of the mean (S.E.M.). Statistical differences were determined by two-tailed Student's t test for two-group comparisons unless otherwise indicated. Data were considered significant if P < 0.05.

Figure 1. Chelating zinc with TPEN facilitates GABAergic IPSCs in granule cells
A, distributions of latencies of PSCs evoked by stimuli delivered in stratum lucidum of the CA3b region (white) or in stratum granulosum near the tip of the 'V' shape formed by the layer (grey). Distributions were fitted with two Gaussian curves (data from 15 neurones). Sample recordings show 20 consecutive PSCs superimposed (grey) as well as the mean PSC (black). B, switching the stimulation frequency from 0.06 Hz to 0.6 Hz depresses evoked PSCs (n = 15). C, example traces from one granule cell recording (averages of 10 consecutive trials) showing the J Physiol 592.1

Extracellular zinc chelation enhances dentate interneurone excitability
Where are the synapses affected by zinc chelators and by which presynaptic neurones are they formed? The facilitation of stratum lucidum evoked IPSCs is consistent with an effect at mossy fibre-CA3 interneurone synapses or, alternatively, synapses made onto dentate interneurones by recurrent mossy fibres and axon terminals from hilar mossy cells. Such interneurones also have excitatory synapses that show the peculiar metabotropic receptor agonist sensitivity of mossy fibres and extend their axons in stratum lucidum (Alle et al. 2001). To identify the location of presynaptic interneurones, we pressure-applied glutamate via a pipette positioned close to a site at which electrical stimulation evoked a di-synaptic IPSC in granule cells. D-APV was omitted from the perfusion solution in order to allow excitation of dendritic NMDA receptors (Weisskopf et al. 1993) in interneurones, whereas NBQX (20 μM) was included to minimize polysynaptic activity. We first performed control experiments in current-clamp to verify that pressure-applied glutamate in stratum lucidum (n = 3) or in the dentate molecular layer (n = 2) could recruit local interneurones (Fig. 3A and B). Further control experiments in granule cells held in voltage-clamp at −70 mV with a Cs-Cl containing pipette revealed IPSCs that reversed at 0 mV in response to glutamate puffs delivered >300 μM in the layer (Fig. 3C and D). These responses are likely to reflect the polysynaptic activation of GABA A receptors in granule cells consecutive to firing in local interneurones. When applied in various locations near the stratum lucidum stimulus site, the glutamate puff failed to elicit IPSCs in granule cells, in contrast to electrical stimuli ( Fig. 4A-C). However, in recordings from the same cells with the puff pipette repositioned in the dentate gyrus near the stratum granulosum stimulus electrode, glutamate puffs elicited bursts of IPSCs, the amplitude of which increased by 14.6 ± 7.1% following the superfusion of TPEN (1 μM, n = 6; P < 0.05) (Fig. 4C  and D). TPEN did not affect the frequency of IPSCs evoked by such glutamate puffs ( Fq = 0.1 ± 1.4 Hz, n = 6; P > 0.05) (Fig. 4E). In a separate set of experiments (n = 4), glutamate was replaced by KCl (3 M) on the assumption that a local build-up in extracellular K + would depolarize mossy fibres and thus mimic electrical stimulation. KCl puffs elicited bursts of IPSCs in granule cells irrespective of the site of application (Fig. 4F) and superfusion of TPEN (1 μM) increased their amplitude (Fig. 4G). A similar enhancement of glutamate-evoked responses by TPEN was obtained when holding granule cells near the reversal potential for glutamate receptors with a Cs-gluconate-based pipette solution (V holding = 0 mV, E Cl = −70.3 mV; n = 4) (Fig. 5). Altogether, these experiments show that dentate interneurones with terminals that release GABA onto granule cells can be recruited by focal glutamate application or action potentials in mossy fibre collaterals and that zinc modulates this process.
Dentate interneurones provide a strong inhibitory input to granule cells and receive zinc-containing boutons (Ribak et al. 1990;Ribak & Peterson, 1991). We recorded from interneurones located at the border between the granule cell layer and the polymorphic layer of the hilus, systematically labelled them with biocytin and analysed the zinc sensitivity of their pharmacologically isolated low-voltage activated Ca 2+ currents. Similar experiments were performed in granule cells for comparison (Fig. 6A). Confocal analysis of Z-stack projections from recorded interneurones revealed a large soma size (35-45 μm), thick B, cumulative distribution of PSC frequency in the same cell before (baseline) and after glutamate puffs in stratum lucidum and stratum granulosum. C, data pooled from six neurones showing the increase in PSC frequency after focal glutamate application in stratum granulosum. D, sample recordings obtained in one neurone (superimpositions of four consecutive trials) showing the facilitation of Glu-evoked PSCs after chelation of zinc with TPEN (1 μM). Glu-evoked PSCs are abolished by the addition of bicuculline (BIC; 10 μM). Selected portions of the traces are shown at higher magnification (bottom). The bar graph summarizes data obtained in six neurones ( * P < 0.05, paired t test). E, the effect of TPEN in the same neurones is not accompanied by a change in glutamate-evoked PSC frequency. F, G, sample traces taken from two recorded neurones showing that focal KCl application (arrows; 50 p.s.i., 50 ms) evokes bursts of PSCs whether delivered in stratum granulosum or stratum lucidum. Focal KCl application in stratum lucidum elicits bursts of PSCs that are enhanced by superfusion of TPEN (1 μM). apical and basal dendritic trunks through the granular region and extending towards the stratum moleculare and the hilus, and axonal projections that ramified in the granular layer, with the exception of one cell, which had axon ramifications in stratum moleculare. Granule cells and dentate interneurones were held at V holding = −70 mV and a 500 ms pre-pulse followed by a 1 s step from −90 mV to +20 mV with 10 mV increments was applied, in the presence of a cocktail of channel blockers and ionotropic receptor antagonists (see Methods). This protocol activated transient inward currents, the amplitude of which peaked and declined when the pipette potential was >0 mV (Fig. 6B and C).

Endogenous zinc broadens the time window for integration of perforant path inputs
Synaptic inhibition in dentate granule cells mainly operates by the shunt of concurrently activated excitatory inputs caused by the reduction in membrane resistivity and dendritic space constant associated with the GABA A receptor-mediated conductance (Rall, 1964;Barrett & Crill, 1974). Furthermore, the reversal Figure 6. Zinc chelation facilitates T-type Ca 2+ currents in fast-spiking interneurones A, digital reconstruction of a granule cell forming a network of mossy fibre collaterals within the hilus (left) and of a dentate interneurone (right), the cell body of which is located at the border between the granular layer and the hilar region. Axonal arbours are coloured in blue. The location of the cells is shown in relation to the different dentate layer boundaries. GCL, granule cell layer; ML, molecular layer. Scale bars: 100 μm. B, examples of I-V relations in a granule cell (left) and a dentate interneurone (right) showing the activation of a transient inward current evoked by depolarizing voltage steps. The voltage step protocol is represented above traces. C, normalized peak current amplitude plotted against the pipette potential for cells shown in B. D, zinc chelation with TPEN (1 μM) selectively enhances the amplitude of low-voltage activated T-type Ca 2+ currents in fast-spiking interneurones (n = 7, right), but not granule cells (n = 8, left) ( * P < 0.05, Student's paired t test). E, F, summary bar graphs and data from individual neurones showing the effects of TPEN on the amplitude of Ca 2+ currents at −50 mV (E) and −40 mV (F) in granule cells (n = 7, left) and interneurones (n = 6, right). Averages of three sweeps at each potential are shown for the control condition and TPEN (traces were truncated for clarity). 105.2 ± 3.9 101.6 ± 5.3 110.9 ± 5.3 108.1 ± 4.8 Firing frequency, max (s −1 ) 100.1 ± 11.2 101.4 ± 9.1 131.0 ± 9.0 130.0 ± 12.0 potential for GABA A receptor activation is ∼16 mV more depolarized than the resting membrane potential (E resting = −82.5 ± 2.7 mV) (see Misgeld et al. 1986;Soltesz & Mody, 1994), which implies that GABAergic events remain largely depolarizing. Because mossy fibre collaterals can rapidly recruit fast-spiking interneurones, ensuring a precise temporal integration of the excitatory input (Geiger et al. 1997;Calixto et al. 2008), we tested the hypothesis that zinc-mediated modulation of interneurone activity regulates synaptic integration in granule cells and ultimately spike routing to CA3. We positioned a tungsten electrode in stratum lucidum to stimulate mossy fibres and another in the outer molecular layer to stimulate the lateral perforant path. When recording from granule cells in current-clamp mode, electrical stimulation delivered in stratum lucidum elicited a depolarizing potential able to initiate action potentials. Superfusion of bicuculline (10 μM, n = 3) abolished this synaptic response and firing (Fig. 8A) in a manner similar to the effect of blocking AMPA/kainate receptors with NBQX (20 μM, n = 3; data not shown) and in good agreement with the data obtained in the voltage-clamp configuration ( Fig. 1C and E). These observations were consistent with di-synaptic GABAergic inputs to granule cells that were conveyed by mossy fibre collateral-interneurone synapses, the depolarizing nature of which could trigger action potentials. They also revealed some degree of feed-forward activation of inhibitory neurones by the outer molecular layer stimulus. When chelating zinc with TPEN (1 μM) the probability for evoking action potentials in granule cells increased from 0.16 ± 0.05 to 0.31 ± 0.08 (n = 4; P < 0.04) without affecting action potentials generated by the activation of lateral perforant path synapses (Fig. 8B). We adjusted the stimulus intensities so that simultaneous activation of the two pathways resulted in an approximately 50% chance of the neurone spiking (Fig. 8C). We then measured the spike probability while systematically varying the inter-stimulus interval.

Figure 7. Zinc chelation modulates the action potential waveform and the offset of the I-O relation in fast-spiking interneurones but not granule cells
A, superimposed action potential waveforms (averages of 20 sweeps) in a fast-spiking interneurone and a granule cell, in the control condition (thin black trace) and after chelation of zinc with TPEN (1 μM; thick grey trace). B, C, bar graphs summarize the broadening of action potentials in all fast-spiking interneurone recordings (n = 8) and no effect in granule cells (n = 9) ( * * P < 0.01, Student's paired t test). D, E, time-derivative of the action potentials displayed in A, plotted against the membrane potential. The thin black trace is the phase plot in control condition and the thick grey trace is that after the addition of TPEN (1 μM). The arrow denotes the shift of the action GABAergic conductance. Although these experiments do not directly link the changes in granule cell excitability to those happening in presynaptic interneurones, they underline the fact that zinc facilitates granule cell excitability depending on the timing of the combined activation of a glutamatergic monosynaptic input and a di-synaptic GABAergic input, respectively. Thus, endogenous zinc regulates information transfer to local targets in the dentate gyrus with significant consequences on spike routing from granule cells to CA3.

Discussion
We found that endogenous zinc depresses mono-and di-synaptic GABAergic signalling in granule cells by acting on T-type Ca 2+ channels. We also found that zinc chelation selectively modulated action potential threshold properties in fast-spiking interneurones and that it narrowed the time window for integration of glutamatergic perforant path inputs in granule cells. It is highly unlikely that our findings of chelation-mediated modulation of dentate granule cell excitability can be explained by an effect on GABA A receptors in somata and dendrites of granule cells themselves. Firstly, their somatic holding current was not affected by zinc chelation, which is consistent with the relatively low zinc sensitivity of GABA A receptors in this cell type at this postnatal age (Buhl et al. 1996;Kapur et al. 1999). Secondly, the facilitation of evoked IPSCs consecutive to zinc chelation was of roughly the same order of magnitude whether a highor low-chloride concentration was used in the pipette solution. Thirdly, zinc-mediated modulation of evoked GABAergic signalling in the dentate gyrus was absent if slices were treated with a T-type Ca 2+ channel antagonist. Thus, zinc inhibits GABAergic signalling via direct binding to GABA A receptors at a monosynaptic input (Ruiz et al. 2004) or by regulating the excitability of presynaptic interneurones at a di-synaptic pathway, as shown here. These findings imply that feed-forward and feedback inhibition in the dentate gyrus is tightly regulated by zinc. However, with the current experiment design, we cannot rule out the possibility that stratum lucidum stimulation recruited a proportion of back-projecting interneurones providing long-range cross-regional inhibition from CA1 to hilar regions (Sik et al. 1994;Szabadics & Soltesz, 2009). Lastly, the ultimate demonstration that the observed effects were directly linked to the vesicular release of zinc at mossy fibre-interneurone synapses would require the use of ZnT3 −/− mice, which lack the fraction of chelatable zinc from synaptic vesicles. Furthermore, we cannot exclude the possibility that non-vesicular release of zinc might contribute to the observed effects.
Our results showing that the T-type Ca 2+ channel antagonists mibefradil and NNC 55-0396 reduced evoked GABAergic synaptic transmission and occluded the enhancing effect of zinc chelation suggests that zinc inhibits low-voltage activated Ca 2+ channels in presynaptic dentate interneurones. In keeping with this, interneurones positioned in the inner molecular and granular layers show immunoreactivity for all Ca v 3 isoforms from the T-type Ca 2+ channel family (McKay et al. 2006;Vinet & Sik, 2006), let alone the dense localization of T-type Ca 2+ channel gene transcripts in this area (Talley et al. 1999) and the rich innervation of parvalbumin-immunoreactive interneurones by Timm-stained mossy fibre collaterals (Blasco-Ibanez et al. 2000;Seress & Gallyas, 2000).
Although T-type Ca 2+ channels are generally not involved in evoked neurotransmitter release, they can initiate slow release from non-axonal sites (Cueni et al. 2009) and recent evidence in dorsal horn neurones (Jacus et al. 2012) and layer III enthorinal cortex (Huang et al. 2011) has shown that they can potently inhibit glutamate release. Ca v 3.1-containing T-type Ca 2+ channels can also modulate the release of quanta from GABAergic synapses formed by parvalbumin-expressing and perisomatic-targeting interneurones onto CA1 pyramidal cells (Tang et al. 2011). However, the precise molecular mechanisms responsible for the role of presynaptic T-type Ca v 3.2 channels and their significance in evoked synaptic transmission would be best evaluated in interneurone-granule cell pairs in slices from transgenic mice that lack these channels.
Analysing spike shape and threshold as well as I-O relations revealed specific effects of zinc chelation in interneurones, but not in granule cells. The positive shift of neuronal offset in fast-spiking interneurones independently of gain was consistent with the removal of a fixed tonic conductance, the effect of which in the baseline potential threshold towards negative values. F, summary data showing the effect of TPEN (1 μM) on the action potential threshold in fast-spiking interneurone recordings (n = 8) and the absence of effect in granule cells (n = 9) ( * P < 0.05, Mann-Whitney U test.) G, examples traces obtained in a fast-spiking interneurone showing voltage deflections in response to depolarizing current steps that are sub-and supra-threshold to action potential initiation (control). A rheobase current of +20 pA elicited two action potentials, whereas injection of +50 pA triggered nine action potentials in the presence of TPEN (1 μM; grey traces). H, example of I-O relation obtained in a fast-spiking interneurone showing a shift in offset but no change in gain. I, effects of TPEN (1 μM) on the gain and offset of the I-O relation in fast-spiking interneurones (n = 7) ( * P < 0.05, Student's paired t test; n.s., non-significant.) J, as in I but expressed as a relative change. J Physiol 592.1 condition dampened neuronal firing over the entire time domain. Whether zinc-mediated modulation of T-type Ca 2+ channel activity caused the offset of the I-O relation in fast-spiking interneurones remains to be elucidated.
Our finding that TPEN narrowed the time window for the integration of excitatory synaptic inputs driven by lateral perforant path synapses has important ramifications for signal integration in the dentate gyrus. We showed that it was possible to position the stimulus electrode in stratum lucidum such that recurrent activation of mossy fibres would depolarize granule cells via di-synaptic GABAergic actions, as shown for monosynaptic connections (Chiang et al. 2012;Sauer et al. 2012). Activation of recurrent mossy fibre synapses and perforant path inputs evoked depolarizing postsynaptic potentials that summated, and yielded a significant increase in spiking probability in granule cells. When zinc was chelated with TPEN, a significant decrease in the contribution of perforant path inputs to the summated postsynaptic potential in favour of inputs activated by recurrent mossy fibres at stimulus intervals of >20 ms was observed. Taken together with the pro-excitatory effect of zinc chelation on granule cell firing (Fig. 6A), these results unravel a powerful homeostatic mechanism that tunes information transfer to dentate and CA3 microcircuits. Further, the modulation of dentate gyrus excitability by endogenous zinc has profound implications for developmental and pathological processes, in particular Figure 8. Zinc chelation narrows the time window for integration of perforant path inputs A, blocking GABA A receptors with bicuculline (10 μM) blocks action potentials evoked by stimuli delivered in stratum lucidum. B, conversely, chelation of zinc with TPEN (1 μM) increases this probability. ( * P < 0.05, Student's paired t test.) TPEN does not affect the probability for evoking action potentials elicited by stimulation of the outer molecular layer (P > 0.05, Student's paired t test). (Data presented from five neurones.) C, representative voltage traces taken from one neurone at different inter-stimulus intervals (−30 ms to +30 ms) in the control condition and after chelation of zinc with TPEN (1 μM). Vertical dashed lines indicate the time reference (t = 0) when electrical stimuli were delivered in the outer molecular layer. X, stratum lucidum stimulation. D, summary histogram showing the narrowing of the window for integration of perforant path inputs in the presence of TPEN (data from five neurones). The probability for evoking action potentials by stimulating the perforant path is reduced for inter-stimulus intervals >10 ms (P < 0.05, ANOVA).

Figure 9. Zinc chelation reduces the contribution of glutamatergic perforant path inputs to summated subthreshold potentials
A, B, summated postsynaptic potentials (averages of five consecutive sweeps) evoked by stratum lucidum stimulation (SL stim, blue arrow) and outer molecular layer stimulation (OML stim, red arrow) at two different inter-stimulus intervals (−30/+30 ms and −20/+20 ms), in control condition (black traces) and in the presence of TPEN (grey traces). Green vertical lines on top traces indicate the peak of the first and summated responses. Bar graphs show amplitudes of individual responses at both pathways expressed as percentages of the summated responses, in the control condition and following chelation of zinc with TPEN ( * P < 0.05, ANOVA). C, summary bar graph and data from individual experiments showing the effect of zinc chelation on individual responses. D, amplitude of the summated potential at different inter-stimulus intervals, in the control condition (white symbols) and after superfusion of TPEN (black symbols). * * P < 0.01, Student's t test). Data presented are from five neurones. J Physiol 592.1 epilepsy, which is associated with an increase in the sensitivity of GABA A receptors to inhibition by zinc, extensive mossy fibre sprouting and loss of interneurone populations.