Introduction

The entorhinal cortex (EC) conveys spatial, limbic and sensory information to the hippocampus, a structure responsible for important activities in the brain including learning, memory and spatial information coding. Excitatory projections from the EC to the dentate gyrus (DG) play a role in memory encoding, and they are affected in humans suffering from Alzheimer Disease (AD) and temporal lobe epilepsy, contributing critically to these pathologies (Paré et al., 1992; Llorens-Martín et al., 2014). In rodents, evidence suggests that the lateral perforant pathway (LPP) conveys more sensory related information, whereas the medial perforant pathway (MPP) conveys information related to spatial location and limbic signals that are involved in attention and motivation. Evidence suggests that LPP- and MPP-GC synapses have distinct electrophysiological properties, these differences potentially affecting the functional processing of information (Krueppel et al., 2011; Petersen et al., 2013; Kim et al., 2018). Indeed, these functional differences probably reflect their distinct inputs, as they contact the outer (LPP) or middle (MPP) third of the molecular layer of the DG (Paré et al., 1992; Burgess et al., 2002). Significantly, MPP-GC synapses are associated with a higher probability of release (Min et al., 1998) and the existence and/or activity of presynaptic NMDARs, which differs between the two types of fibres (Savtchouk et al., 2019). Interestingly, LPP and MPP synapses appear to influence pathological changes distinctly, with the former showing changes in release probability in models of epilepsy (Scimemi et al., 2006), preferential β-amyloid deposition (Reilly et al., 2003), earlier susceptibility to AD (Khan et al., 2014) and reduced plasticity with age (Froc et al., 2003). The mechanistic bases of these differences are virtually unknown and they pose a challenge when attempting to understand circuit-specific biological computations, as well as the susceptibility to pathological insults. In addition, the forms of plasticity at these synapses remain unclear and the limited studies into plasticity available did not distinguish between these two pathways.

Plasticity drives the organization of cortical maps during development, and it underlies learning and memory (reviewed in Malenka and Bear, 2004). The most extensively studied forms of plasticity are long-term potentiation (LTP) and long-term depression (LTD) of synaptic transmission. Spike timing-dependent plasticity (STDP) is a Hebbian form of long-term synaptic plasticity found in all known nervous systems, from insects to humans. Indeed, STDP appears to be a synaptic mechanism that underlies circuit remodelling during development, and it is thought to be involved in learning and memory (Dan and Poo, 2004; Feldman, 2012). In STDP, the relative order and millisecond timing of pre- and postsynaptic action potentials (spikes/APs) determine the direction and magnitude of the synaptic changes. Thus, spike timing- dependent LTP (t-LTP) occurs when a postsynaptic spike follows a presynaptic spike within ms, whereas spike timing-dependent LTD (t-LTD) is induced when this order is reversed (Feldman, 2012). In addition, in vivo studies have suggested modifications to LTP and LTD after mice undertake behavioural tasks (Kemp and Manahan-Vaughan, 2004; Clarke et al., 2010). However, it is surprising that while plasticity of EC-DG granule cell (GC) synapses has been studied using classical protocols (Froc et al., 2003; Harney et al., 2006; Hayashi et al., 2014, Savtchouk et al., 2019), no information exists as to whether STDP or t-LTD can be induced at these synapses.

Here we studied STDP at PP-GC synapses, differentiating the lateral and medial pathways, determining whether these individual synapses are subject to t-LTD and if this form of plasticity at these synapses influences learning, memory or information processing. As a result, we identified two different forms of presynaptic t-LTD at LPP- and MPP-GC synapses. Unlike LPP-GC synapses, the t-LTD at MPP-GC synapses requires ionotropic NMDARs that contain GluN2A subunits, and the t-LTD at each type of synapse require different group I mGluRs, with LPP-GC synapses dependent on mGluR5 and MPP-GC t-LTD requiring mGluR1. In addition, we found that both forms of t-LTD require postsynaptic calcium, endocannabinoid (eCB) synthesis, CB₁ receptor (CB₁R) signalling, astrocytes and glutamate. Hence, we describe here two novel forms of presynaptic t-LTD in the brain at EC-GC synapses with different action mechanisms, probably implicated in the different aspects of information processing.

Materials and Methods

Animals and ethical approval

All animal procedures were conducted in accordance with the European Union Directive 2010/63/EU regarding the protection of animals used for scientific purposes, and they were approved by the Ethics Committee at the Universidad Pablo de Olavide and that of the Andalusian Government. C57BL/6 mice were purchased from Harlan Laboratories (Spain) and P13–21 (P, post-natal day) mice of either sex were used. Animals were kept at temperatures between 18 and 24 °C on a continuous 12 h light/dark cycle, and at 40–60% humidity, with ad libitum access to food and water. In some experiments, dominant-negative (dn) SNARE mice (Pascual et al., 2005) of the same age were used. These mice were not fed with doxycycline from birth, the supplement that drives transgene expression. In these mice, the human glial fibrillary acidic protein (hGFAP) promoter mediates the specific expression of the tetracycline transactivator (tTA) in astrocytes, which in turn activates the tetracycline responsive element operator and drives the cytosolic expression of VAMP2/synaptobrevin II in these cells, along with the enhanced green fluorescence protein (eGFP). Expression of the dnSNARE transgene interferes with the formation of the SNARE complex, blocking exocytosis and impairing vesicle release by astrocytes (Sultan et al., 2015).

Slice preparation

Hippocampal slices were prepared as described previously (Andrade-Talavera et al., 2016; Pérez-Rodríguez et al., 2019; Falcón-Moya et al., 2020). Briefly, mice were anesthetized with isoflurane (2%), decapitated, and their whole brain was removed and placed in an ice-cold ACSF solution (pH 7.2, 300 mOsm*L^-1) containing (in mM): 126 NaCl, 3 KCl, 1.25 NaH₂PO₄, 2 MgSO₄, 2 CaCl₂, 26 NaHCO₃, and 10 glucose. Slices (350 µm thick: Leica VT1000S Vibratome) were maintained oxygenated (95% O₂/5% CO₂) in the same solution for at least 1 h before use. Experiments were carried out at 33-34 °C and during the experiments; the slices were superfused continuously with the same solution indicated above.

Electrophysiological recordings

Whole-cell patch-clamp recordings of DG GCs were obtained under visual guidance by infrared differential interference contrast microscopy. The identity of the neurons was verified by their characteristic voltage response to a current step protocol applied in current-clamp configuration using a patch- clamp amplifier (Multiclamp 700B), and acquiring the data with pCLAMP 10.2 software (Molecular Devices). Patch-clamp electrodes were pulled from borosilicate glass tubes and had a resistance of 4-7 MΩ when filled with the intracellular solution consisting in(in mM, pH 7.2–7.3, 290 mOsm L⁻¹): potassium gluconate, 110; HEPES, 40; NaCl, 4; ATP-Mg, 4; and GTP, 0.3. Only cells with a stable resting membrane potential below −55 mV were assessed, and the cell recordings were excluded from the analysis if the series resistance changed by >15%. All recordings were low-pass filtered at 3 kHz and acquired at 10 kHz. For plasticity experiments, excitatory postsynaptic potentials (EPSPs) were evoked alternately at 0.2 Hz in two input pathways, test and control, by using two monopolar stimulating electrodes placed 200-400 µm from the cell soma. Stimulating electrodes were situated on fibres contacting the distal and medial third of the dendritic arbour of GCs, corresponding to LPP- and MPP-GC synapses, respectively. Stimulation was adjusted (200 µs, 0.1-0.2 mA) to obtain an EPSP peak amplitude of 3-5 mV in control conditions. Pathway independence was ensured by the lack of cross-facilitation when the pathways were stimulated alternately at 50 ms intervals. Plasticity was assessed through the changes in the EPSP slope, measured in its rising phase as a linear fit between time points corresponding to 25-30% and 70-75% of the peak amplitude under control conditions. Miniature responses were recorded in the presence of tetrodotoxin (TTX, 300 nM).

Plasticity protocols

After establishing a stable basal EPSP over 10 min, the test input was paired 100 times with a single postsynaptic spike. The single postsynaptic spike was evoked by a brief somatic current pulse (5 ms, 0.1–0.5 pA), whereas the control pathway was left unstimulated during the pairing period. To induce t-LTD, a postsynaptic AP was evoked 18 ms before the onset of the EPSP. EPSP slopes were monitored for at least 30 min after the pairing protocol, and the presynaptic stimulation frequency remained constant throughout the experiment. In some experiments, “glutamate puffs” were applied using a Picospritzer (Parker Hannifin), expulsing glutamate dissolved in the external solution through a micropipette over an astrocyte in proximity to the LPP-GC or MPP-GC synapses at a pressure of 10 psi for 50–200 ms. In each experiment, 100 glutamate puffs were applied to the recording neuron at 0.2 Hz at baseline, 18 ms before the onset of the EPSP, which did not affect patch clamping. The EPSP slopes were then monitored for 30 min.

Drugs

The following agents were purchased from: Sigma-Aldrich - BAPTA, Zinc chloride, and the compounds used to prepare the ACSF and current clamp internal solutions; and Tocris Bioscience - MK-801 maleate, D-AP5, TTX, PPDA, Ro 25-6981 maleate, L-glutamic acid, MPEP, LY367385, LY341495, AM251, Orlistat (THL), FK506, Nimodipine, Thapsigargin and Ifenprodil. All compounds were dissolved in distilled water except for Nimodipine, Thapsigargin, PPDA, FK506, AM251 and THL, which were dissolved in dimethyl sulfoxide (DMSO).

Data analysis

The data were analysed using the Clampfit 10.2 software (Molecular Devices) and the last 5 min of the recordings were used to estimate the changes in synaptic efficacy relative to the baseline. For the paired-pulse ratio (PPR) experiments, two EPSPs were evoked for 30 s at the baseline frequency, one at the beginning of the baseline recording (40 ms apart) and another 30 min after the end of the pairing protocol. The PPR was expressed as the slope of the second EPSP divided by the slope of the first EPSP. A Coefficient of Variation (CV) analysis was carried out on the EPSP slopes (Rodríguez-Moreno and Paulsen, 2008; Falcón-Moya et al., 2020) and the noise- free CV of the EPSP slopes was calculated as:

where σ² (EPSP) and σ² (noise) are the variance of the EPSP and baseline, respectively. The plot compares the variation in the mean EPSP slope (M) to the change in response variance of the EPSP slope (1/CV²: see Brock et al., 2020 for a comprehensive explanation). Graphs were prepared using SigmaPlot 11.0.

Statistical analysis

Before applying any statistical comparison, the data were subjected to Shapiro-Wilk normality and equal variance tests, applying a confidence interval (CI) of 95%. Unpaired or paired Student’s t-tests were used for comparisons between two groups. For multiple comparisons of more than two groups, one-way ANOVA was used followed by a Holm–Sidak post-hoc test (when necessary). All the data are expressed as the mean ± standard error of the mean (SEM) and P values were considered significant when < 0.05: *p = 0.05, **p = 0.01, ***p = 0.001.

Results

Pairing presynaptic activity with single postsynaptic action potentials at low frequency can induce t-LTD at mouse lateral and medial perforant pathway-dentate gyrus granule cell synapses

We initially determined whether spike timing-dependent LTD can be induced at perforant pathway-DG GC synapses by pairing presynaptic stimulation with single postsynaptic spikes at low frequency (0.2 Hz). We used a post-pre protocol, pairing a single postsynaptic spike followed 18 ms later by presynaptic stimulation 100 times, as described previously at cortical and hippocampal synapses (Andrade-Talavera et al., 2016; Martínez-Gallego et al., 2022a). We monitored the EPSPs evoked by extracellular stimulation of the LPP- and MPP-GC synapses, which make synaptic contacts with the distal and medial third of the dendritic arbour of hippocampal DG GCs, respectively (Petersen et al., 2013; Savtchouk et al., 2019). EPSPs were recorded from GCs using the current clamp mode of the whole-cell patch-clamp configuration (Fig. 1a, k). A post- before-pre pairing protocol induced robust t-LTD at both LPP-GC (63 ± 5 %, n = 14: Fig 1b, c) and MPP-GC synapses (59 ± 5 %, n = 14: Fig. 1l, m), while unpaired control pathways remained unaffected (103 ± 4 % and 105 ± 7 %, respectively: Fig. 1b, c, l, m).

Figure 1.

Presynaptic expression of t-LTD at lateral and medial perforant pathway-dentate gyrus granule cell synapses

We next determined the site where this t-LTD was expressed through several approaches. As we observed failures in synaptic transmission at both types of synapse, we analysed these and found that the number of failures increased consistently after t- LTD at both LPP-GC (from 43 ± 7 % at baseline to 76 ± 6 % after t-LTD, n = 8: Fig. 1d) and MPP-GC synapses (from 40 ± 5 % at baseline to 59 ± 4 % after t-LTD, n = 6: Fig. 1n), suggesting a presynaptic mechanism. We then analysed the PPR at baseline and 30 min after the pairing protocol, identifying a significant increase in the PPR after t-LTD at both LPP-GC (from 1.16 ± 0.11 at baseline to 1.67 ± 0.19 after t-LTD, n = 14: Fig. 1e) and MPP-GC synapses (from 1.17 ± 0.11 at baseline to 1.81 ± 0.15 % after t- LTD, n = 14: Fig. 1o), again suggesting a presynaptic expression of these two forms of t-LTD. Third, we estimated the noise-subtracted CV of the synaptic responses before and after t-LTD induction. A plot of CV^-2 versus the change in the mean evoked EPSP slope (M) before and after t-LTD mainly yielded points below the diagonal line at LPP- GC and MPP-GC synapses (Fig. 1f, p), consistent with a modification of the release parameters (Rodríguez-Moreno and Paulsen, 2008; Brock et al., 2020). Finally, we recorded and analysed the miniature responses (mEPSP) in the presence of tetrodotoxin (300 nM, TTX), before and after t-LTD, adding TTX at the baseline and washing it out before performing the t-LTD experiment, and then adding TTX again after t-LTD (Falcón-Moya et al., 2020). Under these conditions, a t-LTD similar to previous experiments was evident at both types of synapse (LPP-GC synapses 55 ± 11 %, n = 6; MPP-GC synapses 67 ± 8 %, n = 7), and we found that the frequency of mEPSPs decreased at both types of synapse after t-LTD induction: LPP-GC - baseline 1.4 ± 0.13 Hz, after t-LTD induction 0.7 ± 0.09 Hz, n = 6 (Fig. 1g, i); and MPP-GC - baseline 1.2 ± 0.1 H, after t-LTD induction 0.7 ± 0.07 Hz, n = 8 (Fig. 1q, s). No effect of TTX on mEPSP amplitude was evident at any of these synapses: LPP-GC - baseline 1.13 ± 0.09, after t-LTD 1.13 ± 0.02 mV, n = 5 (Fig. 1h, j); MPP-GC - baseline 1.03 ± 0.07, after t- LTD 1.00 ± 0.09 mV, n = 6 (Fig. 1r, t). These results again suggest a presynaptic locus for t-LTD expression. Hence, together these results are consistent with a decrease in the probability of neurotransmitter release and they are indicative of a presynaptic locus for these two forms of t-LTD at both types of synapse.

Presynaptic t-LTD requires NMDARs at MPP-GC but not at LPP-GC synapses

Different forms of t-LTD have been shown to require NMDARs at cortical synapses and in the hippocampus (Bender et al., 2006; Nevian and Sakman, 2006; Rodríguez-Moreno and Paulsen, 2008; Rodríguez-Moreno et al., 2011, 2013; Andrade- Talavera et al., 2016). To determine whether the two forms of t-LTD studied here require NMDARs, we treated slices with the NMDAR antagonists D-AP5 (50 µM) or MK-801 (500 µM), completely blocking NMDARs and NMDAR-mediated currents (from 57 ± 8 pA, n = 5). In these experimental conditions, t-LTD at LPP-GC synapses was not affected by these antagonists (D-AP5 66 ± 4 %, n = 8; MK-801 62 ± 9 %, n = 6; interleaved controls 60 ± 6 %, n = 8: Fig. 2a, b), whereas t-LTD at MPP-GC synapses was blocked in slices treated with either D-AP5 (90 ± 7 %, n = 10) or MK-801 (107 ± 10 %, n = 6) relative to the interleaved controls (60 ± 7 %, n = 8: Fig. 2e, f). When the experiment was repeated by loading the postsynaptic cell with MK-801 (500 µM-1 mM), t-LTD at MPP-GC synapses was not affected (54 ± 7 %, n = 9) relative to the interleaved controls (60 ± 7 %, n = 8: Fig. 2e, f). These results indicate that while t-LTD at MPP-GC synapses requires NMDARs, t-LTD at LPP-GC synapses does not. Moreover, the NMDARs required for t-LTD at MPP-GC synapses are ionotropic and not postsynaptic, consistent with data from neocortical synapses where presynaptic NMDARs are required for t-LTD (Sjöström et al., 2003; Bender et al., 2006; Nevian and Sakmann, 2006; Rodríguez-Moreno and Paulsen, 2008).

Figure 2.

Presynaptic NMDARs containing GluN2B and GluN3A subunits have been found at MPP-GC synapses and they modulate glutamate release (Jourdain et al., 2007; Savtchouk et al., 2019). To determine the subunit composition of the NMDARs involved in t-LTD at MPP-GC synapses, we first repeated the experiments in slices treated with Zn²⁺ (300 nM), an antagonist of NMDARs containing GluN2A subunits (Bidoret et al., 2009, Andrade-Talavera et al., 2016; Prius-Mengual et al., 2019) that prevented t-LTD (95 ± 5 %, n = 7: Fig. 2g, h). We also repeated the experiments on slices in the presence of Ro25-6981 (0.5 µM), an antagonist of NMDARs containing GluN2B subunits (Hu et al., 2009; Mares et al., 2021). This antagonist did not affect t- LTD (55 ± 11 %, n = 6: Fig. 2g, h) and neither did an additional antagonist of NMDARs containing GluN2B subunits, ifenprodil (3 µM, 62 ± 11 %, n = 6: Fig. 2g, h). Finally, t- LTD was not affected in the presence of PPDA (10 µM, 61 ± 7 %, n = 6: Fig. 2g, h) relative to the pooled interleaved controls (63 ± 6 %, n = 9), an antagonist of NMDARs that contain GluN2C/2D subunits (Chen et al., 2021; Wang et al., 2020). These results indicate that the presynaptic NMDARs involved in t-LTD induction at MPP-GC synapses probably contain GluN1/GluN2A subunits.

Presynaptic t-LTD at lateral and medial perforant pathway-dentate gyrus granule cell synapses requires mGluRs

As mGluRs have been implicated in plasticity and t-LTD in different brain regions, and at distinct synapses (Gladding et al., 2009), and they may modulate synaptic transmission at PP-DG synapses (Dietrich et al., 1997), we tested whether the presynaptic forms of t-LTD at LPP- and MPP-GC synapses require mGluRs. Exposure to the broad-spectrum mGluR antagonist LY341495 (100 µM) completely blocked t- LTD at both LPP-GC (100 ± 10 %, n = 8: Fig. 2c, d) and MPP-GC synapses (99 ± 9 %, n = 10: Fig. 2i, j). Indeed, the mGluR5 antagonist MPEP (20 µM) prevented t-LTD at LPP-GC (103 ± 10 %, n = 8: Fig. 2c, d) but not at MPP-GC synapses (61 ± 7 %, n = 6: Fig. 2i, j), whereas the specific mGluR1 antagonist LY367385 (100 µM) did not affect t-LTD at LPP-GC synapses (76 ± 3 %, n = 6: Fig. 2c, d) but it did block t-LTD at MPP- GC synapses (95 ± 7 %, n = 7: Fig. 2i, j) relative to their respective interleaved controls (LPP-GC synapses 68 ± 7 %, n = 9; MPP-GC synapses 65 ± 7 %, n = 9: Fig. 2c, d, i, j). Hence, both LPP-GC and MPP-GC synapses require mGluR activity to induce these presynaptic forms of t-LTD, but whereas t-LTD at LPP-GC synapses requires mGluR5, mGluR1 receptors are required at MPP-GC synapses.

Postsynaptic Ca²⁺ is required for t-LTD at lateral and medial perforant pathway- dentate gyrus granule cell synapses

Both t-LTP and t-LTD require postsynaptic Ca²⁺ at neocortical and hippocampal synapses (Bender et al., 2006; Nevian and Sakmann, 2006; Rodríguez-Moreno et al., 2013; Andrade-Talavera et al., 2016; Inglebert et al., 2021; Martínez-Gallego et al., 2022a), which led us to investigate the postsynaptic Ca²⁺ requirements of t-LTD at PP- DG synapses. As L-type Ca²⁺ channels have previously been implicated in plasticity (Nevian and Sakmann, 2006; Andrade-Talavera et al., 2016; Wiera et al., 2017; Falcón- Moya et al., 2020; Martínez-Gallego et al., 2022a), we assessed whether they are involved in the t-LTD studied here by performing the pairing protocol after bath application of the L-type Ca²⁺ channel blocker, nimodipine (10 µm). This channel blocker did not impede t-LTD induction at LPP-GC (79 ± 4 %, n = 6: Fig. 3a, b) or MPP-GC synapses (67 ± 6 %, n = 6: Fig. 3e, f), indicating that L-type Ca²⁺ channels are not required for these forms of t-LTD. The Ca²⁺ involved in synaptic plasticity may also be released from intracellular stores (Bender et al., 2006; Nevian and Sakmann, 2006; Andrade-Talavera et al., 2016; Martínez-Gallego et al., 2022a) and indeed, t-LTD was prevented at both LPP-GC and MPP-GC synapses when thapsigargin (10 µM) was present in the superfusion fluid (LPP-GC synapses 108 ± 16 %, n = 6; MPP-GC synapses108 ± 6 %, n = 6), a drug that depletes intracellular Ca²⁺ stores (Singh et al., 2021). A similar effect was witnessed when thapsigargin (10 µM) was loaded into the postsynaptic neuron (LPP-GC synapses 109 ± 12 %, n = 9; MPP-GC synapses 107 ± 5 %, n = 7) relative to untreated interleaved controls (LPP-GC synapses 66 ± 7 %, n = 9; MPP-GC synapses 67 ± 7, n = 9: Fig. 3a, b, e, f). Hence, Ca²⁺ release from the intracellular stores appears to be necessary for t-LTD at both LPP and MPP-GC synapses.

Figure 3.

Postsynaptic calcineurin is required for t-LTD induction at LPP-GC but not MPP-GC synapses

To gain better insight into the mechanisms involved in t-LTD, we hypothesized that a Ca²⁺-dependent enzyme might be involved. Since the Ca²⁺-dependent protein phosphatase calcineurin has been implicated in synaptic plasticity in the hippocampus (Mulkey et al., 1994) and neocortex (Torii et al., 1995; Rodríguez-Moreno et al., 2013; Andrade-Talavera et al., 2016), we tested the effect of inhibiting calcineurin on t-LTD. Accordingly, t-LTD induction was prevented at LPP-GC (110 ± 7 %, n = 8) but not MPP-GC synapses (72 ± 7 %, n = 6: Fig. S1) when the calcineurin blocker FK506 (10 µM) was added to the superfusion fluid, relative to the untreated interleaved controls (LPP-GC 66 ± 8 %, n = 7; MPP-GC 71 ± 7%, n = 6: Fig. S1). Subsequently, we loaded the postsynaptic neuron at LPP-GC synapses with FK506 and as t-LTD was no longer detected at these synapses (105 ± 11 %, n = 7: Fig. S1), postsynaptic calcineurin appears to be involved in t-LTD induction at LPP-GC but not MPP-GC synapses.

Endocannabinoid signalling at lateral and medial perforant pathway-dentate gyrus granule cell synapses is required for t-LTD

Postsynaptic Ca²⁺ dynamics may affect the synthesis of eCBs like anandamide and 2-arachidonoylglycerol (2-AG), ligands of CB₁Rs (Laguerre et al., 2021) necessary for t-LTD at neocortical, striatal and hippocampal synapses in rodents (Bender et al., 2006; Andrade-Talavera et al., 2016; Cui et al., 2016; Hogrefe et al., 2022). These receptors may be expressed in neurons and/or astrocytes, where they can mediate the release of gliotransmitters like glutamate, ATP/adenosine and D-serine through exocytotic mechanisms (Covelo and Araque, 2018). The induction of t-LTD at hippocampal and neocortex synapses requires the eCB system and astrocyte signalling, which may in turn drive gliotransmitter release (Min and Nevian, 2012; Andrade- Talavera et al., 2016; Martínez-Gallego et al., 2022a). To investigate the role of eCB signalling in these forms of t-LTD, we loaded the postsynaptic neurons with tetrahydrolipstatin (THL, 5 µM) to inhibit diacylglycerol lipase activity and prevent eCB synthesis. THL prevented t-LTD induction at both LPP-GC (92 ± 7 %, n = 8 versus interleaved controls 67 ± 3% n = 8: Fig. 3c, d) and MPP-GC synapses (92 ± 5 %, n = 6 versus interleaved controls 66 ± 5%, n = 8: Fig. 3g, h). Moreover, the presence of the CB₁R antagonist AM251 (3 µM) also blocked t-LTD at both LPP-GC (102 ± 7 %, n = 14: Fig. 3c, d) and MPP-GC synapses (98 ± 5 %, n = 6: Fig. 3g, h). Thus, eCB synthesis by postsynaptic cells and CB₁R activation of is required for presynaptic t-LTD at LPP- and MPP-DG GC synapses.

Astrocyte signalling and glutamate release at lateral and medial perforant pathway- dentate gyrus granule cell synapses is required for t-LTD

Astrocytes are involved in plasticity and synaptic transmission at hippocampal and cortical synapses (Min and Nevian, 2012; Araque et al., 2014; Andrade-Talavera et al., 2016; Covelo and Araque, 2018; Pérez-Rodríguez et al., 2019; Falcón-Moya et al., 2020; Durkee et al., 2021; Martínez-Gallego et al., 2022a), and at the synapses studied here (Jourdain et al., 2007; Savtchouk et al., 2019; Pérez-Otaño and Rodríguez-Moreno, 2019). To investigate the possible involvement of astrocytes in the induction of t-LTD at the EC-DG synapses studied here, two different approaches were used. First, individual astrocytes were loaded with the Ca²⁺ chelator BAPTA (20 mM) though the patch pipette to inhibit vesicular and Ca²⁺ dependent gliotransmitter release (Parpura and Zorec, 2010), both at LPP-GC and MPP-GC synapses, and in neurons situated at 50-100 µm from the loaded astrocytes (Fig. 4a, f). In these experimental conditions, t- LTD was impaired at both LPP-GC (103 ± 9 %, n = 6: Fig. 4b, c) and MPP-GC synapses (94 ± 6 %, n = 6: Fig. 4g, h). Second, we assessed t-LTD in transgenic mice expressing a selective dnSNARE domain (dnSNARE) in astrocytes, preventing functional vesicular gliotransmitter release from these cells (Pascual et al., 2005; Sultan et al., 2015; Sardinha et al., 2017). In contrast to the typical t-LTD observed in WT mice (LPP-GC synapses 71 ± 5 %, n = 7; MPP-GC synapses 60 ± 5 %, n = 6: Fig. 4b, c, g and h), t-LTD could not be induced in these dnSNARE mice (LPP-GC synapses 102 ± 9 %, n = 9; MPP-GC synapses 112 ± 6 %, n = 7: Fig. 4c, d, g and h). These results clearly indicate that astrocytes are required to induce t-LTD at both LPP-GC and MPP- GC synapses.

Figure 4.

At these synapses it is thought that glutamate is released from astrocytes to modulate synaptic transmission and participate in plasticity (Jourdain et al., 2007; Savtchouk et al., 2019), as indicated for cortical synapses (Min and Nevian, 2012; Martínez-Gallego et al., 2022a). To determine whether glutamate, probably released by astrocytes, is required for t-LTD at LPP- and MPP-GC synapses, we repeated the experiments in slices from dnSNARE mutant mice but applying glutamate puffs (100 µM) in the proximity of the astrocytes and neurons under study rather than pairing. In these experimental conditions, t-LTD, which was not evident in untreated dnSNARE mice (LPP-GC synapses 95 ± 5 %, n = 7; MPP-GC synapses 109 ± 6 %, n = 6) was fully recovered at both LPP-GC (72 ± 11 %, n = 6: Fig. 4d, e) and MPP-GC synapses (70 ± 8 %, n = 6: Fig. 4i, j) relative to the control interleaved slices from WT mice (LPP-GC synapses 62 ± 8 %, n = 7; MPP-GC synapses 64 ± 5 %, n = 7). These results indicate that glutamate, probably released by astrocytes as gliotransmitters, are necessary for the induction of t-LTD at LPP- and MPP-GC synapses.

Discussion

In this study, two new forms of t-LTD were seen to be elicited at LPP and MPP synapses onto DG GCs by pairing presynaptic activity with single postsynaptic APs at low frequency (0.2 Hz). Using different approaches, we found that these two forms of t- LTD are expressed presynaptically, not least through a decrease in the glutamate release probability. Interestingly, t-LTD induction at MPP-GC synapses but not at LPP-GC synapses depends on presynaptic NMDARs containing GluN2A subunits. In addition, mGluRs are required for t-LTD at both types of PP-DG synapses, although LPP-GC synapses require mGluR5 activation and MPP-GC synapses require mGluR1 activity. These two forms of presynaptic t-LTD require postsynaptic Ca²⁺ released from postsynaptic intracellular stores and whereas t-LTD at LPP-GC synapses, requires calcineurin phosphatase activity; this is not the case at MPP-GC synapses. Both these two forms of t-LTD require eCB synthesis and CB₁R activation, and both require astrocytes and astrocyte signalling to promote gliotransmitter release, as well as glutamate that is probably released by astrocytes.

While different time intervals between the postsynaptic spike and presynaptic EPSP of a pairing protocol may induce t-LTD (Feldman, 2012), it is clear here that pairing EPSPs of 3-5 mV size 100 times with single postsynaptic spikes 18 ms later at 0.2 Hz induces robust LTD at both LPP- and MPP-GC synapses. Interestingly, a similar protocol induces t-LTD at hippocampal CA3-CA1 synapses (Andrade-Talavera et al., 2016; Pérez-Rodríguez et al., 2019, Falcón-Moya et al., 2020) and at L4-L2/3 synapses of the somatosensory (Rodríguez-Moreno and Paulsen, 2008; Banerjee et al., 2009; 2014; Martínez-Gallego et al., 2022a), visual (Larsen et al., 2014) and cingulate cortices (Hogrefe et al., 2022). As such, the results presented here extend the existence of t-LTD to LPP and MPP synapses onto hippocampal DG GCs. Although there are several uncertainties associated with quantal analysis in the CNS (see Brock et al., 2020), we used different approaches here to determine the locus of expression of these two forms of t-LTD (failure rate, PPRs, CV, and mEPSP frequency and amplitude analysis), all of which were consistent with presynaptic changes.

NMDARs are required for general synaptic plasticity and for STDP, and in different experimental models (cell cultures, rodent brain slices and human neurons) they are required for t-LTD and t-LTP at different synapses in the hippocampus, neocortex, cerebellum and striatum (Bi and Poo, 1998; Bender et al., 2006; Nevian and Sakmann, 2006; Banerjee et al., 2009, 2014; Fino et al., 2010; Verhoog and Mansvelder, 2011; Kwag and Paulsen, 2012; Andrade-Talavera et al., 2016; Sgritta et al., 2017; Louth et al., 2021; Martínez-Gallego et al., 2022a). NMDARs have been implicated in synaptic plasticity at MPP-GC but not LPP-GC synapses through classical protocols (Dahl et al., 1990; Min et al., 1998; Beck et al., 2000; Hayashi et al., 2014). Our results are consistent with the presence of NMDARs at MPP-GC synapses, in our case participating in LTD. Indeed, t-LTD is not present at MPP-GC synapses in the presence of D-AP5, and the presence of presynaptic NMDARs at MPP-GC synapses was demonstrated previously by immunogold labelling (Jourdain et al., 2007; Savtchouk et al., 2019). At these synapses t-LTD was also blocked when MK801 was present in the bath but not when it was loaded into the postsynaptic neuron, indicating that the NMDARs necessary for t-LTD are not postsynaptic, as seen in L4-L2/3 synapses of the somatosensory (Bender et al., 2006; Nevian and Sakmann, 2006, Rodríguez-Moreno and Paulsen, 2008; Rodríguez-Moreno et al., 2011, 2013) and visual cortex (Larsen et al., 2014). Since the NMDA channel blocker MK801 prevents t-LTD, the NMDARs required for t-LTD appear to be ionotropic, although an additional role for a postsynaptic metabotropic NMDARs cannot be ruled-out (Nabavi et al., 2013). Thus, the involvement of NMDARs is one important difference between LPP- and MPP-GC synapses that may account for at least some of the physiological differences described between these two pathways. By using antagonists with different NMDAR subunit specificities, we also found that the NMDARs required for t-LTD at MPP-GC synapses contain GluN2A but not GluN2B or GluN2C/D subunits. NMDARs that enhance glutamate release or that participate in LTP at PP-GC synapses are thought to contain GluN2B subunits in rats as their effect is blocked by ifenprodil (Jourdain et al., 2007; Savtchouk et al., 2019). However, ifenprodil or Ro25698 had no effect on t-LTD here, such that the preNMDARs containing GluN2B subunits that participate in LTP and enhance glutamate release may differ from those participating in t-LTD, and there may even be differences between the species (i.e.: rat and mouse). Thus, while both forms of t-LTD described here appear to require group I mGluRs, they require different ones: mGluR5 at LPP-GC synapses and mGluR1 at MPP-GC contacts. Nevertheless, the exact location of these mGluRs is not clear at present, which may represent a further difference between LPP- and MPP-GC synapses, and with important behavioural and physiological consequences.

Both, t-LTD and t-LTP require postsynaptic Ca²⁺ at neocortical (Bender et al., 2006; Nevian and Sakmann, 2006; Rodríguez-Moreno et al., 2013; Martínez-Gallego et al., 2022a) and hippocampal synapses (Andrade-Talavera et al., 2016; Falcón-Moya et al., 2020; Mateos-Aparicio and Rodríguez-Moreno, 2020; Ingebert et al., 2021). Moreover, like other forms of t-LTD, that at LPP- and MPP-GC synapses requires a rise in Ca²⁺ in the postsynaptic cell due to Ca²⁺ release from internal stores, as also seen in other situations of hippocampal, neocortical and cerebellar t-LTD (Bender et al., 2006; Nevian and Sakmann, 2006; Andrade-Talavera et al., 2016; Sgritta et al., 2017; Falcón- Moya et al., 2020; Martínez-Gallego et al., 2022a). This is similar to the LTD induced by LFS in DG GCs (Wang et al., 1997). Our data also implies an involvement of calcineurin in t-LTD induction at LPP-GC synapses but not at MPP-GC synapses, an additional striking difference between these two types of synapses. Protein phosphatases, including calcineurin, may be required for different forms of LTD, both in the hippocampus (Mulkey et al., 1994) and neocortex (Torii et al., 1995). Future experiments will determine the exact mechanism by which calcineurin mediates t-LTD at LPP but not at MPP synapses. What is clear for the moment is that while the requirements for Ca²⁺ seem to be similar at both LPP-GC and MPP-GC synapses, the Ca²⁺-dependent protein phosphatase calcineurin is only involved in t-LTD at LPP-GC synapses. Significantly, the functional consequences of this difference remain to be determined. Our results indicate that eCBs and CB₁Rs are involved in the induction of these two different forms of t-LTD, in agreement with previous field recordings showing that the eCB system is involved in synaptic plasticity at PP-DG synapses (Wang et al., 2016; Peñasco et al., 2019; Fontaine et al., 2020). Here we did not elucidate the location of the CB₁Rs mediating t-LTD, such that further work will be required to address this question. However, the demonstration that astrocytes are required for t-LTD is suggestive that astrocytic CB₁Rs might be involved.

Through two different approaches (introducing BAPTA into astrocytes and using mutant dnSNARE mice) we demonstrated the involvement of astrocytes in presynaptic t-LTD, both at LPP- and MPP-GC synapses. Astrocytes participate in synaptic potentiation at PP-GC synapses (Jourdain et al., 2007), although little is known about the contribution of astrocytes to LTD. Thus, this is the first demonstration of the participation of astrocytes in LTD at these synapses. Interestingly, and as found in the somatosensory cortex (Min and Nevian, 2012), glutamate is required to mediate t-LTD induction at both types of PP-GC synapses, and in fact, t-LTD can be recovered by the local application of glutamate at both LPP- and MPP-GC synapses in slices from dnSNARE mice that do not develop t-LTD otherwise. These findings agree with previous studies showing that astrocytes can release glutamate that reaches the NMDARs located in the presynaptic boutons at PP-GC synapses (Bezzi et al., 2004; Jourdain et al., 2007), influencing synaptic transmission at MPP-GC synapses (Savtchouk et al., 2019). It is interesting to note that the presynaptic targets of the glutamate at least in part released by astrocytes are not the same at LPP- and MPP-GC synapses, as they are mGluRs in the case of LPP-GC synapses, and both NMDARs and mGluRs in the case of MPP-GC synapses. While it is possible that glutamate from the presynaptic neuron also activates preNMDARs and mGluRs, the amount of glutamate released by presynaptic terminals may be insufficient to drive t-LTD, and hence, glutamate released from astrocytes may be required for t-LTD.

What is the physiological role of these forms of plasticity? For the moment, the exact role of STDP in the perforant cortex or in the hippocampus is not known, and more work is necessary to specifically determine functions for t-LTP and t-LTD. STDP seems a good candidate to mediate spatial learning in the hippocampus (Bush et al., 2010) and the possible role of t-LTP and t-LTD in forms of learning involving the hippocampus and the perforant cortex will be addressed in future studies. Our studies were done in developing P13-P21 animals indicating its relevance during maturation. The functions of t-LTP and t-LTD during development are most probably related to the refinement of synaptic connections and remodelling of neuronal circuits (Andrade- Talavera et al., 2023). As a Hebbian learning rule, t-LTP should occur when the spike order is pre-before-post, strengthening those connections in which the presynaptic neuron takes part in firing the postsynaptic cell, as predicted by Hebb, whereas t-LTD occurs when the spike order is reversed, so that non-causal spiking weakens the connections involved, possibly as a first step in the elimination of those connections during development as has been suggested (see Caporale and Dan, 2008). Indeed, further studies will be necessary to determine the true influence of STDP in the perforant cortex and in the hippocampus, and the specific developmental role of t-LTD and t-LTP in these circuits. As discussed previously (Pérez-Rodríguez et al., 2019; Falcón-Moya et al., 2020), t-LTD most probably participates in refining synapses, potentially weakening excitatory synapses that are underused or behaviorally irrelevant (Buonomano and Merzenich, 1998; Feldman and Brecht, 2005).

It is interesting to note that with these new results it is clear that both forms of t- LTD may require glutamate from astrocytes. Presynaptic plasticity may involve structural changes and may change the short-term properties of neurotransmitter release, participating in circuit computations, and changing the excitatory/inhibitory balance or sensory adaptations (Monday et al., 2018). Why some synapses, like L4-L2/3 synapses at somatosensory cortex (and as observed in the hippocampus at CA3-CA1 synapses), show pre- and/or post-synaptic plasticity requires further study. Interestingly, in the synapses studied here, t-LTD of LPP-GC and MPP-GC synapses have different requirements, indicating that the presynaptic expression of plasticity is fundamental for the correct functioning of brain circuits and it is possible that different presynaptic forms of t-LTD from synapses onto the same neuron as pre- and postsynaptic forms of LTD (Banerjee et al., 2009, 2014) are regulated differently. Presynaptically expressed forms of t-LTD may in fact control the trial-to-trial reliability, and they may induce a larger change in signal-to-noise ratio than postsynaptic changes alone, as described in auditory cortex (Froemke et al., 2013). Broad-ranging input representations from the EC transmitted via the PP are sparse, and the DG processes this information and converts it into a less well-correlated output to become less similar and more specific (Borzello et al., 2023). Whether t-LTD at LPP and MPP synapses described here are involved in this important process require further research and the different cellular and molecular mechanisms involved in t-LTD at LPP- and MPP-GC synapses may account for some of the difference in their physiological influence. Notwithstanding this, the potential behavioural influence of the presynaptic forms of LTD studied here is still an emerging issue of particular interest in the near future.

Acknowledgements

We thank Dr Joao Oliveira for providing the dnSNARE mice. This work received support from the Spanish Agencia Estatal de Investigación and Fondo Europeo de Desarrollo Regional (FEDER, Grants PID2019-107677GB-I00 and PID2022-136597NB-I00), and from the Junta de Andalucía and Fondo Europeo de Desarrollo Regional (FEDER, Grant P20-0881) to A. R-M. H. C-C was supported by a Visiting Scholarship from the ESN, and I. M-G was supported by a FPU Fellowship from the Spanish Agencia Estatal de Investigación.

Author contributions

H. C-C and I. M-G. performed the experiments and analysed the results. A. R-M. designed and supervised the research, and wrote the paper.